A Galectin-3 Ligand Corrects the Impaired Function of Human CD4 and CD8 Tumor-Infiltrating Lymphocytes and Favors Tumor Rejection in Mice

Most cancer patients mount a spontaneous T-cell response against their tumor, and antitumor T cells can accumulate at metastatic sites (1–4). In several tumor types, the presence of T-cell infiltrates, which may contain tumor-specific lymphocytes, seems to correlate with better prognosis (5–7). In patients with advanced cancer, tumors progress despite the presence of tumor-infiltrating lymphocytes (TIL). This suggests that the antitumoral T cells become ineffective either because tumor cells have become resistant to the immune attack or because TILs have become functionally impaired (8, 9).

Despite this potentially adverse tumoral environment, human TIL can easily be expanded in vitro and the resulting T-cell lines and clones can lyse autologous tumor cells (10, 11). However, reports on the ex vivo function of human TIL are scarce. CD8⁺ T lymphocytes isolated from renal cell carcinomas displayed no cytotoxicity in a redirected killing assay, in contrast with blood CD8⁺ T cells of healthy donors (12). CD8⁺ T cells, which were isolated from a metastatic lymph node of a melanoma patient and recognized the modified Melan-A^MART-1 peptide, failed to produce IFN-γ on short-term restimulation with peptide, as opposed to anti-cytomegalovirus CD8⁺ T cells from the same tumors (13). TILs were also shown to differ from blood T cells for gene expression patterns and for several markers such as granzyme B and perforin, suggesting an impaired function (5, 14–17).

We previously described a physical dissociation of T-cell receptors (TCR) from coreceptors CD8 on human TIL and CTL clones with impaired function (18). A galectin ligand, N-acetyllactosamine (LacNAc), restored the ability of nonfunctional T cells to secrete cytokines, and this increased secretion was associated with an increased proximity of surface TCRs with the coreceptor CD8. We therefore attributed the TCR-CD8 dissociation observed on nonfunctional CTL clones to a reduced mobility of the TCRs, which are trapped in a lattice of glycoproteins clustered by extracellular galectin-3, a lectin abundantly secreted by macrophages and various types of tumor cells (19, 20). Our data suggested that intratumoral galectin-3 could impair T-cell function and that galectin-3 ligands could improve antitumor immunity in vivo.

Several galectin ligands, in particular galectin-3 ligands, have been identified in the past decade, such as polysaccharides extracted from plants or fungi (e.g., ref. 21; a complete list of references is available as Supplementary Data). Synthetic galectin-3 ligands were obtained by decorating lactose and galactose cores with various moieties or by creating branched structures such as glycodendrimers (reviewed in refs. 22, 23). A few of them were reported to reduce the number of tumor metastases in mice (24–29). Only GCS-100, a modified citrus pectin, and DAVANAT, a galactomannan extracted from guar beans, have previously been used in clinical trials (30–32).

We here compared the effect of GCS-100 with that of LacNAc on the cytotoxicity and cytokine secretion by CD8⁺ and CD4⁺ human TIL on ex vivo stimulation, and tested the therapeutic potential of GCS-100 in a murine tumor model.

Patient-derived solid tumor samples, tumor ascites, and blood cells were collected after approval by institutional review boards of all collaborating institutions. Cells from ascites were concentrated by centrifugation and either frozen at −80°C or resuspended in culture medium made of Iscove's modified Dulbecco's medium (Life Technologies) supplemented with 0.24 mmol/L of l-asparagine, 0.55 mmol/L of l-arginine, 1.5 mmol/L of l-glutamine (AAG), 100 μg/mL of streptomycin, 100 units/mL of penicillin, 30 μg/mL of gentamicin (culture medium; Sigma-Aldrich), and 2% human serum (HS) or ascitic fluid (ascites without cells). Ascitic fluid was frozen at −20°C. Solid tumors were mechanically dissociated and either passed through a 0.1-mm sieve before cryopreservation in 10% DMSO/3.6% HS or incubated for 2 to 3 hours in culture medium supplemented with 0.14 units/mL Liberase Blendzymes (Roche). Mononuclear cells were isolated from the blood of hemochromatosis patients, ascites, or solid tumor samples using Lymphoprep (Axis-Shield PoCAS) and usually frozen at −80°C or −196°C. Cells were thawed in culture medium supplemented with 2% to 10% HS, 6 IU/mL of interleukin (IL)-2, and 5 units/mL of DNase I (Sigma-Aldrich). T cells were isolated from mononuclear cells by rosetting with sheep erythrocytes (BioMérieux) previously treated with 2(2-aminoethyl) isothiourea dihydrobromide (Sigma-Aldrich). CD8⁺ and CD4⁺ T cells were isolated by a positive selection strategy (AutoMACS System, Miltenyi Biotec) using magnetic beads coated with anti-CD8. For samples LB2990, LB2974, and LB2912 in Fig. 1A, all the samples in Fig. 1B, sample LB2974 in Fig. 1C, and sample LB2991 in Fig. 2, a depletion strategy using the CD8 untouched isolation kit (Miltenyi Biotec) was used. The purity of the CD4⁺ or CD8⁺ T-cell samples was between 90% and 95%. Human recombinant IL-2 was from Chiron Healthcare SAS, IL-7 was from R&D Systems, LacNAc was from Carbosynth, and unmodified citrus pectin was from Sigma. GCS-100 was provided by Prospect Therapeutics, Inc. (now Solana Therapeutics). All the cell lines were obtained several years ago in our laboratory and described previously, but the cell lines were not authenticated according to guidelines described in http://cancerres.aacrjournals.org/site/misc/ifora.xhtml.

