Regulation of a1,3galactosyltransferase expression in pig endothelial cells Implications for xenotransplantation Dominique Mercier 1,2 , Beatrice Charreau 3 , Anne Wierinckx 2 , Remco Keijser 1 , Lize Adriaensens 1 , Renate van den Berg 1 and David H. Joziasse 1 1 Department of Molecular Cell Biology, Research Institute of Immunology and Inflammatory Diseases and 2 Department of Medical Pharmacology, Research Institute Neurosciences VU, VUmc, Amsterdam, the Netherlands; 3 Institut de Transplantation et de Recherche en Transplantation (ITERT), INSERM U437, Nantes, France The disaccharide galactosea1,3galactose (the aGal epitope) is the major xenoantigen responsible for the hyperacute vascular rejection occurring in pig-to-primates organ trans- plantation. The synthesis of the aGal epitope is catalyzed by the enzyme a1,3-galactosyltransferase (a1,3GalT). To be able to control porcine a1,3GalT gene e xpression specific- ally, we have analyzed the ups tream portion of the a1,3GalT gene, and identified the regulatory sequences. Porcine a1,3GalT transcripts were detected by 5¢ RACE analysis, and the corresponding genomic sequences were isolated from a phage library. The porcine a1,3GalT g ene consists of at least 1 0 d ifferent exons, f our of which c ontain 5 ¢ untranslated sequence. Four distinct promoters, termed A– D, d ri ve a1,3GalT g ene t ranscription in porcine cells. A combination of a lternative promoter usage a nd alternative splicing produces a series of transcripts that differ in their 5¢ portion, but encode the same p rotein. Promoters A–C have been isolated, and functionally characterized using luciferase reporter gene assays in trans- fected porcine endothelial cells (PEC-A). Promoter prefer- ence in porcine endothelial cells was estimated on the basis o f relative transcript levels as determined by real-time quanti- tative PCR. More than 90% of the a1,3GalT transcripts in PEC-A cells originate from promoter B, which has charac- teristics of a housekeeping gene promoter. While promoter preference remains unchanged, a1,3GalT mRNA levels increase by 50% in 12 h upon tumour necrosis factor a-activation of P EC-A cells. However, the magnitude of this change induced by inflammatory conditions could be insufficient to affect cell surface a1,3-galactosylation. Keywords: a1,3galactosyltransferase; promoter; pig endo- thelial cells; regulation; xenotransplantation. The growing disparity between the demand for transplant- able organs and the supply has renewed interest in the possibilities of transplanting animal organs to humans (xenotransplantation) [1]. Several animal species have been evaluated f or their suitability a s organ donor. Currently, pigs are con sidered as the most suitable donor animal because pig organs are physiolo gically similar to human organs and the potential risk of pathogen transmission is low when compared w ith the use of o rgans from species closely related to humans [1, 2]. But when transplanted into humans or nonhuman primates, pig organs are rejected hyperacutely by antibody-mediated complement activation [3–5]. The hyperacute rejection i s i nitiated by the interaction between natural preformed anti-pig Ig (xeno reactive natural antibodies) and carbohydrate epitopes expressed by endo- thelial cells of donor organs. T his results in the activation of the classical complement pathway with concomitant endo- thelial cell activation, which ultimately induces graft failure [4]. A major portion (about 80%) of xenoreactive natural antibodies is directed against a single determinant, the terminal disaccharide structure galactosea1,3galactose (the aGal epitope), present o n the surface of p ig vascular endothelium [6–8]. These anti-Gala1,3Gal Ig, originally identified by Galili et al. [9], are also found in apes and Old World mo nkeys, bu t not in lower primates (e.g. New World monkeys) or nonprimate mammals (including the pig). The latter species express the aGal e pitope, whereas humans and higher primates don’t [9–11]. Xenoantigens that contain the Gal a1,3Gal structure are synthesized by UDP-Gal:Galb1,4GlcNAc a1,3galactosyl- transferase (a1,3GalT). Genes and cDNAs encoding the a1,3GalT e nzyme have been cloned from several species (cow, mouse and pig) [12–16]. In humans, one pseudogene (HGT-10) and one retro-processed pseudogene (HGT-2) have been identified, both containing multiple frame-shift mutations and internal stop codons in the protein coding sequence [17–19]. The absence of a functional a1,3GalT gene copy in humans, and in apes and Old W orld monkeys, explains the absence of aGal epitopes in these species. In Correspondence t o D.Mercier, Department ofMedical Pharmacology, Research Institute Ne urosciences, Vrije Universiteit Medical Center, van der B oechorststraat 7, 1081 BT Amsterdam, the N etherlands. Fax: + 31 20 444 81 00, Tel.: + 31 20 444 80 96, E-mail: d.mercier.pharm@med.vu.nl Abbreviations:AhR,arylhydrocarbonreceptor;Arnt,AhRnuclear translocator; C/EBP, CCAAT/enhancer-binding protein; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; NF-jB, nuclear factor j beta; pPAEC, primary pig aortic endothelial cell; Q-PCR, real-time quantitative PCR; TNFa, tumor necrosis factor a;YY1,YingYang1 transcription factor. (Received 28 September 2001, revised 7 January 2002, accepted 16 January 2002) Eur. J. Biochem. 