Galectin-3 is expressed and secreted by immune cells and has been implicated in multiple aspects of the inflammatory response. It is a glycan binding protein which can exert its functions within cells or exogenously by binding cell surface ligands, acting as a molecular bridge or activating signalling pathways. In addition, this lectin has been shown to bind to microorganisms. In this study we investigated the interaction between galectin-3 and Neisseria meningitidis, an important extracellular human pathogen, which is a leading cause of septicaemia and meningitis. Immunohistochemical analysis indicated that galectin-3 is expressed during meningococcal disease and colocalizes with bacterial colonies in infected tissues from patients. We show that galectin-3 binds to N. meningitidis and we demonstrate that this interaction requiresfull-length, intact lipopolysaccharide molecules. We found that neither exogenous nor endogenous galectin-3 contributes to phagocytosis of N. meningitidis; instead exogenous galectin-3 increases adhesion to monocytes and macrophages but not epithelial cells. Finally we used galectin-3 deficient (Gal-3−/−) mice to evaluate the contribution of galectin-3 to meningococcal bacteraemia. We found that Gal-3−/− mice had significantly lower levels of bacteraemia compared with wild-type mice after challenge with live bacteria, indicating that galectin-3 confers an advantage to N. meningitidis during systemic infection.
Galectins are a family of carbohydrate binding lectins defined by a β-sandwich fold, a signature binding site (Gabius et al., 2011) and affinity for β-galactose-containing molecules. They are key regulators of the immune system, expressed by almost all immune cells and have roles in fundamental biological processes including cell adhesion, chemotaxis, growth regulation and inflammation (Kaltner and Gabius, 2012; Liu et al., 2012). Galectin effector functions are exerted within cells and extracellularly, where they cross-link cell surface or matrix glycoconjugates, leading to adhesion and/or activation of signalling cascades to modulate cell activities and survival (Boscher et al., 2011). Each galectin has a characteristic profile of ligand specificity which can be influenced by clustering of glycans and their structural context (Kopitz et al., 2010; Krzeminski et al., 2011). In addition to cellular ligands, galectins also target sites on the surface of microorganisms (reviewed in Vasta, 2009) and have thus attracted attention for investigation into their roles as potential pathogen receptors and regulators of responses to infection.
Galectin-3 (also known as MAC-2 antigen) is widely expressed in lymphoid and non-lymphoid tissues (Flotte et al., 1983). It is present intracellularly (Wang et al., 2004) and can be secreted via a non-classical pathway (Hughes, 1999). Endogenous galectin-3 is required for efficient phagocytosis of opsonized erythrocytes and apoptotic thymocytes (Sano et al., 2003), although the precise mechanism by which it contributes to this process is not known. Galectin-3 is a monovalent glycan binding protein; however, upon binding to multivalent ligands, it oligomerizes and forms stable complexes. Galectin-3 can target glycans in microdomains with high affinity (Kopitz et al., 2010) and has been shown to have important roles in adhesion or biosignalling. For example, it can directly enhance adhesion of neutrophils to laminin (Kuwabara and Liu, 1996) and endothelial cells (Sato et al., 2002).
Galectin-3 production and secretion is highly upregulated in response to inflammatory stimuli (Henderson and Sethi, 2009), indicating that this protein is important in immune responses to infection. Consistent with this, galectin-3 is a key molecule in host defence against Streptococcus pneumoniae as it enhances recruitment, adhesion and function of neutrophils in the lungs of mice during pneumococcal infection to aid in phagocytosis and clearance of the pathogen (Sato et al., 2002; Farnworth et al., 2008; Nieminen et al., 2008). Interestingly, intracellular galectin-3 is recruited to phagosome membranes of cells infected with Shigella and Listeria (Paz et al., 2010) and accumulates around vacuoles containing Mycobacteria (Beatty et al., 2002); however, in each case, the precise function of galectin-3 recruitment and the effects on bacterial survival have not been established.
Several microorganisms display surface carbohydrates that contain epitopes which are known ligands for galectin-3. For example, through recognition of a unique pathogen-associated glycan, galectin-3 has direct fungicidal activity against Candida albicans by as yet undefined mechanisms (Kohatsu et al., 2006). Alternatively, galectin-3 can act as a cell surface receptor, as shown for Proteus mirabilis on MDCK cells (Altman et al., 2001), Helicobacter pylori on gastric epithelial cells (Fowler et al., 2006) and Pseudomonas aeruginosa on corneal epithelia (Gupta et al., 1997), while addition of exogenous galectin-3 to cultures of Trypanosoma cruzi increases its adhesion to smooth muscle cells (Kleshchenko et al., 2004). Interaction with galectin-3 may thus prove beneficial for pathogens, by acting as a cell surface docking site or as a cross-linking molecule which promotes adhesion. Several of the microbial ligands for galectin-3 have been identified. These include the fimbrial adhesin from P. mirabilis (Altman et al., 2001), glycosylphosphatidylinositols from Toxoplasma gondii (Debierre-Grockiego et al., 2010), phosphatidylinositol mannosides from Mycobacteria (Beatty et al., 2002) and lipopolysaccharide (LPS) from Klebsiella pneumoniae, Salmonella (Mey et al., 1996), H. pylori (Fowler et al., 2006), P. aeruginosa (Gupta et al., 1997), Escherichia coli (Stowell et al., 2010) and Neisseria gonorrhoeae (John et al., 2002).
Neisseria meningitidis colonizes the human respiratory tract and is an important Gram-negative human pathogen, able to cause septicaemia and meningitis (Lo et al., 2009). Along with type four pili, the polysaccharide capsule and LPS are important virulence factors expressed by this bacterium. These key surface structures can influence interaction with host cells and the differences in the composition of the capsule and the LPS can be used to classify strains into serogroups (Craven et al., 1978) and immunotypes (Scholten et al., 1994). Meningococcal invasion activates the complement and coagulation cascades and provokes an excessive inflammatory response, which is largely due to LPS activation via Toll like receptor 4 (TLR4) (Brandtzaeg et al., 1989). In light of the emerging roles of galectins as modulators of inflammatory responses and the evidence for galectin-3 interaction with pathogen surfaces, the goal of this work was to investigate whether galectin-3 could bind to N. meningitidis and to examine whether this could have consequences during meningococcal infection.
