Address correspondence and reprint requests to Professor Gennaro Schettini, Pharmacology and Neuroscience Institute for Cancer Research, c/o Advanced Biotechnology Center, Largo Rosanna Benzi, 10, 16132 Genoa, Italy. E-mail: firstname.lastname@example.org
Chemokines represent key factors in the outburst of the immune response, by activating and directing the leukocyte traffic, both in lymphopoiesis and in immune surveillance. Neurobiologists took little interest in chemokines for many years, until their link to acquired immune deficiency syndrome-associated dementia became established, and thus their importance in this field has been neglected. Nevertheless, the body of data on their expression and role in the CNS has grown in the past few years, along with a new vision of brain as an immunologically competent and active organ. A large number of chemokines and chemokine receptors are expressed in neurons, astrocytes, microglia and oligodendrocytes, either constitutively or induced by inflammatory mediators. They are involved in many neuropathological processes in which an inflammatory state persists, as well as in brain tumor progression and metastasis. Moreover, there is evidence for a crucial role of CNS chemokines under physiological conditions, similar to well known functions in the immune system, such as proliferation and developmental patterning, but also peculiar to the CNS, such as regulation of neural transmission, plasticity and survival.
Since their discovery in 1987, when a new factor with chemotactic activity for neutrophils (later named interleukin IL 8, now called CXCL8) was identified, the chemokine family has grown and more than 50 chemokines that interact with at least 20 receptors have been identified to date (Walz et al. 1987; Yoshimura et al. 1987). Chemokines are a group of structurally and functionally related proteins that exert their biological activity by binding to cell surface receptors in many instances through interaction with sulfated proteins and proteoglycans. Chemokines have been highly conserved during evolution, as indicated by their expression in numerous different species of mammals, birds and fish, and by the amino acid similarities across species. These proteins are produced by many cellular sources and their actions cover a wide range of functions.
Chemokines were initially recognized to play a role in leukocyte communication and migration, both in physiological and pathological contexts. These molecules control immune cell trafficking and recirculation of the leukocyte population between the blood vessels, lymph, lymphoid organs and tissues, a process important in host immune surveillance, and in acute and chronic inflammatory responses. Dysregulation of chemokine expression is associated with chronic inflammatory conditions such as arthritis, atherogenesis, inflammatory bowel syndrome and glomerulonephritis (Feng 2000; Murdoch and Finn 2000; Baggiolini 2001).
However, it has now become evident that chemokines not only play a fundamental role in development, homeostasis, and immune system function, but are also important in the angiogenesis and angiostatic processes, tumor and metastasis progression, and in the CNS (Bajetto et al. 2001b; Belperio et al. 2000; Rossi and Zlotnik 2000).
Numerous viruses, including members of the herpesvirus, poxvirus, retrovirus and lentivirus families, encode molecular analogs of chemokines and chemokine receptors, able to mimic their action, presumably to help them to overcome the immune response (Murphy 2001).
Here we review recent studies on chemokines and chemokine receptors, with particular emphasis on their physiopathological role in the CNS.
Chemokines and chemokine receptors: introduction on their structure and classification
Chemokines are small (8–14 kDa) molecules, mainly basic, that are structurally and functionally related to form a family of proteins subdivided into four groups on the basis of the relative position of their first N-terminal cysteine residues. The CC or β-chemokines, and the CXC or α-chemokines, are the largest groups and contain four conserved cysteines. In the CC family the first two cysteines are adjacent, while in the CXC family they are separated by one amino acid. The CX3C or δ-chemokines include only one member, called CX3CL1/fractalkine, in which the first two cysteines are separated by three amino acids; this molecule exists in both soluble and membrane-bound forms (Bazan et al. 1997; Pan et al. 1997). Finally, the C or γ-chemokines are represented by two chemokines, called XCL1/lymphotactin-α and XCL2/lymphotactin-β, that contain only two of the four conserved cysteines found in the other subfamilies (Kelner et al. 1994; Kennedy et al. 1995; Yoshida et al. 1995) (Table 1). The CXC chemokine group can be further subclassified into those that contain a conserved tripeptide motif glutamic acid–leucine–arginine (ELR) at the N-terminal of the protein, before the CXC domain, and those that do not. The ERL motif is not simply structural but appears to be linked to function, giving specificity for neutrophil chemotaxis and angiogenesis (Strieter et al. 1995a; Strieter et al. 1995b).
Table 1. Chemokine classification according to the new nomenclature
In general, chemokine subfamilies show similar, often overlapping, specificity. Chemokines belonging to the CC family attract monocytes, basophils, eosinophils and Tlymphocytes, but have little or no action on neutrophils; CXC-ELR-negative chemokines attract lymphocytes and monocytes but are poor chemoattractants for neutrophils, while the CXC-ELR-positive chemokines act essentially on neutrophils. In addition, a more recent chemokine classification distinguishes between ‘inflammatory’, also called inducible, chemokines and ‘homeostatic’ or constitutive chemokines (Moser and Loetscher 2001).
Chemokines exert their biological activity by binding to cell surface receptors that belong to the superfamily of seven-transmembrane domain receptors that signal through coupled heterotrimeric G-proteins. Although multiple chemokines can often bind the same receptor and a single chemokine can bind several receptors, the chemokine–chemokine receptor interactions are almost always restricted within a single subclass. Chemokine receptor nomenclature is therefore based on the chemokine group to which their ligand(s) belong. However, monogamous chemokine–receptor relationships exist; those confirmed by knockout studies are CXCL12/SDF-1–CXCR4 and CX3CL1/fraktalkine–CX3CR1, while other monogamous interactions, yet to be formally demonstrated, appear to be the couples CCL20/LARC–CCR6, CCL25/TECK–CCR9, CXCL13/BCA-1–CXCR5 and CXCL16–CXCR6 (Fig. 1). It should be underlined that these monogamous ligand–receptor relationships are provisional.
