Galectin-1: a small protein with major functions

Abstract

Galectins are a family of carbohydrate-binding proteins with an affinity for β-galactosides. Galectin-1 (Gal-1) is differentially expressed by various normal and pathological tissues and appears to be functionally polyvalent, with a wide range of biological activity. The intracellular and extracellular activity of Gal-1 has been described. Evidence points to Gal-1 and its ligands as one of the master regulators of such immune responses as T-cell homeostasis and survival, T-cell immune disorders, inflammation and allergies as well as host–pathogen interactions. Gal-1 expression or overexpression in tumors and/or the tissue surrounding them must be considered as a sign of the malignant tumor progression that is often related to the long-range dissemination of tumoral cells (metastasis), to their dissemination into the surrounding normal tissue, and to tumor immune-escape. Gal-1 in its oxidized form plays a number of important roles in the regeneration of the central nervous system after injury. The targeted overexpression (or delivery) of Gal-1 should be considered as a method of choice for the treatment of some kinds of inflammation-related diseases, neurodegenerative pathologies and muscular dystrophies. In contrast, the targeted inhibition of Gal-1 expression is what should be developed for therapeutic applications against cancer progression. Gal-1 is thus a promising molecular target for the development of new and original therapeutic tools.

Galectins: an overview

Galectins are a phylogenetically conserved family of lectins defined in 1994 as a shared consensus of amino-acid-sequences of about 130 amino acids and the carbohydrate recognition domain (CRD) responsible for β-galactoside binding (Barondes et al., 1994). Fifteen mammalian galectins have been identified to date. While some of these galectins contain one CRD and are biologically active as monomers (galectins-5, -7, -10), as homodimers (galectins-1, -2, -11, 13–14, -15) or as oligomers that aggregate though their non-lectin domain (galectin-3); others contain two CRDs connected by a short linker peptide (galectins-4, -6, -8, -9, -12). While the CRDs of all the galectins share an affinity for the minimum saccharide ligand N-acetyllactosamine—a common disaccharide found on many cellular glycoproteins—individual galectins can also recognize different modifications to this minimum saccharide ligand and so demonstrate the fine specificity of certain galectins for tissue- or developmentally-specific ligands (Ahmad et al., 2004). Location studies of galectins have established that these proteins can segregate into multiple cell compartments in function of the status of the cells in question (Danguy et al., 2002; Liu and Rabinovich, 2005). Although galectins as a whole do not have the signal sequence required for protein secretion through the usual secretory pathway, some galectins are secreted and are found in the extracellular space (Hughes, 1999). While the intracellular activity of galectin-1 (Gal-1) is mainly independent on its lectin activity, its extracellular activity is mainly dependent on it.

Gal-1: molecular structures at gene and protein levels

The first protein discovered in the family was Gal-1. As reported by the MapViewer program and the Entrez genome database on the NCBI website (http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=Genome&itool=toolbar), Gal-1 is encoded by the LSGALS1 gene located on chromosome 22q12 (Figure 1). The 0.6 kb transcript (GenBank: NM_002305) is the result of the splicing of four exons encoding a protein with 135 amino acids (GenPept : NP_002296, SwissProt: P09382) (Figure 2). Gal-1 occurs as a monomer as well as a non-covalent homodimer consisting of subunits of one CRD (Gal-1, ∼29 kDa) (Barondes et al., 1994; Cho and Cummings, 1995). Each form is associated with different biological activities, as detailed below.

The overall folding of Gal-1 involves a β sandwich consisting of two anti-parallel β-sheets (Figure 2). This jellyroll topology of the CRD constitutes the typical folding patterns of galectins. Human Gal-1 exists as a dimer in solution (Lopez-Lucendo et al., 2004). The integrity of this dimer is maintained principally by interactions between the monomers at the interface and through the well-conserved hydrophobic core, a factor which explains the observed stability of the dimer in molecular terms (Figure 2) (Lopez-Lucendo et al., 2004). Nevertheless, one of the main characteristics of the homodimeric Gal-1 protein is that it spontaneously dissociates at low concentrations (K_d ∼ 7 µM) into a monomeric form that is still able to bind to carbohydrates (Cho and Cummings, 1995), but with a lower level of affinity (Leppanen et al., 2005). In the present paper Gal-1 will be used as the general term for both the monomeric and the dimeric product of the LGALS1 gene. We will clearly state monomeric (mGal-1) or dimeric galectin-1 (dGal-1) when reporting results specific to the monomer or the dimer form respectively. Gal-1 can also exist in an oxidized form, that is, a form that lacks lectin activity (Outenreath and Jones, 1992).

The regulation of Gal-1 expression

A small region spanning the initial transcription start-up site (–63/+45) is sufficient in the promoter region of the Gal-1 gene for its transcriptional activity in mice (Chiariotti et al., 2004) (Figure 1). Both an upstream and a downstream position-dependent cis-element are necessary for efficient transcriptional activity; an additional start-up site has been mapped at position-31, and an Sp1-binding site (–57/–48) and a consensus initiator (Inr) element (which partially overlaps a non-canonical TATA box) direct RNA initiation (Chiariotti et al., 2004) (Figure 1). The upstream transcripts contribute to more than half of the Gal-1 mRNA population (Figure 1). The 5′- end of this transcript is extremely GC-rich and may fold into a stable hairpin structure which could influence translation (Chiariotti et al., 2004). The approximate position of the other putative and/or characterized regulatory elements is indicated in Figure 1 and relates to the CAAT box, to the nuclear factor kappaB-binding site (NF-κB) and to the sodium butyrate and retinoic acid (RA) response sequences. The various physicochemical agents already known as being able to modulate the expression of Gal-1 are listed in Table I. The methylation status of the Gal-1 promoter is also a very important mechanism that controls the expression of the gene (Chiariotti et al., 2004).

