Rat encodes the paralogous gene equivalent of the human histo-blood group ABO gene. Association with antigen expression by overexpression of human ABO transferase.

We cloned a rat ABO homologue and established human A- and B-transferase transgenic rats. A DNA fragment corresponding to exon 7 of the human ABO gene was amplified from Wistar rat genomic DNA and sequenced. Using the amplified fragments as a probe for Southern blotting, multiple hybridized bands appeared on both EcoRI- and BamHI-digested genomes of seven rat strains, which showed variations in the band numbers among the strains. Four cDNAs were cloned from a Wistar rat, three of which showed A-transferase activity and one of which showed B-transferase activity. These activities were dependent on the equivalent residues at 266 and 268 of human ABO transferase. Wild Wistar rats expressed A-antigen in salivary gland, intestine, and urinary bladder tissue, but B-antigen was not stained in any organs studied, whereas a transcript from the ABO homologue with B-transferase activity was ubiquitous. Human A-transferase and B-transferase were transferred into Wistar rats. A-transgenic rats expressed A-antigen in ectopic tissue of the brain plexus, type II lung epithelium, pancreas, and epidermis. B-antigen in the B-transgenic rat was expressed in the same organs as A-transgenic rats. These results may shed light on the function and evolution of the ABO gene in primates.

Histo-blood group ABH antigens are important not only for blood transfusion but also for organ transplantation. The structures of the epitopes have been defined as trisaccharide determinants (1), GalNAc␣1Ј3(Fuc␣1Ј2)Gal-for A, Gal␣1Ј3-(Fuc␣1Ј2)Gal-for B, and Fuc␣1Ј2Gal-for H. The determinants are synthesized by the products of the ABO locus, which encodes three alleles. The A allele produces ␣1,3-N-acetylgalac-tosamine transferase (A-transferase), 1 which catalyzes the transfer of GalNAc residues from the UDP-GalNAc donor nucleotide to the Gal residues of the acceptor H-antigen. The B allele encodes ␣1,3-galactosaminyl transferase (B-transferase), which catalyzes the transfer of Gal residues from the UDP-Gal donor nucleotide to the Gal residues of the acceptor H-antigen. The O allele lacks both enzymatic activities for the one base deletion causing a frameshift.
Natural antibodies against both A-and B-antigens are produced in all individuals with type O blood and the antibodies against A or B are produced in individuals with type B or type A blood, respectively. Because most of the serum antibodies are composed of IgM, ABO-mismatched transfused red blood cells or endothelium from transplanted organs reacts with recipient antibodies and the antibody activates a compliment cascade. Therefore, if anti-A or anti-B antibodies in the recipients are not sufficiently removed by immunoadsorbence or a plasma exchange procedure, the graft organs are rejected in a hyperacute mode (2,3). To improve the clinical management of transplantation, development of ABO-mismatched animal models is desirable. Rats are more suitable for ordinary examination of organ transplantation because of the larger body size compared with mice (4). To evaluate the availability of rats as an animal model, we studied the rat ABO homologue and established human A-and B-transferase transgenic rats.
Besides the clinical importance of ABH antigens, those physiologic functions are still under investigation. The ABH antigens are distributed in the human body in a tissue-specific manner (5). During fetal development, marked changes in antigen expression have been observed (6). The ABH antigen alteration in cancer cells associate with the patient prognosis. An ABH sugar chain affects cell-to-cell interaction through modification of integrin receptor function on human cancer cells (7,8). Besides human, rats are the most investigated animals for ABH antigen expression. ABH polymorphism among the strains (9), tissue distribution (10,11), changes during the development (10), changes by oncogenesis (12,13), and changes by parasites infection (14) have been reported. However, the molecular basis of them has not been defined. In the present study, multiple rat orthologues are demonstrated to be a paralogous gene family with distinct expression profiles. Until now, ABO orthologues cloned in other animals have been * This work was supported by Scientific Research from the Ministry of Education, Science and Culture of Japan Grants-in-Aids 13670435 (to S. I.) and 13557038 (to E. K.) and a grant from Research on Health Sciences focusing on Drug Innovation (to E. K.). 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  a single-locus gene. The antigen expression in wild rats and the effect of exogenous human genes will shed light on the function and the evolution of the ABO gene in primates.

