TNF-α increases the carbohydrate sulfation of CD44: induction of 6-sulfo N-acetyl lactosamine on N- and O-linked glycans

Glycobiology vol. 12 no. 10 pp. 613 622, 2002 TNF-α increases the carbohydrate sulfation of CD44: induction of 6-sulfo N-acetyl lactosamine on N- and O-linked glycans Marc Delcommenne 3, Reiji Kannagi 4, and Pauline Johnson 1,2 3 Department of Microbiology and Immunology, 6174 University Boulevard, University of British Columbia, Vancouver, British Columbia, V6T 1Z3 Canada, and 4 Program of Experimental Pathology, Aichi Cancer Center, Nagoya 464, Japan Received on May 8, 2002; revised on June 18, 2002; accepted on June 23, 2002 CD44andsulfationhavebothbeenimplicatedinleukocyte adhesion. In monocytes, the inflammatory cytokine tumor necrosis factor α (TNF-α) stimulates CD44 sulfation, and this correlates with the induction of CD44-mediated adhesion events. However, little is known about the sulfation of CD44 or its induction by inflammatory cytokines. We determined that TNF-α inducesthe carbohydrate sulfation of CD44. CD44 was established as a major sulfated cell surface protein on myeloid cells. In the SR91 myeloid cell line, the majority of CD44 sulfation was attributed to the glycosaminoglycan chondroitin sulfate. However, TNF-α stimulation increased CD44 sulfation two- to threefold, largely attributed to the increased sulfation of N- and O-linked glycans on CD44. Therefore, TNF-α induced a decrease in the percentage of CD44 sulfation due to chondroitin sulfate and an increase due to N- and O-linked sulfation. Furthermore, TNF-α induced the expression of 6-sulfo N-acetyl lactosamine (LacNAc)/Lewis x on these cells, which was detected by a monoclonal antibody after neuraminidase treatment. This 6-sulfo LacNAc/Lewis x epitope was induced on N-linked and (to a lesser extent) on O-linked glycans present on CD44. This demonstrates that CD44 is modified by sulfated carbohydrates in myeloid cells and that TNF-α modifies both the type and amount of carbohydrate sulfation occurring on CD44. In addition, it demonstrates that TNF-α can induce the expression of 6-sulfo N-acetyl glucosamine on both N- and O-linked glycans of CD44 in myeloid cells. Key words: carbohydrate sulfation/cd44/glycosylation/ inflammatory cytokines Introduction Sulfation is a biosynthetic modification occurring on proteins, lipids, polysaccharides, and glycosaminoglycans (GAGs). Sulfation creates neoepitopes by adding negative charges and can confer novel physical and chemical properties to molecules, 1 Present address: Section of Bone Marrow Transplantation, Rush Presbyterian St. Luke s Medical Center, Chicago, IL 60612, USA 2 To whom correspondence should be addressed; E-mail: pauline@interchange.ubc.ca which can contribute to their biological function. Sulfated GAGs are important structural constituents of basement membranes and various types of connective tissues (reviewed in Kjellen and Lindahl, 1991). Sulfated cell surface glycoproteins and proteoglycans have been implicated in a large range of functions, including cell adhesion, growth factor presentation, cell signaling, and development. In Drosophila, the heparan sulfate sulfotransferase mutant, sulfateless (sfl), impairs wingless (Lin and Perrimon, 1999) and fibroblast growth factor-mediated signaling (Lin et al., 1999) and the sulfotransferase mutant, pipe, blocks the formation of embryonic dorsoventral polarity (Sen et al., 1998). Sulfated carbohydrates have been shown to mediate cell adhesion events in several systems (reviewed in Brockhausen and Kuhn, 1997; Bowman and Bertozzi, 1999), but its role is perhaps best understood in the immune system, where cell surface molecules decorated with sulfated carbohydrates mediate leukocyte adhesion and lymphocyte homing (reviewed in Rosen, 1999; Hemmerich and Rosen, 2000). Sulfation of L-selectin ligands, such as GLYCAM-1 (Imai et al., 1993), CD34 (Baumheter et al., 1993), and podocalyxin-like protein (Sassetti et al., 1998), occurs constitutively on high endothelial venules (HEV) to facilitate the initial attachment and tethering of lymphocytes via L-selectin. A monoclonal antibody (mab), MECA-79, recognizes sulfated O-linked oligosaccharides present on HEV and blocks leukocyte adhesion (Hemmerich et al., 1994). Tyrosine sulfation, together with recognition of a sialyl Lewis x like epitope, is required for P-selectin binding to its ligand, PSGL-1, which facilitates activated endothelial cell neutrophil interactions (Pouyani and Seed, 1995; Sako et al., 1995; Wilkins et al., 1995). Tyrosine sulfation has also been reported on the chemokine receptors CCR5 and CXCR4 and on the C5a anaphylatoxin receptor, where it has been implicated in binding ligand and in the attachment of human immunodeficiency virus to chemokine receptors (Farzan et al., 1999, 2001). Therefore, there is increasing evidence for the sulfation of cell surface molecules and a role for this posttranslational modification in cell cell or cell ligand interactions, particularly in cells associated with immune cell functions. CD44 is a widely expressed cell adhesion receptor that binds hyaluronan (HA), a nonsulfated GAG present in the extracellular matrix (ECM), and mediates cell cell and cell-ecm interactions (reviewed in Lesley et al., 1997; Lesley and Hyman, 1998; Siegelman et al., 1999; Johnson et al., 2000). CD44 has been implicated in leukocyte adhesion to HEV (Jalkanen et al., 1986, 1987) and extravasation at inflammatory sites (Camp et al., 1993; DeGrendele et al., 1997). Interestingly, CD44 isolated from peripheral blood leukocytes is sulfated (Jalkanen et al., 1988). Although CD44 can exist as several isoforms due to alternative splicing, in leukocytes it is predominantly present as the standard isoform of ~85 kda, 2002 Oxford University Press 613

