Integrins are cell surface glycoproteins that mediate cell-cell and cell-extracellular matrix (ECM) interactions (reviewed in Hynes, 1992; Hemler, 1999). They regulate many aspects of cell behavior including survival, proliferation, migration, and differentiation. Integrins achieve their effects by two means: first, they provide a physical transmembrane link between the ECM and the cytoskeleton, and second, they transduce bi-directional signals across the cell membrane. Integrins are expressed as cell surface heterodimers consisting of α and β subunits, noncovalently linked. At present, 16 different α and eight different β mammalian integrin subunits have been identified, which associate to form 22 recognized αβ heterodimers and which fall into three basic classes: β1, β2, and αv (Fig. 1). Each integrin recognizes specific ligands, which are either molecules of the ECM (e.g., laminin and fibronectin) or other cell surface counter-receptors of the immunoglobulin superfamily (e.g., intercellular adhesion molecule-1 or ICAM-1). Some integrins bind only one specific ligand while others are more promiscuous, binding several different ligands, and conversely some ECM ligands also bind to multiple integrins. This system allows for a great deal of diversity but also permits specificity because the same ECM ligand can trigger distinct cellular responses via different integrins. Within the central nervous system (CNS), integrins of the β1 and αv classes are expressed on a variety of different cell types, including neurons, glial cells, meningeal cells, and endothelial cells, and this expression is subject to both regional and developmental regulation (Jones, 1996; Pinkstaff et al., 1999). β2 integrins are expressed specifically by leukocytes, and within the CNS are found on microglia, and on infiltrating leukocytes (Springer, 1990; Jones, 1996).
Integrin Functions During Neural Development in the CNS
Historically, integrins have been shown to be critical determinants of basic morphogenesis in many developmental events including implantation, gastrulation, and formation of the cardiovascular system (Hynes, 1996). One aspect of CNS development that has been extensively studied is the migration of neurons during establishment of the cerebral cortex (reviewed in Hatten, 1999). In this process, neurons are born in the ventricular zone, and after exiting the cell cycle they attach to radial glial fibers to migrate radially into more superficial positions (Fig. 2). At the cortical plate they detach from the glial fibers and form into separate layers. Later-born neurons migrate to more superficial positions in the cerebral cortex, which results in an “inside-out” laminar organization. Several studies support the notion that β1 integrins may be important in mediating the adhesion and migration of neuronal cell populations within the developing CNS. First, neuronal migration in the developing chick tectum is markedly reduced after retroviral infection with antisense mRNAs of the β1 or α6 integrin subunits (Galileo et al., 1992; Zhang and Gallileo, 1998). Second, neuronal migration in α6 integrin null mice is also perturbed, though interestingly rather than seeing reduced cell migration, the α6 integrin null neurons actually over-migrate, and invade the cortical marginal zone, a structure they would normally not penetrate (Georges-Labouesse et al., 1998). Third, inhibition of the α3β1 integrin reduces neuronal migration along radial glia in vitro (Anton et al., 1999), and the α3β1 integrin has been identified as a cell surface receptor for the ECM protein Reelin (Dulabon et al., 2000), which is present in the marginal zone and is essential for normal cortical development (Rice and Curran, 2001). Taken together, this evidence supports a role for β1 integrins in regulating neuronal migration during CNS development, most likely by mediating the adhesive interactions between neurons and radial glial fibers.
