Galectins act intra- and extracellularly. As known so far they are secreted via a non-classical mechanism which is not fully understood yet. They lack classical signalling sequences for specific localisation but can be found in the outer cellular space as well as inside the cells even located in the nucleus (Hughes, 1999). Although the complex regulation of secretion remains still elusive some explanations have been found: Galectin-1 secretion depends on the binding to a counter-receptor molecule and does not involve plasma membrane blebbing (Seelenmeyer et al., 2005; Seelenmeyer et al., 2008). Galectin-3 secretion seems also to be regulated by binding to other proteins such as chaperons and subsequent vesicular secretion (Hughes, 1999; Mehul & Hughes, 1997). The N-terminal-domain of galectin-3 is important for subcellular translocation and secretion of the protein (Gong et al., 1999).

#### **2.3.2 Galectin-1: Importance of reducing conditions**

The lectin activity of galectin-1 depends on reduced cysteine residues. Oxidised galectin-1 has no lectin activity but functions in the regeneration of nerve axons (Horie et al., 2004). Galectin-1 has six cysteine residues which are accessible to the solvent (see Fig. 2). The removal of the most accessible cysteine (Cys2) (Lopez-Lucendo et al., 2004) - or better all cysteine residues - enhances protein stability under both reducing and non-reducing conditions significantly (Cho & Cummings, 1995; Nishi et al., 2008), while none of them is necessary for lactose binding as shown by site directed mutagenesis and x-ray crystallography (Hirabayashi & Kasai, 1991; Lopez-Lucendo et al., 2004).

#### **2.3.3 Galectin-3: The only known chimera type galectin**

Galectin-3 has some specific properties due to its unique structure. Galectin-3 consists of three parts: 1) a N-terminal 12 amino acid leader sequence containing two phosphorylation sites, 2) a proline and glycine rich collagen like domain necessary for oligomerisation and 3) the carbohydrate recognition domain (Ahmad et al., 2004a; Dumic et al., 2006; Kubler et al., 2008; Mehul & Hughes, 1997; Nieminen et al., 2008). The first few amino acids forming the leader peptide are important for the subcellular localisation and secretion of the protein (Gong et al., 1999). Moreover phosphorylation of Ser6 seems to regulate affinity for different ligands and thereby cellular activity of galectin-3 (Dumic et al., 2006; Mazurek et al., 2000; Szabo et al., 2009; Yoshii et al., 2002). Galectin-3 can be cleaved by different proteases such as metalloproteinases-2 and –9 (gelatinases A and B respectively), metalloproteinase-13 (collagenase-3) and with low activity metalloproteinase-1 (collagenase-1) separating the fulllength CRD from the N-terminal extension (Guévremont et al., 2004; Ochieng et al., 1994). The main cleavage position is located between Ala62 and Tyr63 while other cleaving sites are only recognised by some specific proteases to lesser extend (Dumic et al., 2006;

Galectins: Structures, Binding Properties and Function in Cell Adhesion 9

assays and surface plasmon resonance (Appukuttan, 2002; Blixt et al., 2004; Bohorov et al., 2006; Ideo et al., 2003; Munoz et al., 2010; Song et al., 2009b; Stowell et al., 2008a). For glycan arrays it can be important which linker is used to bind the glycan epitopes to the surface (length, chemical structure). Moreover it is possible to use chemically and/or enzymatically produced ligands as well as glycans from natural compounds like glycopeptides and glycolipids (Blixt et al., 2004; Bohorov et al., 2006). The latter allows the analysis of complex and even unknown glycan structures of different cells (Song et al., 2009a; Song et al., 2009b; Song et al., 2010). Immobilised galectins are for example used in frontal affinity approaches

Variations of binding assays with immobilised partners are assays in which the surface binding is inhibited by a soluble ligand. Such inhibition studies of surface interactions allow a direct read-out of IC50 values and thereby the direct comparison of relative affinities (Sörme et al., 2002). For the calculation of affinity constants assumptions have to be made to simplify calculations which might not be correct for each single interaction measurement.

Most assays with one immobilised component as well as some direct interaction assays are based on the read-out of a fluorescence signal or other labels. Therefore either the galectin or the glycan structures have to be labelled. This leads to some additional problems: If the glycan is chemically labelled the linker or label itself can alter the binding affinity with specific effects for different galectins (Sörme et al., 2004). Therefore the affinity constants measured do not exactly fit to the unmodified glycan structures. Moreover the labelling of glycans is time-consuming. The labelling of galectins can also alter the binding specificities. It is in most cases done by random chemical modification of specific functional groups such as amino or thiol functionalities (Carlsson et al., 2007; Patnaik et al., 2006; Rapoport et al., 2008; Salomonsson et al., 2010; Song et al., 2009b; Stowell et al., 2008a; Stowell et al., 2008b). Although this labelling is assumed not to influence binding specificity or inactive galectins are removed after the labelling reaction, binding and oligomerisation still might be slightly affected. Moreover lot-specific aberrations between different labelling reactions occur. Labelled galectins are for example used for glycan arrays and ELISA-type set-ups (Blixt et al., 2004; Carlsson et al., 2007; Rapoport et al., 2008; Salomonsson et al., 2010; Song et al., 2009b; Stowell et al., 2008a) while fluorescence labelled glycans are used in frontal affinity chromatography or fluorescence polarisation (Carlsson et al., 2007; Hirabayashi et al., 2002;

Assays using both binding partners in its soluble form overcome most of the mentioned problems. But although those assays have different advantages the results cannot directly be compared with the natural set-up in which the glycans are immobilised on glycoproteins or glycolipids and thereby multivalently presented. Fluorescence polarisation is one of these methods measuring direct interactions of ligands in solution, but facing negative side effects of glycan labelling. Similarly, titration calorimetry also measures the interaction of two soluble binding partners. As for titration calorimetry no labelling reaction has to be performed this assay set-up might be considered as the one with fewest problems. But needed galectin concentrations for this test are usually (but not always) in high micromolar ranges and therefore above the physiological range. In this concentration range galectins tend to oligomerise, aggregate or precipitate (Ahmad et al., 2004b; Bachhawat-Sikder et al., 2001; Cho & Cummings, 1995; Dam et al., 2005; Sörme et al., 2004). Moreover titration

and ELISA assays (Hirabayashi et al., 2002; Sörme et al., 2002).

Salomonsson et al., 2010; Sörme et al., 2004).

Additional the problems mentioned before still persist (Sörme et al., 2004).

Guévremont et al., 2004; Ochieng et al., 1994). The single CRD is mainly described to have an increased affinity for different carbohydrates such as *N*-acetyllactosamine, the glycoprotein asialofetuin or glycans presented on endothelial cells but to have less biological activity as it looses the ability to form oligomers. This reveals the possible regulatory function of galectin-3 cleavage (Dam et al., 2005; Dumic et al., 2006; Ochieng et al., 1998a; Shekhar et al., 2004). In terms of this regulation it is suggested that the single galectin-3-CRD binds with high affinity to glycans on cell surfaces thereby blocking these interaction partners for full-length galectin-3 binding. After this blockage the full-length protein cannot perform its physiological functions anymore. In this way galectin-3 cleavage could act as down-regulation of galectin-3 function (John et al., 2003; Shekhar et al., 2004).
