Glycobiology Advance Access originally published online on February 16, 2006
Glycobiology 2006 16(7):584-593; doi:10.1093/glycob/cwj090
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Glycosyltransferases involved in type 1 chain and Lewis antigen biosynthesis exhibit glycan and core chain specificity
Division of Clinical Immunology, Karolinska Institutet, Karolinska University Hospital, Huddinge, S-141 86 Stockholm, Sweden
1 To whom correspondence should be addressed; e-mail: jonas.lofling{at}ki.se
Received on October 19, 2005; revised on January 30, 2006; accepted on February 10, 2006
 |
Abstract |
|---|
 Top Abstract IntroductionResultsDiscussionMaterials and methodsConflict of interest statementAcknowledgmentsReferences |
|---|
Sialyl Lewis A (SLea), Lewis A (Lea), and Lewis B (Leb) have been studied in many different biological contexts, for example in microbial adhesion and cancer. Their biosynthesis is complex and involves ß1,3-galactosyltransferases (ß3Gal-Ts) and a combined action of
2- and/or 4-fucosyltransferases (Fuc-Ts). Further, O-glycans with different core structures have been identified, and the ability of ß3Gal-Ts and Fuc-Ts to use these as substrates has not been resolved. Therefore, to examine the in vivo specificity of enzymes involved in SLea, Lea, and Leb synthesis, we have transiently transfected CHO-K1 cells with relevant human glycosyltransferases and, on secreted reporter proteins, detected the resulting Lewis antigens on N- and O-linked glycans using western blotting and Le-specific antibodies. ß3Gal-T1, -T2, and -T5 could synthesize type 1 chains on N-linked glycans, but only ß3Gal-T5 worked on O-linked glycans. The latter enzyme could use both core 2 and core 3 precursor structures. Furthermore, the specificity of FUT5 and FUT3 in Lea and Leb synthesis was different, with FUT5 fucosylating H type 1 only on core 2, but FUT3 fucosylating H type 1 much more efficient on core 3 than on core 2. Finally, FUT1 and FUT2 were both found to direct 2-fucosylation on type 1 chains on both N- and O-linked structures. This knowledge enables us to engineer recombinant glycoproteins with glycan- and core chain-specific Lewis antigen substitution. Such tools will be important for investigations on the fine carbohydrate specificity of Leb-binding lectins, such as Helicobacter pylori adhesins and DC-SIGN, and may also prove useful as therapeutics. Key words: blood group antigens / cancer associated epitopes / fucosyltransferase / Lewis antigens / ß3-galactosyltransferase
|
Introduction |
|---|
|
TopAbstractIntroductionResultsDiscussionMaterials and methodsConflict of interest statementAcknowledgmentsReferences |
|---|
The carbohydrate epitopes Lewis A (Lea) and Lewis B (Leb) have been studied in many different biological contexts, for example in microbial adhesion (Boren et al., 1993
; Greenwell, 1997; Mitchell et al., 2002; Harrington et al., 2004) and cancer (Hakomori and Andrews, 1970; Blaszczyk et al., 1985; Orntoft, Greenwell et al., 1991; Greenwell, 1997; Lopez-Ferrer et al., 2000). Additionally, Lea has been implicated in the adhesion of sperm to egg via zona pellucida glycoprotein 3 (Kerr et al., 2004). The sialylated version of Lea, Sialyl Lewis A (SLea, also known as the CA19-9 antigen), has been shown to be a selectin ligand in some model systems (Zhang et al., 1994; Alon et al., 1995; Zhang et al., 1996). SLea can also be considered a cancer-associated antigen (Magnani et al., 1983). However, the physiological role, if any, of type 1 chain-based Lewis epitopes is still not clear.
Lea and Leb are based on the type 1 chains, which are synthesized by ß1,3-galactosyltransferases (ß3Gal-Ts). Data from enzymatic studies in vitro from H. Clausen’s group suggested that type 1 chain biosynthesis by ß3Gal-T1 was restricted to glycolipids, by ß3Gal-T2 to N-linked glycans and that type 1 chain biosynthesis by ß3Gal-T5 occurred almost exclusively on O-linked glycans (Amado et al., 1998; Amado et al., 1999). On the other hand, Zhou et al. (1999) showed that ß3Gal-T1 worked on ovalbumin, which carries N-glycans, much better than ß3Gal-T5 did. Salvini et al. (2001) found that mainly ß3Gal-T5 was responsible for elongation of N-linked glycans on carcinoembryonic antigen in CHO-cells.
In nature, O-linked glycans with different core structures have been identified (Brockhausen, 1999), and the ability of the different ß3Gal-Ts to use these as substrates has not been resolved, although ß3Gal-T5 has been claimed to be responsible for type 1 chain elongation of core 3 O-glycans (Zhou et al., 1999).
From the type 1 chain precursor, the biosynthesis of Leb and Lea has been considered (Clausen and Hakomori, 1989; Brockhausen, 1999) to proceed via FUT2 and FUT3, respectively. FUT5 has also shown 4-fucosylation activity in vitro (Weston et al., 1992; Nguyen et al., 1998; Vo et al., 1998; Dupuy et al., 1999; de Vries et al., 2001) and in vivo (Legault et al., 1995). However, somewhat conflicting results exist (de Vries et al., 1995; Xu et al., 1996), and most importantly, it is not clear on which glycans (N-, O-linked, or glycolipid) the synthesis may take place in vivo. Furthermore, FUT1 has been seen as an additional candidate for 2 fucosylation of type 1 chains in vitro and in vivo (Liu et al., 1998; Mathieu et al., 2004). To complicate matters even more, it has been demonstrated in vitro that the specificity of the fucosyltransferases differs; FUT1 and FUT5 prefer type 2 to type 1 and FUT2 and FUT3 prefer type 1 to type 2 (Oriol et al., 1986; Oriol and Mollicone, 2002). The biosynthesis of SLea proceeds via the fucosylation of its precursor, DUPAN-2 (Sia3Galß3GlcNAc), and it has been suggested that there is a competition between the synthesis of DUPAN-2, H type 1, and Lea (Greenwell, 1997; Brockhausen, 1999).
