Tumor necrosis factor alpha increases the expression of glycosyltransferases and sulfotransferases responsible for the biosynthesis of sialylated and/or sulfated Lewis x epitopes in the human bronchial mucosa.

There is increasing evidence that inflammation may affect glycosylation and sulfation of various glycoproteins. The present study reports the effect of tumor necrosis factor alpha (TNF-alpha), a proinflammatory cytokine, on the glycosyl- and sulfotransferases of the human bronchial mucosa responsible for the biosynthesis of Lewis x epitope and of its sialylated and/or sulfated derivatives, which are expressed in human bronchial mucins. Fragments of macroscopically normal human bronchial mucosa were exposed to TNF-alpha at a concentration of 20 ng/ml. TNF-alpha was shown to increase alpha1,3-fucosyltransferase activity as well as expression of the two alpha1,3-fucosyltransferase genes expressed in the human airway, FUT3 and FUT4. It had no influence on alpha1,2-fucosyltransferase activity or FUT2 expression. It also increased alpha2,3-sialyltransferase activity and the expression of ST3Gal-III and, more importantly, ST3Gal-IV and both N-acetylglucosamine 6-O-sulfotransferase and galactose 3-O-sulfotransferase. These results are consistent with the observation of oversialylation and increased expression sialyl-Lewis x epitopes on human airway mucins secreted by patients with severe lung infection such as those with cystic fibrosis, whose airways are colonized by Pseudomonas aeruginosa. However, other cytokines may also be involved in this process.

Human bronchial mucins represent a very broad family of polydisperse high molecular weight glycoproteins that are part of the innate airway immunity. Apomucins, which correspond to their peptide part, are encoded by at least six mucin genes (MUC1, MUC2, MUC4, MUC5B, MUC5AC, and MUC7). Bronchial mucins are highly glycosylated and contain from one single to several hundreds of carbohydrate chains. These carbohydrate chains that cover the apomucins are extremely diverse, adding to the complexity of these molecules. Currently, more than 150 different O-linked carbohydrate chains have been described, and the mucins from a single individual probably contain a few hundred different carbohydrate chains (1). Due to their wide structural diversity forming a combination of carbohydrate determinants, as well as their location at the surface of the airways, mucins may be involved in multiple interactions with microorganisms and are very important in the protection of the underlying airway mucosa.
The biosynthesis of these carbohydrate chains is a stepwise process involving many glycosyl-and sulfotransferases. The only structural element shared by all mucin O-glycan chains is an N-acetylgalactosamine residue linked to a serine or threonine residue of the apomucin. The nonreducing end of the chains, which corresponds to the termination of the biosynthetic process, may bear different carbohydrate structures, such as blood groups A, B, or O determinants; H and sulfated H determinants; and Lewis a, Lewis b, Lewis y, and various derivatives of the Lewis x epitope. The synthesis of these different terminal determinants involves different pathways utilizing a variety of transferases (1,2).
Although bronchial mucins secreted by patients suffering from cystic fibrosis or chronic bronchitis share various epitopes such as the sialyl-Le x 1 and sulfo-Le x determinants (4 -8), the sulfo-sialyl-Le x epitope has been mostly characterized in mucins secreted by patients suffering from cystic fibrosis (8), in agreement with the oversulfation of mucins observed in this disease.
Recently, differences have been observed in the glycosylation of bronchial mucins secreted by patients suffering from bronchial diseases according to the severity of bacterial infection (9). These observations suggest that chronic and severe inflammation of the airway mucosa may be responsible for increased sialylation and expression of sialyl-Le x epitope on the carbohydrate chains of these mucins.
