Structure and expression of H-type GDP-L-fucose:beta-D-galactoside 2-alpha-L-fucosyltransferase gene (FUT1). Two transcription start sites and alternative splicing generate several forms of FUT1 mRNA.

The expression of the ABO antigens on erythrocyte membranes is regulated by H gene (FUT1)-encoded a(1,2)fucosyltransferase activity. We have examined the expression of the FUT1 in several tumor cell lines, including erythroid lineage and normal bone marrow cells, by Northern blot and/or reverse transcription-polymerase chain reaction (RT-PCR) analyses. RT-PCR indicated that bone marrow cells, erythroleukemic cells (HEL), and highly undifferentiated leukemic cells (K562) that have erythroid characteristics expressed the FUT1 mRNA while four leukemic cell lines did not. The FUT1 mRNA was also demonstrated in gastric, colonic, and ovarian (MCAS) cancer cell lines by RT-PCR. Northern blot analysis indicated that a 4.0-kilobase FUT1 transcript was expressed in some of these tumor cell lines. Rapid amplification of 5* cDNA end (RACE) analysis suggested that the FUT1 transcript had several forms generated by two distinct transcription start sites and alternative splicing. The results of RT-PCR using specific primers for each starting exon suggested that two transcription initiation sites (exon 1A and exon 2A) of the FUT1were identified in gastric cancer cells and in ovarian cancer cells. Only exon 1A was identified as a transcription start site in another gastric cancer cell line, two colonic cancer cell lines, and in K562 cells, whereas only exon 2A was identified in HEL cells and in bone marrow cells. These two transcription start sites were located 1.8 kilobases apart. Therefore, two distinct promoters appeared to be present in the FUT1. The distinct promoters of the FUT1 and alternative splicing of the FUT1 mRNA may be associated with timeand tissue-specific expression of the FUT1.

The expression of the ABO antigens on erythrocyte membranes is regulated by H gene (FUT1)-encoded ␣(1,2)fucosyltransferase activity. We have examined the expression of the FUT1 in several tumor cell lines, including erythroid lineage and normal bone marrow cells, by Northern blot and/or reverse transcription-polymerase chain reaction (RT-PCR) analyses. RT-PCR indicated that bone marrow cells, erythroleukemic cells (HEL), and highly undifferentiated leukemic cells (K562) that have erythroid characteristics expressed the FUT1 mRNA while four leukemic cell lines did not. The FUT1 mRNA was also demonstrated in gastric, colonic, and ovarian (MCAS) cancer cell lines by RT-PCR. Northern blot analysis indicated that a 4.0-kilobase FUT1 transcript was expressed in some of these tumor cell lines. Rapid amplification of 5 cDNA end (RACE) analysis suggested that the FUT1 transcript had several forms generated by two distinct transcription start sites and alternative splicing. The results of RT-PCR using specific primers for each starting exon suggested that two transcription initiation sites (exon 1A and exon 2A) of the FUT1 were identified in gastric cancer cells and in ovarian cancer cells. Only exon 1A was identified as a transcription start site in another gastric cancer cell line, two colonic cancer cell lines, and in K562 cells, whereas only exon 2A was identified in HEL cells and in bone marrow cells. These two transcription start sites were located 1.8 kilobases apart. Therefore, two distinct promoters appeared to be present in the FUT1. The distinct promoters of the FUT1 and alternative splicing of the FUT1 mRNA may be associated with time-and tissue-specific expression of the FUT1.
The ABO(H) histoblood group antigens are oligosaccharides (1), and their biosynthesis is regulated by several glycosyltransferases that add monosaccharides to a precursor molecule in a sequential fashion (2,3). The ␣(1,2)fucosyltransferase 1 that forms the H antigen, an essential precursor of the A and B antigens, plays a regulatory role in the tissue expression of ABO antigens. These antigens are found not only on erythrocyte membranes but also in most epithelial cells and in body fluids. Several lines of evidence have indicated that at least two distinct ␣(1,2)fucosyltransferases are present in human tissues (4 -9). One is the H gene (FUT1)-encoded ␣(1,2)fucosyltransferase (H enzyme) and the other is the Secretor gene (FUT2)encoded ␣(1,2)fucosyltransferase (Se enzyme). The H enzyme regulates the expression of the H antigen mainly on erythrocyte membranes, while the Se enzyme regulates the expression of the H antigen mainly in epithelial cells and in body fluids such as saliva (2,3,10).
