Recently, cDNAs encoding human chondroitin 4-O-sulfotransferase-1 and -2 (C4ST-1 and C4ST-2) were cloned based on their similarity to HNK-1 sulfotransferase (HNK-1ST) (Hiraoka, N., Nakagawa, H., Ong, E., Akama, T.O., Fukuda, M.N., and Fukuda, M.  Molecular cloning and expression of two distinct human chondroitin 4-O-sulfotransferasesthat belong to the HNK-1 sulfotransferase gene family. J. Biol. Chem., 275, 20188–20196). In the present study, we identified two additional novel sulfotransferases by searching the expression sequence tag and genomic DNA database for enzymes similar to C4ST-1 and C4ST-2. These newly cloned enzymes, termed GalNAc4ST-1 and GalNAc4ST-2, belong to the HNK-1ST gene family having 40–42% identity with C4ST-1. GalNAc4ST-1 and -2 do not add sulfate to HNK-1 precursor glycans, chondroitin, or desulfated dermatan sulfate. Instead, both enzymes can transfer sulfate to the 4-position of GalNAc in the context of GalNAcβ1→4GlcNAcβ1→R attached to both N-linked and core 2 branched O-linked oligosaccharides. GalNAc4ST-1 and -2 transcripts are highly expressed in the pituitary gland and trachea, respectively, and GalNAc4ST-1 and -2 transcripts are reciprocally expressed in other tissues as well. Moreover, both enzymes can transfer sulfate to lutropin, a pituitary glycoprotein hormone. These combined results indicate that GalNAc4ST-1 and -2 play critical roles in forming sulfo→4GalNAcβ1→4GlcNAcβ1→R in both N-glycans and O-glycans in a tissue-specific manner.
Received on December 22, 2000; revised on January 19, 2001; accepted on January 23, 2001.
Sulfate groups in carbohydrates play important roles in conferring highly specific functions on glycoproteins, glycolipids, and proteoglycans (Hooper et al., 1996). For example, abolition of a specific sulfate group was shown to result in abnormal embryonic development in both mouse and Drosophila (Bullock et al., 1998; Lin and Perrimon, 1999). Binding of growth factors, such as basic fibroblast growth factor (FGF), and adhesion of herpes simplex virus I to the cell surface were also shown to be dependent on sulfate groups (Rapraeger et al., 1991; Shukla et al., 1999). These results demonstrate that the sulfation of carbohydrates plays a critical role in cell–cell interaction, signal transduction, and embryonic development.
Sulfation of carbohydrates is carried out by a sulfotransferase that catalyzes a specific reaction on specific acceptors. These sulfotransferases have type II membrane topology and almost exclusively function in the Golgi apparatus. Molecular cloning of these sulfotransferases was initially successful for chondroitin 6-O-sulfotransferase and keratan sulfate 6-O-sulfotransferase based on the amino acid sequences of purified enzymes (Fukuta et al., 1995, 1997). Although it was not apparent that these sulfotransferases share homologous sequences, molecular cloning of HNK-1 sulfotransferase (HNK-1ST) revealed that there is a homologous sequence motif among the sulfotransferases cloned to date (Ong et al., 1998). This motif, ZZRDPXXXZ, where X and Z denote any amino acid and a hydrophobic amino acid, respectively, was then found to be a part of the 3′-phosphate binding site for the donor 3′-phosphoadenosine 5′-phosphosulfate (PAPS) (Kakuta et al., 1997, 1998; Ong et al., 1999). Crystallographic studies also demonstrated that the amino acid sequences responsible for binding to 5′-phosphosulfate are conserved among different sulfotransferases (Kakuta et al., 1998).
Based on the presence of the weak but discernible similarity among different sulfotransferases, additional sulfotransferases have been identified by their similarity to previously cloned sulfotransferases. One such example is the molecular cloning of an L-selectin ligand sulfotransferase (LSST or HEC-GlcNAc6ST) that adds a sulfate to the 6-position of N-acetylglucosamine, leading to the formation of 6-sulfo sialyl Lewis x, NeuNAcα2→3Galβ1→4[Fucα1→3(sulfo→6)]GlcNAcβ1→R (Hiraoka et al., 1999; Bistrup et al., 1999), which functions as an L-selectin ligand. This cloning was based on its similarity to chondroitin 6-O-sulfotransferase (Fukuta et al., 1995) and keratan sulfate 6-O-sulfotransferase (Fukuta et al., 1997).