M. Delcommenne et al. referred to as CD44H, containing none of the alternatively spliced exons. CD44 can be modified by GAGs, including chondroitin sulfate (CS), heparan sulfate, and keratan sulfate (Takahashi et al., 1996; Greenfield et al., 1999). This typically results in the expression of higher-molecular-mass species of CD44 in the range of 150 200 kda (Jalkanen et al., 1988). GAG-modified CD44 has been shown to bind components of the ECM, fibronectin (Jalkanen and Jalkanen, 1992) and collagen (Carter and Wayner, 1988; Ehnis et al., 1996) as well as certain chemokines (Tanaka et al., 1993) and growth factors (Bennett et al., 1995), and can promote cell migration (Faassen et al., 1992). CD44 is also modified with N- and O-linked glycans, and these oligosaccharides have been shown to play an important role in modulating the ability of CD44 to bind HA (Kincade et al., 1997; Lesley et al., 1997). However, the mechanisms regulating CD44 carbohydrate remodeling are still largely unknown. A multitude of factors can affect the HA binding ability of CD44, such as cytoskeletal associations, oligomerization, isoform expression, and posttranslational modifications, such as N- and O-linked carbohydrate, GAG addition, sialylation, and sulfation (reviewed in Isacke, 1994; Kincade et al., 1997; Lesley and Hyman, 1998; Johnson et al., 2000). However, the contribution and importance of these factors in the induction of HA binding by CD44 in response to physiological stimuli remain poorly defined. As with other leukocyte adhesion molecules, the ability of CD44 to bind its ligand, HA, is tightly regulated. Although CD44 is expressed in all leukocytes, most unstimulated cells do not bind HA. HA binding can be induced on cell activation by cytokines or antigen (Murakami et al., 1990; Lesley et al., 1994; Levesque and Haynes, 1997). In monocytes, tumor necrosis factor α (TNF-α) can induce HA binding by CD44, and this has been correlated with changes in CD44 isoform expression, glycosylation and GAG addition, sialylation, and sulfation (Levesque and Haynes, 1996, 1999; Maiti et al., 1998; Katoh et al., 1999; Jones et al., 2000; Brown et al., 2001). We have shown previously that TNF-α can induce CD44 sulfation in monocytes and in the SR91 myeloid cell line, and this correlated with the induction of HA binding and SR91 cell adhesion to an endothelial cell line (Maiti et al., 1998; Brown et al., 2001). Herein we identify the sulfated moiety of CD44 and determine the changes occurring to the sulfation of CD44 in response to the inflammatory cytokine TNF-α. Results CD44 is a major sulfated cell surface protein present on myeloid cells To determine the prevalence of sulfated membrane proteins in the SR91 myeloid cell line, plasma membrane proteins were biotinylated and isolated with neutravidin-coated beads after labeling the cells with [ 35 S]sulfate. Autoradiography after sodium dodecyl sulfate polyacrylamde gel electrophoresis (SDS PAGE) of the biotinylated membrane proteins indicated that there were very few sulfated proteins (Figure 1). In fact, only one major sulfated band was observed at approximately 85 kda. Immunoprecipitation of CD44 revealed that the 85-kDa form of CD44, CD44H, was sulfated to low levels in unstimulated SR91 cells and was increased on TNF-α stimulation (Figure 1). Immunoprecipitation of β 1 integrins, LFA-1, and ICAM-1 614 Fig. 1. CD44 is a major sulfated membrane protein present in unstimulated and TNF-α-stimulated SR91 cells. Cells were labeled with [ 35 S]sulfate, then cell surface molecules were biotinylated and immunoprecipitated with either anti-cd44 mab IM7 or streptavidin-coated beads (see Materials and methods). Immunoprecipitates were electrophoresed, transferred to PVDF membranes, and subjected to autoradiography (top). The middle panel is an anti-cd44 immunoblot of the same membrane. The membrane was stripped and reprobed with streptavidin-hrp to detect biotinylated cell surface proteins (bottom). The molecular mass markers are indicated on the left in kda. from SR91 cells revealed a low level of sulfate incorporation in β 1 integrins and ICAM-1 but no significant sulfate incorporation in LFA-1 (data not shown). CD44 was also one of the major sulfated membrane proteins present in another myeloid cell line, KG1a (data not shown). Therefore, sulfation of membrane proteins in myeloid cells is a selective event, and CD44 is a major sulfated membrane protein. The majority of sulfate incorporated into CD44 is present on O-linked posttranslational modifications To gain insight into the type of CD44 sulfation occurring in unstimulated and TNF-α-stimulated SR91 cells, we assessed whether CD44 sulfation was affected by removal of both O-linked glycans and GAGs by β-elimination. Figure 2 indicates that the majority of sulfation was lost from CD44 on unstimulated cells after β-elimination, whereas the amount of sulfation present on CD44 isolated from TNF-α-stimulated SR91 cells was significantly reduced but not abolished. Although a time course for β-elimination was performed, it was difficult to ascertain if the reaction had gone to completion because prolonged incubation resulted in the loss of protein. Because antibody reactivity was lost after β-elimination, the amount of CD44 protein was monitored by parallel labeling of the samples with [ 35 S]methionine-cysteine (Figure 2). When data from several experiments were compared, 77 ± 18% (N =6) loss of [ 35 S]sulfate from CD44 occurred from unstimulated cells and 73 ± 11% (N =6)of[ 35 S]sulfate was removed by

TNF-α induces changes to the carbohydrate sulfation of CD44 Fig. 2. β-elimination significantly reduces the sulfation of CD44 immunoprecipitated from unstimulated and TNF-α-stimulated SR91 cells. Cells were labeled with [ 35 S]sulfate or [ 35 S]methionine-cysteine for 24 h, as indicated by SO4 and Met, respectively. CD44 was immunoprecipitated with IM7-Sepharose, electrophoresed in duplicate on 7.5% SDS PAGE under nonreducing conditions, then transferred to PVDF membrane. One half of the blot was β-eliminated by treatment with 0.1 M NaOH at 45 C for 16 h. Control andnaoh-treatedpvdfmembranesweresubsequentlyexposedtox-ray film. Molecular mass markers are indicated on the left in kda. β-elimination from TNF-α-stimulated cells. The low level of sulfation present on CD44 in unstimulated SR91 cells resulted in increased variability but implied that the majority of CD44 sulfationonunstimulatedsr91cells(59 95%)occurredon O-linked posttranslational modifications: either O-linked glycans and/or GAGs. The results from TNF-α-stimulated SR91 cells indicated that approximately 75% of the sulfation of CD44 could be removed by β-elimination, but not all the sulfation was removed. Overall, this indicated that the majority (approximately 75%) of CD44 sulfation in both unstimulated and TNF-α-stimulated SR91 cells is attributable to O-linked posttranslational modifications. TNF-α induces the sulfation of N-linked glycans on CD44 To establish the nature of the remaining sulfation on CD44 from TNF-α-stimulated SR91 cells, we assessed whether CD44 was sulfated on N-glycans. Overnight digestion of immunoprecipitated and eluted CD44 with peptide:n-glycosidase F (PNGase F) reduced the apparent molecular mass of CD44 and decreased the sulfation of CD44 (Figure 3). CD44 sulfation was decreased by 29 ± 13% (N = 8) indicating that ~30% of CD44 sulfation was incorporated into N-glycans after TNF-α stimulation. The low level of sulfation on CD44 in unstimulated SR91 cells made it difficult to obtain an accurate value for the percentage loss of