The subcellular distribution and the export of Gal-1

Gal-1 is present both inside and outside cells and has both intracellular and extracellular functions (Figure 3). Gal-1 shows the characteristics of typical cytoplasmic proteins as well as an acetylated N-terminus and a lack of glycosylations (Clerch et al., 1988) (Figure 2); it has been described in cell nuclei and cytosols and also translocates to the intracellular side of cell membranes (Figure 3). Nevertheless, even though Gal-1 lacks recognizable secretion signal sequences and does not pass along the standard endoplasmic reticulum/Golgi pathway (Hughes, 1999), it is well-known that it is secreted and can be found on the extracellular side of all cell membranes as well as in the extracellular matrices of various normal and neoplastic tissues (Cooper and Barondes, 1990; van den Brule et al., 1997, 2003; Clausse et al., 1999; Camby et al., 2002; Danguy et al., 2002; von Wolff et al., 2005). There is evidence that this protein is secreted in a manner similar to fibroblast growth factor-2 (FGF-2) (Nickel, 2005) via inside-out transportation involving direct translocation across the plasma membrane of mammalian cells and requiring as yet unidentified integral membrane proteins and cytosolic factors (Nickel, 2005). The β-galactoside-binding site may constitute the primary targeting motif for galectin export machinery using β-galactoside-containing surface molecules as export receptors for intracellular Gal-1 (Nickel, 2005). As a consequence there is a quality control mechanism present since the export machinery recognizes only properly folded Gal-1. The similarity of the FGF-2 and Gal-1 export pathways suggests an important role for the sodium pump (the Na⁺/K⁺–ATPase) in their export features because ouabain, a selective inhibitor of the sodium pump, inhibits these export processes (Nickel, 2005).

Gal-1 binding partners

Although galectins in general, and Gal-1 in particular, were first described as lectins that bind β-galactosides, it is now clear from the literature that as well as being a lectin, Gal-1 is also engaged in protein–protein interactions (Table II). Interestingly enough, in most cases the lectin activity of Gal-1 is observed when it is extracellular, while the protein–protein interactions of Gal-1 concern its intracellular functions. The fact nevertheless remains that one can wonder what is the real biological relevance of such a high number of Gal-1 binding partners (Table II). Binding affinity studies are warranted in order to determine the potential binding partners which are real for Gal-1 and the ones which are not because they are associated with excessively weak affinities.

Protein–carbohydrate partnering

The lectin activity of Gal-1 relates to its carbohydrate-binding site (Figure 2). Sugar binding is enthalpically driven, and this supports the notion that van der Waals contacts and hydrogen bonds constitute the main forces driving and/or stabilizing complex formations (Lopez-Lucendo et al., 2004). The published dissociation constant of dGal-1 with various glycoproteins is about 5 µM (Symons et al., 2000). Although dGal-1 binds preferentially to glycoconjugates containing the ubiquitous disaccharide N-acetyllactosamine (Gal-β1–3/4 GlcNAc also known as LacNAcII or type 2 saccharide), its binding to individual lactosamine units is characterized by relatively low levels of affinity (K_d ∼ 50 µM) (Schwarz et al., 1998; Ahmad et al., 2004). It is the arrangement of lactosamine disaccharides in multiantennary repeating chains (up to three branches) that increases the binding avidity (K_d ∼ 4 µM) (Schwarz et al., 1998; Ahmad et al., 2004). In contrast, there is no increase in avidity when the recognition unit is repeated in a string (poly-N-lactosamine) (Ahmad et al., 2004). In polysaccharides, dGal-1 does not bind glycans that lack a terminal non-reducing unmodified N-acetyllactosamine (Di Virgilio et al., 1999; Stowell et al., 2004). Although terminal galactose residues are important for dGal-1 recognition, dGal-1 binds similarly to α3-sialylated (created by ST3Gal III sialyltransferase) and α2-fucosylated (created by fucosyltransferase) terminal N-acetyllactosamine, but not to α6‐sialylated or α3-fucosylated terminal N-acetyllactosamine (Amano et al., 2003; Leppanen et al., 2005). Whether extended or otherwise, free ligands in solution bind dGal-1 with a relatively low level of affinity (Leppanen et al., 2005); in contrast, the avidity of dGal-1 for extended glycans is enhanced when it is surface-bound as on cell surfaces or in extracellular matrix (ECM) (He and Baum, 2004).

In the ECM

Gal-1 binds to a number of ECM components in a dose-dependent and β-galactoside-dependent manner in the following order: laminin > cellular fibronectin > thrombospondin > plasma fibronectin > vitronectin > osteopontin (Moiseeva et al., 2000; Moiseeva, Williams, and Samani, 2003). Laminin and cellular fibronectin are glycoproteins which are highly N-glycosylated with bi- and tetra-antennary poly-N-lactosamines (Carsons et al., 1987; Fujiwara et al., 1988). Gal-1 is also involved in ECM assembly and remodeling: it inhibits the incorporation of vitronectin and chondroitin sulphate B into the ECM of vascular smooth muscle cells (Moiseeva, Williams, and Samani, 2003). The interaction of Gal-1 with vitronectin seems to depend on vitronectin conformation since it preferentially recognizes unfolded vitronectin multimers rather than inactive folded monomers (Moiseeva, Williams, and Samani, 2003).

Cell surface-binding partners

Various membrane glycoproteins have been identified as the binding partners of Gal-1 for the mediation of cell–cell or cell–ECM adhesion (Table II). We detail below some of the major cell surface-binding partners of Gal-1.

