DNA methylation is involved in cellular development, differentiation and transformation 1. In different types of tumours, aberrant methylation of CpG islands in the promoter region has been observed for many differentiation- and cancer-related genes resulting in the silencing of their expression 2. Therefore, over the past decade, there has been increasing interest in the use of demethylating agents to induce the differentiation or the apoptosis of cancer cells 34. Treatment of the cells with the pyrimidine analogue 5-AzaCytidine (AzaC), which inhibits methylation of cytosine residues during replication in the newly synthesised DNA, has been demonstrated to reactivate the expression of many silenced genes, as well as the expression of the silenced retro viral genomes 5. Silencing of one of the most important cell cycle regulatory proteins p16^INK4a by methylation of the CpG islands in the promoter region has been found to be a common event in tumours 67. Protein p16 suppresses S-phase entry by antagonising the cyclin-dependent kinases CDK4 and CDK6 8.

Deciphering the molecular mechanisms underlying the phenotypic effects of the treatment with demethylating drugs is a crucial step in understanding what genes may be interesting targets for chemotherapy. The available data on the mechanism of action of these drugs strengthen the idea that it is different from that of agents that act primarily via their cytotoxic effects, such as Arc-C 9. Several lines of evidence suggest that galectin-1 (Gal-1), a 14 kDa galactoside-binding protein distributed widely in immune cells, could be involved in these mechanisms. Several members of the galectin family have been found to modulate cell differentiation and cell survival 101112131415. Early studies demonstrated that the expression of Gal1 could be induced in cultured hepatoma-derived cells by treatment with AzaC 16. Chiariotti and co-workers showed that reactivation of the silent Gal1 alleles is accompanied by a transition from a fully methylated to a fully unmethylated state of several CpG dinucleotides in the promoter region 17. In addition, nonexpressing tissues exhibited highly heterogeneous methylation profiles 18. Gal is considered to be a typical cytosolic protein, lacking a signal peptide for membrane translocation 19. However, most of the functions assigned to galectins are confined to the cell surface or extracellular milieu 102021, consistent with evidences for extracellular roles of Gal1 in regulation of cellular differentiation and proliferation. It is clear that Gal1 can be specifically secreted and targeted by an infrequent mechanism 222324. The constitutive expression as well as the secretion of Gal1 dramatically depend on cell types 25 and are responsive to developmental events 202223.

An example is found during erythroid differentiation of the K562 human leukaemia cell line. During differentiation induced by erythropoietin and deprivation of granulocyte-macrophage colony-stimulating factor, the cells empty their cytoplasmic content of endogenous Gal1 into the external medium where it is bind to cell surface receptors 24. The synthesis and secretion of Gal1 by leukocytes are of interest because lactosaminoglycans present at the leukocyte cell surface may be physiologically significant galectin receptors that could mediate autocrine or paracrine functions. Several lines of evidence indicate that Gal1 may function as an autocrine negative growth regulator or as a pro-apoptotic factor 262728. We have recently demonstrated that Gal1 binding to Burkitt lymphoma cells results in an intracellular signal, with inhibition of the tyrosine phosphatase activity of CD45 and therefore phosphorylation of Lyn kinase 2930.

In this work, we study the effect of AzaC treatment on the lymphoma cell line BL36. As p16^INK4a gene has been found to be downregulated by hypermethylation at high frequency in different types of tumours 67, it is used as a control of AzaC effect. Then, the phenotypic effects of AzaC are compared to those obtained by the addition of exogenous Gal1. The findings that we report here lend further support to a potential role for Gal1 in the AzaC-induced pathway of differentiation in hematopoietic cells.

The BL36 B lymphoma cell line 31, a gift from Pr. Lenoir (CIRC, Lyon, France) to Pr. M. Raphael, was maintained in a complete medium of RPMI 1640, containing 10% heat-inactivated foetal calf serum, 2 mM L-Glutamine, 1 mM sodium pyruvate (complete medium). Treatment with AzaC was conducted as follows: 24 hours after seeding 5, 10 or 50 μM of AzaC were added to 5 ml of complete medium 1632. The cells were feeded with the appropriate medium during the time of the experiments.