sulfation in this case, but from the average of four experiments there was no significant loss of [ 35 S]sulfate. That is, the average percentage loss was estimated to be less than 10% (data not shown). This suggests that TNF-α increases the amount of N-linked sulfation present on CD44 from less than 10% to ~30%. When CD44 from TNF-α-stimulated SR91 cells was treated with PNGase F and β-elimination, the sulfate label was virtually eliminated (Figure 3), indicating that Fig. 3. Combined PNGase F treatment and β-elimination of CD44 immunoprecipitated from TNF-α-stimulated SR91 cells abolishes the sulfation of CD44. Cells were labeled with [ 35 S]sulfate or [ 35 S]methioninecysteine for 24 h, as indicated by SO4 and Met, respectively. CD44 was immunoprecipitated with IM7-Sepharose, eluted, acetone precipitated, treated for 16 h in the presence or absence of 500 U of PNGase F, and then subjected to 7.5% SDS PAGE and transferred to PVDF membrane (see Materials and methods).onehalfoftheblotwasβ-eliminated by treatment with 0.1 M NaOH at 45 C for 16 h. Control and NaOH-treated PVDF membranes were subsequently exposed to X-ray film (upper panel). The relative amount of sulfate present per unit CD44 was calculated by comparing [ 35 S]sulfate incorporation with CD44 levels assessed by immunoblotting the membrane with anti-cd44 mab, 3G12, before and after PNGase F treatment (lower panel) or with [ 35 S]methionine-cysteine incorporation into CD44 after β-elimination. The molecular mass markers are indicated on the left in kda. CD44 sulfation on TNF-α-stimulated cells can be accounted for by the sulfation of N- and O-linked posttranslational modifications. In this set of experiments, β-elimination released 70 ± 9% (N =5)of[ 35 S]sulfate from TNF-α-stimulated SR91 cells, and after PNGase F digestion, the release of sulfate was 91 ± 6% (N = 8). This indicates that on TNF-α stimulation, approximately 30% of CD44 sulfation is attributable to N-linked glycans and approximately 70% to O-linked posttranslational modifications: either O-linked glycans and/or GAGs. CD44H is modified by CS, and TNF-α reduces the percentage of CD44 sulfation due to CS To identify the type of sulfated polysaccharides that were β-eliminated from CD44, CD44 was digested with various glycosidases, and the remaining level of sulfation was assessed. Keratanase or heparinase did not affect CD44 sulfation on unstimulated or TNF-α-stimulated SR91 cells (data not shown). Treatment of CD44 immunoprecipitates with chondroitin ABC lyase reduced the sulfation level of the 85-kDa form of CD44 (CD44H) from both unstimulated and TNF-α-stimulated SR91 cells (Figure 4). Chondroitin ABC lyase treatment of CD44 from unstimulated SR91 cells resulted in a decrease in [ 35 S]sulfate of 57 ± 16% (N = 14), whereas in TNF-α-stimulated 615

M. Delcommenne et al. Fig. 4. Chondroitin ABC lyase treatment of CD44 immunoprecipitates after [ 35 S]sulfate labeling of unstimulated and TNF-α-stimulated SR91 cells. Unstimulated and TNF-α-stimulated (10 ng/ml) SR91 cells were labeled with [ 35 S]sulfate for 24 h. CD44 was immunoprecipitated from SR91 cells with IM7-Sepharose and was treated in duplicate for 16 h in the presence or absence of 2 mu chondroitin ABC lyase (Chabc lyase; see Materials and methods). Immunoprecipitates were subjected to 7.5% SDS PAGE under nonreducing conditions and transferred to PVDF membrane. The amount of [ 35 S]sulfate present on CD44 was determined by autoradiography (upper panels). The amount of CD44 present in the immunoprecipitate was estimated by immunoblotting the membranes with the anti-cd44 mab 3G12 (lower panels), and relative values were determined by densitometric scanning. Molecular mass markers are indicated on the left in kda. SR91 cells, the decrease in sulfation was 33 ± 14% (N = 14). This shows that TNF-α treatment of SR91 cells results in a decrease in the percentage of sulfate due to CS, which conversely indicates an increase in other forms of sulfation. It should be noted that both the level of CD44 and its sulfation are increased in TNF-α-treated SR91 cells. There is approximately 2.5 ± 0.8 (N = 23) times more [ 35 S]sulfate incorporated into equivalent amounts of CD44 from TNF-α-stimulated SR91 cells than from unstimulated SR91 cells. Assessment of the increase in sulfation per unit CD44 after treatment of chondroitin ABC lyase indicated a 3.6 ± 1.5 (N = 14) fold increase in non-cs, whereas there was no significant increase in sulfation that was attributable to CS, 1.4- ± 1.1-fold (N = 14). This suggests that there is no significant increase in the level of CS per unit CD44 on TNF-α stimulation, but there is an increase in other forms of CD44 sulfation. It also suggests that the two- to threefold increase in CD44 sulfation on TNF-α treatment of SR91 cells is largely due to forms of CD44 sulfation other than CS, such as the sulfation of O- and N-linked glycans. Because it is always difficult to verify that the chondroitin ABC lyase has removed all the CS side chains, we used a second approach to examine the contribution of CD44 associated GAGs to the overall sulfation level of CD44. p-nitrophenyl β-d-xylopyranoside (xyloside), a competitive substrate for the addition of xylose to O-linked proteoglycans, was added to the SR91 cells to inhibit the addition of GAGs. The treatment of unstimulated SR91 cells with xyloside significantly reduced the incorporation of sulfation into CD44 by 85 ± 7% (N =5), see Figure 5. This percentage is higher than with chondroitin ABC lyase alone (~57 ± 16%), suggesting that chondroitin ABC lyase was not removing all the CS from CD44 or that other GAGs are present on CD44. Because we found no decrease in sulfation after keratanase or heparinase treatment, we have no evidence to support the latter possibility. Xyloside treatment of TNF-α-stimulated SR91 cells reduced CD44 616 Fig. 5. Effect of xyloside and chondroitin ABC lyase treatment on the sulfation of CD44 in SR91 cells. Unstimulated and TNF-α-stimulated (10 ng/ml) SR91 cells were labeled with [ 35 S]sulfate for 24 h in the presence or absence of 2 mm xyloside. CD44 was immunoprecipitated with IM7-Sepharose; each sample wasdividedintwoandincubatedfor16hat37 C in the presence or absence of 2 mu of chondroitin ABC lyase (Chabc lyase). Immunoprecipitates were subject to 7.5% SDS PAGE under nonreducing conditions and transferred to PVDF membrane. The amount of [ 35 S]sulfate present in CD44 immunoprecipitates was determined by autoradiography (upper panel). The amount of CD44 present in the immunoprecipitates was assessed by densitometry after immunoblot analysis using the anti-cd44 mab, IM7 (lower panel). Molecular mass markers are indicated