Alterations in protein glycosylation in PMA-differentiated U-937 cells exposed to mineral particles.

Carbohydrate moieties of cell glycoconjugates play a pivotal role in molecular recognition phenomena involved in the regulation of most biological systems and the changes observed in cell surface carbohydrates during cell activation or differentiation frequently modulate certain cell functions. Consequently, some aspects of macrophage response to particle exposure might conceivably result from alterations in glycosylation. Therefore, the effect of mineral particles on protein glycosylation was investigated in phorbol myristate acetate (PMA)-differentiated U-937. Jacalin, a lectin specific for O-glycosylated structures, showed a global increase in O-glycosylation in particle-treated cells. In contrast, no significant modifications were observed with concanavalin A, a lectin that recognizes certain N-glycosylated structures. The sialic acid-specific lectins Sambucus nigra agglutinin and Maackia amurensis agglutinin and the galactose-specific lectin Ricinus communis agglutinin revealed a complex pattern of alterations in glycoprotein glycosylation after crystalline silica or manganese dioxide treatments. Expression of sialyl Lewis(x), a glycosylated structure implicated in leukocyte trafficking, could not be detected in control or treated cells. This finding was consistent with the decrease in sialyl Lewis(x) expression observed during PMA-induced differentiation. In conclusion, various treatments used in this study induced quantitative as well as qualitative changes in protein glycosylation. Whether these changes are due to glycosidase release or to an alteration in glycosyltransferase expression remains to be determined. The potential functional implications of these changes are currently under investigation.


Introduction
It is now widely accepted that the carboregulation of leukocyte trafficking, which hydrate moieties of glycoproteins might act relies on the specific interaction between as ligands in biological recognition and for membrane-bound lectins, selectins, and this reason are implicated in many biologithe oligosaccharidic structure sialic acidcal processes such as cell-cell interaction, a [2,3]galactase-P [1,4] (fucose-c4 [1,3])Ncell migration, and proliferation (1)(2)(3)(4). acetylglucosamine (sialyl Lewisx) that This concept is well illustrated by the allows leukocytes to roll along the blood This paper is based on a presentation at The Sixth International Meeting on the Toxicology of Natural and Man-Made Fibrous and Non-Fibrous Particles held 15-18 September 1996 in Lake Placid, New York. Manuscript was received at EHP26 March 1997; accepted 22 April 1997.
Authors wish to thank M.-C. Jaurand for helpful discussions and C. Vaslin for her excellent secretarial assistance. This work was supported by institutional funding from Institut National de la Sante et de la Recherche M6dicale and European Union grant EV5V CT94 0399.
Address correspondence to Dr. Y. Pilatte, Institut National de la Sant6 et de la Recherche Medicale U139, vessels. This process, which results in activation and extravasation of inflammatory cells to the injured site, is of paramount importance in inflammation (4).
Several lines of evidence further support the biological importance of carbohydrates. For instance, alterations in the glycosylation of cell surface and/or soluble glycoconjugates were reported in many diseases (5,6). This is particularly evident in cancer, where these alterations are believed to be involved in the disturbance of cell adhesiveness, which leads to metastasis (7)(8)(9)(10). Alternatively, carbohydrates and especially sialic acids are essential in self/nonself discrimination by the immune system (11). Consequently, uncontrolled alterations in cell surface glycosylation may ultimately lead to disease (11,12).
The interaction between inhaled partides and alveolar macrophage(s) (AM) is the first step in macrophage activation, which is recognized as the origin of later dysfunctions in pneumoconiosis (13). Many studies have demonstrated the diversity of macrophage responses to mineral particles and it is likely that other aspects of this response remain to be determined (13). In this context, there is evidence suggesting that carbohydrates might be important in macrophage biology. For instance, glycosylation patterns were altered in murine macrophages upon activation (14), and similar observations were made in AM from patients with interstitial lung diseases such as sarcoidosis and idiopathic pulmonary fibrosis (15).
Given the potential physiopathological interest of carbohydrates, this study was designed to determine whether alterations in protein glycosylation occur in macrophages following mineral particle exposure. The effect of particles on protein glycosylation was investigated in U-937 cells, a human myelomonocytic cell line that can be induced to differentiate into monocyte/ macrophage by a phorbol ester (phorbol myristate acetate [PMA]) (16). Several reasons dictated the choice of this cell line. First, it was essential to use human cells because glycosylation is a species-specific parameter (2,3) that frequently makes studies on animal cells difficult to extrapolate to man. Second, although normal human AM would constitute a more realistic model, the use of such cells in experimental studies is limited by ethical considerations. Finally, PMA-differentiated U-937 cells were frequently used as a Environmental Health Perspectives * Vol 105, Supplement 5 * September 1997 I 11l53 TRABELSI ETAL. macrophage model in immunological studies (16), and in our research they were an interesting surrogate for investigating AM responses to mineral particles (Trabelsi et al., in preparation).
Cell Culture U-937 cells (a myelomonocytic cell line derived from a patient with histiocytic lymphoma) were maintained as a suspension culture at 37°C in a humidified chamber in the presence of 5% CO2. They were grown in RPMI 1640 medium supplemented with 10% heat-inactivated fetal calf serum, 2 mM glutamine, 50 U/ml penicillin, and 50 pg/ml streptomycin. Routinely, they were seeded at 1 x 105 cells/ml and readjusted at this concentration with fresh complete medium three times a week. In the experiments described herein, U-937 was differentiated with 0.162 pM PMA over 4 days, as previously described by Oberg et al. (16).