269, 1464–1473 (2002) Ó FEBS 2002 mouse a nd pig, the c oding r egion of the a1,3GalT gene is distributed o ver s ix exons that span 24 kb of genomic DNA (18 k b in mouse) [13, 20, 21]. In addition, the gene contains several 5¢ noncoding exons. A suppression of the a1,3-galactosylation of the donor organ will possibly overcome hyperacute rejection, and t hus facilitate xenotransplantation. Therefore, efforts have been directed at targeting t he porcine a1,3GalT gene [22–24], but thus far, no viable knockout pigs have b een produced. The design of more subtle methods to down-regulate a1,3GalT expression in a time- and tissue-dependent fashion requires an understanding of the regulation o f a1,3GalT g ene transcription. Therefore we have a nalyzed a1,3GalT regu- lation in pig endothelial cells. We have set out to isolate the sequences upstream of the 5¢ untranslated exons. T hree of the four putative promoter regions were isolated from a pig genomic library, and functionally characterized using gene reporter assays. When t ransiently transfected into pig aortic endothelial cells, all three putative promoter regions were able to drive luciferase transcription. Relative importance of the different promoters was determined in resting and tumor nec rosis fa ctor a (TNFa) stimulated e ndothelial cells using real-time quantitative P CR (Q-PCR). Results indica- ted that more than 90% of the a1,3GalT gene expression in pig e ndothelial cells was associated with only one of the four putative promoters (promoter B). The modest effect of TNFa treatment on a1,3GalT transcription suggests that the various promoters are only weakly s ensitive to inflam- matory conditions. MATERIALS AND METHODS Cells and cell lines COS7 cells were obtained from the Netherlands Cancer Institute (Amsterdam, the Netherlands). The pig kidney cells (PK15) were obtained from A. Roos (Department of Nephrology, Leiden University Med ical C enter, the Neth- erlands). Pig aortic endothelial cells (PEC-A [25]), and pig primary aortic endothelial cells (pPAECs) were provided by J. Holgersson (Karolinska Institute, Huddinge, S weden) and B. Charreau (Institut de Transplantation et de Recher- che en Transplantation, Nantes, France), respectively. All cells were cultured in Dulbecco’s modified Eagle’s medium supplemented w ith 10% of fetal bovine serum and 100 units of penicillin/streptomycin (Life Technologies). Probe preparation Pig genomic DNA was prepared from PK15 cells (2–3 · 10 6 cells) using proteinase K ( Boehringer-Mann- heim) treatment [26] and phenol/chloroform extraction. PCR using pig genomic DNA (100 ng) and primers p1 and p2 (Table 1) hybridizing to 5¢ untranslated exons 2 and 3, respectively, w as carried out under the following conditions: one cycle of 2 min a t 94 °C, 15 s at 68 °Cand5minat 72 °C; 35 cycles of 10 s at 94 °C, 15 s at 65 °C, and 1 min at 72 °C i ncremented with 1 s at each cycle, followed by a final extension at 72 °C for 7 min. The amplified fragment (538 bp) was purified from a 2% agarose gel (Nucleotrap nucleic acid purification kit, Clontech), cloned into the vector pGEM-T easy (pGEM-T easy vector system, Promega) and sequenced (T7 sequencing kit, USB) using primers ( SP6 a nd T7) flanking the c loning site. This fragment was used as a probe to screen for phage clones containing the A /C region. A second probe w as generated from PK15 cDN A using primers p3 and p 4 ( Table 1) under the conditions described previously and was used to screen for ph age clones containing the p utative p romoter B region. Isolation of promoter regions B and A/C A pig genomic library (Clontech) was screened by a standard plaque hyb ridization m ethod, according to the manufac- turer’s recommendations. Two probes were used in these experiments. The first one, d esigned to isolate the promoter C region, contained the pig a1,3GalT exon 2/intron 2/exon 3 (generated as described above) and the second one, suitable for t he cloning o f promoter B region, contained exon 1. After plating (2.5–3 · 10 5 plaques, 5 · 10 4 plaques per Table 1. Sequences of oligonucleotides used. Primer name Sequence (5¢-3¢) Localization Target p1 TCAAACAGAACAACTTCTGAAGCC Exon 2 Promoter A p2 GCTCTGCTCTGCAGAAGGAGGC Exon 3 Promoter A p3 GCCACTGTTCCCTCAGCCGAG Exon 1 Promoter B probe p4 CTGATCGGCAGAAGCTGGGTG