Regulation of I-Branched Poly-N-Acetyllactosamine Synthesis

I-branched poly-N-acetyllactosamine is a unique carbohydrate composed of N-acetyllactosamine branches attached to linear poly-N-acetyllactosamine, which is synthesized by I-branching β1,6-N-acetylglucosaminyltransferase. I-branched poly-N-acetyllactosamine can carry bivalent functional oligosaccharides such as sialyl Lewisx, which provide much better carbohydrate ligands than monovalent functional oligosaccharides. In the present study, we first demonstrate that I-branching β1,6-N-acetylglucosaminyltransferase cloned from human PA-1 embryonic carcinoma cells transfers β1,6-linked GlcNAc preferentially to galactosyl residues ofN-acetyllactosamine close to nonreducing terminals. We then demonstrate that among various β1,4-galactosyltransferases (β4Gal-Ts), β4Gal-TI is most efficient in adding a galactose to linear and branched poly-N-acetyllactosamines. When a β1,6-GlcNAc branched poly-N-acetyllactosamine was incubated with a mixture of β4Gal-TI and i-extension β1,3-N-acetylglucosaminyltransferase, the major product was the oligosaccharide with one N-acetyllactosamine extension on the linear Galβ1→4GlcNAcβ1→3 side chain. Only a minor product contained galactosylated I-branch withoutN-acetyllactosamine extension. This finding was explained by the fact that β4Gal-TI adds a galactose poorly to β1,6-GlcNAc attached to linear poly-N-acetyllactosamines, while β1,3-N-acetylglucosaminyltransferase and β4Gal-TI efficiently add N-acetyllactosamine to linear poly-N-acetyllactosamines. Together, these results strongly suggest that galactosylation of I-branch is a rate-limiting step in I-branched poly-N-acetyllactosamine synthesis, allowing poly-N-acetyllactosamine extension mostly along the linear poly-N-acetyllactosamine side chain. These findings are entirely consistent with previous findings that poly-N-acetyllactosamines in human erythrocytes, PA-1 embryonic carcinoma cells, and rabbit erythrocytes contain multiple, short I-branches.

Poly-N-acetyllactosamines are often modified to express differentiation antigens and functional oligosaccharides. Among those oligosaccharides, sialyl Lewis x (Le x ) 1 and its sulfated forms are ligands for E-, P-, and L-selectin (12)(13)(14)(15)(16). During inflammation, E-and P-selectin expressed on activated endothelial cells bind to sialyl Le x oligosaccharides present on granulocytes and such initial binding leads to the extravasation of granulocytes. L-selectin on lymphocytes, on the other hand, recognizes sulfated sialyl Le x expressed in high endothelial venules of blood vessels (15,16). This L-selectin/counterreceptor interaction allows lymphocytes to migrate into lymphoid system, allowing lymphocytes to circulate fully in the body.
The acquisition of I-branches is important, since two of the N-acetyllactosamine side chains can have functional terminal structures. It has been demonstrated that multivalent sialyl Le x poly-N-acetyllactosamines inhibit L-selectin-mediated binding and the rejection of organ transplants with much better efficacy than monovalent sialyl Le x poly-N-acetyllactosamines (23,24). Similarly, blood group H antigens present at both termini in branched poly-N-acetyllactosamines were shown to have much better avidity to anti-ABO antibodies than linear poly-N-acetyllactosamines containing single antigenic structures (25). It was suggested that expression of i antigen in fetal erythrocytes minimizes a detrimental immune response when mother and fetus have incompatible blood group antigens (25).
In the present study, we first describe how the IGnT cloned from PA-1 cells (27) adds I-branches to linear poly-N-acetyllactosamines containing multiple acceptor sites. We then demonstrate that ␤4Gal-TI is responsible for galactosylation in the synthesis of both linear and branched poly-N-acetyllactosamines. Finally, we reconstituted the synthesis of I-branched poly-N-acetyllactosamine, the structure of which resembles that present in human erythrocytes (5), PA-1 human embryonic carcinoma cells (34), and rabbit erythrocytes (10). The results demonstrate an intricate interaction between acceptor substrates and these glycosyltransferases.