Galectin-3 is expressed during meningococcal infection
Meningococcal infection is characterized by a marked inflammatory response that contributes to the severity of the disease (Stephens et al., 2007). We therefore investigated the presence of different galectins in tissues of mice challenged with N. meningitidis. Based on the topology of lectin site presentation, galectins have been classified into three groups: proto-type, tandem-repeat and chimera-type galectins (Kasai and Hirabayashi, 1996). We chose a representative galectin from each of these three subgroups and, using specific antibodies (Kaltner et al., 2002; Lohr et al., 2007; 2008) and immunohistochemistry, we analysed the presence of the proto-type galectin-1, the tandem-repeat-type galectin-4 and chimera-type galectin-3 in spleen tissue of mice infected with serogroup B isolate MC58 for 24 h, compared to tissues from uninfected animals. Galectin-1 and galectin-4 were detected only at low levels in the spleen from both uninfected and infected animals (Fig. 1A). In contrast, while low levels of galectin-3 were present in tissue of control mice, intense staining for galectin-3 was detected in tissue from infected mice (Fig. 1A). Similar results were obtained in other organs such as the heart, lungs, kidneys and the meninges (data not shown).
To determine whether galectin-3 is also detected in human disease, we performed immunohistochemical analysis of spleen tissue from a patient with meningococcal disease (Fig. 1B). As a control, spleen tissue from an uninfected patient was examined. Galectin-3 could be observed within cells and extracellularly in splenic tissue from infected patients but not in control tissue. The distribution and shape of the galectin-3 positive cells indicates that they are most likely macrophages or monocytes.
Galectin-3 binds to N. meningitidis
Since galectin-3 can be secreted by phagocytes and is able to bind to galactoside-containing molecules on the surface of cells or of microorganisms, we sought to establish whether galectin-3 can bind to N. meningitidis. Closer inspection of galectin-3 staining in sections of spleen tissue from patients with meningococcal disease revealed colocalization of galectin-3 staining with bacterial colonies (Fig. 2A). Serial sections of tissue were examined for presence of galectin-3 or for meningococci which were detected using a polyclonal antibody against N. meningitidis. Accumulation of galectin-3 in foci corresponding to bacterial colonies was evident, suggesting that galectin-3 could be bound to N. meningitidis within the tissues.
We therefore analysed the binding of galectin-3 to serogroup B N. meningitidis strain MC58 using fixed, whole bacteria and recombinant human galectin-3. For comparison, we also tested human galectin-1 and galectin-4 that belong to the other two galectin subgroups, have overlapping but not identical specificities/affinities, and which were only detected at low level in our immunohistochemical analysis of N. meningitidis-infected murine tissue. Flow cytometry analysis revealed that galectin-3 binds to N. meningitidis (Fig. 2B). The lack of detectable binding of the other galectins tested indicates this is a particular feature of galectin-3 and prompted us to further characterize the interaction.
Full-length galectin-3 is required for interaction with N. meningitidis
The biological activities of galectin-3 are mainly dependent on its interaction with various ligands via its C-terminal carbohydrate recognition domain (CRD) (Ochieng et al., 2004). In addition, the N-terminal collagenase-sensitive region of galectin-3 can interact with non-carbohydrate ligands (Mey et al., 1996) and can mediate oligomerization and the formation of pentamers in presence of multivalent ligands (Hsu et al., 1992; Ahmad et al., 2004). To characterize the interaction of N. meningitidis and galectin-3, we first analysed the binding in presence of lactose, a pan-galectin ligand which acts as a competitive inhibitor of galectin-carbohydrate interactions via the CRD (Sparrow et al., 1987; Seetharaman et al., 1998). Consistent with a carbohydrate-dependent binding of galectin-3 to N. meningitidis we showed that the interaction is inhibited by lactose. Fixed bacteria were incubated with galectin-3 in absence or presence of 100 mM lactose. Addition of lactose partially inhibited lectin association (Figs 3A and S1), resulting in a reduction of up to 75% in galectin-3 binding (P = 0.0062), while addition of sucrose, a disaccharide that does not inhibit galectin-carbohydrate interactions, had no inhibitory effect (Fig. S1).
In addition, to identify which domains of galectin-3 are required for interaction, we examined the binding of truncated galectin-3 which lacks the N-terminal region following proteolytic treatment with collagenase D, as described previously (Kopitz et al., 2001). This leaves the CRD of the protein intact and still able to bind to endogenous ligands on cell surfaces (Kopitz et al., 2001). Flow cytometry was performed using the full-length protein (FL Gal-3) and the truncated protein (Tr Gal-3) and binding to N. meningitidis strain MC58 was analysed. As shown in Fig. 3B, the proteolytic removal of the N-terminal collagen-like repeats of galectin-3 almost completely abolished binding to MC58, indicating that the CRD is insufficient to support galectin-meningococcal interaction and that the N-terminal region of the protein is also required.
Full-length LPS is required for galectin-3 binding to N. meningitidis
In almost all Gram-negative bacteria which have been reported to bind galectin-3, the target ligand has been identified as LPS (reviewed in Vasta, 2009). We therefore examined whether meningococcal LPS is involved in the attachment of galectin-3 to the bacterial surface. Meningococcal LPS (Fig. 4A) consists of lipid A containing a disaccharide of glucosamine residues, which anchors the molecule in the outer membrane. The lipid A region of the LPS molecule is linked via two 2-keto-3-deoxy-octulosonic acid residues (Kdo) to a core oligosaccharide with an inner core di-heptose-N-acetylglucosamine backbone, comprising two heptose residues (HepI and HepII). This backbone provides a point of attachment for a variety of outer core oligosaccharides which leads to expression of different immunotypes of LPS (Jennings et al., 1999). Furthermore, N-acetylneuraminic acid (Neu5Ac, Fig. 4A) can be attached to LPS via the terminal galactose of specific LPS immunotypes. We investigated the contribution of N. meningitidis LPS to galectin-3 binding using meningococcal strains with defined truncations of LPS in order to assess the importance of the saccharide units to galectin recruitment. We examined binding to wild-type strain MC58 and isogenic strains carrying mutations in genes encoding glycosyltransferases involved in LPS biosynthesis (Fig. 4A). The lgtA and lgtB mutants lack enzymes responsible for the biosynthesis of the N-acetyl-lactosamine (LacNAc), while the galE mutant lacks the lacto-N-neo-tetraose (LNnT) structure (Jennings et al., 1995a); icsB lacks the enzyme responsible for the addition of glucose to HepI (van der Ley et al., 1997) and the lsi-1 mutant lacks the enzyme responsible for the addition of HepII to HepI but has also been shown to lack the oligosaccharide portion of the LPS molecule except the first heptose (Jennings et al., 1995b). Analysis of galectin-3 binding to N. meningitidis by flow cytometry demonstrated that all mutants expressing a truncated LPS molecule showed a significant reduction in the binding of galectin-3 compared to the wild-type strain (Fig. 4B) demonstrating that full-length LPS is necessary for galectin-3 attachment to the meningococcal surface. The fact that mutation of lgtB alone was sufficient to abolish galectin-3 binding indicates that the terminal galactose and an intact LacNAc moiety are required for interaction.