To date, six CXC receptors have been identified, named from CXCR1 to CXCR6, and 11 CC receptors, CCR1 to CCR11, along with a single receptor for fractalkine and one for lymphotactin α/β, called CX3CR1 and XCR1, respectively.
Despite the structural similarity and the apparent redundancy, the chemokine receptors differ in their capacity to activate signal transduction pathways. Their stimulation can lead to diverse responses, as expression of the pleiotropic biological effects carried out by these proteins (Fig. 2).
The most characteristic response is the stimulation of cell migration, which appears to require functional coupling of the receptor to Gαi as migration is completely inhibited by treatment with Bordetella pertussis toxin (PTX), even if the Gαi itself may not be necessary for cell migration (Neptune et al. 1999). In fact, the bulk of the signal of chemokine receptors is reported to be carried out by the βγ subunits of the G-protein, and in particular it appears that only the βγ subunits released from Gi-coupled receptors, but not those released by the Gs- or Gq-coupled receptors, can mediate cell migration (Neptune and Bourne 1997).
Chemokine receptor stimulation causes the rapid activation of phospholipase C, which leads to inositol-1,4,5-triphosphate generation and transient elevation of cytosolic Ca2+ concentrations. This is one of the most characteristic effects of chemokine stimulation and is used to test the responsiveness of chemokine receptors to different chemokines (Bokoch 1995; Premack and Schall 1996; Bajetto et al. 1999; Thelen 2001). Chemokine-mediated activation of phospholipase C not only results in inositol-triphosphate production, but also leads to the formation of diacylglycerol and subsequent activation of protein kinase C (Bokoch 1995; Premack and Schall 1996). However, protein kinase C activation is not a characteristic event in chemokine receptor transduction, but is generally induced by the activation of most surface receptors.
Regulation of other ion fluxes after chemokine receptor activation has also been described (Hegg et al. 2000; Liu et al. 2000). In microglial cells, the activation of CCR3 by CCL5/RANTES or CCL11/eotaxin induced a rapid calcium influx from the extracellular environment that was sensitive to the dihydropyridine derivative nifedipine (Hegg et al. 2000). Although this observation suggests a modulation of L-type voltage-sensitive calcium channels, the lack of voltage-dependent ion fluxes in microglial cells indicates that this effect is probably mediated by a different class of yet unidentified nifedipine-sensitive channels (Hegg et al. 2000). The stimulation of CCR5 and CXCR4 by CCL4/MIP-1β and CXCL12/SDF-1, respectively, has been observed to induce the activation of both calcium-activated potassium channels and chloride channels in primary cultures of human macrophages (Liu et al. 2000).
Other important players in chemokine signaling are phosphoinositide 3-kinases (PI-3Ks) and the subsequent activation of Akt, the mitogen-activated protein kinase (MAPK) cascade, especially the pathway involving extracellular signal-regulated kinase (ERK) 1/2 activation, as well as the phosphorylation of the cytoskeletal-associated kinases focal adhesion kinase and protein tyrosine kinase 2 (Ganju et al. 1998a; Ganju et al. 1998b; Sotsios et al. 1999; Tilton et al. 2000; Bajetto et al. 2001a).
Through the G-protein βγ subunits, chemokines can directly stimulate PI-3Kγ with consequent phosphatidyl inositol (PIP3) formation and Akt activation, as demonstrated by the lack of PIP3 production and Akt activation in response to chemokine stimulation in leukocytes isolated from mice in which the PI-3Kγ subunit was deleted (Sasaki et al. 2000). Neutrophils from these animals also showed a marked attenuation, but not complete inhibition, of chemotaxis invitro (Hirsch et al. 2000). Other studies demonstrated that the total inhibition of PI-3Ks did not completely block chemokine-induced chemotaxis, suggesting the existence of another signal transduction pathway linked to chemotaxis (Thelen 2001).
The janus activated kinase–signal transducer and activator of transcription (JAK/STAT) pathway is also involved in both α- and β-chemokine receptor signaling. Activation of different STAT transcription factors in T cells through CCR2 in response to stimulation with CCL2/MCP-1, and through CCR5 in response to CCL5/RANTES, has been reported, while CXCL12/SDF-1 can activate, through CXCR4, the JAK/STAT pathway in a PTX-independent manner (Mellado et al. 1998; Wong and Fish 1998; Vila-Coro et al. 1999; Wong et al. 2001).
Mellado et al. (2001) reported that chemokines induce receptor homo–heterodimerization which results in the activation of the JAK/STAT pathway (Rodriguez-Frade et al. 1999; Vila-Coro et al. 1999; Vila-Coro et al. 2000), and that receptor heterodimerization may increase the cell sensitivity to chemokine stimulation. However, the activation of this pathway by chemokines may not be a general principle, and its occurence could depend both on the receptor and on the cell type involved. Alhough described in hematopoietic progenitors and T cells, CXCR4 signaling through STAT proteins was not found in astrocytes (Han et al. 2001), while nuclear translocation of STAT in response to RANTES has been observed in both T cells and astrocytes (Bakhiet et al. 2001). Moreover, it has been reported that in an ovarian cancer cell line, CXCR1/2 stimulation by CXCL8/IL-8 induced epidermal growth factor receptor phosphorylation, indicating that a ‘cross-talk’ between chemokine and growth factor signaling pathways exists. Such cross-talk has been reported for other G-protein-coupled receptors whose signal transduction is linked to tyrosine kinase receptors, such as epidermal growth factor receptor and platelet-derived growth factor receptor (Venkatakrishnan et al. 2000).