Polyclonal stimulations were performed in round-bottomed microwells with 10,000 T cells in 200 μL of culture medium, containing 0–6 IU/mL of IL-2, with 3,000 (ELISA and Bioplex) or 10,000 (ELISpot) beads coated with anti-CD3/CD28 antibodies (Dynabeads, Invitrogen). IFN-γ secreted after overnight coculture was measured by ELISA using Biosource Cytoset reagents (Invitrogen) or by ELISpot (ELISpot^PLUS kit ALP; Mabtech) using the AID-ELISpot classic reader software 5.0 (Autoimmun Diagnostika). In the cytotoxicity assay, the target cells were murine P1.aza^r cells derived from mastocytoma cell line P815 (33). Target cells were labeled for 1 hour with 50 μCi of Na[⁵¹Cr]O₄, washed, and incubated for 15 minutes at room temperature with anti-CD3 monoclonal antibody (OKT3, 1 μg/mL; Mabtech). CD8⁺ T cells were added to the targets, and chromium release was measured after 4 hours of incubation. For the degranulation assay, CD8⁺ T cells (75,000) were stimulated for 5 hours in Iscove's modified Dulbecco's medium 2% HS AAG with 150,000 mouse cells (P815) incubated with anti-CD3 antibody or 75,000 CD3/CD28 beads. Brefeldin-A (GolgiPlug, Becton Dickinson) and FITC-coupled anti-CD107a and anti-CD107b (1:100), or the control isotype (IgG₁, Becton Dickinson), were also added. After 5 hours of stimulation, cells were washed, labeled for 15 minutes at 4°C with peridinin chlorophyll protein (PerCP)-coupled anti-CD3 (1:40, SK1; Becton Dickinson) and allophycocyanin-coupled anti-CD8 (1:40, RPA-T8; Becton Dickinson) antibodies, washed, and fixed in PBS and 1% paraformaldehyde (PFA). Cells were analyzed on a FACSCalibur (Becton Dickinson). The percentage of CD107⁺ cells was estimated for the CD3⁺CD8⁺ T cells.

Cells from ascites were treated for 15 minutes at 4°C with Fc blocking reagent (Miltenyi Biotec) before treatment with anti–galectin-1 or anti–galectin-3 antibodies. Melanoma cell line MZ2-MEL.43 and cells from ascites were incubated for 2 hours at 37°C under agitation with LacNAc (5 mmol/L), GCS-100 (25 or 5 μg/mL), or B2C10 (mouse IgG₁; Becton Dickinson); fixed for 5 minutes at room temperature in PBS with 0.5 mg/mL dithio-bis-succinimidyl propionate (Thermo Scientific); washed; incubated for 20 minutes at room temperature with 25 mmol/L of glycine; washed again; and incubated with 5 μg/mL of either anti–galectin-3 antibody M3/38 (rat IgG_2a; BioLegend) or polyclonal anti–galectin-1 rabbit IgG (Abcam). Cells were further incubated at 4°C for 15 minutes, washed, and incubated for 15 minutes at 4°C with relevant anti-immunoglobulin secondary antibodies coupled to Alexa Fluor 488 (Invitrogen), together with anti–CD3-PerCP (1:40, SK7; Becton Dickinson) and anti–CD8-Alexa Fluor 647 (1:40, 3B5; Invitrogen). After a final washing step, cells were fixed with 2% PFA in PBS and analyzed on a FACSCalibur. We attributed a molecular weight of 120 kDa (average) to GCS-100, according to the molecular weight described for another modified pectin by Miller and colleagues (34).

