Biochemical Journal

Research article

Galectin-8 tandem-repeat structure is essential for T-cell proliferation but not for co-stimulation

Valentina Cattaneo, María V. Tribulatti, Oscar Campetella


Gal (galectin)-8 is a tandem-repeat Gal containing N-CRDs (Nterminal carbohydrate-recognition domains) and C-CRDs (C-terminal carbohydrate-recognition domains) with differential glycan-binding specificity fused by a linker peptide. Gal-8 has two distinct effects on CD4 T-cells: at high concentrations it induces antigen-independent proliferation, whereas at low concentrations it co-stimulates antigen-specific responses. Associated Gal-8 structural requirements were dissected in the present study. Recombinant homodimers N–N (two N-terminal CRD chimaera) and C–C (two C-terminal CRD chimaera), but not single C-CRDs or N-CRDs, induced proliferation; however, single domains induced co-stimulation. These results indicate that the tandem-repeat structure was essential only for the proliferative effect, suggesting the involvement of lattice formation, whereas co-stimulation could be mediated by agonistic interactions. In both cases, C–C chimaeras displayed higher activity than Gal-8, indicating that the C-CRD was mainly involved, as was further supported by the strong inhibition of proliferation and co-stimulation in the presence of blood group B antigen, specifically recognized by this domain. Classic Gal inhibitors (lactose and thiodigalactoside) prevented proliferation but not co-stimulatory activity, which was inhibited by 3-O-β-D-galactopyranosyl-D-arabinose. Interestingly, Gal-8 induced proliferation of naïve human CD4 T-cells, varying from non- to high-responder individuals, whereas it promoted cell death of phytohaemagglutinin or CD3/CD28 pre-activated cells. The findings of the present study delineate the differential molecular requirements for Gal-8 activities on T-cells, and suggest a dual activity relying on activation state.

  • galectin alternative inhibitor
  • glycan binding
  • lattice formation
  • T-cell activation


Gals (galectins) constitute a family of mammalian lectins that are involved in a broad spectrum of cellular responses such as proliferation, apoptosis, differentiation, adhesion, migration and cytokine secretion [1,2]. Many of them exert immunoregulatory functions, playing critical roles in immune system homoeostasis [25]. They are characterized by the presence of conserved CRDs (carbohydrate-recognition domains) and are classified into three subgroups based on their structure: prototype (one CRD, such as Gal-1, -2 and -7, among others), tandem-repeat (two linked CRDs, such as Gal-4, -8 and -9) and chimaera (one CRD fused to a non-lectin domain, only Gal-3). Prototypical and chimaeric Gals can form homodimers or multimers respectively. Thus, acting as bi- or multi-valent agents, Gals participate in many cellular processes such as cell–cell and cell–matrix interactions. Another outstanding feature of this family of proteins is the formation of ordered cell-surface Gal–glycan structures termed lattices, which engage specific cell-surface glycoconjugates by traditional ligand–receptor interactions. These structures are involved in the control of receptor endocytosis, host–pathogen interaction and immune system homoeostasis [6].

Gal-8, which belongs to the tandem-repeat group, is intrinsically a heterodimer because it has two distinct N- and C-terminal CRDs (N-CRD and C-CRD) joined by a hinge linker peptide of variable length. Gal-8 is expressed in different organs and tissues under physiological or pathological conditions, and is also found in several human cancers [712]. Although both CRDs of Gal-8 are structurally homologous [13], they present different sugar-binding specificity. In fact, the affinity with which each domain recognizes different glycan ligands has been extensively studied through multiple approaches. Several groups have reported the preference of sialylated and sulfated glycans for the N-CRD; and blood group antigens and poly-N-acetyl-lactosamine glycans for the C-CRD [1417]. Various functions exerted by Gal-8, such as the modulation of cell adhesion, are mediated by cross-linking of integrins and concomitant activation of the associated signalling pathways, where the Gal tandem-repeat structure is always required [1719].

