Galectin-1, an Endogenous Lectin Produced by Thymic Epithelial Cells, Induces Apoptosis of Human Thymocytes

Galectin-1 was prepared by the method of Couraud et al. (23) in Escherichia coli strain BL21 (DE3) transformed with the expression vector pT7IML-1 (gift of Incyte Pharmaceuticals, Inc., Palo Alto, CA).

Human thymocytes were isolated from surgical specimens and passaged over nylon wool as previously described (24). Nylon wool passaged thymocytes were cultured overnight at 37°C in serum-free IMDM with 1% dialyzed BSA, and 85 μg/ml human transferrin (AT-IMDM). The following day, 10⁶ cells in 0.2 ml of AT-IMDM, 1.1 mM dithiothreitol (DTT) were incubated for 5 h at 37°C with 20 μM recombinant, human galectin-1. DTT was added to maintain the carbohydrate binding activity of galectin-1. Samples were adjusted to 0.1 M β-lactose to dissociate aggregated thymocytes, washed in 10 mM phosphate buffer, pH 7.4, 140 mM NaCl (PBS), then either surface stained or fixed with 1% paraformaldehyde for terminal deoxynucleotidyl transferase–mediated dUTP biotin nick end-labeling (TUNEL) to detect apoptotic cells. Samples were surface stained as previously described (11). PE-conjugated CD3 antibody was used in addition to fluorescein-conjugated antibody to discriminate more easily between cells lacking CD3 expression (CD3⁻) and cells expressing low levels of CD3 (CD3^lo) (10). The following mouse anti–human antibodies, directly conjugated to fluorochromes were used: CD3–FITC, CD4–PE, and CD8–peridinin chlorophyll protein (PerCp; Becton Dickinson, San Jose, CA), CD3–PE, CD8–FITC, and CD69–PE (Caltag, South San Francisco, CA). Isotype-matched controls were included for all reagents. After surface staining and washing, cells were either immediately subjected to flow cytometry, or fixed in 1% paraformaldehyde and stored at 4°C for flow cytometry, which was performed within 24 h. All flow cytometry was performed using a Becton Dickinson FACScan^® with analysis using Lysis II^\xa9 or CELLQuest^\xa9 software.

Thymocytes (2.5 × 10⁶ cells/ml) were incubated with 2 μg/ml of soluble antibody to CD3ε (UCHT-1, Dako Corp., Carpinteria, CA), with 20 μM dexamethasone (Calbiochem, La Jolla, CA), or with medium alone at 37°C. After 16 h, galectin-1 or buffer was added. The samples were then incubated, stained, and fixed as described above.

DNA content and apoptosis were assessed by TUNEL using a modification of the method of Gorczyca (25). Paraformaldehyde-fixed cells were washed in PBS, resuspended in cold (−20°C) 80% ethanol and kept on ice for 30 min. Cells were washed in PBS, then incubated at 37°C for 1 h with terminal deoxytransferase (TdT, 120 U/ml; Promega Corporation, Madison, WI), 1× TdT reaction buffer (provided with enzyme), and biotin–16-2′-deoxyuridine5′-triphosphate (biotin–dUTP, 10 nmol/ml; Boehringer Mannheim, Indianapolis, IN) in a total volume of 50 μl. After washing, incorporated biotin–dUTP was stained with streptavidin–FITC (10 μg/ml, Boehringer Mannheim) in PBS with 1% BSA, 0.05% Triton X-100, and 0.01% NaN₃ for 1 h at room temperature. Samples were washed in the same buffer, then in PBS, 1% BSA, and stained for DNA content with 1 μg/ml 7-aminoactinomycin D (7-AAD; Molecular Probes, Eugene, OR), or propidium iodide (PI, Calbiochem, Inc., San Diego, CA) with 0.01% RNAase A, 30 min before flow cytometry was performed. Control samples were treated as above, except the TdT enzyme was omitted.

We incubated nylon wool passaged thymocytes with and without galectin-1. Galectin-1 induced apoptosis of a significant fraction of thymocytes (Fig. 1). In galectin-1–treated samples, 37 ± 5.9% (SEM) of the cells were apoptotic with a range of 21–41%. In control samples, treated with ATIMDM, 1.1 mM DTT, an average of 22.2 ± 5.9% of the cells (range, 11–33.5%) were apoptotic. Thus, galectin-1 treatment consistently induced apoptosis of ∼15% of the total thymocyte population.