Recent studies demonstrated that chondroitin 4-O-sulfotransferases (C4ST) could be cloned based on their similarity to HNK-1ST (Hiraoka et al., 2000; Yamauchi et al., 2000). HNK-1ST adds a sulfate to the 3-position of glucuronic acid (GlcA), which is in turn linked to the 3-position of galactose in the context of N-acetyllactosamine. On the other hand, C4ST adds a sulfate to the 4-position of N-acetylgalactosamine, which is linked to the 4-position of GlcA. Moreover, HNK-1ST adds a sulfate to GlcA at the nonreducing end, while C4ST apparently adds sulfate groups to a chondroitin polymer. These results indicate that sulfotransferases, utilizing entirely different acceptors, can be related to each other. Similarly, a novel chondroitin 6-O-sulfotransferase, chondroitin 6-O-sulfotransferase-II (Kitagawa et al., 2000), was found to be related more to GlcNAc-6-O-sulfotransferase, which adds a sulfate at C-6 of the nonreducing terminal N-acetylglucosamine (Uchimura et al., 1998), than the previously cloned chondroitin 6-O-sulfotransferase (Fukuta et al., 1995).
During the molecular cloning of C4ST-1 and C4ST-2, we noticed that there are two additional DNA sequences in the expression sequence tag (EST) and genomic DNA database that are related to HNK-1ST, C4ST-1, and C4ST-2. Expression of full-length cDNAs revealed that they do not encode HNK-1ST or C4ST. Instead, we herein demonstrate that these cDNAs encode novel N-acetylgalactosamine 4-O-sulfotransferases that add a sulfate to the 4-position of N-acetylgalactosamine in GalNAcβ1→4GlcNAcβ1→R attached to pituitary hormones (Green et al., 1985; Skelton et al., 1991; Smith et al., 1993). Moreover, we found that these two GalNAc-4-O-sulfotransferases (GalNAc4ST-1 and GalNAc4ST-2)can add sulfate to GalNAcβ1→4GlcNAcβ1→R in both N-glycans and core 2 branched O-glycans. The expression of these two enzymes exhibits reciprocal tissue distribution, indicating that GalNAc4ST-1 and -2 play complementary roles in different tissues.
Isolation of cDNA encoding GalNAc-4-O-sulfotransferase (GalNAc4ST)
By searching the EST database for a novel cDNA related to HNK-1ST, three distinct cDNA sequences were found to have homology to the HNK-1ST sequence. The first two were found to encode chondroitin 4-O-sulfotransferase-1 and -2 as reported previously (Hiraoka et al., 2000). The third gene (EST # h15485 and AA417127) encodes a part of the cDNA sequence. The full-length cDNA was obtained by 5′-rapid amplification of cDNA ends (RACE), which encodes an open reading frame of 1329 base pairs, predicting a 443-amino acid residue protein (52,036 Da). Because this enzyme is highly related to the recently reported GalNAc4ST-1 (Okuda et al., 2000; Xia et al., 2000), we termed this new enzyme GalNAc4ST-2 (Figure 1).
By searching the human genomic DNA database for a novel cDNA related to GalNAc4ST-2, one genomic sequence (AC067911) was found to have homology to GalNAc4ST-2 and HNK-1ST. Based on the genomic sequence and cDNA prepared from poly(A)+RNA isolated from human pituitary gland (Clontech), the full-length cDNA was obtained. This cDNA encodes an open reading frame of 1272 base pairs, predicting a 424 amino acid residue protein (48,834 Da) and was named GalNAc4ST-1. This enzyme is identical to that reported by others (Okuda et al., 2000; Xia et al., 2000). The nucleotide sequences have been deposited in GenBank with the accession numbers AF305981 for GalNAc4ST-1 and AF239821 for GalNAc4ST-2.
Comparison of the amino acid sequences
The comparison of the amino acid sequences of GalNAc4ST-1, GalNAc4ST-2, C4ST-1, C4ST-2, and HNK-1ST is shown in Figure 2. The amino acid sequence of GalNAc4ST-1 is more homologous to GalNAc4ST-2 (63.7% identity in 278 amino acids) than that between C4ST-1 and C4ST-2 (41.8% identity). Both GalNAc4ST-1 and -2 is related slightly more to C4ST-1 (40–42% identity) than HNK-1ST (32–37% identity) or C4ST-2 (33–36% identity). Figure 2 also illustrates that all members of the HNK-1ST gene family are highly homologous to each other in the catalytic domains. As described previously (Hiraoka et al., 2000), the 5′-phosphosulfate binding and 3′-phosphate binding sites are well conserved. Moreover, three additional regions (A, B, and C) are conserved among all these sulfotransferases, whereas the fourth, fifth, and sixth homologous regions (D, E, and F) are only conserved between GalNAc4ST-1 and -2 (Figure 2). These conserved regions (A–F) likely correspond to domain structures that are important for the proper sterical structure maintained in all of the HNK-1ST gene family (A, B, and C) and both GalNAc4ST-1 and -2 (D, E, and F).