There is increasing evidence for a link between inflammation and glycosylation. Since the original work describing increased activity of several glycosyltransferases of the rat liver during turpentine-induced inflammation (10), many modifications in the glycosylation of acute-phase glycoproteins have been ob-served during acute and chronic inflammation. In humans, inflammation induces modifications of the N-glycan chains of different acute-phase glycoproteins, such as ␣1-acid glycoprotein (11)(12)(13), haptoglobin (14,15), or ␣1-antitrypsin (16). These modifications may differ according to the early or late phase of acute inflammatory response (17) or the stage of the disease (18). Increased fucosylation and increased expression of sialyl-Le x epitopes have been observed in ␣1-acid glycoprotein (11,17,19): these modifications seem to be related to an induction of the fucosyltransferase FUT6 in the liver (20). Monokines may regulate the glycosylation of acute-phase proteins (21). As a matter of fact, IL-1␤ and TNF-␣ affect the glycosylation of rat ␣1-acid glycoprotein. IL-1 and IL-6 change the glycosylation of human ␣1-acid glycoprotein produced by hepatocytes (22,23), whereas transforming growth factor ␤ has an effect opposite to that produced by IL-6 (24).
TNF-␣ is also an important factor in airway mucosa inflammation, acting as an initial inflammatory cytokine that subsequently regulates both early neutrophil infiltration and eosinophil recruitment into the lung and airspace (31). TNF-␣, as other cytokines, is found in the airways of patients suffering from bronchial diseases such as chronic bronchitis or cystic fibrosis (32,33) and might affect the glycosylation of airway mucins.
The present work was designed to determine the effect of TNF-␣ on (i) the ␣2,3-sialyltransferases, ␣1,3-fucosyltransferases, galactose-3-O-sulfotransferase, and N-acetylglucosamine-6-O-sulfotransferase activities involved in the biosynthesis of Le x , sialyl-Le x , sulfo-Le x , and sulfo-sialyl-Le x determinants by the human bronchial mucosa and (ii) the expression of the different genes possibly encoding these enzymes.

EXPERIMENTAL PROCEDURES
Explant Culture-Tissues were collected in macroscopically healthy areas of the bronchial tree from patients undergoing surgery for bronchial carcinoma. They were immersed in Leibovitz L15 medium (Invitrogen), immediately transported on ice to the laboratory, and then processed to isolate the mucosa. Mucosa (2-3 cm 2 ) were cut into 1-mm 2 pieces and suspended in CMRL-1066 medium (Invitrogen) complemented with 0.2 mM L-glutamine (34). They were maintained at 37°C for 4 or 16 h in the presence or absence (controls) of 20 ng/ml TNF-␣ (Prepro Tech, London, United Kingdom).
Enzyme Preparation-After incubation with or without TNF-␣, explants were disintegrated with a glass-Teflon homogenizer (1,400 rpm, five strokes) in 50 mM Tris-HCl buffer, pH 7.4, containing 25 mM potassium chloride, 250 mM saccharose, 5 mM ␤-mercaptoethanol, and 5 mM magnesium acetate (34). The mixtures obtained were subjected to 16,000 ϫ g centrifugation for 20 min at 4°C. The supernatants were ultracentrifuged further at 180,000 ϫ g for 1 h at 10°C. The resulting pellets containing microsomal fractions were stored at Ϫ80°C until use (34), as were the 180,000 ϫ g supernatants.
Genotyping of Fucosyltransferases FUT2 and FUT3-Before assaying ␣1,2or ␣1,3-fucosyltransferase activities, blood samples were obtained from patients to define their secretor and Lewis status. Methods based upon polymerase chain reaction and restriction fragment length polymorphism were used to detect mutations of the FUT2 and FUT3 genes (35,36).
GlcNAc-6-O-sulfotransferase and Galactose-3-O-sulfotransferase Assays-The GlcNAc-6-O-sulfotransferase assay was performed as described previously (37). The incubation mixture contained 50 -100 g of microsomal protein, 0.  After incubation for 1 h at 30°C, the reactions were stopped by the addition of ice-cold methanol. The mixtures were kept at 4°C overnight, and the resulting precipitates were eliminated by centrifugation at 10,000 ϫ g for 20 min. The pellets were washed twice with ice-cold methanol and centrifuged. The supernatants were pooled, evaporated to dryness, and then directly subjected to HPAEC-PAD (34).