To determine the stage of differentiation at which thymocytes were most susceptible to galectin-1, we examined the viable cells in galectin-1–treated and control samples for expression of a number of cell surface markers. We observed that galectin-1 treatment eliminated a fraction of the CD3⁻ population, as indicated by comparing absolute numbers of cells within the region defined by M1 in Fig. 2,A. We also consistently saw a decrease in the population expressing intermediate levels of CD3 (CD3^int, Fig. 2,A; M3). The same samples were stained by three-color staining using CD3–FITC, CD4–PE, and CD8–PerCP. The results shown in Fig. 2,B indicate that galectin-1 selectively eliminated a subset of the CD4⁺ CD8⁺ double-positive (DP) cells that were dim for both markers. CD3 staining in these samples (data not shown) was consistent with that shown in Fig. 2, A and C. By comparing absolute numbers of cells in control and galectin-1–treated samples, we determined that none of the CD4⁻CD8⁻ double-negative (DN) cells were lost (data not shown). This indicates that the CD3⁻ cells that were susceptible to galectin-1 were expressing CD4 and CD8, and were not in the triple negative fraction. Galectin-1 also eliminated thymocytes from the CD5^lo (data not shown) and CD69⁻ (Fig. 2 C) populations. In summary, the thymocyte populations most susceptible to galectin-1–induced apoptosis were CD3^int or CD3⁻ and CD4^loCD8^loCD5^loCD69⁻.

The apoptotic effects of CD3 stimulation and steroids antagonize each other when the treatments are given simultaneously (26, 27). To determine whether galectin-1 induces apoptosis via a pathway that interacts or overlaps with that involved in apoptosis induced by steroids or CD3 antibody, we pretreated thymocytes for 16 h with dexamethasone, or with CD3 antibody before exposure to galectin-1 (Fig. 3). As expected, the level of apoptosis in dexamethasone pretreated samples was higher than in the nontreated controls (19% and 11%, respectively). However, after correcting for background apoptosis, dexamethasone pretreatment did not alter the fraction of the total population that was susceptible to galectin-1 (10% as compared with 10.9% in controls). These data indicate that galectin-1–induced apoptosis of thymocytes was independent of steroid pretreatment.

In contrast, pretreatment of the thymocytes with soluble antibody to CD3 did not alter the background level of apoptosis, but did result in a fourfold increase in the fraction of cells susceptible to galectin-1–induced apoptosis (40% above control levels as compared with 10% without CD3 antibody pretreatment). These results demonstrate that engagement of the TCR with CD3 antibodies before galectin-1 exposure increased the sensitivity of thymocytes to galectin-1.

Previous results from our laboratory showed that galectin-1 induces apoptosis of peripheral T cells that are activated and proliferating, whereas resting peripheral T cells are not susceptible to galectin-1–induced apoptosis (12). Others have reported cell cycle–dependent differences in the susceptibility of cells to apoptosis in a number of systems (25, 28, 29, 30). To determine at which stage(s) of the cell cycle thymocytes were undergoing apoptosis, we examined the DNA content of galectin-1–treated and –untreated thymocytes. By selective gating of TUNEL labeled thymocytes, we found that in untreated samples, the majority of apoptotic thymocytes (74%) had a DNA content greater than 2N, characteristic of S, G₂, or M phase cells (data not shown). When a greater fraction of the total population was induced to undergo apoptosis by galectin-1, a comparable percentage (71%, Fig. 4) of the TUNEL-positive (apoptotic) cells had a DNA content greater than 2N. In contrast, 29% of the total sample and 16% of the TUNEL-negative cells had a DNA content greater than 2N (Fig. 4). DNA content was not altered by the TUNEL labeling procedure (data not shown). This observation strongly suggests that thymocytes that are progressing through the cell cycle are most sensitive to apoptosis and most susceptible to galectin-1.

We have investigated the effects of galectin-1 on thymocyte survival and have examined the surface phenotype of thymocyte populations before and after galectin-1 treatment. Our results indicate that, after correction for background apoptosis, exogenous galectin-1 induced apoptosis of 15% of human thymocytes (Fig. 1). This is a significant fraction when one considers that the thymocytes were exposed to galectin-1 in vivo less than 48 h before use in this study. Additionally, we found that the susceptible thymocytes were all of an immature phenotype.

