Galectins at a glance

Galectins were discovered based on their galactoside binding activity, in a quest to find proteins that decode complex cell-surface glycans. They were defined as a protein family based on conserved β-galactoside-binding sites found within their characteristic ∼130 amino acid (aa) carbohydrate recognition domains (CRDs) (Barondes et al., 1994). However, it was soon realised that galectins are synthesised as cytosolic proteins, reside in the cytosol or nucleus for much of their lifetime and reach their galactoside ligands only after non-classical secretion that bypasses the Golgi complex (Cummings et al., 2015). Indeed, galectins also interact with various non-galactose-containing binding partners, and their CRDs (or other parts) also have non-carbohydrate binding sites (Kim et al., 2013; Li et al., 2013) (see poster). While all cells express galectins, sometimes at cytosolic concentrations as high as 5 µM (Lindstedt et al., 1993), the expression pattern varies between cell types and tissues (Table 1).

Galectins are small soluble proteins that contain one or two CRDs but no other types of folded protein domains (Hirabayashi and Kasai, 1993). A core set of mammalian galectins are conserved from earlier vertebrate species, these comprise the ‘prototype’ galectin-1, -2 and -7, which contain one CRD, the chimeric galectin-3, and galectin-4, -8, -9 and -12, which are tandem repeat galectins with two CRDs (Houzelstein et al., 2004). Other galectins and galectin-related proteins are listed in Table 1.

Here, and in the accompanying poster, we exemplify the binding activities of galectins and their cytosolic and nuclear functions, as well as their capacity to: (1) recognise disrupted endosomal membranes; (2) be secreted via the non-classical pathway; and (3) modulate the activity of glycoproteins at the cell surface and inside the lumen of intracellular compartments. Because of space limitations, the focus of this article is on their cellular activities with the goal of stimulating interest in the interplay between intra- and extracellular galectin activities, and on the best studied galectins, mainly in human and mouse. We highlight how the understanding of molecular mechanisms will be critical to obtain confirmed models of galectin function. The organismal functions of galectins are only briefly summarised, including their tissue distribution (Table 1) and small molecule inhibitors (Box 1). A more detailed description of the carbohydrate structures to which galectins bind (Boxes S1 and S2) and of the main defects that are observed in galectin knockout mice (Table S1) are presented in the supplementary material.

The galectin CRD folds as a slightly bent β-sandwich (Di Lella et al., 2011; Leffler et al., 2004) (see poster). The concave S side forms a shallow groove that can approximately hold a tetrasaccharide (labelled A-B-C-D on the poster and Fig. S1). The defining galactose-binding site (C) that is conferred by a conserved sequence motif of about seven amino acids is localised within this groove. Weaker sites on either side of the galactose-binding site (including site E shown on the poster) enhance or decrease binding, thereby giving each galectin CRD its unique specificity and higher affinity for larger glycoconjugates (Hirabayashi et al., 2002; Kamili et al., 2016; Leffler et al., 2004; Salomonsson et al., 2010a). Thus, while galectin CRDs typically bind galactose by themselves with affinities in the millimolar range, they bind the common disaccharide N-acetyl-lactosamine with a mid-micromolar affinity and may bind to their preferred glycoprotein with sub-micromolar affinity (see Boxes S1 and S2 for a more detailed discussion).

Of note, binding sites on the CRD for non-carbohydrate-containing ligands, such as those found in the cytosol and nucleus (described below), have also been identified. The best characterised example is the C-terminal CRD of galectin-8, where a binding site on the F-side (see poster) confers a specific affinity for the protein NDP52 (also known as CALCOCO2) (Kim et al., 2013; Li et al., 2013). A similar site in galectin-3 binds to its own N-terminal domain (Ippel et al., 2016). Sites at the edge of the CRDs confer the dimerisation of galectin-1, -2 and -7.