Two Gal-8 isoforms differing only in the length of the linker region are expressed in the mouse thymus and spleen [20]. We have previously found that Gal-8 exerts two different activities on peripheral CD4 T-cells. At high concentrations it induces strong antigen-independent CD4 T-cell proliferation, whereas at low concentrations it co-stimulates T-cells in the presence of APCs (antigen-presenting cells) and the corresponding antigen [21]. These activities are mediated by the interaction with the T-cell-surface glycoprotein CD45 and involve the activation of Zap-70 and MAPK (mitogen-activated protein kinase) signalling activation pathways [21]. The proliferative and co-stimulatory activities exerted by Gal-8 on T-cells are particularly interesting since no similar effects have been reported for other Gals, and suggest a possible involvement of this lectin in inflammatory and autoimmune processes. The aim of the present study was to investigate the molecular basis of Gal-8 activation of mouse and human T-cells, focusing on the tandem-repeat structure of this lectin and the differential ligand-specificity of each CRD.


Expression of recombinant proteins

Recombinant N-CRD, C-CRD, N–N (two N-CRD chimaera) and C–C (two C-CRD chimaera) proteins were generated using a mouse Gal-8L-encoding plasmid as the template and following different PCR strategies (for details see [22]). Mutations in the N-CRDs or C-CRDs of human Gal-8M were generated by replacing the corresponding codons as R69H (protein Gal-8N69) and R233H (protein Gal-8C233) [23] in synthetic constructions. Conditions for protein expression and purification by lactosyl-Sepharose (Sigma) followed by immobilized metal-affinity chromatography (GE Healthcare) were as described for mouse Gal-8 [20]. Lectin activity of these recombinant proteins was tested by haemagglutination assays.

Inhibition of haemagglutination activity

Haemagglutination assays were carried out as described previously [20]. To screen for putative Gal-8 inhibitors, the lectin was pre-incubated with 5 mM of several commercially available lactose-related sugars obtained from Sigma as shown in Supplementary Figure S1 at To test for inhibition of other mouse Gal activity, Gal-8 (70 μg/ml), Gal-3 (100 μg/ml) or Gal-1 (50 μg/ml) were pre-incubated with TDG (D-galactopyranosyl-β-D-thiogalactopyranoside) or Gal-Ara (3-O-β-D-galactopyranosyl-D-arabinose) (concentrations ranging from 5 to 30 mM) in a final volume of 50 μl before addition of a 4% (v/v) suspension of PBS-washed mouse RBCs (red blood cells).

Antibodies and reagents

Affinity-purified rabbit anti-(mouse Gal-8) antibodies were obtained as described previously [21]. An FITC-conjugated anti-human CD45 monoclonal antibody was from Becton Dickinson. Vectashield mounting medium was from Vector Laboratories. The inhibitor of ERK (extracellular-signal-regulated kinase) 1/2 phosphorylation U0126 was from Cell Signaling Technology. The CD45 PTPase (protein tyrosine phosphatase) inhibitor N-(9,10-dioxo-9,10-dihydro-phenanthren-2-yl)-2,2-dimethyl-propionamide [24] and lactose were from Calbiochem. TDG and Gal-Ara were from Sigma. Galα1–3[Fucα1–2]Galβ1–4GlcNAcβ (Te 223, β tetra type 2) and Neu5Acα2–3Galβ1–3GlcNAcβ (Tr 34, 3′SLec) were obtained from The Glycoconsortium for Functional Glycomics resources. PHA (phytohaemagglutinin) was from Invitrogen.

Mice, cell lines and cell purification.

C57BL/6J and C.Cg-Tg(DO11.10)10Dlo/J (DO11.10) breeding pairs were obtained from The Jackson Laboratory (Bar Harbor, ME, U.S.A.) and bred in our facilities. The human Jurkat T-cell line was obtained from the American Type Culture Collection. For mouse splenocyte purification, spleens from 4–8-week-old animals were removed and disrupted against a stainless steel mesh in RPMI 1640 medium (Invitrogen). The cell suspension was washed and incubated with RBC lysis buffer (Sigma) and washed again with medium. Freshly isolated human PBMCs (peripheral blood mononuclear cells) were purified from whole blood samples obtained by venous puncture from healthy donors. Samples were drawn in sodium citrate-containing polypropylene tubes and lymphocytes were purified using Ficoll–Paque PLUS (GE Healthcare). For CD4 T-cell purification from human PBMCs, paramagnetic beads (Myltenyi Biotec) were used. Cell purity was tested by FACS using PE (phycoerythrin)-labelled anti-CD4 and PerCP (peridinin chlorophyll protein complex)-labelled anti-CD3ϵ monoclonal antibodies from BD Biosciences, rendering 99.6% purity. All experiments involving animals or human donors were approved by the Ethics Committee of our Institution (Instituto de Investigaciones Biotecnológicas, Universidad Nacional de San Martin, Buenos Aires, Argentina), and human donors provided informed consent.