Functional expression of GalNAc4ST-1 and GalNAc4ST-2
To determine the acceptor specificity of GalNAc4ST-1 and GalNAc4ST-2, pcDNA3.1-GalNAc4ST-1, pcDNA3.1-GalNAc4ST-2, and control pcDNA3.1 were separately transfected into Chinese hamster ovary (CHO) cells. Sulfotransferase activity on cell extracts from the transfected cells was assayed using various acceptors. First, neither GalNAc4ST-1 nor GalNAc4ST-2 exhibited activity toward an HNK-1 precursor acceptor GlcAβ1→4Galβ1→4GlcNAcβ1→octyl. Neither GalNAc4ST-1 nor GalNAc4ST-2 incorporated [35S]-sulfate into chondroitin, chondroitin sulfate, or desulfated dermatan sulfate in contrast to a strong activity by C4ST-1 (data not shown).
We then tested GalNAcβ1→4GlcNAcβ1→2Manα1→6Manβ1→octyl as an acceptor, because GalNAc4-O-sulfotransferase and chondroitin 4-O-sulfotransferase catalyze similar reactions on similar acceptors. As shown in Figure 3, both GalNAc4ST-1 and GalNAc4ST-2 exhibited significant activity on this acceptor. Similarly, GalNAc4ST-1 and -2 added sulfate to GalNAcβ1→4GlcNAcβ1→6Manα1→6Manβ1→octyl. Moreover, both enzymes exhibited strong activity toward a core 2–based O-glycan, GalNAcβ1→4GlcNAcβ1→6(Galβ1→3) GalNAcα1→octyl (Figure 3). In contrast, none of the oligosaccharides containing galactose or N-acetylglucosamine as terminal sugars served as an acceptor. These results indicate that GalNAc4ST-1 and -2 act on GalNAcβ1→4GlcNAcβ1→R in both N- and O-glycans.
GalNAc4ST-2 exhibited a much stronger in vitro activity than GalNAc4ST-1 after expression in CHO cells (Figure 3). Although such a difference could reflect the difference in catalytic efficiency between these two enzymes, it is more likely that it is due to a difference in the efficiency in translation or posttranslational modification such as disulfide bond formation.
Identification of reaction products
To formally demonstrate that GalNAc4ST-1 and GalNAc4ST-2 add a sulfate to the 4-position of GalNAc residues, GalNAcβ1→4GlcNAcβ1→2Manα1→6Manβ1→octyl was incubated with [35S]-PAPS and the enzyme preparation. The reaction product was purified by a Sep-Pak C18 cartridge column and then subjected to Bio-Gel P-4 gel filtration or QAE-Sephadex column chromatography. As shown in Figure 4A, the product eluted as a monosulfated compound on QAE-Sephadex column chromatography. The same product eluted at an elution position close to that of GalNAcβ1→4GlcNAcβ1→2Manα1→6Manβ1→octyl in Bio-Gel P-4 gel filtration (Figure 4B). Previous experiments demonstrated that a tetrasaccharide elutes at almost the same position regardless of whether it is monosulfated or not (Hiraoka et al., 1999). The purified product was also subjected to mild acid hydrolysis to release sulfated monosaccharide. On high-performance liquid chromatography (HPLC) analysis, the majority of 35S-sulfated monosaccharide eluted at the same elution position as that of 4-O-sulfated GalNAc, separating from 6-O-sulfated GalNAc, 3-O-sulfated GalNAc, or 4,6-di-O-sulfated GalNAc (Figure 4C). These results demonstrate that GalNAc4ST-1 and -2 transfer a sulfate group to C-4 of GalNAc in the context of GalNAcβ1→4GlcNAcβ1→R.
Incorporation of 35S-sulfate into lutropin by GalNAc4ST-1 and GalNAc4ST-2
We then tested whether GalNAc4ST-1 and -2 can add sulfate to lutropin. After incubating lutropin with GalNAc4ST-1, GalNAc4ST-2, and [35S]-PAPS, the products were subjected to SDS–polyacrylamide gel electrophoresis (PAGE) followed by fluorography. The results shown in Figure 5 demonstrate that both GalNAc4ST-1 and -2 can transfer 35S-sulfate to lutropin. The same experiments also showed that 35S-sulfate was incorporated into N-glycans as expected from the previous structural analysis (Green et al., 1985).
GalNAc4ST-1 and GalNAc4ST-2 are differentially expressed in various tissues
Northern blot analysis demonstrates that GalNAc4ST-1 transcripts (∼2.4 kb) are expressed highly in fetal brain, adult brain, and adult spinal cord and moderately in adult liver, kidney, and heart (Figure 6). In addition, skeletal muscle apparently expresses low molecular weight (∼1.0 kb) transcripts. It is not clear if the ∼1.0 kb transcript produces a functional GalNAc4ST-1. In dot blot analysis, the GalNAc4ST-1 transcripts are strongly expressed in heart, pituitary gland, kidney, liver, small intestine, and placenta (Figure 7). In contrast, the transcripts of GalNAc4ST-2, mostly 2.4 kb in size, are expressed predominantly in trachea, pancreas, and testis and moderately in spleen, prostate, fetal lung, and fetal kidney (Figure 6). In particular, the transcripts of GalNAc4ST-2 in trachea were strongly detected on the dot blot (Figure 7).