Isolation of Radiolabeled Products by HPAEC-PAD-Dry samples of sulfated, fucosylated, or sialylated radiolabeled products were dissolved in water and injected directly into a CarboPac PA-100 column (4 ϫ 250 mm) for HPAEC (Dionex Corp.). The elution of neosynthesized products was monitored both by pulsed amperometric detection (PAD 2 model; Dionex Corp.) and by radioactivity on line (high performance liquid radioactivity detector LB 506 C-1; EG & G, Berthold, Wildbad, Germany).
Elution of sulfated products was performed at alkaline pH, with a flow rate of 1 ml/min in 0.05 M NaOH/0.2 M sodium acetate with a linear gradient of sodium acetate to 0.05 M NaOH/0.3 M sodium acetate at 22 min and 0.05 M NaOH/0.95 M sodium acetate at 24 min, followed by isocratic elution with 0.05 M NaOH/0.95 M sodium acetate for 10 min (gradient I).
Elution of fucosylated or sialylated products was performed at alkaline pH at a flow rate of 1 ml/min in 0.1 M NaOH for 10 min, followed by a linear gradient of sodium acetate to 0.1 M NaOH/0.07 M sodium acetate at 16 min, 0.1 M NaOH/0.1 M sodium acetate at 30 min, and 0.1 M NaOH/0.45 M sodium acetate at 80 min (gradient II).
The activity of the different enzymes was expressed as femtomoles of radiolabeled nucleotide sugar transferred/min/mg of protein for ␣1,2and ␣1,3-fucosyltransferases and as picomoles of radiolabeled nucleotide sugar transferred/min/mg of protein for ␣2,3-sialyltransferases and sulfotransferases.
Protein Determination-The protein content of microsomal fractions was determined by BCA Protein Assay (Pierce) (40).
RNA Isolation and cDNA Synthesis-Total RNA was isolated from fragments of bronchial mucosa and incubated with and without TNF-␣ for 4 h, using the guanidinium thiocyanate/CsCl method according to Ref. 41. Isolated RNA was pretreated with DNase I (Invitrogen) and subjected to Expand TM reverse transcriptase (Roche Molecular Biochemicals) in the presence of oligodeoxythymidilic acid 12-18 primer (Invitrogen) for cDNA synthesis according to the manufacturer's instructions. After 16 h of incubation at 37°C, there was no expression of glycosyltransferases in the control explants.
Semiquantitative PCR Analysis of Glycosyltransferase Expression-The oligonucleotides used as primers for the PCR reactions are given in Table I. They were obtained from Genset (La Jolla, CA) and have been described previously (35,(42)(43)(44)(45)(46). PCR of mRNAs from FUT1, FUT2, FUT3, FUT4, and FUT5 was carried out under the conditions described previously (47). For FUT6 and FUT7, we used the same conditions used for other fucosyltransferases, with a different number of amplification cycles (35 cycles for FUT6 and 37 cycles for FUT7). The conditions used for amplification of each ␣2,3-sialyltransferase mRNA were as follows: 94°C for 2 min (1 cycle); 94°C for 1 min, 58°C for 1 min, and 72°C for 1 min 30 s (35 cycles); and 72°C for 7 min. The conditions used for ␤-actin are described in Ref. 42.
In preliminary experiments, optimized conditions were determined to obtain PCR products proportional to the transcript of interest. The expression of glycosyltransferase genes was compared with the expres- sion of the ␤-actin gene. ␤-Actin was expressed equally in all samples, indicating that the glycosyltransferase mRNAs were obtained from similar quantities of total RNA. The amplified fragments were verified by digestion with restriction enzymes from Invitrogen (43). To determine whether the amplified fragments were FUT3 or FUT5 products, a digestion was performed with NaeI, which cleaves the fragments from FUT5 but not those from FUT3. Eleven l of PCR products were subjected to electrophoresis (220 V, constant-voltage field) on a 2% agarose gel equilibrated in Tris-borate electrophoresis buffer containing ethidium bromide (1 g/ml). Gels were photographed under UV light and analyzed by computerized scanning of the image using the Gel Analyst 3.01 program. Sizes of the amplified fragments were checked according to the migration of DNA ladders (Roche Molecular Biochemicals).