Integrins

The activity of integrin adhesion receptors is essential for normal cellular function and survival (Frisch and Ruoslahti, 1997; Stupack and Cheresch, 2002). N-glycosylations of β-integrins regulate β1 integrin functions by modulating their heterodimerization with α chains and ligand-binding activity (Gu and Taniguchi, 2004). Numerous variants of integrin glycoforms have been described in many normal and pathological cell types. Via its direct binding to β1 integrins (without cross-linking them) dGal-1 increases the amounts of partly activated β1 integrins, but does not induce dimerization with α subunits (Moiseeva, Williams, and Samani, 2003). In the case of vascular smooth muscle cells this interaction of Gal-1 with the α1β1 integrin has been reported both as transiently phosphorylating the focal adhesion kinase (FAK) and as modulating the attachment of cells and their spreading and migration on laminin, but not on cellular fibronectin (Moiseeva et al., 1999; Moiseeva, Williams, and Samani, 2003). Gal-1 is secreted during skeletal muscle differentiation and accumulates with laminin in the basement membrane surrounding each myofiber (Gu et al., 1994). The coincidence of Gal-1 secretion with the onset of myoblast differentiation and fusion and the transition in myoblast adhesion and mobility on laminin are regulated by the interaction of Gal-1 with laminin and the α7β1 integrin (Gu et al., 1994). As a consequence Gal-1 inhibits the association of the α7β1 integrin with laminin, and is thus able to prevent and dissociate the interaction of cells with laminin in a dose-dependent fashion (Gu et al., 1994). In contrast, Gal-1 does not affect the binding of cells on fibronectin via the same α7β1 integrin (Gu et al., 1994). Fischer et al. (2005) have recently shown on a panel of epithelial cell carcinomas that Gal-1-induced growth inhibition requires functional interactions with the α5β1 integrin.

As far as other integrins are concerned, Gal-1 from mouse macrophages has been found to specifically associate with the α_Mβ2 integrin (the complement receptor 3, CR3) (Avni et al., 1998).

CD2, CD3, CD7 CD43, CD45

A number of T-cell glycoproteins from MOLT-4 and Jurkat human T cells have been shown to be specific receptors for mammalian Gal-1 binding: CD45, CD43, CD7 (Pace et al., 1999; Walzel et al., 1999; Symons et al., 2000; Fajka-Boja et al., 2002). The functions of these receptors are detailed further in the review.

GM1 ganglioside

Gal-1 is a major receptor for the carbohydrate portion of the ganglioside GM1 exposed on the surface of human neuroblastoma cells (Kopitz et al., 1998; Andre et al., 2004). Cell confluence increases the surface presentation of dGal-1. Under these circumstances Gal-1 acts as a negative growth regulator of neuroblastoma cells, though without being pro-apoptotic (Kopitz et al., 1998, 2001).

Protein–protein partnering

The proteins identified so far that interact in a carbohydrate-independent manner with Gal-1 are not structurally related to each other and do not seem to share any common domains or motifs (Table II). The galectin sites that are involved in these interactions have not yet been established.

Gemin4

Gemin4 is found in the cytoplasm as well as in the nucleus of cells both as a member of the survival of motor neuron protein (SMN) complex and as the miRNP particle (microRNA [MiRNA] ribonucleoprotein [RNP]). The cytoplasmic SMN complex plays a role in the biogenesis of snRNPs in the cytoplasm before their entry into the nucleus (Paushkin et al., 2002). Nuclear SMN-containing complexes are thought to recycle and/or resupply snRNPs to the early (H/E) complexes in the spliceosome assembly pathway and so to be involved in the processes that direct pre-mRNA splicing (Paushkin et al., 2002). Thus, the findings that nuclear Gal-1 interacts with Gemin4 and is co-immunoprecipitated with nuclear SMN complexes (Park et al., 2001) offer mechanistic insights into its potential role in the splicing pathway (Vyakarnam et al., 1997) by involving the H/E complex as the locus of action of Gal-1 in the spliceosome assembly (Figure 3).

Ras

Gal-1 interacts in a lactose-independent manner with H-Ras-guanosine triphosphate (H-Ras-GTP) through its farnesyl cystein carboxymethylester (Paz et al., 2001; Rotblat et al., 2004) and so strengthens its membrane association (Paz et al., 2001). The binding of Gal-1 to Ras is one of the most interesting and potentially significant functions of Gal-1. We will therefore detail these functions in the following section.

The effect of Gal-1 on cell signaling pathways

Regulation of cell growth

While extracellular Gal-1 has no effect on the growth rates of naïve T cells (Endharti et al., 2005) or of astrocytic (Camby et al., 2002) or colon (Hittelet et al., 2003) tumor cell lines, Gal-1 is mitogenic for various types of normal or pathological murine and human cells, that is, murine Thy-1-negative spleen or lymph node cells (Symons et al., 2000), mammalian vascular cells (Sanford and Harris-Hooker, 1990; Moiseeva et al., 2000), and hepatic stellate cells (Maeda et al., 2003). Gal-1 inhibits the growth of other cell types such as neuroblastoma (Kopitz et al., 2001) and stromal bone marrow cells (Andersen et al., 2003). Interestingly enough, it has been reported that depending on the dose involved, Gal-1 causes the biphasic modulation of cell growth. While high doses (∼1 µM) of recombinant Gal-1 inhibit cell proliferation independently of Gal-1 sugar-binding activity, low doses (∼1 nM) of Gal-1 are mitogenic and are susceptible to inhibition by lactose (Adams et al., 1996; Vas et al., 2005). While the knock-down of Gal-1 expression in murine melanomas (Rubinstein, Alvarez et al., 2004) and human glioma cells (our unpublished data) does not affect their growth rate in vitro, it does decrease it in 9L rat gliosarcomas (Yamaoka et al., 2000). Furthermore, Gal-1 can also regulate cell cycle progression in human mammary tumor cells (Wells et al., 1999). The seemingly paradoxical positive and negative effects of Gal-1 on cell growth are highly dependent on cell type and cell activation status, and might also be influenced by the relative distribution of monomeric versus dimeric, or intracellular versus extracellular, forms.

Regulation of Cell Migration Processes

While cell migration is the net result of adhesion, motility, and invasion (Lefranc et al., 2005; Decaestecker et al., forthcoming), Gal-1 modifies each of these three cell migration-related processes.