Even though galectins are largely monomeric in solution at physiological concentrations (Lepur et al., 2012; Salomonsson et al., 2010b), most are able to crosslink glycoconjugate ligands and cause their aggregation. For example, galectin-1 forms a weak dimer (K_d in low micromolar range) (Di Lella et al., 2011; Salomonsson et al., 2010b) and dimerisation may be enhanced upon ligand encounter. Galectin-3 is monomeric up to high concentrations (100 µM), but can aggregate with ligands and agglutinate cells at much lower concentrations, which is promoted by its N-terminal non-CRD region (Hsu et al., 1992; Massa et al., 1993) via an oligomerisation mechanism that is not yet completely understood (Ahmad et al., 2004; Barboni et al., 1999; Birdsall et al., 2001; Kuklinski and Probstmeier, 1998; Lepur et al., 2012; Nieminen et al., 2007; Yang et al., 1998) (see poster and Box S3). Galectins with two different CRDs may form non-covalent dimers to achieve an increase in valency (Kamili et al., 2016).

Multivalency is required for many of the biological effects of galectins, especially when they encounter the outside of cells, as it permits not only the crosslinking of ligands to induce signalling pathways and the formation of cell surface lattices, but also ligand endocytosis (see poster). Affinity enhancement by multivalency can be very strong for other lectin families and is termed the cluster glycoside effect (Lee and Lee, 2000; Lundquist and Toone, 2002). However, this effect appears to be modest or absent for galectins (Leppänen et al., 2005; Salomonsson et al., 2010b; Stowell et al., 2009).

It has been suggested that galectins may participate in liquid-liquid phase-separated structures (droplets) at the cell surface or in intracellular locations, based on weak interactions such as those found with terminal saccharide motifs (Banani et al., 2017; Dennis, 2015). Another interesting proposed role of weak binding is for rapid sampling of receptors (Kasai, 2006).

Galectins interact with a panoply of cytosolic and nuclear ligands that do not appear to be functionally linked (Haudek et al., 2010) (see poster). Galectin-1 and galectin-3 interact with the small GTPases HRAS (Paz et al., 2001) and KRAS (Shalom-Feuerstein et al., 2008), respectively. In one proposed mechanism, galectin-1 binds to and stabilises RAS dimers to positively regulate GTP-HRAS nanoscale signalling hubs (Blaževitš et al., 2016).

Galectin-3 associates with basal bodies and centrosomes, and its absence results in disorganisation of microtubules and perturbation of epithelial morphogenesis of primary cilia (Koch et al., 2010) or motile respiratory epithelial cilia (Clare et al., 2014). Moreover, galectin-3 associates with nuclear mitotic apparatus protein (NuMa), a key mitotic regulator required for the establishment and cohesion of the spindle pole (Magescas et al., 2017). Expression of mutant NuMA blocked in O-GlcNAc glycosylation results in aberrant spindle pole formation. This mutant does not co-immunoprecipitate galectin-3, which suggests that the galectin-3–NuMA interaction depends on O-GlcNAc glycosylation (Magescas et al., 2017). The exact mechanism remains unclear as O-GlcNAc itself does not bind galectin-3.

Galectin-3 also interacts with the endosomal sorting complex required for transport (ESCRT). Yeast two-hybrid analysis revealed interaction of the N-terminal domain of galectin-3 with the ESCRT component ALG-2-interacting protein X (Alix) (Chen et al., 2009), which promotes T-cell receptor downregulation, affects intracellular trafficking of the epidermal growth factor receptor (Liu et al., 2012) and release of human immunodeficiency virus (HIV)-1 (Wang et al., 2014). ESCRT components have fundamental roles in the biogenesis of multivesicular bodies (MVBs), which may provide opportunities to recruit galectin-3 into exosomes that are released from MVBs. Their presence in exosomes has been shown for galectin-3 (Fei et al., 2015; Thery et al., 2001; Welton et al., 2010) and galectin-9 (Keryer-Bibens et al., 2006).