on the left in kda. sulfation to a lesser extent, to 46 ± 12% (N = 5). Once again, this is a higher percentage than that reported by chondroitin ABC lyase treatment (33 ± 14%), raising the possibility that the values obtained using chondroitin ABC lyase may be an underestimate. However, the trend is the same in both cases, in unstimulated SR91 cells the majority of sulfation is contributed by CS (57 85%), whereas after TNF-α stimulation this percentage is significantly reduced (33 46%). This further supports the premise that TNF-α stimulates an increase in forms of non-gag sulfation on CD44. In these experiments, the overall increase in sulfation of CD44 in response to TNF-α was 1.8 ± 0.7 (N =5),whereas there was no significant increase in GAG sulfation, 1.0 ± 0.6 (N = 5), but there was a significant increase in the amount of non-gag sulfation of 7.0- ± 2.0-fold (N = 5). Once again, this supports the notion that TNF-α is increasing the amount of non-gag sulfation of CD44 several fold, which in turn is decreasing the percentage of CD44 sulfation due to CS. We have already established that the sulfation of N-linked glycans present on CD44 account for approximately 30% of the total sulfation levels of CD44 from TNF-α-stimulated SR91 cells. If GAG sulfation accounts for approximately 46% and O-linked carbohydrate modifications account for approximately 75%, this suggests that approximately 30% may be attributable to the sulfation of O-linked glycans. TNF-α increases the presence of sulfated O-linked glycans on CD44 To directly assess the amount of sulfation due to O-linked glycosylation, the SR91 cells were treated with the O-linked glycosylation inhibitor, benzyl 2-acetamido-2-deoxy-α-D-

TNF-α induces changes to the carbohydrate sulfation of CD44 galactopyranoside (BG), in the presence and absence of TNFα. Initial results indicated no significant decrease in the overall sulfation of CD44, so the experiments were repeated in the presence of 2 mm xyloside to prevent GAG addition to the O-linked sites. Results are shown in Figure 6 and demonstrate that treatment with BG resulted in a slight decrease in apparent molecular mass of CD44. In the presence of xyloside, treatment with BG resulted in an additional decrease in the incorporation of [ 35 S]sulfateintoCD44isolatedfromTNF-α-treated cells of 30 ± 19% (N = 4). Once again, the low level of incorporation of [ 35 S]sulfateintoCD44fromunstimulatedSR91cellsmade it difficult to assess the percentage change in sulfation, but from four experiments, no significant loss in sulfation was observed after BG treatment of unstimulated cells. This provides direct evidence for the presence of sulfated O-linked glycans on CD44 isolated from TNF-α-stimulated cells and provides data to support the notion that TNF-α enhances the sulfation of O-linked glycans, as well as the sulfation of N-linked glycans, on CD44 in SR91 cells. Considered together, the data indicate that CD44 is primarily sulfated on CS in unstimulated SR91 cells and that TNF-α increases the sulfation of both O- and N-linked glycans on CD44 and thus decreases the contribution of CS to the overall sulfation of CD44. TNF-α induces the expression of 6-sulfo N-acetyl lactosamine/ Lewis x on N- and O-linked glycans of CD44 To further identify the carbohydrate determinants that are sulfated on CD44 in SR91 cells, we first tested the reactivity of a battery of mouse mabs that recognize different sulfated terminal sugars by fluorescence-activated cell sorter (FACS) analysis. MECA-79, a mab that binds 6-sulfo N-acetyl lactosamine (LacNAc) and L-selectin ligands (Bruehl et al., 2000) and has recently been shown to recognize a 6-sulfo LacNAc determinant on extended core 1 O-glycans (Yeh et al., 2001), did not bind to SR91 cells (data not shown). Likewise, mab G72 (anti-6-sulfo sialyl LacNAc/Lewis x), mab G270 16 (anti-6,6 -disulfo sialyl LacNAc/Lewis x), mab L4L4-8 (anti- 6,6 -disulfo LacNAc), and mab SU59 (anti-3 -sulfo Lewis x) showed no significant reactivity on unstimulated or TNF-αstimulated SR91 cells (data not shown). In contrast, the mab 2F3, which requires fucose for reactivity with sialyl Lewis x and sialyl 6-sulfo Lewis x, was strongly positive for both unstimulated and TNF-α-stimulated SR91 cells (Figure 7). The mab AG107, which recognizes 6-sulfo LacNAc/Lewis x, did not bind to SR91 cells. However, if these cells were pretreated with neuraminidase to remove terminal sialic acid residues, then AG107 bound to a percentage of TNF-αinduced SR91 cells but did not bind to unstimulated SR91 cells (Figure 7). In each case, 2F3 reactivity was lost after neuraminidase treatment, verifying the loss of sialic acid. The AG107 epitope was always induced in TNF-α-stimulated SR91 cells, becoming evident after neuraminidase treatment, but the percentage of strongly positive cells varied between experiments. An average representation is shown in Figure 7. Treatment of SR91 cells with sodium chlorate, an inhibitor of sulfation, did not affect the binding of mab 2F3 (data not shown) but did abolish binding of AG107 (Figure 7), indicating that 2F3 was recognizing the sulfate-independent epitope sialyl Lewis x on SR91 cells, whereas the AG107 epitope required sulfation for reactivity. This demonstrates that TNF-α can induce the expression of 6-sulfo N-acetyl glucosamine (6-sulfo Fig. 6. Effect of the O glycosylation inhibitor BG on xyloside-treated unstimulated and TNF-α-stimulated SR91 cells. Unstimulated and TNF-α-stimulated (10 ng/ml) SR91 cells were labeled with [ 35 S]sulfate for 24 h in the presence of 2 mm xyloside and the presence or absence of 2 mm BG. CD44 was immunoprecipitated with IM7-Sepharose and subjected to 7.5% SDS PAGE under nonreducing conditions and transferred to PVDF membrane. The amount of [ 35 S]sulfate present in CD44 immunoprecipitates was determined by autoradiography (upper panel). The amount of CD44 present in the immunoprecipitates was assessed by densitometry after immunoblot analysis using the anti-cd44 mab, IM7 (lower panel). Molecular mass markers are indicated on the left in kda. Fig. 7. Flow cytometric analysis of the expression of sulfated or sialylated LacNAc/LewisxonSR91cells.SR91cellsculturedfor24hat10 6 cells/ml in thepresenceorabsenceoftnf-α (10 ng/ml) and sodium chlorate (50 mm) were treated as indicated at the left. Both untreated and neuraminidase treated (0.1 U/ml for 2 h at 37 C) cells were labeled with mab 2F3, anti-sialyl Lewis x; mab AG107, anti-6-sulfo LacNAc/Lewis x; or anti-cd44 mab, IM7. Antibody binding was detected by flow cytometry using a fluoresceinated secondary antibody (see Materials and methods). The negative control for mabs 2F3 and AG107 was an irrelevant IgM with labeled secondary antibody, whereas the negative control for IM7 was secondary antibody alone. The y-axis is cell number and x-axis fluorescence intensity (log scale). 617