A detailed protocol is available as Supplementary Data. The following antibodies were used: anti-TCRβ (1:40, IP-26, mouse IgG₁; eBioscience), anti-CD8α (1:40, UCHT-4, mouse IgG_2a; Sigma-Aldrich), anti-CD4 (1:40, RPA-T4, mouse IgG₁; eBioscience), anti–IgG_2a-Alexa Fluor 488 (1:400, donor, green; Invitrogen), and anti–IgG₁-Alexa Fluor 568 secondary antibodies (1:400, acceptor, red; Invitrogen).

P815 clone 3 cells were expanded in DMEM supplemented with AAG, 10% FCS, glucose, and 75 μmol/L of mercaptoethanol (33). The recombinant adenovirus construct contains the coding sequence of a fusion protein combining the first 81 amino acids of the mouse invariant chain with the first 83 amino acids of P1A, a protein containing an antigenic peptide, which is a tumor rejection antigen of P815 (35).

CD8⁺ TILs were isolated to >90% purity from 18 tumors: 5 solid melanomas and 1 melanoma ascites, 1 breast carcinoma pleural effusion, and 11 carcinoma ascites from different histologic origins. The lymphocytes were incubated with GCS-100 or LacNAc for 2 to 20 hours and then stimulated overnight by the addition of beads coated with anti-CD3 and anti-CD28 antibodies (CD3/CD28 beads). IFN-γ secretion by nontreated TIL isolated from four of five solid tumors was undetectable, whereas secretion by nontreated TIL isolated from ascites was more variable. The presence of GCS-100 boosted IFN-γ secretion by >3-fold in 13 of 18 cultures, and the average increase in IFN-γ secretion for all cultures was 4-fold (Fig. 1A). Treatment with LacNAc instead of GCS-100 had a similar effect (Fig. 1A). When the same GCS-100 or LacNAc treatment was applied to blood CD8⁺ T cells isolated from patients with hemochromatosis, the increase in IFN-γ secretion was marginal (Fig. 1B). We concluded that CD8⁺ TILs, compared with normal blood CD8⁺ T cells, are often impaired in their ability to secrete IFN-γ on stimulation. However, this defect can be reverted, at least in part, with the addition of GCS-100 or LacNAc.

In our previous report (18), CD4⁺ TILs were not analyzed. CD4⁺ TILs were therefore isolated from 13 of the 18 TIL samples shown in Fig. 1. They were also incubated with GCS-100 or LacNAc and then stimulated by the addition of CD3/CD28 beads. In the presence of GCS-100, IFN-γ secretion was increased by >3-fold in 6 of 13 samples (Fig. 1C). We conclude that the impaired secretion of IFN-γ can be corrected by GCS-100 and LacNAc not only for CD8⁺ TIL but also for CD4⁺ TIL.

To estimate the number of IFN-γ–secreting T cells, we performed ELISpot experiments with a few TIL and blood samples. Isolated CD8⁺ and CD4⁺ T cells were incubated overnight with GCS-100 or LacNAc and then stimulated by the addition of CD3/CD28 beads. After sugar treatment, the number of IFN-γ–secreting cells increased by a factor of 2 for the TIL, whereas the sugar treatment had a marginal effect on the blood CD8 T cells (Fig. 1D).

To increase TIL function in patients, GCS-100 has to be effective in a tumor environment, despite the presence of suppressive factors found in ascitic fluid and cells that could inhibit TIL, such as tumor cells or regulatory T cells. To mimic these situations, two types of experiments were conducted. First, nine CD8⁺ TIL populations were incubated with galectin ligands as above and stimulated in medium supplemented with 20% to 25% of cell-free carcinomatous ascitic fluid. IFN-γ secretion increased >3-fold in five of nine samples (Fig. 2A). Second, the effect of LacNAc and GCS-100 was tested on either CD2⁺ cells (including natural killer cells, CD8⁺, and CD4⁺ T cells) or on the total cell population, further containing macrophages and tumor cells. The cells were resuspended either in culture medium or in ascitic fluid. CD8⁺ TILs were purified after a first GCS-100 or LacNAc treatment and then stimulated with CD3/CD28 beads in the presence of LacNAc or GCS-100 (Fig. 2B). The results suggest that both LacNAc and GCS-100 can boost IFN-γ secretion by CD8⁺ TIL even in the presence of either carcinomatous ascitic fluid or other cells present in ascites (Fig. 2B).