Other examples of cytosolic functions for galectins include the association of galectin-3 with β-catenin and the Wnt signalling pathway (Shimura et al., 2005) and with Bcl2 for the regulation of apoptosis (Harazono et al., 2014b). Here, the NWGR motif in galectin-3 has been recognised to be anti-apoptotic and implicated in its association with both Bcl2 and Ras (Harazono et al., 2014a). However, it is important to note that this sequence is also part of the canonical galactose-binding site (Salomonsson et al., 2010a).

Galectin-3 shuttles between the cytoplasm and the nucleus. Nuclear localisation sequences were identified in murine galectin-3 [(253)ITLT(256); Davidson et al., 2006] and human galectin-3 [(223)HRVKKL(228); Nakahara et al., 2006], and the latter was shown to bind to importin-α for passage through nuclear pores. In addition, a leucine-rich nuclear export signal (NES) is found at positions 241-249 of murine galectin-3 (Tsay et al., 1999). All these sequences are within the CRD. Yet, the N-terminus of galectin-3 may also contribute to active nuclear transport. Indeed, phosphorylation of galectin-3 at Ser6 by casein kinase 1 (Huflejt et al., 1993) plays a role in its nuclear export and its anti-apoptotic activity (Haudek et al., 2010).

In the nucleus, galectin-1 and galectin-3 are involved in nuclear splicing of pre-mRNAs and in gene expression (Dagher et al., 1995; Vyakarnam et al., 1997). Both galectins have been shown to be functionally redundant splicing factors for mRNA processing and nuclear export (Wang et al., 2006). Galectin-3 assembles with small nuclear ribonucleoprotein (snRNP) particles in nuclear extracts. Here, it was identified in the earliest detectable mammalian prespliceosome complex (E-complex) (Haudek et al., 2016), and the heterogeneous nuclear ribonucleoprotein complex (H-complex) (Fritsch et al., 2016). These complexes initiate spliceosome assembly and the splicing cascade (Michaud and Reed, 1991). In the cytosol of pancreatic cancer cells, galectin-3 also stabilises MUC4 mRNA by interacting with the heterogeneous nuclear ribonucleoprotein L (Coppin et al., 2017). Galectin-3 has furthermore been implicated in the regulation of transcription through various mechanisms (see Kang et al., 2016 and references therein).

Galectins rapidly accumulate around disrupted endocytic vesicles or compartments. This appears to be due to the binding of cytosolic galectin to glycans, which are normally present in the lumen and, upon disruption, are exposed to the cytosol. This phenomenon was first discovered for galectin-3 on Shigella-containing vacuoles that undergo lysis (Paz et al., 2010) (see poster). Subsequently, it was also observed for galectin-8 on Salmonella-containing vacuoles (Thurston et al., 2012), galectin-3 on adenovirus-containing endosomes (Maier et al., 2012), and for galectin-1 and galectin-9 (Thurston et al., 2012). Galectins also accumulate around endosomes that have been disrupted in other ways, for instance by protein aggregates in neurodegenerative disease (Flavin et al., 2017; Jiang et al., 2017), by chemicals (Aits et al., 2015) or by nucleic acid transfection reagents (Wittrup et al., 2015). One function of this phenomenon is to link vesicle disruption to selective autophagy; here, the N-terminal CRD of galectin-8 interacts with glycoconjugates, whereas the C-terminal CRD interacts with the pathogen-specific autophagy receptor NDP52 (Boyle and Randow, 2013; Kwon and Song, 2018). Galectin-3 accumulation has been proposed to inhibit the galectin-8-mediated link to autophagy (Cheng et al., 2017). Galectins also interact with the tripartite motif-containing (TRIM) proteins, which are another class of receptors for selective autophagy (Chauhan et al., 2016).