M. Delcommenne et al. GlcNAc), which is integrated into a larger glycan structure involving either lactosamine or a Lewis x like epitope. To determine if the 6-sulfo LacNAc/Lewis x determinant induced by TNF-α was present on CD44, immunoprecipitates of cell surface CD44, isolated from unstimulated and TNF-αstimulated SR91 cells that had been incubated in the presence or absence of 50 mm sodium chlorate were treated with neuraminidase, subjected to SDS PAGE, transferred to polyvinylidene difluoride (PVDF) membrane and subsequently probed with mab AG107. Results showed that this epitope was not detected on CD44 from unstimulated SR91 cells but appeared on TNF-α stimulation and was abrogated by treatment of the cells with sodium chlorate (Figure 8). To determine if the 6-sulfo LacNAc/Lewis x epitope was present on N- or O-linked CD44 carbohydrate, western blots were performed with mab AG107 after treatment of CD44 immunoprecipitates with PNGase F. Figure 8 demonstrates that PNGase F treatment of CD44 immunoprecipitates resulted in a substantial decrease in antibody reactivity (70 ± 10%, N = 3). This indicates that the 6-sulfo LacNAc/Lewis x determinant is present on CD44 and is expressedtoagreaterextentonn-linkedglycansthanono-linked glycans. Discussion Here we have established that CD44 is a major sulfated membrane protein in the myeloid cell line SR91 and that this sulfation occurs on the carbohydrate moieties of CD44. We have shown that on unstimulated SR91 cells, CD44 sulfation was primarily due to the addition of CS, whereas on TNF-α stimulation, CD44 sulfation was increased two- to threefold and was due not only to CS but also to the sulfation of both N- and O-linked glycans. PNGase F digestion and β-elimination of TNF-α-stimulated SR91 cells indicated that approximately 30% of CD44 sulfation was associated with N-linked glycans and ~75% with O-linked glycans and GAGs. Xyloside treatment of TNF-α-treated cells indicated that ~46% of CD44 sulfation is due to GAGs, and BG treatment indicated that ~30% of non-gag sulfation (i.e., approximately 15% of total CD44 sulfation) is due to the sulfation of O-linked glycans. This is illustrated diagrammatically in Figure 9. Although there was a significant decrease in the percentage of sulfation from CS (due to significant increases in N and O-linked glycan sulfation), on TNF-α stimulation the amount of sulfate incorporated as CS per unit CD44 did not change significantly (sulfation levels due to GAG sulfation increased 1.4- ± 1.1-fold). This suggests that TNF-α may primarily affect the glycan sulfation of CD44 without substantially affecting GAG addition or the degree of GAG sulfation. CD44 has six N-linked glycosylation sites that are located primarily at the amino-terminal region that binds HA. In contrast, the majority of O-linked glycosylation and GAG addition sites are located in the membrane proximal region of CD44. Human CD44 contains four serine-glycine motifs in the membrane proximal region of all isoforms, which are potential GAG addition sites, and additional sites are present in the alternatively spliced exons, V3 and V10. Analysis of the sites for GAG attachment in CD44-Rg proteins produced in COS cells indicated that only two sites are used, one in V3 that supports heparan sulfate and CS attachment and the second in exon 5 that supports only CS attachment (Greenfield et al., 1999).In somecells,cd44canalsobemodifiedbykeratansulfate (Takahashi et al., 1996). Here wefindthatthe standard85-kda form of CD44 is modified by CS. Chondroitin or heparan sulfate modified forms of CD44 typically result in the expression of heterogenous higher-molecular-mass species on SDS PAGE of up to 200 kda (Jalkanen et al., 1988). Therefore, it was surprising to find that a significant amount of the 85-kDa form of CD44 (CD44H) was modified by CS, yet the molecular weight did not increase significantly. Only a small amount of highermolecular-mass material (90 180 kda) was observed in CD44 immunoprecipitates after [ 35 S]sulfate labeling (Figures 4 and 5). Removal of CS by chondroitin ABC lyase or xyloside did not result in a significant change in apparent molecular mass of 85 kda, implying that the CS side chains must be relatively small in length. The biological significance of the addition of Fig. 8. TNF-α induced expression of a 6-sulfo LacNAc/Lewis x epitope on CD44 from neuraminidase treated SR91 cells. Cells (10 7 SR91 cells) were cultured for 24 h at 10 6 cells/ml in the presence or absence of TNF-α (10 ng/ml) or 50 mm sodium chlorate. Cells were treated for 2 h with 0.1 U/ml neuraminidase, and cell surface CD44 was immunoprecipitated with mab IM7 and Protein G Sepharose. CD44 was eluted from the beads, precipitated with acetone, and incubated at 37 C for 16 h with or without 500 U PNGase F (see Materials and methods). Samples were run on 7.5% SDS PAGE and transferred to PVDF membrane. The membrane was incubated first with mab AG107 and HRP-conjugated anti-mouse IgM antiserum and antigens were visualized by ECL (upper panel). After stripping, the membrane was reprobed with anti-cd44 mab, 3G12 (lower panel). 618 Fig. 9. Diagrammatic representation of distribution of sulfate on CD44 from unstimulated andtnf-α-stimulated SR91 cells. Approximate percentages and values were taken from experiments using xyloside, PNGaseF, and BG. The average increase in CD44 sulfation is indicated on the left. N and O represent the sulfation attributed to N- and O-linked glycans, respectively.

TNF-α induces changes to the carbohydrate sulfation of CD44 relatively small CS chains to the 85-kDa form of CD44 remains to be established. We have been able to identify the different types and relative amounts of CD44 sulfation occurring on TNF-α stimulation. However, there was a relatively large variability of [ 35 S]sulfate incorporation into CD44 in response to TNF-α (2.5- ± 0.8-fold increase), as well as some variation in the percentage of sulfate due to GAGs and N- and O-linked glycans. This resulted in quite a range of percentages: 20 40% for N-linked glycan sulfation, 5 25% for O-linked glycan sulfation, and 34 58% for GAG sulfation. This variation may reflect a normal cellular variation in sulfation, glycosylation, and GAG synthesis and may also reflect variations in culture conditions. For example, the glucose levels present in the media have been reported to affect CD44 glycosylation (Zheng et al., 1997) and ammonia, which can be released in the media from glutamine degradation, can