Cloning and characterization of a new human UDP-N-acetyl-alpha-D-galactosamine:polypeptide N-acetylgalactosaminyltransferase, designated pp-GalNAc-T13, that is specifically expressed in neurons and synthesizes GalNAc alpha-serine/threonine antigen.

To date, 10 members of the UDP-N-acetyl-alpha-d-galactosamine:polypeptide N-acetylgalactosaminyltransferase (pp-GalNAc-T) family have been cloned and analyzed in human. In this study, we cloned and analyzed a novel human pp-GalNAc-T from an NT2 cell cDNA library, and we named it pp-GalNAc-T13. In amino acid sequences, pp-GalNAc-T13 was highly homologous, showing 84.3% identity, to pp-GalNAc-T1. Real time PCR analysis revealed pp-GalNAc-T13 to be highly and restrictively expressed in the brain and present at very low or undetectable levels in other tissues, in contrast to the ubiquitous expression of pp-GalNAc-T1. pp-GalNAc-T13 was abundantly expressed in all neuroblastoma cells examined and primary cultured neurons but not in glioblastoma cells and primary cultured astrocytes. pp-GalNAc-T13 exhibited much stronger activity to transfer GalNAc to mucin peptides, such as Muc5Ac and MUC7, than did pp-GalNAc-T1. In addition, pp-GalNAc-T13 differed in substrate specificity to pp-GalNAc-T1. pp-GalNAc-T13 was able to form a triplet Tn epitope, three consecutive GalNAc-Ser/Thr structures, on peptides encoded in syndecan-3, a proteoglycan expressed in neurons. pp-GalNAc-T13-deficient mice have been established in a previous work. Immunohistochemical study showed a remarkable decrease in Tn antigen expression in the cerebellum of the pp-GalNAc-T13 knockout mouse. pp-GalNAc-T13 would be a major enzyme responsible for the synthesis of O-glycan and specifically the Tn antigen epitope in neurons.


To date, 10 members of the UDP-N-acetyl-␣-D-galactosamine:polypeptide N-acetylgalactosaminyltransferase (pp-GalNAc-T) family have been cloned and analyzed in human.
In this study, we cloned and analyzed a novel human pp-GalNAc-T from an NT2 cell cDNA library, and we named it pp-GalNAc-T13. In amino acid sequences, pp-GalNAc-T13 was highly homologous, showing 84.3% identity, to pp-GalNAc-T1. Real time PCR analysis revealed pp-GalNAc-T13 to be highly and restrictively expressed in the brain and present at very low or undetectable levels in other tissues, in contrast to the ubiquitous expression of pp-GalNAc-T1. pp-GalNAc-T13 was abundantly expressed in all neuroblastoma cells examined and primary cultured neurons but not in glioblastoma cells and primary cultured astrocytes. pp-Gal-NAc-T13 exhibited much stronger activity to transfer GalNAc to mucin peptides, such as Muc5Ac and MUC7, than did pp-GalNAc-T1. In addition, pp-GalNAc-T13 differed in substrate specificity to pp-GalNAc-T1. pp-Gal-NAc-T13 was able to form a triplet Tn epitope, three consecutive GalNAc-Ser/Thr structures, on peptides encoded in syndecan-3, a proteoglycan expressed in neurons. pp-GalNAc-T13-deficient mice have been established in a previous work. Immunohistochemical study showed a remarkable decrease in Tn antigen expression in the cerebellum of the pp-GalNAc-T13 knockout

mouse. pp-GalNAc-T13 would be a major enzyme responsible for the synthesis of O-glycan and specifically the Tn antigen epitope in neurons.
O-Linked glycosylation of mucin is initiated by the transfer of N-acetylgalactosamine with an ␣-linkage to a serine or threonine residue in protein. The GalNAc residue attached to the peptide is usually extended to form more complex O-glycan structures by the action of multiple glycosyltransferases. The addition of GalNAc is controlled by multiple members of the pp-GalNAc-T 1 (EC 2.4.1.41) family. To date, 10 distinct members have been identified in human (pp-GalNAc-T1, -T2, -T3, -T4, -T6 to -T9, -T11, and -T12) (1-10) and 7 in rodent (ppGaN- Tase-T1, -T2, -T3, -T4, -T5, -T7, -T10 2 ) (11)(12)(13)(14)(15)(16)(17). All share a highly homologous primary sequence, 40 -60% homology at the amino acid level, to each other particularly in the predicted catalytic domain. Each member exhibits different substrate specificity toward peptide sequences. Thus, the positions of O-glycan in proteins are determined by the substrate specificity of each pp-GalNAc-T. The genes encoding these enzymes are distributed at different genomic localizations on chromosomes and have distinct genomic structures (4).
pp-GalNAc-T1, which was the first pp-GalNAc-T to be cloned from bovine tissue (1), is the best characterized of the members. It shows activity toward Muc1, Muc2, Muc5Ac, and Muc7 peptides (4,18). Peptide sequences derived from these mucins were used to determine the site of the O-glycosylation of the Muc2 and Muc5Ac mucin tandem repeat region by pp-GalNAc-T1 (19 -22).
Many pp-GalNAc-Ts, including pp-GalNAc-T1, were found to * This work was performed as a part of the R&D Project of the Industrial Science and Technology Frontier Program (R&D for Establishment and Utilization of a Technical Infrastructure for Japanese Industry) supported by the New Energy and Industrial Technology Development Organization. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AB078142 and AB082928.