MATERIALS AND METHODS
Probe Preparation and Southern Hybridization-Rat genomic DNA was extracted from the following rat strains: outbred Wistar, inbred DA/Slc, inbred LEW/Crj, inbred PVG/Sea, inbred ACI/N Slc, inbred SD, and inbred BN/Crj, which were maintained at Charles River Japan, Inc. (Yokohama, Japan). A DNA fragment corresponding to exon 7 of the human ABO gene was amplified from Wistar rat genomic DNA by PCR using a primer pair (sense, CAGAGCAGCACTTCATGGTGG, homologous to nucleotides 408 to 428, and antisense, GGGCCAGCCCAG-CAGCTGCTGGTCCCA, complementary to nucleotides 970 to 996 of the human ABO cDNA). The PCR product was subcloned into pCR2.1 vector (Invitrogen, Carlsbad, CA) and sequenced using a Dye Terminator Cycle Sequencing kit (Beckman, Fullerton, CA) by a capillary sequencer CEQ2000 (Beckman). The rat genome DNA from the seven strains was digested by restriction enzymes, EcoRI or BamHI. The genome DNA fragments separated by agarose gel were blotted onto a membrane and hybridized with the 32 P-labeled PCR products.
cDNA Cloning of Rat ABO Homologue-According to the results of the genome sequence, four clone-specific reverse primers were prepared for 5Ј-rapid amplification of a cDNA ends (RACE) procedure. Sense primers for 3Ј-RACE were then designed on sequences of the 5Ј-RACE. The RACE procedures were carried out using a Marathon TM cDNA amplification kit (Clontech Laboratories, Inc., Palo Alto, CA) and rat intestinal poly(A) ϩ RNA with the following primers: antisense, ATGG-ACACGTCCTGCCAGCGA, CACACCAGGTAATCCACTTCATA, TAA-AGGACTCCCGTTGGCTACT, and CCCATATAGTAAAAGTCACCCT-GG, and sense, CTATGGATTCCTGAGCCACAGAA, GTTCCTGAGCC-ACAGATTCCAT, GTTATGGGTTCCTGAGCCACAA, and TTTGCTTC-GTGTGCCTGAGCCTCAG. The PCR products were subcloned, and 18 to 34 clones of each fragment were sequenced on both strands. The sequences were aligned by DNASIS-Mac software (Hitachi Software Engineering Co., Ltd., Hitachi, Japan) and GENETYX-Mac (Software Development Co., Ltd., Tokyo, Japan).
The organ distribution of rat ABO homologues was determined by multiplex RT-PCR with the primers for the RACE procedure and rat glyceraldehyde-3-phosphate dehydrogenase. The total template RNAs were prepared from 15 Wistar rat organs: brain, submandibular gland, heart, lung, stomach, small intestine, large intestine, liver, pancreas, spleen, kidney, urinary bladder, testis, skin, and bone marrow. The PCR products were semiquantified by gel electrophoresis followed by ethidium bromide staining.
Analysis of Enzymatic Activity of the ABO Homologue-Full-length cDNAs of rat ABO homologues were prepared by high fidelity RT-PCR with primers specific for 5Ј and 3Ј ends of each clone. Rat ABO homologues, human A-cDNA, and human B-cDNA were inserted into pCR3 (Invitrogen), and H-fructosyltransferase (FUT1) cDNA was inserted in pSSR␣Ϫbsr vector (15). The enzymatic activities of four ABO homologues were assessed by the expression of A-or B-antigen on HeLa cells transfected by the cDNAs with or without co-transfection of FUT1 cDNA. The cDNA transfection was carried out by an electroporation procedure using GenePulser II (Bio-Rad). Twenty-four hours after transfection, HeLa cells were harvested and stained by anti-A, anti-B, or anti-H mouse monoclonal antibodies (DACO, Carpinteria, CA), and a second antibody of fluorescein isothiocyanate-labeled anti-mouse IgM goat serum (Cappel, Aurora, OH). The stained cells underwent flow cytometry (Cytronabsolute, Ortho Clinical Diagnostics, Raritan, NJ).