In addition to IFN-γ secretion, several other TIL functions were also tested (i.e., secretion of other cytokines, cytotoxicity, and degranulation). In comparison with untreated cells, GCS-100– and LacNAc-treated cells also secreted more IL-2 and tumor necrosis factor-α, but no IL-4 (Supplementary Table S1). With some TIL samples, we observed slightly more secretion of IL-10 and IL-17 after treatment. Cytokine secretion was similar for treated and untreated blood CD8⁺ T cells.

We estimated the cytotoxicity of CD8⁺ TIL using mouse target cells coated with an anti-CD3 antibody (Fig. 3A). Only the CD8⁺ TIL treated with GCS-100 had a significant cytotoxicity, whereas the cytotoxicity of untreated CD8⁺ TIL was minimal. The cytotoxicity of CD8⁺ peripheral blood lymphocytes (PBL) was higher than that of TIL, but similar for treated and untreated cells. We also measured degranulation, a prerequisite for cytolysis. The percentages of degranulating cells were identified by surface expression of CD107a and CD107b, which reside in cytolytic granule membranes within the cytoplasm and are mobilized to the cell surface following activation-induced granule exocytosis (Fig. 3B). The percentage of CD107⁺ cells was much higher in treated TIL compared with untreated TIL, whereas this percentage was equivalent for treated and untreated blood CD8⁺ cells. Finally, we measured the intracellular content of perforin and granzyme B as phenotypic markers of cells that might have a potential to kill. Both treated and untreated TILs have a similar content of the two enzymes, indicating that the boosting effect of GCS-100 and LacNAc treatment was not due to an effect on perforin or granzyme content (Supplementary Fig. S1).

The toxicity of GCS-100 was evaluated with human melanoma cells and T-cell clones. Their proliferation was measured for at least 4 weeks in the presence of various concentrations of GCS-100. Growth inhibition was observed starting at day 7 at a concentration of 125 μg/mL (Supplementary Fig. S2). At 5 μg/mL, which was efficient in augmenting cytokine secretion by TIL, GCS-100 had no effect on proliferation.

Our working hypothesis for impaired IFN-γ secretion by CD8⁺ TIL on ex vivo stimulation is that galectin-3, which can be secreted by tumor cells and macrophages, binds to glycoproteins and oligomerizes on the T-cell surface, resulting in galectin–glycoprotein lattices that decrease TCR mobility (36). This hypothesis was also based on the dissociation of TCR from CD8 coreceptors observed on nonfunctional CTL clones and the increased TCR/CD8 colocalization after LacNAc treatment (18). However, because GCS-100 and LacNAc could compete with several galectins, we tested the effect of an anti–galectin-3 monoclonal antibody, B2C10, which binds to the NH₂-terminal domain of galectin-3 and presumably affects the oligomerization of the lectin: blocking this process should therefore decrease the T-cell inhibitory activity of galectin-3 (37). CD8⁺ TILs were incubated and stimulated in the presence of either antibody B2C10, a control antibody as negative control, or GCS-100 as a positive control (Fig. 4A). B2C10 boosted IFN-γ secretion to a comparable extent as GCS-100.

Other galectins (i.e., galectin-1 and galectin-9) were detected in human and/or murine tumors and thought to play a role in antitumor immunity (reviewed in ref. 19). We failed to detect galectin-8, galectin-9, or MGL (a lectin implicated in the regulation of T-cell function) on TIL (38). However, in addition to galectin-3, we detected galectin-1 (Fig. 5). We compared GCS-100 and LacNAc for their ability to detach galectin-3 and galectin-1 from melanoma cells that carry both galectins (Supplementary Fig. S3). The concentrations of GCS-100 that were able to detach 50% (EC₅₀) of the surface galectin-3 and galectin-1 were estimated at 1.3 μmol/L (159 μg/mL) and 2 μmol/L, respectively. EC₅₀s were estimated for LacNAc at 9 and 18 μmol/L, respectively. Different modified citrus pectins and synthetic ligands were previously shown to detach galectin-3 in a solid-phase assay with EC₅₀ as low as 1 mmol/L (22, 39–42). The most effective synthetic galectin-3 ligand seems to be a thiodigalactoside derivative that has a K_d of 33 nmol/L (39).