EXPERIMENTAL PROCEDURES
Isolation of cDNA Encoding iGnT and IGnT-cDNA encoding iGnT was cloned into pcDNA3.1, resulting in pcDNA3.1-iGnT, as described previously (26). pcDNAI-A, harboring cDNA encoding a signal sequence and an IgG binding domain of Staphylococcus aureus protein A, was constructed as described before (35). The catalytic domain of iGnT was cloned into this vector, resulting in pcDNAI-A⅐iGnT.
cDNA encoding IGnT was cloned from a cDNA library constructed from human PA-1 embryonic carcinoma cells, resulting in pcDNAI-IGnT, as described previously (27). A catalytic domain of IGnT was prepared by polymerase chain reaction using pcDNAI-IGnT as a template. 5Ј-and 3Ј-primers for this polymerase chain reaction were 5Ј-GCGGATCCAAGCTTCCAAAGGCTA-3Ј and 5Ј-GGCTCGAGCTCAAA-AATACCAGCTGGGT-3Ј (BamHI and XhoI sites are underlined). The polymerase chain reaction product encoding amino acid residues 30 -400 of the IGnT was digested with BamHI and XhoI and then cloned into the same sites of pcDNAI-A, resulting in pcDNAI-A⅐IGnT.
Expression of the Protein A-iGnT and Protein A-IGnT Fusion Protein-pcDNAI-A, pcDNAI-A⅐iGnT, and pcDNAI-A⅐IGnT were separately transfected with Lipofectamine Plus (Life Technologies) into COS-1 cells as described previously (36). The chimeric enzyme released into serum-free Opti-MEM was used after adsorbing the protein A chimeric enzymes to IgG-Sepharose 6FF (Amersham Pharmacia Biotech) as described previously (37). Alternatively, the culture medium was concentrated 100-fold by a Centricon 10 concentrator (Amicon) and directly used as an enzyme source. In most of the studies, the concentrated culture medium was used for iGnT, since IgG-Sepharose-bound enzymes had a low activity as seen for other glycosyltransferases (38,39). Typically, the activities of iGnT and IGnT in the incubation mixture were 38.0 nmol/h/ml using 0.5 mM Gal␤134Glc␤3p-nitrophenol (Toronto Research Chemicals) and 40.0 nmol/h/ml using 0.5 mM Gal␤134GlcNAc␤133Gal␤134GlcNAc␤136Man␣136Man␤3octyl (see below) as acceptors, respectively.
The medium from mock-transfected COS-1 cells contained less than 1 ⁄5 of iGnT activity as described (26) or less than 1 ⁄24 of IGnT activity compared with that derived from pcDNAI-A⅐iGnT-or pcDNAI-A⅐IGnTtransfected COS-1 cells.
␤4Gal-TV (42) was cloned and expressed in COS-1 cells as described previously (36). The supernatant from the transfected COS-1 cells was concentrated 100-fold as described above and used as an enzyme source. For comparing the enzymatic activities of different ␤4Gal-T samples, the final concentration of ␤4Gal-TI, -TII, -TIII, -TIV, and -TV was adjusted to 38.0 nmol/h/ml as measured using 0.5 mM GlcNAc␤3pnitrophenol (Sigma) as an acceptor.