Meningococcal strain MC58 expresses an L3,7,9 LPS immunotype which can be modified by the addition of an α2,3-linked sialic acid to the terminal galactose of the alpha-chain through the activity of a specific sialyltransferase (Lst) (Mandrell et al., 1990; 1991; Gilbert et al., 1997). The presence of sialic acid on terminal residues of glycans can modulate binding of some galectins, although this was shown to have minor effect on galectin-3; in fact galectin-3 is known to bind to LacNAc repeats with terminal α2,3-sialylation and α2,6-sialylation (Ahmad et al., 2002; Stowell et al., 2008). To examine whether sialylation of LPS alters galectin-3 binding to the surface of N. meningitidis we compared the binding to wild-type strain MC58 and an isogenic mutant lacking lst. In addition, we examined the effect of the presence of the polysaccharide capsule which is expressed on the bacterial surface and protects the bacterium from complement mediated lysis and phagocytosis (Hill et al., 2010). The gene siaD encodes the α2,8-polysialyltransferase required for chain elongation during capsule synthesis and siaC is necessary for sialic acid biosynthesis (Edwards et al., 1994); therefore, MC58ΔsiaC is unable to produce a capsule or sialylate LPS without exogenous 5′cytidine monophospho-N-acetyl-neuraminic acid (CMP-NANA). We analysed galectin-3 binding to MC58 and the isogenic mutant strains MC58ΔsiaC, MC58ΔsiaD, and MC58Δlst (Exley et al., 2005). All three mutants exhibited elevated levels of galectin-3 binding, although these increases were not significantly different to the parental strain MC58 (Fig. 4C). These data indicate that the presence of surface sialic acids (capsule and terminal sialylation of LPS) does not affect galectin-3 association with the bacterial surface.
Finally, to ascertain whether galectin-3 binding was restricted to MC58 we analysed the binding to other strains of N. meningitidis. We chose a small representative panel of disease causing isolates that belong to different serogroups and have different LPS immunotypes (Table 1). Analysis by flow cytometry demonstrated that, while there was some variation in the relative level of binding, strains belonging to serogroups A, B and C were able to bind galectin-3; only one strain, F8238, had a significantly reduced level of galectin-3 binding compared to MC58 (Fig. 4D). Interestingly, all the strains which bind galectin-3 are able to express an LPS immunotype containing a terminal LacNAc; Z2491 is a serogroup A isolate which expresses an L9 immunotype (Choudhury et al., 2008) while serogroup B strains MC58 and H44/76 and serogroup C strains 8013 and FAM18 (Plested et al., 1999; Geoffroy et al., 2003; and this study, data not shown) all express L3,7,9 immunotype. F8238 has an LPS immunotype L10 (Maslanka et al., 1997) that does not include the LacNAc and LNnT moieties (Kim et al., 1994), consistent with these structures being required for galectin-3 binding.
Table 1. N. meningitidis strains used in this study
Taken together these data demonstrate that full-length galectin-3 is able to bind to N. meningitidis and that LPS molecules on the meningococcal surface are required for this interaction. We therefore chose to examine the impact of galectin-3 binding to N. meningitidis on bacterial interaction with cells.
Endogenous galectin-3 does not alter the interaction of N. meningitidis with phagocytes
Galectin-3 is highly expressed by immune cells such as activated macrophages (Liu et al., 1995) and endogenous galectin-3 has been shown to be important for phagocytosis of opsonized erythrocytes and apoptotic cells in vitro and in vivo (Sano et al., 2003). In the first instance we examined whether endogenous galectin-3 is important for binding and phagocytosis of N. meningitidis. As a relevant cell model, human THP-1 (Human acute monocytic leukaemia) cells were differentiated into macrophages using phorbol 12-myristate 13-acetate (PMA) and transfected with small interfering RNA (siRNA) designed to impair galectin-3 expression. Knock-down was confirmed by Western blot analysis (Fig. 5A). Cells were challenged with N. meningitidis strain MC58 at a multiplicity of infection (MOI) of 30 and bacteria were recovered 1 h later. Internalization of N. meningitidis was analysed after gentamicin treatment to kill extracellular bacteria. In parallel we analysed the phagocytosis of N. meningitidis by THP-1 macrophages subjected to transfection in the same conditions as the siRNA treated cells but in absence of siRNA oligonucleotides as controls. Neither the total number of cell-associated bacteria (Fig. 5B) nor the number of internalized bacteria (Fig. 5C) (each expressed as percentage of the inoculum) was affected by knock-down of galectin-3. These data suggest that endogenous galectin-3 is not required for the binding and uptake of N. meningitidis by macrophages.