ʈʈ To whom correspondence should be addressed: Glycogene Function Team, Research Center for Glycoscience, National Institute of Advanced Industrial Science and Technology, Central-2, Open Space Laboratory, 1-1-1 Umezono, Tsukuba, Ibaraki 305-8568, Japan. Tel.: 81-298-61-3200; Fax: 81-298-61-3201; E-mail: h.narimatsu@aist.go.jp. be distributed in digestive organs, lymphatic organs, and peripheral blood cells (3)(4)(5) that produce various species of O-glycosylated protein. pp-GalNAc-T9 is an exception in that it is mainly expressed in the brain; however, its activity has yet to be examined (8). Very recently, it was reported that certain proteins in neurons, such as a neurite outgrowth receptor, human ␣-synuclein related to Parkinson's disease, and human syndecan-3, are modified by O-linked glycosylation (23,24). It is of interest to identify the pp-GalNAc-T(s) responsible for the O-linked glycosylation of proteins in neurons.
The shortest sequence motif, XTPXP, required for high level mucin-type O-glycosylation, was discovered using semi-purified pp-GalNAc-T from bovine colostrum (25). However, it has not been determined whether this motif is suitable for all pp-GalNAc-Ts, some of which were not cloned until the motif was found. Thus, it is now important to determine the peptide motif specifically recognized by each pp-GalNAc-T.
pp-GalNAc-T13 has a very high level of sequence identity to pp-GalNAc-T1. This resulted previously in the establishment of mice lacking an intact gene for pp-GalNAc-T13 by using the pp-GalNAc-T1 gene as a probe for cloning, and prior to evidence for multiple pp-GalNAc-T isozymes (26,27). In the previous study (27), the complete gene encoding pp-GalNAc-T13, which was partially identified and tentatively designated mouse pp-GalNAc-T8, was never cloned and characterized. In this report, we cloned and characterized a pp-GalNAc-T13 that is expressed exclusively in neuronal cells. We examined the substrate specificities of both enzymes, pp-GalNAc-T13 and -T1, toward peptides encoded in the sequences of O-glycosylated proteins in digestive organs and neuronal cells.

EXPERIMENTAL PROCEDURES
Tumor Cell Lines-The tumor cell lines were cultured in RPMI 1640 medium (Invitrogen) supplemented with 10% fetal bovine serum (FBS). The cell lines were donated or else purchased from American Tissue Culture Collection, Riken Cell Bank, Japan Cell Research Bank, Immunobiological Laboratory, or Dai-Nippon Pharmaceutical Co., Ltd.
Isolation of Human pp-GalNAc-T13 cDNA and Mouse pp-GalNAc-T13 cDNA-In the data base of the full-length human cDNA sequencing project supported by the New Energy and Industrial Technology Development Organization, we found a cDNA encoding a full-length sequence of pp-GalNAc-T13, which is highly homologous to human pp-Gal-NAc-T1 (2). In the full-length human cDNA sequence project, an oligocapped, full-length enriched cDNA library of human NT2RI cells was constructed using an expression vector pME18SFL3 according to the oligo-capping method (28). This novel sequence (2428 bp) of pp-GalNAc-T13 encoded a full open reading frame from bp 527 to 2198. The full-length sequence of pp-GalNAc-T13 was deposited into GenBank TM / EBI Data Bank with the accession number AB078142.
In a mouse EST data base, we found four EST sequences (Gen-Bank TM accession numbers BF467081, BE859530, BB623792, and BB304998) being highly homologous (92.5% homology at the nucleic acid level and 99.3% at the amino acid sequence level) to human pp-GalNAc-T13. In this study, the full-length sequence of mouse pp-GalNAc-T13 was registered in the GenBank TM /EBI Data Bank with the accession number AB082928.
Expression and Purification of Recombinant pp-GalNAc-T13 Expressed in Insect Cells-The method used to express the soluble form of the enzyme was almost the same as that described in our previous paper (29). Truncated coding regions of pp-GalNAc-T13 and -T1, starting at the same amino acid, residue 40, were constructed downstream of the human immunoglobulin secretion signal and the FLAG epitope tag in the vector pFBIF. An expression construct of pp-GalNAc-T13 was prepared by PCR using primer RO17F (5Ј-GGGGACAAGTTTGTACA-AAAAAGCAGGCTTCAGATCTCTGCTGCCTGCATTGAGG-3Ј) and O17R (5Ј-GGGGACCACTTTGTACAAGAAAGCTGGGTCCTATGTGC-CCAAGGTCATGTTCCTTAG-3Ј). The amplified fragment was cloned into pFBIF, and the construct was used to produce recombinant virus, which was in turn used to infect Sf21 (Pharmingen) or High Five TM insect cells (Invitrogen) and amplified. The infected Sf21 or High Five TM insect cells were incubated at 27°C until the survival rate of the cells was less than 30 -50% to yield conditioned media containing recombinant pp-GalNAc-T13 protein fused with the FLAG peptide. The recombinant protein was readily purified using anti-FLAG M1 antibody resin (Sigma) and eluted with a 50 mM Tris-buffered saline (50 mM Tris-HCl, pH 7.4, 150 mM NaCl) buffer containing 2 mM EDTA. The FLAG-tagged protein purified from the mock-transfected cells did not show any activity (data not shown).