Binding assays

Assays were performed as described previously [21]. Briefly, mouse splenocytes were treated in the presence of 0.1 μM Gal-8, N-CRD, C-CRD, N–N or C–C proteins and washed with ice-cold PBS, or PBS plus 100 mM lactose. After incubation with rabbit anti-Gal-8 and FITC-conjugated anti-rabbit IgG antibodies (Molecular Probes), cells were fixed in 2% paraformaldehyde in PBS and analysed by flow cytometry.

CD45 surface localization

Experiments were performed as described in [25]. Briefly, 5×105 Jurkat T-cells were treated with 5 μM human Gal-8 for 10 min on ice followed by 20 min at 37 °C to allow migration of counter-receptors on the cell surface. Cells were then cooled to 4 °C and bound Gal-8 was dissociated on ice-cold 100 mM lactose. The cells were immediately fixed with 2% paraformaldehyde for 30 min and the reaction was quenched with 0.2 M glycine for 5 min on ice. After staining for 1 h with FITC-conjugated anti-CD45 antibody, cells were washed and dropped on to poly-L-lysine (Sigma)-coated glass microscope slides. Labelled cells were mounted using Vectashield mounting medium and analysed with the ×100 objective on an epifluorescence microscope (Nikon).

Cell proliferation and co-stimulation assays

Proliferation assays on mouse cells were performed as described previously [21]. Briefly, splenocytes (5×105 cells) from C57BL/6J mice were cultured at 37 °C in 5% CO2 for 48 h in flat-shaped 96-well plates in 0.2 ml of RPMI 1640 medium in the presence of 10% FBS (fetal bovine serum, Invitrogen), 2 mM glutamine and 5 μg/ml gentamicin (complete medium). For co-stimulatory assays, splenocytes (3×105 cells) from DO11.10 mice were cultured for 48 h in 0.2 ml of complete medium in the presence of the cognate ovalbumin OVA-(323–339) peptide at 0.4 μg/ml (Sigma–Genosys). For human cell-proliferation assays, freshly isolated human PBMCs or purified CD4 T-cells (5×105 cells) were cultured for 72 h in U-shaped 96-well plates (Nunc) in 0.2 ml of complete medium. PHA was included as a control for cell proliferation. For T-cell activation, PBMCs (5×105 cells) were seeded on to anti-CD3ϵ (BD Biosciences)-coated plates (0.1 μg/ml in PBS) in the presence of 0.5 μg/ml anti-CD28 antibody (BD Biosciences). After 3 days of culture, Gal-8 (5 μM) was added and cells were incubated for an additional 24 h. Parallel cultures were performed and fixed with ice-cold 70% ethanol in PBS in order to asses hypoploidy by FACS (see below).

In all proliferation assays, 1 μCi of [3H]methylthymidine (NEN) was added to each well 16 h before harvesting. TDG (30 mM), lactose (50 mM) and CD45 PTPase inhibitor (0.5 μM) were added 30 min before the addition of recombinant proteins, except for the U0126 inhibitor (10 μM) which was added 1 h before. Basal c.p.m. corresponding to unstimulated cells ranged from 200 to 1000 and was discounted in all experiments. Assays were performed in quadruplicate. The PI (proliferation index) was calculated as the treated/untreated cells c.p.m. ratio.

Cell-death assays

Jurkat T-cells (5×105 cells) were cultured in 0.5 ml of complete medium in the presence of 2.5–5 μM human Gal-8. Cells were starved for 2 h in serum-depleted medium before Gal addition. For the cell-death positive control, Jurkat T-cells were kept for 4 days in RPMI 1640 medium plus 0.1% FBS. After 18 h of incubation at 37 °C in 5% CO2, cells were washed with 100 mM lactose in PBS to detach them from the plate bottom. To assess hypoploidy, cells were fixed in ice-cold 70% ethanol in PBS, washed and dispersed in propidium iodide solution [5 μg/ml in 0.1% Triton X-20, 200 μg/ml RNAse A and 0.1% sodium citrate (all from Sigma)] until FACS analysis.