Recently, an important role for protein tyrosine phosphatases (PTPs) in the regulation of signals generated by chemokines has emerged. It has been reported that src homology 2 domain-containing phosphatase (SHP)1, a PTP expressed in normal and injured CNS, besides hematopoietic cells, and src homology 2 domain-containing inositol phosphatase (SHIP) act as negative regulators of chemokine signaling, in contrast to the ubiquitously expressed PTP, SHP2, that appears to enhance chemokine signaling (Kim et al. 1999a; Kim et al. 1999b; Chernock et al. 2001).
The chemotactic responses to CXCL12/SDF-1, CXCL12/SDF-1-dependent actin polymerization and ERK1/2 activation were all enhanced in macrophages, T and B cells derived from SHP1-deficient mice, and chemotaxis, calcium influx and actin polymerization were enhanced in response to CXCL12/SDF-1 in both immature and mature hematopoietic cells derived from SHIP-deficient mice (Kim et al. 1999a; Kim et al. 1999b). CXCL12/SDF-1 stimulation of Tcells enhanced the tyrosino-phosphorylation of SHP2 that is constitutively associated with the CXCR4 receptor (Chernock et al. 2001). In addition, the β-chemokine receptor CCR5 signals through SHP1 and SHP2 (Ganju et al. 2000).
Chemokine and chemokine receptors expression in the brain: physiological roles
Chemokines have been associated with the trafficking of leukocytes in physiological immune surveillance as well as with inflammatory cell recruitment in different diseases including those that develop in the CNS, but little is known about their physiological roles in the CNS. The expression of chemokines and chemokine receptors has been described in the CNS by several authors through in vitro and in vivo studies (Table 2). Among the chemokine receptors, several members of the CCR family (all except CCR6 and CCR7) have been reported to be expressed at the CNS level, as well as all the CXCR members (CXCR1–CXCR6) and CX3CR1. In addition, several reports describe the expression of the relative CXC, CC and CX3C chemokines. The cells expressing and producing chemokines and chemokine receptors in the CNS are represented by astrocytes, microglial cells, oligodendrocytes, neurons and endothelial cells (for a review see Dorf et al. 2000; Gebicke-Haerter et al. 2001). Their expression has been described in the brain in several pathological conditions such as inflammatory and neurodegenerative diseases (multiple sclerosis (MS), experimental autoimmune encephalitis, Alzheimer's disease (AD), acquired immune deficiency syndrome (AIDS) dementia complex (ADC), brain injury and tumors), but also in normal brain and, in particular, during brain development.
Table 2. Reports on constitutive chemokine receptor expression in the CNS and in CNS-derived cells
This table is not intended to give an exhaustive review of the literature on chemokine receptor expression in the CNS. For brevity only the first report on the expression of a chemokine receptor in a given context was taken into consideration. aH, human; Mac, macaque; M, mouse; R, rat. bA, astrocytes; E, brain endothelium; M, microglia; N, neurons; NE, embryonic neuroepithelium; O, oligodendrocytes; SC, spinal cord.
The significance of the presence of these molecules in the brain, although still poorly understood, has begun to emerge in the past few years. A few studies tend to ascribe a classic chemotactic activity to neural cells (Bolin et al. 1998; Hesselgesser et al. 1998; Lazarini et al. 2000), but the involvement of chemokines in directing CNS cell migration during development remains to be elucidated, with the exception of strong evidence concerning CXCL12/SDF-1 and its receptor CXCR4, emphasized by studies in vivo in knockout mice. The lack of either of these molecules results in an anomalous cerebellar development, characterized by the disruption of the regular laminar architecture, probably due to a premature and disorganized inward migration of external granular layer cells (Nagasawa et al. 1996; Ma et al. 1998; Zou et al. 1998). Other aspects of cerebellar development appeared to be normal in these mice, including the development and structure of the Bergmann glia (the cerebellar radial glial cells), which have an important role in the migration of external granular layer cells (Ma et al. 1998; Zou et al. 1998). In addition, CXCL12/SDF-1 is abundantly and selectively expressed in the developing and mature brain, and the developmental pattern of expression supports a role for this molecule during migration and neurogenesis. At birth, CXCL12/SDF-1 is expressed in cerebellum and olfactory bulbs, and progressively decreases within 2 weeks, in contrast to other brain regions (cortex, thalamus and hippocampus) where its expression, low at birth, progressively increases during the first 2 weeks postnatally. The spatiotemporal distribution of CXCL12/SDF-1 transcripts correlates with granule cell migration across the molecular layer in the cerebellum (Tham et al. 2001).
Chemokines may also be involved in the regulation of cell proliferation in the developing CNS. Some studies have reported that chemokines induce proliferation in glial cells in vitro which, in some cases, has been correlated with their expression pattern. CXCL12/SDF-1 induces invitro astrocyte proliferation through ERK1/2 activation, and its expression, in the rat brain, correlates with that of CXCR4, suggesting paracrine and/or autocrine signaling (Bajetto et al. 2001a; Tham et al. 2001). CXCL1/GROα enhances in vitro the proliferative response of immature spinal cord oligodendrocyte precursors to their major mitogen, platelet-derived growth factor, but has no effect on more mature oligodendrocyte precursors, and its expression pattern in postnatal spinal cord, developmentally regulated, is consistent with the pattern of emergence of oligodendrocytes (Robinson et al. 1998). Similarly, the ability of CCL5/RANTES to differentially promote proliferation or survival in human fetal forebrain astrocytes, depending on the age of the fetuses from which the cells were isolated, suggested that this chemokine could be involved in regulation of the expansion of astrocytes in the earlier phases of development, and then promotion of differentiation and maintainance of the astrocyte population (Bakhiet et al. 2001). Finally, recombinant CCL2/MCP-1 and CCL3/MIP-1α induce proliferation in glial cultures (Rezaie et al. 2002).