To compare the activity of pp-GalNAc-T13 to those of human pp-GalNAc-T1 and -T9, recombinant forms of the latter two were also prepared. For pp-GalNAc-T1, the sequence was cloned into pFBIF by PCR with the primer RT1F120 (5Ј-GGGGACAAGTTTGTACAAAAAA-GCAGGCTTCAGAGGACTTCCTGCTGGAGATGTT-3Ј) and OT1R (5Ј-GGGGACCACTTTGTACAAGAAAGCTGGGTCCTAGAATATTTCTG-GCAGGGTGACGTT-3Ј) and then expressed in the High Five insect cells. A soluble form of human pp-GalNAc-T9 was produced in the same way. The primers used to clone pp-GalNAc-T9 were RT9F115 (5Ј-GG-GGACAAGTTTGTACAAAAAAGCAGGCTTCCGCATCGTGAGCGGC-GACCGCCGGT-3Ј) and OT9R (5Ј-GGGGACAAGTTTGTACAAAAAAG-CAGGCTTCGAAGGAGATAGAACCATGGCGGTGGCCAGGAAGATC-CGA-3Ј). Recombinant pp-GalNAc-T13, -T1, and -T9 proteins were separated by 10% SDS-PAGE and transferred to a polyvinylidene difluoride (PVDF) membrane (Bio-Rad). Bands of interest were detected using anti-FLAG monoclonal antibody with Western blotting to determine the relative amounts of recombinant enzymes.
Standard assays were performed in a 20-l reaction mixture containing 25 mM Tris-HCl, pH 7.4, 10 mM Mn 2 Cl, 0.2% Triton X-100, 250 M UDP-GalNAc (Sigma), and 0.01-2 mM of acceptor peptide. The enzyme reaction was performed for an appropriate period at 37°C and terminated by boiling. The reaction mixture was analyzed by HPLC (Shi- madzu, Kyoto, Japan) using a Cosmosil 5C 18 -AR reverse phase column (4.6 ϫ 250 mm) with a flow rate of 1 ml/min and a linear gradient of 0 -50% acetonitrile, 0.05% trifluoroacetic acid in 30 min at 40°C. FAMlabeled glycopeptides were detected by fluorescence detector (RF-10A XL , Shimadzu) at an emission wavelength of 520 nm. K m values for acceptor substrates were calculated using peptides with concentrations from 0.01 to 2 mM in the presence of 0.5-1.0 mM UDP-GalNAc. The Michaelis-Menten equation was applied to the initial rate data using a non-linear least square regression program (34). Quantitative Analysis of pp-GalNAc-T13 and -T1 Transcripts in Human Tissues and Cell Lines by Real Time PCR-For the quantification of pp-GalNAc-T13 and -T1 transcripts, we employed the real time PCR method, as described in detail previously (29,35,36). Marathon Ready TM cDNAs derived from various human tissues were purchased from Clontech. cDNAs derived from many different cells were prepared as in our previous study (37). Standard curves for the pp-GalNAc-T13 and -T1 genes and the endogenous control, glyceraldehyde-3-phosphate dehydrogenase (GAPDH), were generated by the serial dilution of pBluescript II SK(ϩ) vectors containing the pp-GalNAc-T13 and -T1 genes and a pCR2.1 vector (Invitrogen) containing the GAPDH cDNA, respectively. The PCR mixture was prepared using a TaqMan Universal Master Mix (PE Applied Biosystems). Primer sets and probes were as follows: for pp-GalNAc-T13 gene, a forward primer (T13-F) encoded in exon 6 -7, 5Ј-ATCTTGAAATGTCTTTTAGGATTTGGC-3Ј; a reverse primer (T13-R) encoded in exon 8 -9, 5Ј-TGGTTTCAACATTTCT-TATCTCACCA-3Ј; and a probe (T13-P) in exon 7, 5Ј-CCTGGTG-GCACTGGTCATGTCATCAACAAG-3Ј with a minor groove binder (38); for pp-GalNAc-T1 gene (GenBank TM accession number X85018), a forward primer (T1-F) in exon 1-2, 5Ј-GGACTTCCTGCTGGAGATGT-TCTA-3Ј; a reverse primer (T1-R) in exon 3-4, 5Ј-CTAAAGGCCTTT-TCAAAAAGTCTCTTTC-3Ј; and a probe (T1-P) in exon 3, 5Ј-CAATGAGGCTTGGAGCACACTTCTG CG-3Ј with the minor groove binder. Each real time PCR was performed for 50 cycles (95°C for 30 s, 60°C for 40 s, and 72°C for 30 s) using a real time PCR system (ABI PRISM 7700 Sequence Detection System, Applied Biosystems).