Adhesion

Gal-1 has been shown to increase the adhesion of various normal and cancer cells to the ECM via the cross-linking of glycoproteins (integrins) exposed on the cell surfaces with carbohydrate moieties of ECM components such as laminin and fibronectin (Ellerhorst, Nguyen, Cooper, Lotan et al., 1999; Moiseeva et al., 1999; van den Brule et al., 2003). In addition, Gal-1 can also mediate homotypical cell interaction, so favoring the aggregation of human melanoma cells (Tinari et al., 2001) and heterotypical cell interactions such as the interaction between cancer and endothelial cells, which, in its turn, favors the dispersion of tumor cells (Clausse et al., 1999; Glinsky et al., 2000).

Motility

Gal-1 causes the increased motility of glioma cells and the reorganization of the actin cytoskeleton associated with an increased expression of RhoA, a protein that modulates actin polymerization and depolymerization (Camby et al., 2002) (Figure 3). Conversely, the knock-down of Gal-1 expression in glioma cells reduces motility and adhesiveness (Camby et al., 2002, 2005). Oxidized Gal-1 (see Gal-1 in pathological nervous systems) stimulates the migration of Schwann cells from both the proximal and the distal stumps of transected nerves and promotes axonal regeneration after peripheral nerve injury (Fukaya et al., 2003). In colon carcinomas a Gal-1-enriched ECM decreases colon carcinoma cell motility (Hittelet et al., 2003).

Invasion

Using a proteomic approach based on the comparison of highly and poorly invasive mammary carcinoma cell lines, Harvey et al. (2001) identified the membrane expression of Gal-1 as a signature of cell invasiveness.

Interaction between Gal-1 and Ras

Ras genes which are frequently mutated in human tumors promote malignant transformation (Paz et al., 2001). Ras transformation requires membrane anchorage, which is promoted by Ras farnesylcysteine carboxymethylester and a second signal. The overexpression of Gal-1 increases membrane-associated Ras, Ras-GTP, and active ERK results in cell transformations, which are blocked by dominant negative Ras (Paz et al., 2001). Gal-1 antisense RNA inhibits transformations by H-Ras and abolishes the membrane anchorage of green fluorescent protein (GFP)-H-Ras, but not of GFP-H-Ras wild-type, GFP-K-Ras, and GFP-N-Ras (Paz et al., 2001). Thus, H-Ras–Gal-1 interactions establish an essential link between two proteins associated with cell transformation and human malignancies that can be exploited to selectively target oncogenic Ras proteins. In fact, H-Ras-GTP recruits Gal-1 from the cytosol to the cell membrane with the resulting stabilization of H-Ras-GTP, the clustering of H-Ras-GTP and Gal-1 in non-raft microdomains (Prior et al., 2003), the subsequent binding to Raf-1 (but not to PI3Kinase), the activation of the ERK signaling pathway and, finally, increased cell transformation (Elad-Sfadia et al., 2002) (Figure 3). So, in addition to increasing and prolonging H-Ras activation, the Gal-1-H-Ras complex renders the activated molecule selective toward Raf-1, but not toward PI3K (Ashery et al., forthcoming) (Figure 3). Fischer et al. (2005) have observed that the antiproliferative potential of Gal-1 in a number of carcinoma cell lines requires functional interaction with the α5β1 integrin. Antiproliferative effects result from the inhibition of the Ras-MEK-ERK pathway and the consecutive transcriptional induction of p27, whose promoter contains two Sp1-binding sites crucial for Gal-1 responsiveness (Fischer et al., 2005). The inhibition of the Ras-MEK-ERK cascade by Gal-1 increases Sp1 transactivation and DNA binding due to the reduced threonine phosphorylation of Sp1. In addition, Gal-1 induces p21 transcription and selectively increases p27 protein stability, while the Gal-1-mediated accumulation of p27 and p21 inhibits cyclin-dependent kinase 2 activity, a process which ultimately results in G1 cell cycle arrest and growth inhibition (Fischer et al., 2005).

Rotblat et al. (2004) have identified a hydrophobic pocket in Gal-1 analogous to the Cdc42 geranylgeranyl-binding cavity in RhoGDI. This pocket possesses homologous isoprenoid-binding residues including the critical L11, whose RhoGDI L77 homologue changes dramatically on Cdc42 binding. By substituting L11A, Rotblat et al. (2004) obtained a dominant interfering Gal-1 that possesses a normal carbohydrate-binding ability but inhibits H-Ras GTP-loading and extracellular signal-regulated kinase activation, dislodges H-Ras from the cell membrane and attenuates H‐Ras fibroblast transformation and PC12-cell neurite outgrowth. Thus, whereas Gal-1 cooperates with Ras independently of carbohydrate binding, Gal-1 (L11A) inhibits it.

Gal-1 in embryonic and adult tissue development and differentiation

Menstrual cycle, early gestation and embryogenesis

Gal-1 expression has been reported in male and female gonads (Wollina et al., 1999; Timmons et al., 2002; Dettin et al., 2003), and exogenously added Gal-1 has an inhibitory effect both on the steroidogenic activity of Leydig cells in the testicles (Martinez et al., 2004) and on the granulosa cells in the ovary (Jeschke et al., 2004; Walzel et al., 2004).

In the uterus Gal-1 expression is restricted to the endometrium (Maquoi et al., 1997) and varies during the menstrual cycle and the early phases of gestation (von Wolff et al., 2005). The expression of Gal-1 increases significantly in the late secretory phase endometrium and in decidual tissue (Maquoi et al., 1997; von Wolff et al., 2005), and shows a specific pattern of expression in trophoblastic tissue (Maquoi et al., 1997; Vicovac et al., 1998).

During the first trimester of human embryogenesis Gal-1 is expressed in connective tissue, in smooth and striated muscles, and in some epithelia such as the skin, the gonads, the thyroid gland, and the kidneys (van den Brule et al., 1997; Savin et al., 2003; Hughes, 2004; von Wolff et al., 2005).