A growing body of evidence suggests that chemokines participate in the regulation of neuronal signaling in various ways. The first suggestions came from the observation that transgenic mice expressing high levels of KC (the murine homolog of CXCL1/GROα) frequently developed a progressive neurological dysfunction, characterized by ataxia, postural instability and rigidity. These mice did not show significant damage to neurons, myelin or axons, so the syndrome could be explained only by hypothesizing direct receptor-mediated effects (Tani et al. 1996). Indeed it was later demonstrated that in mouse Purkinje neurons CXCL1/GROα, as well as CXCL8/IL-8, through the activation of CXCR2, is able not only to generate calcium transients, but also to enhance synaptic activity by increasing neurotransmitter release and to suppress the induction of long-term depression (LTD) (Giovannelli et al. 1998). The same authors noted that CXCL12/SDF-1 also induces an enhancement in the spontaneous synaptic activity and a slow inward current in Purkinje neurons; these effects were reduced by ionotropic glutamate receptor antagonists, but not by tetrodotoxin (TTX), and thus were most likely the consequence of extrasynaptic glutamate released by surrounding cells (Limatola et al. 2000). The observation that CXCL12/SDF-1 stimulated the release of glutamate from astrocytes, and influenced the synaptic activity independently of the presence of its receptor on neurons, confirmed these hypotheses. This chemokine was then proposed to intervene in the communication between glia and neurons, in particular in the glia-mediated regulation of synaptic transmission (Bezzi et al. 2001). However, an inhibitory effect on neuronal transmission has also been noted: several CC and CXC chemokines and soluble CX3CL1/fractalkine were able to reduce calcium oscillations in synaptically coupled hippocampal neurons in vitro by decreasing glutamate release from the presynaptic neuron (Meucci et al. 1998), while CXCL8/IL-8 reduces calcium currents in cholinergic neurons through the G-protein-mediated inhibition of L- and N-type calcium channels (Puma et al. 2001).
Taken together, these findings indicate that some chemokines could be included in the large group of molecules that act as neuromodulators; their mechanism of action and their cellular target can vary significantly, since they act both at presynaptic and postsynaptic levels, as well as on surrounding glia, mostly through the regulation of neurotransmitter release, but also through direct modulation of ion channel activity.
Chemokines and chemokine receptor involvement in brain pathologies
MS is a demyelinating disease associated with an autoimmune response directed against myelin proteins within the CNS. MS lesions are characterized by chronic inflammation, with a progressive immune-mediated destruction of the myelin sheath and a recruitment of immune cells into the CNS, mainly represented by T cells and monocytes (Bar-Or et al. 1999). The inflammatory process and the infiltration of the immune cells into the CNS are occur in response to chemotactic signals such as chemokines.
Several groups have investigated the role of chemokines in MS pathogenesis and in experimental autoimmune encephalomyelitis (EAE), an animal model of MS. Chemokine expression in MS has been observed both in the CSF andby in situ and immunohistochemical analysis, in MSlesions.The presence of CCL2/MCP-1, CCL8/MCP-2 and CCL7/MCP-3 was described in acute and chronic active MS lesions from autopsied brains (McManus et al. 1998). In addition to CCL2/MCP-1, the β-chemokines CCL5/RANTES, CCL3/MIP-1α, and CCL4/MIP-1β were observed in MS lesions. CCL3/MIP-1α expression was predominantly associated with glial cells, CCL4/MIP-1β with macrophages/microglia, CCL5/RANTES with perivascular leukocytes and CCL2/MCP-1 with macrophages and astrocytes in chronic active MS lesions (Simpson et al. 1998; Balashov et al. 1999). Consistent with these findings is the presence of the cognate receptors CCR2, CCR3 and CCR5 on foamy macrophages, activated microglia in the lesions. CCR2 and CCR5 were also present on infiltrating T cells, while CCR3 and CCR5 were expressed on reactive astrocytes (Balashov et al. 1999; Simpson et al. 2000b). The α-chemokines also appear to have a role in the pathogenesis of MS. CXCL10/IP-10 and CXCL9/Mig, in addition to CCL5/RANTES, have been detected in the CSF during MS attacks. Expression of the corresponding chemokine receptors CXCR3 and CCR5 was found in lymphocytes in the perivascular inflammatory infiltrate, and in lymphocytes, macrophages and microglia in MS lesions, respectively (Balashov et al. 1999; Sorensen et al. 1999; Simpson et al. 2000a).
The EAE is considered an excellent animal model for MS. EAE is induced by immunization with antigens derived from myelin, such as proteolipidic protein and myelin oligodendrocyte glycoprotein (MOG). The inflammatory state generated in this model is characterized by multifocal perivascular CNS inflammatory infiltrates primarily consisting of T cells, both antigen specific and antigen non-specific, and monocytes, with little or no polymorphonuclear cells. Using this model, several studies have indicated that the expression of CCL5/RANTES, CCL3/MIP-1α, CCL4/MIP-1β, CCL2/MCP-1, CXCL10/IP-10 mRNA and proteins correlated with the inflammatory lesions (Berman et al. 1996; Hulkower et al. 1993; Ransohoff et al. 1993; Godiska et al. 1995; Glabinski et al. 1997; Jiang et al. 1998; Kennedy et al. 1998).