It is well known that the glycosylation patterns including ABH antigens are changed during embryonic development, cell maturation, and malignant transformation (1,11). In early embryos, the ABH antigens are expressed on cell surfaces of red blood cells and of endothelial and epithelial cells of most organs. The surface expression of the ABH antigens on epithelial cells reaches a maximum at about nine weeks and thereafter decreases. Some cells such as the neurons, muscle cells, and bone cells completely lose the capacity to synthesize ABH antigens. On digestive mucosa, a decrease in the cell surface expression of ABH antigens coincides with the onset of mucus secretion (12). One of the well examined digestive mucosa for the expression of ABH and related antigens is colorectal mucosa. While the ABH, Lewis y, and Lewis b are expressed in fetal distal colorectal mucosa, these antigens disappear in adult distal colorectal mucosa (13). However, these antigens are re-expressed in colorectal carcinoma (13). The expression of H, Lewis y, and Lewis b antigens in colorectal carcinoma are thought to be regulated by both H and Se enzymes (13), and the level of the FUT1 transcript was increasing during malignant transformation in colorectal mucosa (14). Therefore, analyses of the gene structure and promoter region that regulate the expression of the ␣(1,2)fucosyltransferase genes are important for understanding stage-and tissue-specific expression of the H and H-related antigens.
Recently, the FUT1, FUT2, and a FUT2-pseudogene (Sec1) have been isolated (15,16). These genes and a pseudogene share a high degree of DNA sequence homology and are located within a 100-kb region on chromosome 19q13.3 (17), suggesting that they were generated by gene duplication from the same ancestor gene and then subsequently diverged. Recently, Kelly et al. (18) reported the gene structure and transcription initiation site of the FUT1 in A431 cells. Here, we have examined the expression of the FUT1 and found several forms of FUT1 * This work was supported by a Grant-in-aid for Scientific Research from the Ministry of Education, Science and Culture, Japan. 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 DDBJ and GeneBank TM /EBI Data Bank with accession numbers D87935-D87941, D87943, and D87944.

MATERIALS AND METHODS
Cell Lines, RNA Isolation and Northern Blot Analysis-Human cancer cell lines MCAS (ovarian cancer), COLO201 and WiDr (colon cancer), KATOIII and MKN74 (gastric cancer), and HEL (erythroleukemia) cells were obtained from the Health Science Research Resources Bank, Osaka, Japan. Human hematopoietic cell lines K562 (chronic myelogenous leukemia, blast crisis), HL60 (promyelocytic leukemia), U937 (histiocytic lymphoma), BALL-1 (B-cell lymphoblastic leukemia), and MOLT-4 (T-cell lymphoblastic leukemia) cells were a kind gift from Dr. K. Sagawa (Kurume University). Total cytoplasmic RNA was isolated from these cells using the acid guanidinium thiocyanate/phenol/chloroform method (19). For Northern blot analysis, total RNA (20 g) was denatured, divided into 10 g/lane, and separated by 1.2% formaldehyde-agarose gel electrophoresis and then transferred onto a nylon membrane (Hybond Nϩ, Amersham International, Tokyo) (20,21). The membrane was divided to prepare duplicate membranes. One membrane was stained by 0.04% methylene blue in 0.5 M sodium acetate (pH 5.2) (22). The other membrane was hybridized with digoxigenin-labeled FUT1 antisense RNA probe for the total protein coding region. The digoxigenin-labeled RNA probe was prepared using the DIG RNA labeling kit (Boehringer Mannheim, Mannheim, Germany). Hybridization was carried out at 68°C with 5 ϫ SSC, 50% formamide, 2 ϫ Denhardt's reagent, and 0.5% SDS overnight. The final wash was carried out at 68°C with 0.2 ϫ SSC and 0.1% SDS for 40 min. Detection of hybridized bands was performed using a DIG luminescent detection kit for nucleic acids (Boehringer Mannheim).
Reverse Transcription PCR (RT-PCR)-Synthesis of single strand cDNA of several kinds of cultured cells was performed on total RNA (2 g) using SUPERSCRIPT preamplification system (Life Technologies, Inc.) according to the manufacturer instructions. 1 l (total 20 l) of resultant single strand cDNA was used as the template for PCR. The RT-PCR primers for the amplification of FUT1 (15) and glyceraldehyde-3-phosphate dehydrogenase (G3PDH) gene (24) are listed in Table I. The temperature profile of all RT-PCRs was as follows: denature at 94°C for 20 s, annealing at 65°C for 1 min, extension at 72°C for 1 min, and 25 cycles for G3PDH and 30 cycles for FUT1. The starting exonspecific RT-PCR was performed using each distinct starting exon-specific primer (Table II) and the 3Ј-FUT1 primer (Table I) (denature at 98°C for 10 s, annealing at 68°C for 1 min, extension at 72°C for 2 min, and 30 cycles). For identification of starting exons of the FUT1 mRNA in the normal erythroid progenitor cells, first strand cDNA synthesis was performed on 500 ng of human bone marrow poly(A) ϩ RNA (Clontech, Palo Alto, CA), as mentioned above. Then the starting exonspecific PCR was performed. The products were analyzed by 1.0% agarose gel electrophoresis and stained with ethidium bromide. No PCR product was amplified in the control reactions without reverse transcriptase.