Addition of N-Acetylglucosamine by iGnT and IGnT-To assay the transfer of N-acetylglucosamine residues by the iGnT, the reaction mixture was exactly the same as described previously (36). As acceptors, lacto-N-neo-tetraose, lacto-N-neo-hexaose, lacto-N-hexaose, pyridylaminated lacto-N-neo-hexaose (PA-lacto-N-neo-hexaose), and lacto-N-neo-tetraose (PA-lacto-N-neo-tetraose) were used. The assay products derived from the above oligosaccharides were purified by QAE-Sephadex A-25 gel and subjected to Bio-Gel P-4 gel filtration as described previously (36). The radioactivity of the aliquots was determined by a scintillation counter. The products derived from the PA-oligosaccharides were filtered through Ultrafree-MC (10-kDa cut; Millipore Corp.) and applied to the same ODS-80TS column and eluted as described above. Since PA-oligosaccharides can be detected by fluorescence, nonradioactive UDP-GlcNAc was used as a donor for the experiments using PA-oligosaccharides.

Addition of I-branch to Various Poly-N-Acetyllactosaminyl
Side Chains-Recently, it was reported that the IGnT cloned from PA-1 cells exclusively adds ␤1,6-linked N-acetylglucosamine to a galactose residue in a central position as seen in Gal␤134GlcNAc␤133Gal␤134Glc(NAc)3 R (the underlined galactose is the accepting galactose (28)). To determine if the IGnT can add N-acetylglucosamine residues far from nonreducing terminals, acceptors with various numbers of N-acetyllactosamine repeats were used. As shown in Fig. 1, B and D, the oligosaccharide containing two potential acceptor sites, (Gal␤134GlcNAc␤133) 2 Gal␤134GlcNAc␤136Man3 R, incorporated slightly more N-acetylglucosamine than that containing only one acceptor site. When the products obtained from (Gal␤134GlcNAc␤133) 2 Gal␤134GlcNAc␤136Man3 R were analyzed by endo-␤-galactosidase digestion, the majority of singly branched products contained ␤1,6-linked GlcNAc close to the nonreducing terminal ( Fig. 1F and Fig. 2, compound e).
These results indicate that the IGnT displays a preference for a galactose residue separated by one N-acetyllactosamine unit from the nonreducing terminal (Fig. 2, compounds b, e, and g). The results shown in Fig. 1, C and D, also indicate that the addition of two I-branches to neighboring galactose residues (compounds d and g in Fig. 2) occurs less efficiently than the addition of one branch. Fig. 1, A and C, illustrate that oligosaccharides containing N-acetylglucosamine at nonreducing terminals can also serve as acceptors. In particular, the results shown in Fig. 1A demonstrate that the IGnT can add ␤1,6-linked GlcNAc to a peridistal galactose (Fig. 2, compound a), thus containing dIGnT activity. This finding is consistent with the results reported in our recent study (29). When two galactoses are available in the oligosaccharide containing nonreducing terminal N-acetylglucosamine, all of the singly branched product(s) remained radioactive after endo-␤-galactosidase digestion (Fig. 1E), indicating its structure as GlcNAc␤133Gal␤134GlcNAc␤133(GlcNAc-␤136)Gal␤134GlcNAc␤136Man3 R (Fig. 2, compound c). Peridistal galactose was utilized only when the central galactose was utilized (Fig. 2, compound d). These results indicate that a peridistal galactose is the least favored by the cIGnT cloned from PA-1 cells. When the preferred galactose is missed by the IGnT, the enzyme can still add an I-branch to that galactose after one N-acetylglucosamine (Fig. 2, compounds c and d) or N-acetyllactosamine is added (Fig. 2, compounds f and g).
The above results were obtained using the recombinant IGnT bound to IgG-Sepharose, the activity of which was 3.6 nmol/h/ml. Almost identical results were obtained when a concentrated culture medium from IGnT-transfected cells, of which activity was 40.0 nmol/h/ml, was used.
␤4Gal-TI Is Responsible for Galactosylation of Linear and Branched Poly-N-acetyllactosamine Synthesis-In our previous study, we demonstrated that ␤4Gal-TI is involved in N-acetyllactosamine formation in N-glycans, while ␤4Gal-TIV forms N-acetyllactosamine in core 2 branched O-glycans (36).