Exon 1 Promoter B probe p5 CCAAGGGTGGTGGCTGTCCCTC Exon 3 Promoter A p6 TGTCCCTGCTAGTTGTCATTTGG Intron 2 Promoter A p7 ACGACCACTTTGTCAAGCTCATT GAPDH p8 TGAGGTCCACCACCCTGTTG GAPDH p9 TCCTGAAACGCCTTCGGAAGAG E-selectin p10 CCATTGGGTTGAAGGCATTCG E-selectin p11 ACAAGGCCCCTGGCTGCT Exon 3 a1,3GalT-5¢-A p12 CCTGTCAAAAGAATAAACAGCGGTT Exon 3 a1,3GalT-5¢-A p13 CACTGTTCCCTCAGCCGAGGAC Exon 1 a1,3GalT-5¢-B and -E p14 CCAACTCCTGATCGGCAGAAGC Exon 1 a1,3GalT-5¢B and E p15 ACTTCTGAAGCCTAAAGGATGCGA Exon 2 a1,3GalT-5¢-C p16 AGGCAGGGCTGGGAGGAA Exon 3 a1,3GalT-5¢-C p17 TTGCTGTCGGAAGATACATTGAG Exon 8 a1,3GalT-coding region p18 CTTTGTGGCCAACCATGAAGTA Exon 9 a1,3GalT-coding region Ó FEBS 2002 Regulation of a1,3GalT expression in pig endothelial cells (Eur. J. Biochem. 269) 1465 plate), the plaques were transferred t o H ybond N + (Amersham) membranes (two membranes per plate) and hybridized in 40 m M Na 2 HPO 4 ,7%SDS,1m M EDTA with the labeled probe (10 6 c.p.m. per membrane) labeled with [a- 32 P]dCTP (3000 CiÆmmol )1 , Amersham) using the Prime-a-gene labeling system (Promega). Double-positive clones were purified by secondary and tertiary r ounds of screening and genomic DNA inserts were purified using poly(ethylene glycol) precipitation [26]. The phage inserts were mapped by restriction digestion and Southern blot analysis. Hybridizing DNA fragments were then subcloned in pBluescript SK + (pBS, Stratagene) and sequenced. Generation of a1,3galactosyltransferase promoter luciferase constructs All the promoter luciferase constructs were subcloned i n t he reporter plasmid pGL3-Enhancer (Promega). pGL3- Enhancer contains SV40 enhancer sequences, and is used rather than pGL3-Basic in view of the relatively low transcriptional activity of the a1,3GalT promoter. Clone A represents vector pGEM-T-easy containing the whole PCR fragment (nucleotides 1249–1786) correspond- ing to the putative promoter A region (see above). For construct A.1 (promoter A , construct 1), the P CR fragment (nucleotides 1249–1786) was excised from clone A with EcoRI. Construct A.2 (nucleotides 1388–1786) was gener- ated by PCR (primers p6 and p2, Table 1), in which clone A DNA was the template. In a s imilar w ay, const ruct A.3 (nucleotides 1483–1786) was also made by PCR, using primers p5 and p2 (Table 1). For construct A.4, the 5¢ part (nucleotides 1249–1611) of the insert w as delet ed from the clone A by Hi n cII digestion . For construct A .5, a Sty I- internal fragment was deleted from the clone A (Dnt1484– 1590). Construct A.6 was made using a blunted StyI-fragment of clone A (nucleotides 1484–1590). Construct A .7 consisted of a HincII fragment ( nucleotides 1249–1611) of clone A. For promoter B sequences, a BamHI fragment was used that originated from the hybridization-positive phage clone 2.3.1, isolated from the genomic library. This 2.7-kb DNA fragment was ligated with plasmid pBS. Part of the intronic sequence (intron 1, nucleotides 1385–2695) was deleted by SmaI digestion and re-closure, and the clone thus obtained (pBS clone B) was used for further constructs. Construct B.1 contained a Bgl II/HindIII fragment of 1.2 kb ( nucleotides 175–1385). An antisense construct (B.2, nucleotides 1385– 175) was also made using a BglII/SmaI digestion. The two other constructs (B.3 and B.4) contained SacIfragment nucleotides 29–1099 cloned in sense (B.3) or antisense (B.4) orientations. Constructs for the promoter C were made using the 1.8-kb BamHI fragment which was isolated from hybrid- ization-positive phage clone 2.1.3, and subcloned in pBS. Constructs C.4 and C.5 consisted of a BglII fragment (nucleotides 712–1219) and a SacI/Bgl II fragment (nucleo- tides 1–711), r espectively. Construct C .1 ( nucleotides 1–1219) was made from construct C.5 linearized with BglII by insertion of the BglII fragment from C.4 (nucleo tides 712–1219). Taking advantage of unique restriction sites in the insert of C.1 and in the pGL3-Enhancer plasmid, NdeI (Dnt1–164) a nd PvuII (Dnt1–475) fragments were also deleted, yielding constructs C.2 and C.3, respectively. The presence of the insert and its orientation in the plasmids was checked using restriction enzyme mapping and sequencing. For the position of the relevant restriction sites see b elow. Transient transfections and luciferase assays COS7 and PEC-A cells were grown in complete medium as described above (10 5 cells per well plated in 24-well plates). The DNA transfection complex was p repared by m ixing 0.5 lg of luciferase construct a nd 0.25 lgofpCH110 plasmiddilutedin30 lL of serum-free m edium a nd 3 lLof SuperFect reagent (Qiagen) per well. After 5–10 m in of incubation at room temperature, the mixture was diluted with 170 lL o f Dulbecco’s modified Eagle’s medium containing 10% of fetal bovine serum and added t o the cells. After 3 h of incubation