RESULTS
TNF-␣ has a concentration-dependent effect on the secretion of mucins by human airway organ explant cultures (48). At a concentration of 20 ng/ml in these cultures, it increases MUC2 mRNA levels and, within 8 h, more than doubles the airway mucin secretion. TNF-␣ also increases MUC2 and MUC5B expression by cancer cell line LS180 (49) and MUC2 expression in middle ear epithelium (50).
The effect of increasing concentrations of TNF-␣ on the ␣1,3fucosyltransferase activity of a bronchial mucosa explant was checked. After a 4-h incubation, this activity (using 5 and 10 ng/ml TNF-␣) was not different from that of the control (93.8 and 99.9 fmol radiolabeled fucose transferred/min/mg microsomal protein, respectively, as compared with 98 fmol radiolabeled fucose transferred/min/mg microsomal protein), in contrast to the increase observed at 20 ng/ml (159.7 fmol radiolabeled fucose transferred/min/mg microsomal protein).
Effect of TNF-␣ on ␣1,2-Fucosyltransferases-When microsomes from bronchial mucosa explants were incubated with [ 3 H]GDP-Fuc and Gal␤1-4GlcNAc, the HPAEC elution profile showed one characteristic peak at 7 min that was absent when incubations were performed without any substrate. When the radiolabeled products were injected with nonlabeled Fuc␣1-2Gal␤1-4GlcNAc, the neosynthesized fucosylated product coeluted exactly with this standard, showing the presence of ␣1,2-fucosyltransferase activity in the microsomal fractions. A similar elution profile was obtained with microsomes from bronchial mucosa explant treated for 4 h with TNF-␣. The incorporation of [ 3 H]GDP-Fuc in Gal␤1-4GlcNAc was similar to that of the control, indicating that TNF-␣ had no effect on ␣1,2-fucosyltransferase activity (Table II). When the explants were maintained for 16 h at 37°C, no ␣1,2-fucosyltransferase activity was recovered in the microsomes of controls or in those of TNF-␣-stimulated explants. No ␣1,2-fucosyltransferase activity was detected in the 180,000 ϫ g supernatant.
The expression of FUT2 mRNA in explants incubated for 4 h was not significantly modified by TNF-␣ (Table III). The FUT1 mRNA was not detected, even when PCR experiments were performed with a higher number of cycles.
A similar elution profile was obtained when incubations were performed with microsomes from bronchial mucosa explant treated with TNF-␣. However, the incorporation of radiolabeled fucose from [ 3 H]GDP-Fuc in Gal␤1-4GlcNAc was increased in microsomes from explants treated for 4 h with TNF-␣ as compared with control (Table II). When the incubation of explants with TNF-␣ was extended to 16 h, the incorporation of radiolabeled fucose was even higher as compared with the control (Table II). After incubation with TNF-␣, no ␣1,3-fucosyltransferase activity was detected in the 180,000 ϫ g supernatant.
Incubation of explants with TNF-␣ for 4 h stimulated the expression of FUT3 and FUT4 mRNAs ( Fig. 3 and Table III). The primers used to amplify FUT3 cDNA (447 bp) also amplified FUT5 cDNA (486 bp). In our experiments, a single band appeared after migration on an 8% polyacrylamide gel; this reverse transcription-PCR product was not cleaved by NaeI digestion (43), indicating that only FUT3 mRNA had been amplified. The expression of the mRNA from FUT3 was also significantly modified after incubation with TNF-␣ for 4 h (Fig.  3 and Table III). FUT5, FUT6, and FUT7 are not normally expressed in the human bronchial explants, and they were not induced by TNF-␣.