The expression and biological significance of CCL3/MIP-1α and CCL2/MCP-1 during acute and relapsing EAE have been studied in vivo using neutralizing antibodies. These studies suggest that CCL3/MIP-1α controls mononuclear cell accumulation during acute EAE, while CCL2/MCP-1 controls mononuclear cell infiltration during relapsing EAE (Karpus et al. 1995; Kennedy et al. 1998).
The involvement of the chemokines found to be mainly expressed in MS and EAE in the pathogenesis of EAE has been further demonstrated by the generation of mice deficient in CCL3/MIP-1α, CCL2/MCP-1, and CCR1, CCR2 and CCR5 (Fife et al. 2000; Izikson et al. 2000; Rottman et al. 2000; Tran et al. 2000). CCR2-deficient mice appear to be resistant to EAE, fail to develop mononuclear infiltrates and display a decreased antigen-specific pro-inflammatory response in the secondary lymphoid organs (Fife et al. 2000; Izikson et al. 2000). In agreement with these observations, mice deficient in CCL2/MCP-1 are resistant to MOG-induced EAE (Huang et al. 2001), while CCL3/MIP-1α- and CCR5-deficient mice are not resistant to the MOG-induced disease, although CCR1-deficient mice had less severe clinical signs (Rottman et al. 2000; Tran et al. 2000). The existence of apparently contradictory findings in chemokine expression patterns and responses to MOG in different animal models could reflect heterogeneity in the pathogenesis of MS.
Recently two groups have provided further evidence for the importance of CXCL10/IP-10 in the pathogenesis of MS, using neutralizing antibodies into two different animal models: intracerebral infection of mice with mouse hepatitis virus, which results in an acute encephalomyelitis followed by a chronic demyelination disease with clinical and histological similarities with MS, and the EAE model (Fife et al. 2001; Liu et al. 2001). Antibodies against CXCL10/IP-10 decreased the clinical and histological incidence of disease and the accumulation of inflammatory mononuclear cells during the pathogenesis of EAE, as well as in the mouse hepatitis virus model (Fife et al. 2001; Liu et al. 2001).
AD is a common form of dementia characterized by the progressive loss of selected neuronal populations, the presence of abnormal neurons containing tangles of tau protein and the formation of extracellular β-amyloid (Aβ) peptide plaque deposits that are surrounded by activated astrocytes and microglia. Aβ peptides have been shown to be potent activators of microglia and macrophages. Although abnormal or excessive migration of inflammatory cells into the CNS in AD is not so relevant as in other neurological disorders, inflammation and leukocyte infiltration clearly occurs in pathologically vulnerable regions of the AD brain (Akiyama et al. 2000). Damaged neurons, insoluble Aβ peptide deposits and neurofibrillary tangles can behave as stimuli for inflammation. There is growing evidence that chemokines and their receptors are up-regulated in resident CNS cells in the AD brain, and that chemokines may contribute to plaque-associated inflammation and neurodegeneration. Both α- and β-chemokines have been reported to be expressed in the AD brain.
The expression of CXCR2, CXCR3, CCR2 and CCR5 and their ligands has been observed in AD tissues. Xia et al. (1997) reported that CXCR2 is expressed in hippocampal and cortical neurons in normal brain and also in dystrophic neurites that are associated with a subset of senile plaques. CCL2/MCP-1 has been localized to mature senile plaques and reactive microglia, but was not found in immature senile plaques (Ishizuka et al. 1997). CCR3 and CCR5 were present on microglia of both control and AD brains, with an increased expression in reactive microglia in AD. CCL3/MIP-1α was reported to be constitutively expressed at a low level by neurons and microglia, whereas MIP-1β was predominantly expressed by a subpopulation of reactive astrocytes. Many CCR3- and CCR5-reactive microglial cells and CCL4/MIP-1β-reactive astrocytes were found to be associated with amyloid deposits (Xia et al. 1998). The expression of CXCR3 by neurons and neuronal processes in various cortical and subcortical regions has been reported, while its ligand, CXCL10/IP-10, has been detected in a subpopulation of astrocytes in the normal brain and is markedly elevated in the AD brain (Xia et al. 2000). In addition it was reported, by in vitro studies, that the CXCR3 ligands CXCL10/IP-10 and CXCL9/Mig activated the ERK1/2 pathway in mouse cortical neurons (Xia et al. 2000). Other in vitro studies have focused on the ability of Aβ peptides to contribute to inflammation and leukocyte infiltration in the AD brain by stimulating chemokine production. Aβ peptide stimulation induced the production and release of CCL2/MCP-1 and CCL5/RANTES by astrocytes and oligodendrocytes, and CXCL8/IL-8, CCL2/MCP-1, CCL3/MIP-1α and CCL4/MIP-1β expression in human monocytes (Meda et al. 1995; Johnstone et al. 1999; Akiyama et al. 2000; Prat et al. 2000). Thus the chemokines produced by the plaque-surrounding cells could play a role in the recruitment and accumulation of astrocytes and microglia in senile plaques and in the associated inflammation.
Brain ischemia and trauma
Stroke and trauma are two pathological conditions that elicit robust inflammation in the brain. Increasing evidence indicates that ischemic brain injury secondary to arterial occlusion is characterized by acute local inflammation. Reperfusion following transient ischemia is a clinical condition that often causes greater tissue damage than persistent ischemia, which is characterized by leukocyte infiltration into the damaged brain. Chemokines have been shown to play a crucial role in leukocyte accumulation into ischemic lesions. In experimental cerebral ischemia, generated by occlusion of common carotid artery and middle cerebral artery, the expression of chemokines such as CXCL1/KC, CXCL10/IP-10, CCL2/MCP-1, CXCL8/IL-8, CCL7/MCP-3, CCL3/MIP-1α, CCL4/MIP-1β and the chemokine receptor CXCR3 have been observed by several authors (Che et al. 2001; Wang et al. 1995; Matsumoto et al. 1997; Wang et al. 1998; Gourmala et al. 1999; Wang et al. 1999; Wang et al. 2000). In addition, it has been reported that anti-CXCL8/IL-8 neutralizing antibodies significantly reduced brain edema and infarct size, suggesting that chemokines could be considered as novel potential therapeutic targets for stroke and neurotrauma (Matsumoto et al. 1997).