at 37 °C, the m ixture w as replaced by 400 lL o f c omplete m edium. The b-galactosi- dase plasmid pCH110 (Amersham), containing the SV40 early promoter, served as an internal control for transfection efficiency. pGL3-Control ( i.e. pGL3-Enhancer plas mid containing the early promoter of SV40, Promega) was used as a positive control, and empty pGL3-Enhancer plasmid a s the n egative c ontrol. All the constructs were tested in two or three i ndependent experiments, each performed in triplicate. Luciferase reporter and b-galactosidase assays for cell extracts were performed 48 h after the start of the transfec- tion. Luciferase activity was measured using the Luciferase assay system (Promega) and 5 lL ( out of 60 lL) of cell extract in a BioOrbit-1250 luminometer (BioOrbit). b-Galactosidase activity was assayed using ortho-nitrophe- nyl-b- D -galactopyranoside as t he substrate, and t he am ount of reaction product was determined from the absorbance a t 420 nm. TNFa activation of pig endothelial cells and cDNA synthesis pPAEC and PEC-A cells were cultured as described above and were stimulated with 100 U ÆmL )1 of recombinant human TNFa (hTNFa; CLB, the Netherlands) added to the medium for different periods o f time (1, 2, 4, 8 , 12, 24, 48 and 72 h). After activation, cells were washed in phosphate buffered saline (NaCl/P i ),lysedinSVRNAlysisbuffer (175 lL per well), and total RNA was extracted using the SV total RNA isolation system (Promega) according to manufacturer’s recommendations. One microgram of t otal RNA was reverse transcribed using the R everse transcrip- tion system (Promega). Transcript quantification cDNAs w ere quantified using Q -PCR. Samples were run on the ABI PRISM 7700 sequence detector system using SYBR green PCR core reagents (PE Applied Biosystems). Q-PCR was carried out in a volume of 25 lL containing 12.5 ng of cDNA, 2.5 m M MgCl 2 ,0.2m M dATP, dCTP, dGTP and dTTP, 0.35 U Ampli-Taq Gold DNA and 0.14 U AmpErase uracil-N-glycosylase. The PCR condi- tions were as follows: 4 0 cycles of 1 5 s at 95 °Cand1minat 60 °Ceach. Primers (Table 1) f or pig glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (p7 and p8), E-selectin (p9 and 1466 D. Mercier et al. (Eur. J. Biochem. 269) Ó FEBS 2002 p10), a1,3GalT transcripts 5¢-A (p11 and p12), 5¢-B (p13 and p14), 5¢-C (p15 and p16), and total amount of pig a1,3GalT m RNA (p17 and p18) were used with cDNA from pPAECs or PEC-A cells (either activated or not with human TNFa) a s the template. GAPDH was used to normalize the quantity of cDNA used for each a ssay, and background due to primer dimerization was c hecked with nontemplate controls (reaction without cDNA). The activ- ation e fficiency of the endothelial c ells was tested by quantification of E-selectin transcripts (GenBank accession number L 39076) as a control. Ct values, corresponding to the cycle number required for fluorescence intensity to exceed an arbitrary threshold in t he exponential phase of the amplification (0.3 arbitrary units), were determined for all the samples and the gene to be analyzed. In addition, to quantify the mRNA copy numbers standard curves were generated. Plasmids containing exon 2-intron 2-exon 3 (pGEM-T-easy clo ne A), exon 1 (pBS clone B), exons 8 and 9 (PCR product obtained with primers p17 and p18, Table 1 , subcloned in pGEM-T-easy) or pig E-selectin cDNA, corresponding to transcripts 5¢-A and -C, 5¢-B and -E, a1,3GalT coding sequence and E-selectin, respect- ively, were selected. Various amounts of these different plasmids (from 10 3 to 10 6 copies per reaction) were used in Q-PCR assays, and data obtained for each concentration (2 Ct ) were plotted against the amounts. RESULTS The organization of the pig a1,3GalT gene Using 5¢ RACE analysis, we confirmed the occurrence in porcine endothelial cells of four transcripts 5 ¢-A, -B, - C and -E, described earlier by K atayama et al.[20].Inorderto complete the model of the pig a1,3GalT gene organization, we have compared the structure of the various a1,3GalT transcripts with p artial maps of the gene o rganization as established by Koike et al . [21] and Katayama et al. [20]. Transcripts A–E encode the same protein, but differ in the structure of their 5¢ ends by the presence or absence of one or more untranslated exons. These exons were mapped onto genomic s equences us ing long-distance PCR, w hich allowed us to establish the order of t he exons, and also to estimate intron sizes. A model of the a1,3GalT genomic structure, showing how the various transcripts are formed by a combination of alternative start site usage and alternative splicing, is presented in Fig. 1. By 5¢ RACE analysis of primary porcine endothelial cell ( pPAEC) cDNA we have identified a n additional, sixth transcript, termed 5¢-F. This transcript contains untranslated