To determine whether GalNAc4ST-2 is also expressed in pituitary gland, reverse-transcription polymerase chain reaction (RT-PCR) using pituitary poly(A)+RNA was performed. The results show that both GalNAc4ST-2 and GalNAc4ST-1 transcripts can be detected in pituitary gland (Figure 8). These results indicate that GalNAc4ST-1 is highly expressed in pituitary gland and GalNAc4ST-2 is expressed in pituitary gland at minute but detectable levels.
Chromosome localization of the GalNAc4ST-1 and GalNAc4ST-2 gene
The genomic sequence of GalNAc4ST-1 can be found in registered genes (#AC067911, AC005615, AC008994, AC007205). The cluster of these genomic sequences has been mapped to human chromosome 19 band q13.1. The genomic sequence of GalNAc4ST-2 can be found in registered gene #AP001272. This gene has been mapped to chromosome 18, band q11.2. The genomic sequence of HNK-1ST can be found in registered gene #AC012493, which has been mapped to chromosome 2. Previously, we reported that the C4ST-1 and C4ST-2 genes are mapped to chromosome 12q23 and chromosome 7p22, respectively (Hiraoka et al., 2000). These results indicate that all of the genes belonging to the HNK-1 sulfotransferase gene family reside in different chromosomes.
The present study describes the isolation of novel cDNAs encoding N-acetylgalactosamine 4-O-sulfotransferase by searching the EST and genomic DNA database for a novel cDNA homologous to the human HNK-1ST (Ong et al., 1998). Previously, we have cloned cDNAs encoding C4ST-1 and -2 using the same approach (Hiraoka et al., 2000). The proteins, encoded by novel cDNAs cloned in the present study, however, did not add a sulfate to chondroitin or HNK-1 precursor oligosaccharides. Because 4-O-sulfation in N-acetylgalactosamine was known in pituitary hormone and other glycoproteins (Green et al., 1985; Skelton et al., 1991; Smith et al., 1993), we tested the enzymatic activity toward those acceptors. The present study thus demonstrated that both GalNAc4ST-1 and -2 add a sulfate to the 4-position of GalNAc in GalNAcβ1→4GlcNAcβ1→R. This acceptor and the acceptor for C4ST, GalNAcβ1→4GlcAβ1→R, are similar, and it is not unreasonable that these four enzymes share conserved amino acid sequences. On the other hand, HNK-1ST adds a sulfate to the 3-position of GlcA in GlcAβ1→3Galβ1→4GlcNAcβ1→R. It is thus not clear why these five sulfotransferases are related to each other, forming the HNK-1 sulfotransferase gene family (see also Discussion in Hiraoka et al., 2000). It has been shown that the activity of GalNAc4ST in bovine pituitary membranes can be inhibited more significantly by chondroitin sulfate A (Skelton et al., 1991), which contains 4-O-sulfated GalNAc, than chondroitin sulfate C, which contains 6-O-sulfated GalNAc. This finding is consistent with the results obtained in the present study that GalNAc4STs are related more to chondroitin 4-O-sulfotransferase than chondroitin 6-O-sulfotransferase.
Comparison of the amino acid sequences of GalNAc4ST-1, GalNAc4ST-2, and the other three enzymes that belong to the HNK-1ST gene family showed that these five enzymes significantly share conserved sequences in the 5′-phosphosulfate and 3′-phosphate binding sites (Figure 3). Two additional homologous regions (A and B in Figure 3) were also noticed in the previous studies when C4ST-1, C4ST-2, and HNK-1ST were compared (Hiraoka et al., 2000). Moreover, GalNAc4ST-1 and 2 share conserved sequences in the vicinity of the 3′-phosphate binding site and the COOH-terminal (D, E, and F in Figure 3). These results indicate that a subfamily of sulfotransferases may be recognized by the presence of conserved sequences that are not apparent in other members of the same gene superfamily. It is important to note, however, that GalNAc4ST-1 and -2 clearly belong to the HNK-1ST gene family based on a phylogenetic tree analysis carried out previously (Hiraoka et al., 2000).
Sulfation at C-4 of N-acetylgalactosamine in glycoproteins has been extensively studied (Green et al., 1985; Fiete et al., 1991; Skelton et al., 1991; Stockell Hartree and Renwick, 1992; Siciliano et al., 1993, 1994; Bergwerff et al., 1995; Manzella et al., 1996). These studies indicate that sulfo→4GalNAcβ1→4GlcNAcβ1→R is present in the N-glycans of pituitary hormones, such as luteinizing hormone, proopiomelanocortin, and thyroid-stimulating hormone. This sulfated terminal structure in these hormones is recognized by a receptor present in hepatic endothelial cells. It has been suggested that the rise and fall in circulating luteinizing hormone levels is regulated by its uptake through the recognition of sulfo→4GalNAcβ1→4GlcNAcβ1→R by a mannose-binding protein on endothelial cells (Fiete et al., 1991, 1997). Because GalNAc4ST-1 is highly expressed in the pituitary gland, it is likely that this enzyme is responsible for the 4-O-sulfation of pituitary hormones.