It is also evident that ␤4Gal-TI, among these ␤4Gal-Ts, acts most efficiently on a linear poly-N-acetyllactosamine acceptor, GlcNAc␤133Gal␤134Glc3 PA (Fig. 3B, Table I). In both experiments, similar results were obtained when the concentration of these enzymes was increased 5-fold or decreased 5-fold. These results indicate that ␤4Gal-TI is mostly responsible for galactosylation of both branched and linear poly-N-acetyllactosamines.
Addition of N-Acetylglucosamine to I-branched Acceptors-To determine how iGnT adds N-acetylglucosamine residues to I-branched oligosaccharides, the incorporation of N-acetylglucosamine was compared between different acceptors including linear and branched oligosaccharides. The results shown in Fig. 4 demonstrate that lacto-N-neo-hexaose did not incorporate twice the amount of N-acetylglucosamine compared with lacto-N-neo-tetraose, despite the fact that the former contains two acceptor sites (A and B). Moreover, lacto-Nhexaose containing only one acceptor galactose in the I-branch incorporated N-acetylglucosamine as much as lacto-N-neohexaose (Fig. 4C). The results strongly suggest that the addi-  tion of N-acetylglucosamine to one side chain precludes the addition of another GlcNAc to the other side chain in branched structures.
To determine the structures of products, the enzymatic reaction products derived from PA-lacto-N-neo-hexaose were subjected to HPLC using an ODS column. Two products, peak A and B, were then separately digested with endo-␤-galactosidase followed by exo-␤-N-acetylglucosaminidase treatment (Fig.  5). The results indicate that peak A, the major product (75% of the total), contains N-acetylglucosamine in the I-branch, while peak B contains N-acetylglucosamine at the linear side chain (Fig. 6). These results indicate that iGnT prefers the Gal␤13 4GlcNAc␤136 branch over Gal␤134GlcNAc␤133, which was originally part of linear poly-N-acetyllactosamine. More importantly, no product containing two N-acetylglucosamines at both terminals was formed, supporting the above conclusion that N-acetylglucosamine can be added to only one of the terminals. This result indicates that a branched acceptor probably becomes a competitive inhibitor for iGnT as soon as one N-acetylglucosamine is added.
Elongation of N-Acetyllactosamine Units on Branched Poly-N-acetyllactosamine-To determine how N-acetyllactosamine elongation takes place after the IGnT adds a ␤1,6-linked Nacetylglucosamine, Gal␤134GlcNAc␤133(GlcNAc␤136)Gal-␤134Glc3 PA was incubated with a mixture of ␤4Gal-TI and the iGnT, and the products were separated by reverse phase ODS column as shown in Fig. 7.
The minor peak of the products, peak F, eluted shortly after the starting material and corresponds to Gal␤134GlcNAc␤133-(Gal␤134GlcNAc␤136)Gal␤134Glc3 PA, which represents the addition of one galactose to the I-branch in the acceptor substrate. This structure was further confirmed by endo-␤galactosidase digestion and exoglycosidase digestion (data not shown). The major peak (84% of the total products), peak E, was digested by endo-␤-galactosidase, followed by ␤-N-acetylglucosaminidase (Fig. 7, C and D). This digested material was eluted at the position corresponding to Gal␤134GlcNAc␤136-Gal␤134Glc3 PA (Fig. 7D), which was also obtained in the above experiment (Fig. 5H). These results indicate that the products are either those containing one N-acetyllactosamine extension at Gal␤134GlcNAc␤133Gal␤134Glc side chain (compound E in Fig. 8) or lacto-N-neo-hexaose (compound F in Fig. 8).
Since the majority of the product was compound E, the extension of N-acetyllactosamine units along the linear poly-Nacetyllactosamine side is favored over galactosylation of the I-branch forming compound F. These results are entirely consistent with the previous findings that I-branches are usually composed of only one N-acetyllactosamine unit in erythrocytes (5) and human PA-1 embryonic carcinoma cells (34) from which the cIGnT was cloned.