Measurement of pp-GalNAc-T13 and -T1 Transcripts by Semi-quantitative RT-PCR in Primary Cultured Neurons and Astrocytes from
Mouse Brain-cDNAs derived from primary cultured neurons and astrocytes from mouse were prepared as follows: E17 and P0 wild-type mouse brain was used for the primary culture of neurons and astrocytes, respectively, according to the method described previously (39). For the neuron culture, the cells were seeded to the poly-L-lysine-coated wells of 24-well plates at a density of 5 ϫ 10 5 cells/well in minimum essential medium (Invitrogen) containing 2% FBS. The neurons were cultured for 5 days at 37°C in 5% CO 2 . For the astrocyte culture, the cells were seeded to 75-cm 2 flasks at a density of 2.0 ϫ 10 7 cells/bottle in Dulbecco's modified essential medium containing 10% FBS and cultured at 37°C in 5% CO 2 with a change of medium twice a week. After 1 week, non-astrocytes were removed by shaking (40). The resultant astrocytic monolayer was removed and dissociated by trypsin treatment. The cell suspension was seeded onto 24-well plates at a density of 2 ϫ 10 5 /well or onto 75-cm 2 culture flasks at a density of 2 ϫ 10 7 cells. The astrocytes were cultured for a further 5 days.
The purity of the neurons and the astrocytes in each primary culture was ϳ97 and over 98%, respectively, as determined by immunostaining with anti-neurofilament antibody (Dako, A/S, Denmark) for neurons and with anti-glial fibrillary acidic protein antibody (Dako) for astrocytes.

Matrix-assisted Laser Desorption/Ionization-Time of Flight Mass Spectrometry (MALDI-TOF-MS) and
Amino Acid Sequencing of Glycopeptide Products-In order to determine the glycosylation sites of products of the pp-GalNAc-T13 and -T1 reactions, glycopeptides isolated by HPLC were subjected to MALDI-TOF-MS using REFLEX TM IV (Bruker Daltonics, Tsukuba, Japan) and Edman degradation on a protein sequencer PPSQ-23A (Shimadzu, Kyoto, Japan). Mass spectra were externally calibrated unless otherwise stated. The glycopeptides to be examined were dissolved in 0.1% trifluoroacetic acid to a concentration of 2 pmol/l. One microliter of the sample solution was mixed with 1 l of the matrix solution (2,5-dihydroxybenzoic acid, 10 mg/ml in 30% acetonitrile containing 0.1% trifluoroacetic acid) and then applied directly to the target. Simultaneously, 50 -100 pmol of the same sample was dried on PVDF membrane with Prosorb Sample Preparation Cartridge (Applied Biosystems) and spotted on BioBrene Plus (Applied Biosystems). Phenylhydantoin (PTH)-amino acid mixed standard (Wako, Osaka, Japan) was chromatographed using the isocratic program and monitored for absorbance at 269 nm. The PTH-Thr-O-GalNAc diastereomers were detected in two peaks at 3.07 and 3.25 min, which were near the position of dithiothreitol. The PTH derivative of Ser-O-GalNAc (Calbiochem) was identified as a peak at 2.96 min the same as the retention time of PTH-Glu.
Immunohistochemistry with HB-Tn1 Antibody-Brain tissues from pp-GalNAc-T13 null-mice and wild-type mice described in a previous study (26) were fixed in 10% buffered formalin for 24 h and transferred to 70% ethanol for the subsequent dehydration and paraffin-embedding process. Paraffin sections were prepared with a 5-m thickness, then deparaffinized, and rehydrated. After the blocking of endogenous peroxidase activity with 0.3% hydrogen peroxide for 30 min, the serial tissue sections were incubated overnight at 4°C with HB-Tn1 (1:50) antibody (Dako). After being rinsed with Tris-buffered saline, the tissue sections were incubated with Envisionϩ (Dako) for 30 min at room temperature. After another rinse, the peroxidase activity was visualized with the aid of a diaminobenzidine/hydrogen peroxide solution.
Counterstaining was carried out with hematoxylin. Anti-mouse IgM (Dako) was used for control experiments; no specific staining was found (data not shown).

RESULTS
cDNA Cloning and Sequence Analysis of Human and Mouse pp-GalNAc-T13-The alignment of the nucleotide sequence of human pp-GalNAc-T13 and mouse pp-GalNAc-T13 is shown in Fig. 1. The open reading frame sequences of pp-GalNAc-T13 in human and mouse exhibit 92.5% homologies. Both encode proteins of 556 amino acids, respectively, and only 4 amino acids are different from each other. This indicates that mouse pp-GalNAc-T13 (formerly designated mouse pp-GalNAc-T8) is the orthologous gene of human pp-GalNAc-T13.
The alignment of the amino acid sequences of human pp-GalNAc-T13 and -T1 using ClustalW is shown in Fig. 2A. The putative amino acid sequence of human pp-GalNAc-T13 encoded a 556-amino acid protein, which is predicted to be a type II membrane protein. The protein consists of an 8-amino acid Nterminal cytoplasmic region and a 20-amino acid transmembrane segment, followed by a stem region, a putative catalytic region, and a C-terminal region that possesses structural homology to the plant lectin ricin (42). As shown in Fig. 2A, pp-GalNAc-T13 has high sequence identity to pp-GalNAc-T1, 84.3% at the amino acid level, and retained both the DXH motif (position 209 -211), probably essential for divalent cation binding, and the DXXXXXWGGENXE motif (position 310 -322) in the catalytic domain. Both motifs are conserved in the other pp-GalNAc-T members. Three possible N-glycosylation sites were found in the primary sequence of pp-GalNAc-T13, two of which were also conserved at the same positions of pp-GalNAc-T1. We compared the genome structures of two genes, pp-Gal-NAc-T13 and -T1 (Fig. 2B). The cDNA sequence of pp-GalNAc-T13 was compared with a draft genome sequence (GenBank TM accession number NT_005311), which is localized on chromosome 2, encoding its genome sequence. Both genes contain at least 11 exons, the lengths of which were exactly the same between the genes. The pp-GalNAc-T13 gene spanned over 506 kb in the NT_005311 clone and is almost 10 times longer than the pp-GalNAc-T1 gene spanning about 49 kb in the NT_011029 clone which is localized to chromosome 18. For chromosome locations, mouse pp-GalNAc-T13 and -T1 are also localized to chromosomes 2 and 18, respectively.