Association of meningococci with phagocytic cells is increased in presence of exogenous galectin-3
Our initial analysis of the role of galectin-3 in interaction with phagocytes showed that depletion of endogenous galectin-3 from THP-1 macrophages did not affect the adhesion or uptake of meningococci; however, phagocytes also secrete galectin-3 upon activation, and inflammation is characterized by galectin-3 in the microenvironment. We therefore examined whether the binding of exogenous galectin-3 to N. meningitidis had an impact on the bacterial-phagocyte interaction. We analysed the effect of pre-incubating N. meningitidis with galectin-3 on association with and uptake into THP-1 macrophages and primary human monocytes obtained from whole blood of healthy donors. Wild-type THP-1 cells were differentiated into macrophages using PMA for 72 h and infected at an MOI of 30 for 1 h with wild-type N. meningitidis pre-incubated with 3.3 μM galectin-3 or PBS. Pre-incubation of meningococci with recombinant galectin-3 resulted in a significant increase (1.7-fold, P = 0.0201) in the total number of bacteria associated with THP-1 cells (expressed as a percentage of the inoculum, Fig. 6A). Similar results were obtained after infection for 6 h (Fig. S2) and with primary human monocytes (twofold increase, P = 0.0002) (Fig. 6B). On the other hand, we observed that the percentage of bacteria which are internalized by monocytes and THP-1 cells is comparable for bacteria pre-incubated with PBS or galectin-3 (Figs 6C and S3), suggesting that exogenously added galectin-3 increases the adhesion of bacteria to these cells, with no net effect on bacterial internalization.
Finally, we examined whether presence of LacNAc in the bacterial LPS was required for the galectin-3-mediated enhanced association to monocytes, using the lgtB mutant that fails to bind galectin-3 (Fig. 4B). Addition of exogenous galectin-3 had no effect on the adhesion of the lgtB mutant (Fig. 6D), indicating that a full-length LPS is required for the galectin-3 enhanced interaction of N. meningitidis with phagocytes.
Galectin-3 expression in epithelial cells is unaltered upon infection with N. meningitidis
Several pathogens can induce the upregulation and secretion of galectin-3 from epithelial surfaces. For example, this protein is released in soluble form into the alveolar fluid during pneumococcal infection in mice (Sato et al., 2002) and H. pylori induces galectin-3 upregulation in and release from human gastric epithelial cells (Fowler et al., 2006). The first cells that N. meningitidis encounters in the human host are nasopharyngeal epithelial cells (Carbonnelle et al., 2009). We therefore analysed galectin-3 mRNA level in the human pharyngeal cell line, Detroit 562, with or without infection with N. meningitidis (Fig. 7A). Levels of galectin-3 mRNA were not altered 4 h or 24 h after challenge and there was no obvious change in galectin-3 protein production as shown by Western blot (Fig. 7B). Using fluorescence microscopy we also assessed whether galectin-3 is expressed on the cell surface, where it could potentially serve as a receptor for N. meningitidis, as it is known that other bacteria bind to galectin-3 on epithelial cell surfaces (Gupta et al., 1997; Fowler et al., 2006). We were able to detect surface galectin-3 on uninfected cells or cells infected with MC58 (Fig. 7C). While we did not observe any striking change in the intensity of staining of surface galectin-3 or localization in infected cells, in some instances we could detect colocalization of galectin-3 with adherent meningococci (Fig. 7D). Finally, given that pre-incubation of N. meningitidis with galectin-3 led to a higher number of bacteria associated to the surface of phagocytes we examined whether this was also observed upon interaction with epithelial cells; however, the preincubation of bacteria with galectin-3 did not affect their association with Detroit 562 cells (Fig. 7E).
Meningococcal bacteraemia is reduced in galectin-3 deficient mice
To gain insight into the role of galectin-3 in meningococcal infection in vivo we compared the levels of bacteraemia using wild-type and galectin-3 deficient mice, both in a C57BL/6 genetic background (Fig. 8). Although N. meningitidis is a human-specific pathogen, murine galectin-3 binds the meningococcus (Fig. S5) and the murine model has been used in previous studies to examine virulence (Exley et al., 2005), assess efficacy of vaccine antigens (Li et al., 2009) and analyse the contribution of innate immune components to bacterial survival (Pluddemann et al., 2009a). Groups of 8-week-old mice were challenged by intraperitoneal injection of N. meningitidis and bacteraemia was measured at seven and 24 h post challenge by counting the number of bacteria in blood after serial dilution and plating. At 7 h, the average bacteraemia from wild-type and Gal3−/− mice was 1.2 × 107 cfu ml−1 of blood and 4.4 × 106 cfu ml−1 of blood respectively (Fig. 8; P < 0.0001). A significant difference in levels of bacteraemia between the wild-type and Gal3−/− mice was also evident at 24 h (Fig. 8; P = 0.087). This demonstrates that galectin-3 promotes bacteraemic meningococcal disease in this model.
Neisseria meningitidis is a frequent commensal of the upper respiratory tract, but can also cause serious blood and central nervous system infections manifested as septicaemia and meningitis. A characteristic of meningococcal disease is the profound inflammatory response that is largely responsible for the often devastating outcomes of this infection (Stephens et al., 2007). Galectin-3 production and secretion is highly upregulated in response to inflammatory stimuli (Liu et al., 1995), thus meningococci are likely to encounter this molecule in several sites where cells have been activated during invasive disease. Here we demonstrate for the first time that galectin-3 binds to N. meningitidis in vitro and show a striking colocalization of meningococci and galectin-3 in tissue from patients with meningococcal infection by immunohistochemistry. This interaction relies on the presence of full-length LPS molecules containing LacNAc and we show that galectin-3 binding enhances the association of meningococci to the surface of monocytes and macrophages.
Galectin-3 expression is associated with inflammatory responses to different parasites and bacterial pathogens. It is found in lung tissue from mice infected with S. pneumoniae (Farnworth et al., 2008) and in patients with disseminated C. albicans infection (Kohatsu et al., 2006). Consistent with this, by immunohistochemistry we found that galectin-3 is present in spleen tissue from mice and patients infected with N. meningitidis, and is evident within cells and in the extracellular compartment. Interestingly, we did not detect a similar presence of proto-type galectin-1, which has been shown to share some functions with galectin-3 in inflammatory responses (Almkvist and Karlsson, 2004; Liu et al., 2012), or tandem-repeat type galectin-4 which has been proposed to contribute to the innate immune response to bacterial infection (Stowell et al., 2010).