Effect of TNF-␣ on 2,3-Sialyltransferases-When microsomes from bronchial mucosa explants were incubated with [ 14 C]CMP-NeuAc and Gal␤1-4GlcNAc, the HPAEC elution profile showed one characteristic peak at 22 min 36 s that was absent when incubations were performed without the substrate. When the radiolabeled products were injected with nonlabeled NeuAc␣2-3Gal␤1-4GlcNAc, the neosynthesized sialylated product co-eluted exactly with this standard, showing the presence of ␣2,3-sialyltransferase activity in the microsomal fractions (Fig. 4). TNF-␣ increased incorporation of radiolabeled NeuAc in Gal␤1-4GlcNAc at 4 h and increased it even more at 16 h (Table II). After a 16-h incubation, some sialyl- transferase activity was observed in the 180,000 ϫ g supernatants of both control and TNF-␣ explants (0.79 Ϯ 0.04 and 1.6 Ϯ 0.14 pmol/min/mg protein, respectively). The expression of the mRNAs from ST3Gal-III and ST3Gal-IV was significantly increased by TNF-␣ ( Fig. 3 and Table III).  (Table III). When microsomes were obtained from bronchial mucosa explant treated with TNF-␣ for 4 h, both sulfotransferases were significantly increased. These microsomal activities were lost when the explants were maintained for 16 h. After a 16-h incubation, some galactose-3-O-sulfotransferase and GlcNAc-6-O-sulfotransferase activities were observed in the 180,000 ϫ g supernatants of both controls (0.03 Ϯ 0.01 and 0.06 Ϯ 0.01 pmol/min/mg protein, respectively) and TNF-␣ explants (0.08 Ϯ 0.1 and 0.17 Ϯ 0.005 pmol/ min/mg protein, respectively). DISCUSSION TNF-␣ is a multifunctional proinflammatory cytokine able to activate diverse target genes, and its actions are mediated by several signal transduction pathways.

Effect of TNF-␣ on Galactose 3-O-sulfoltransferase and N-
The present study demonstrates its influence on the expression of different glycosyl-and sulfotransferases involved in the biosynthesis of various Lewis x epitopes by the human bronchial mucosa.
TNF-␣ may influence the expression of some mucin genes and transferases involved in posttranslational modification of the peptides. At a concentration of 20 ng/ml in human airway organ cultures, it increases the expression of a mucin gene, MUC2, and, within 8 h, more than doubles the airway mucin secretion (48). The present study indicates that, at a similar concentration, TNF-␣ also influences glycosylation and sulfation. It increases the expression of ␣2,3-sialyltransferase, ␣1,3-fucosyltransferases, galactose-3-Osulfotransferase, and N-acetylglucosamine-6-O-sulfotransferase, which are involved in the biosynthesis of Lewis x, sialyl-Lewis x, sulfo-Lewis x, and sulfo-sialyl-Lewis x determinants by the human bronchial mucosa (Fig. 1), but it does not influence the ␣1,2-fucosyltransferases.
Human bronchial mucosa expresses ␣1,2-fucosyltransferase b Amplified products were quantified in each sample using the Gel Analyst 3.01 software. The densitometric values of cDNA products corresponding to the expression of ␤-actin and transferase genes were calculated for each reaction. The results were expressed as relative expression units (transferase/␤-actin scores) TNF-␣ . FUT-1, -5, -6, and -7 were not expressed in these experiments, and they were not induced by TNF-␣. c The p value was calculated using the Wilcoxon signed-rank test. d n.s., nonsignificant (p Ͼ 0.05).

FIG. 2. HPAEC elution profile of mixtures containing radiolabeled products synthesized when equal amounts of microsomal proteins from bronchial explants treated with TNF-␣ (B) or not treated with TNF-␣ (A) were incubated with [ 3 H]GDP-Fuc and
Gal␤1-4GlcNAc. The elution was monitored by radioactivity detection (solid line) for the synthesized products and by PAD (dashed line) for the nonlabeled standard, Gal␤1-4[Fuc␣1-3]GlcNAc. Elution was performed with gradient II (described under "Experimental Procedures").  (Table II). This activity is due to the fucosyltransferase FUT2, or secretor enzyme, responsible for the secretor status of human airway mucins (9). This activity is lost or degraded when the explants are kept for 16 h. TNF-␣ has no effect on the expression of FUT2 (Table II). Interestingly, the FUT1 enzyme is not expressed in any of the adult mucosa studied thus far, in contrast to cultures of human airway cells, where its expression appears after 3 weeks (47). The unique expression of FUT2 is consistent with the observation that human airway mucins secreted by nonsecretor individuals belonging to blood group O never express H determinants (9).