Flow cytometry

The FlowMax cytometer PASIII (Partec) and WinMdi 2.9 software were used.

Statistical analysis

A Student's t test was used.


The Gal-8 bivalent structure is critical for the induction of cell proliferation

When assayed at relatively high concentrations (0.5–2 μM), Gal-8 is able to induce strong mouse CD4 T-cell proliferation in the absence of antigen [21]. To analyse the dependence on tandem-repeat structure for this activity, we generated mouse recombinant Gal-8 N-CRDs and C-CRDs, as well as chimaeras containing two equal domains linked by the hinge region, named N–N and C–C (see the scheme in Figure 1A). Neither the addition of 1 μM of separate N-CRDs or C-CRDs, nor their equimolar mixture, was able to mimic the strong cell proliferation achieved with the entire Gal-8 protein (Figure 1B). These recombinant proteins also failed to induce proliferation even when tested at 2 μM (results not shown). These results suggest that bivalency was therefore an absolute requirement to induce antigen-independent proliferation. To test this hypothesis, 1 μM N–N and C–C chimeras were then assayed. N–N induced proliferation at a rate similar to that of Gal-8 but, strikingly, C–C was even more efficient, rendering a 4-fold increased proliferation rate (Figure 1C). These results confirm that bivalent structure is essential to trigger T-cell proliferation, and also suggest that the C-CRD is mainly responsible for this activity. The involvement of each CRD was further tested by pre-incubating cells in the presence of blood group B (Galα1–3[Fucα1–2]Galβ1–4GlcNAcβ) or sialylated N-acetyl-lactosamine (Neu5Acα2–3Galβ1–3GlcNAcβ) that are specifically recognized by the C-CRD and N-CRD respectively. As shown in Figure 1(D), only blood group B was able to preclude proliferation, supporting the hypothesis that the C-CRD is the major domain involved in Gal-8-induced proliferation.

Figure 1 Tandem-repeat structure of Gal-8 is necessary to induce proliferation

(A) Schematic diagram of the recombinant proteins used throughout the present study. (BD) Mouse splenocyte proliferation assays. Splenocytes from C57BL/6J mice were cultured for 48 h in the presence of (B) 1 μM of Gal-8, N-CRD, C-CRD or an equimolar mixture of N-CRD+C-CRD proteins or (C) 1 μM of Gal-8, N–N and C–C proteins. Cells were pre-incubated for 30 min with 5 mM Te 223 (Galα1–3[Fucα1–2]Galβ1–4GlcNAcβ) or Tr 34 (Neu5Acα2–3Galβ1–3GlcNAcβ), before Gal-8 addition (D). (E) Flow-cytometry histograms showing the binding capacity of Gal-8, N-CRD, C-CRD, N–N and C–C to mouse splenocytes. The control was purified rabbit anti-Gal-8 IgG in the absence of recombinant proteins. Lact, cells were washed with 100 mM lactose after treatment with recombinant proteins. (F) CD45 surface localization on Gal-8-treated cells. Jurkat T-cells were incubated with 5 μM of Gal-8 (II) or PBS (I), stained with FITC-conjugated anti-CD45 and analysed by epifluorescence microscopy. For the control, cells were incubated with lactose before Gal-8 addition (III). Displayed images were observed in more than 75% cells in each case. Proliferation assays are representative of three independent experiments *P< 0.05; #P< 0.0001.

Next, we analysed whether the binding ability of these recombinant proteins to mouse splenocytes was related to their distinct proliferative activity. Although the homodimeric chimaeras N–N and C–C bound at a rate similar to Gal-8, the binding ability of single CRDs was reduced, probably reflecting a reduced avidity that precludes lattice formation (Figure 1E). The observation that lactose could displace Gal-8 and, although to a lesser extent, C–C, but not N–N, binding (Figure 1E, I–III) is consistent with the inhibitory capacity of the lactose analogue TDG over the proliferative activity of these proteins, as shown in Figure 1(C).