While we were preparing this manuscript, the cloning of GalNAc4ST-1 was reported by others (Okuda et al., 2000; Xia et al., 2000). Although the results obtained in our study are very similar to those reported, there is a slight difference in the expression of GalNAc4ST-1 transcripts in different tissues. We and Okuda et al. (2000) showed that GalNAc4ST-1 is expressed widely in different tissues including the pituitary gland (Figures 6 and 7), while the expression of the same transcripts were more restricted to the pituitary gland in the studies by Xia et al. (2000). This difference may be due to a difference in the blots used. Nevertheless, all of these studies agree that GalNAc4ST-1 is highly expressed in the pituitary gland, indicating its role in the sulfation of pituitary gland glycoproteins.
GalNAc4ST-2, on the other hand, is highly expressed in the trachea, where a large amount of mucin-type glycoproteins are synthesized. This expression profile is the same as that recently reported for GalNAc4ST-2 (Kang et al., 2001). It has been shown that proopimelanocortin contains sulfo→4GalNAcβ1→4GlcNAcβ1→6(Galβ1→3)GalNAc as a major O-glycan (Siciliano et al., 1994), in addition to the same sulfated terminal in N-glycans (Siciliano et al., 1993; Skelton et al., 1992). This sulfated core2 branched O-glycan may play a role in determining the turnover of proopiomelanocortin following its secretion (Siciliano et al., 1994). Because proopimelanocortin is synthesized in the pituitary gland, GalNAc4ST-1 is mainly responsible for sulfation in O-glycans, although GalNAc4ST-2 must contribute to sulfation of N- and O-glycans in the same tissue, to some extent. Other reports demonstrate that human urokinase and Tamm-Hosfall glycoprotein, mainly synthesized in kidney cells, also contain sulfo→4GalNAcβ1→4GlcNAcβ1→R (Hard et al., 1992; Bergwerff et al., 1995). Our results show that GalNAc4ST-1 transcripts are apparently expressed more in human adult kidney than are GalNAc4ST-2 transcripts, suggesting that GalNAc4ST-1 may be mainly responsible for the sulfation of these glycoproteins in adult kidney cells. On the other hand, GalNAc4ST-2 is probably involved in the sulfation of tissue factor pathway inhibitor synthesized in the human embryonic kidney 293 cell line (Smith et al., 1992), because the transcripts of GalNAc4ST-2 are expressed in fetal kidney.
The present study also demonstrated that a side chain derived from the 6-position of α2,6-linked mannose was as good an acceptor as a side chain from the 2-position of α2,6-linked mannose for both GalNAc4ST-1 and GalNAc4ST-2 (Figure 5). In nature, a side chain derived from the 6-position of α2,6-linked mannnose is present in tetra-antennary N-glycans. The structural analyses thus far reported, however, show that sulfo→4GalNAcβ1→4GlcNAcβ1→R in pituitary gland glycoproteins are present in biantennary N-glycans and core 2 branched O-glycans (Green et al., 1985; Skelton et al., 1991, 1992; Stockell Hartree and Renwick, 1992; Siciliano et al., 1993, 1994; Smith et al., 1993; Bergwerff et al., 1995; Manzella et al., 1996). These results suggest that the N-acetylgalactosaminyltransferase that adds β1,4-linked GalNAc to GlcNAc may have a strict acceptor requirement for carbohydrate moieties as well as proteins that attach those carbohydrates (Mengeling et al., 1995). This may also be the reason why luteinizing hormone but not chorionic gonadotropin contains a sulfo→4GalNAcβ1→4GlcNAc terminal structure (Endo et al., 1979; Smith et al., 1993). In contrast, human urinary kallidinogenase synthesized in the kidney contain GalNAcβ1→4GlcNAcβ1→R in side chains derived from the 6- or 2-position of α-mannose (Tomiya et al., 1993). These findings are consistent with the studies showing that there are at least two different N-acetylgalactosaminyltransferases: one in the pituitary gland that recognizes a specific amino acid motif, and another in other tissues such as the kidney that does not (Dharmesh et al., 1993). On the other hand, mouse LSST preferentially adds a sulfate to core 2 branched O-glycans and less efficiently to N-glycans (Hiraoka et al., 1999). These combined results indicate that sulfation of oligosaccharides can be regulated by acceptor substrate specificity of glycosyltransferases which form a precursor or that of sulfotransferase itself.