Galactosylation of ␤1,6-GlcNAc Branch Is a Rate-limiting Step-The above results demonstrate that I-branch formed by cIGnT is not extended further and that N-acetyllactosamine extension takes place preferentially at linear poly-N-acetyllactosamine side chain. To understand how this is achieved, K m and V max values of ␤4Gal-TI and iGnT were obtained for linear and branched oligosaccharide acceptors. As shown in Table I, ␤4Gal-TI exhibits much lower affinity toward the branched acceptor Gal␤134GlcNAc␤133(GlcNAc␤136)Gal␤134Glc3 PA (K m ϭ 2.49 mM) than its linear counterpart GlcNAc␤-133Gal␤134Glc3 PA (K m ϭ 0.31 mM). In contrast, iGnT exhibits higher affinity toward the branched acceptor Gal␤134GlcNAc␤133(GlcNAc␤136)Gal␤134Glc3 PA (K m ϭ 0.52 mM) than the linear acceptor Gal␤134GlcNAc␤13 3Gal␤134Glc3 PA (K m ϭ 1.09 mM) (Fig. 3, C and D, Table II).
These results indicate that ␤1,3-linked GlcNAc is added to the Gal␤134GlcNAc␤133Gal side chain before galactosylation of ␤1,6-linked GlcNAc branch (Fig. 8C). This reaction is, most likely, immediately followed by galactosylation of the GlcNAc␤133Gal␤134GlcNAc␤133Gal moiety (Fig. 8D), considering that the additions of GlcNAc and Gal to Gal␤134-GlcNAc␤133Gal are favored by these two enzymes (Tables I  and II). As the last step, galactosylation of I-branch takes place (Fig. 8E). Only as a minor biosynthetic pathway, galactosylation of I-branch precedes other reactions forming compound F in Fig. 8, which may lead to more complex poly-Nacetyllactosamines. DISCUSSION The present study demonstrated that I-branch formation takes place more efficiently at sites closer to nonreducing termini than at internal sites, suggesting that I-branch is formed preferentially at the end of elongating poly-N-acetyllactosamine ( Figs. 1 and 2). If this potential site is missed, the IGnT still utilizes the site with a lower efficiency after Nacetylglucosamine (Fig. 2, c and d) or N-acetyllactosamine is added (Fig. 2, f and g). It is noteworthy that the addition of I-branches is not facilitated by having two potential acceptor sites. Rather, the total amount of N-acetylglucosamine transferred is close to that in the compound containing one acceptor site (Fig. 2, b and e-g). The addition of one branch appears to inhibit the addition of another branch on the same side chain. These results strongly suggest that the second branch is usually formed when two acceptor sites are separated by more than one N-acetyllactosamine unit.
These results are consistent with the structures of branched poly-N-acetyllactosamines determined on human adult band 3 and PA-1 cells (5, 34). First, the majority of nonreducing termini contain I-branches. Second, two I-branches are mostly was cloned from a cDNA library of human PA-1 embryonic carcinoma cells (27). These results suggest that human and rabbit erythroid precursor cells contain IGnT that is the same as or similar to the IGnT cloned from PA-1 cells. Hog small intestine may have similar poly-N-acetyllactosamine structures as seen in erythrocytes and PA-1 cells, since a cIGnT was purified from this tissue (46).
The present study also demonstrated that intrinsic properties of ␤4Gal-TI and iGnT are critical in forming short I-branches. The addition of a galactose residue to ␤1,6-linked N-acetylglucosamine branch is a much slower process than the addition of N-acetylglucosamine or galactose to an elongating Gal␤134GlcNAc␤133 side chain, as demonstrated by kinetic data (Table I and II). The addition of ␤1,3-linked N-acetylglucosamine to the Gal␤134GlcNAc␤133 side chain is even more efficient when this acceptor has a ␤1,6-linked GlcNAc branch (Table II). Combined together, these results indicate that galactosylation of a GlcNAc␤136 branch takes place as a ratelimiting step in the synthesis of branched poly-N-acetyllactosamines. This is most likely the reason why no elongation of I-branch was observed in the present study when the acceptor containing ␤1,6-linked GlcNAc was incubated with a mixture of iGnT and ␤4Gal-TI (Figs. 7 and 8).