The results of the present study up until now demonstrate that the bivalent structure of Gal-8 is strictly required to induce T-cell proliferation, raising the idea that it could be acting by cross-linking of counter-receptors. Since CD45 was identified as a Gal-8 counter-receptor [21], Jurkat T-cells were incubated in the presence of Gal-8, fixed and stained for CD45. As shown in Figure 1(F), Gal-8 induced CD45 cluster formation at the cell surface, an effect that relys upon lectin–glycan interaction as lactose prevented the interaction, suggesting that its activity on T-cells involves lattice formation.

Both N-CRDs and C-CRDs are able to induce T-cell co-stimulation

At suboptimal concentrations to induce proliferation, Gal-8 is able to co-stimulate T-cell antigen-specific responses [21]. To analyse whether the tandem-repeat structure is also required for the co-stimulatory effect, splenocytes from TCROVA transgenic DO11.10 mice were incubated with 0.1 μM of the N-CRD, C-CRD, N–N or C–C proteins, together with the cognate OVA peptide. Interestingly, all of these proteins were able to co-stimulate T-cells but, surprisingly, the C–C chimaera displayed the strongest effect (Figure 2A). In fact, C–C induced co-stimulation at doses as low as 0.025 μM (results not shown). From these results it can be concluded that, in strong contrast with the proliferative activity, the Gal-8 bivalent structure is not essential to promote co-stimulation. This effect may be elicited through an agonistic induction pathway that does not precisely require lattice formation. Nevertheless, in both Gal-8-induced properties the C-CRD seems to be the main CRD involved. To gain further insights into the biological activity of each domain on the co-stimulatory activity, blood group B and sialylated N-acetyl-lactosamine were also tested. Consistently, despite both sugars being able to inhibit co-stimulation, blood group B (specifically recognized by C-CRD) was more effective than the sialylated compound (Figure 2B). As we have previously observed for Gal-8 [21], the inhibition of ERK phosphorylation or CD45 PTPase activities prevented the co-stimulation induced by all Gal-8-derived recombinant proteins (results not shown). On the other hand, the addition of TDG only partially inhibited Gal-8 co-stimulation and showed no effect on single or homodimeric CRDs (Figure 2A). Because Gal-8 exerts its co-stimulatory effect at very low concentrations, it could be feasible that high-affinity interactions are responsible for this phenomenon.

Figure 2 N-CRD, C-CRD, N–N and C–C recombinant proteins are able to induce a T-cell co-stimulatory effect, with C–C being the strongest activator

Co-stimulatory assays using the DO11.10 TCROVA model. (A) Splenocytes from DO11.10 mice were cultured for 48 h in the presence of the cognate OVA-(323–339) peptide, together with 0.1 μM Gal-8, N-CRD, C-CRD, N–N or C–C proteins. (B) Cells were pre-incubated for 30 min with 5 mM Te 223 (Galα1–3[Fucα1–2]Galβ1–4GlcNAcβ) or Tr 34 (Neu5Acα2–3Galβ1–3GlcNAcβ), before Gal-8 addition. Results are representative of three independent experiments. *P< 0.05; #P< 0.0001.

Gal-Ara inhibits the Gal-8 co-stimulatory effect

Although lactose and TDG almost abrogated Gal-8-induced proliferation, they were unable to fully inhibit the co-stimulatory effect seen in the DO11.10 model (see Figure 2A). A feasible explanation for this could be the fact that co-stimulation involves high-affinity agonistic interactions, since it can be achieved even with low concentrations of single CRDs. In a search for putative inhibitors of the Gal-8 co-stimulatory effect, several commercially available lactose-related sugars were tested for their ability to inhibit Gal-8-induced haemoagglutination (Supplementary Figure S1). Gal-Ara and 4-O-α-D-galactopyranosyl-D-glucitol (lactitol) were selected for further assays, based on their availability and low cost. Similarly to lactose, these sugars were able to reduce Gal-8 binding to splenocytes (results not shown) and to fully inhibit the proliferative activity of Gal-8 (Figure 3A). Strikingly, Gal-Ara completely inhibited the Gal-8 or its single CRD co-stimulatory effect in contrast with that observed with lactose, TDG or lactitol (Figures 3B and 3C). Any possible deleterious effect of Gal-Ara on cell viability was ruled out by its inability to alter PHA-induced T-cell proliferation (results not shown). Therefore these results suggest existing differences at the sugar recognition level on the two Gal-8 biological activities on T-cells.