Differentiation of the myogenic lineage

During the course of myoblast differentiation intracellular Gal-1 is externalized as myoblasts fused into myotubes (Cooper and Barondes, 1990). The role of Gal-1 in the case of myoblast fusion may be explained by the fact that the adherence of the myoblast to the extracellular component laminin is disrupted in the presence of Gal-1 (Cooper et al., 1991) via the selective modulation by Gal-1 of the interaction between the α7β1 integrin and fibronectin and laminin (Gu et al., 1994). Although the exact role of Gal-1 in myogenesis remains to be seen, this galectin has been shown to induce non-committed myogenic cells in the dermis to express myogenic markers. It increases the terminal differentiation of committed myogenic cells and has a role to play in the development and regenerative ability of muscles (Cooper and Barondes, 1990; Cooper et al., 1991; Harrison and Wilson, 1992; Goldring et al., 2002). Gal-1 may thus be regarded as a potentially important tool in the treatment of cases of human muscular dystrophy (Goldring et al., 2002).

Differentiation of the hematopoietic lineage

Mesenchymal cells give rise to the stromal marrow environment that supports hematopoiesis. These cells constitute a wide range of differentiation potentials (e.g., adipocytes, osteoblasts, chondrocytes, lymphocytes, erythrocytes, macrophages) as well as a complex relationship with hematopoietic and endothelial cells. Numerous studies have demonstrated that Gal-1 may be a key element in the course of hematopoietic cell differentiation (Lutomski, Fouillit et al., 1997; Andersen et al., 2003; Silva et al., 2003; Wang et al., 2004; Vas et al., 2005). The K562 human leukemia cell line expresses Gal-1 in the cytosol, but upon treatment with erythropoietin these cells develop an erythroid phenotype that leads to the externalization of cytosolic Gal-1 (Lutomski, Fouillit et al., 1997). Similarly, Gal-1 is externalized during adipocyte differentiation (Wang et al., 2004) and is able to modulate osteoblastic differentiation (Andersen et al., 2003) as well as the proliferation and death of hematopoietic stem and progenitor cells (Vas et al., 2005).

Nerve structure development

Gal-1 is widely distributed in the central and peripheral nervous systems of rodents during their development. Although it has been shown that Gal-1 plays a number of important roles in the formation of the neural network of the olfactory bulb of mice (Puche et al., 1996), there are no reports on its role in other regions. Gal-1 homozygous null mutant (Gal-1–/–) mice are viable and can grow into adults without any obvious phenotypical abnormalities except for a deficiency in the olfactory network (Poirier and Robertson, 1993; Tenne-Brown et al., 1998) and a reduced thermal sensitivity (McGraw, Gaudet, Oschipok, Steeves et al., 2005). In these mice the neuronal subpopulation in the olfactory bulb, which normally expresses Gal-1, does not reach the appropriate targets in the olfactory glomeruli (Puche et al., 1996). During its development into adulthood, a rat’s sensory neurons from the dorsal root ganglion express Gal-1, as do some spinal motor neurons (Regan et al., 1986). The initial expression in the sensory neurons begins as they finish their final mitotic division and begin their growth toward their targets in the dorsal horn of the spinal cord. When Gal-1-expressing neurons reach their targets Gal-1 expression remains high, albeit at lower levels (Regan et al., 1986; Hynes et al., 1990; Sango et al., 2004). In addition to neurons, Gal-1 mRNAs are also detected in the non-neuronal cells such as the pia mater, the choroid plexus, and the pineal gland as well as in reactive astrocytic and Schwann cells (Akazawa et al., 2004; Sango et al., 2004; Egnaczyk et al., 2003).

Gal-1 and the immune system

Galectins in general, and Gal-1 in particular, are known to be deeply involved in the initiation, amplification, and resolution of inflammatory responses (Figure 4) (Almkvist and Karlsson, 2004).

T-cell homeostasis and survival

A growing body of evidence indicates that Gal-1 functions as a homeostatic agent by modulating innate and adaptative immune responses. Gal-1 induces the inhibition of cell growth and cell-cycle arrest (Blaser et al., 1998; Rabinovich, Ramhorst et al., 2002) and promotes the apoptosis of activated, but not resting, immune cells (Perillo et al., 1995; Rabinovich et al., 1998; Chung et al., 2000; He and Baum, 2004). This said, resting T cells are sensitized to CD95/Fas-mediated cell death by Gal-1 (Matarrese et al., 2005). Furthermore, it has been shown that the Gal-1 expressed by thymic epithelial cells promotes the apoptosis of immature cortical thymocytes in vitro (Perillo et al., 1997), so suggesting a potential role for this protein in the processes of positive and/or negative selection within the thymic microenvironment. Gal-1 also suppresses the secretion of the pro-inflammatory cytokine interleukin-2 (IL-2) (Rabinovich, Ariel et al., 1999) and favors the secretion of the anti-inflammatory cytokine IL-10 (van der Leij et al., 2004) (Figure 4). All of these activities have been demonstrated by adding a relatively high concentration (µM range) of exogenous Gal-1 to T cells in vitro. In this context Bättig et al. (2004) have shown that the irreversibly dimeric form of Gal-1 is a dramatically more potent inducer of apoptosis in T cells than wild-type Gal-1. One concern regarding the proapoptotic activity of Gal-1 is whether high levels of soluble protein can be achieved in vivo. Recent evidence indicates that the amount of Gal-1 secreted by different cell types in the ECM is sufficient to kill T cells (Perillo et al., 1995; Chung et al., 2000; He and Baum, 2004). The effects of Gal-1 on immune and inflammatory cells are likely to be due to the binding and cross-linking of cell-surface glycoproteins on these cells (Galvan et al., 2000) (Figure 4). As a bivalent dimer, Gal-1 binds to the glycoproteins (including CD2, CD3, CD7, CD43, and CD45) on the cell surface of T cells in a carbohydrate-dependent manner (Pace et al., 1999). The regulated expression of glycosyltransferases—leading to the creation of N-acetyllactosamine ligands—during development and activation may determine T-cell susceptibility to Gal-1-induced cell death (Galvan et al., 2000; Amano et al., 2003; Carlow et al., 2003). Sezary cells—the malignant T cells in cutaneous T-cell lymphomas (the Sezary syndrome or mycosis fungoides)—resist a variety of apoptosis-inducing agents including Gal-1, because of the loss of CD7 expression and altered cellular glycosylations (Rappl et al., 2002; Roberts et al., 2003).