Similar to findings with the gonococcus (John et al., 2002), we showed that human galectin-3 binds to N. meningitidis and that this is dependent on the presence of LacNAc in the outer core of the alpha-chain of LPS. Although the lsi-1 mutant showed a significant reduction in galectin-3 binding, this could be due to the interdependence of synthesis of the alpha- and beta-chains; there is no discernible oligosaccharide extension in the lsi-1 mutant (Jennings et al., 1995b). It is also possible that the inner core is important for binding, by allowing appropriate positioning or presentation of the molecule and/or by supporting multiple interactions and permitting galectin-3 oligomerization, as was recently described for glycoprotein ligands of galectin-3 (Ahmad et al., 2004; Krzeminski et al., 2011). Importantly, since galectin-1 and galectin-4 did not react, the mere presence of the LacNAc residue which is recognized by all three galectins with similar affinity (Dam et al., 2005) is not sufficient for binding and highlights the specific nature of the interaction between galectin-3 and N. meningitidis.
The ability of lactose to inhibit binding of galectin-3 to N. meningitidis shows that the interaction relies largely on the CRD of galectin-3, although we also observed some residual binding to N. meningitidis in the presence of lactose. In addition, the CRD is not sufficient for interaction as we also found that there was almost complete abolition of binding between meningococci and the truncated form of galectin-3 devoid of the N-terminal section. This may reflect a requirement for both the N- and C-terminal regions of galectin-3 to bind different bacterial targets, for example, the N-terminal region of galectin-3 has been shown to bind to lipid A of Salmonella minnesota in a manner which is not inhibited by lactose (Mey et al., 1996). In addition, this could also reflect that oligomerization of galectin-3 is important for binding. It has been shown that the association via the non-CRD region of galectin-3 leads to concentration- and time-dependent formation of oligomers, increasing the affinity of this protein compared to the monomeric form and leading to positive cooperativity (Hsu et al., 1992; Massa et al., 1993; Ahmad et al., 2004). We propose that binding of galectin-3 to N. meningitidis is achieved through multivalent interactions with bacterial LPS molecules and with itself, a hypothesis which is supported by recent evidence that LPS from E. coli can induce galectin-3 oligomerization (Fermino et al., 2011).
The presence of terminal sialic acid on cell surface glycans can modulate and even block galectin interactions, although it is thought to have little effect on galectin-3 binding (Ahmad et al., 2002; Stowell et al., 2008). Consistent with this, we found that removal of the sialic acid capsule and/or the terminal sialylation of LPS did not significantly increase the binding of galectin-3 to N. meningitidis. We observed that the propensity of different strains of N. meningitidis to bind galectin-3 was associated with their LPS immunotype. Meningococcal LPS can be classified into different immunotypes based on variations in composition that alter its structure and antigenic properties (Tsai et al., 1983; Plested et al., 1999). All the strains which bind galectin-3 are able to express an LPS immunotype containing LNnT (and LacNAc); the only strain which did not show similar levels of binding was F8238, a serogroup A isolate which expresses LPS immunotype L10 (Maslanka et al., 1997) that is proposed to lack this structure (Kim et al., 1994). Interestingly, through phase variation, individual meningococcal strains have the potential to express alternative outer core structures (Jennings et al., 1999; Berrington et al., 2002). For example, phase variation leading to switching off expression of the lgtA gene results in loss of LacNAc and complete LnNT epitopes and a switch from immunotype L3 to L8 (Jennings et al., 1995a). Our results using a defined LPS mutant lacking lgtA suggest that this phase variation event would abolish galectin-3 binding. Of note, expression of an L3,7,9 immunotype is associated with invasive disease isolates (Mackinnon et al., 1993). To date this has been largely associated with the ability of this structure to undergo sialylation which enhances serum resistance (Moran et al., 1994) but further elucidation of the consequences of galectin-3 binding may reveal that the capacity to bind this lectin also confers a survival advantage to the meningococcus during infection.
Galectin-carbohydrate interactions have been proposed as a mechanism of pathogen recognition and attachment (Sato et al., 2009; Vasta, 2009). Binding of galectin-3 to several microorganisms has been described (Altman et al., 2001; John et al., 2002; Fowler et al., 2006; Stowell et al., 2010) and cell surface or extracellular galectin-3 can be involved in pathogen binding. N. meningitidis is largely an extracellular pathogen but a key event in meningococcal infection is the interaction with host cells (Carbonnelle et al., 2009). N. meningitidis encounters different cell types from the state of colonization to invasion and the onset of systemic disease. These include epithelial cells in the nasopharynx to which the bacteria must bind in order to establish carriage, resident macrophages in the submucosa during the early invasive process, endothelial cells during traversal and infiltration of the systemic circulation and monocytes in the bloodstream. To investigate the possible effects of galectin-3 binding to N. meningitidis, we used siRNA knock down and relevant human cell infection models in vitro to investigate the consequences of the interaction on the association of bacteria with different cell types.
Pathogens such as H. pylori and P. aeruginosa (Gupta et al., 1997; Fowler et al., 2006) bind galectin-3 on the surface of epithelial cells and it has been proposed as a receptor for N. gonorrhoeae as it is constitutively expressed at the apical side of the non-ciliated fallopian tube epithelial cells which selectively bind gonococcus (John et al., 2002). Galectin-3 expression has also been demonstrated in human nasopharyngeal epithelial tissues (Saussez et al., 2008). We found that Detroit 562 human pharyngeal epithelial cells express galectin-3 and expression levels are unchanged upon infection with N. meningitidis. We also found that the protein is detectable on the surface of Detroit 562 cells and confocal microscopy indicated that in some instances galectin-3 colocalizes with meningococci. It is therefore possible that galectin-3 on the surface of these cells could contribute to meningococcal attachment, similar to other pathogens. Although our experiments did not address this directly, we found that pre-incubation of bacteria with galectin-3 did not result in a reduction in the association of the bacteria, which might be expected if surface galectin was a direct target for meningococcal LPS binding. To conclusively address the role of epithelial cell surface galectin-3 as a receptor for N. meningitidis, more detailed studies are required.