Mineral Partides
In this study we used three reference particles: DQ12, with a diameter of 2.2 pm and specific area of 3 m2/g; MnO2, with a diameter of 3.7 pm and specific area of 59 m2/g; and TiO2, with a diameter of 0.33 pm and specific area of 7.5 m2/g. Particle-specific area was determined as previously described (23). Particles were heated for 2 hr at 200°C to inactivate bacterial endotoxins, then sonicated for 5 min (50 kHz, 20 W) just before use. Cell Treatments After 4 days of PMA differentiation, U-937 cell lines were exposed to particles at 50 pg/cm2 over 48 hr. When sialyl Lewisx expression was investigated during U-937 differentiation, cells were harvested 1, 2, 3, or 4 days after addition of PMA.

Western Blotting
After cells were extensively washed with phosphate-buffered saline (PBS), they were homogenized at 40 x 106 cells/ml in PBS supplemented with 1% Nonidet P40, 10 mM EDTA, and 25 pM of the serine protease inhibitor p-nitrophenyl-p'-guanidobenzoate (Sigma, l'Isle d'Abeau, France). Proteins (10 pg for each sample) were first separated on 10% polyacrylamide discontinuous minislab gels according to the method of Laemmli (24), then electrophoretically transferred onto Immobilon membrane according to the method of Towbin et al. (25). Prestained molecular weight markers (Bio-Rad) were used to assess transfer efficiency and determine the size of blotted proteins. After its transfer, the Immobilon membrane was blocked overnight at room temperature in PBS supplemented with 2% polyvinylpyrolidone (PVP) (wt/vol) and 0.1 mM CaCl2 (PVP buffer), then incubated with various biotinylated lectins at 10 pg/ml in PVP buffer. The specificities of the lectins used in this study are described elsewhere (26)(27)(28). After their extensive wash, we incubated the membranes with peroxydaseconjugated streptavidin. After their wash, the lectin-labeled proteins were visualized on a Hyperfilm-ECL (Amersham) with an ECL Western blotting detection kit (Amersham) as recommended by the manufacturer. When the mouse monoclonal antibody anti-sialyl Lewisx was used as a carbohydrate-specific reagent, the Immobilon membrane was blocked overnight at room temperature with PVP buffer, then incubated with the monoclonal antibody at 0.5 pig/ml for 1 hr at room temperature. After the membrane was washed with PVP buffer, it was incubated with gold-labeled rabbit antibodies to mouse IgM for 2 hr at 37°C, washed again, and developed with the IntenseBL silver enhancement kit as described by the manufacturer Jansenn Life Sciences). Protein Asay Protein concentrations were estimated by a simplified Lowry's method (Bio-Rad) that was specially designed for detergentcontaining samples. Bovine serum albumin was used as standard.