Recently, an elegant study by Grauss-Porta et al. (2001) sought to evaluate the role of β1 integrins in neurodevelopment by deleting β1 integrins specifically from neurons and glia (Grauss-Porta et al., 2001). This was achieved by creating one strain of transgenic mice expressing the Cre recombinase under the control of the nestin promoter, and crossing these animals with another transgenic strain in which the β1 integrin subunit had been flanked by two LoxP sites. This study revealed very interesting and surprising results; development of the cerebral cortex is perturbed, but not in the way anticipated. Surprisingly, developing neurons still form adhesive interactions with the radial glial fibers, and are able to migrate along the fibers, and then detach to take up their positions in the appropriate layer. However, the most superficial layer of the cortex, the marginal zone, appears to be grossly disorganized. In the normal CNS the radial glial fibers terminate on the basement membrane produced by meningeal cells, and a specific cell population called the Cajal-Retzius cells form a chain immediately below the basement membrane, and secrete the ECM molecule Reelin (Rice and Curran, 2001). In β1-deficient mice, the glial fibers fail to form contacts with the meningeal basement membrane, and terminate at random positions throughout the different cortical layers. In addition, the linear chain of Cajal-Retzius cells is disrupted and the cells are arranged in clusters, and expression of both the basement membrane and Reelin show an abnormal discontinuous distribution. As a result, the migrating neurons invade the marginal zone at breaks in the Cajal-Retzius cell layer, to produce aberrant cellular organization in this zone. The clear message from this elegant study is that β1 integrins are important for establishing the marginal zone, in particular mediating adhesive contacts between glial endfeet and the meningeal basement membrane, and the subsequent formation of the Cajal-Retzius cell layer and the zone of Reelin. However, contrary to expectations, neuron-glial interactions and neuronal migration and positioning within the developing cerebral cortex do not require β1 integrins. In agreement with these findings, disorganization of the cerebral cortex is also seen in humans and mice with mutations in the laminin α2 chain, a component of the basement membrane (Miyagoe-Suzuki et al., 2000). Thus, glial-meningeal adhesive interactions, mediated by β1 integrins and laminin at the molecular level, are essential for normal assembly of the cerebral cortex.
Integrin Involvement in Synaptogenesis
It is well established that cell adhesion molecules such as integrins play an important role in building and maintaining synaptic structure during CNS development (Benson et al., 2000). Perhaps one of the most exciting developments in this field is the realization that integrins may be instrumental in modulating long-term potentiation (LTP), by representing a vital molecular mechanism responsible for the maintenance of synaptic changes within the adult CNS. In vitro studies on hippocampal slices have shown that peptide inhibitors of integrin-ECM interactions have no effect on the induction of LTP, but cause the LTP to decay relative to control conditions (Bahr et al., 1997). This decay is only observed when the integrin inhibitors are applied within 25 minutes of the LTP induction; after this time they have no effect (Staubli et al., 1998). This implies that integrin-mediated events are important for stabilizing LTP immediately after induction, thereby bringing about the permanent molecular changes which underlie this synaptic remodeling event. At the molecular level, integrins could be mediating this effect by changing cellular adhesive events, or by modulating intracellular signaling. Recent work on hippocampal slice cultures has shown that increased synaptic activity triggers maturation of synapses, and that this is associated with reduced glutamate release and a switch in the subunit composition of the postsynaptic N-methyl-D-aspartate (NMDA) receptors (Chavis and Westbrook, 2001). Chronic blockade of synapses with peptide inhibitors of integrins (RGD peptides) prevents both the reduction in glutamate release and the NMDA receptor subunit switch, and this effect is also observed with antibodies against the β3 integrin subunit. In addition, β3 integrins appear to be preferentially expressed on mature synapses. Therefore, integrin-mediated signaling is seemingly essential for maturation of CNS synapses. Based on these observations, it will be interesting to see whether β3 integrin-null mice (which are viable and fertile) have any deficits in learning or memory formation.
A role for integrins in mediating synaptic plasticity has also been substantiated using an entirely different approach. Using the powerful technique of random mutagenesis in Drosophila, genes were identified which when mutated, produce severe defects in learning and memory. One of these genes, Volado, encodes an integrin α subunit, also known as αPS3 (Grotwiel et al., 1998). Expression of Volado is highly enriched in mushroom bodies, synapse-dense structures that act as insect olfactory memory centers. Volado mutants show a 50% reduction in short-term memory, and this memory loss can be rescued by conditional Volado expression 3 hours before training. Further analysis has revealed that synapses in Volado mutants are structurally enlarged, suggesting that Volado negatively regulates synaptic sprouting and growth (Rohrbough et al., 2000). In addition, mutant synapses show increased evoked synaptic currents, and reduced Ca2+ dependence of transmission. Significantly, many aspects of activity-dependent synaptic plasticity in Volado mutants are reduced or absent. All abnormalities in transmission and plasticity are rescued by conditional expression of Volado, and application of peptide integrin inhibitors phenocopies all the features of mutant transmission and plasticity. Taken together with the findings from the mammalian hippocampus, these studies in Drosophila strongly suggest that integrins play an important role in learning and memory formation by affecting both synaptic architecture and functional transmission properties of the synapse.