Figure 3 Gal-Ara inhibits Gal-8-mediated proliferative and co-stimulatory effects

(A) Proliferation assays using splenocytes from C57BL/6J mice in the presence of 2 μM of Gal-8. (B and C) Co-stimulatory assays using splenocytes from DO11.10 mice in the presence of the cognate OVA peptide together with 0.1 μM Gal-8 (B); or 0.1 μM N-CRD or C-CRD proteins (C). In all cases, cell cultures were pre-incubated for 30 min with 30 mM Gal-Ara or lactitol, before addition of recombinant proteins. *P< 0.01.

Because the finding of Gal inhibitors constitutes an area of growing interest, Gal-Ara was also assayed with Gal-1 and Gal-3. This sugar prevented the haemoagglutination induced by these Gals (results not shown) indicating that Gal-Ara might be used to inhibit not only Gal-8 functions, but also those mediated by other prominent members of the Gal family.

Dual activity of Gal-8 on human PBMCs

The findings on the effect of Gal-8 on mouse T-cells prompted us to extend our studies to human cells. Proliferation was tested in PBMCs from normal donors with 5 μM recombinant human Gal-8. Responses ranging from non-responder (considered as those whose PI<10) to high-responder individuals (PI>70) were observed (Figure 4A), an effect that was abrogated by the presence of lactose. The differences observed between responders compared with non-responders could not be ascribed to gender (n=17; 9 women and 8 men), age (33±5, range 24–39 years) or blood group, and were consistently reproduced after collection of different samples from the same individuals. This also did not correlate with a differential binding capacity of Gal-8 at the cell surface (results not shown). To further characterize PBMCs from non-responder donors, increasing doses of Gal-8 (from 0.1 to 40 μM) were tested. Only the higher doses resulted in a mild increase in the PI, thus confirming that these cells were actually reluctant to Gal-8-induced proliferation (Figure 4B). No significant increase in the percentage of hypoploid cells was observed after incubation with Gal-8 (2 or 5 μM) for 24–72 h as assessed by propidium iodide staining (results not shown), ruling out the possibility that cell death was responsible for the absence of proliferation. Since CD4 T-cells are the main target cells of Gal-8-induced proliferation in mice [21], highly purified CD4 T-cells were obtained from responsive donors and cultured in the presence of this lectin. Strong proliferation of this T-cell subpopulation was recorded (Figure 4C). Since other researchers have reported a pro-apoptotic effect of Gal-8 on human Jurkat T-cells and activated PBMCs [26] that seems to be in contrast with the proliferative response reported in the present study, we searched for this activity. In agreement with results previously reported by Lu et al. [27], we found no significant apoptotic effect on Jurkat T-cells incubated in the presence of Gal-8 (Supplementary Figure S2 at However, a reduced PI was observed when Gal-8 (5 μM) was assayed on either PHA- or CD3/CD28-stimulated PBMCs (Figure 4D), a finding that correlated with the increased number of hypoploid cells (Figure 4E). Therefore a dual activity on T-cells arises because Gal-8 provides a stimulatory effect on naïve T-cells, but it can also lead activated cells to death, an observation that supports a regulatory role of Gal-8 in the immune response.

Figure 4 Gal-8 stimulates freshly isolated PBMCs, whereas it induces cell death of the pre-activated PBMCs

(A) Individual variation to Gal-8-induced proliferation. PBMCs from healthy donors were incubated with 5 μM human Gal-8. After 72 h, cell proliferation was quantified and the PI was calculated as described in the Experimental section, allowing the discrimination between responsive and non-responsive individuals. (B) PBMCs from an unresponsive donor were incubated for 72 h in the presence of 0.1–40 μM human Gal-8, and proliferation was quantified. (C) Freshly purified CD4 T-cells from PBMCs or the remanent cells that passed through the column (‘flow-through’) were assayed as described in (A). (D) PBMCs from a responder donor were stimulated either with PHA or CD3/CD28 stimuli for 72 h, and then 5 μM Gal-8 was added and incubated for an additional 24 h. Parallel cultures were assayed for hyploidy by propidium iodide staining and flow cytometry analysis (E). Findings for Gal-8-induced cell death of pre-activated PBMCs were similar for responders and non-responder individuals. Results are representative of three independent experiments. *P< 0.01.