Most studies on the enzymes involved in Lea and Leb synthesis have been carried out in vitro. Very few have addressed this issue in a physiological context, such as in cultured cells. Therefore, we have transfected CHO-K1 cells with relevant human glycosyltransferases and detected the resulting Lewis antigens on secreted reporter proteins carrying N- or O-linked glycans, respectively. All three studied ß3Gal-Ts could synthesize type 1 chains on N-linked glycans, but only ß3Gal-T5 worked on O-linked glycans. The latter enzyme could use both core 2 and core 3 precursor structures. Remarkably, even though both ß3Gal-T1 and T2 produced about the same amount of type 1 chain available for direct fucosylation to Lea by 4Fuc-Ts, there was a marked difference in the conversion to Leb, irrespective of the 2Fuc-T used, with ß3Gal-T1 promoting more Leb synthesis than did ß3Gal-T2. Furthermore, the specificity of FUT5 and FUT3 in Leb synthesis was different, with FUT5 fucosylating H type 1 on core 2, but FUT3 almost only on core 3. Finally, FUT1 and FUT2 were both found to direct 2-fucosylation on type 1.
|
Results |
|---|
|
TopAbstractIntroductionResultsDiscussionMaterials and methodsConflict of interest statementAcknowledgmentsReferences |
|---|
Staining of serum albumin neoglycoproteins carrying defined Lewis antigens confirmed the specificity of antibodies used
Because there have been several studies showing cross-reactivity of antibodies claimed to be specific for Lewis antigens, we first performed western blotting control experiments. We tested mAb specificity using bovine serum albumin (BSA) or human serum albumin (HSA) neoglycoproteins carrying defined Lewis determinants (data not shown). The Lea mAbs T174 and 78FR2.3 did not cross-react with H type 1, Leb, Lex, SLex, or Ley. T218 (BG-6) reacted only with Leb, whereas 96FR2.10 stained Leb- and H type 1-carrying BSA. It has been shown to be common for anti-Leb-antibodies to cross-react also with Ley, but we did not see any binding to Ley-HSA by any of the anti-Leb mAbs used. The SLea antibody 1116-NS-19–9 was tested against Ley, SLex, and Leb conjugates of BSA and no binding was observed. No Lea-conjugate was tested. To exclude nonspecific staining of glycans on fusion proteins, we performed control transfections in each experiment in which one or several glycosyltransferase cDNAs needed for the synthesis of the respective epitopes were left out. In no case did we find any cross-reactivity of the specific mAbs used.
ß3Gal-T1, -T2, and -T5 are all able to make type 1 chain on N-linked glycans in CHO cells
Lea was found in equal amounts on 1-acid glycoprotein (AGP)/mIgG from cells expressing either ß3Gal-T1 or T2 together with FUT3, whereas the staining was much more intense on AGP/mIgG from cells transfected with ß3Gal-T5 and FUT3 cDNAs (data not shown). This staining remained even if the cells were also transfected with cDNA encoding an 2Fuc-T (Figure 1, panel A and B). Interestingly, the Leb expression pattern was different from the Lea-pattern, with the strongest staining of AGP/mIgG from cells transfected with plasmids encoding ß3Gal-T5 (Figure 1, panel C, lane ß3Gal-T 5), an intermediate staining following ß3Gal-T1 cDNA transfection (Figure 1, panel C, lane ß3Gal-T 1), and the weakest staining of AGP/mIgG from cells expressing ß3Gal-T2 (Figure 1, panel C, lane ß3Gal-T 2) in combination with FUT3 and FUT2. All western blots on fusion proteins were also probed with an anti-mouse IgG Ab. This was to establish that the difference in staining of Le-epitopes was not due to a difference in the amount of fusion protein loaded on the gel (Figure 1, panel D). The relative Leb staining intensity obtained with the different ß3Gal-Ts was the same, irrespective of the Leb mAb used (data not shown). This suggests a higher enzymatic activity of ß3Gal-T5 compared with the other two ß3Gal-Ts on precursors carried by N-glycans. Thus, ß3Gal-T1, -T2, and -T5 can all form type 1 structures on N-linked oligosaccharides, albeit with different efficacy.
|
Only ß3Gal-T5 can make type 1 on O-linked glycans in CHO cells
We tested the ability of ß3Gal-T1, -T2, and -T5 to direct biosynthesis of type 1 on core 2 and core 3 O-glycans carried by the CD43/mIgG2b reporter protein. On O-linked oligosaccharides, in contrast to on N-linked glycans, only ß3Gal-T5 and not ß3Gal-T1 or ß3Gal-T2 was shown to produce type 1, as detected by anti-Lea and Leb mAbs following co-transfection with FUT3 alone (Figure 2, panel A) or in combination with FUT2 cDNAs (Figure 2, panel B). Furthermore, ß3Gal-T5 galactosylates GlcNAc branches linked ß1,6 and ß1,3 to GalNAc, that is it supports type 1 chain elongation on both core 2 and 3 O-glycans (Figure 2, panels A and B).