Results
Western blotting using lectins or carbohydrate-specific antibodies is an easy and convenient way to analyze glycoprotein glycosylation (29). This technique was then used to investigate the possible alterations in PMA-differentiated U-937 cell line glycoproteins following phagocytosis of three reference particles: DQI2, MnO2, and TiO2. Special attention has been paid to sialic acid and fucose (Fuc) residues, which are located on the nonreducing termini of the saccharide chain, and to galactose (Gal) residues, which can be unmasked after enzymatic desialylation. The levels of 0-and N-glycans were also evaluated.

N-and aGlycosylation
Two major types of glycans are carried by glycoproteins; the first is attached through an asparagine side chain (N-glycans or Nglycosylation), and the second through a threonine or serine side chain (O-glycans or O-glycosylation). ConA, which binds certain mannosylated structures exclusively found in N-glycans, and Jacalin, which binds the disaccharide galactose-3(1,3)Nacetylgalactosamine; (Gal-f(1,3)GalNac) found exclusively in O-glycans, represent interesting tools to investigate N-and O-glycosylations (29).
As shown in Figure 1, more bands are observed with ConA than with Jacalin, which confirms that N-glycosylation is more frequent than O-glycosylation in mammalian cells. No significant changes were observed with ConA after particle treatments. The labeling profile and band intensity were similar in treated and control cells, with one exception ( Figure IA). Only one band could be visualized at about 50 kDa in MnO2-treated cells, whereas a doublet was observed at this level in all other samples. This band corresponded to the larger component of the doublet and its intensity was slightly higher. More dramatic changes were observed in O-glycosylation, as assessed by Jacalin labeling (Figure 1 B). Electrophoretic profiles were similar in all analyzed samples. However, labeling intensity was increased in the three treatments. This increase was slight but significant in TiO2-, moderate in MnO2-, and more pronounced in DQl2-treated cells.

Sialic Add Expression
Sialic acids occur as terminal su and a(2,6) linkages to Gal. T lectins able to discriminate b two linkages, which recognize a(2,6) and a(2,3), respective in this study ( Figure 3C). It should be noted that PNA is specific for the same disaccharide Gal-0(1,3)GalNac as Jacalin, but unlike Jacalin, it does not bind the sialylated form of this tgars in a(2,3) disaccharide (28). Iherefore, two ietween these > sialic acid in ly, were used (27). In the in Figure 2A Expression of sialyl Lewisx, a carbohydrate structure that is the common counter ligand of selectins, has been investigated by using a monoclonal antibody. We found no evidence of sialyl Lewisx expression in treated or untreated cells (not shown). Because U-937 is a cell line known to express this structure (30), we hypothesized that it might have been lost during PMA-induced differentiation. Protein extracts were prepared from U-937 cells after various times of differentiation and analyzed. After 24 hr differentiation, only the larger band was visible and its intensity faded gradually during differentiation (Figure 4). These observations are consistent with the absence of labeling initially observed in particle-treated and control cells, as cell extracts were prepared 6 days Environmental Health Perspectives * Vol 105, Supplement