In addition to the secretor enzyme, human bronchial mucosa expresses an ␣1,3-fucosyltransferase activity that is increased after 4 h of incubation with TNF-␣ and increased even more after 16 h of incubation (Table II). A small part of this activity is released in the 180,000 ϫ g supernatant during microsome preparation. In the bronchial mucosa, ␣1,3-fucosyltransferases are considered to be active in the biosynthesis of Lewis x or sialylated Lewis x determinants. Different genes (FUT3, FUT4, FUT5, FUT6, and FUT7) might be responsible for such an activity. As a matter of fact, the human adult bronchial explants only express FUT3 and FUT4 (Fig. 3). This expression of FUT4 may explain the observation that human airway mucins from nonsecretor individuals, which do not express active FUT3, are nevertheless fucosylated (9), Moreover, because FUT7 is expressed in activated Th1 CD4 T cells (26), its lack of expression in the respiratory explants is evidence that the ␣1,3-fucosyltransferase activity measured in these explants is not due to the few T cells present in this tissue.
The expression of both FUT3 and FUT4 genes in the explants analyzed in the present study was increased by TNF-␣. Therefore, the increased ␣1,3-fucosyltransferase activity by TNF-␣ in the human airway mucosa is most probably due to an increased expression of FUT3 and FUT4, although an overexpression of some still unknown fucosyltransferase cannot be excluded. The stimulatory effect of TNF-␣ on the ␣1,3-fucosyltransferase activity even increases when the incubation is extended to 16 h. These results are most probably related to an indirect effect of TNF-␣. TNF-␣ is considered to be an early marker of inflammation responsible for a cytokinetic cascade. The induction of other cytokines might explain the difference between the increased activity at 4 h and 16 h.
These results are in agreement with the observations that inflammation, and particularly TNF-␣, changes ␣3-fucosylation and induces sialyl-Lewis x expression on several glycoproteins (23,51,52). ␣1,3-Fucosyltransferases are required after ␣2,3-sialyltransferases in the generation of the sialyl-Le x determinants. However, the enzymes involved in this process differ from one tissue to another: FUT6 in the liver (20), FUT7 in activated Th1 CD4 T cells (24), and FUT3 and FUT4 in the airway mucosa. In the latter case, there is no expression of any other known ␣1,3-fucosyltransferase. From a general point of view, it is also interesting to note that IL-4, a Th2-polarizing cytokine, is known to inhibit FUT7 expression and binding to vascular selectins (53), suggesting that its action on other ␣1,3-fucosyltransferases should also be studied.
The human bronchial mucosa also has ␣2,3-sialyltransferase activity, which increases after incubation with TNF-␣ for 4 h and increases even more after 16 h of incubation with TNF-␣ (Table II). There are large differences in the levels of these fucosyl-and sialyltransferase activities that may be due to the fact that the oligosaccharidic substrate Gal␤1-4GclNAc used in the present study is not the natural substrate. Among the human ␣2,3-sialyltransferases already cloned, ST3Gal-III and ST3Gal-IV use this disaccharide sequence as an acceptor to synthesize sialyl-Lewis x (54). These two enzymes are expressed in the human bronchial mucosa, and their expressions increase after treatment with TNF-␣ (Table III). The ST3Gal-IV involved in the biosynthesis of sialyl-Le x is also up-regulated in activated Th1 CD4 T cells (26).
This increased activity is in agreement with the hypersialylation and overexpression of sialyl-Le x in mucins from severely infected patients as compared with mucins from noninfected patients (9). However, other cytokines might be overexpressed in severe inflammation of the respiratory mucosa and induce similar effects.