|
Anti-Lea monoclonal antibodies (mAbs) exhibit core chain-dependent Lea reactivity on O-linked glycans
Following co-transfection of ß3Gal-T5 and FUT3 cDNAs with the core 2 ß6GlcNAc-T1 or the core 3 ß3GlcNAc-T6 cDNAs, the Lea mAb T174 reacted only with Lea on core 2 O-glycans of the reporter protein CD43/mIgG (Figure 2, panel A and Figure 3, panels A and B). This reactivity did not correlate with that of the Leb mAbs after co-transfecting the FUT3 and FUT2 gene cDNAs. Then a much stronger staining of Leb on core 3 structures was detected with both Leb mAbs (96FR2.10, data not shown, and T218, Figure 3, panels C and D). Since Leb could be detected on core 3, type 1 precursors do exist on core 3 and the absence of such cannot explain the lack of Lea reactivity on core 3 structures using the T174 mAb. Interestingly, another anti-Lea mAb, 78FR2.3, strongly stained Lea on core 3, and also, albeit much weaker, Lea on core 2 (Figure 3, panels E and F). Similar to what was found for N-linked glycans of the AGP/mIgG fusion protein (see above), the anti-Lea staining was diminished, but was not completely abolished on O-glycans of reporter proteins carrying Leb following co-transfection of the FUT2 gene 1,2-fucosyltransferase (cf. Figure 3, panels A and E to panels B and F). To verify that the epitopes were located on O-linked and not N-linked glycans, we also performed peptide: N-glycosidase F (PNGaseF)-treatment of the proteins (Figure 3). No reduction in staining intensity was observed following this treatment, which suggests that the epitopes are located on O-glycans (Figure 3, + for PNGaseF treatment, – for no treatment).
|
FUT5 supports type 1 Lewis antigen biosynthesis and has a different preference for core 2 and core 3 O-glycans compared with FUT3
Following transient transfection of CHO-K1 cells with plasmids encoding FUT1, ß3Gal-T5, ß6GlcNAc-T1, or ß3GlcNAc-T6, FUT5, and either CD43/mIgG (Figure 4, panel A) or PSGL-1/mIgG (Figure 5, panel A), Leb was detected only on core 2. On the other hand, when FUT3 was used instead of FUT5, Leb could almost only be detected on the core 3 and not on the core 2 branches (Figures 4 and 5, panel A). Thus, it appears as if FUT3 4-fucosylates type 1 chain extensions on core 3 whereas FUT5 4-fucosylates type 1 chains on core 2 O-glycans. The same result was also seen with FUT2 (data not shown). The absence of Lea could not be explained by SLea production, as revealed by staining with an anti-SLea mAb (Figures 4 and 5, panel C). No Lea reactivity was detected on CD43/mIgG reporter proteins isolated from cells transfected with FUT5 with (Figures 4 and 5, panel B) or without any 2Fuc-T (data not shown) suggesting that FUT5 requires H type 1 structures as precursors for type 1 Lewis antigen biosynthesis.
|
|
Because the expression of FUT1 might compete with the formation of DUPAN-2, we also examined SLea staining of CD43/mIgG and PSGL-1/mIgG purified from cells co-transfected with ß3Gal-T5, ß6GlcNAc-T1, or ß3GlcNAc-T6, FUT3 or FUT5 without any 2Fuc-T. The same preference of FUT3 and FUT5 was seen, that is FUT3 made SLea mostly on core 3 (Figure 6). A weak SLea staining of PSGL-1 carrying core 2 was seen in cells transfected with FUT5 (Figure 6).
|
Both FUT1 and FUT2 produce 2-fucosylated type 1 structures and compete with the formation of SLea
On AGP/mIgG (data not shown), PSGL-1/mIgG, and CD43/mIgG (Figure 7, panel A) fusion proteins from cells transfected with FUT1 and FUT3 in addition to the appropriate ß3Gal-Ts and core enzymes, Leb was found. This supports previous studies proposing that FUT1 accepts type 1 chain precursors (Liu et al., 1998; Mathieu et al., 2004). In fact, the staining intensity suggested that FUT1 was more efficient than FUT2 in supporting Leb antigen biosynthesis on both N- and O-linked glycans. This difference was not dependent on whether Leb was situated on core 2 (data not shown) or 3 (Figure 7). The intensity of the Leb staining was inversely correlated to that of both SLea and Lea (Figure 7, panels A–C). In fact, on PSGL-1/mIgG from cells transfected with FUT1, no SLea was seen (Figure 5, panel C and Figure 7, panel C).
|
|
Discussion |
|---|
|
TopAbstractIntroductionResultsDiscussionMaterials and methodsConflict of interest statementAcknowledgmentsReferences |
|---|
By using transient transfections in this study, we have examined in CHO-K1 cells the biosynthesis of Lea and Leb on recombinant reporter proteins carrying O- and N-glycans, respectively. CHO-K1 cells express only core 1 O-glycans (Liu et al., 2005
; Olson et al., 2005). Thus, co-expression of core 2 ß6GlcNAc-T1 or core 3 ß3GlcNAc-T6 enabled us to assess the core chain specificity of the ß3Gal-Ts and Fuc-Ts involved in type 1 chain and Lea/Leb antigen biosynthesis, respectively.
FUT3 has been considered as the main enzyme in humans producing 4-fucosylation (Oriol et al., 1986; Clausen and Hakomori, 1989). Nevertheless, there is some evidence suggesting that FUT5 also can fucosylate type 1 precursors in vitro (de Vries et al., 1995; Xu et al., 1996; Vo et al., 1998; de Vries et al., 2001) and in vivo (Orntoft, Holmes, et al., 1991; Serpa et al., 2003). In light of this, the most surprising and important observation was that, although both FUT3 and FUT5 were able to mediate Leb synthesis when co-expressed with an 2-fucosyltransferase, their preference for O-glycan core structures was clearly different. FUT3 preferred to make Leb on core 3 chains on CD43/mIgG and PSGL-1/mIgG, whereas FUT5 used core 2 for Leb biosynthesis. This was also true for the synthesis of SLea (Figures 4–
6). This finding was quite unexpected, but in keeping with this, it has been shown that FUT5 in vitro fucosylates type 2 on core 2 in favor of core 3 (Pykari et al., 2000). It is remarkable that FUT3 and FUT5, even though 91% identical (Weston et al., 1992), show a different preference for both core 2 and core 3 and for both type 1 and type 2. We cannot at this stage exclude that this specificity of the enzymes is due to a different Golgi localization of the enzymes (de Graffenried and Bertozzi, 2003, 2004), rather than a direct difference in specificity for the core structures (Pykari et al., 2000), but we are currently investigating that issue.