Several independent methods were used to assess the proliferative versus death states of the cells. For direct determination of cell number, at the indicated time point the cells were harvested and counted using a Coulter Counter (Beckman Coulter France, Villepinte, France). Determination of the viable cell numbers used propidium iodide (PI) (Sigma, Saint-Quentin Fallavier, France). To determine the percent cell death including both apoptotic and necrotic cells, AzaC-treated cells were analyzed for AnnexinV binding and PI uptake as described 33 using the ApoDetect AnnexinV-FITC kit (Zymed). Briefly, after washing of cultured cells (3.10⁵) with PBS and resuspension in binding buffer, the cells were stained with 10 μl FITC-labelled AnnexinV and 10 μl PI (20 μg/ml). After 5 min of incubation at room temperature in the dark, again 200 μl of binding buffer was added and cells were analyzed by flow cytometry. The percent cell death was calculated by determining the percent of viable cells:

The MTS assay (CellTiter 96™, Promega, Charbonnières, France) 34 was used to compare the percentage of metabolically active cells in treated cells vs. untreated cells, as previously described 14. To study the cell cycle, 2.10⁶ BL36 cells were pelleted at 1000 g for 5 minutes and the pellet was incubated in a mixture of ethanol-PBS (70/30 v/v). The cells were pelleted another time and suspended in 800 μl PBS. One hundred microliters of 1 mg/ml RNase was added to the cell suspension. One hundred microliters of 400 μg/ml PI was added to the solution to stain the nuclear DNA. The DNA content of the cells was determined by a flow cytometer, FACScan (Becton Dickinson, Mountain View, CA), and the percentages of cells in G0+G1, S, and G2+M phases of the cell cycle were analyzed by a polynomial model (SFIT, Becton Dickinson).

To determine the cell phenotypes, cells were suspended in PBS supplemented with 2% BSA, and incubated in suspension for 30 min with the fluorescently tagged primary antibody or negative control, or incubated with untagged antibody, washed, and incubated again for 30 min with FITC-conjugated second antibody. All incubations were performed at 4°C. Flow cytometry was performed using a Coulter Epics Elite ESP instrument.

Human recombinant Gal1 (rGal1) was obtained as described elsewhere 14. The protein was purified by affinity chromatography on a lactosyl-divinylsulfone-agarose column. Antibodies against human Gal1 were generated as described in 35 and 36. The immunolocalization of Gal1 was carried out as described for K562 cells 24.

For the preparation of cell extracts, a total of 17.10⁶ cells were solubilized in 1 ml of extraction buffer: 50 mM Tris pH 7.4, 150 mM NaCl, 5 mM EDTA, 0.1% SDS (w/v), 0.5% sodium deoxycholate (w/v), 0.5% Nonidet NP40 (v/v). Finally, 1 tablet of antiprotease cocktail (Roche, Meylan, France) was added to 10 ml of buffer. The cells were sonicated in ice three times. The lysat was centrifuged at 100, 000 g for 15 min at 4°C. The supernatant was collected and stored at -20°C until used. Proteins were resolved by discontinuous SDS-PAGE on 1.5 mm gel (T: 6–18%) according to the method of Laemmli. 37. For Western blotting, proteins separated by SDS-PAGE were electrotransferred onto Immobilon-P membrane (Millipore). Blots were incubated with 1:50 anti-Gal1 mAb, or 1:500 anti-pl6 mAb (Pharmingen) for 2 h at room temperature. The blots were developed with anti-mouse Ig antibody-HRP diluted 1/10,000, followed by incubation in the Amplified Opti-4CN kit substrate (Bio-Rad). The image of the membranes was acquired from GS-700 Densitometer 4, and analyzed with Molecular Analyst Software (Bio-Rad).

During the four days of treatment with AzaC, cell proliferation appeared to slow (Fig. 1A). When BL36 cells were cultured with AzaC at the high concentration of 50 μM, cell proliferation was strongly inhibited. BL36 cells exposed to 5 μM AzaC also exhibited a significant reduced growth rate but these cells were >95% viable by propidium iodide test even at day 4 (data not shown), and were found metabolically active by MTS assay as described below. According to these results, the following experiments addressing metabolic activity, death, and cell differentiation were performed.

The proportion of metabolically active cells was determined using the MTS test in cultures treated with AzaC in comparison with untreated cells (Fig. 1B). The cellular conversion of MTS to the ultraviolet-absorbing formazan product has been demonstrated to be directly proportional to cellular metabolism resulting in the formation of reducing equivalents such as NADH or NADPH 38. A drop in the % of metabolically active cells of about 36% was observed over the 24 h period following the addition of 10 μM AzaC, and went up to 42% in cells treated with 50 μM AzaC. Cell death was confirmed by an increase in annexinV/PI staining (Fig. 2). Within 48 h of incubation with AzaC at 10 μM, the number of AnnexinV⁺ PI^- apoptotic cells and AnnexinV⁺ PI⁺ necrotic cells increased. For longer incubation times, induced cell death increased dramatically.