Establishment of A-and B-Transferase Transgenic Rats-
The human A-transferase cDNA isoform, which includes intron 6 (FY-66-1), was kindly provided by Dr. Hakomori (16). B-transferase cDNA was generated by replacement of the SacII/SalI fragment of FY-66-1 with that of B-transferase gene exon 7. The cDNA was inserted into the downstream of the chicken ␤-actin promoter of the pCAGGS vector and injected into Wistar rat pronuclei. The microinjected fertilized eggs were then transferred to pseudopregnant recipient rats. Transgenic founders were detected by PCR and RT-PCR of tail genomic DNA and total RNA, and stable lines were generated by breeding the founders.
Immunohistochemistry and Immunoblotting-Wild, A-transgenic, and B-transgenic rats were sacrificed at 3 and 5 months of age. The 15 organs were removed from the animals and embedded in compound. Ten-m thick frozen sections were fixed in 4% paraformaldehyde in 0.1 M phosphate buffer and placed in blocking buffer (phosphate-buffered saline, pH 7.4, 1% bovine serum albumin, 0.3% Triton X-100). The blocked sections were or were not reacted with anti-A, anti-B, or anti-H antibody. After washing, the sections were reacted with biotin-labeled anti-mouse goat antibody followed by Texas Red-labeled avidin staining. Counterstaining was performed with 4Ј,6-diamidino-2-phenyindole (Sigma).
Small intestinal specimens from wild, A-transgenic, and B-transgenic rats were homogenized in 0.25 M sucrose solution and solubilized in SDS sample buffer. Ten micrograms per lane of solubilized small intestinal proteins were separated on a 5% polyacrylamide gel in denaturing buffer. After electrophoresis, proteins were electroblotted on a polyvinylidene difluoride membrane (Amersham Biosciences). The blot membrane was separated into three strips and the strips were stained individually with anti-A, anti-B, or anti-H antibody. Antigen-antibody complexes were visualized with peroxidase-conjugated goat anti-mouse IgM and West Pico chemiluminescent substrate (Pierce).

RESULTS
Southern Blot Analysis-To clone the rat ABO homologue gene, partial sequences were amplified from genome DNA, which corresponded to exon 7 of the human gene. Used primer sequences were chosen from the fragments highly conserved between human and mouse (17). The PCR product was subcloned into a plasmid and the individual clones were sequenced. Twenty sequenced clones were divided into four groups according to the base substitutions and tentatively named pCR2.61, pCR2.66, pCR2.6B, and pCR3.47. The Southern blot membrane of the rat genome was probed by these fragments. The genome DNA digested by EcoRI showed variations in fragment number and size: four fragments in Wistar, DA/Slc, and ACI/N Slc; three fragments in LEW/Crj and BN/ Crj; two fragments in PVG/Sea; and five fragments in SD rat (Fig. 1). DNA fragments that appeared in BamHI-digested genomes also showed polymorphisms among the strains: three fragments in Wistar, DA/Slc, ACI/N Slc, and SD rats; and two fragments in LEW/Crj, PVG/Sea, and BN/Crj rats. These findings indicate that the rat ABO homologue consists of multicopy genes and the copy number possibly varies among strains.