These results are entirely consistent with the structures determined on poly-N-acetyllactosamines from human erythrocyte band 3 (5), human PA-1 embryonic carcinoma cells (34), and rabbit erythrocytes (10). Only one N-acetyllactosamine unit is present in each I-branch of these glycans.
The results obtained in the present study predict that it takes longer for additional modifications of formed I-branch than the extension of Gal␤134GlcNAc␤133 side chain. If this is the case, I-branches in internal positions contain fewer modifications than I-branches at nonreducing termini. Similarly, additional modification at elongating linear poly-N-acetyllactosamines should take place faster than at I-branches even at nonreducing termini. In fact, more ␣1,2-fucosylated I-branch was found in those branches at nonreducing termini than those in internal positions in human band 3 (5). Moreover, Fuc␣132Gal␤134GlcNAc␤133(Gal␤134GlcNAc␤136)Gal␤-13 R but not Gal␤134GlcNAc␤133(Fuc␣132Gal␤134Glc-NAc␤136)Gal␤13 R was found in monofucosylated termini of human band 3 (5). Thus, all of these structural data are consistent with the results predicted from the present study, demonstrating that galactosylation of I-branch is a rate-limiting step.
It is noteworthy that the iGnT can act very efficiently on Gal␤134GlcNAc␤133(GlcNAc␤136)Gal␤134Glc3 PA ( Fig.  3C and Table II) but not on Gal␤134GlcNAc␤133-(GlcNAc␤133Gal␤134GlcNAc␤136)Gal␤134Glc3 PA (Fig.  5, peak A). This result suggests that short GlcNAc␤136 branch may not be recognized by the iGnT, while the terminal GlcNAc residue in the extended GlcNAc␤133Gal␤134GlcNAc␤136 branch may be recognized by the iGnT, preventing the addition of GlcNAc to the other side chain due to a substrate inhibition. It is thus tempting to speculate that the short GlcNAc␤136 branch attached to linear poly-N-acetyllactosamines may be difficult for ␤4Gal-TI to recognize because of its conformation. In fact, NMR studies on the pentadecaglycolipid indicate that the anomeric proton of ␤1,6-GlcNAc linked to the internal galactose is not detected, suggesting that it is conformationally inaccessible (10). These results, combined together, indicate that intricate interaction between these glycosyltransferases and I-branched acceptors play a critical role in the synthesis of I-branched poly-N-acetyllactosamines.
Previously, it was determined that core 2 branch GlcNAc␤-136(Gal␤133)GalNAc␣3 R is galactosylated most efficiently FIG. 7. HPLC analysis of the products derived from Gal␤134GlcNAc␤133(GlcNAc␤136)Gal␤134Glc3PA after incubation with iGnT and ␤4Gal-TI. Gal␤134GlcNAc␤133-(GlcNAc␤136)Gal␤134Glc3 PA was incubated with the iGnT, ␤4Gal-TI, UDP-GlcNAc, and UDP-Gal (A). Peak E was purified (B) and then sequentially digested with endo-␤-galactosidase (C) and exo-␤-N-acetylglucosaminidase (D). Peak D was eluted at the same position as Gal␤134GlcNAc␤136Gal␤134Glc3 PA shown in Fig. 5H. Peaks S and F correspond to the starting material and PA-lacto-N-neo-hexaose, respectively. by ␤4Gal-TIV (36). Other galactosyltransferases such as ␤4Gal-TI, -TII, -TIII, and -TV exhibit a substrate inhibition toward the core 2 acceptor, probably because of competition between ␤-galactose in the acceptor and the donor substrate, UDP-Gal. Similarly, ␤4Gal-TII, -TIII, and -TV exhibited a substrate inhibition toward the I-branched acceptor (Fig. 3A). ␤4Gal-TI and ␤4Gal-TIV did not show a substrate inhibition toward the I-branched acceptor. However, the kinetic efficiency (V max /K m ) for ␤4Gal-TIV is less than a half of that for ␤4Gal-TI (Table I), indicating that ␤4Gal-TI is probably