Studies investigating the mechanisms whereby B lymphoma cells are induced to undergo apoptosis demonstrated that an arrest in the cell cycle preceded apoptosis 39. To determine whether AzaC modified the cell cycle distribution of BL36 cells, the DNA content of AzaC-treated cells was analyzed by PI staining. Cells were exposed to 10 μM AzaC for 2, 4 and 7 days. After 7 days, AzaC-treated cells exhibited a relative increase of cells in G0/G1 (77 %) in comparison with controls (56%). In the same time, the percentage of cells in the S phase compartment drops from 38 to 18%, suggesting an accumulation of cells into the G1 phase of the cell cycle (Fig. 3). However, there was no change in the relative cell cycle distribution at 2 or 4 days after AzaC treatment. By this time, a significant population of cells has undergone apoptosis. This suggests that, at least during the first days of treatment, inhibition of growth is uncoupled with an arrest in the cell cycle.

The effects of different concentrations of AzaC (5, 10, and 50 μM) on the expression of various surface antigen of BL36 cells were investigated. AzaC increased the cell population that expressed the cell-surface maturation marker CD23, CD30 present on activated B-lymphocytes 40 and CD138 (Syndecan-1). After the treatment with 50 μM AzaC some changes were observed that were not detected at lower concentrations: an increase of the cell population that expressed CD21, and a decrease of the cell population that expressed CD19 that is lost on maturation to plasma cells 41, and CD71 that is expressed on proliferating cells 42 (Table 1).

P16 is a biochemical marker of cell cycle progression 43 and is also commonly utilized as a marker of demethylation reaction 44. Consequently, we examined the effects of AzaC treatment on expression of p16 in BL36 cells. As expected, when these cells were treated with AzaC for 48 h, we detected the presence of p16 by Western Blotting (Fig. 4A). Thus, taken together the data showed that AzaC treatment induced an inhibition of cell growth related to an arrest at G0/G1 phase of the cell cycle, and confirmed the efficiency of AzaC treatment for expressing the tumour suppressor gene p16.

To determine whether AzaC is able to induce the expression of Gal1 in B lymphoma cells, protein extracts prepared form cells treated with 5–10 μM AzaC and from untreated cells were separated by SDS-PAGE and analyzed by immunoblotting with anti-Gal1 antibody. Fig. 4B,C shows the rate of variation in Gal1 expression induced by the differentiating agent. The galectin was not detected either in untreated BL36 cells or BL36 cells treated with AzaC for 24 or 48 h. The galectin became detectable in AzaC-treated cells by 96 h. Then, to determine whether the newly synthesised galectin molecules in AzaC-treated BL36 cells are expressed on the cell surface, the binding of anti-Gal1 Ab to untreated and treated BL36 cells was studied by immunocytochemistry. To visualize only extracellular Gal1, cultures of BL36 cells were incubated with anti-Gal1 antibody without permeabilization of their membranes. Over the 4-day time-course, AzaC-induced Gal1 did progressively localise to the cell surface. However, Gal1 did not exclusively localise to the cell surface but was also distributed throughout the cytoplasm (Fig. 5).

Several of the changes in AzaC-treated BL cells might be attributed directly to Gal1 induction, as Gal1 has been implicated in differentiation and growth inhibition 111213294546. A series of experiments were thus performed to address whether changes observed in BL36 were direct AzaC effects and/or secondary effects related to Gal1 expression induced by AzaC treatment. To this end, BL36 was treated with recombinant Gal1 (rGal1). Slower growth rates were confirmed for the rGal1-treated cultures. A drop in the % of metabolically active cells of about 28% was observed over the 24 h period following the addition of 700 nM rGal1. However the proportion of metabolically active cells in cultures treated with rGal1 vs. the proportion in untreated cultures was not modified after 4 days of treatment (Fig. 1C). Moreover the variation of cell death detected in treated-cells relative to the control cultures was discrete (<10%) and limited to the first day after the treatment (Fig. 2). To determine whether the inhibition of proliferation could be correlated to a modified phenotype, the effects of 700 nM rGal1 on the expression of various surface antigens shown to be modified or not by AzaC were investigated. The populations of cells that expressed CD19, CD23, CD45RO and CD45RA were unchanged. On the other hand, 24 h rGal1 treatment increased the cell population expressing CD138 (Fig. 6), a marker for plasma cells, while it decreased the expression of CD71. These modifications were reinforced after 48 h rGal1 treatment (Table 2).