Finally, TNF-␣ also increases the activities of both galactose-3-O-sulfotransferase and N-acetylglucosamine-6-O-sulfotransferase, which can lead to various sulfated epitopes of bronchial mucins such as the 3-sulfo-Lewis x or the 6-sulfo-sialyl-Lewis x determinants (Fig. 1). These activities were increased after a 4-h incubation with TNF-␣ but were not observed after 16 h, probably indicating that, as seen for the ␣1,2-fucosyltransferase, they were degraded after 16 h of incubation. A cathepsin D-like proteinase has been reported previously to release an ␣2,6-sialyltransferase during acute-phase response in rat Golgi liver (55).
There are several observations indicating abnormal sulfation in CF. The sulfate content of bronchial mucins secreted by CF patients is increased (56 -59). Altered sulfation and glycosylation of glycoproteins secreted by CF cells in culture (60 -62), as well as hypersulfation of CF human airway mucins secreted by a xenograft model of CF airway mucosa, have also been reported (63). Because there is no bacterial infection in the xenograft model, a link between hypersulfation of CF mucins and the primary defect of the disease has been envisaged (63).
The airways of patients suffering from CF are usually heavily infected by Pseudomonas aeruginosa, and the relations between infection and inflammation have been questioned by investigators who observed the precocity of lung inflammation in CF patients, possibly before colonization by P. aeruginosa (64 -67). Although CF mice have no spontaneous airway infection by P. aeruginosa, they have an excessive inflammatory response of the airways when they are challenged with P. aeruginosa (68).
In a recent work, mucins secreted by patients suffering from either cystic fibrosis or chronic bronchitis, with or without a severe infection, were compared for their sialic acid and sulfate contents, as well as for sialyl-Lewis x expression (9). As already mentioned, this study described the hypersialylation and overexpression of the sialyl-Lewis x epitope in mucins from severely infected patients and confirmed the higher sulfation already reported in cystic fibrosis (56 -58). Interestingly, the sulfate content of the mucins from the infected patients was also higher than that of the mucins from the noninfected patients, raising the question of a possible influence of severe inflammation on the sulfation process of the bronchial mucosa (9).
In airway inflammation, the release of TNF-␣ and other exokins induce the expression of E-selectin and adhesion molecules such as intercellular adhesion molecule 1 on microvascular endothelial cells, allowing the migration of leukocytes through the endothelium. Moreover, TNF-␣ up-regulates the expression of intercellular adhesion molecule 1 by airway epithelial cells (69) and favors the subsequent migration of leukocytes across the epithelium (70) and their retention in the airways (69).
Simultaneously, the up-regulation of the different fucosyltransferases, sialyltransferases, and sulfotransferases of mucin-secreting cells by TNF-␣ may increase the synthesis of Lewis x epitopes, especially sialyl-Le x , sulfo-Le x , and sulfosialyl-Le x , on airway mucins, allowing the attachment of leukocytes to the mucus film covering the airway lumen.
These epitopes are largely expressed on CF mucins (8) and are also possible sites of attachment for P. aeruginosa (71,72). Therefore, by offering a large array of possible ligands for leukocytes and for the adhesins of P. aeruginosa, they may contribute to the chronicity of airway infection in cystic fibrosis and in other chronic bronchial diseases with severe inflammation. Moreover, P. aeruginosa interacting with mucins may escape opsonophagocytic killing by human polymorphonuclear leukocytes (73).
In conclusion, the action of TNF-␣ on the glycosylation and sulfation process of the human bronchial mucosa raises the more general question of the influence of inflammation on posttranslational modifications of proteins, such as glycosylation and sulfation. In the future, it will be necessary to find out whether other pro-inflammatory cytokines (IL-1 and, more specifically, IL-6 and IL-8, which are secreted in abundance by CF cells (74)) may also alter the glycosyltransferase and sulfotransferase expression pattern of the human bronchial mucosa. In human epithelial cells, signaling induced by TNF-␣ was recently described as occurring through nuclear factor B (69, 75, 76) and activator protein 1 (76). In the future, it will be important to study the regulatory mechanisms of glycosyltransferase and sulfotransferase genes by TNF-␣.