Chemokines play another important, although indirect, role in the pathogenesis of stroke, as there is accumulating evidence that these proteins have a pivotal role in atherosclerosis. CXCL8/IL-8, CXCL12/SDF-1, CXCL10/IP-10, CCL1/I-309 and CXCR2 have all been associated with atherosclerotic lesions in animal models, and in particular CCL2/MCP-1 and CCR2 (Mach et al. 1999; Abi-Younes et al. 2000; Boisvert et al. 2000; Haque et al. 2001). It has been reported that CCL2/MCP-1-deficient mice have less arterial lipid deposition in hypercholesterolemia models such as low-density lipoprotein receptor deficiency or apoB over-expression (Gosling et al. 1999; Gu et al. 1998). Similarly, CCR2–/– mice showed a reduction in disease in an apoE deficiency model (Boring et al. 1998; Dawson et al. 1999).
Traumatic injuries in the CNS are characterized by activation of astrocytes and migration of mononuclear inflammatory cells in the CNS parenchyma, and are most frequently studied using a rodent brain stab wound animal model.
Several groups have reported that an increase in CCL2/MCP-1 expression follows brain injury. It is produced by reactive astrocytes after both stab injury, nitrocellulose stab or implants in adult mice (Glabinski et al. 1996). CCL2/MCP-1 is also the only chemokine produced after sterile injury, in contrast to CCL3/MIP-1α, CCL4/MIP-1β, CCL5/RANTES, and CXCL10/IP-10 whose expression was increased in the brain after cortical injury and addition of endotoxin (Hausmann et al. 1998). Temporal and regional distribution of CCL2/MCP-1 expression following a visual cortical lesion demonstrated the presence of CCL2/MCP-1 in the thalamus before any sign of retrograde neuronal degeneration (Muessel et al. 2000).
Peripheral nerve axotomy induces an increase in CX3CL1/fractalkine and CX3CR1 in facial motor nucleus with parallel cellular and morphological changes in microglia (Harrison et al. 1998). However, the hypothesized role of CX3CL1/fractalkine as messenger between injured neurons and microglia has been ruled out by studies on knockout mice, in which the absence of CX3CR1 did not result in impaired activation, proliferation, differentiation and recruitment of microglial cells (Jung et al. 2000). In contrast, in the CNS the expression of CX3CL1/fractalkine mRNA was unaffected by stimuli including induction of EAE or experimental cerebral ischemia in vivo, although in vitro this chemokine was released from cell membranes after excitotoxic stimulus (Chapman et al. 2000).
The relevance of chemokines in the pathology of AIDS emerged when CCL5/RANTES, CCL3/MIP-1α, CCL4/MIP-1β and CXCL12/SDF-1 were found to act as suppressive factors of human immunodeficiency virus (HIV) 1 infection (Cocchi et al. 1995; Bleul et al. 1996; Oberlin et al. 1996). Among the receptors for these chemokines, CCR5 and CCR3 are key co-receptors for M tropic viruses, while CXCR4 is the co-receptor for T tropic viruses. It appears that these receptors are required for viral entry, with CD4 interactions with the HIV-1 coat protein gp120 altering its conformation, allowing co-receptor binding and subsequent fusion with the membrane and HIV-1 entry into the cells (Bleul et al. 1996; Cocchi et al. 1995; Feng et al. 1996; Oberlin et al. 1996). Other receptors, such as CCR2b, CCR8, CXCR6 and CX3CR1, have similar HIV co-receptor functions, but their role in HIV infection is less prominent, suggesting that viral envelop proteins have lower affinity for these receptors (Murphy 2001). However, the role of chemokines in AIDS does not seem to be limited to virus targeting and entry. The ADC identifies a progressive encephalopathy that occurs in 25–50% of HIV-infected children and adults, and is characterized by cognitive, motor and sensory impairment (Kaul et al. 2001). Neuronal death in patients with ADC is not the consequence of a viral infection of neurons, since HIV-1 in the CNS only infects microglia and macrophages, and, non-productively, astrocytes and oligodendrocytes. Thus, the neuronal damage is probably the result of an indirect mechanism mediated by cells that have been infected or indirectly activated. Brains from patients with ADC show increased levels of several factors, including arachidonic acid metabolites, free radicals, nitric oxide, platelet-activating factor, pro-inflammatory cytokines tumor necrosis factor-α, α- and β-interferon, IL-1β and IL-6, that may be released from activated microglia and astrocytes, that are known to be a potential cause of neuronal death (Kaul et al. 2001). Some viral proteins shed by infected cells, such as the viral envelope glycoprotein gp120 and the regulatory protein Tat, seem to be the best candidates as initiators of the cascade process that leads to neurotoxicity, as demonstrated by several in vitro and in vivo studies (Bagetta et al. 1996; Scorziello et al. 1997; Bonavia et al. 2001; Kaul et al. 2001). Transgenic mice that constitutively express gp120 in astrocytes show brain damage similar to that observed in patients with ADC, with gp120 expression sufficient to induce pathological effects on neurons, astrocytes and microglia (Toggas et al. 1994).