In order to investigate the role of galectin-3 in meningococcal interactions with phagocytes, we used differentiated THP-1 cells as a source of macrophage-like cells and employed siRNA technology to gain insights into galectin-3 function in adhesion/uptake. Galectin-3 appears to be important for phagocytosis of cells (Sano et al., 2003); however, we found that siRNA knock-down of galectin-3 in THP-1 macrophages did not alter the capacity of these cells to bind or phagocytose N. meningitidis. This indicates that neither surface expressed nor intracellular galectin-3 is necessary for meningococcal recognition and uptake by macrophages in vitro. Galectin-3 accumulations have been observed in macrophages infected with Shigella, Listeria and Mycobacteria (Beatty et al., 2002; Paz et al., 2010), but using macrophages from galectin-3 null mice, no role for intracellular galectin-3 in mycobacterial or Salmonella uptake and survival was identified (Beatty et al., 2002; Li et al., 2008). In fact, we are not aware of evidence for endogenous galectin-3 having a role in phagocytosis of any bacterium by macrophages.
Interestingly, pre-incubation of N. meningitidis with exogenous galectin-3 led to a significant increase in bacterial adhesion to monocytes and macrophages but did not alter internalization. This increased adhesion was dependent upon expression of full-length meningococcal LPS molecules, which permit galectin-3 binding. Recent data has shown that pre-incubation of exogenous galectin-3 with purified LPS dramatically increases LPS binding to neutrophil surfaces (Fermino et al., 2011) and our observations with whole bacteria and macrophages or monocytes are consistent with this. The increased attachment of the meningococcus to the surface of macrophages and monocytes but not to epithelial cells suggests this role is restricted to phagocyte interactions and may be related to different repertoires of receptors or the different outcomes of interaction with these cell types. Previous work has shown that human macrophages can bind, phagocytose and at least partially kill N. meningitidis (Read et al., 1996) and receptors that have been implicated in meningococcal interaction with phagocytes include TLRs, macrophage receptor with collagenous domain (MARCO), scavenger receptors (Pluddemann et al., 2009b) and siglecs (Jones et al., 2003). How pre-incubation of N. meningitidis with galectin-3 leads to increased adhesion without affecting uptake remains to be elucidated. Taking into account the known properties of this protein, we speculate that decoration of the bacterial surface with galectin-3 could enhance adhesion through cross-linking of surface glycans, receptor rearrangement, or modification of signalling pathways, in a way that leads to bacteria being maintained on the surface.
Our data from in vivo infection experiments are consistent with galectin-3 being beneficial for meningococcal survival, as we found that the bacterial load in the blood was significantly lower in mice lacking galectin-3. Given the pleiotropic roles of both endogenous and exogenous galectin-3 in response to infection (Vasta, 2009), there are several potential explanations for this finding. One possibility is that the increased association to macrophages and monocytes we observed in vitro could lead to enhanced meningococcal survival, for example by allowing escape from phagocytosis by neighbouring cells, enhancing dissemination or affecting the activity of the phagocyte through altered TLR ligand presentation. Gal3−/− mice have been shown to be more susceptible to LPS-induced shock yet are relatively resistant to infection with Salmonella (Li et al., 2008). Bacteria may therefore use galectin-3 to mask LPS on the surface, promoting bacterial replication. We are currently analysing macrophage and monocyte cytokine responses to N. meningitidis in presence or absence of galectin-3 to elucidate how this interaction affects the induction of host immune responses to LPS during infection.
Overall, our data reveal an interaction between N. meningitidis and galectin-3. We propose that this interaction could occur during meningococcal infection at sites where local concentrations of galectin-3 secreted by activated immune cells could accumulate and bind to the surface of bacteria. As a consequence of this interaction, bacterial adhesion to the surface of phagocytes is increased and this could eventually lead to enhanced meningococcal survival. Further elucidation of the roles of galectin-3 during meningococcal infection will provide important insights into the contribution of this protein to bacterial survival and disease progression.
Bacterial strains and growth conditions
Neisseria meningitidis was grown on Brain Heart Infusion (BHI) agar (1.5% wt/vol, Oxoid) at 37°C in the presence of 5% CO2. MC58 is a serogroup B isolate of N. meningitidis. The mutants MC58ΔsiaD, MC58ΔsiaC and MC58Δlst, MC58ΔlgtA, MC58ΔlgtB, MC58ΔgalE, MC58Δlsi-1, MC58ΔicsB have been described previously (Table 1).
Galectin-3 and antibodies
The lectins used in this work were either purchased as recombinant purified protein from R&D Systems (human galectin-3) or purified after recombinant production by affinity chromatography on lactosylated Sepharose 4B. LPS contamination was removed by a further chromatographic step as previously described (Sarter et al., 2009). Purity was ascertained by one- and two-dimensional gel electrophoreses and mass spectrometry, which was also performed after proteolytic truncation with collagenase at Tyr106/Ile107 and Glu229/Ile230 peptide bonds and biotinylation was carried out under activity preserving conditions (Kubler et al., 2008). Activity controls were run by solid-phase and cell assays using the microtitre plate-based system and asialofetuin as matrix, as described in (Andre et al., 2009; Leyden et al., 2009). Polyclonal antibodies against full-length galectin-1, galectin-3 and galectin-4 were made and checked for specificity and absence of cross-reactivity to other galectins as previously described (Kaltner et al., 2002; Lohr et al., 2007). Rabbit poly-clonal anti-human galectin-3 antibody was also purchased from Santa Cruz Biotechnology.
Challenge of mice with live N. meningitidis
All animal experiments were carried out under protocols reviewed and approved by the Home Office, UK. For immunohistochemistry, groups of ten 6-week-old female BALB/c mice (Harlan) were housed under pathogen-free conditions, acclimatized for 1 week and randomly distributed in two groups. Animals were challenged by intraperitoneal injection of N. meningitidis MC58. Bacteria were grown overnight on BHI plates and then harvested into PBS. The number of colony-forming units was estimated by measuring the absorbance at 260 nm of a lysate of the suspension in P2 lysis buffer (Qiagen), and the number of viable bacteria was confirmed by plating to solid media. Each animal in the challenge group received 1 × 106 cfu of MC58 in 0.5 ml of BHI medium containing 0.5% (vol/vol) iron dextran (Sigma, Poole, UK). Control groups consisted of mice given BHI medium with iron dextran alone (uninfected). After 24 or 48 h, the animals were sacrificed and multiple organs were collected, fixed in 10% formalin for 1 week and embedded in paraffin using routine methods. Five micrometer thickness sections were cut from each block for analysis.