dominant in I-branch formation. ␤4Gal-TI is most efficient in galactosylation of a linear poly-N-acetyllactosamine as well ( Fig. 3B and Table I). Moreover, ␤4Gal-TI is most efficient in the synthesis of N-glycan poly-N-acetyllactosamine as shown in the previous study (36). Overall, these results indicate that ␤4Gal-TI plays a major role in poly-N-acetyllactosamine extension and I-branch formation in N-glycans. It is noteworthy that ␤4Gal-TI knock-out mice survive during development (47), and those mutant mice express polysialic acid (35) and the HNK-1 carbohydrate epitope (48) in brain glycoproteins (49). These results suggest that a ␤4Gal-T other than ␤4Gal-TI partly compensates for the loss of ␤4Gal-TI in the knock-out mice and is possibly involved in N-acetyllactosamine synthesis under normal conditions as well.
The experiments carried out in the present study were designed to mimic cellular biosynthetic pathways. The biosynthetic oligosaccharide products are also a result of the balance between the amount of glycosyltransferases present and the movement of glycoproteins in the Golgi apparatus during biosynthesis (50). For large scale synthesis of oligosaccharides in vitro, however, enzymatic synthesis can be achieved despite the fact that such a reaction is unlikely in vivo. For example, Renkonen et al. (51) synthesized highly branched poly-Nacetyllactosaminyl oligosaccharides containing four sialyl Le x termini using the cIGnT. In this oligosaccharide, every possible acceptor site was occupied by I-branch and all of the I-branches contained sialyl Le x . It was also reported that galactosylation of core 2 branch GlcNAc␤136(Gal␤133)GalNAc could be achieved using an excess amount of ␤4Gal-TI (36, 52), although ␤4Gal-TI is unlikely to be involved in its galactosylation in vivo. These results strongly suggest that the results obtained by in vitro studies need to be evaluated regarding how these findings reflect the biosynthesis taking place in cells.
The present study reveals the biosynthetic pathway involving the cIGnT that adds ␤1,6-linked GlcNAc to central galactose residues. It has been demonstrated that there is an additional IGnT, dIGnT, which adds ␤1,6-linked GlcNAc at peridistal galactose residues, forming GlcNAc␤133(GlcNAc␤-136)Gal3 R at nonreducing termini (30 -33). In this situation, galactosylation at I-branch may not be a rate-limiting step, since no substrate inhibition takes place. Recently, we have cloned a novel ␤1,6-N-acetylglucosaminyltransferase that has more dIGnT activity than cIGnT activity (29). Future studies will be of significance to determine the structures of I-branched poly-N-acetyllactosamines synthesized by this newly cloned enzyme.
FIG. 8. Proposed biosynthetic steps of I-branched poly-N-acetyllactosamine. ␤1,6-Linked N-acetylglucosamine is first added to a central galactose by cIGnT (B). This is followed by the addition of ␤1,3-linked N-acetylglucosamine by iGnT (C) and ␤1,4-linked galactose by ␤4Gal-TI (D), adding N-acetyllactosamine to the linear poly-N-acetyllactosamine side chain. This is followed by galactosylation of ␤1,6-linked N-acetylglucosamine, forming I-branch (E). As a minor biosynthetic pathway, galactosylation of I-branch may take place as soon as ␤1,6linked N-acetylglucosamine is added (F). If compound F is formed, ␤1,3-linked Nacetylglucosamine is preferentially added to I-branch by iGnT, potentially leading to more complex poly-N-acetyllactosamines. E and F correspond to peaks E and F in Fig. 7, respectively. This biosynthetic pathway is based on the results obtained in the present study.