The accumulation of galectins around disrupted vesicles is very rapid (within minutes) and easy to detect with immunohistochemistry or when using GFP-conjugated galectins (Paz et al., 2010). Hence, galectins have become a useful tool for detecting vesicle disruption in cell culture and in vivo (Aits et al., 2015), which can be exploited to assess the intracellular activity of galectin inhibitors (Stegmayr et al., 2016) (see also Box 1).

The non-classical secretion of galectins was shown to require their galactoside-binding activity (Seelenmeyer et al., 2005); this has led to the suggestion that accumulation around disrupted vesicles and binding to luminal glycans might provide a mechanistic link between autophagy and galectin secretion (Hughes, 1999). However, in another study, maturation of N-linked glycoprotein was not required for galectin-3 transport from the cytosol to the extracellular space, and secreted galectin-3 was predominantly free and not packaged into extracellular vesicles (Stewart et al., 2017), indicating that further studies are required to settle this matter. For galectin-4 secretion, a link to Src-mediated tyrosine-phosphorylation has been proposed (Ideo et al., 2013). Thus, the mechanism for the non-classical secretion of galectins remains unclear.

Upon their non-classical secretion, galectins enter the extracellular space, where they encounter ligands such as galactose-containing glycoproteins and glycolipids (see poster). Galectins can re-enter cells by endocytosis (Furtak et al., 2001) and populate endocytic and recycling compartments (Stechly et al., 2009; Straube et al., 2013).

According to the glycolipid-lectin (GL-Lect) hypothesis, binding of galectin to glycosylated cargo proteins, such as CD44 and α5β1 integrin, triggers galectin oligomerisation, which then enables their interaction with glycosphingolipids (Box S2) and the generation of membrane curvature, leading to the formation of tubular endocytic pits from which clathrin-independent endocytic carriers are formed (Johannes et al., 2016; Lakshminarayan et al., 2014) (see poster).

Galectins have also been described to counteract endocytic uptake by forming lattices at the plasma membrane (Brewer et al., 2002; Demetriou et al., 2001; Dennis, 2015; Dennis and Brewer, 2013; Dennis et al., 2009; Di Lella et al., 2011; Gupta et al., 1996; Kamili et al., 2016; Sacchettini et al., 2001; reviewed in Nabi et al., 2015). The balance between stimulation of endocytosis or the formation of an inhibitory lattice may depend on the specificity, structure and concentration of both the galectins themselves, and their glycoprotein and glycolipid ligands. A highly interesting link to the metabolic state of cells has been proposed, which is mediated by the intracellular levels of GlcNAc and UDP-GlcNAc that affect quantity and branching of complex N-glycans and so regulate galectin binding and the galectin lattice (Lau et al., 2007).

The T168N mutation in the interferon-γ receptor (IFN-γR), which causes Mendelian susceptibility to mycobacterial diseases syndrome (MSMD) (Kong et al., 2010), illustrates how abnormal interactions with galectins affect receptor function. This mutation creates a new N-glycosylation site, in addition to the six that are normally present; this draws the receptor from lipid nanodomains into actin nanodomains, thereby reducing its signalling function (Blouin et al., 2016). This also demonstrates that a dysregulation of the galectin system may predispose to disease. A similar interplay between actin and galectins has also been described for the T-cell receptor (Chen et al., 2007).