cDNA Cloning of Rat ABO Homologue-The RACE procedure was used to obtain the full-length transcript sequences of the ABO homologues. Sixty-eight 5Ј-RACE and 48 3Ј-RACE clones were sequenced and realigned into four clones: rat-2.66 (GenBank TM accession number AB081649), rat-2.61 (AB081650), rat-2.6B (AB081651), and rat-3.47 (AB081652). These sequences were verified by a standard RT-PCR procedure with their 5Ј and 3Ј specific primers. The open reading FIG. 1. Southern blot analysis of rat genomic DNA. Ten micrograms of genome DNA from the seven rat strains were digested by EcoRI or BamHI. DNA fragments separated through a 0.8% agarose gel were blotted on a nylon membrane and probed by PCR fragments from a Wistar rat corresponding to a part of exon 7 of the human ABO gene. frames estimated were aligned with human and mouse transferases ( Fig. 2A). The catalytic domain was conservative, but the regions following to the N-terminal transmembrane domain were variable. The rat-3.47 clone lost the fragment that corresponded to exon 4 of the human ABO gene and was also lost in mouse cDNA. The overlined amino acids in Fig. 2A show the four residues that are considered the amino residues responsible for the sugar nucleotide donor selection of human A-transferase (Arg-175, Gly-234, Leu-266, and Gly-268) and B-transferase (Gly-175, Ser-234, Met-266, and Ala-268) (18). Among these residues, the last two have been shown to be particularly important for substrate selection. Three cDNAs, rat-2.66, rat-2.61, and rat-2.6B, encoded Ala and Gly at equivalent residues of human 266 and 268, which are similar to those of human A-transferase. The cloned rat-3.47 encoded both Met and Ala, which are identical to those of B-transferase.
Based on the maximum matching analysis, a phylogenetic tree was drawn using the neighbor joining method (Fig. 2B). The four rat cDNAs were clustered in a branch next to the mouse ABO, suggesting that they are paralogous genes.
Enzymatic Activity of Rat ABO Homologues-Four rat cDNAs of the ABO homologues were inserted in a eukaryotic expression vector and their enzymatic activities were assayed by introducing the constructs into HeLa cells. The transfected HeLa cells were stained by anti-A, anti-B, and anti-H antibodies and subjected to flow cytometry (Table I). Three cDNA clones, rat-2.66, rat-2.61, and rat-2.6B, expressed A-antigen on HeLa cells when they were co-transfected with FUT1 cDNA, but clone rat-3.47 did not. Reciprocally, rat-3.47 expressed B-antigen but rat-2.66, rat-2.61, and rat-2.6B clones did not. Apparent expression of A-or B-antigens was not observed on the cells transfected with only ABO homologues (pSSR␣-bsr).  2. Comparisons of the predicted amino acid sequences. A, multiple alignment of the coding regions among rat, mouse (AB041038), and human A-transferase (AF134412), and B-transferase (AF134414) genes. Asterisks under the rat sequences denote invariant residue positions in rat four homologues, and the asterisks at the bottom denote those in the seven sequences. Four rat ABO homologues conserve the sequence with 97 to 83% homology, whereas rat-2.66 was 72% identical to human A transferase, and rat-3.47 was 83% identical to mouse ABO homologue. The overlined amino acids show the four residues that are responsible for the substrate selection of human A-and B-transferase. Arrowheads indicate the boundaries of exons in human, suggesting the gap in rat-3.47 and mouse ABO corresponds to exon 4 of the human ABO gene. The rectangle indicates predicted transmembrane domains. B, full-length of the amino acid sequences of the rat, mouse, pig (AAC68840), crab-eating macaque (AAF04732), rhesus monkey (AAD56306), and human were aligned and a phylogenetic tree was drawn. The neighbor joining method was used for Kimura's distance. Mouse ␣-galactosyltransferase (M85153) was used as an outgroup gene. The numbers on two interior branches are boot strap probabilities.
These results indicate that rat homologues also select H-antigen as an acceptor substrate and transfer GalNAc or Gal to it, depending on the residues responsible for substrate selection.