exons 0, 1 and 3, which confirms that exon 0 previously identified by Katayama et al. [20] is indeed authentic. In addition, PCR analysis of porcine genomic DNA showed that the s tart sites that give rise to transcripts 5¢-A and -C are closely spaced. An intron of only 427 bp separates exon 2 from exon 3. In fact, transcript A starts within this intron 2, 94 bp upstream of the start of exon 3. Similarly, sequences upstream of exon 1 either serve as intron (intron 0), or are retained in the processed mRNA, depending on the start site used. More recently, Koike et al . [21] detected two more transcripts starting in intron 0 (cf. Fig. 1). BasedonRT-PCR,piga1,3GalT is expressed in l ung and in all cell types investigated so far (kidney PK15 cells, hepatocytes, endothelial cells). Transcripts 5¢-B and/or -E have been detected in all samples, and 5¢-A in most of them with the exception of hepatocytes . The 5¢-C and - F transcripts are present in pPAECs. Transcript 5¢-D was not detected in any of the samples studied here. Cloning and sequence analysis of pig a1,3GalT promoter regions The available a1,3GalT cDNA sequences (this paper and [20]) were used to generate DNA probes by PCR, with the aim t o i solate relevant 5¢ flanking sequences of the g ene from a genomic library. As start sites A and C had been found to be closely spaced, a single probe was sufficient to screen the library for their individual regulatory sequences. To isolate the genomic region upstream of the 5¢-A transcript, we performed PCR o n pig genomic DNA using primers p1 and p2 that hybridize with exons 2 and 3, respectively. A fragment of 538 bp was obtained ( clone A), which contains e xon2-intron2-exon3 sequences that overlap with the putative promoter A region (Fig. 2B, GenBank accession number: AF415202). This fragment was used as a probe to screen the pig genomic library. A single hybrid- ization-positive clone (phage 2.1.3) containing an insert of 14 kb was isolated. Southern blot analysis of the phage DNA confirme d t hat a major portion of the probe sequence is included in a 1.6-kb BamHI fragment (Fig. 2A). This DNA fragment contains 1.2 k b of sequence upstream of start site C (Fig. 2B, GenBank accession number: AF415202), as well as 430 bp of clone A. Sequences thus obtained were scanned for putative transcription factor binding sites using TRANSFAC. E xon 3 is preceded by a well-conserved pyrimidine-rich acceptor splice site which is functional i n all transcripts 5¢-B, - C and -F. The 538-bp sequence of clone A does not contain TATA o r CAAT- boxes, but multiple putative transcription factor binding ATGExon 0 Exon 1 Exon 2 Exon 3 Exon 4 5'-A 5'-B 5'-E 5'-C 5'-D 5'-F 16 kb0.5 kb n.d. n.d. D B C A Koike et al [21] Fig. 1. Schematic representation of the a1,3galT gene and transcript structures. Exon numbers are indicated above the correspo nding exons on the gene structure . The sizes of introns 2 and 3, determined using long-PCR, are indicated below the gene stru cture. T he length of introns 0 and 1 could not be determine d (n.d.). Untranslated exonic sequence is indicated b y white or gray boxes, co ding sequence is indicated in black . Gray b oxes are fl anked on b oth sides by consensus splice sites. Structures of the 5¢ e nd of the different transcripts identified in this paper (5¢-A, -B, -E and -F) or by Katayama et al.(5¢-A to -E [20]), and/or Koike et al. [21] are drawn below the gene structure. Ó FEBS 2002 Regulation of a1,3GalT expression in pig endothelial cells (Eur. J. Biochem. 269) 1467 sites that could be important for promoter activity in pig cells, such a s GATA- and GC-boxes, AP-1, Inr and YY1 are present (Fig. 2B). Analysis of the sequences upstream o f t he exon 2 transcriptional start site revealed the presence o f four putative NF-jB binding sites (Fig. 2B) located at nucleo- tides 86, 167, 371 and 552, respectively. A dditional potent ial transcription factor binding sites such as Oct-1, AP-1, GATA- a nd GC-boxes are distributed all along the promoter C sequence, and a TATA-box is present 16 bp downstream of the transcriptional start site of exon 2 (Fig. 2B). In order to clone the putative promoter B region, the genomic library was screened with a probe corresponding t o exon 1. Phage clone 2.3.1 thus isolated contained an 11-kb insert; BamHI digestion of the DN A produced a 2.7-kb fragment that hybridized with the p robe (Fig. 3A, GenBank accession number A F415201). This f ragment contains 1.18 kb of sequence upstream of exon 1, exon 1 itself, and 1.36 kb of intron 1. The GC c ontent of the whole fragment is about 60%, and in the 1.6-kb region between nucleotides 724 a nd nucleotide 233 5 (Fig. 3B) it reaches 68%. Associ- ated with the high GC content of this r egion, 12 putative Sp1 binding sites are present. In addition, the promoter B region