down-regulate sialylation (Yang and Butler, 2000). We have previously shown that CD44 is sulfated in SR91 cells and in normal peripheral blood monocytes, and overall CD44 sulfation is increased on exposure to TNF-α in both (Maiti et al., 1998; Brown et al., 2001). Here we show that CS and O- and N-linked glycans contribute to the overall sulfation of CD44 in SR91 cells and TNF-α can affect the amount and percentage of sulfation due to CS and N- and O-glycan sulfation in these cells. We have also found that CD44 is a major sulfated cell surface protein in another myeloid cell line, KG1a. Here, sulfation of CD44 was atttributed primarily to sulfation of N- and O-linked glycans (~30 and ~70% respectively, data not shown). In peripheral blood monocytes, preliminary data indicate the presence of CS and sulfated glycans on CD44 (Brown and Johnson, unpublished data). This indicates that the carbohydrate sulfation of CD44 is not restricted to SR91 cells but also occurs in myeloid cells. We had previously correlated the overall sulfation of CD44 with its ability to bind HA. However, here we have established that multiple sulfated moieties are present on CD44, indicating that the initial correlation was somewhat simplistic. It will now require further analysis to dissect the role and contribution of each sulfated moiety toward HA binding. Further analysis using sulfated carbohydrate specific mabs indicated that TNF-α-induced the expression of a 6-sulfo LacNAc or Lewis x determinant on the N-linked and (to a lesser extent) on the O-linked glycans of CD44. However, at present we do not know how much this type of sulfation contributes to the overall sulfation of CD44 glycans. In addition, because no antibodies were available that could detect 6-sulfo galactose or galactosamine, we cannot rule out the possibility that this type of sulfation could also be occurring. However, we have established the presence of 6-sulfo GlcNAc on both N- and O-linked CD44 glycans, and this raises the intriguing possibility that TNF-α may induce the expression or activation of a GlcNAc 6-O-sulfotransferase (GlcNAc6ST). However, TNF-α may also induce changes to CD44 glycosylation, which in turn may affect their sulfation, as changes in glycosylation have been reported to occur with CD44 upon cytokine treatment (Levesque and Haynes, 1999; Cichy and Pure, 2000). There are now several glycosyl sulfotransferases identified, and most have very restricted substrate specificities (Hemmerich and Rosen, 2000). AG107 reactivity of isolated CD44 indicates that the 6-sulfo LacNAc/Lewis x epitope is present on N-linked glycans and, to a lesser extent, on O-linked glycans, implicating a GlcNAc6ST that can act on both N-linked and O-linked glycans. Putative candidates include the broadly expressed GlcNAc6ST-1 (Uchimura et al., 1998) also known as CHST-2 or GST-2, and GlcNAc6ST-4 (Uchimura et al., 2000) also known as C6ST-2 (Kitagawa et al., 2000) or GST-5 (Bhakta et al., 2000), which can transfer sulfate to both GlcNAc-Man and core 2 structures (Uchimura et al., 2000), and can also sulfate 6-N-acetyl galactosamine to low levels (Kitagawa et al., 2000). The nomenclature for the sulfotransferases is as described in Fukuda et al. (2001). The MECA-79 mab recognizes sulfated determinants present on HEV and has recently been shown to recognize the terminal epitope Galβ1-4(sulfo-6)GlcNAcβ1-3Galβ1-3GalNAc present on core 1 but not core 2 mucin type O-glycans or N-glycans (Yeh et al., 2001). TNF-α-stimulated SR91 cells did not bind the MECA-79 mab (data not shown), indicating that the sulfated determinant present on CD44 is distinct from that on HEV, which is recognized by MECA-79. AG107 reactivity to TNF-α-stimulated SR91 cells required the prior removal of sialic acid residues, implying that sialylation is occurring on the 6-sulfo LacNAc/Lewis x structures. However, the mab G72, which recognizes α2 3 linked monosialylated 6-sulfo LacNAc/Lewis x, did not bind to TNF-α-stimulated SR91 cells. This suggests an alternative linkage for the terminal sialic acid or polysialylation, or alternatively, AG107 binding may be prevented by interference from neighboring sialic acid residues. Here we have demonstrated the induction of a 6-sulfo LacNAc or Lewis x epitope on CD44 on SR91 cells in response to TNF-α, and preliminary data suggest that the epitope is also induced on peripheral blood monocytes (Brown and Johnson, unpublished data). Sulfated sialyl Lewis x epitopes have been implicated in the initial adhesion or rolling of leukocytes on endothelial cells (Rosen, 1999; Hemmerich and Rosen, 2000), and sulfated CD44 has been implicated in regulating its adhesion to HA and to an endothelial cell line (Maiti et al., 1998; Brown et al., 2001). It is therefore possible that the induction of sulfated carbohydrates on CD44 augments leukocyte adhesion via either CD44 or a selectin ligand interaction. Now the sulfated moieties of CD44 have been identified, their role in leukocyte adhesion can be further explored. SR91 cells express surface markers characteristic of myeloid cells and their progenitors. Early hematopoietic progenitor cells adhere to the bone marrow stroma, and CD44 has been implicated in this process (Legras et al., 1997). In particular, CS-modified CD44 and α 4 β 1 cooperate to mediate progenitor cell adhesion to fibronectin, an ECM component of the bone marrow stroma (Verfaillie et al., 1994). Anti-CD44 antibodies can inhibit lympho- and myelopoiesis (Miyake et al., 1990; Moll et al., 1998), and CD44 knockout mice show anomalous myeloid progenitor distribution (Schmits et al., 1997). It has also been shown that sulfated fucans can mobilize hematopoietic progenitor cells from the bone marrow (Sweeney et al., 2000). Thus, it is possible that remodeling of CD44 by sulfated carbohydrates contributes to the regulation of myeloid cell adhesion during both development and inflammation. 619