Bacteraemia was assessed in wild-type and galectin-3 deficient (Gal3−/−) C57BL/6 mice which were generated by targeted disruption of the galectin-3 gene (Hsu et al., 2000). Groups of eight or ten 8-week-old mice were challenged by intraperitoneal injection of N. meningitidis MC58 grown overnight at 37°C in 5% CO2 on BHI plates. Each animal received 1 × 108 cfu in 0.5 ml of BHI medium containing 0.5% (vol/vol) iron dextran and bacterial dose was verified by serial dilution and plating to BHI at the time of infection. Blood was obtained from mice 7 and 24 h after challenge, and serial dilutions were plated to BHI agar to quantify bacteria after overnight growth at 37°C, 5% CO2. Experiments were performed on two independent occasions. Counts for individual animals were plotted and results were analysed using a two-tailed, unpaired t-test.
The presence of galectins-1, -3 and -4 was assessed in different organs from mice challenged with N. meningitidis strain MC58 or uninfected animals and presence of galectin-3 was assessed in spleen tissue sections from patients with systemic meningococcal infection and control tissue from patients without bacterial infection. Tissues were obtained during a research study in 1991–1992 (Lucas et al., 1993). Paraffin-embedded sections were deparaffinized in xylene, and rehydrated in serially diluted ethanol (100%, 95%, 75%, 50% and 0%). Following washes with distilled water for 5 min, sections were incubated with 3% methanol for 10 min to quench endogenous peroxidase. For antigen retrieval specimens were boiled in citrate buffer under full vapour pressure for 2 min, and washed in distilled water and in PBS/0.1% Tween-20. Following incubation with 150 μl of protein block (Dako) for 5 min, samples were incubated overnight at 4°C with 200 μl of polyclonal anti-galectin-3 antibody (2 μg ml−1) or 200 μl of a 1/500 dilution of polyclonal antibody against N. meningitidis (Fitzgerald Industries International). The following day, sections were washed with PBS/0.1% Tween-20 and incubated with 100 μl of goat anti-rabbit antibody conjugated with horseradish peroxidase (Dako) or alkaline phosphatase (Dako) for 30 min. After washing with PBS/0.1% Tween-20, galectins in the tissues were visualized by incubation in diaminobenzidine (DAB) for horseradish peroxidase or permanent red for alkaline phosphatase. Finally tissue sections were counterstained with Mayer's hematoxylin, mounted and analysed with a light microscope. The immunohistochemical analysis for different galectins was repeated at least three times.
The monocytic cell line THP-1 and the nasopharyngeal epithelial cell line Detroit 562 were purchased from ATCC. THP-1 cells were grown in RPMI 1640 medium (Gibco, Invitrogen) with 2 mM glutamine (Invitrogen) supplemented with 10% foetal calf serum (PAA Laboratories), 10 mM Hepes, 1 mM sodium pyruvate, 1.25 g d-glucose, 0.05 mM β-mercaptoethanol. THP-1 cells (1 × 106) were differentiated into macrophages in 24-well plates containing 1 ml RPMI 1640 medium with 100 ng ml−1 PMA for 72 h prior to experiments. Detroit 562 epithelial cells were cultured in Minimum Essential Medium (Eagle) with non-essential amino acids (Invitrogen). Detroit 562 medium was supplemented with 2 mM GlutamaxTM (Invitrogen), 10% foetal calf serum, 1 mM sodium pyruvate and lactalbumin hydrolysate (0.1%). Cells were cultured to a density of 1–5 × 105 cells ml−1 and maintained at 37°C with 5% CO2.
Transfection of THP-1 cells was performed using Thermo Scientific Dharmacon On Target plus SMARTpool, a four siRNA oligo system designed specifically for galectin-3. The RNA sequences targeted are: GGAGAGUCAUUGUUUGCAA, GUACAAUCAUCGGGUUAAA, GGCCACUGAUUGUGCCUUA, CGGUGAAGCCCAAUGCAAA. Briefly, 1 × 106 cells were transfected with 3 μg of siRNA using Nucleofector solution V (Dharmacon) according to manufacturer's instructions. Negative controls were carried out using only the Nucleofector buffer and electroporation was carried out with a Nucleofector machine (Amaxa) using the program V-01/V-001 (for high transfection efficiency). Cells were then cultured in culture medium supplemented with PMA for 24 h to allow attachment. The medium was then replaced with fresh medium to remove dead cells and the cells were cultured for a further 48 h to allow differentiation into macrophages. To confirm silencing of galectin-3, Western blot analysis was performed on the cells 72 h after the initial transfection. All the experiments were performed on cells with no detectable galectin-3.
Preparation of human monocytes
Peripheral blood was extracted from healthy consenting volunteers in accordance with local research committee guidelines and density centrifugation of blood-EDTA on Histopaque 1077 gradient was performed as per the manufacturer's instructions (Sigma Aldrich). Peripheral blood mononuclear cells were gently washed off and seeded at 3 × 105 cells ml−1 in 24-well plates in RPMI medium without serum and positively selected by plastic adherence for 3 h at 37°C, after which residual non-adherent B and T cells were washed away. Cell viability was estimated by Trypan blue exclusion and was higher than 95%. Monocytes were infected with N. meningitidis as described for adhesion/internalization assays.
For analysis by flow cytometry, bacteria were grown overnight on BHI agar and then harvested into PBS. The number of bacteria was estimated by measuring the OD at 260 nm of an aliquot of the bacterial suspension lysed in 200 mM NaOH/1% SDS (v/w). Bacteria (2 × 108 cfu) were fixed in 3% paraformaldehyde for 2 h at room temperature, washed three times in PBS, and stored at −80°C in PBS/15% glycerol. To measure galectin binding, 2 × 107 fixed bacteria were incubated with 30 μl of label free or biotinylated galectins (Andre et al., 2006) at a final concentration of 3.3 μM (100 μg ml−1) for 1 h at 37°C and when required lactose was added at a concentration of 100 mM. After two washes with PBS/Tween 0.1% binding was detected following incubation with a rabbit anti-human galectin antibody (Lohr et al., 2008) for 45 min at 4°C or a mouse anti-biotin antibody. The cells were washed twice in PBS/Tween 0.1%, then resuspended in 50 μl of PBS with a polyclonal donkey anti-rabbit IgG-Cy2 conjugate (1/200 dilution in PBS; Jackson ImmunoResearch Laboratories) or a polyclonal donkey anti-mouse IgG FITC conjugate (1/200 dilution; Jackson ImmunoResearch Laboratories) and incubated for 30 min on ice. Galectin binding was quantified using a flow cytometer (Calibur FACScan; BD Biosciences), and at least 4 × 104 events were recorded. Galectin-3 binding was expressed by calculating the Fluorescence Index [% positive gated bacteria multiplied by the geometric mean fluorescence (Findlow et al., 2006)]. Unless otherwise stated, data shown is the mean and standard deviation and the number of experiments is specified in the figure legends.