It has been demonstrated that sulfo→4GalNAcβ1→4GlcNAcβ1→2Man can be bound to the mannose/GalNAc-4-SO4-receptor (Fiete et al., 1997). It has not been determined, however, if the mannose/GalNAc-4-SO4 receptor also binds to sulfo→4GalNAcβ1→4GlcNAcβ1→R in mucin-type O-glycans. Future studies will need to determine whether the same acceptor recognizes sulfo→4GalNAcβ1→4GlcNAcβ1→R in both N- and O-glycans.
Materials and methods
Isolation of cDNAs encoding GalNAc-4-O-sulfotransferases
In HNK-1ST, the conserved motif IVRDPFERL is present in amino acid residues 187–195 (Bakker et al., 1997; Ong et al., 1998). The amino acid sequence of residues 165–230 was thus used as a probe to search the dbEST using the TBLASTN program. This resulted in two overlapping genes, h15485 and AA417127, which were derived from human infant brain and testis, respectively. The sequence analysis indicated that both lacked 5′-end sequences, although the clone h15485 contained longer 5′-region sequences.
To obtain 5′-sequences, 5′-RACE was carried out using poly(A)+RNA isolated from human lymph node (Clontech). For reverse transcription, an anti-sense primer corresponding to nucleotides 565–546 (nucleotides 1–3 encode the initiation methionine) was used under the conditions described previously (Hiraoka et al., 1999). PCR was then carried out using a nested 3′-primer corresponding to nucleotides 524–503. PCR products were cloned into pBluescript II SK(+) that had been digested with EcoRV and reacted with Taq polymerase, and colony hybridization was performed to identify the 5′-RACE products using a 32P-labeled DNA fragment encompassing nucleotides 182–524. The latter probe DNA was obtained by PCR. The nucleotide sequence of the PCR product revealed a mismatch during PCR, resulting in a frame shift. To obtain the correct sequence in this region, RT-PCR was carried out to amplify nucleotides 227–948 using poly(A)+RNA from human lymph node. The PCR product was excised with HindIII and EcoRI, replacing the corresponding sequence in the first product of 5′-RACE. The resultant 5′-RACE product was digested with NotI and EcoRI, and this fragment was appended to the 5′-end EcoRI site of AA417127 (encompassing nucleotides 671–1966), producing a full length cDNA encoding GalNAc4ST-2. The ligated cDNA cloned into the NotI sites of pcDNA3.1/Hygro (Invitrogen) resulted in pcDNA3.1–GalNAc4ST-2 (the name of GalNAc4ST-2 was given after the determination of its acceptor specificity).
The second gene was identified by searching the human genomic DNA database using the amino acid sequence for GalNAc4ST-2 as a probe. Comparison of the identified genomic sequence, registered as ACO67911, with that of GalNAc4ST-2 revealed that the upstream sequence from the 5′-phosphosulfate binding site to the stop codon appeared to be encoded in one exon. Accordingly, RT-PCR was carried out using a 5′-primer (nucleotides 540–559), a 3′-primer (nucleotides 1052–1032), and Marathon cDNA prepared from human pituitary gland (Clontech) as a template. Because the product of this reaction showed an expected sign of 512 base pairs, 5′-RACE and PCR of the 3′-region of the cDNA were carried out using the above Marathon pituitary cDNA as a template. Finally, PCR was carried out using the above Marathon cDNA from human pituitary gland as a template, a 5′-primer (nucleotides –3 to 16), and a 3′-primer (nucleotides 1280–1261), which were synthesized based on the sequences obtained by the above 5′-RACE and PCR experiments. These primers also contained EcoRI and XbaI sites, and the PCR product was cloned into the same sites of pcDNA 3.1/Hyg, resulting in pcDNA3.1–GalNAc4ST-1.