Incubation of TIL for 2 hours with 5 mmol/L of LacNAc, 25 μg/mL of GCS-100, or 5 μg/mL of GCS-100 resulted in a decrease of galectin-3 staining by 67%, 40%, and 24%, respectively (Fig. 5). Staining of galectin-1 did not decrease after GCS-100 treatment, whereas it decreased by 38% after LacNAc treatmen. Treatment with anti–galectin-3 antibody B2C10 detached galectin-3 but not galectin-1 (Fig. 4B). We failed to detach galectin-1 by treating TIL with three different anti–galectin-1 antibodies, and these treatments did not boost IFN-γ secretion either (data not shown). Because the anti–galectin-3 antibody seemed unable to detach galectin-1 while boosting TIL function, we conclude that detaching galectin-3 from TIL is sufficient to restore function while not excluding the contribution of other galectins.

We have previously used a flow cytometry–based fluorescence resonance energy transfer approach to show a loss of colocalization of TCR and CD8 at the surface of CD8⁺ TIL compared with blood CD8⁺ T cells (18). We examined here the effect of GCS-100 or LacNAc on this colocalization in both CD8⁺ and CD4⁺ TIL using a microscopy-based fluorescence resonance energy transfer approach known as “acceptor photobleaching.” TILs were treated for 2 hours with GCS-100 or LacNAc, attached to coverslips, and labeled with an anti-TCRβ antibody coupled to an acceptor fluorochrome and anti-CD8α or anti-CD4 antibodies coupled to a donor fluorochrome. Upon excitation at donor wavelength, energy could be transferred from the donor to the acceptor if closer than ∼10 nm. In these conditions, full acceptor bleaching abrogates the energy loss, thus increasing donor emission. For each analyzed cell, three regions of the membrane were bleached and donor emission increases were measured. Data are shown in Table 1 for two CD8⁺ and two CD4⁺ TIL samples and for two cells in each condition.

When CD8⁺ TILs were not treated with LacNAc or GCS-100, no increase in donor emission was detected upon acceptor photobleaching, indicating poor TCR-CD8 colocalization (Table 1). For TIL VUB147, increases in donor emission were 20% and 14% after incubation for 2 hours with GCS-100 and LacNAc, respectively, thus indicating increased TCR-CD8 colocalization after treatment. Lower but significant increases were observed with TIL LB3022. Although modest at first sight, these increases are well within the reported range in the literature and similar to our previously reported value (26%) with a functional CTL clone (18, 43, 44).

For CD4⁺ TILs not treated with LacNAc or GCS-100, there was also no increase in donor emission detected upon acceptor photobleaching (Table 1). After 2 hours of treatment of TIL VUB129, increases were 11% with GCS-100 and 11% with LacNAc. Comparable values were obtained with TIL LB3022. For both CD4⁺ and CD8⁺ TILs treated with GCS-100 or LacNAc, increased TCR/CD8 or TCR/CD4 colocalization was paralleled by increased secretion of IFN-γ upon ex vivo restimulation (Table 1).

We tested the therapeutic effect of GCS-100 in tumor-bearing mice. Forty mice were injected s.c. in the flank with 2 × 10⁶ P815 mastocytoma cells. On day 4, half of the animals were vaccinated with an adenovirus encoding P815 tumor antigen P1A (Fig. 6A; ref. 35). On day 10, at which time all mice, vaccinated or not, bore a tumor of ∼70 mm², treatments with either PBS or GCS-100 (200 μg s.c. close to the tumor and 500 μg i.p.) were initiated. Mice were injected three times a week. Three weeks later, the tumor had become undetectable in 6 of the 10 vaccinated mice treated with GCS-100, of which 5 were still alive after another 3 months. Vaccine alone had no effect on tumor growth, but slightly increased survival, with a median life span 10 days longer than in nonvaccinated mice. In nonvaccinated mice, GCS-100 had no effect by itself, either on tumor size or on animal survival (Fig. 6B and C).