However, the exact mechanism by which gp120 causes neuronal damage is still unknown, though some findings indicate that the neurotoxic process could arise from the gp120–chemokine receptor interaction. This hypothesis is supported by the ability of gp120 to bind and, more importantly, to generate an intracellular signal through chemokine receptors, even independently of CD4 (Davis et al. 1997; Hesselgesser et al. 1997; Meucci et al. 1998; Scorziello et al. 1998; Bajetto et al. 1999; Zheng et al. 1999b; Lazarini et al. 2000). Gp120 could cause an abnormal stimulation of chemokine receptors, or, alternatively, interfere with the physiological function of chemokines inthe brain. Among the chemokine receptors, CXCR4, expressed in the CNS by neurons, astrocytes, microglia and endothelial cells (Bajetto et al. 1999; Tanabe et al. 1997b; Ohtani et al. 1998; Westmoreland et al. 1998), has been considered the best candidate as mediator of gp120 toxicity by several authors, although there is still discussion as to whether gp120 exerts its toxicity directly by binding CXCR4 on neurons, or indirectly by binding the receptor expressed on glial cells and thus triggering neurotoxin release. Meucci et al. (1998) have shown that several chemokines, including CCL22/MDC, CCL5/RANTES, CXCL12/SDF-1α and soluble fractalkine, block gp120IIIB (T-tropic strain)-induced apoptosis in hippocampal neurons; they proposed that gp120 interferes with the trophic function of chemokines by competing at their binding sites, at least for CXCL12/SDF-1α. In contrast, Kaul and Lipton (1999) reported that CCL5/RANTES and CCL4/MIP-1β prevented apoptosis induced by gp120SF2 (another T-tropic strain) in mixed cerebrocortical cultures, while CXCL12/SDF-1α/β not only failed to protect from gp120, but in fact induced neurotoxicity on its own. Other authors have reported that the over-stimulation of CXCR4 by gp120 and progeny virions, as well as by its natural ligand CXCL12/SDF-1, can induce apoptosis in human and rat neurons in vitro and in vivo (Hesselgesser et al. 1998; Zheng et al. 1999a; Corasaniti et al. 2001). This apparently ambiguous role of CXCL12/SDF-1 might be understood in the light of the discovery that CXCR4 activation on astrocytes induces a rapid and dose-dependent release of glutamate through autocrine/paracrine secretion of tumor necrosis factor-α and prostaglandin synthesis (Bezzi et al. 2001). A perturbation of this pathway could lead to the release of toxic amounts of glutamate, as expected in glial–neuronal cultures over-stimulated by CXCL12/SDF-1 or gp120. Indeed, the authors reported that glutamate release could be induced by gp120, and this effect, as well as gp120-induced neuronal death, that were both strongly enhanced by the presence of activated microglia, was inhibited by the CXCR4 blocker.
Neoplasia represents another important pathological condition in which chemokines play a very important role through multiple mechanisms. Chemokines can influence tumor cell proliferation, regulate the angiogenic/angiostatic process, control cell migration and metastasis, and finally regulate the infiltration of immune cells into the tumor mass.
CXCL8/IL-8, one of the most studied chemokines, has been shown to contribute to human cancer progression, acting as mitogenic, angiogenic and motogenic factor (for areview see Xie 2001). CXCL8/IL-8 expression has been described in several human cancer tissues and cell lines, and this chemokine can act as an autocrine growth factor in different tumors including human melanoma, pancreatic cancer, colon cancer and malignant mesothelioma (Xie 2001). Its expression is increased in glioma after anoxia, and the h-CXCL8/IL-8 antisense expression vector inhibited the growth of human glioma cells in vitro (Yamanaka et al. 1995; Desbaillets et al. 1999). CXCL8/IL-8 was identified, by RT–PCR and histologically, within tumor cells in human pituitary adenomas and in activated and neoplastic astrocytes (Kasahara et al. 1991; Suliman et al. 1999). In addition, several studies have demonstrated that CXCL8/IL-8 shows angiogenic properties, and is involved in both vascular endothelial cell proliferation and tumor neovascularization (Xie 2001).
Pathological angiogenesis is a hallmark of cancer, as tumors require the presence of constant neovascularization to guarantee an adequate supply of oxygen and nutrients. Angiogenesis is regulated by an opposing balance of angiogenic and angiostatic factors (Belperio et al. 2000). CXC chemokines can be either angiogenic or angiostatic factors. Chemokines containing an ELR motif before the first cysteines, including CXCL1/GROα, CXCL5/ENA-78, CXCL7/NAP-2 and CXCL8/IL-8, show angiogenic activity through interaction with CXCR2. In contrast, the ELR-negative CXC chemokines CXCL9/MIG-1 and CXCL10/IP-10 exert an angiostatic activity. CXCL12/SDF-1 has been described to induce vascularization, despite the lack of the ELR motif (Salcedo et al. 1999; Belperio et al. 2000). Recently, it has been reported that CCL2/MCP-1, a member of the CC chemokine family, can also act as a direct mediator of angiogenesis (Salcedo et al. 2000). Chemokines may have therapeutic effects through their angiostatic action or by boosting the immune response against the tumor through their ability to induce migration of T, natural killer cells, dendritic cells and macrophages. It has been reported, for example, that CXCL10/IP-10 and CXCL9/MIG have shown an anti-tumor action generated by both the inhibition of angiogenesis and an increased T cell recruitment. These chemokines are induced by γ-interferon and are believed to be responsible, at least in part, for the anti-tumor effects ofIL-12 through a γ-interferon-dependent mechanism (Kanegane et al. 1998; Tannenbaum et al. 1998). Other chemokines that have shown anti-tumor activity include CCL7/MCP-3, CCL3/MIP-1α, CCL2/MCP-1 and CCL1/I-309 (Rossi and Zlotnik 2000).