5Ј-RACE (Rapid Amplification of cDNA Ends)-Poly(A) ϩ RNA was isolated from total cellular RNA (100 -500 g) by Oligotex-dT30 (Takara). Double-stranded cDNA synthesis and adapter ligation were performed using the Marathon cDNA amplification kit (Clontech). 5 l of the double strand cDNA (250-fold diluted) were used as the template for the RACE analysis. The first PCR was performed using 3Ј-FUT1 primer (Table I) and the AP 1 primer (provided by the supplier). Nested PCR was performed using FUT1 nest primer (Table I) and the AP 2 primer (provided by the supplier). The temperature profile of all PCR for RACE analysis was as follows: denature at 98°C for 10 s, annealing and extension at 68°C for 3 min, and 30 cycles. The 5Ј-RACE product was cloned into pGEM using the pGEM-T vector system I (Promega, Madison, WI) for sequencing.
PCR Amplification of the Promoter Regions and the Genomic DNAs of the FUT1-The genomic DNA of the FUT1 (from exon 1 to exon 3) was amplified by PCR using 100 ng of genomic DNA as the template. The 5Ј-Ex1 (upper primer, Table II) and FUT1 nest (lower primer, Table I, and see Fig. 3C) primers were used for amplification of the FUT1 (denature at 98°C for 10 s, annealing and extension at 68°C for 4 min, and 30 cycles). The PCR product of the FUT1 genomic DNA between exons 1 and 3 was cloned into pGEM for DNA sequencing.
Construction of Luciferase Reporter Gene Plasmid-To create constructs of the luciferase fusion genes, a 0.7-kb PvuII-AatI fragment (see Fig. 5) of pGEM containing the 2.8-kb PCR fragment between exons 1 and 3 of the FUT1 was subcloned into pGL2-enhancer vector (Promega). The plasmid DNA was purified with Qiagen tips (Qiagen Inc., Chatsworth, CA) and transfected into MCAS cells using Trans-IT (Takara) as described previously (20,21). After 48 h, the cells were lysed, and the luciferase activity was measured using the luciferase assay system (Promega) and a luminometer (Lumat LB9501, Berthold, Wildbad, Germany). The relative promoter activity was indicated by relative light units obtained from transfection with pGL2-control vector as 100%. Transfection efficiency was normalized as described previously (21).
DNA Sequencing-Double-stranded plasmid DNA containing the FUT1 fragment was sequenced in both orientations using an AutoRead DNA sequencing kit and an ALF DNA sequencer (Pharmacia, Uppsala, Sweden) or using an ABI PRISM dye terminator cycle sequencing ready reaction kit and an ABI 373 sequencer (Applied Biosystems).

Expression of the FUT1 in Several Tumor
Cell Lines-Northern blot analysis indicated that about 4.0 kb of the transcripts of the FUT1 were detected in total RNA from HEL and MCAS cells but not from WiDr cells (Fig. 1A). The transcript with the same size was also detected in total RNA from KATOIII cells but not from COLO201 and MKN74 cells (data not shown). However, the FUT1 transcript was detected in all these cell lines (Fig. 1B) and in undifferentiated leukemic K562 cells by RT-PCR analysis, whereas other leukemic cells (HL60, U937, BALL-1, and MOLT-4) tested did not express the FUT1 (data not shown). The results indicated that epithelial cancer cells expressed FUT1 mRNA, while many leukemic cells did not express the mRNA. However, erythroid lineage HEL cells (25) and undifferentiated leukemia K562 cells, which has erythroid characteristics (26), expressed the FUT1 mRNA.