In 1994, Chiariotti et al. 16 first reported experimental data showing that the expression of Gal1 can be induced in cultured hepatoma-derived cells by treatment with AzaC. Interestingly, Gal1 and AzaC individually have been shown to affect similar cell processes in cancer cells, including differentiation, and growth inhibition 121314152729464748495051. Available data are consistent with the suggestion that the expression of Gal1, accompanied by its secretion and its binding to cell surface receptors, could be involved in the AzaC observed effects in hematopoietic cells where Gal1 modulates differentiation or apoptosis. However, the mechanisms of these effects on hematopoietic cells are unclear yet.

In the present study, the effects of AzaC, on the cell phenotype, cell differentiation and cell death of BL36 cells were analyzed. The effects on cellular growth were time- and dose-dependent. Five μM AzaC caused significant inhibition on cellular growth, but the cell viability remained practically unchanged. AzaC increased the cell fraction in the G0/G1 phases, suggesting that AzaC inhibits cell division, which may be one of critical mechanisms of cell modulation by AzaC. AzaC-treated cells initially may become arrested at the G1 phase and then may either escape to the cycle arrest or die due to mechanisms leading to programmed cell death.

We also show that incubation of BL36 with AzaC induces expression of Gal1. AzaC is thought to exert its effects as a competitive inhibitor of cytosine methylation, resulting in the expression of silenced genes. The gene for Gal1 is one whose expression is possibly enhanced in this manner. In the results reported here, we found that Gal 1 was detected in cytosol after 120 h of treatment by 10 μM AzaC and then Gal1 was externalized and bound to cell surface receptors 24 h later. A key to understanding the extracellular biological functions of Gal1 is how its secretion appears to be regulated and re-directed during development and differentiation. Gal1 is likely released form vesicles close to the plasma membrane. On the basis of the data, we propose that the released Gal1 be immediately recruited to modulate cell activity. Gal1 may do this by interacting with and modulating cell receptors via its carbohydrate recognition domains because the Gal1-receptor interaction is abrogated by thiodigalactoside 29.

Others and we have previously reported that Gal1 binds to T and B lymphoblastoid cells 295253. Other studies have demonstrated that galectins are immunosuppressive, in animal models of autoimmune diseases 545556. Whereas the full role of Gal1 in modulating immune function is not yet understood, the increase in Gal1 expression by AzaC in BL cells suggests that Gal1 may play a role in the behaviour of normal leukocytes and of tumour cells.

What is the underlying mechanism? Although the regulatory machinery triggered by demethylating stimulus and resulting in phenotype modifications is not yet elucidated, it probably involves the stimulation of a signalling cascade that regulates cell proliferation and viability. A recently proposed model for such a cascade suggests the involvement of a cytoplasmic protein, AZ2 57. The amino-terminal part of the AZ2 protein is homologous to the previously reported TANK and I-TRAF, which participate in the signal transduction cascade from the TNF-receptor to the transcription factor NFkappaB. Demethylating stimulus may also modify a pathway activated by the membrane-anchored protein-tyro sine phosphatase CD45. Engagement of CD45 is known to regulate Src tyrosine kinases phosphorylation, phospholipase Cγ regulation, inositol phosphate production, diacylglycerol production, PKC activation, and calcium mobilisation 585960.

Increased synthesis and secretion of Gal1 by the cell could account for part of the phenotypic alterations detected in AzaC treated cells. Gal1-induced dimerisation and/or segregation might inhibit the catalytic site in CD45, thereby blocking tyrosine phosphates activity. Because Gal1 binding to cell surface receptors results in tyrosine phosphorylation 2961, it may allow a kinase-dependent signal to be transduced. Several studies have linked Gal1 expression with growth inhibition 27 and cell death 2662. However, the reports that some growth inhibitory agents did not induce Gal1 expression indicated that Gal1 expression is not dependent on the cell's growth state in general, through it may be involved in growth suppression 63. Moreover, it is likely that Gal1 acts in a manner to regulate specific signal transduction processes that is determined by the cell type and by the state of cell differentiation. In this work, exogenous rGal1 added to BL cells inhibited cell growth. Moreover, Gal1 as well as AzaC induced an expression of the cell surface plasma cell antigen, CD138, a phenotypic marker that identify cells with plasmacytic differentiation 64. This is consistent with the hypothesis that AzaC and Gal1 share similar signals for differentiation, however, since there was a significant difference in the expression of CD19 and CD23 after AzaC or Gal1 treatments it is likely that some pathways are specifically modified by AzaC. The mechanisms involved in these different pathways, important in clinical therapy, remain to be elucidated in the future. Ongoing studies are aimed at identifying as globally as possible the modifications resulting from AzaC treatment by using proteomics 65.