RNA procedures and RT-PCR
The RNA samples were isolated by using RNeasy mini kit (QIAGEN) and genomic DNA was removed by treatment with DNase (QIAGEN). 400 ng RNA was reverse transcribed to cDNA using the QuantiTect reverse transcription kit (QIAGEN). Transcript levels were measured by qrtRT-PCR using QuantiFast with SYBR green detection (QIAGEN) on a thermal cycler (Rotor-Gene 3000; Corbett Research). Results for the transcription of galectin-3 were normalized with levels for the housekeeping gene β-actin (primers 5′-CTCTTCCAGCCTTCCTTCCT-3′ and 5′-GCACTGTGTTGGCGTACAG-3′). The following primers were used to monitor transcript levels: forward 5′-TGTGCCTTATAACCTGCCTTT-3, reverse 5′-TTAAAGTGGAAGGCAACATCA-3′. Data were analysed using the Comparative Quantitation method by Rotor-Gene software (version 6.0; Corbett Research). Controls included reactions with no template and samples of RNA that had not been treated with reverse transcriptase (RT). qrtRT-PCR was performed in triplicate on cDNA samples derived from three independent assays.
Western blot analysis
Cell pellets (1 × 106 cells) were resuspended in SDS-PAGE loading buffer (50 mM Tris-HCl, 2% SDS, 100 μM β-mercaptoethanol, 10% glycerol, 0.1% bromophenol blue), protein samples were separated under reducing conditions by 12% SDS-PAGE, then transferred to PVDF (polyvinylidene fluoride) membranes (Millipore) for 1 h in cold transfer buffer (25 mM Tris, 190 mM glycine). After blocking overnight in 5% milk in PBS, galectin-3 was detected by incubation of the membrane with anti-galectin-3 (300 ng ml−1; Santa Cruz Biotechnology) or anti-actin (Sigma) for 1 h at room temperature. Binding of the primary antibody was detected using an anti-rabbit-IgG Horse Radish Peroxidase-conjugated secondary antibody (1:1000 dilution; Santa Cruz Biotechnology), and ECL detection kit (GE Healthcare).
Fluorescence and confocal microscopy
For fluorescence and confocal microscopy, Detroit cells were seeded on glass coverslips and infected with MC58 at an MOI of 30 for 4 h at 37°C. Samples were washed twice with PBS, fixed for 20 min at room temperature in 3% paraformaldehyde in PBS, and then washed three times in PBS. For labelling of bacteria the cells were incubated for 1 h with an anti-LPS antibody (α-L3,7,9 NIBSC) and with an anti-galectin-3 antibody (Santa Cruz Biotechnology), after which the cells were washed twice in PBS then incubated for 30 min with an anti-mouse Alexa Fluor 488-conjugated antibody and anti-human Alexa Fluor 555-conjugated antibody (Invitrogen) to label bacteria and cells respectively. DAPI staining was also performed to visualize nuclei. After labelling, cells were washed twice in PBS and then double distilled water, mounted in Aqua Poly/Mount (Polysciences), and analysed by fluorescence microscopy (BX40, Olympus) or by laser-scanning confocal microscopy (Zeiss Axiovert LSM510).
Adhesion and internalization assays
For adhesion and internalization assays Detroit 562 cells, THP-1 cells differentiated with PMA and human primary monocytes from healthy volunteers were used. Bacteria were harvested following overnight culture on solid media and resuspended in 400 μl of PBS. The concentration of the bacterial suspension was adjusted in culture media to give the desired MOI. PMA differentiated THP-1 and Detroit 562 were infected at an MOI of 30 for 1 h at 37°C, while human primary monocytes were infected at an MOI of 50 for 90 min at 37°C, in each case following pre-incubation of bacteria with 30 μl of recombinant galectin-3 (3.3 μM) or PBS for 1 h at 37°C. For adhesion assays, cells were washed extensively with PBS to eliminate non-adherent bacteria, then lysed with 1% saponin for 10 min at 37°C to recover cell associated bacteria. For internalization assays, cells were washed with PBS to eliminate non-adherent bacteria and exposed to gentamicin (100 μg ml−1) for 15 min to kill extracellular bacteria. Cells were washed extensively with PBS after gentamicin treatment to remove the antibiotic and dead extracellular bacteria, and subsequently lysed or reincubated in medium for 1 or 2 h. In both adhesion and internalization assays serial dilutions of bacteria were plated following cell lysis or gentamicin treatment and colonies were counted the following day. Bacteria recovered from cells without gentamicin treatment represented the total number of bacteria associated with cells; bacteria recovered from cells treated with gentamicin represent the number of internalized bacteria.
The statistical significance of galectin-3 binding or interaction with cells was determined with the Graph Pad Prism 5 Software by using the Student's unpaired t-test for flow cytometry data analysis, the one-tailed, paired Student's t-test for cell adhesion or internalization assay data analysis and a two-tailed, unpaired t-test for analysis of levels of bacteraemia. Statistical significance is indicated with asterisks where *P < 0.05, **P <0.01 and ***P <0.001 and ****P <0.0001. Exact P-values are shown in the figure legends.
We thank David Holden and his group for helpful comments and are grateful to Antonio Santos for assistance with confocal microscopy. M. H. was supported by the DFG (SFB643), the Interdisciplinary Center of Clinical Research (IZKF) at the University Hospital of the University of Erlangen-Nuremberg and the K&R Wucherpfennigstiftung. H. J. G. wishes to acknowledge the generous funding by the EC research program GlycoHIT and inspiring discussions with Dr B Friday. Work in C. T.'s lab is funded by the MRC and Wellcome Trust.