M. Delcommenne et al. Materials and methods Cell lines SR91 is a myeloid cell line derived from a human leukemia that is positive for the myeloid-specific marker CD33 and negative for the T cell markers CD2 and CD7. It was obtained from H. Klingemann (see Klingemann et al., 1994). Cells were maintained in Roswell Park Memorial Institute (RPMI) medium supplemented with 2 mm glutamine, 1 mm sodium pyruvate and 10% fetal calf serum (Invitrogen, Burlington, ON or HyClone, Logan, UT). Antibodies and reagents The rat mab IM7 (IgG2b, Trowbridge et al., 1982), reactive against mouse and human CD44, was from J. Lesley and B. Hyman. The mouse mab 3G12, directed against human CD44 (Dougherty et al., 1994), was from G. Dougherty. The rat IgM mab MECA-79 (Streeter et al., 1988), was from S. Hemmerich. The following mouse IgM mabs were also used: L4L4-8, anti-6,6 -disulfo LacNAc; AG107, anti-6-sulfo LacNAc/Lewis x; G72, anti-sialyl 6-sulfo LacNAc/Lewis x (Uchimura et al., 1998); G270-16, anti-sialyl 6,6 -disulfo LacNAc/Lewis x; 2F3, anti-sialyl Lewis x, anti-sialyl 6-sulfo Lewis x, SU59, anti 3 -sulfo Lewis x (Ohmori et al., 1993; Mitsuoka et al., 1998; Izawa et al., 2000). Fluoresceinated- or horseradish peroxidase (HRP)-conjugated anti-mouse IgG, anti-rat IgG, anti-mouse IgM, and streptavidin were from Jackson ImmunoResearch (West Grove, PA). Cell surface biotinylation and neuraminidase treatment Cell surface proteins were labeled using 0.5 mg NHS-LC-biotin (Pierce, Rockford, IL) for 3 10 7 cells in 1 ml phosphate buffered saline (PBS) 1% glucose for 30 min 1 h at 4 C. Excess biotin was quenched with 1 mm lysine in PBS, and cells were washed three times in this buffer. Cells were lysed and processed for immunoprecipitations as described later. Alternatively, 10 7 cells/ml were incubated in the presence of 0.1 U/ml Vibrio cholerae neuraminidase (Roche Diagnostics, Laval, Quebec) in buffered medium (RPMI 1640/PBS, 1:1, ph 6.8) for 1 2 h at 37 C, washed, and then analyzed for antibody binding by flow cytometry or incubated at 4 C for2hwith1mlcd44mab IM7 tissue culture supernatant prior to cell lysis and CD44 immunoprecipitation with Protein G Sepharose (Amersham Pharmacia Biotech, Baie d Urfé, Quebec) for 2 h at 4 C. Biosynthetic labeling of SR91 with [ 35 S]sodium sulfate or [ 35 S]methionine-cysteine Cells (1 10 6 /ml) were incubated (20 24 h) in sulfate-free RPMI medium (cell culture facility, University of California SanFrancisco)supplementedwith2mMglutamine,1mM sodium pyruvate, and 10% fetal calf serum in the presence of 100 µci/ml [ 35 S]Na 2 SO 4 (~1000 Ci/mmol, ICN Biomedicals, St Laurent, Quebec) or 40 µci/ml [ 35 S]methionine-cysteine protein labeling mix (1175 Ci/mmol; Easy Tag Express, NEN DuPont Canada, Markham, Ontario) in the culture medium containing 90% methionine-cysteine free RPMI 1640 medium (ICN) and 10% standard RPMI 1640 medium. In some cases, SR91 cells were also incubated for 20 24 h with 10 ng/ml TNF-α (R&D Systems, Minneapolis, MN) in the presence or absence of 50 mm sodium chlorate or 2 mm xyloside (Sigma) or both 2 mm xyloside and the O-glycosylation inhibitor, 2 mm BG (Sigma). 620 Immunoprecipitation Labeled cells were rinsed three times in cold PBS before lysis in radio-immunoprecipitation assay lysis buffer containing the protease inhibitors (purchased from Sigma) phenylmethylsulfonyl fluoride (1 mm), leupeptin (1 µg/ml), aprotinin (1 µg/ml), and pepstatin (1 µg/ml). After centrifugation to remove particulate material, the resulting supernatants were precleared with Sepharose CL-4B beads (Sigma). The lysates were incubated withanti-cd44mabim7coupledtocyanogenbromide activated Sepharose beads (Amersham Pharmacia; 4 mg/ml IM7-Sepharose, 50 µl/10 7 cells),for2hat4 C. UltraLinkimmobilized Neutravidin (Pierce) was used to precipitate biotinylated cell surface proteins. The beads were washed five times with lysis buffer, and immunoprecipitated material was eluted by boiling beads in sample buffer, resolved directly on 7.5% SDS PAGE, transferred to PVDF membrane (Millipore Canada, Mississauga, Ontario) and exposed to Kodak BioMax MR film (Interscience, Markham, Ontario) for approximately 4 days. Films were scanned and band intensities were determined with Alphaimager software (Alpha Innotech, San Leandro, CA). Relative sulfation levels were expressed as the ratio of sulfated CD44 to immunoprecipitated CD44, which was determined by western blot analysis or in some cases by autoradiography of [ 35 S]methionine-cysteine-labeled CD44. Enzymatic treatment of immunoprecipitated CD44 CD44 from [ 35 S]sulfate-labeled cells was immunoprecipitated, eluted from beads by boiling in sample buffer, precipitated with six volumes of acetone at 20 C, spun for 30 min at 16,000 g, washed twice with acetone, and then resuspended in appropriate glycosidase buffer. CD44 immunoprecipitated from ~2 10 6 cells was incubated for 16 h at 37 Cin20µl50mM sodium phosphate buffer, ph 7.5, containing 0.5% SDS, 1% NP40, and 1% β-mercaptoethanol, with 500 U of PNGase F (New England Biolabs, Mississauga, Ontario) to remove N-linked glycans. Alternatively, CD44 immunoprecipitated from ~2 10 6 cells was incubated with 2 mu Proteus vulgaris chondroitin ABC lyase (Sigma) in 20 µl 50 mm Tris HCl, 50 mm sodium acetate, ph 8.0, for 16 h at 37 C toremovecswith2u/mlof keratanase from Pseudomonas sp. (Seikagaku America, Falmouth, MA) for 17 h at 37 C in50mmtris HCl,pH7.4, to remove keratan sulfate or with 0.4 U/ml of heparinase I or III from Flavobacterium heparinum (Sigma) in 25 mm sodium acetate, 100 mm NaCl, 5 mm calcium acetate, ph 7.0, for 16 h at 37 C to remove heparan sulfate. Samples were resolved on 7.5% SDS PAGE, transferred to PVDF membrane, and exposed to X-ray film. β-elimination on PVDF membrane [ 35 S]sulfate-labeled CD44 was immunoprecipitated, separated by SDS PAGE, and transferred to PVDF membrane. CD44 associated GAGs and O-linked sugars were subjected to β-elimination by incubating the PVDF membrane in 0.1 M NaOH for 16 h at 45 C (Duket al., 1997). Membranes were rinsed twice in water, dried, and exposed to X-ray film. Western blot The dried PVDF membranes were incubated with 3G12, IM7, or AG107 hybridoma supernatants diluted 1:10 in 150 mm NaCl, 10 mm Tris, ph 7.5, 0.05% Tween 20 (TBST)

TNF-α induces changes to the carbohydrate sulfation of CD44 containing 5% skim milk powder for 3G12 and IM7 and 3% bovine serum albumin for AG107. The blots were washed with TBST and incubated with the appropriate HRP-conjugated secondary antibody at 1:10,000 and washed again several times; proteins were detected by enhanced chemiluminescence (ECL; Amersham Pharmacia). In some cases, antibodies were stripped from membranes by incubation in 1% NP40, 0.1 M glycine, ph 3, for 1 h. Membranes were rinsed in TBST, dried, and stored until reprobed. When biotinylated cell surface antigens were used for immunoprecipitation, the blots were blocked for 1 h in 3% bovine serum albumin in TBST, washed with TBST, and incubated with 1:10,000 HRP-conjugated streptavidin (Jackson) for 30 min. Membranes were washed in TBST and the proteins detected by ECL. Immunofluorescence staining and flow cytometry analysis Cells were washed in ice-cold PBS containing 2% fetal calf serum and 0.1% NaN 3 (FACS buffer), and 3 10 5 cells were incubated on ice for 30 min with 50 µl of 1:16 dilution of all hybridoma supernatants except IM7 and 3G12 supernatants, which were used undiluted, and the MECA-79 mab, which was used at 