Thus, a number of T-cell glycoproteins from human MOLT-4 and Jurkat T cells have been shown to be specific receptors for mammalian Gal-1 binding and to be involved in Gal-1-mediated T-cell death: CD45, CD43, CD7 (Pace et al., 1999; Walzel et al., 1999; Symons et al., 2000; Fajka-Boja et al., 2002). However, although the deletion mutants of the glycoproteins confirm their importance in the apoptotic response to Gal-1 (Pace et al., 1999, 2000; Perillo et al., 1995), the role of CD45 in T-cell apoptosis mediated by Gal-1 remains controversial since Gal-1 induces apoptosis in CD45-deficient T cells (Walzel et al., 1999; Fajka-Boja et al., 2002). As for CD7, it seems that only specific spliced isoforms or glycoforms of CD45 may be important in signaling Gal-1-induced cell death (Nguyen et al., 2001; Xu and Weiss, 2002; Amano et al., 2003; Lanteri et al., 2003).

The signal transduction events that lead to galectin-induced cell death in activated T cells involve several intracellular mediators including the induction of specific transcription factors (i.e., NFAT, AP-1), the activation of the Lck/ZAP-70/MAPK signaling pathway, the modulation of Bcl-2 protein production, the depolarization of the mitochondrial membrane potential and cytochrome c release, the activation of caspases and the participation of the ceramide pathway (Rabinovich, Alonso et al., 2000; Walzel et al., 2000; Hahn et al., 2004; Ion et al., 2005; Matarrese et al., 2005) (Figure 4). However, a recent study has shown that Gal-1-induced apoptosis in human T leukemia MOLT-4 cells deficient in Fas-induced cell death is not dependent on the activation of caspase-3 or on cytochrome c release—two hallmarks of apoptosis—but involves the rapid nuclear translocation of EndoG from mitochondria (Hahn et al., 2004), so implying that Gal-1-induced cell death might also relate to one of the other types of cell death (Broker et al., 2005). Furthermore, recent evidence also indicates that whereas dGal-1 can induce the exposure of phosphatidylserine (an early apoptotic marker involved in the phagocytosis of apoptotic cells) on the plasma membrane of human T leukemia MOLT-4 cells, this does not result in cell death on activated neutrophils and on the promyelocytic cell line, but prepares the cells for phagocytic removal (Dias-Baruffi et al., 2003). At low concentrations (the nM range) Gal-1 has been shown in vitro to inhibit T‐cell adhesion to ECM and to modulate the tumor necrosis factor alpha (TNFα) as well as the interferon gamma (IFNγ) secretion from activated T cells (Allione et al., 1998) (Figure 4). In addition, in vivo studies on experimental autoimmunity models have revealed the ability of Gal-1 to skew the balance toward a T2-type cytokine response by reducing the levels of IFNγ, TNFα, IL-2, and IL-12 and increasing the level of IL-5 secretion (Santucci et al., 2000, 2003; Baum et al., 2003) (Figure 4).

T-cell immune disorders and chronic inflammation

In vivo, Gal-1 has powerful immunoregulatory effects through its ability to inhibit T-cell effector functions (van der Leij et al., 2004; Figure 4). Gal-1 treatment has resulted in improvements and even in cases of prevention in a number of experimental models of autoimmune diseases (Table III). The in vivo administration of Gal-1 prevents the development of chronic inflammation and impairs the ongoing disease in experimental models of autoimmune encephalomyelitis (EAE) (Offner et al., 1990), arthritis (Rabinovich, Daly et al., 1999), colitis (Santucci et al., 2003), hepatitis (Santucci et al., 2000), and chronic pancreatitis (Wang et al., 2000). The ability of Gal-1 to suppress the allogenic T-cell response through apoptotic and non-apoptotic mechanisms suggests its potential use for immunosuppression in organ transplantation and graft versus host disease (GVHD) (Baum et al., 2003).

Acute inflammation and allergy

In addition to its role in adaptative immune responses and chronic inflammation, Gal-1 also participates in innate immunity and acute and allergic inflammation (Liu, 2000). In phospholipase A2 (PLA2)-induced hind paw edema in rats the transmigration of both neutrophils and mast cells into the tissue is reduced in the presence of Gal-1 (Rubinstein, Ilarregui et al., 2004), which is also responsible for the inhibition of the release of arachidonic acid from lipopolysaccharide (LPS)-stimulated macrophages, mast-cell degranulation, and eosinophil migration (Rabinovich, Sotomayor et al., 2000; Delbrouck et al., 2002; La et al., 2003). This suggests that signals generated by Gal-1 binding inhibit rather than promote the migration of inflammatory cells.

Gal-1 is also able to induce an oxidative burst in neutrophils that have extravasated into tissue, but not in the case of peripheral blood neutrophils (Almkvist et al., 2002). This enhancement of cellular activity in exudated neutrophils is referred to as the neutrophil priming that might occur through the interaction of Gal-1 with α_Mβ₂ integrins expressed at the neutrophil cell surface (Almkvist et al., 2002), as has been described with respect to macrophages (Avni et al., 1998).