Galectins regulate the trafficking of intravesicular glycoprotein cargoes. Galectin-3 and galectin-4 accumulate in endosomal organelles following their internalisation from the cell surface (Stechly et al., 2009; Straube et al., 2013). Alternatively, a direct route from the cytosol into the lumen of endosomes could be possible, but this has not yet been shown. These endosomes serve as platforms for the sorting of cargo molecules into transport vesicles destined for the apical membrane of epithelial cells. Galectin-4 specifically interacts with lipid raft-associated glycoproteins, whereas galectin-3 binds to the non-raft fraction (Delacour et al., 2006, 2005). Depletion of the two galectins results in basolateral mis-sorting of the corresponding glycoproteins, suggesting an involvement of galectins in apical protein sorting (Delacour et al., 2006, 2005). Galectin-3-mediated apical sorting comprises pH-dependent cargo binding (von Mach et al., 2014) and cargo crosslinking into high-molecular weight clusters in a post-Golgi compartment (Delacour et al., 2007). Galectin-4-dependent sorting depends on the exposure of complex N-glycans on apical glycoproteins (Morelle et al., 2009). Both galectins escort their ligands to the apical cell surface, where they enter the apical milieu and are then reinternalised into endosomal organelles for a new round of apical protein sorting (Stechly et al., 2009; Straube et al., 2013) (see poster). Such a pathway is further supported by evidence for retrograde trafficking of galectin-9 from the plasma membrane to the Golgi (Mishra et al., 2010), which suggests participation of this galectin (and possibly others) in sorting events at the trans-Golgi network (TGN).

Extracellular galectins also induce signals from the cell surface and modify interactions with and within the extracellular matrix, as well as with microbes (Arthur et al., 2015; Di Lella et al., 2011; Gordon-Alonso et al., 2017; Thiemann and Baum, 2016). Most studies described here have considered only the common galactoside binding ability of galectins, but not the finer details in their specificity (described in Box S1). The role of N-glycan branching and extension with poly-N-acetyl lactosamines in lattice formation has been analysed and modelled in detail (Kamili et al., 2016; Nabi et al., 2015). The role of higher N-glycan branching for galectin-3 binding and intracellular routing of transferrin has also been demonstrated (Carlsson et al., 2013). Sialylation at the 3-position of galactose, which gives high affinity binding to galectin-8, directs the intracellular trafficking of this galectin (Carlsson et al., 2007). The inhibition of galectin-1 binding by sialylation at the 6-position of galactose has been suggested to make Th1 cells resistant to apoptosis (Di Lella et al., 2011).

The cytosolic and nuclear localisation and functions of galectins (Haudek et al., 2010) have been known almost as long as the extracellular ones (Di Lella et al., 2011), and this raises the question of whether the intra- and extracellular functions of galectins are connected by a regulatory system. Early evidence for this came from the observation that galectin-1 switches from a predominantly intracellular localisation in cultured myoblasts to a predominantly extracellular localisation upon their differentiation into myotubes (Cooper and Barondes, 1990). In developing mammary gland epithelial cells, galectin-1 was found mainly in the cytosol and nucleus of cells of the end bud, whereas it localised to the cell surface or extracellularly in duct cells (Bhat et al., 2016). Furthermore, galectin-1 binding to glycoconjugates was increased in the duct cells, with evidence that this interaction could ‘draw’ the galectin to an extracellular location (Bhat et al., 2016). Such a mechanism could provide a regulatory loop, whereby extracellular galectin-1-binding glycoconjugates reduce galectin concentration and associated functions in the cytosol and/or nucleus.

A highly interesting and plausible picture of the physiological and pathophysiological galectin functions in inflammation, immunity, cancer and many other diseases (Sciacchitano et al., 2018) has emerged from biochemical and cellular studies as described above, as well as from evidence from whole organisms (Table 1), in which galectins or their ligands were manipulated using genetic techniques, galectin inhibitors or galectins themselves (Table S1). As a detailed discussion is beyond the scope of this review, we pose here three fundamental questions.