Expression Levels of pp-GalNAc-T13 and -T1 Transcripts in Various Human Tissues and Cell Lines-The expression levels of pp-GalNAc-T13 and -T1 transcripts in various adult and fetal human tissues and tumor cell lines were determined by the real time PCR method. The pp-GalNAc-T1 transcripts were ubiquitously expressed in all tissues examined as reported previously (3) (Fig. 3). It was notable that it was almost 10 times more abundant than the pp-GalNAc-T13 transcript (the scale bar of the transversal axis indicates a 10-fold difference). However, the expression level of pp-GalNAc-T13 was highest in the fetal brain, followed by the adult cerebellum, cerebral cortex, and whole brain. The transcript was negligibly expressed or undetectable in the other tissues (Fig. 3). pp-GalNAc-T1 was also ubiquitously expressed in all human cell lines examined, whereas pp-GalNAc-T13 was expressed exclusively in neuroblastoma cells being undetectable in all other cell lines including glioblastoma cells (Fig. 4).
Exclusive Expression of Mouse pp-GalNAc-T13 Transcripts in Neurons-As same as the expression pattern in human, the mouse pp-GalNAc-T13 transcript was expressed only in brain, which was determined by the Northern blot analysis with total RNA from various mouse tissues (data not shown). To confirm that pp-GalNAc-T13 is expressed exclusively in neuronal cells, we employed a primary culture system of mouse neurons and astrocytes. Expression levels of mouse pp-GalNAc-T13 and -T1 in primary cultured neurons and astrocytes were determined semi-quantitatively by RT-PCR. As seen in Fig. 5, the amount of pp-GalNAc-T13 expressed in neurons was less than that of pp-GalNAc-T1 consistent with the results in the previous experiments. The pp-GalNAc-T13 transcript was not detected in astrocytes, whereas pp-GalNAc-T1 was abundantly expressed in both neurons and astrocytes.
These results suggested that the expression of pp-GalNAc-T13 is limited to neuronal cells and not to glial cells in the brain.
Difference of Substrate Specificity between pp-GalNAc-T13 and -T1-As shown in Fig. 6A, the FLAG-fused truncated forms of both pp-GalNAc-T13 and -T1 were detected by Western blotting with anti-FLAG antibody. The intensity of the bands corresponding to a molecular mass of 64.4 kDa was measured by densitometer to determine the relative amounts of each enzyme. The amounts of each enzyme were made equal for assaying the pp-GalNAc-T activity. To compare the substrate specificity and the kinetic properties of pp-GalNAc-T13 and -T1 for the GalNAc-transfer, we first employed a FAMlabeled Muc5Ac peptide (FAM-GTTPSPVPTTSTTSA) as an acceptor substrate. This substrate contains 9 threonine and serine residues, which are potential sites for the GalNAc transfer. The reaction products were analyzed by HPLC at different time points during the 24-h reaction. The HPLC patterns of reaction products with a 50% amount of substrate remaining in each enzyme reaction were compared and shown in Fig. 6B. pp-GalNAc-T13 and -T1 produced a peak (P1) as the first product at the same retention time (20.98 min). Subsequently, pp-GalNAc-T13 and -T1 produced second peaks, P2 and P2Ј, which showed different retention times, 20.53 and 20.28 min, respectively. Although we could not determine the GalNAc positions by peptide sequencer because of N-terminal block by FAM, peaks were determined as a mono-glycosylated Muc5Ac (P1) and as di-glycosylated Muc5Ac peptides (P2 and P2Ј) by MALDI-TOF MS (data not shown). These results demonstrate that two highly homologous enzymes, pp-GalNAc-T13 and -T1, show the different substrate specificities for GalNAc transfers to Muc5Ac. A significant difference between the activities of the two enzymes was found when a FAM-labeled Muc7 peptide (PTP-SATTPAPPSSSAPPETTAAK-FAM) was used as an acceptor substrate (Fig. 7A). pp-GalNAc-T1 produced the first distinct peak (P1) at around 0.5 h of incubation but no additional peaks even after 96 h, whereas pp-GalNAc-T13 showed activity to produce P1 which appeared after 10 min of incubation, followed by the production of additional peaks. The P1 peaks of both the pp-GalNAc-T1 and -T13 products were isolated by HPLC. Each P1 peak was subjected to mass spectrometry and protein sequencing analyses and was determined to have a single Gal-NAc residue at the 7th threonine of the peptide, i.e. mono-GalNAc/T 7th -Muc7 (Table II). To ascertain the secondary activity of both enzymes, the mono-GalNAc/T 7th -Muc7 (P1) produced by pp-GalNAc-T1 was isolated and used as the acceptor substrate for fresh enzyme reaction again. When fresh enzyme was added, pp-GalNAc-T1 did not produce any new peaks even after prolonged incubations (Fig. 7B). However, pp-GalNAc-T13 exhibited apparent and strong secondary and tertiary activities producing multiple new peaks. In addition, the activities of both enzymes remaining after 96 h of incubation were examined by the addition of fresh acceptor and donor substrates. As the substrate peak area decreased with the concom- itant increase of the product peak area during 24 h of incubation, the enzymes were proved to be still active even after the first 96 h of incubation (data not shown). These results indicated that pp-GalNAc-T1 exhibits the primary activity to transfer GalNAc to the 7th threonine but not secondary activity for additional GalNAc transfers. In contrast, pp-GalNAc-T13 exhibits apparent secondary and tertiary activities to produce multiple peaks following the primary GalNAc transfer to the 7th threonine.