Microglia Use Integrins as Potent Effector Molecules
The role of integrins in mediating leukocyte extravasation during CNS inflammation has been extensively studied and many excellent reviews are available on this subject (for example, Jones, 1996; Archelos et al., 1999). The principal immune effector cells within the CNS are microglial cells, and there are extensive data suggesting that integrins are important for modulating microglial behavior. Following injury or infection within the CNS, the normally quiescent microglia are transformed into highly activated cells, which migrate to the site of injury, proliferate, and phagocytose both micro-organisms and host tissue (Kreutzberg, 1996). Microglia express integrins from each of the three classes: β1, β2, and αv, and expression of specific integrins is upregulated following microglial activation, both in vitro and in vivo. In vitro experiments show that microglial activation is associated with a morphological switch from a resting ramified cell to an amoeboid form, and that this is accompanied by increased expression of the α4β1 and lymphocyte function-associated antigen (LFA-1) integrins (Hailer et al., 1996). Inflammatory cytokines and lipopolysaccharide (LPS) also induce the amoeboid morphology and increased expression of α4β1, α5β1, and Mac-1 integrins (Yu et al., 1998; Kloss et al., 2001a). In vivo, integrin expression is increased on activated microglia in Alzheimer's disease (Akiyama and McGeer, 1990), after lesion of the entorhinal cortex or facial nucleus (Hailer et al., 1997; Kloss et al., 1999), after infusion of LPS (Kloss et al., 2001a), and in multiple sclerosis lesions (Bo et al., 1996).
Evidence describing a functional role for microglial integrins has recently begun to emerge. First, inhibition of LFA-1 integrin expression by antisense oligonucleotides markedly attenuates microglial migration and activation during neuroinflammation (Ullrich et al., 2001). Second, RGD peptide inhibitors of integrins effectively block microglial phagocytosis of apoptotic neurons (Witting et al., 2000). Third, a ligand for LFA-1 has been identified on forebrain neurons, and microglia cultured on this ligand, telencephalin, exhibit clustering of LFA-1 integrins, a step leading to integrin signaling (Mizuno et al., 1999). The observation that the ECM influences microglial behavior also lends weight to the idea that integrin-mediated adhesion and signaling regulate microglial function. Laminin and fibronectin have been shown to have opposing effects on microglial morphology, phagocytosis, and the generation of nonspecific esterase and superoxide free radicals (Chamak and Mallat, 1991). In addition, synthesis and secretion of amyloid precursor protein (APP) by microglial cells is modulated by the ECM; fibronectin promotes increased secretion of APP, while laminin and collagen have the opposite effect (Monning et al., 1995). Recently, we investigated adhesive interactions between microglia and some of the potential substrates they will encounter in vivo, including astrocytes and the associated ECM (Milner and Campbell, 2002a). These studies illustrated that microglia attach well to fibronectin and vitronectin, but only weakly to laminin and astrocyte ECM, and that laminin exerts a dominant anti-adhesive effect on microglial adhesion. Furthermore, microglial adhesion to laminin and astrocyte ECM is increased by proinflammatory cytokines such as tumor necrosis factor (TNF) and interferon-gamma (IFN-γ) and this is counteracted by transforming growth factor beta-1 (TGF-β1). These cytokine effects are mediated by parallel changes in the activation state of the α6β1 integrin. In light of the evidence that laminin expression is upregulated after CNS injury (Liesi et al., 1984), these findings suggest that cytokines may regulate the extent of microglial infiltration into the site of damage, by altering integrin-mediated adhesion.