The concentration of galectins in serum or plasma is typically in the high picomolar range, whereas most cellular studies have been performed at much higher concentrations. Thus, it is not likely that galectins act at a distance, similar to circulating hormones. Instead, their presence in serum may result from leakage from tissue, as has been found for many other proteins (Landegren et al., 2018) and serum levels of galectins could therefore be useful as diagnostic markers for disease. A correlation between galectin-3 levels and heart disease has indeed recently emerged (Gehlken et al., 2018). Furthermore, galectin-driven homotypic aggregation and/or adhesion of cancer cells to endothelial cells has been proposed to lead to increased metastasis (Glinsky et al., 2001; Sindrewicz et al., 2016). It has also been shown that at concentrations as low as 0.3 nM, galectin-3 stimulates sugar-dependent CD44 endocytosis (Lakshminarayan et al., 2014), whereas galectin-7 affects the endocytosis of E-cadherin in a sugar-independent manner (Advedissian et al., 2017). These findings suggest that some galectin-mediated cellular activities might indeed be sufficiently sensitive to respond to serum levels of galectins.

Adoptive transfer of bone marrow between galectin knockout and wild-type mice clearly showed that galectin-1 from cancer cells has angiogenic and immunosuppressive effects on the surrounding cells (Méndez-Huergo et al., 2017). Galectin-3 from myeloid cells also affects the growth of cancer cells and induces insulin resistance in fat cells (Li et al., 2016). Hence, galectins can have both autocrine and paracrine functions.

Most of the descriptions of galectin function in whole organisms consider the carbohydrate-binding activity of galectins, and have been suggested to take place either within the extracellular space, at the cell surface and/or in intracellular trafficking pathways (Blidner et al., 2015; Thiemann and Baum, 2016). However, as discussed above, galectins are rapidly endocytosed and directly traffic along both endocytic and exocytic pathways (Furtak et al., 2001; Lakshminarayan et al., 2014; Schneider et al., 2010; Straube et al., 2013). Thus, the cellular expression and release of galectins can modify these trafficking properties, possibly by setting an optimum level of galectin in the combined intra- and extracellular compartments of an organism. This may result in direct galectin-induced signalling (e.g. induction of apoptosis in T-cells) or galectin-regulated cell surface exposure and lateral organisation of glycosylated receptors, such as T-cell, VEGF, TGFβ and insulin receptors, along with their subsequent effects, including induction of angiogenesis, lymphangiogenesis, fibrosis in chronic inflammation and other mainly immune-regulatory functions (Blidner et al., 2015; Thiemann and Baum, 2016) (Table S1).

By contrast, the contribution of the cytosolic and nuclear activities of galectins to their organismal functions have been considered to a much lesser extent. The accumulation of galectins around damaged vesicles has been shown to occur also on cells within the organismal context (Aits et al., 2015), and thus may be a useful marker for vesicle-disrupting activities in vivo, as observed in neurodegenerative disease (Flavin et al., 2017; Jiang et al., 2017). Another elegant example is the role of galectin-3 in organising airway cilia and other cilia within the living organism (Clare et al., 2014), demonstrating an organismal role for the association of galectin-3 with centrosomes. These roles of galectins may also be connected to their extracellular activities, as described above, and have stimulated the development of galectin inhibitors as therapeutics (Box 1). Such drugs have been used to inhibit the immunosuppressive and angiogenesis-inducing effects of galectin-1, galectin-3 and galectin-9 in cancer (Blidner et al., 2015), to prevent unwanted angiogenesis in the eye (Chen et al., 2017) and to reduce fibrosis in chronic inflammation (Delaine et al., 2016). Anti-inflammatory galectin inhibitors have already reached the clinical development stage (https://galecto.com/).

Although the galectin field has come a long way in the past decade, a number of aspects remain incompletely understood, such as the molecular mechanisms of their actions in intra- and extracellular space and the relationship between these activities, as well as any associated regulatory loops. Furthermore, the structural aspects of galectin function, notably those linked to their oligomerisation, are also not fully known and require further work. Another exciting challenge in the field of cellular glycobiology is to elucidate the quantitative aspects of galectin distribution in cells and tissues, and the redundancies and specificities of galectins in health and disease. Addressing these issues will help to demystify the many cellular and organismal functions of galectins and pave the way to future therapeutic strategies for associated diseases and conditions.

H.L. is co-founder and shareholder in Galecto Biotech AB.