Identification of the 5Ј cDNA End of the FUT1 Transcript by RACE Analysis-To isolate the 5Ј-end of FUT1 cDNA, RACE was performed using 1 g of either poly(A) ϩ RNA from HEL or MCAS cells or of human bone marrow Marathon-Ready cDNA (Clontech). Several different sizes of 5Ј-RACE products of the FUT1 were amplified from cDNA prepared from bone marrow, HEL, and MCAS cells (Fig. 2). DNA sequence analysis of 15 clones of the 5Ј-RACE products of the FUT1 from each of the cells indicated that the bone marrow and the HEL cells had two and four different forms by alternative splicing with a single transcription initiation site, respectively, while the MCAS cells had seven forms with two distinct transcription initiation sites and alternative splicing (Fig. 3, B and C). Although we could not identify accurately the 5Ј-ends of the FUT1 by RACE analysis, the longest 5Ј-ends of each starting exon are shown in Fig. 3C. To analyze the transcription initiation sites in each cell line, RT-PCR was carried out using a primer specific for each of the two distinct starting exons (5Ј-Ex1 primer for exon 1A or 5Ј-Ex2 primer for exon 2A, see Table II). The results also indicated that both exon 1A and exon 2A were transcription start sites in MCAS and KATOIII cells, while only exon 1A was a start site in MKN74, WiDr, and COLO201 cells (Fig. 4). In hematopoietic cells, only exon 1A was used as a transcription start site in K562 cells, while only exon 2A was used in HEL cells (Fig. 4). As in HEL cells, exon 2A but not exon 1A was a transcription start site of the FUT1 mRNA in normal bone marrow cells (Fig. 4). Since the mature peripheral leukocytes (data not shown) and four leukemic cell lines did not express FUT1 mRNA by RT-PCR analysis, the results suggested that normal erythroid progenitor cells in bone marrow used only one promoter present in the 5Ј-flanking region of FUT1 exon 2.
Identification of the FUT1 Gene Structure-To identify the FUT1 gene structure, the genomic DNA sequence between exons 1 and 3 of the FUT1 was amplified by PCR. DNA sequence analysis indicated that the first and second introns of the FUT1 were 1654 and 202 bp, respectively, and the all exon/intron junctions of these genes were compatible with GT/AG rule (Fig. 3C). The gene structure of the FUT1 is shown in Fig. 3A.
Analyses of DNA Sequence and Promoter Activity of the 5Јflanking Region of the Exon 2 of the FUT1-Since the 5Јflanking region of exon 2 (or first intron) of FUT1 appeared to act as a promoter, we isolated this region and DNA sequence analysis was performed. The 5Ј-flanking region of the exon 2 of the FUT1 contained a TATA-like sequence (27), three possible Sp 1 binding sites, two possible Ap 2 binding sites, two GATA consensus sequences, and a Myc consensus sequence (28) (Fig. 5).
The pGL2-enhancer vector containing 5Ј-flanking regions of the exon 2 of the FUT1 (between nucleotides Ϫ662 and 45 bp of the exon 2 of the FUT1, Fig. 5) showed promoter activity about 20-fold that of pGL2-control vector with SV40 early promoter and enhancer after transfection of each plasmid into MCAS cells (data not shown). The results suggested that this region acted as a promoter and that the exon 2A was one of transcription initiation sites of the FUT1 in MCAS cells. DISCUSSION In the present study, we examined the structures of cDNA and genomic DNA of the FUT1. The FUT1 cDNA had several forms created by two distinct transcription initiation sites and alternative splicing of 5Ј-untranslated exons. Kelly et al. (18) have reported the transcription start site of FUT1 gene of A431 cells using the primer extension method. However, the transcription start site reported previously was present within exon 3C in this study, and the site was unlikely to be a transcription start site in MCAS and in HEL cells. The reason for this discrepancy is unknown, but one possibility may be that it was due to a difference in cell types.
A recent study has suggested that the re-expression of the H and H-related antigens such as Lewis b and Lewis y in colorectal tumors was regulated by H and Se enzymes (13). In addition, the expression of the FUT1 transcript was increasing during malignant transformation in colorectal mucosa (14). Our results indicated that the FUT1 transcript was expressed in not only colonic cancer cells but also gastric cancer cells. The expression of the H and H-related antigens in normal digestive mucosa, such as gastric and proximal colonic mucosa, are thought to be regulated by the Se enzyme but not the H enzyme (1). Our results suggest that the expression of the FUT1 mRNA and thereafter the expression of the H enzyme are increasing during malignant transformation in digestive mucosa.