Discussion
In this study we present the that quantitative as well a changes may affect the c moiety of cell glycoprotei exposure to mineral particles. particle induced a typical p; changes. TiO2 (17,18) showed or erate effect on protein glycosyl contrast, MnO2 (19,20) and DQ induced more significant alterat of the more prominent changes following treatment with these cles was an increase in O-gly structure expression detected b  (31). Moreover, these fairly simil PNA. Ten microues indicate that this parameter :h sample.
was not involved in the particl profiles of glycosylation alteratior It is interesting that the maji ifferentiation alterations were observed followin tS.
to MnO2 and DQ12, which related to the damage caused to t these particles, which are signifil first evidence toxic than TiO2 on PMA-diffi .s qualitative U-937 (Trabelsi et al., in prel arbohydrate However, this assumption shoul ns following pered by the fact that these partid The observed opposite effects on the level of gal structures recognized by RCA, whereas they induced similar cytotoxicity. Alternatively, phagocytosis of mineral partides is the first step of complex processes that lead to macrophage activation. Because the form of activation likely is dependent on the type of partides, the different effects of MnO2 and 97 kDa DQ12 on protein glycosylation might be related to the fact that they trigger distinct activation pathways. Similar but less pronounced results were obtained when the cells were exposed to partides at 25 instead 50 jig/cm2 or over 24 instead of 48 hr (data not shown). However, the fact that a 48-hr incubation seems optientiation on mum to detect glycosylation alterations I l s were har-does not rule out the possibility that signif-PMA (0.162 icant changes may occur during the early subjected to steps of particle phagocytosis. Further ected with a experiments are needed to clarify this n micrograms point. Similarly, although the doses of particles used in this study were in the range of those frequently used in in vitro experic, as each ments, the pathophysiological relevance of attern of the data reported in this paper remains to nly a modbe fully ascertained, as such high doses of lation. In particles are unlikely to be encountered in 12 (21,22) vivo. Along this line, alterations in sialic ions. One acid expression similar to those observed observed here with DQ12 were found in AM from two parti-individuals occupationally exposed to silica cosylated (Matrat et al., in preparation). )y Jacalin.
Several mechanisms can account for with other these glycosylation changes. For instance, bserved in differences in glycosylation are frequently a(2,3) in attributable to differences in the level of ted cells, expression of specific glycosyltranferases in a (2,6) (2,32). Therefore, the increased level of a )2-treated glycosylated structures detected by Jacalin in galactosy-particle-treated cells might be due to the RCA was upregulation of the N-acetyl galactosaminyl-12-treated transferase and the galactosyltransferase ated cells. involved in the synthesis of 0-linked glycan res for the core structures. Mineral particles may also ells ranged interfere with other aspects of glycoprotein vn), which glycan biosynthesis. Oligosaccharide chains ining fluid are assembled while the newly synthesized lar pH valproteins are moving through a specific r probably pathway from endoplasmic reticulum to the le-specific trans-Golgi network. Every compartment IS. along this pathway is equipped with a speor cellular cific set of enzymes to carry out the g exposure required reactions (33). Therefore, any dismight be turbance of this membrane flow may generthe cells by ate alterations in protein glycosylation. antly more Such disturbances are likely to be encounerentiated tered in cells treated with high doses of parparation). ticles, as decreases in organelle motion Id be tem-resulting from possible effects on the [es induced cytoskeleton were observed in AM in which actosylated the volume occupied by ingested particles Environmental Health Perspectives * Vol 105, Supplement 5 * September 1997 A Control Ti2 MnO2 DQ12 ON exceeded 7 to 8% of normal cell volume (34). Alternatively, as a given oligosaccharide can be carried by different glycoproteins (4), some of the changes observed for certain structures may be attributable to changes in the level of expression of glycoprotein scaffolds that display these structures. Finally, differentiated U-937 cells readily phagocytize the mineral particles used in this study (Trabelsi et al., in preparation), and phagocytosis is frequently associated with an important release of lysosomal glycosidases (35) that are able to degrade oligosaccharides. Therefore, the involvement of these glycosidases cannot be excluded when a band disappears or when its intensity decreases following treatment. The functional significance of particleinduced alterations in glycoprotein glycans remains to be determined, but the reduced sialylation observed following DQ 2 and MnO2 treatment might be of importance. This observation is consistent with the concept that mineral particles activate macrophages; reduction in glycoprotein sialylation is known to occur during the priming step of macrophage activation (14). However, sialic acid is an important element in cell reactions with and perception by its external environment (11). Consequendy, any qualitative or quantitative changes in cell sialylation may modulate or alter certain cell functions (11). For instance, interactions of macrophages with various subsets of T lymphocytes are influenced by the level of macrophage cell surface sialylation (11). Likewise, particle-induced changes in terminal Gal expression observed in this study are also of interest. Indeed, these galactosylated structures are likely ligands for mammalian membrane-bound and soluble Gal-specific lectins (11,36) and for naturally occurring anti-Gal antibodies (11,12). The biological significance of these Gal-specific molecules has yet to be fully understood but there is increasing evidence that they play an important role in multiple biological processes through interaction with specific ligands (37).
In conclusion, we show that mineral particles induce specific patterns of changes in the carbohydrate moiety of glycoproteins. Given the biological implications of carbohydrates, these changes may be critical in modulating macrophage functions. Identification of the proteins affected by these changes, as well as elucidation of the mechanisms that cause them, might provide interesting clues in our understanding of macrophage response to mineral particles.