In cells expressing an 2Fuc-T and an 4Fuc-T, Leb is produced on the PSGL-1/ and CD43/mIgG reporter proteins if co-expressed with a combination of ß3Gal-T5 and a core 2 or core 3 enzyme. However, under these circumstances, the PSGL-1/ and CD43/mIgG proteins are still reactive with anti-Lea mAbs, even though the reactivity is less intense than on proteins produced in CHO cells expressing FUT3 or FUT5 as the only fucosyltransferase (Figures 2–5). This indicates that the 2Fuc-Ts has a lower Km for type 1 chain precursors than FUT3 and FUT5 leading to the biosynthesis of H type 1 structures, and subsequently Leb, rather than Lea. This is also supported by previous studies which suggest that H-type 1 structures are much more easily fucosylated by FUT3 and FUT5 than are the corresponding non-fucosylated type 1 precursors (Xu et al., 1996; Dupuy et al., 2002). We cannot exclude that some of the Lea is converted to Leb even though this is unlikely to be the case considering all the available in vitro data. It is also not probable that the Lea Abs used were cross-reacting with Leb, because no staining of the Leb-BSA conjugate was observed with the two Lea mAbs used. Curiously, we could not detect any Lea following FUT5 gene transfection (Figures 4 and 5), even though this has been reported earlier (de Vries et al., 1995, 2001; Dupuy et al., 2002). Instead, a small amount of SLea is produced on PSGL-1 (Figure 6) and Dupuy et al. (2002) have shown that there is more expression of SLea than Lea in cells expressing FUT5. A possible explanation is also that FUT5 needs H type 1 or DUPAN-2, rather than type 1, as precursors in vivo as suggested by Xu et al. (1996). In line with the latter hypothesis, available in vitro data show a clearer preference for 4-fucosylation by FUT5 on H type 1 than on non-fucosylated type 1 (Xu et al., 1996; Costache et al., 1997; Dupuy et al., 2002). Another possibility is that FUT5 can make on Lea, but not on O-glycans in our cell-based system.
From gene expression investigations and functional studies, the in vivo relevance of FUT5 is still not clear (Cameron et al., 1995; Serpa et al., 2003). However, several investigations have shown the existence of 4-fucosylated structures in FUT3-negative individuals (Orntoft, Holmes et al., 1991; Henry et al., 1997; Candelier et al., 2000; Angstrom et al., 2004). A recent study on 47 patients who were Lewis negative on RBC showed that 36 were still Le positive in the gastric mucosa supporting the notion that 4-FucTs other than FUT3 are operational at this anatomical site (Serpa et al., 2003). It might potentially also be ascribed to a leaky expression of FUT3 (Nishihara et al., 1999a,b).
It is intriguing to speculate that the ability to make 4-fucose might not have occurred through divergent evolution (Costache et al., 1997; Dupuy et al., 2002), but rather that the ancestor of vertebrates had this ability (Gustafsson, Kacskovics, et al., 2005) and that it was subsequently lost in some lineages. Interestingly, the Lea found in frogs is situated on a core 6 structure (GlcNAcß6GalNAc), which is more similar to core 2 than to core 3 (Guerardel et al., 2003).
There are so far seven confirmed ß3Gal-Ts cloned in humans (Amado et al., 1998; Kolbinger et al., 1998; Amado et al., 1999; Isshiki et al., 1999; Zhou et al., 1999). This study was carried out on ß3Gal-T1, -T2, and -T5 because they are considered to be the enzymes in humans capable of type 1 chain synthesis (Amado et al., 1999; Hennet, 2002). Earlier investigations on the enzymatic specificity, using different saccharides and glycoproteins as acceptors, have given contradictory results. Amado et al. (1999) found that ß3Gal-T2, but not T1 or T5 galactosylated ovalbumin—a glycoprotein carrying N-linked glycans. Zhou et al. (1999) however, found that ß3Gal-T1, and to a certain extent also T5, could galactosylate ovalbumin. In addition, they showed that ß3Gal-T5 worked best on O-glycans and not very well on N-glycans. On the other hand, Salvini et al. (2001) found that ß3Gal-T5 acted very well on N-linked glycans in vivo, much better than ß3Gal-T1 or T2, and that this activity also blocked polylactosamine chain elongation.
Amado et al. (1998) found that a core 3-based substrate could not be an acceptor for ß3Gal-T1 and T2. It was also reported that core 2 O-glycans could not act as acceptors for any of the ß3Gal-Ts—T1, T2, or T5 (Amado et al., 1999). However, in vitro data from the same group showed that GlcNAcß6Man-1-OMe could be galactosylated by ß3Gal-T1 (Amado et al., 1998).
To examine the ability of the various ß3Gal-Ts to synthesize type 1 chains on N-linked glycans in intact cells, we used an 1-acid glycoprotein immunoglobulin fusion (AGP/mIgG2b) protein as a reporter that carries only N-linked glycans (Fournier et al., 2000). Because no good type 1 chain core-specific mAbs were available, the presence of type 1 chains was indirectly verified by Lea or Leb reactivity following co-transfection of an 4-fucosyltransferase cDNA (FUT3 or FUT5) alone or in combination with an 2-fucosyltransferase cDNA (FUT1 or FUT2). We, in agreement with the findings of Salvini et al. (2001), found that ß3Gal-T5 was effective in making type 1 on N-glycans. What is more surprising in our results is that ß3Gal-T1 made a substantial, whereas ß3Gal-T2 made a very small, contribution to the synthesis of type 1 chain on N-linked glycans (Figure 1, panels A–C).
