The β1,3-Galactosyltransferase β3GalT-V Is a Stage-specific Embryonic Antigen-3 (SSEA-3) Synthase*

We have previously reported the molecular cloning of β1,3-galactosyltransferase-V (β3GalT-V), which catalyzes the transfer of Gal to GlcNAc-based acceptors with a preference for the core3 O-linked glycan GlcNAc(β1,3)GalNAc structure. Further characterization indicated that the recombinant β3GalT-V enzyme expressed in Sf9 insect cells also utilized the glycolipid Lc3Cer as an efficient acceptor. Surprisingly, we also found that β3GalT-V catalyzes the transfer of Gal to the terminal GalNAc unit of the globoside Gb4, thereby synthesizing the glycolipid Gb5, also known as the stage-specific embryonic antigen-3 (SSEA-3). The SSEA-3 synthase activity of β3GalT-V was confirmed in vivo by stable expression of the human β3GalT-V gene in F9 mouse teratocarcinoma cells, as detected with the monoclonal antibody MC-631 by flow cytometry analysis and immunostaining of extracted glycolipids. The biological relation between SSEA-3 formation and β3GalT-V was further documented by showing that F9 cells treated with the differentiation-inducing agent retinoic acid induced the expression of both the SSEA-3 epitope and the endogenous mouse β3GalT-V gene. This study represents the first example of a glycosyltransferase, which utilizes two kinds of sugar acceptor substrates without requiring any additional modifier molecule.

Glycosyltransferase enzymes mediate the synthesis of glycosylated structures by catalyzing the transfer of sugar units to various acceptor molecules utilizing nucleotide-activated sugars as donor substrates. In most cases, a single glycosyltransferase enables the transfer of one type of carbohydrate donor to one type of carbohydrate acceptor, although some exceptions have been reported. The lgtA glycosyltransferase from Neisseria meningitidis represents such an exception, as it can transfer two donors, GlcNAc and GalNAc, to Gal-based acceptors (1). Another notable exception is represented by the ␤1,4-galactosyltransferase-I enzyme, which is able to change its acceptor substrate specificity from GlcNAc to Glc by interacting with ␣-lactalbumin (2,3).
Previous work from this group and other laboratories has revealed a family of structurally related ␤1,3-glycosyltransferase enzymes (4 -10). During the course of our investigation, we have found that one member of this family, the ␤1,3-galactosyltransferase (␤3GalT) 1 -V enzyme that normally act on Gl-cNAc-based acceptors, can catalyze the transfer of galactose to GalNAc presented in the context of the Gb4 globoside structure. The resulting product, the Gb5 globoside, is also known as the stage-specific embryonic antigen (SSEA)-3 (11), which is found in the 4 -8 cell stage mouse embryo as well as some adult tissues. We have established the SSEA-3 synthase activity of ␤3GalT-V both in vitro and in vivo in F9 mouse teratocarcinoma cells as a model system.
Transfection of F9 Teratocarcinoma Cells-The mouse teratocarcinoma cell line F9 was obtained from ATCC (CRL-1720) and grown in Dulbecco's modified Eagle's medium with 10% fetal calf serum on tissue culture plates precoated with 0.1% gelatin (Sigma). The protein-coding region of the human ␤3GalT-V gene (10) was ligated into the BamHI and XhoI sites of the pcDNA3.1 ϩ expression vector (Invitrogen). 5 g of the resulting vector were introduced into F9 cells by LipofectAMINE (Life Technologies, Inc.)-mediated transfection. G418 (Life Technologies, Inc.) was added 24 h post-transfection at 600 g/ml, and surviving cell clones were picked after 10 days.
Glycolipid Analysis-Glycolipids were extracted from NCCIT (ATCC, CRL-2073) and ␤3GalT-V expressing F9 cells as described previously (13). Lipids were separated into neutral and ganglioside * This work was supported by the Swiss National Science Foundation Grant 31-58577.99 and by United States Public Health Service, National Institutes of Health Grants R01-NS24405 and PO1-HD05505. 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) AF254738.
Retinoic Acid Treatment-F9 cells were seeded at low density (5,000 cells/80-cm 2 dish) and treated with 10 Ϫ7 M all-trans-retinoic acid (Sigma) for 10 days. Cells were analyzed for SSEA-3 expression by flow cytometry as described above.