We isolated each peak of the pp-GalNAc-T13 products and determined the number and positions of GalNAc residues transferred on the peptide by mass spectrometry and protein sequencing, respectively (Table II). Mass spectrometric analysis revealed that P1, P2, P3, P4, and P5 possess one, two, three, four, and five GalNAc residues on the peptide, respectively. The order in which GalNAc was added to the Muc7 peptide by pp-GalNAc-T13 is summarized in Fig. 7C based on the results of the protein sequencing of each peak. The GalNAc residue was first transferred to the 7th threonine showing P1. P2 was determined to have two GalNAc residues at the 7th threonine and the 13th serine and was followed by the appearance of P3, which possess three GalNAc residues at the 2nd threonine in addition to the two GalNAcs of P2. P4 has four GalNAc residues and the new GalNAc addition was determined to be at the 14th serine. Finally, GalNAc was added to the 19th threonine resulting in the appearance of P5.
Comparison of Kinetic Properties of Enzymic Activity between pp-GalNAc-T13 and -T1-To investigate the effects of the concentration of the acceptor peptide, we determined several kinetic parameters of pp-GalNAc-T13, -T1, and -T9. As summarized in Table I, the K m value of purified recombinant pp-GalNAc-T13 was found to be slightly lower than that of pp-GalNAc-T1 for Muc1aЈ and similar for Muc7, but the V max values of pp-GalNAc-T13 for both acceptor peptides were found to be much higher than those of pp-GalNAc-T1. These results demonstrated that pp-GalNAc-T13 has stronger activity toward Muc1aЈ and Muc7 than pp-GalNAc-T1. The kinetic parameters of pp-GalNAc-T1, such as K m values for Muc1aЈ and Muc7, were similar to those in previous reports (4,18). The subtle difference in K m values for Muc1aЈ between our experiments (0.23 mM shown in Table I) and theirs (0.66 mM, (18)) may be due to the sequence of the peptide or FAM modification. However, the V max of pp-GalNAc-T1 was slightly lower than that of a previous report (maximum 20%), which should be caused by the different expression and assay systems. In contrast, the K m and V max values of pp-GalNAc-T9 for either Muc1aЈ or Muc7 were lower than those of pp-GalNAc-T13 and -T1.
Ability of pp-GalNAc-T13 to Transfer GalNAcs at Three Consecutive Amino Acid Positions on Human Syndecan-3-Human syndecan-3 is an integral membrane proteoglycan associated largely with epithelial cells and also known to be mostly expressed in the nervous system (43). As described in the previous section, the expression of pp-GalNAc-T13 is restricted to neurons. In this sense, syndecan-3 may be a natural substrate for pp-GalNAc-T13. It has been reported that syndecan-3 carries an antigenic epitope that reacts with an anti-Tn antibody, MLS128, and the reactivity is regulated during development in the mouse brain (24,44,45). The epitope structure recognized by MLS128 has also been reported to compose three or four consecutive sequences of GalNAc-Ser/Thr (46,47). It is of interest to determine which pp-GalNAc-T is responsible for the synthesis of the MLS128 epitope, comprising the clustered GalNAc residues, on syndecan-3. A search for potential Oglycosylation sites in the human syndecan-3 amino acid sequence (GenBank TM accession number AF248634) was made in the NetOGlyc 2.0 data base. This search identified three potential sites for the consecutive GalNAc attachment and additional O-glycosylated sites having the mucin box motif (25). Three peptides that contain the potential sites for the consecutive GalNAc attachment were prepared as acceptor substrates and named SDC106, SDC155, and SDC165 in which numbers indicate the position of the N-terminal amino acid of the peptide in human syndecan-3. One peptide, SDC284, having a typical T1 mucin box (XTPXP), and two peptides, SDC222 and SDC238, having neither a mucin box nor potential triplet-Tn epitope sites, were also prepared for controls.
pp-GalNAc-T13 was incubated for 24 h with SDC106 having a triplet of STT. As seen in Fig. 8A, pp-GalNAc-T13 gave three peaks of the reaction products, P1, P2, and P3. Each peak was isolated and subjected to MALDI-TOF MS and protein sequencing. P1, P2, and P3 were identified to have a molecular mass of 1546.3, 1749.3, and 1952.3 m/z, respectively, which indicated that they were mono-, di-, and tri-glycosylated SDC106 peptides, respectively (Fig. 8B).