Functional C-CRD is required for human T-cell proliferation

To address the involvement of each CRD on human T-cell proliferation, we assayed two recombinant human Gal-8 proteins with individually mutated CRDs such that the carbohydrate-recognition capacity was abolished (named Gal-8N69 or Gal-8C233) [17,28,29]. Gal-8N69, containing an intact C-CRD, induced proliferation at an even higher rate than Gal-8 itself. In strong contrast, Gal-8C233, containing an intact N-CRD, was completely inactive (Figure 5), indicating that proliferation of human PBMCs depends on a functional C-CRD. It is worth noting that both mutated proteins are able to dimerize via the N-CRD, independently of intact sugar recognition ability [17]. In fact, Gal-8N69, but not Gal-C223, retained haemoagglutinatory activity (results not shown). Taken together, our findings are in agreement with that reported by Stowell et al. [17], where human Gal-8 actually exists as a dimer via N-CRD dimerization, exposing both C-CRDs to accomplish activation. In conclusion, these results establish a correlation with mouse cells where the C-CRD and the ability to build lattices were required to induce proliferation (Figure 1).

Figure 5 Proliferative effects of Gal-8-derived proteins with mutated CRDs

PBMCs from a responsive donor were cultured for 72 h in the presence of 5 μM human Gal-8 or the mutant proteins Gal-8N69 and Gal-8C233. Results are representative of three independent experiments. P< 0.005; #P< 0.00001. Lact, lactose.


Gals arise as important regulators of the immune response and inflammation. Gal-8 is widely expressed and exerts several activities on cells from the immune system, such as apoptosis, proliferation and lymphocyte co-stimulation. To gain further knowledge of the molecular basis of these Gal-8 activities, we designed homodimeric and single-domain Gal-8-derived recombinant proteins. In contrast with single-domain proteins that were unable to trigger cell proliferation, the N–N and C–C homodimers induced strong proliferative activity. In fact, the C–C chimaera was even more effective than Gal-8. The absolute requirement of the tandem-repeat structure for the induction of cell proliferation supports the idea of the building of a lattice at the cell surface. This provides a plausible explanation for the high lectin concentrations required. It has been reported recently that the joining of two molecules of the prototype Gal-1 with a linker peptide confers increased potency in cell-death induction, by forming higher-order multimers with increased valency [30]. In this context, it is to be noted that Gal-8 tends to form dimers and that this dimerization occurs through the N-CRD, then rendering two C-CRDs to be exposed [17]. This structure might explain the increased proliferative effect of the C–C protein, an observation that correlates with the strong inhibition of Gal-8 activities by the blood group B (Te 223) that is recognized by the C-CRD. A similar situation takes place in human cells where the disruption of the C-CRD sugar-recognition site, but not that from the N-CRD, abolishes the proliferative activity of the lectin. Notably, the inactivation of C-CRD leads to the loss of cell adhesion induced by Gal-8 in Jurkat T-cells [28] and neutrophils [29]. By building lattices, Gals induce the aggregation of glyconjugates that would be otherwise scattered on the cell surface [31]. This effect may lead to the formation of rafts that, in turn, can either facilitate or hamper cell signalling. By means of this mechanism, cross-linking of the CD45 by Gal-3 can reduce T-cell activation [32]. On the other hand, by mediating a similar lattice formation mechanism, Gal-8 promoted the clustering of CD45 on T-cells, but leading to activation instead. In fact, CD45 PTPase activity was shown to be involved in Gal-8 T-cell proliferative and co-stimulatory effects, by lowering the TCR activation threshold [21].

In strong contrast with the requirement of a lattice formation to induce cell proliferation, the bivalent structure of Gal-8 was not required to exert its co-stimulation property. This, together with the low amount of protein required, suggests a stronger and selective, and therefore more specific, interaction between Gal-8 CRDs and cell-surface counter-receptors. Both N-CRDs and C-CRDs were able to co-stimulate mouse T-cells, thus indicating that different counter-receptors are involved. The activities found for Gal-8 single domains suggest that prototype Gals, such as Gal-1 or Gal-2, might eventually co-stimulate T-cells. It is to be noted that Gal-8 single-domain-encoding mRNAs have been isolated from human cancer cells [33]. The co-stimulatory property of Gal-8 was not prevented by lactose, but could be inhibited in the presence of Gal-Ara. Moreover, this sugar was able to fully inhibit the co-stimulatory activity of either the N-CRD or C-CRD. We have recently demonstrated that Gal-8 is expressed by human platelets, and that this lectin is exposed at the platelet surface after thrombin activation [22]. Gal-8 triggered several activation responses by interacting with the platelet-surface integrin GPIb–IX–V complex [22]. Interestingly, these effects were achieved by the single N-CRD, indicating that bivalency is not essential in this case either. It is worth noting the fact that the platelet surface is heavily sialylated [34] and that the N-CRD recognizes sialylated glycans with high affinity. Recently, a central role of platelets in the inflammatory processes was demonstrated [35]. Therefore an association between the presence of high amounts of Gal-8 in the inflamed synovia [36] and its activity on platelets arises [22], suggesting that Gal-8 exposure at the inflammatory foci is fueling the response.