Cloning and Expression of the Mouse ␤3GalT-V Gene-A 129/SvJ mouse genomic -library (Stratagene) was screened as described previously (4) using the human ␤3GalT-V gene as a probe. The genomic region, including the sequence encoding the catalytic domain of the ␤3GalT-V enzyme, was amplified by PCR with the primers 5Ј-CAGCA-TGAATTCTTTCAGAGAACTCC-3Ј and 5Ј-GCAAGAGGATCCCAGAT-CGTCACAAA-3Ј and subcloned into the pFLAG-CMV-1 vector (Sigma). The medium of COS-7 cells transiently transfected with the ␤3GalT-V vector was incubated with 20 l of anti-FLAG M2-agarose beads (Sigma) for 2 h at 4°C. After washing three times with phosphatebuffered saline, the beads were directly used for galactosyltransferase assays. For metabolic labeling, transfected cells were starved for 2 h in methionine-free Dulbecco's modified Eagle's medium, then incubated in the presence of 0.1 mCi of [ 35 S]methionine for 1 h. After replacement with complete medium, cells were further incubated for 4 h at 37°C. After this time, the cell supernatant was collected and incubated with anti-FLAG M2-beads as described above.
Reverse Transcription (RT)-PCR-Total RNA was extracted from F9 cells using the procedure of Chomczynski and Sacchi (15). RNA samples were digested with RNase-free DNase-I (Sigma) before RT-PCR. The mouse ␤3GalT-V RNA (5 g) was reverse-transcribed using 1 unit of enhanced avian myeloblastosis virus reverse transcriptase (Sigma) and the primer 5Ј-AGTCCTGTTCTTTCGAGTTCTC-3Ј. Samples in a final volume of 20 l were incubated for 2 h at 42°C. PCR amplification was performed using 4 units of Taq polymerase (Sigma) with 10 l of the RT reaction product and the primers 5Ј-CAAGGATGCCTACTTCAA-CCTG-3Ј, 5Ј-CCAGGGAAGAAGGTCTGTTTGG-3Ј for 35 cycles at 95°C for 45 s, 55°C for 30 s, and 72°C for 60 s. Using the same RT-PCR conditions, as an internal control the mouse GAPDH mRNA was reverse-transcribed with the primer 5Ј-CCCTGTTGCTGTAGCCGTAT-TC-3Ј and amplified with primers 5Ј-TTGGCATTGTGGAAGGGCTCA-T-3Ј and 5Ј-CCCTGTTGCTGTAGCCGTATTC-3Ј. The human ␤3GalT-V RNA was detected in the NCCIT cell line (ATCC, CRL-2073) by RT-PCR with the RT primer 5Ј-GAAAGGATTTAGACTGTACATGC-3Ј and the PCR primers 5Ј-GAAAGGATTTAGACTGTACATGC-3Ј and 5Ј-GTGAA-TTCCTCTTTCTCTCTGCTG-3Ј using the same conditions as described above.

RESULTS
During our investigation of the donor and acceptor substrate specificity of the ␤3GalT-V enzyme, we have found that this enzyme exhibited a preference for GlcNAc-based acceptors (10). However, when assaying glycolipid acceptors, the ␤3GalT-V enzyme showed activity toward the lacto-series glycolipid Lc3Cer and toward the globoside Gb4 (Fig. 1). The latter activity was surprising as the ␤3GalT-V enzyme displayed no activity toward the monosaccharide GalNAc(␣/␤-O-benzyl) acceptors (10). In addition, the Gb5 synthase activity of ␤3GalT-V did not require any modifier protein, since a FLAG-tagged soluble form of the ␤3GalT-V enzyme was capable of catalyzing the formation of Gb5 when affinity-purified on anti-FLAG M2agarose beads (Fig. 2). By comparison, the ␤3GalT-I, -II, -III enzymes characterized previously (4, 5, 7) did not display any dual activity toward GlcNAc and GalNAc-based acceptors (data not shown).
To establish whether the ␤3GalT-V enzyme is capable of catalyzing the synthesis of the Gb5 structure in vivo, we chose to overexpress the human ␤3GalT-V gene in F9 teratocarcinoma cells. This cell line is known to express the Gb4 globoside but lacks the Gb5 synthase activity (16). F9 cells stably expressing the human ␤3GalT-V gene were first analyzed for the presence of the Gb5 product by flow cytometry. As shown in Fig. 3A, F9 cells expressing the human ␤3GalT-V gene showed increased levels of the SSEA-3 epitope at their surface when compared with F9 cells transfected with a mock vector. For a comparison, the human teratocarcinoma cell line NCCIT showed a high level of SSEA-3 expression as reported earlier (16). Immunostaining of glycolipids extracted from ␤3GalT-Vtransfected F9 cells also clearly indicated the synthesis of Gb5 structure in vivo (Fig. 3B). The globo-series ganglioside GL7, also termed SSEA-4, was also detected in the upper Folch fraction of the glycolipid extract in the transfected cells by immunostaining with MC-631 antibody (data not shown), thus indicating that the Gb5 product was further sialylated in F9 cells.