As seen in Fig. 9, the amino acid sequencing demonstrated that P1 had a GalNAc residue at the 5th threonine, P2 had two GalNAc residues at the 5th and 4th threonines, and P3 had three GalNAc residues at the 5th and 4th threonine and the 3rd serine. The above results strongly demonstrated that pp-GalNAc-T13 is able to synthesize the triplet cluster of GalNAcs on syndecan-3, and the order in which GalNAc is transferred first at the 5th threonine, then at the 4th threonine, and finally at the 3rd serine. pp-GalNAc-T13 also synthesized a triplet cluster of GalNAcs on SDC155 (QRATTVS*T*T*MAK-FAM; asterisks indicates the positions of GalNAc transferred) and SDC165 (ATTAAT*S*T*GDPTVAK-FAM) as well as on SDC106 (data not shown). As shown in Table I, the K m and V max values of pp-GalNAc-T13 for transferring the first GalNAc, i.e. producing the P1 peak (Fig. 8A), to SDC106, SDC155, and SDC165 were determined. pp-GalNAc-T13 exhibited strong activity for the transfer of GalNAc to SDC106, -155 and -165, and quite strong activity toward SDC284 but not toward SDC222 and SDC238.
In contrast, pp-GalNAc-T1 did not transfer GalNAc to any peptides except for SDC284. pp-GalNAc-T9, which has been reported to be expressed in neuron (8), was examined for activity to synthesize the Tn epitope, because pp-GalNAc-T9 is also probably co-localized with syndecan-3 in neuronal cells. However, pp-GalNAc-T9 had no activity for the transfer of  Table II. The order of GalNAc transfers to the Muc7 peptide was indicated by circled numbers.
Immunohistochemical Detection of Tn Antigen in Brain Tissue of pp-GalNAc-T13 Null Mouse-In a previous study performed prior to direct evidence of multiple pp-GalNAc-T isozymes, mice lacking an intact pp-GalNAc-T13 gene, subsequently termed pp-GalNAc-T8, were established using the bovine pp-GalNAc-T1 gene as a probe during genomic cloning in the mouse. The high level of sequence identity between the two clones suggested an orthologous relationship (26,27). In this paper, however, the tentative mouse pp-GalNAc-T8 gene was revealed as the orthologous gene of human pp-GalNAc-T13. To investigate the ability of pp-GalNAc-T13 to synthesize Tn antigen in the brain, we performed an immunohistochemical study on the brain of pp-GalNAc-T13 null mice using HB-Tn1 antibody. So far, the expression of Tn antigen on syndecan-3 had been investigated in the developing mouse brain using MLS128 antibody (24,44). It was reported that the triplet Tn antigen is present in late embryonic brain with an expression peak in the early postnatal brain (postnatal 3-7 days, P3-P7), and immunoreactivity for MLS128 was restricted to the cell bodies of Purkinje cells in the P7 mouse brain (44). P7 mouse cerebellum and HB-Tn1 antibody were employed for immunohistochemical analysis in this experiment. As seen in Fig. 10A, the immunoreactivity of HB-Tn1 was detected in Purkinje cells of the P7 wild-type mouse, consistent with a previous report (44). In addition, cell bodies of neuronal cells in the internal granular layer also exhibited positive reactivity. According to the previous studies (24,44), internal granular layer cells were not likely stained with MLS128, and this was different from our present results using HB-Tn1. In contrast, the pp-GalNAc-T13 null mouse showed a remarkable decrease in HB-Tn1 reactivity both in Purkinje cells and in internal granular layer cells (Fig. 10B).
HB-Tn1 is commercially available, but the epitope recognized by HB-Tn1 has not been determined. To explain the  above immunohistochemical results, the reactivity of HB-Tn1 against glycopeptides was examined. As shown in Fig. 11A, mono-GalNAc/T-Muc1aЈ, mono-GalNAc/T-SDC106, di-Gal-NAcs/TT-SDC106, and tri-GalNAcs/STT-SDC106 were isolated by HPLC and used as antigens. The amount of each glycopeptide was made equal by monitoring the amount of FAM. As seen in Fig. 11B, HB-Tn1 positively reacted with all glycopeptides examined and LSC cell homogenates as a positive control. It did not react to the peptides without GalNAc residues. This demonstrated that HB-Tn1 can recognize not only Tn antigen of the mono-GalNAc structure but also multi-GalNAc Tn antigen including a di-GalNAcs/TT or tri-GalNAcs/STT structure. DISCUSSION The initiation of O-glycosylation is catalyzed by a family of pp-GalNAc-Ts. Studies over the past decade have found a family of pp-GalNAc-Ts that display differences in both their substrate specificity and their patterns of tissue distribution. In the present study, we found a novel isoform of the pp-GalNAc-T family, designated pp-GalNAc-T13. The pp-GalNAc-T13 sequence has very high similarity to the pp-GalNAc-T1 sequence, both at the nucleic acid and the amino acid level throughout the coding region. A comparison of gene structure revealed that both genes compose at least 11 exons, and the length of each exon is the same between the genes. However, the total length of the pp-GalNAc-T13 gene, which is located on chromosome 2, is almost 10 times that of the pp-GalNAc-T1 gene which is on chromosome 18. The longer intron lengths of pp-GalNAc-T13 may reflect different evolutionary rates between the two genes after they diverged.