Our data is strongly suggestive of a role for ß3Gal-T5 in ß1,3-galactosylation of core 2 O-glycans (Figures 2–6). To our knowledge, this is the first time that type 1 chain elongation of a core 2 structure has been definitely demonstrated in vivo, although a few studies have provided indirect evidence for this (Amano and Oshima, 1999; Amano et al., 2001; Salvini et al., 2001). This shows that in vitro data using purified enzymes or cell extracts need to be confirmed using controlled experiments in cells. For this purpose, CHO-K1 cells are a good choice for several reasons. First, they are well characterized regarding their glycan repertoire (Gustafsson, Hultberg, et al., 2005; Liu et al., 2005; Olson et al., 2005). Second, their glycan profile seems to be rather small, as they are not known to express any 2 or 3Fuc-T, nor any enzymes of the ABO blood group family (Lofling et al., 2002).
Curiously, even though both ß3Gal-T1 and T2 expression resulted in the same amount of Lea-staining on AGP/mIgG following co-transfection with FUT3, there was clearly a difference in Leb-staining when FUT2 (Figure 1, panels A–C) was co-expressed. These results were not dependent on which 2Fuc-T was used (data not shown). The most plausible interpretation of these experiments is that ß3Gal-T1 and -T2 must galactosylate N-glycans differently, and that this difference is only utilized by 4Fuc-Ts in conjunction with an 2Fuc-T, perhaps due to a different branch localization of the resulting H type 1 structure similar to what we see for O-glycan synthesis.
In O-glycans, the type 1 and 2 chains can be found on different core structures (Clausen and Hakomori, 1989; Brockhausen, 1999). The most common in mammals are core 1 and 2, which are found in many tissues of the human body, and core 3 and 4, which are mainly found in tissues of endodermal origin (Brockhausen, 1999). ß6GlcNAc-T1 (Bierhuizen and Fukuda, 1992) is expressed ubiquitously and we therefore chose this core 2 enzyme for our study. So far, only one core 3 producing enzyme, ß3GlcNAc-T6, has been described (Iwai et al., 2002). There is also a recently cloned core 1 elongation enzyme (Yeh et al., 2001), which has been shown to be involved in the biosynthesis of selectin ligands and the MECA-79 epitope (Yeh et al., 2001; Mitoma et al., 2003). It might be that choosing different core enzymes could give different results, and that merits further investigation.
A caveat is that CD43 and PSGL-1 in normal tissues so far have mainly been found on leukocytes; cells that normally do not express core 3. However, reports exist for CD43 expression in colorectal adenomas, adenocarcinomas, and on cell lines of diverse origin (Sikut et al., 1997; Fernandez-Rodriguez et al., 2002). Further, to our knowledge, there is only one investigation that has studied the expression of PSGL-1 (Laszik et al., 1996). The authors, however, did not systematically examine malignant tissues outside the hematopoietic system. In fact, PSGL-1 was recently shown to be expressed in human prostate tumor cells (Dimitroff et al., 2005). This indicates that CD43 and/or PSGL-1 may very well in vivo be found in cells that express ß3GlcNAc-T6. Furthermore, since the ß3GlcNAc-T6 works on both CD43 and PSGL-1 (Figures 2–7), we believe our results not to be an artifact due to the potentially differential expression pattern of ß3GlcNAc-T6 on the one hand and PSGL-1 and CD43 on the other.
It has been unclear whether both FUT1 and FUT2 work on type 1 structures (Oriol et al., 1986; Clausen and Hakomori, 1989; Liu et al., 1998; Oriol and Mollicone, 2002; Mathieu et al., 2004). The Km value for FUT2 on type 1 has been shown to be lower than for FUT1 on type 1 (Oriol and Mollicone, 2002). We found that FUT1, together with both FUT3 and FUT5, seemed to be even better in producing Leb than is FUT2 (Figure 7), opposite to what was predicted by in vitro data but in agreement with Liu et al. (1998). This was the case for both N- and O-glycans. Also, the synthesis of H type 1 directly influenced the synthesis of precursors for SLea production (Figures 6 and 7, panels A and C).
In summary, we have found that the specificities of FUT3 and FUT5 seem to complement each other, not only with regard to the preference for type 1 and type 2 but also for the core structure the H type 1 chain is situated on, with FUT3 preferring core 3 and FUT5 core 2. We also found that ß3Gal-T1, -T2, and -T5 all support type 1 chain synthesis on N-linked glycans, whereas only ß3Gal-T5 can make type 1 on both core 2 and core 3 O-glycans in vivo. Remarkably, even though both ß3Gal-T1 and T2 produced about the same amount of type 1 chain on N-glycans available for conversion to Lea by 4Fuc-Ts, there was a marked difference in the amount of Leb produced, with ß3Gal-T1 promoting more Leb synthesis than did ß3Gal-T2 irrespective of the 2Fuc-T used. Finally, confirming previous investigations, both FUT1 and FUT2 are able to use type 1 chains. It will be of great interest to further examine the importance of these findings in relation to, for example, evolution, antibody, and lectin binding specificity, bacterial adhesion, fertilization, and dendritic cell modulation via DC-SIGN (Young et al., 1983; Bianchet et al., 2002; Kojima et al., 2002; Mitchell et al., 2002; Frison et al., 2003; Griffitts et al., 2005; Terada et al., 2005).
|
Materials and methods |
|---|
|
TopAbstractIntroductionResultsDiscussionMaterials and methodsConflict of interest statementAcknowledgmentsReferences |
|---|
Plasmid construction
Fusion proteins.