Host–pathogen (bacteria–virus–parasites) interactions

Taking recombinant Gal-1 as a basis, it has been shown that this galectin influences the ability of macrophages to control intracellular infections by inhibiting microbicidal activity, by promoting parasite replication, or by inducing host-cell apoptosis (Zuniga et al., 2001). While a biphasic modulation has been reported in Trypanosoma cruzi replication and cell viability, low concentrations (3 nM) of Gal‐1 increase parasite replication and do not affect macrophage survival, higher concentrations (300 nM) are able to condemn cells to apoptosis and to inhibit parasite replication. The expression of Gal-1 is markedly upregulated after parasite or virus infection (Giordanengo et al., 2001; Zuniga et al., 2001; Lim et al., 2003). It has been shown on the basis of experiments using exogenously added recombinant protein that Gal-1 acts as a soluble host factor that promotes HIV-1 infectivity though the stabilization of virus attachment to host cells (Ouellet et al., 2005), and that the altered T-cell surface glycosylations in HIV-1 infection results in an increased susceptibility to Gal-1-induced cell death (Lanteri et al., 2003). In contrast, in the case of the Nipah virus infection (responsible for severe, often fatal, febrile encephalitis) of endothelial cells, dGal-1 inhibits the cell fusion of the envelope glycoproteins of the Nipah virus with the host cells and favors the secretion of proinflammatory cytokines by dendritic cells (Levroney et al., 2005).

Gal-1 involvement in tumor progression and tumor immune-escape

From all the studies reported in the literature and summarized in Table IV it is reasonable to assume that Gal-1 expression or overexpression in a tumor or the tissue surrounding a tumor (stroma) must be considered as a sign of the tumor’s malignant progression and, consequently, of a poor prognosis for patients. This prognosis is often related to tumor immune-escape, to the long-range dissemination of tumoral cells (metastasis), or to their presence in the surrounding normal tissue, as is discussed below and illustrated in Figure 4.

Gal-1 could be involved in tumor angiogenesis because both vascular smooth muscle and endothelial cells express it (Moiseeva et al., 2000; Moiseeva, Williams, and Samani, 2003). Although the vessel walls of normal lymphoid tissues do not express Gal-1, the blood vessel walls of lymphomas do so in relation to their vascular density (D’Haene et al., 2005).

A number of mechanisms have been described that potentially contribute to tumor cell evasion from an anti-tumoral immune response (Zou, 2005). Together with the abundance of pro-apoptotic Gal-1 in privileged immune sites such as the placenta (Hirabayashi et al., 1989), the brain (Joubert et al., 1989), and the reproductive organs (Wollina et al., 1999) the fact that a number of studies have highlighted the expression of Gal-1 in the stromal tissue around tumors (Gillenwater et al., 1996; Sanjuan et al., 1997; Berberat et al., 2001; Shimonishi et al., 2001; van den Brule et al., 2001, 2003) or in the endothelial cells from capillaries infiltrating them rather than in those in the adjacent non-tumoral stroma (Clausse et al., 1999) suggest that Gal-1 might trigger the death of infiltrating T cells and protect these sites from the tissue damage induced by T-cell-derived proinflammatory cytokines (Figure 4). Le et al. (2005) have recently demonstrated a significant relation in head and neck squamous cell carcinomas (HNSCCs) between Gal-1 expression and the presence of both hypoxia markers and an inverse correlation with T-cell infiltration, a fact which suggests that hypoxia can affect malignant progression by regulating the secretion by tumor cells of proteins (like Gal-1) that modulate immune privilege. The immunomodulatory effects of Gal-1 and the correlation between Gal-1 expression in cancer cells and their aggressiveness (as described in the previous section) suggest the hypothesis that tumor cells may impair T-cell effector functions through the secretion of Gal-1, and that this mechanism may contribute toward tilting the balance in favor of an immunosuppressive environment at a tumor site (Figure 4). The blockade of the biological activity of Gal-1 in melanoma tissue results in a reduced tumor mass and stimulates the in vivo generation of a tumor-specific T-cell response (Rubinstein, Alvarez et al., 2004). These observations support the idea that Gal-1 may contribute to the immune privilege of tumors by modulating the survival or polarization of effector T cells (Figure 4).

Gal-1 in pathological nervous systems

Nerve regeneration

Gal-1 in its oxidized form—a form that lacks lectin activity—promotes neurite outgrowth (Outenreath and Jones, 1992) and enhances axonal regeneration in peripheral (Horie et al., 1999; Inagaki et al., 2000; Fukaya et al., 2003; Kadoya et al., 2005) and central (McGraw, McPhail et al., 2004; Rubinstein, Ilarregui et al., 2004; McGraw, Gaudet, Oschipok, Kadoya et al., 2005) nerves even at relatively low concentrations (picoM range) (Horie and Kadoya, 2000). The marked axonal regeneration-promoting activity of oxidized Gal-1 is likely to be paracrine (Figure 5). Indeed, Gal-1 is expressed in dorsal root ganglion neurons and motor neurons, with immunoreactivity restricted to the neuronal cell bodies, the axons, and the Schwann cells of adult rodents (Regan et al., 1986; Hynes et al., 1990; Horie et al., 1999; Fukaya et al., 2003) (Figure 5). After axonal injury, cytosolic reduced Gal-1 is likely to be externalized from growing axons and reactive Schwann cells to an extracellular space where some of the molecules may be converted into an oxidized form and may enhance axonal regeneration (Horie and Kadoya, 2000; McGraw, McPhail et al., 2004; Rubinstein, Ilarregui et al., 2004; Miura et al., 2004; Sango et al., 2004) (Figure 5). Miura et al. (2004) have recently identified a novel, naturally occurring, N-terminally processed form of Gal-1 that lacks the six amino-terminal residues of full length Gal-1. This isoform of Gal-1, which is monomeric under both reducing and oxidizing conditions, promotes axonal regeneration (Miura et al., 2004). Since oxidized Gal-1-induced neurite outgrowth is not observed on isolated neurons (Horie et al., 1999), the secreted Gal-1 probably influences the non-neuronal cells surrounding the axons, including the Schwann cells (Fukaya et al., 2003), and in so doing recruits macrophages, fibroblasts, and perineuronal cells (Horie et al., 2004) (Figure 5). In this respect, macrophages are potential candidates since they secrete an axonal regeneration-promoting factor when stimulated by oxidized Gal-1 (Horie et al., 2004). A preclinical study using rats with surgically transected sciatic nerves has recently shown that the administration by an osmotic pump of oxidized Gal-1 at the site of surgery restores nerve function (Kadoya et al., 2005).