10 µg/ml. The cells were washed with FACS buffer, incubated for 30 min with 1:100 fluoresceinated secondary antibody, and washed with FACS buffer. Labeled cells were resuspended in FACS buffer with propidium iodide (2 µg/ml), and flow cytometry analysis on gated live cells was performed on a FACScan (Becton Dickinson, Mississauga, Ontario) using CellQuest software. Acknowledgments We thank Drs. J. Lesley, B. Hyman, G. Dougherty, and S. Hemmerich for kindly providing CD44 mabs and Dr. H. Klingemann for the SR91 cell line. We thank K. Brown and Dr. R. Lian for critical reading and help with the manuscript. This work was supported by funds from the Canadian Institutes of Health Research (CIHR). Abbreviations BG, benzyl 2-acetamido-2-deoxy-α-D-galactopyranoside; CS, chondroitin sulfate; ECL, enhanced chemiluminescence; ECM, extracellular matrix, FACS, fluorescence-activated cell sorter; GAG, glycosaminoglycan; HA, hyaluronan; HEV, high endothelial venules; HRP, horseradish peroxidase; mab, monoclonal antibody; PBS, phosphate buffered saline; PNGase F, peptide:n-glycosidase F; PVDF, polyvinylidene difluoride; RPMI, Roswell Park Memorial Institute; SDS PAGE, sodium dodecyl sulfate polyacrylamide gel electrophoresis; TBST, Tris buffered saline + Tween; xyloside, p-nitrophenyl β-d-xylopyranoside. References Baumheter, S., Singer, M.S., Henzel, W., Hemmerich, S., Renz, M., Rosen, S.D., and Lasky, L.A. (1993) Binding of L-selectin to the vascular sialomucin CD34. Science, 262, 436 438. Bennett, K.L., Jackson, D.G., Simon, J.C., Tanczos, E., Peach, R., Modrell, B., Stamenkovic, I., Plowman, G., and Aruffo, A. (1995) CD44 isoforms containing exon v3 are responsible for the presentation of heparin-binding growth factor. J. Cell Biol., 128, 687 698. Bhakta,S.,Bartes,A.,Bowman,K.G.,Kao,W.M.,Polsky,I.,Lee,J.K., Cook, B.N., Bruehl, R.E., Rosen, S.D., Bertozzi, C.R., and Hemmerich, S. (2000) Sulfation of N-acetylglucosamine by chondroitin 6-sulfotransferase 2(GST-5).J. Biol. Chem., 275, 40226 40234. Bowman, K.G. and Bertozzi, C.R. (1999) Carbohydrate sulfotransferases: mediators of extracellular communication. Chem. Biol., 6, R9 R22. Brockhausen, I. and Kuhn, W. (1997) Role and metabolism of glycoconjugate sulfation. Trends Glycosci. Glycotech., 9, 379 398. Brown, K.L., Maiti, A., and Johnson, P. (2001) Role of sulfation in CD44- nediated hyaluronan binding induced by inflammatory mediators in human CD14+ peripheral blood monocytes. J. Immunol., 167, 5367 5374. Bruehl, R.E., Bertozzi, C.R., and Rosen, S.D. (2000) Minimal sulfated carbohydrates for recognition by L-selectin and the MECA-79 antibody. J. Biol. Chem., 275, 32642 32648. Camp, R.L., Scheynius, A., Johansson, C., and Pure, E. (1993) CD44 is neccessary for optimal contact allergic responses but is not required for normal leukocyte extravasation. J. Exp. Med., 178, 497 507. Carter, W.G. and Wayner, E.A. (1988) Characterization of the class III collagen receptor, a phosphorylated, transmembrane glycoprotein expressed in nucleated human cells. J. Biol. Chem., 263, 4193 4201. Cichy, J. and Pure, E. (2000) Oncostatin M and transforming growth factor-beta 1 induce post-translational modification and hyaluronan binding to CD44 in lung-derived epithelial tumor cells. J. Biol. Chem., 275, 18061 18069. DeGrendele, H.C., Estess, P., and Siegelman, M.H. (1997) Requirement for CD44 in activated T cell extravasation into an inflammatory site. Science, 278, 672 675. Dougherty, G.J., Cooper, D.L., Memory, J.F., and Chiu, R.K. (1994) Ligand binding specificity of alternatively spliced CD44 isoforms recognition and binding of hyaluronan by CD44R1. J. Biol. Chem., 269, 9074 9078. Duk, M., Ugorski, M., and Lisowska, E. (1997) Beta-elimination of O-glycans from glycoproteins transferred to immobilon P membranes: method and some applications. Anal. Biochem., 253, 98 102. Ehnis, T., Dieterich, M., Bauer, M., Vonlampe, B., and Schuppan, D. (1996) A chondroitin/dermatan sulfate form of CD44 is a receptor for collagen xiv (undulin). Exp. Cell Res., 229, 388 397. Faassen,A.E.,Schrager,J.A.,Klein,D.J.,Oegema,T.R.,Couchman,J.R.,and McCarthy, J.B. (1992) A cell surface chondroitin sulfate proteoglycan, immunologically related to CD44, is involved in type I collagen-mediated melanoma cell motility and invasion. J. Cell Biol., 116, 521 531. Farzan, M., Mirzabekov, T., Kolchinsky, P., Wyatt, R., Cayabyab, M., Gerard, N.P., Gerard, C., Sodroski, J., and Hyeryun, C. (1999) Tyrosine sulfation of the amino terminus of CCR5 facilitates HIV-1 entry. Cell, 96, 667 676. Farzan, M., Schnitzler, C.E., Vasilieva, N., Leung, D., Kuhn, J., Gerard, C., Gerard, N.P., and Choe, H. (2001) Sulfated tyrosines contribute to the formation of the C5a docking site of the human C5a anaphylatoxin receptor. J. Exp. Med., 193, 1059 1065. Fukuda, M., Hiraoka, N., Akama, T.O., and Fukuda, M.N. (2001) Carbohydratemodifying sulfotransferases: structure, function, and pathophysiology. J. Biol. Chem., 276, 47747 47750. Greenfield, B., Wang, W.C., Marquardt, H., Piepkorn, M., Wolff, E.A., Aruffo, A., and Bennett, K.L. (1999) Characterization of the heparan sulfate and chondroitin sulfate assembly sites in CD44. J. Biol. Chem., 274, 2511 2517. Hemmerich, S. and Rosen, S.D. (2000) Carbohydrate sulfotransferases in lymphocyte homing. Glycobiology, 10, 849 856. Hemmerich, S., Butcher, E.C., and Rosen, S.D. (1994) Sulfation-dependent recognition of high endothelial venules (HEV)-ligands by L-selectin and MECA 79,and adhesion-blocking monoclonal antibody. J. Exp. Med., 180, 2219 2226. Imai, Y., Lasky, L.A., and Rosen, S.D. (1993) Sulphation requirement for GlyCAM-1, an endothelial ligand for L-selectin.Nature, 361, 555 557. Isacke, C.M. (1994) Commentary: the role of the cytoplasmic domain in regulating CD44 function. J. Cell Sci., 107, 2353 2359. Izawa, M., Kumamoto, K., Mitsuoka, C., Kanamori, C., Kanamori, A., Ohmori, K., Ishida, H., Nakamura, S., Kurata-Miura, K., Sasaki, K., and others. (2000) Expression of sialyl 6-sulfo Lewis X is inversely correlated with conventional sialyl Lewis X expression in human colorectal cancer. Cancer Res., 60, 1410 1416. Jalkanen, S. and Jalkanen, M. (1992) Lymphocyte CD44 binds the COOH-terminal heparin-binding domain of fibronectin. J. Cell Biol., 116, 817 825. Jalkanen, S., Jalkanen, M., Bargatze, R., Tammi, M., and Butcher, E.C. (1988) Biochemical properties of glycoproteins involved in lymphocyte recognition of high endothelial venules in man. J. Immunol., 141, 1615 1623. Jalkanen, S.T., Bargatze, R.F., de los Toyos, J., and Butcher, E.C. (1987) Lymphocyte recognition of high endothelium: antibodies to distinct epitopes of an 85 95 kd glycoprotein antigen differentially inhibit 621