contains numerous putative t ranscription factor binding sites including GATA-boxes, Oct-1, e ts-1, AP-1, NF-jB and C/EBP sites (Fig. 3B). Unfortunately, out of the 6 · 10 5 plaques s creened with a probe corresponding to exon 0, no positive genomic clones containing exon 0 upstream sequences (promoter D region) were isolated. The stro ng homology of e xon 0 sequences with a portion of the porcine invariant chain gene [21] could be responsible of the isolation of the false positive clones from the l ibrary. To search for preferred transcriptional start s ites, a RNA polymerase II context analysis was performed on promoter regions A, B and C using PROMOTERINSPECTOR software (Genomatix). Results indicated that promoter region B contains two putative RNA polymerase II binding regions located at nucleotides 677–1296 (upstream of exon 1) and nucleotides 2 201–2392 (within intron 1), r espectively, whereas promoters A and C do not seem to contain such regions. Functional characterization of the porcine a1,3GalT promoters Luciferase reporter gene assays were performed to test the ability of the clon ed sequences to drive transcription. To characterize the promoters in more detail, a deletion analysis was carried out. The various test fragments were placed upstream of the luciferase gene, a nd the resulting plasmid constructs were transiently transfected into cultured cells. In view of the relatively low efficiency of transfection of primary endothelial c ells, these experiments were p er- formed using the e stablished cell line PEC-A [25]. Construct A.1 that contains the entire 5 38-bp promoter A fragment is able to drive luciferase gene transcription in PEC-A pig endothelial cells (Fig. 4A, open bars). Deletion of the 140 bp 5¢ portion of the fragment t o give A.2 did not affect activity, but deletion of an additional 9 6 bp ( construct A.3) re sulted in a fivefold lower luciferase a ctivity (Fig. 4A). A segment of 1.2-kb containing putative promoter C regulatory regions was also analyzed. The full 1.2-kb sequence (construct C.1, containing nucleotides 1–1219) was able to drive transcription in PEC-A cells (Fig. 4 B, open bars). Deletion of a 164-bp (C.2) or 719-bp (C.4) 5¢ fragment resulted in a fourfold and t wofold reduction in luciferase activity, respectively. ForpromoterB,afragmentof1.2 kb (nucleotides 175– 1385 in Fig. 3B) containing most of the GC-rich region was studied. When transiently transfected into PEC-A cells, construct B.1 produced a luciferase a ctivity ninefold greater than negative control ( Fig. 4C). Deletion of the 3 ¢ 286-bp portion (construct B.3) did not change the activity, which 5'-C [20] A 1 2 3 4 6 3 1.5 0.5 Size in kb AP-1 Oct-1 NF-κB GATA Nde I Pvu II GATA AP-1 BamH I NF- κB NF-κB 1 50 100 150 200 250 300 350 400 450 500 ets-1GATA GATASp1 Bgl II C/EBPNF-κB Sp1 Sp1 Sp1 501 550 600 650 700 750 800 850 900 950 1000 Sp1 TATA AP-1 Bgl II Ap-1 p1 ets-1 GATA YY1 GATA Sty I exon 2 NF-κB 1001 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500 GATA Sp1 Sty I Hinc II BamH I 5'-A [20] GATAAp-1 GATA Inr p2 Sp1 exon 3 1501 1550 1600 1650 1700 1750 1800 B Fig. 2. Schematic representation of the a1,3GalT promoter A an d C regions. (A) Southern blot analysis. DNA prepared from phage 2.1.3 isolated from the pig genomic library was digested with various restriction enzymes, and hybridized with the clone A DNA fragment (see Materials a nd m eth ods). Siz es o f t he d ifferent bands of t he DN A marker are indicated on the left of the figure. Lane 1: PstI, lane 2: HindIII, lane 3: EcoRI , lane 4: BamHI. (B) Schematic representation of th e a1,3GalT promoter A and C regions. Positions of primers used to gen erate the 5 38-bp fragment ( Table 1) a re positioned un der the sequence and are indicated by horizontal arrows. Restriction enzyme sites used to generate the different reporter gene con structs are indi- cated by vertica l lines ab ove the sequen ce. Exonic sequences (exon 2 and 3) a re indicated b y gray boxes. The sta rt sites of transcript 5¢-A and -C are in dicated [20]. Putative transcription factor b inding sites detected using TRANSFAC are indicated above the seque nce a nd are represented by ho rizon tal lines. 1468 D. Mercier et al. (Eur. J. Biochem. 