Our previous observations that human CD8 TIL increased their secretion of IFN-γ after ex vivo treatment with disaccharide LacNAc prompted us to test GCS-100, a polymeric galectin ligand approved for clinical trials. The results reported here indicate that GCS-100 is as efficient as LacNAc in boosting TIL to be cytotoxic and to secrete different cytokines, not only CD8⁺ but also CD4⁺ TIL. This effect was obtained at concentrations (5 μg/mL) much lower than those used to induce apoptosis of tumor cells (>100 μg/mL), the activity that prompted its administration to cancer patients (30).

Why do galectin-3 ligands improve human TIL function? Our working hypothesis is that TILs have been stimulated by antigen recently and that activation could modify the expression of enzymes of the N-glycosylation pathway and change the structure of N-glycans exposed at the cell surface, as shown for murine T cells (45). We surmise that the recently activated TILs, compared with resting T cells, harbor a set of glycans that are either more numerous or better ligands for galectin-3. Galectin-3 is abundant in many solid tumors and carcinomatous ascites and can form lattices with surface glycoproteins that would thereby reduce TCR mobility. This could explain the absence of colocalization of TCR and CD4/CD8 coreceptors, the impaired function of TIL, and the boosting effect of galectin ligands.

Several of our new results support this hypothesis. First, treatment with an anti–galectin-3 antibody that was able to detach galectin-3 from TIL is as efficient as GCS-100 in increasing IFN-γ secretion by TIL. Second, both GCS-100 and LacNAc can detach galectin-3 from the surface of cells. Third, treatment of TIL with GCS-100 or LacNAc increased both IFN-γ secretion and colocalization of TCR with coreceptors CD8 or CD4. In this respect, the increased TCR-CD4 colocalization after GCS-100 or LacNAc treatment suggests a common mechanism behind impaired function of both CD4⁺ and CD8⁺ TIL, at least for those cells responding to galectin ligands.

This hypothesis agrees well with data from two other groups. Demetriou and colleagues observed that T cells from mice deficient in N-acetylglucosaminyl-transferase V, an enzyme of the N-glycosylation pathway, have a lower activation threshold than T cells from wild-type mice. The authors were the first to propose that reduced N-glycosylation, resulting in looser glycoprotein/galectin lattices, could improve TCR clustering at the immune synapse (46). In addition, Kuball and colleagues (47) observed an increased frequency of cells able to secrete cytokines upon stimulation, if they were expressing a transfected TCR lacking a glycosylation site. They proposed therefore that N-glycosylation modifies TCR avidity.

Other galectins (i.e., galectin-1 and galectin-9) were detected in human and/or murine tumors and thought to play a role in antitumor immunity (reviewed in ref. 19). In humans, exposure to galectin-1 induced a shift from Th1 to Th2 cytokine response, promoted the expansion of CD4⁺CD25^high FOXP3⁺ T cells, and increased the tolerogenic potential of dendritic cells (41, 48–50). Galectin-1 was reported to induce T lymphocyte apoptosis (48). In mice, tumor growth and immune dysfunction correlated with the presence of galectin-1 (51). On the contrary, injections of galectin-9 were reported to improve the survival of tumor-bearing mice (52). Here, we show that, in addition to galectin-3, we also detected galectin-1 on TIL. At the concentration used for treating TIL, LacNAc but not GCS-100 was able to detach galectin-1 from treated TIL. Because the anti–galectin-3 antibody did not detach galectin-1 from TIL, we conclude that the mere detachment of galectin-3 could result in a boosted secretion of IFN-γ by stimulated TIL.

Impaired TIL function most probably results from a combination of mechanisms, all ultimately depending on tumor cells. It is therefore remarkable that a few hours of treatment with GCS-100 or LacNAc was sufficient to strongly increase IFN-γ secretion by stimulated TIL. However, the respective contribution of other galectins, transforming growth factor-β, IL-10, PD-1/PD-L1, or regulatory T cells, remains important to clarify in order to define whether different lymphocyte populations are impaired through different mechanisms and whether several inhibitory pathways add up in the same cells.

The identification of antigens recognized by T lymphocytes on human tumor cells has resulted in numerous clinical trials involving the vaccination of tumor-bearing patients with defined tumor antigens. Among vaccinated metastatic melanoma patients, ∼5% show a complete or partial clinical response, whereas an additional 10% show some evidence of tumor regression without clear clinical benefit. Analyses of the T-cell responses of melanoma patients suggest the following scenario (1, 53). Spontaneous tumor-specific T-cell responses are frequent. Although they could control early tumors, they eventually fail to do so and become impaired in their function, most probably due to immunosuppressive factors present in the tumor. Galectin-3 could be one of them. Thus, the tumors of the patients about to receive the vaccine already contain functionally impaired T cells directed against tumor antigens. It is therefore possible that, in the few patients who show tumor regression following vaccination, some T cells generated by the vaccine reach the tumor and succeed in reversing the local immunosuppression.