Expression Profile of Rat ABO Homologues-To discriminate the expression of the four rat ABO homologues, multiplex RT-PCR was used for the templates originating from the 15 organs. Simultaneous amplification in a tube and gel electrophoresis showed exclusive and relatively high level expression of rat-2.66 transcript in salivary glands, small intestine, and urinary bladder, and weak expression in the stomach and colon (Fig. 3). Weak bands from rat-2.61 and rat-2.6B appeared in the large intestine. Transcripts of rat-3.47 were ubiquitous.

Establishment of A-and B-Transferase Gene Transgenic Rats-Human
A-transferase or B-transferase cDNA in pCAGGS vector was introduced in rat fertilized eggs by microinjection, and the founders were screened by PCR of tail DNA and RNA. Fifty-six rats were born from 346 eggs of the injected A-transferase gene. Genome integration was observed in 10 founders of the 56 rats. Six of the 10 rats were also RT-PCR positive. Two B-transferase transgenic (B-Tg) founders were identified in the same way as A-Tg rats. A-or B-antigen expression was assayed by an immunohistochemical technique. RT-PCR positive rats expressed A-or B-antigen on the epidermal cell surface, hair follicle epithelium, and sweat glands but not on red blood cells, vascular endothelium, connecting tissue, or bone (Fig. 4A). Two founders of each transgenic rat were mated and homozygous transgenic rats were established. After 1 year of observation, no macroscopic and microscopic phenotype alterations were observed in the founder or F1 and F2 transgenic rats. Although the differences in the embryo have not been analyzed, these findings suggest that enhanced expression of A-transferase or B-transferase activity has no effect on rat growth and development. Fig. 4C and representative results are shown in Fig. 4B. Wild rats expressed A-antigens obviously on the acinus of salivary glands, mucosal crypts, intestinal mucus, and epithelium of the urinary bladder and weakly on stomach epithelium, which corresponded to the results of rat-2.66, rat-2.61, and rat-2.6B mRNA distribution. Enhanced expression of human A-transferase under the actin promoter resulted in the ectopic expression of A-antigen in the brain plexus, type II lung epithelium, the exocrine gland of the pancreas, and the epidermis, and enhanced expression in the epithelium of the stomach. Apparent staining by anti-B antibody was not observed in any wild and A-Tg rat organs studied, whereas weak expression of 3.47 mRNA was ubiquitous. Only B-Tg rats expressed B-antigen on the same organs as A-Tg rats, despite the fact that the staining was relatively weak compared with that in the A-Tg rats. The salivary glands, stomach, intestines, and bladder of B-Tg rats were still positive for anti-A antibody.

A-or B-Antigen Expression in Wild, A-Tg, and B-Tg Rats-To access the basal expression of A-and B-antigens in wild rats and determine the influence of overexpression of human A-transferase or B-transferase, systemic organs were stained by an immunohistochemical procedure. The staining profiles by anti-A or anti-B antibodies are listed in
Western Blot Analysis of ABH Antigen Expression in the Intestine-To evaluate quantitatively the expression of ABH antigens in the intestine, the membranes blotted with small intestinal proteins of the three rats were stained by anti-A, anti-B, or anti-H antibody. It appeared that the bands with molecular weights around 216,000 and 78,000 were the nonspecific immune complexes or intrinsic peroxidase activity. Monoclonal antibody against A-antigen showed equivalent smear bands in all three rats. Antibody against B-antigen stained only B-Tg rat intestinal proteins. Anti-H antibody stained all rats weakly, but the smear band in wild rat was most prominent, which might be the result of masking of residual H-antigen by the overexpression of human A-transferase or human B-transferase (Fig. 5). DISCUSSION We cloned four paralogous gene family members equivalent to the ABO glycosyltransferase gene from a Wistar rat. Three of the genes expressed A-antigen and one expressed B-antigen on  The expressions of four rat ABO homologues were assessed by multiplex RT-PCR with an internal control glyceraldehyde-3-phosphate dehydrogenase (GAPDH). One microgram of total RNA was reverse transcribed into cDNA and amplified by PCR with specific primers for each cDNAs. PCR products were visualized on 2% agarose gel and stained with ethidium bromide.