Potential Roles of Endothelial Integrins During CNS Angiogenesis and Inflammation
One exciting new area where integrins are thought to play an important role is the establishment and stabilization of the endothelium in blood vessels. Within the CNS, this structure is unique from that of other organs because it exhibits high electrical resistance, and together with astrocyte endfeet, constitutes the blood-brain barrier (BBB; Rubin and Staddon, 1999). In the adult CNS, it is well established that β1 integrins are expressed predominantly by endothelial cells lining blood vessels (Milner and Campbell, 2002b), and that this expression is tightly regulated. In many different pathological conditions, β1 integrin expression on endothelial cells is significantly altered. In animal models of cerebral ischemia, several studies have shown marked loss of the α1, α6, and β1 integrin subunits on vascular endothelial cells (Kloss et al., 2001b; Tagaya et al., 2001). During degeneration of the facial motor nucleus, blood vessels display increased expression of the α5, α6, and β1 integrin subunits (Kloss et al., 1999). In a study of tissue taken from multiple sclerosis patients, blood vessels demonstrated reduced levels of the α6 and β1 integrin subunits in acute lesions, while in chronic lesions, the levels of α6 and β1 were restored to normal levels (Sobel et al., 1998). The observation that β1 integrins are expressed at high levels by endothelial cells within the CNS, and that this expression is so tightly regulated during pathological conditions, strongly suggests that integrins are playing a critical role in regulating the function of endothelial cells within the adult CNS.
More recently, the role of integrins in blood vessel development (angiogenesis) in the CNS has been investigated, and has revealed interesting and surprising results. First, a large body of data, based largely on antibody and peptide inhibition of integrin function, has suggested that αv integrins play an essential role in angiogenesis, by promoting endothelial cell survival during blood vessel growth (Eliceiri and Cheresh, 1999). However, contrary to expectations, mice null for αv integrins show no perturbation in angiogenesis, but do show two distinct lethal phenotypes (Bader et al., 1998). Eighty percent of embryos die in utero due to a failure in implantation. The other 20% are born, but die within hours due to hemorrhage, specifically within the CNS. This study demonstrates that αv integrins are not essential for angiogenesis. More importantly, it shows that αv integrins play an essential role in contributing to the stabilization of the cerebral vascular endothelium, and that these molecules may be critical for establishing the unique properties of the BBB.
Second, another study has revealed that during CNS angiogenesis, endothelial cells markedly upregulate β1 integrin expression, and also show a clear developmental switch in the β1 integrins they express (Milner and Campbell, 2002b). In the early postnatal CNS, when angiogenesis is ongoing, endothelial cells express high levels of the α4 and α5 integrins, but very low levels of α1 and α6. Maturation of the CNS is associated with downregulation of α4 and α5 integrin expression on endothelial cells, and upregulation of α1 and α6, and this is accompanied by a parallel switch in the appropriate ECM ligands for these integrins; loss of fibronectin, and upregulation of laminin. Therefore, blood vessel maturation in the CNS is associated with a switch from fibronectin-mediated signaling early in angiogenesis, to laminin-mediated signaling at later stages. Taken together, these studies suggest that integrins are important, both for the establishment and maintenance of blood vessels within the CNS, and furthermore imply that specific integrins are important at distinct stages of this process.
Conclusions and Future Directions
Clearly, integrins and their ligands represent an important group of molecules that influence many fundamental processes within the CNS, both during development and in the adult. With the advent of more complex transgenic technology, such as the Cre/Lox system, it has now become possible to directly test the function of individual integrins within specific cell types in the CNS. Particularly exciting will be the challenge to define the mechanisms whereby integrins regulate changes in synaptic plasticity, and how this process alters with age. It will also be vital to uncover how integrins regulate angiogenesis in the CNS; this will have implications for the treatment of diseases such as ischemia (stroke) and cerebral tumors that involve abnormal blood vessel growth. In addition, it will be important to define the role that integrins play in stabilizing the BBB, with the spin-off that this knowledge may prove useful to therapeutically regulate the permeability of the BBB, both during inflammation and in scenarios requiring enhanced drug delivery to the CNS.
R.M. was the recipient of a Wellcome Trust International Prize Traveling Research Fellowship. The studies in the authors' laboratory were supported by NIH grants MH62231 and DA12444 (to I.L.C.). This is manuscript no. 14970-NP from the Scripps Research Institute.