Synthesis of acceptor oligosaccharides
Galβ1→4GlcNAcβ1→2Manα1→6Manβ1→octyl, GlcNAcβ1→2Manα1→6Manβ1→octyl, Galβ1→4GlcNAcβ1→6Manα1→6Manβ1→octyl, GlcNAcβ1→6Manα1→6Manβ1→octyl, GlcAβ1→3Galβ1→4GlcNAcβ1→octyl, and Galβ1→4GlcNAcβ1→O(CH2)8COOCH3 were synthesized as described previously (Tsuboi et al., 1996; Ding et al., 1998; Ujita et al., 1998; McAuliffe et al., 1999). Galβ1→4GlcNAcβ1→6(Gal-β1→3)GalNAcα1→octyl and GlcNAcβ1→ 6(Galβ1→ 3)GalNAcα1→octyl were synthesized as described below. The suitably protected Galβ1→3GalNAcα1→octyl (compound 1) was synthesized by the coupling of methyl 2,3,4,6-tetra-O-benzoyl-1-thio-β-d-galactopyranoside with octyl 2-acetamido-4,6-O-benzylidene-2-deoxy-α-d-galactopyranoside in a dimethyl(methylthio)sulfonium triflate promoted reaction (68% yield) (Fügedi and Garegg, 1986). Benzylidene cleavage under acidic conditions then yielded the disaccharide diol having two free hydroxyl groups at C-4 and C-6 of the GalNAc residue (compound 1). In parallel, ethyl 3,6-di-O-benzyl-2-deoxy-2-phthlimido-1-thio-β-d-glucopyranoside was coupled with 2,6-di-O-acetyl-3,4-di-O-chloroacetyl-α-d-galactopyranosyl chloride in the presence of AgOTf to give the required bifunctional N-acetyllactosamine donor (compound 2) in 73% yield. Reaction of the disaccharide donor (compound 2) with the core 1 disaccharide acceptor (compound 1) in the presence of dimethyl(methylthio)sulfonium triflate gave the tetrasaccharide (compound 3) in 61% yield. Dechloroacetylation of compound 3 using hydrazinedithiocarbonate in 2,6-lutidine-HOAc (3:1) then furnished the tetrasaccharide triol in 74% yield. Conventional deprotection of the tetrasaccharide triol produced Galβ1→4GlcNAcβ1→6(Galβ1→3)GalNAcα1→octyl (compound 4) with a yield of 56% after purification on Sephadex LH-20. GlcNAcβ1→6(Galβ1→3)GalNAcα1→octyl (compound 5) was obtained on a 5 mg scale from compound 4 by digestion with β-galactosidase from jack beans as described elsewhere (McAuliffe et al., 1999).
Partial NMR (500 MHz; D2O) data are as follows: 4: δ 4.83 (d, J1,2 = 3.6 Hz, 1H, H-1), 4.52 (d, J1¢¢,2¢¢ = 7.8 Hz, 1H, H-1″), 4.42 (2d, J1¢,2¢¢ = 7.8 Hz and J1¢¢¢,2¢¢¢ = 7.8 Hz, 2H, H-1′ and H-1″′), 4.26 (dd, 1H, H-3), 4.17 (bs, 1H, H-4), 2.0 and 2.02 (2s, each 3H, 2 NHAc); 13C NMR: δ 105.4, 103.7, 102.2 and 97.5. 5: δ 4.86 (d, J1,2 = 3.0 Hz, 1H, H-1), 4.52 (d, J1¢,2¢ = 8.5 Hz, 1H, H-1¢), 4.46 (d, J1¢¢,2¢¢ = 8.0 Hz, 1H, H-1″), 4.30 (dd, 1H, H-3), 4.21 (bs, 1H, H-4), 2.02 and 2.01 (2s, each 3H, 2 NHAc); 13C NMR: δ 105.3, 102.1, 97.2. m/z (matrix-assisted laser desorption ionization-time of flight mass spectrometry [MALDI-TOF]) are: 4, Calculated for C36H64O21N2 (M+Na+) 883.3894, found 883.3910; 5, calculated for C30H54O16N2 (M+Na+) 721.3366, found 721.3365. Detailed procedures of the synthesis will be published elsewhere (Misra et al., 2001).
Addition of β1,4-linked GalNAc to GlcNAc-terminated compounds was carried out as a secondary reaction of milk β1,4-galactosyltransferase (β4Gal-TI) as described previously (Palcic and Hindsgaul, 1991; Do et al., 1995). α-Lactalbumin (10 mg/ml) was added to facilitate N-acetylgalactosaminylation (Do et al., 1995). The reaction products were separated from UDP-GalNAc using Sep-Pak C-18 cartridge column chromatography and then fractionated by HPLC using an AX-5 NH2-bonded column (4.0 mm × 30 cm) (Bierhuizen and Fukuda, 1992). The column was initially equilibrated with solution A (15% H2O/85% acetonitrile) and then eluted by a linear gradient to 95% solution A and 5% solution B (60% 15 mM KH2PO4/40% acetonitrile) in 10 min. The column was then eluted by a linear gradient to 50% solution A and 50% solution B in 55 min. Under these conditions, GlcNAcβ1→2Manα1→6Manβ1→octyl and GalNAcβ1→4GlcNAcβ1→2Manα1→6Manβ1→octyl eluted at 31 min and 40 min, respectively. Similarly, GalNAcβ1→4GlcNAcβ1→6Manα1→6Manβ1→octyl and GalNAcβ1→4GlcNAcβ1→6(Galβ1→3)GalNAcα1→octyl were separated from GlcNAcβ1→6Manα→6Manβ1→octyl and GlcNAcβ1→6(Galβ1→3)GalNAcα1→octyl, respectively.
CHO cells were transfected with pcDNA3.1-GalNAc4ST-1 or pcDNA3.1-GalNAc4ST-2 as described previously (Hiraoka et al., 2000). Sixty-two hours after the transfection, the cells attached to plates were washed with phosphate-buffered saline, scraped, and homogenized; the supernatant after brief centrifugation was used as an enzyme preparation, as described previously (Hiraoka et al., 2000).