Neural diseases

As mentioned above, Gal-1 is an endogenously expressed protein that is important in the embryonic development of primary sensory neurons and their synaptic connections in the spinal cord (Puche et al., 1996; Tenne-Brown et al., 1998). Various reports suggest a relation between Gal-1 expression (or altered expression) and neurological diseases. Gal-1 expression by neuronal and glial cells is closely correlated with regenerative success after injury (Wada et al., 2003; McGraw, Gaudet, Oschipok, Kadoya et al., 2005), and the level of autoantibodies to Gal-1 is significantly higher in patients with neurological disorders than in healthy controls (Lutomski, Joubert-Caron et al., 1997). In cases of neurodegenerative amyotrophic lateral sclerosis (ALS) diseases Gal-1 accumulates in the neurofilamentous lesions (Wada et al., 2003). In a recent report, Chang-Hong et al. (2005) show the neuroprotective effect of oxidized Gal-1 on a transgenic murine ALS model: the administration of oxidized Gal-1 to the mice delayed the onset of their disease, prolonged their life spans, and improved their motor functions.

In the rat brain hippocampus the expression of fosB (an early gene that directs the synthesis of the FosB and DeltaFosB proteins from the AP-1 complex of transcription factors) is induced immediately after ischemia and is accompanied by an increased expression of Gal-1, especially in neurons resistant to the injury (Kurushima et al., 2005). Gal-1 induces astrocyte differentiation, with the subsequently differentiated astrocytes greatly enhancing their production of the brain-derived neurotrophic factor (BDNF) that, in turn, plays an important role in the survival, differentiation, and synaptic plasticity of neurons (Sasaki et al., 2004) (Figure 5). In contrast to the effect of oxidized Gal-1 on axonal regeneration reported above, the effects of Gal-1 on astrocyte differentiation and BDNF production depend on carbohydrate-binding activity and are astrocyte-specific since no effects on neurons have been observed (Sasaki et al., 2004). The Gal-1-triggered astrocyte differentiation occurs predominantly via a tyrosine dephosphorylation pathway that remains to be elucidated (Sasaki et al., 2004). In this context Gal-1 may thus be considered as a means for the prevention of neuronal loss in cases of injury to the central nervous system (Egnaczyk et al., 2003).

Conclusions and prospects for therapeutical applications

Gal-1 plays a number of crucial roles in (1) neuronal cell differentiation and survival in both the central and the peripheral nervous systems, and (2) the establishment and maintenance of T-cell tolerance and homeostasis in vivo. Furthermore, it is reasonable to state that Gal-1 expression or overexpression in tumors or the tissue surrounding them must be considered as a sign of their malignant progression, which is often related to the long-range dissemination of tumoral cells (metastasis), to their dissemination into the surrounding normal tissue, and to tumor immune-escape. Its increased expression is therefore associated with poor prognoses for large numbers of cancer patients.

Gal-1 could constitute a target for the setting up of novel treatments for a number of different diseases. The targeted overexpression (or delivery) of Gal-1 should be considered as a novel approach for the treatment of inflammation-related diseases including GVHD (Baum et al., 2003), arthritis (Rabinovich, Daly et al., 1999), colitis (Santucci et al., 2003), and nephritis (Tsuchiyama et al., 2000), for example. It could be also viewed as a potential therapeutic target in some neurodegenerative pathologies (Chang-Hong et al., 2005; Kadoya and Horie, 2005) and muscular dystrophies (Goldring et al., 2002). In the fight against cancer progression what should be developed for therapeutical applications is the targeted inhibition of Gal-1 expression. Indeed, the knock-down of the expression of Gal-1 in migrating tumor cells, or at least in gliomas (Lefranc et al., 2005; Camby et al., 2006) and melanomas (our unpublished results), could impair malignancy development in different ways, including, for example, a delay in cancer cell migration within the host tissue (as for gliomas in the brain) or at a distance (melanoma metastases), and the sensitization of migrating cancer cells to apoptosis. Indeed, migrating cancer cells are protected against apoptosis (Lefranc et al., 2005; Decaestecker et al., forthcoming). Restricting the migration of cancer cells by down-expressing the Gal-1 in them restores a certain level of sensitivity to cell death, and so to cytotoxic drugs (Lefranc et al., 2005; Camby et al., 2006; Decaestecker et al., forthcoming). Anti-Gal-1 compounds are thus required to combat migrating cancer cells and several groups (Andre et al., 2001; Nangia-Makker et al., 2002; Sorme et al., 2003), including our own (Ingrassia et al., 2006), are engaged in this quest. The possibility exists that such anti-Gal-1 compounds could be assayed in clinical trials (in association with cytotoxic agents) in the near future.

Conflict of interest statement

None declared.

Acknowledgments

This work enjoys the support of the “Fonds de la Recherche Scientifique Médicale” (FRSM, Belgium) and the “Fonds Yvonne Boël.” Funding to pay the Open Access publication charges for this article was provided by the “Fonds Yvonne Boël.” ML is the holder of a “Grant Télévie” from the “Fonds National de la Recherche Scientifique” (FNRS, Belgium); FL is a Clinical Research Fellow with the FNRS and RK a Director of Research with the FNRS.

References

Author notes

2Laboratory of Toxicology, Institute of Pharmacy, Free University of Brussels (ULB), Brussels; 3XPeDoc sprl, rue Halvaux 37, 7090 Ronquières; and 4Department of Neurosurgery, Erasmus University Hospital, Free University of Brussels (ULB), Brussels, Belgium