The ABH antigens are expressed in embryonic but not in adult muscular, bone, and neuronal tissues (12). Recently, as well as the FUT1 gene, some other glycosyltransferase genes with distinct promoters and alternative 5Ј-ends have been re-  1. Expression of the FUT1 mRNA. A, Northern blot analysis. 20 g of total RNA prepared from each tumor cell line was divided into 10 g/lane, electrophoresed, and transferred onto a nylon membrane. The membrane was then divided to prepare duplicate membranes. One membrane was used for Northern blot analysis, and the other membrane was stained by methylene blue. The positions of 28 and 18 S ribosomal RNAs are indicated by arrows. B, RT-PCR analysis. Total RNA (2 g) prepared from each tumor cell line was reverse-transcribed with oligo(dT) primer, and the resulting single strand cDNA from each cell line was used as a template for PCR analysis. The FUT1 amplification was performed for 30 cycles and the G3PDH for 25 cycles. PCR products were electrophoresed in a 1.2% agarose gel and stained by ethidium bromide. Endonuclease StyI-digested Lambda DNA (/S) was used as a molecular size marker. ported, such as the ␣(2,6)sialyltransferase gene (29 -31), murine ␤(1,4)galactosyltransferase gene (32), human N-acetylglucosaminyltransferase V gene (33), ␣(1,3)fucosyltransferase (34) gene, and the human ␤(1,4)-N-acetylgalactosaminyltransferase gene (35). Kozak (36) discussed that transcription of a single gene by multiple promoters may provide additional flexibility in the regulation of gene expression. Such promoters could have tissue-and developmental stage-specific activity. In fact, as shown in this study, the promoter usage and splicing patterns of the FUT1 mRNA in erythroleukemic cells (HEL) and in normal erythroid progenitor cells were different from those in undifferentiated leukemic cells (K562), which has erythroid characteristics, suggesting that changes in promoter usage and splicing patterns appear during differentiation of the erythroid lineage. Thus, the alternative use of multiple promoters and alternative splicing of the glycosyltransferase genes may be associated with tissue-and stage-specific glycosylation patterns.
The tissue expression pattern of the ABO antigens is differ-ent among vertebrate species (2). The expression of these antigens in the digestive mucosa has been observed from amphibians to higher mammals, while only human and some higher anthropoid primates but not old world monkeys express these antigens on erythrocyte membranes. Recently, rabbit homologues of human FUT1, FUT2, and a pseudogene of FUT2 (Sec1) have been isolated (38,39). In addition, we have had indications that the old-world African green monkey had three functional ␣(1,2)fucosyltransferase genes that were homologous to human FUT1, FUT2 and Sec1. 2   sible reason for this phenomenon may be a difference in the promoter characteristics. The present study suggested that the FUT1 had two distinct promoters and that normal erythroid lineage cells and erythroleukemia HEL cells used only one promoter present in the 5Ј-flanking region of exon 2. Thus, it is conceivable that an erythroid-specific promoter of the FUT1 became functional in cells of erythroid lineage after separation from the ancestors of apes to that of old world monkeys. Rare individuals (Bombay and para-Bombay phenotypes) fail to express the ABO antigens on their erythrocyte membranes because of a lack in H enzyme activity (1). Although non-sense and mis-sense mutations have been found in the coding region of the FUT1 of Caucasian and Japanese H-deficient individuals (18,37), no DNA sequence difference was observed in the coding region of the FUT1 of Indian Bombay phenotype (40). The erythroid-specific promoter may be inactivated in Indian H-deficient individuals. Thus, analyses of the promoter present in the 5Ј-flanking region of exon 2 of Indian Bombay individuals or a corresponding region of the monkey may provide further information for understanding tissue-and species-specific expression of the H and ABO antigens.
In the present study, we identified several forms of the FUT1 transcript generated by two transcription start sites and alternative splicing in 5Ј-untranslated exons. Our results suggested that dual promoters regulated the stage-and tissue-specific expression of the FUT1 transcript and thus regulated the expression of ABO and related histoblood antigens in many human tissues.

FIG. 4. RT-PCR analysis for detection of the FUT1 starting exon in various tumor cells and in normal bone marrow cells.
Total RNA (2 g) prepared from each tumor cell line or poly(A) ϩ RNA (500 ng) from bone marrow cells was reverse-transcribed with oligo(dT) primer and used for PCR analysis. The FUT1 amplification was performed using a set of each starting exon-specific primer (5Ј-Ex1 or 5Ј-Ex2 shown on the right) and 3Ј-FUT1 primer (Table I). PCR products were electrophoresed in a 1.2% agarose gel and stained by ethidium bromide. StyI-digested Lambda DNA (/S) was used as a molecular size marker.
FIG. 5. DNA sequences of a part of the first intron of the FUT1. The possible binding sites for transcription factors are indicated by underlining. Some restriction endonuclease sites are also indicated by underlining. The first nucleotide of exon 2 of the FUT1 is numbered as ϩ1.