269) Ó FEBS 2002 underlines the importance of the 5¢ part of the fragment. The same fragments in antisense orientation did not differ significantly from the negative control (cf. B.2 and B.4). To determine whether the constructs mediated cell-type specific expression, they were also transfected into African green monkey COS7 c ells. Each of the co nstructs A.1, B.1 and C.1 was able to drive transcription, and, generally, activities observed in COS7 cells were higher than those in PEC-A cells. Deletion of the 140-bp 5¢ portion of A.1 (construct A.2) resulted in a 10-fold increased luciferase activity (Fig. 4 A). Upon deletion o f an a dditional 96-bp (construct A.3) activity decreased to a value close to that of A.1. Deletion constructs A.4 to A.7, even w hen containing fragment nucleotides 1388–1483, were inactive (Fig. 4 A). For p romoter C, a ctivity i n COS7 c ells seems to be associated with the 0.5-kb 3¢ segment o f C.1, as the 0.7-kb 5¢ portion by itself (C.5 in Fig. 4B) is 10 times less active than the full-length fragment C.1. C onstruct B .1 produced a luciferase activity four times higher than in P EC-A cells. The 2 BamH I Bgl IINde I Pvu II Bgl II 0 50 100 150 200 1000 6000 C.3 123 C.5 12 5136 48 pGL3-Control C.1 158 85 C.2 123 20 C.4 120 48 pGL3-Enhancer 5 5 Normalized Luciferase Activity (mV) B 0 50 100 150 200 1000 6000 B.1 202 45 B.2 58 5.5 B.3 159 42 1 Bgl IISac I Sac I Sma I HinD III B.4 39 7.5 pGL3-Control 5136 48 pGL3-Enhancer 5 5 Normalized Luciferase Activity (mV) C 5136 2 3 Sty I Sty IEcoR I EcoR IHinc II Hinc II A.4 0 A.5 0 A.6 0 A.7 0 0 50 100 150 200 1000 6000 A.1 113 199 A.3 165 40 pGL3-Enhancer 5 pGL3-Control 5 48 1170 A.2 186 Normalized Luciferase Activity (mV) A Fig. 4. Transcriptional activity of a1,3GalT promoter constructs in COS7 and PEC-A cells. Theleftpartofthefigureshowsthestructure of constructs made for p romoter A (A), C (B) and B (C), and their relative positions in th e a1,3 GalT gene. Exonic (gray boxes), and intronic (solid lines) sequences and r estriction enzyme sites are indicated as well as s equences derived from the plasmid pBS (vertical lines). For each construct, the segment of genomic sequence tested in luciferase assay is i ndic ated by horizontal gray b ars. The right part of eac h panel shows the results of transfection experiments for each construct; values (in mV) are the means of three or fo ur separate experim ents, per- formed in triplicate, ± SEM. Luciferase activities are normalized on b-galactosidase activity from a cotransfected vector (see Materials and methods). Solid and open b ars correspond to COS7 a nd PEC-A transfection, respectively. Constructs A.4 to A.7, C.3 and C.5 were only tested in COS7. A 6 3 1.5 0.5 Si ze in kb 1 2 3 4 5 B GATA GATA GATA Bgl II Koike et al [21] Ap -1 GATA BamH I 1 50 100 150 200 250 300 350 400 450 500 Oct -1 Oct -1 ets-1 Oct-1 Sp1 GATA GATA NF- κB 501 550 600 650 700 750 800 850 900 950 1000 Koike et al [21] Sp1 Sp1 Sp1 Sp1 Sp1 Sp1 Sp1 AhR/Arnt Sp1 Sac I 5'-E [20] GATA exon 1 10011050 1100 1150 1200 1250 1300 1350 1400 1450 1500 Sp1 5'-B [20] GATA Sma I Sp1 GATA GATA GATASp1 Arnt CREBets-1 1501 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000 Sp1 Sp1 C/EBP Oct-1 GATA Sp1 GATA ets-1 NF- κB 2001 2050 2100 2150 2200 2250 2300 2350 2400 2450 2500 C/EBP C/EBP BamH I 2501 2550 2600 2650 2700 GATA GATA Sac I Fig. 3. Schematic representation of the a1,3GalT promoter B region. (A) Southern b lot analysis. D NA of p hage 2.3.1, isolated from th e pig genomic library, w as digested with various restriction enzymes, a nd hybridized with a probe corresponding to exon 1 (see Mater ials and methods). Sizes of the different bands of the DNA marker a re indi- cated on the left of the figure. Lane 1: SacI, lane 2: PstI, lane 3: HindIII, lane 4: EcoRI, lane 5: BamHI. (B) Schematic representation of the promoter B region. Restriction enzym e sites used to generate the different constructs are indicated by vertical lines above the sequence. Exon 1 sequences are i ndicated by a gray box. The main transcrip- tional start site used in t ranscripts produced f rom promoter B is indicated [20]. Putative t ranscription factor binding sequences are indicated above the sequence and are represented by ho riz ontal lin es. The GC-rich region of pro moter B i s indicate d by a thi ck line. Ó FEBS 2002 Regulation of a1,3GalT expression in pig endothelial cells (Eur. J. Biochem. 269) 1469 same fragment inserted in pGL3-enhancer in antisense orientation was 3.5-fold less active (Fig. 4 C). Values obtained f or the 3¢ truncated fragments B.3 and B.4 followed the same pattern as those observed for PEC-A cells (Fig. 4C), in that the ability to drive transcription is orientation dependent, and that 3¢ deletion of 286 b p did not significantly alter activity. Relative levels of a1,3GalT transcripts in pig endothelial cells Levels o f a1,3GalT transcripts in pPAEC and in a PEC-A were measured using Q -PCR in order t o establish t he relative importance, within the context of the full gene, of the three different promoters identified above. Taking advantage o f the sequence differences between the 5¢ regions of the promoter specific transcripts, and assuming that the differ- ences observed in terms of transcript levels are proportional to promoter activity, specific primers ( Table 1) were designed to follow variations of a1,3GalT g ene expression in