The relation between ␤3GalT-V and the induction of SSEA-3 was further investigated in wild-type F9 cells. Treatment of F9 teratocarcinoma cells with retinoic acid is known to promote cell differentiation and to induce the expression of the SSEA-3 epitope (17) (Fig. 4A). To determine whether the increase in SSEA-3 levels is paralleled by the induction of ␤3GalT-V gene transcription, we first cloned and expressed the mouse ␤3GalT-V (GenBank TM accession number AF254738). The mouse protein is 308 amino acid long and shows 71% identity to the  ␤3GalT-V as a SSEA-3 Synthase human ␤3GalT-V at the amino acid level. The mouse ␤3GalT-V enzyme exhibited a galactosyltransferase activity when transiently expressed in COS-7 cells (data not shown). The presence of ␤3GalT-V mRNA in F9 cells was analyzed by RT-PCR. As shown in Fig. 4B, F9 cells treated with retinoic acid for 10 days were positive for ␤3GalT-V transcript, whereas untreated F9 cells remained negative. As a comparison, human ␤3GalT-V mRNA was also detected in NCCIT cells, which constitutively express SSEA-3 at their surface (see Fig. 3A). This experiment demonstrated that the expression of the ␤3GalT-V gene correlates with the induction of the SSEA-3 epitope. DISCUSSION In the present study, we report a dual acceptor specificity for the ␤3GalT-V enzyme by establishing its ability to transfer Gal to the terminal GalNAc residue of Gb4 in addition to its activity toward GlcNAc-based acceptors (9,10). This represents one of the few glycosyltransferases reported to date, which are capable of transferring a donor sugar molecule to different terminal monosaccharide acceptors. The best documented case is the ␤1,4-galactosyltransferase-I enzyme, which is able to use both GlcNAc and Glc as substrates (2,3). However, whereas the ␤1,4-galactosyltransferase-I enzyme must interact with another modifier molecule, ␣-lactalbumin, to switch the substrate specificity from GlcNAc to Glc, the ␤3GalT-V enzyme does not require any additional modifier to switch from one acceptor substrate to the other.
␤3GalT-V is the first member of the recently cloned ␤3GalT family that utilizes Gb4 as an acceptor. While the tissue distribution pattern of the SSEA-3 epitope and its biochemical nature have been characterized for many years, a distinct ␤3GalT enzyme with SSEA-3 synthase activity has yet to be cloned. It remains unclear whether ␤3GalT-V is the only enzyme that can perform this role in vivo. Also, it must be established whether the SSEA-3 synthase activities from human teratocarcinoma cells (18) and from mouse kidney (19) represent orthologous proteins.
The human and mouse ␤3GalT-V genes displayed differences in their tissue expression patterns. The human ␤3GalT-V gene was mainly expressed in small intestine, pancreas, and testis (10), and mouse ␤3GalT-V transcripts were mainly detected in brain and kidney but not in testis (data not shown). Noteworthy, expression of SSEA-3 has been reported to be confined to the kidneys in adult mice (20). However, it is difficult to relate the pattern of expression of the ␤3GalT-V gene with the distribution of SSEA-3 among tissues, since additional factors, such as the availability of the Gb4 acceptor, will also affect the presence or absence of the antigen on specific cell types.
The parallel induction of SSEA-3 and ␤3GalT-V expressions in F9 cells treated with retinoic acid supports the concept of a direct relation between both events. Nevertheless, we cannot exclude the existence of other ␤3GalT enzymes, which may participate to the formation of SSEA-3 in vivo. On the other hand, it is also possible that the SSEA-3 synthase activity indeed represents a second activity of the type-1 chain synthase ␤3GalT-V enzyme. In this case, the enzyme may function either way depending on the levels of the respective acceptor available in cells. The targeted inactivation of the mouse ␤3GalT-V gene will provide answers to this question and may represent a valuable model to address the biological functions associated to SSEA-3 during embryonic development. ␤3GalT-V as a SSEA-3 Synthase