Susceptibility to galectin-1 correlated with a number of events that occur during thymocyte differentiation. The phenotype of galectin-1–sensitive cells and how galectin-1 susceptibility relates to thymocyte differentiation is shown schematically in Fig. 5. Examination of the surface phenotype of thymocytes before and after galectin-1 treatment suggests galectin-1 may be involved in apoptosis of both nonselected and negatively selected thymocytes.

There were primarily two galectin-1–susceptible populations as determined by flow cytometry. One of the two populations was CD3⁻CD4^loCD8^loCD69⁻ (Fig. 2). The expression of CD4 and CD8 normally occurs during thymocyte ontogeny subsequent to the expression of the TCR β chain (31, 32). Therefore, those DP-susceptible thymocytes that were detected as CD3⁻ by flow cytometry had already attempted TCR rearrangement. The population that carries this surface phenotype has been shown to express the intracellular regulator of apoptosis, Bcl-x_L, but not Bcl-2 (33, 34). Within this group may be cells that express the pre-T α chain but have unsuccessfully rearranged a TCR β chain (35). Because this population did not express detectable levels of surface CD3, these cells could not undergo negative selection. It is reasonable to propose that these cells represent the nonselected population, i.e., those thymocytes that fail to express a functional TCR and die because they are not positively selected. Our observation that DP, CD3⁻ thymocytes are sensitive to galectin-1 suggests a role for galectin-1 in apoptosis due to nonselection.

The second susceptible population was CD3^intCD4^lo CD8^loCD69⁻. This same phenotype precedes apoptosis after negative selection (5). This population may also include thymocytes that are nonselected because they express a nonfunctional TCR. Although the viable CD69⁺ population was not reduced by galectin-1 treatment, 58% of the apoptotic cells were found to express low levels of CD69 (CD69^lo, data not shown). This fraction of CD69^lo apoptotic thymocytes is similar to that observed by Kersh and Hedrick (5) for murine thymocytes undergoing negative selection. Our data suggest that a fraction of the CD69⁻ thymocytes became CD69^lo upon treatment with galectin-1. This issue warrants further investigation.

Treatment of thymocytes in vitro with antibody specific for the CD3ε chain mimics antigen stimulation and is used as a model for negative selection (36). CD3 antibody treatment of thymocytes increases the CD3^intCD4^loCD8^lo population, as does antigen engagement during negative selection (5, 6). Using in vitro models, specific peptide and MHC binding or CD3 antibody stimulation alone was not sufficient to induce apoptosis, and a second, unknown, signal was required as well (6, 7). This second signal can be delivered by some antigen-presenting cell lines, thymic epithelial cells, or dendritic cells (6, 7). Previous attempts to identify a molecule that mediates this second signal have yielded conflicting results (6, 36, 37). Galectin-1 is potentially a candidate molecule as a provider of a second (apoptotic) signal needed for negative selection. Our observations that CD3^intCD4^loCD8^lo cells were eliminated by galectin-1 treatment and that galectin-1 susceptibility was enhanced by CD3 engagement (Fig. 3) support this hypothesis, and suggest a role for galectin-1 in negative selection.

If galectin-1 induces apoptosis after negative selection, then thymic epithelial (TE) cells may participate in negative selection to a greater extent than has previously been proposed. It is known that TE cells synthesize galectin-1 (11), can induce apoptosis of T cells (7, 38) and can mediate negative selection (39, 40). In our laboratory, we observed that 53% of MOLT-4 cells that bound to TE cells on glass cover slips were apoptotic after 4 h at 37°C, whereas only 7% of CEM cells bound to TE cells were apoptotic (data not shown). This is intriguing in light of our previous observation that MOLT-4 cells are sensitive to galectin-1–induced apoptosis, whereas CEM cells are not (12). Although TE cells can mediate negative selection, dendritic cells have been reported to do this more efficiently (2). If galectin-1 is required for negative selection, one would predict that dendritic cells may also synthesize galectin-1, or else acquire it from the extracellular milieu. This remains to be investigated.

Three distinct pathways leading to apoptosis of thymocytes have been identified (41). The first requires the expression of the tumor suppressor protein p53 and can be induced by DNA damage. The other two pathways, one induced by steroid treatment, the other by TCR engagement, are p53 independent (41–43). It has been reported that DP thymocytes do not express p53 (31). Our data showing that only DP thymocytes are susceptible to galectin-1 suggest that galectin-1 also induces apoptosis by a p53 independent pathway.