GalNAc-4-O-sulfotransferase activity was measured in the reaction mixture (50 µl) containing 50 mM imidazole-HCl, pH 6.8, 0.015% protamine chloride, 40 mM 2-mercaptoethanol, 0.1% Triton X-100, 10 mM NaF, 2 mM ATP, an acceptor oligosaccharide (0.5 mM), 0.2 nmole [35S]-PAPS (1.97 Ci/mmol), and 25 µl of the enzyme solution. The composition of this reaction mixture was determined based on the results shown by Skelton et al. (1991). After incubation for 2 h at 28°C, the reaction mixture was boiled for 2 min, adjusted to 0.25 M ammonium formate, pH 4.0 (Ong et al., 1998), and then applied to a Sep-Pak C-18 cartilage column. The column was washed with the same solution, then the product was eluted with 30% acetonitrile in water and the radioactivity was measured by scintillation counting.
To characterize the product sulfated by GalNAc4ST, 35S-sulfate labeled product was prepared using GalNAcβ1→4GlcNAcβ1→2Manα1→6Manβ1→octyl as an acceptor. The product, purified by a Sep-Pak C-18 cartilage column, was then applied to a column (1.0 × 120 cm) of Bio-Gel P-4 equilibrated with 0.1 M ammonium acetate buffer (pH 6.7) as described elsewhere (Hiraoka et al., 1999). Separation of oligosaccharides with different numbers of sulfate groups was achieved by QAE-Sephadex A-25 column chromatography. The column (0.7 × 0.8 cm) was equilibrated with 10 mM pyridine-acetate buffer (pH 5.5) and stepwisely eluted with 70 mM, 120 mM, 140 mM, 500 mM, and 1 M NaCl in the the same buffer. Mono- and disulfated oligosaccharides elute with 70 mM and 120 mM NaCl, respectively (Hiraoka et al., 1999).
To determine the position of a sulfate attached to GalNAc, the purified 35S-sulfated product was hydrolyzed in 40 mM HCl at 100°C for 120 min (Habuchi and Conrad, 1985). The hydrolysate was directly subjected to HPLC using a Whatman Partisil SAX-10 column (4.6 mm × 25 cm) equilibrated with 10 mM KH2PO4 at room temperature. In the first 20 min, the column was isocratically eluted with 10 mM KH2PO4 and then eluted with a linear gradient from 10 mM KH2PO4 to 400 mM KH2PO4 over the next 20 min. Standard sulfated monosaccharides were purchased from Sigma.
Incorporation of 35S-sulfate into glycoprotein acceptors
Thirty micrograms of human lutropin (Sigma) was incubated with the enzyme preparation for GalNAc4ST-1 or GalNAc4ST-2. After incubation for 18 h at 28°C under the same conditions described above, a portion of the product was digested with N-glycanase. The digested and undigested products were subjected to SDS–PAGE and fluorography.
Northern blot analysis
Northern blots of multiple human tissues (Clontech) or human RNA Master Blot™(Clontech) were hybridized with cDNA fragments isolated from pcDNA3.1–GalNAc4ST-1 and pcDNA3.1–GalNAc4ST-2 after 32P-labeling using a nick translation kit (PrimIt-RmT) from Stratagene.
One hundred nanograms of poly(A)+RNA were reverse transcribed by using MMLV reverse transcriptase (Stratagene) with 300 ng of oligo dT in 50 µl of reaction mixture. One microliter of RT mixture was used as a template for PCR. To detect GalNAc4ST-2 transcripts, samples were denatured for 2 min at 94°C, followed by 35 cycles of 1 min at 94°C, 30 s at 56°C, and 30 s at 72°C. For GalNAc4ST-1, the annealing temperature was raised to 60°C. Oligonucleotide primers were used for specific amplification of human GalNAc4ST-2 (nucleotides 710–729 for 5′-primer and nucleotides 962–940 for 3′-primer) and human GalNAc4ST-1 (nucleotides 540–559 for 5′-primer and nucleotides 838–819 for 3′-primer).
The authors wish to thank Drs. Yili Ding and Joseph McAuliffe for useful discussion, Dr. Edgar Ong for critical reading of the manuscript, and Joseph P. Henig and Risa Tabata for organizing the manuscript. This work was supported by grants RO1 CA33895, PO1 CA71932 and in part by RO1 CA48737 awarded by the National Cancer Institute.
C4ST, chondroiting 4-O-sulfotransferase; CHO, Chinese hamster ovary; EST, expression sequence tag; GlcA, glucuronic acid; HPLC, high-performance liquid chromatography; LSST, L-selectin ligand sulfotransferase; PAGE, polyacrylamide gel electrophoresis; PAPS, 3′-phosphoadenosine 5′-phosphosulfate; RACE, rapid amplification of cDNA ends; RT-PCR, reverse-transcription polymerase chain reaction; ST, sulfotransferase.
To whom correspondence should be addressed