The cDNA encoding the extracellular part of CD43 was amplified by PCR (see Table I for primer sequences) from an expression plasmid encoding full length CD43 (a kind gift of Prof. Brian Seed, Department of Molecular Biology, MGH, Boston, MA, USA), and was subcloned into the mouse IgG2b Fc expression cassette using HindIII and BamH1. Similarly, the AGP-coding sequence was PCR amplified, excluding the stop codon and the leader peptide (Table I), from a human liver cDNA library. The AGP cDNA was fused in frame with the cDNA encoding the CD5 leader sequence upstream and the Fc portion of mouse IgG2b downstream using the NheI and BamHI sites in the expression cassette. The same vector backbone was used for all fusion protein constructs.
|
Core enzymes.
The core 2 ß6GlcNAc-T1 cDNA was constructed as described before (Liu et al., 2003). The core 3 ß3GlcNAc-T6 (Liu et al., 2005) was PCR amplified from human stomach cDNA and subcloned into CDM8 using HindIII and XbaI.
ß3Gal-Ts.
ß3Gal-T1 and -T2 were amplified by PCR using a human tonsil stroma cDNA library as template together with the primers shown in Table I, and were subsequently subcloned into the CDM8 expression plasmid using HindIII and NotI. ß3Gal-T5 was amplified using Expand Long Template PCR (Boehringer Mannheim, Mannheim, Germany) on human placental genomic DNA. The PCR product was subcloned into CDM8 using HindIII and NotI.
2Fuc-Ts.
The cDNAs of the blood group H gene (FUT1) encoding FUT1 and of the Se gene (FUT2) encoding FUT2 were amplified and subcloned as described (Lofling et al., 2002).
4Fuc-Ts.
The Lewis gene (FUT3) encoded 3/4fucosyltransferase expression plasmid was a kind gift of Prof. Brian Seed. The FUT5 cDNA was amplified in one amino- and one carboxy-terminal fragment by PCR using placental genomic DNA as template and internal overlapping primers containing a NaeI site (Table I). The two pieces were cloned into a modified CDM8 vector having NaeI in the polylinker, but lacking the M13 origin of replication and its NaeI site (Table I).
Albumin conjugates
All BSA conjugates were purchased from Dextra Laboratories, Reading, UK. Ley-HSA was from Isosep, Tullinge, Sweden.
Transfections and cell culture
Recombinant proteins were produced by transient transfections of CHO-K1 cells in 25 cm2 T-flasks (Falcon). Plasmids encoding protein/IgG, glycosyltransferases, and, in necessary cases, empty vector were used at a total of 10 µg of plasmid and 20 µL of Lipofectamine 2000 (Invitrogen, Paisley, UK). Cells were cultured and supernatants harvested as previously described (Lofling et al., 2002).
Antibodies
All antibodies were diluted in 3% BSA in phosphate buffered saline (PBS) with 0.05% Tween 20 (PBST) or only PBST. Anti-Lea [T174 (Sakamoto et al., 1986), IgG1; Calbiochem, San Diego, CA, USA and Signet Laboratories, Dedham, MA, USA] was used at a dilution of 1:40–80. Anti-Lea (78FR2.3, IgM; Diamed, Cressier, Switzerland) was a kind gift from Dr A. Shanwell at Blodcentralen, Karolinska University Hospital, Huddinge. It was used at a dilution of 1:200. Anti-Leb [BG-6 (Sakamoto et al., 1986), IgM; Signet] was used at a dilution of 1:200. Anti-Leb (96FR2.10, IgM; Biotest, Dreieich, Germany) was used at a dilution of 1:200. Anti-SLea [1116-NS-19–9 (Magnani et al., 1983; Rye et al., 1998), IgG1; Serotec, London, England] was used at a dilution of 1:40. Horse radish peroxidase (HRP)-conjugated goat anti-mouse IgM (Pierce, Rockford, IL, USA) and goat anti-mouse IgG F(ab')2 were used at a dilution of 1:80–160,000 and anti-mouse IgG at a dilution of 1:10–40,000.
Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS–PAGE) and western blotting
The recombinant proteins were purified by immunoprecipitation as before (Lofling et al., 2002). After immunoprecipitation and three washes in PBS, the agarose beads were mixed with 50 µL of 2x LDS-sample buffer (Invitrogen, Paisley, UK) and heated at 70°C for 10 min. Samples, typically 10 µL, were loaded on a 4–12% NUPAGE-gel (Invitrogen). Electrophoresis was run at 200V, for about 60 min. For western blotting, the samples were blotted onto 0.2 µm nitrocellulose membranes (Invitrogen) at 100V between 60 and 90 minutes in a Mini Protean II transfer system (Bio-Rad, Hercules, CA, USA). Membranes were blocked with 3% BSA/PBST overnight at 4°C. All incubations with antibodies were done for an hour. Washing steps in between all incubations were performed with an initial quick rinse using washing buffer followed by three changes of buffer, 5 min for each change. Thereafter, membranes were developed using ECL plus (Amersham Biosciences, Uppsala, Sweden) following the manufacturer’s instructions. In all lanes, about the same amount of protein was loaded, as detected with an anti-mIgG antibody.
|
Conflict of interest statement |
|---|
|
TopAbstractIntroductionResultsDiscussionMaterials and methodsConflict of interest statementAcknowledgmentsReferences |
|---|
None declared.
|
Acknowledgments |
|---|
|
TopAbstractIntroductionResultsDiscussionMaterials and methodsConflict of interest statementAcknowledgmentsReferences |
|---|
We thank Drs Jining Liu and Rosey She for kindly sharing expression plasmids. We are also grateful to Dr Agneta Shanwell at the Blodcentralen, Huddinge University Hospital, for providing us with antibodies.