Galectin-1 may be a component of TCR-mediated apoptosis. Ligation of CD3 primed thymocytes for galectin1–induced apoptosis and resulted in an increased fraction of cells that were susceptible to galectin-1 (Fig. 3). This data is similar to that found using murine thymocytes treated with CD3 antibody and galectin-1 (Vespa, G.N.R., and M.C. Miceli, manuscript in preparation). This synergism between CD3 antibody and galectin-1 suggests that galectin-1 may be a normal component in the process by which TCR engagement induces apoptosis. Engagement of the TCR, as occurs during negative selection, may induce intracellular expression of apoptotic machinery, which is then activated by galectin-1. This hypothesis is supported by the kinetics of CD3-mediated apoptosis, which is slow (10–18 h) when compared with galectin-1–induced apoptosis, which occurs rapidly (1–4 h).

Galectin-1–induced apoptosis was independent of that induced by steroids. In contrast with the effects of CD3 antibody pretreatment, we observed no effect of dexamethasone pretreatment on galectin-1 susceptibility (Fig. 3). Our observation that dexamethasone neither antagonizes, nor augments galectin-1–mediated apoptosis indicates that the mechanisms leading to steroid and galectin-1–mediated apoptosis can operate independently.

Galectin-1 susceptibility may be regulated by entry of a vulnerable cell into the cell cycle. We observed that galectin-1 preferentially kills proliferating cells. This observation is consistent with a previous proposal that the susceptibility of thymocytes to apoptosis is controlled by entry or exit from the cell cycle (44). We found that only two subsets of the DP thymocytes were susceptible to galectin-1 (Fig. 2). Whereas the majority of DP thymocytes are not dividing, 10–15% of the total thymocyte population is composed of DP cells that are undergoing cell division (45, 46). This number corresponds to the average fraction of thymocytes that we found were susceptible to apoptosis by galectin-1 (15%). Thus, the previously reported size of the population of proliferating DP thymocytes is sufficient to account for the fraction of galectin-1–sensitive thymocytes that we observed.

Entry into the cell cycle is unlikely to be the only factor that determines galectin-1 susceptibility. Thymocytes, which are in direct contact with TE cells during maturation, may be in continuous contact with galectin-1 in vivo (11). However, proliferating thymocytes do not all undergo apoptosis simultaneously in vivo or in vitro. A number of additional factors potentially influence the response of thymocytes to galectin-1. These factors include (a) the expression of specific cell surface glycoprotein counter receptors recognized by galectin-1; (b) the glycosylation of those specific counter receptors; (c) the expression of Bcl-2 family members; and (d) the presence of intracellular apoptotic machinery. The latter may be influenced by TCR engagement, which occurs during negative and positive selection. All of these factors would affect galectin-1 binding and signal transduction, and most likely combine to determine the susceptibility of specific thymocyte populations to galectin-1. It is important to note that lactosamine, the carbohydrate ligand for galectin-1, can be found in varying amounts on more than one cell surface glycoprotein counter receptor. In addition, the dimeric nature of galectin-1 would permit intermolecular and intramolecular cross-linking of cell surface glycoproteins. Thus, a response to galectin-1 may require the expression of more than a single glycoprotein species on the cell surface.

Based on the data presented above and the constitutive expression of galectin-1 in normal thymus, we hypothesize that galectin-1 is a participant in the elimination of nonselected and possibly negatively selected cells during normal thymocyte maturation. A galectin-1 knock out mouse has been generated (47). This model may be useful for further investigation of the role of galectin-1 in thymocyte development.

We wish to thank Drs. H. Spits, M.C. Miceli, S.M. Hedrick, and L. Goodglick for critical review of the manuscript. We also wish to thank Drs. M.C. Miceli and G.N.R. Vespa for very helpful discussions. We are grateful to the staff of the Jonsson Cancer Center Flow Cytometry core laboratory for technical assistance.

This work was supported by training grant no. CA09056 from USPHS to N.L. Perillo, by National Institutes of Health grant no. AG104015 through the Claude D. Pepper Older American Independence Center, and grants from the Concern Foundation and the University of California Cancer Research Coordinating Committee to L.G. Baum, and by National Institutes of Health grant no. HD29341 to C.H. Uittenbogaart.