Astrocytomas, glioblastomas and meningiomas have all been shown to express CCL2/MCP-1, a chemokine isolated in the supernatants of a cultured glioma cell line whose inhibition by antibodies was able to block monocyte chemotaxis induced by tumor fluids of glioblastomas and astrocytomas (Yoshimura et al. 1989; Leung et al. 1997). Recently, it has been reported that the expression of CX3CL1/fractalkine, a chemokine that induces chemotaxis of monocytes and cytotoxic T cells and that is largely expressed in the CNS, was induced by ectopic expression of p53, involving this chemokine in the immune control to prevent cells from undergoing malignant transformation (Shiraishi et al. 2000).
Another important chemokine involved in tumor progression is CXCL12/SDF-1. This chemokine and its receptor CXCR4 have been identified in different human tumors including those developing in the CNS as glioblastomas and neuroblastomas (Mitra et al. 1999; Koshiba et al. 2000; Geminder et al. 2001; Muller et al. 2001; Scotton et al. 2001). CXCR4 expression was up-regulated in human glioblastoma cell lines, in which its inhibition blocked cell proliferation (Sehgal et al. 1998b; Sehgal et al. 1998a). CXCR4 expression can be enhanced by cytokines such as tumor necrosis factor-α and IL-1α (Oh et al. 2001), whose levels are raised in astroglioma tumors and in some cases could be up-regulated by CXCL12/SDF-1 itself (Han et al. 2001). In addition, CXCR4 and CXCL12/SDF-1 expression has been described in human gliobastoma tissues and directly correlated with tumor malignancy grade (Rempel et al. 2000). We observed that CXCL12/SDF-1 was produced and released after lipopolysaccharide stimulation by cultured astrocytes and that this chemokine directly stimulated astrocyte proliferation, suggesting that, like CXCL8/IL-8, CXCL12/SDF-1 could act as a growth factor and that the concomitant expression of CXCR4 and CXCL12/SDF-1 should lead to autocrine and paracrine regulation of cell growth, a control mechanism that may well be altered during glioblastoma tumor progression (Bajetto et al. 1999; Bajetto et al. 2001a). CXCR4 stimulation in astroglioma cells induces expression of both CXCL10/IP-10 and CXCL8/IL-8 and enhances the constitutive expression of CCL2/MCP-1 through the ERK1/2 pathway (Oh et al. 2001). Although the real relevance of this induction to tumor progression in vivo is still to be investigated, it could be a step in a cascade process that, by virtue of the above-mentioned properties of these chemokines, could have repercussions on tumor cell proliferation, on angiogenesis and on tumor infiltration by macrophages.
In addition to their role in cell proliferation and angiogenesis, chemokines and their receptors also seem to be involved in the process of tumor cell migration, invasion and metastasis. Recently, an exciting paper by Muller et al. (2001) demonstrated a critical role for chemokines in organ- specific metastatic deposition of tumor cells. They observed that CXCL12/SDF-1 and CCL21/SLC are over-expressed in organs that are preferentially colonized by breast cancer metastasis, and that the respective receptors were expressed on the tumor cells. Further, the authors showed that they could block metastasis to regional lymph nodes and lung by blocking CXCR4 with a neutralizing antibody in vivo. Although the metastatic process is probably more complex and other molecules may be involved, these results indicate that chemokines and their receptors have a pivotal role in the invasion of tumor cells into other tissues. A role for CXCL12/SDF-1 in the metastatic process has been alsoproposed in neuroblastoma and ovarian cancer based on invitro studies (Geminder et al. 2001; Scotton et al. 2001).
Kaposi sarcoma-associated herpesvirus or human herpes virus 8 has provided additional evidence for a role of chemokines in neoplastic transformation. OFR74 of this virus shares a high degree of similarity with CXCR2 and encodes for a constitutively activated viral chemokine receptor. When transfected into human fibroblasts, it showed increased cellular proliferation and enhanced expression of vascular endothelial growth factor, a potent angiogenic factor (Bais et al. 1998). The relevance of this receptor to Kaposi sarcoma development has been demonstrated by the presence of Kaposi sarcoma-like lesions in transgenic mice expressing the receptor within hematopoietic cells (Yang et al. 2000).
Chemokines are key factors in the immune response as chemoattractants, and they show the same function in many pathological contexts in the CNS where they regulate leukocyte traffic across the blood–brain barrier. They are important regulators of immune system cell development and differentiation, and a similar role has been proposed, and in some cases demonstrated, on neurons and glia. In addition to these similarities with their function outside the nervous system, chemokines show brain-specific activities. Studies on their constitutive physiological functions in the CNS suggest an important role for cytokines in brain migration and patterning, and regulation of neuronal signaling; at present this is a field open to further development. Numerous reports indicate that they behave as double-edged weapons in neurodegenerative processes both as promoters of neuronal survival and as potential mediators of excitotoxic insults. Further studies on brain chemokines could significantly contribute to the knowledge of diseases such as ADC, AD, MS and brain tumors, and provide the basis for the development of new therapeutic approaches to their treatment.
This work was supported by Istituto Superiore Sanità (ISS) AIDS grant 30C.68, Associazione Italiana Ricerca Sul Cancro (AIRC) 2001 and Progetti finalizzati Ministero della Sanità (MISAN) 2001 to GS. SB was supported by a fellowship from Fondazione Italiana Ricerca Sul Cancro (FIRC). The authors thank Dr Douglas M. Noonan for the helpful discussion of this manuscript.