This work was supported by the Swedish Research Council. J.H. and J.L. were both supported by the program "Glycoconjugates in Biological Systems" financed by the Swedish Foundation for Strategic Research in the early phase of this study.
|
Abbreviations |
|---|
AGP/mIgG,
1-acid glycoprotein/mouse immunoglobulin G; BSA, bovine serum albumin; FBS, fetal bovine serum; Fuc-T, fucosyltransferase; Gal-T, Galactosyltransferase; GlcNAc-T, N-acetylglucosaminyltransferase; HSA, human serum albumin; HRP, horse radish peroxidase; (m)IgG, (mouse) IgG; (m)IgM, (mouse) IgM; mAb, monoclonal antibody; PBS(T), phosphate-buffered saline (with 0.05% Tween-20); PNGaseF, peptide N-glycosidase F; PSGL-1, P-selectin glycoprotein ligand-1; SDS–PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; (S)Le, (Sialyl) Lewis |
References |
|---|
|
TopAbstractIntroductionResultsDiscussionMaterials and methodsConflict of interest statementAcknowledgmentsReferences |
|---|
Alon, R., Feizi, T., Yuen, C.T., Fuhlbrigge, R.C., and Springer, T.A. (1995) Glycolipid ligands for selectins support leukocyte tethering and rolling under physiologic flow conditions. J. Immunol., 154, 5356–5366.[Abstract]
Amado, M., Almeida, R., Carneiro, F., Levery, S.B., Holmes, E.H., Nomoto, M., Hollingsworth, M.A., Hassan, H., Schwientek, T., Nielsen, P.A., and others. (1998) A family of human beta3-galactosyltransferases. Characterization of four members of a UDP-galactose: beta-N-acetyl-glucosamine/beta-nacetyl-galactosamine beta-1,3-galactosyltransferase family. J. Biol. Chem., 273, 12770–12778.
Amado, M., Almeida, R., Schwientek, T., and Clausen, H. (1999) Identification and characterization of large galactosyltransferase gene families: galactosyltransferases for all functions. Biochim. Biophys. Acta, 1473, 35–53.[Medline]
Amano, J., Kobayashi, K., and Oshima, M. (2001) Comparative study of glycosyltransferase activities in Caco-2 cells before and after enterocytic differentiation using lectin-affinity high-performance liquid chromatography. Arch. Biochem. Biophys., 395, 191–198.[CrossRef][Medline]
Amano, J. and Oshima, M. (1999) Expression of the H type 1 blood group antigen during enterocytic differentiation of Caco-2 cells. J. Biol. Chem. 274, 21209–21216.
Angstrom, J., Larsson, T., Hansson, G.C., Karlsson, K.A., and Henry, S. (2004) Default biosynthesis pathway for blood group-related glycolipids in human small intestine as defined by structural identification of linear and branched glycosylceramides in a group O Le(a-b-) nonsecretor. Glycobiology, 14, 1–12.
Bianchet, M.A., Odom, E.W., Vasta, G.R., and Amzel, L.M. (2002) A novel fucose recognition fold involved in innate immunity. Nat. Struct. Biol., 9, 628–634.[ISI][Medline]
Bierhuizen, M.F. and Fukuda, M. (1992) Expression cloning of a cDNA encoding UDP-GlcNAc: Gal beta 1–3-GalNAc-R. (GlcNAc to GalNAc) beta 1-6GlcNAc transferase by gene transfer into CHO cells expressing polyoma large tumor antigen. Proc. Natl Acad. Sci. USA, 89, 9326–9330.
Blaszczyk, M., Pak, K.Y., Herlyn, M., Sears, H.F., and Steplewski, Z. (1985) Characterization of Lewis antigens in normal colon and gastrointestinal adenocarcinomas. Proc. Natl Acad. Sci. USA, 82, 3552–3556.
Boren, T., Falk, P., Roth, K.A., Larson, G., and Normark, S. (1993) Attachment of Helicobacter pylori to human gastric epithelium mediated by blood group antigens. Science, 262, 1892–1895.
Brockhausen, I. (1999) Pathways of O-glycan biosynthesis in cancer cells. Biochim. Biophys. Acta, 1473, 67–95.[Medline]
Cameron, H.S., Szczepaniak, D., and Weston, B.W. (1995) Expression of human chromosome, 19p alpha(1,3)-fucosyltransferase genes in normal tissues. Alternative splicing, polyadenylation, and isoforms. J. Biol. Chem., 270, 20112–20122.
Candelier, J.J., Mollicone, R., Mennesson, B., Coullin, P., and Oriol, R. (2000) Expression of fucosyltransferases in skin, conjunctiva, and cornea during human development. Histochem. Cell Biol., 114, 113–124.[ISI][Medline]
Clausen, H. and Hakomori, S. (1989) ABH and related histo-blood group antigens; immunochemical differences in carrier isotypes and their distribution. Vox Sang, 56, 1–20.[ISI][Medline]
Costache, M., Apoil, P.A., Cailleau, A., Elmgren, A., Larson, G., Henry, S., Blancher, A., Iordachescu, D., Oriol, R., and Mollicone, R. (1997) Evolution of fucosyltransferase genes in vertebrates. J. Biol. Chem., 272, 29721–29728.
Dimitroff, C.J., Descheny, L., Trujillo, N., Kim, R., Nguyen, V., Huang, W., Pienta, K.J., Kutok, J.L., and Rubin, M.A. (2005) Identification of leukocyte E-selectin ligands, P-selectin glycoprotein ligand-1 and E-selectin ligand-1, on human metastatic prostate tumor cells. Cancer Res., 65, 5750–5760.
Dupuy, F., Germot, A., Marenda, M., Oriol, R., Blancher, A., Julien, R., and Maftah, A. (2002) Alpha1,4-fucosyltransferase activity: a significant function in the primate lineage has appeared twice independently. Mol Biol. Evol., 19, 815–824.