HeLa cells with the help of FUT1. The amino acid residues responsible for the selection of the nucleotide sugar donor in humans (L266M and G268A) were Ala and Gly in rat-2.66, rat-2.61, and rat-2.6B clones and resembled human A-transferase, and Met and Ala in rat-3.47 clones were identical to human B-transferase. These results showed that the corresponding residues of human 266 and 268 in rat clones are involved in the nucleotide sugar donor selection, UDP-GalNAc or UDP-Gal. B-antigen was undetectable in any organs studied from wild Wistar rats, whereas rat-3.47 mRNA for B-transferase activity was expressed ubiquitously in rat organs. Enforced expression of human B-transferase, however, resulted in the expression of B-antigen, although the level was much less than that of A-antigen. Therefore, the low transcript level of rat B-transferase may be the main reason for the deletion of B-antigen in wild Wistar rats. Another possible reason is competition for the substrate of UDP-galactose with ␣1,3-galactosyltransferase, which was discussed in a mouse homologue with cis-AB activity (17). or B-transgenic rats were stained with anti-A or anti-B antibodies, then with biotin-labeled anti-mouse goat antibody, and finally with Texas Red-labeled avidin (red). Nuclei were stained by 4Ј,6-diamidino-2-phenyindole (blue). A, A-and B-antigen expression in the tails. A-antigens were detected on the cell surface of the epidermis, hair follicle epithelium, and sweat glands of A-transferase transgenic rats but not in B-transferase transgenic rats. B-antigen was detected in the same tissue only in B-transferase transgenic rats. Bouhours et al. (9) reported that rat large intestinal glycosphingolipids showed genetic polymorphisms, in which some rat strains lacked A-antigen, whereas all strains expressed B-antigen active glycosphingolipid. By breeding and backcrossing of the strains, those investigators also revealed that the polymorphism was not the result of allelic genes but was controlled by closely linked multigenes. During the preparation of this paper, Olson et al. (19) reported on two rat A-transferases they cloned as a pair of alleles (AF469945 and AF469946). The nucleotide and amino acid sequences of AF469946 are identical to those of rat-2.61, and AF469945 is 95% homologous to rat-2.66 and rat-2.61 clones. However, our four cDNA clones were identified from one animal and each mRNA expression pattern is distinctive from the others. Furthermore, multiple bands in a Southern blotting study supported a conclusion that four Wistar rat genes equivalent to the human histo-blood group ABO gene are not the allelic genes but members of the paralogous gene family.
The predicted amino acid sequences of rat-2.66, rat-2.61, rat-2.6B, and rat-3.47 showed 97 to 83% homology among them. The neighbor joining method revealed that the rat orthologue members form one cluster, indicating they evolved from an ancestral gene. However, multiple alignment of the four predicted peptides showed sharing of amino acid residues among them, i.e. between rat-2.66 and rat-2.61 and between rat-2.66 and rat-2.6B. This indicated that the amino acid differences among the genes did not simply result from base substitutions but recombination by repeated gene shuffling among them. It has been postulated that tandem copy paralogous genes evolve through interlocus nucleotide shuffling. For example, RHD and RHCE genes, encoding the rhesus blood group system, donate or accept gene fragments from each other and develop highly polymorphic antigen systems (20,21). Rat ABO homologues might have evolved in this manner, and some strains lack A-transferase activity in the large intestine. The multicopy status of the ABO homologue appears not to be restricted to rats. Multiple fragments in the Southern blotting study probed by the human A-transferase gene were found in dogs, cats, and rabbits in addition to rats (22). Expression of both A-and B-antigens in the intestinal mucosa of rabbits has also been reported (23). Saitou and Yamamoto (24) argued that the seven nucleotide polymorphisms between A and B alleles in human are more excessive than expected as simple base substitutions and that positive selection pressure affected the evolution of higher primate ABO genes. However, if the mammalian ABO ancestral genes consisted of multicopy genes with A-and B-transferase activity like rats, they must donate or accept gene fragments from each other and may produce a locus composed of A and B alleles. The multicopy genes may possibly gather and bundle into a single gene through unequal crossover. Southern blotting studies in seven rat strains suggest deletion or acquisition of some loci. This estimation must be resolved by further genome projects in rats and other mammals.