Although pp-GalNAc-T13 and -T1 share highly homologous sequences, they showed different substrate specificities toward the acceptors examined. pp-GalNAc-T13 exhibited stronger activity toward all acceptor substrates examined than pp-Gal-NAc-T1. When Muc7 peptide was used as a substrate, a striking difference in substrate specificity was observed between the two enzymes. Edman degradation revealed that both enzymes produce P1 as the primary product on Muc7 by transferring GalNAc to the 7th threonine of the peptide. After producing P1, pp-GalNAc-T1 did not yield any other products even in prolonged reaction. In contrast, pp-GalNAc-T13 produced P1 more efficiently than did pp-GalNAc-T1 and catalyzed the conversion of P1 to P2 at a rather high rate, which was followed by the production of P3, P4, and P5. pp-GalNAc-T1 was confirmed not to have secondary activity toward Muc7 having a GalNAc residue by the experiment using the isolated P1 (Fig. 7B) as a substrate. It was of interest that pp-GalNAc-T1 exerted secondary activity toward some peptides, such as Muc5Ac as seen in Fig. 6B, and Muc2 (data not shown), but not toward Muc7. The consensus peptide sequence required for the secondary activity of pp-GalNAc-T1 will be examined more extensively in the future. Determining the order in which GalNAc was added by pp-GalNAc-T13 to the Muc7 peptide revealed two interesting points. First, GalNAc was first attached to the C-terminal threonine of the TT doublets. The activity to add GalNAc to the N-terminal threonine of the TT doublets was very weak. Second, the middle serine of the SSS triplet was the first of the three serine residues to be glycosylated. This was in contrast to the order in which GalNAc was added to form the Tn epitope on syndecan-3 by the same enzyme.
In the present study, we found that pp-GalNAc-T13 is expressed exclusively in the brain and neurons in human and mouse. To date, 10 human pp-GalNAc-Ts and 5 mouse pp-GalNAc-Ts have been analyzed as to their tissue distribution by Northern analysis (3-7, 13, 15-17). pp-GalNAc-T1 is expressed in all tissues both in human and rodent (3,13). The others, human pp-GalNAc-T2, -T4, -T7, -T8, and T12 and rodent -T7 and -T10 are also known to be expressed ubiquitously. However, the expression pattern of mouse pp-GalNAc-T4 is different from humans, and it is restrictively expressed in the digestive tissues (13). Mouse pp-GalNAc-T5 is also restrictively expressed in the digestive tissues (16). Human pp-GalNAc-T3 and -T6 is highly or restrictively expressed in pancreas and testis, and in placenta and trachea, respectively (6). pp-Gal-NAc-T11 is restrictively expressed in kidney (9). pp-GalNAc-T9 was found to be expressed mostly in brain (8). All of the tissuespecific pp-GalNAc-Ts could be considered to contribute to the O-glycosylation of specific substrates in certain tissues. However, their functions toward the tissue-specific substrates have not been identified. In this sense, it is noteworthy that pp-GalNAc-T13 apparently synthesized the triplet-Tn epitope on syndecan-3, but the other enzymes, pp-GalNAc-T1 and -T9, could not. It is suggested in the present study that a specific substrate was modified through GalNAc transfer by a pp-Gal-NAc-T in neuron. However, direct and careful elucidation should be required as done by Nehrke et al. (14).
The Tn antigen was first reported as a tumor-associated carbohydrate antigen (48 -50). The triplet Tn antigen was identified to be expressed on syndecan-3 spatiotemporally in the developing mouse brain. The expression patterns of seven pp-GalNAc-T family members, pp-GalNAc-T1, -T2, -T3, -T4, -T5, -T7, and -T10, 2 were investigated from gastrulation through organogenesis in mice (51). Except for -T1 and -T2, pp-Gal-NAc-Ts could not be clearly detected in the central nervous tissue. Although pp-GalNAc-T9 was identified in human brain tissue (8), it could not synthesize the triplet Tn antigen as demonstrated in the present study.
HB-Tn1 is a monoclonal antibody, which was raised against asialo-ovine submaxillary mucin as immunogen. It reacts with most breast, lung, gastric, and colon carcinomas, which express the Tn antigen (48,(52)(53)(54). So far, no one has investigated the epitopic structure of HB-Tn1 antibody. In the present study, we confirmed that HB-Tn1 reacts not only to the Tn antigen having a single GalNAc but also to the Tn antigen having multiple GalNAcs. In light of these results, both the triplet of GalNAcs on syndecan-3 and a single GalNAc in a mucin box must be recognized by HB-Tn1. The dramatic decrease in HB-Tn1 reactivity in the cerebellum of the pp-GalNAc-T13 null mouse indicates that pp-GalNAc-T13 is a major enzyme responsible for the synthesis of Tn antigen in neuronal cells. The residual faint reactivity of HB-Tn1 in the knockout mouse cerebellum is probably directed by the other pp-GalNAc-Ts such as pp-Gal-NAc-T1 and -T9. Together with the in vitro data, it is suggested that expression of Tn antigen having not only single GalNAc but also triplet GalNAcs may be abolished in knockout mouse.
The pp-GalNAc-T13 null mouse shows a normal outlook, and their behavior is also normal. However, neurological examination of the pp-GalNAc-T13 knockout mice would be highly required in the future.