The relevance of combining vaccination and galectin ligands is supported by the results obtained with GCS-100 in vaccinated mice bearing P815 tumors. Previous results indicated that prophylactic vaccination with an adenovirus encoding P815 tumor antigen P1A induced an anti-P1A T-cell response that conferred protection against tumor challenge in 57% of mice (54). However, therapeutic vaccination was ineffective at inducing tumor rejection in tumor-bearing mice.⁹ Injection in mice bearing a subcutaneous P815 tumor of a Semliki Forest virus encoding P1A was therapeutic for 4 of 10 animals, provided that tumors were small (≤25 mm²) and that intratumoral injections were used (55). In the present report, injection of another P1A adenoviral construct conferred only a mild survival advantage. However, in combination with GCS-100 injections, half of the mice rejected their tumor and appeared free of disease on day 130. The vaccination of tumor-bearing mice with P1A was essential for tumor rejection, and we are tempted to believe that GCS-100 positively affected the anti-P1A T cells.

These experiments, however, do not provide any evidence that GCS-100 exerts its therapeutic effects in mice by acting on murine T cells or by disrupting galectin-glycan lattices. Galectins were shown to play pleiotropic roles in tumor metabolism. Myeloma cell growth was inhibited in vitro in the presence of 500 μg/mL of GCS-100, which also triggered apoptosis and decreased cell migration in response to vascular endothelial growth factor (30, 56). Modified citrus pectin inhibited the homoagglutination of melanoma cells, adhesion of tumor cells to extracellular matrix and endothelial cells, as well as the formation of capillary tubes in the presence of galectin-3 (reviewed in ref. 57). Although treatment of nonvaccinated, tumor-bearing mice with GCS-100 did not provide a survival benefit compared with control mice, GCS-100 could have increased the apoptosis of tumor cells and inhibited neoangiogenesis and metastasis formation and therefore participated in the therapeutic effect independently of an effect on the immune system.

It might seem risky to inject a galectin-3 ligand systemically, considering the pleiotropic role of galectin-3 (58, 59). However, no life-threatening side effect was observed in a clinical trial with GCS-100. GCS-100 has been administered to cancer patients based on its potential to induce apoptosis of myeloma cells in vitro (30). Twenty-four patients with chronic lymphocytic leukemia, who had relapsed after one or two therapies, received 160 mg/m² of GCS-100 i.v. for 5 days every 21 days. Six of them experienced a partial clinical response. The treatment was generally well tolerated with minimal hematologic toxicity. Skin rash was observed on two patients that resolved with cessation of treatment or steroids.

Taken together, our observations suggest that treatment of cancer patients with galectin ligands could correct an impaired function of TIL. As observed in mice, it is possible that vaccination combined with local and/or systemic injection of a galectin ligand would be more effective at producing tumor regressions than vaccination alone.

No potential conflicts of interest were disclosed.

We thank Prospect Therapeutics (now Solana Therapeutics), and in particular F. Tao, for providing us with GCS-100; D. Latinne and A. Cornet for ELISpot experiments; D. Godelaine, P. Coulie, T. Boon, and B. Van den Eynde for critical reading; C. Uyttenhove for helpful suggestions; G. Warnier for providing the recombinant adenovirus; E. Marbaix for help in statistical analysis; N. Krack for editorial assistance; and L. Vanbiervliet, S. Ottaviani, C. Wildmann, V. Ha Thi, and T. Lac for their assistance.

Grant Support: Fondation contre le Cancer (Belgium), Fonds de la Recherche Scientifique Médicale (Belgium), European Community Sixth Framework Programme, LIFESCIHEALTH-5-IP (contract number 518234), and Delori family. P. van der Bruggen is a “fellow du fonds Allard-Janssen pour la recherche sur le cancer.” G. Wieërs is supported by a grant from the Fonds National de la Recherche Scientifique (Belgium). The cell imaging platform was funded by FNRS, Région Bruxelloise, Région Wallonne, Loterie Nationale, Interuniversity Attraction Poles, and Université Catholique de Louvain (Belgium).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.