A-Tg rats expressed A-antigen in the brain plexus, lung type II epithelium, pancreas acinus, and skin in addition to digestive mucosa, salivary gland, and bladder, in which A-antigen was expressed in wild rats also. B-antigen in B-Tg rats was detected in the same tissue as A-antigen in A-Tg rats. These results indicate that the organs that ectopically expressed Aantigen provided the substrates for the transferase, H-antigen, but lacked the endogenous ABO homologues and resulted in negative staining against anti-A antibody. H-antigen in rodents is restricted to epithelial organs, whereas they also have two ␣1,2-fucosyltransferase genes like human (25,26). The rodent orthologue of human FUT1 lacks the promoter that transactivates the gene in the erythroid of hominoids, so that rodents do not express H-antigen in the organs of mesothelial origin (27,28). Ubiquitous and overexpression analysis of A-transferase and B-transferase clearly showed the absence of H-antigen in the endothelium but its presence in epidermis, brain plexus, lung, and pancreas. Further transgenesis of the human FUT1 gene will mimic ABH expression in human organs and will be useful for the simulation of the ABOmismatched vascular graft.
Holmes et al. (12) reported that precancerous liver and hepatoma in Fischer 344 rats induced by the chemical carcinogen N-2-acetylaminofluorene expressed B-antigen as the ␣-galactosyl-␣-fucosyl-GM 1 form. Precancerous liver and hepatome were induced to express ␣-fucosyltransferase (29). The synthesized ␣-fucosyl-GM 1 was the substrate for normally expressing ␣-galactosyltransferase and was converted into B-determinants. The ABO homologue that liver is expressing is the rat-3.47 gene with B-transferase activity. Then, the normally existing ␣-galactosyltransferase is estimated to have been the product of the rat-3.47 gene. In human colorectal, lung, and bladder cancer, the deletion of A-or B-antigen in cancer cells associate with poor prognosis of the cancer patients (7). In 25% of primary bladder tumors, loss of heterozygosity of the ABH locus was observed (30). Rat chemical-induced colorectal cancer also changed the ABH antigen expression (13). The ABH gene expression was down-regulated in A-negative human colorectal cancer cells by the methylation of the promoter and the reduced activity of the enhancer element of the gene (31,32). The presented cDNA data should contribute for investigation in alteration of the rat ABO homologue expression in the chemical oncogenesis. Furthermore, the established Tg rats express Aor B-transferase independently from the native genes and will be useful to reveal whether the ABH antigen affect directly the cancer virulence.
ABH antigen expression in human and rat fetuses is under strict control by stage of development and it is organ-specific (6,33), whereas its exact meaning for fetal development is still under investigation. ABH antigen was found in early embryos (week 5) in the cardiovascular endothelium, the epithelial cells of all organ rudiments, and the erythropoietic cells in blood islands (6). After recession of the epithelial cell wall antigens at the end of the first trimester of pregnancy, ABH antigen expression increases as the respective organs mature, e.g. mucous secretion in digestive organs. A-antigen and B-antigen expression in transgenic rats should be controlled by FUT1 or FUT2 gene expression, which might permit the usual embryogenesis of the transgenic rats. Immunohistochemical follow-up of A-antigen expression in wild and transgenic rat fetuses may provide further information on the function of the ABH antigen in embryogenesis.