The human population used for the proliferation induced by Gal-8 was very heterogeneous and thus further research is needed to understand the possible biochemical basis and physiological implications. However, responder individuals showed requisites, including the involvement of the C-CRD, similar to those of the mouse model. Findings reported in the present paper clearly indicated a proliferative response to this lectin in naïve mouse and human CD4 T-cells. Our previous observation that Gal-8 induces apoptosis in the CD4highCD8high thymocytes [20] is actually included in this phenomenon because this subpopulation is prone to enter apoptosis when prompted to proliferate as a way to avoid self-reactivity. A pro-apoptotic effect of Gal-8 on human normal or activated PBMCs and Jurkat T-cells has been described previously [26]. It should be noted that in those studies, Norambuena et al. [26] used a glutathione transferase–Gal-8 fusion protein. The multimerization induced by the glutathione transferase moiety might lead the fused Gal-8 to build an aberrant lattice, which alters the biological activity of the lectin. However, the same laboratory had also reported in earlier work [37] that Gal-8 was able to induce asymmetric lamellipodia formation in Jurkat cells, and considered it as a feature of cellular activation. In agreement with Lu et al. [27], we found that Jurkat cells were not induced to apoptosis by the addition of the lectin. However, we also found that both PHA- and CD3/CD28-pre-stimulated PBMCs became apoptotic after re-stimulation with Gal-8. These findings, together with those reported in the present paper and in our previous studies [20,21], suggest that Gal-8 can be postulated as a natural chimaeric protein with multifunctional properties on T-cells, associated with its C-CRDs or N-CRDs. In contrast with other Gals, Gal-8 induces proliferation and co-stimulation rather than apoptosis of naïve peripheral T-cells, thus playing a role as an inflammation activator. However, Gal-8 can also induce apoptosis of activated cells, a property that might help in the control of T-cell population expansion. In fact, given its high production by several tumours, it can be postulated that Gal-8 could be involved in the escape of the immune response [712]. By these dual activities, Gal-8 arises as a molecule able to shape the immune response by either stimulating otherwise borderline activating antigens or modulating ongoing responses, thus displaying interesting pharmacological properties.


Valentina Cattaneo and María Tribulatti performed the laboratory work. Valentina Cattaneo, María Tribulatti and Oscar Campetella designed the experimental studies and wrote the manuscript.


This work was supported by the Agencia Nacional para la Promoción Científica y Tecnológica (ANPCyT), Argentina [grant numbers PICT 2006-0168, PICT 2008-0369]. O.C. is a Researcher and V.C. and M.V.T. are fellows from the Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET), Argentina.


We thank the Consortium for Functional Glycomics Grant number GM62116 for oligosaccharides Te 223, Tr 34 and Te 100. The careful reading of the manuscript by Dr A. Cassola is also appreciated.

Abbreviations: CRD, carbohydrate-recognition domain; C-CRD, C-terminal CRD; C–C, two C-CRD chimaera; ERK, extracellular-signal-regulated kinase; FBS, fetal bovine serum; Gal, galectin; Gal-Ara, 3-O-β-D-galactopyranosyl-D-arabinose; N-CRD, N-terminal CRD; N–N, two N-CRD chimaera; OVA, ovalbumin; PBMC, peripheral blood mononuclear cell; PHA, phytohaemagglutinin; PI, proliferation index; PTPase, protein tyrosine phosphatase; RBC, red blood cell; TDG, D-galactopyranosyl-β-D-thiogalactopyranoside


View Abstract