Dupuy, F., Petit, J.M., Mollicone, R., Oriol, R., Julien, R., and Maftah, A. (1999) A single amino acid in the hypervariable stem domain of vertebrate alpha1,3/1,4-fucosyltransferases determines the type 1/type 2 transfer. Characterization of acceptor substrate specificity of the Lewis enzyme by site-directed mutagenesis. J. Biol. Chem., 274, 12257–12262.
Fernandez-Rodriguez, J., Andersson, C.X., Laos, S., Baeckstrom, D., Sikut, A., Sikut, R., and Hansson, G.C. (2002) The leukocyte antigen CD43 is expressed in different cell lines of nonhematopoietic origin. Tumour Biol., 23, 193–201.[CrossRef][Medline]
Fournier, T., Medjoubi, N.N., and Porquet, D. (2000) Alpha-1-acid glycoprotein. Biochim. Biophys. Acta, 1482, 157–171.[CrossRef][Medline]
Frison, N., Taylor, M.E., Soilleux, E., Bousser, M.T., Mayer, R., Monsigny, M., Drickamer, K., and Roche, A.C. (2003) Oligolysine-based oligosaccharide clusters: selective recognition and endocytosis by the mannose receptor and dendritic cell-specific intercellular adhesion molecule, 3 (ICAM-3)-grabbing nonintegrin. J. Biol. Chem., 278, 23922–23929.
de Graffenried, C.L. and Bertozzi, C.R. (2003) Golgi localization of carbohydrate sulfotransferases is a determinant of L-selectin ligand biosynthesis. J. Biol. Chem., 278, 40282–40295.
de Graffenried, C.L. and Bertozzi, C.R. (2004) The stem region of the sulfotransferase GlcNAc6ST-1 is a determinant of substrate specificity. J. Biol. Chem., 279, 40035–40043.
Greenwell, P. (1997) Blood group antigens: molecules seeking a function? Glycoconj J., 14, 159–173.[CrossRef][ISI][Medline]
Griffitts, J.S., Haslam, S.M., Yang, T., Garczynski, S.F., Mulloy, B., Morris, H., Cremer, P.S., Dell, A., Adang, M.J., and Aroian, R.V. (2005) Glycolipids as receptors for Bacillus thuringiensis crystal toxin. Science, 307, 922–925.
Guerardel, Y., Petit, D., Madigou, T., Guillet, B., Maes, E., Maftah, A., Boujard, D., Strecker, G., and Kol, O. (2003) Identification of the blood group Lewis (a) determinant in the oviducal mucins of Xenopus tropicalis. FEBS Lett., 554, 330–336.[CrossRef][ISI][Medline]
Gustafsson, A., Hultberg, A., Sjostrom, R., Kacskovics, I., Breimer, M.E., Boren, T., Hammarstrom, L., and Holgersson, J. (2005) Carbohydrate-dependent inhibition of helicobacter pylori colonisation using porcine milk. Glycobiology, 16, 1–10.
Gustafsson, A., Kacskovics, I., Breimer, M.E., Hammarstrom, L., and Holgersson, J. (2005) Carbohydrate phenotyping of human and animal milk glycoproteins. Glycoconj J., 22, 109–118.[CrossRef][ISI][Medline]
Hakomori, S. and Andrews, H.D. (1970) Sphingoglycolipids with Leb activity, and the co-presence of Lea-, Leb-glycolipids in human tumor tissue. Biochim. Biophys. Acta, 202, 225–228.[Medline]
Harrington, P.R., Vinje, J., Moe, C.L., and Baric, R.S. (2004) Norovirus capture with histo-blood group antigens reveals novel virus–ligand interactions. J. Virol., 78, 3035–3045.
Hennet, T. (2002) The galactosyltransferase family. Cell Mol. Life Sci., 59, 1081–1095.[CrossRef][ISI][Medline]
Henry, S., Jovall, P.A., Ghardashkhani, S., Elmgren, A., Martinsson, T., Larson, G., and Samuelsson, B. (1997) Structural and immunochemical identification of Le (a), Le (b), H type, 1, and related glycolipids in small intestinal mucosa of a group O Le (a-b-) nonsecretor. Glycoconj J., 14, 209–223.[CrossRef][ISI][Medline]
Isshiki, S., Togayachi, A., Kudo, T., Nishihara, S., Watanabe, M., Kubota, T., Kitajima, M., Shiraishi, N., Sasaki, K., Andoh, T., and Narimatsu, H. (1999) Cloning, expression, and characterization of a novel UDP-galactose: beta-N-acetylglucosamine beta1,3-galactosyltransferase (beta3Gal-T5) responsible for synthesis of type 1 chain in colorectal and pancreatic epithelia and tumor cells derived therefrom. J. Biol. Chem., 274, 12499–12507.
Iwai, T., Inaba, N., Naundorf, A., Zhang, Y., Gotoh, M., Iwasaki, H., Kudo, T., Togayachi, A., Ishizuka, Y., Nakanishi, H., and Narimatsu, H. (2002) Molecular cloning and characterization of a novel UDP-GlcNAc: GalNAc-peptide beta1,3-N-acetylglucosaminyltransferase (beta 3Gn-T6), an enzyme synthesizing the core 3 structure of O-glycans. J. Biol. Chem., 277, 12802–12809.
Kerr, C.L., Hanna, W.F., Shaper, J.H., and Wright, W.W. (2004) Lewis X-containing glycans are specific and potent competitive inhibitors of the binding of ZP3 to complementary sites on capacitated, acrosome-intact mouse sperm. Biol. Reprod., 71, 770–777.
Kojima, N., Seino, K., Sato, Y., and Mizuochi, T. (2002) Carbohydrate carriers affect adhesion of H. pylori to immobilized Leb-oligosaccharide. FEBS Lett.<