resting and TNFa-stimulated pig endothelial cells. The strong homology of exon 0 to a portion of the porcine invariant chain gene did not allow to d esign p rimers specific of exon 0 and to quantify t he 5¢-D/5¢-F transcripts. Additional primers were designed to determine expression of GAPDH (normalization of cDNA quantities used in Q-PCR) and E-selectin (control o f TNFa stimulation) genes. Before activation of pPAEC, E-selectin transcripts w ere present at  2 · 10 5 copies per lg of total RNA, and a1,3GalT transcripts at 1.1 · 10 6 copiesÆlg )1 . A mounts of 10 4 ,10 6 and 2 · 10 4 copiesÆlg )1 of total RNA were measured for 5¢-A, 5¢-B and 5¢-C transcripts, respectively. During TNFa-induced activation, E-selectin transcript levels rapidly increased, reaching a maximum of abo ut 60 · 10 6 copiesÆlg )1 after only 2 h of TNFa-treatment (340-fold increase, Fig. 5A). A second peak was observed after 24 h of induction (21 · 10 6 copiesÆlg )1 ,Fig.5A), clearly indicating that the cells were properly activated in this experiment. Total mRNA levels of a1,3GalT were also checked during the time course of TNFa induction, as well as the levels of 5¢-A, 5¢-B and 5 ¢-C transcripts. Total a mount of a1,3GalT transcripts started to rise 2 h after the addition of TNFa, to reach a plateau after 4 h of activation (55% increase, to 1.8 · 10 6 copies per lg of total RNA, or  35 copies per cell). After 12 h the amount began to decrease to reach 0.3 · 10 6 copiesÆlg )1 ( 6 copies p er cell) after 72 h of stimulation. The 5 ¢-A and 5 ¢-B transcript le vels varied in parallel with the total amount of a1,3GalT transcripts, whereas quantities of 5¢-C transcripts are regulated differently with two peaks of transcription, a first one after 2 h ( increase of 82%) a nd a second one at 12 h (119% increase). At any activation time point studied, 5¢-B/E transcripts (which could not be distinguished by t he method as used ) were f ound to correspond to 92–97% of the total amount of a1,3GalT mRNA. Unlike the pPAECs, PEC-A cells were poorly activated by recombinant human TNFa (E-selectin increased only 2.5-fold, Fig. 5B). Nevertheless, quantities detected for each of the a1,3GalT transcripts were similar to those observed i n pPAECs. The total amount of a1,3GalT increased signifi- cantly and reached a maximum of 141 % after 1 8 h of TNFa activation. Furthermore, 5¢-B was found to be also the most expressed transcript in PEC-A cells (80–90%) and transcripts 5¢-A and 5¢-B followed t he same pattern of variation as total a1,3GalT mRNA. But in contrast to pPAECs expression, after longer periods of activation no decrease was o bserved for 5¢-A , 5¢-B and total a1,3GalT mRNAs. Both 5¢-A and 5¢-C transc ript quantities seem to be higher in PEC-A cells (3% and 7% of the total amount of a1,3GalT mRNAs, respectively) than in pPAECs (1% for both). Lastly, the 5¢-C transcript presented a different expression profile with only one peak reached after 4 h of activation followed by a fourfold down-regulation after longer stimulation. DISCUSSION Differences in cell surface glycosylation between hu man and pigs form a major hurdle in organ transplantation from pig Fig. 5. Kine tics of a1,3GalT mRNA isoforms induction in pPAEC (A) andPEC-AcellsduringTNFa activation (B). The expression of a1,3GalT after stimulation with TNFa was quantified using Q-PCR. Primers specific of the 5¢ region of the different i soforms were u sed t o quantify specifically each transcript sp ecies (5¢-A, 5¢-B and 5 ¢-C). Total amount of a1,3GalT transcripts was estimated using primers binding to the cod ing sequence (exons 8 and 9). E ffective activation of th e cells was verified b y amplifi cation of E-selectin mRNA. The nu mb er o f transcripts in the experimental samples was calculated from a c alib- ration curve obtained by v arying the number of copies of a plasmid containing the fragment to be amplified, and normalized based on GAPDH levels. 1470 D. Mercier et al. (Eur. J. Biochem. 269) Ó FEBS 2002 to man. Modification of porcine glycosylation has been considered as one strategy to facilitate xenotransplantation. In this respect, it will be important to know the mechanism of regulation of porcine terminal glycosylt ransferases. Research focuses on a1,3GalT in p articular, as the latter enzyme produces the Gala1,3Gal struct ure, the m ajor porcine xenoantigen with a role both i n hyperacute r ejection and in delayed vascular r ejection. Here we have assembled the full structure of the 5¢ flanking regions of the porcine a1,3GalT g ene, completing partial structures as reported by K atayama et al. [20] and Koike et al. [21]. The gene consists of 10 exons, four of whichcontain5¢ untranslated sequence and six coding sequence. The exact structure of th e 5 ¢ flanking regions of the a1,3GalT gene has been unclear. Koike et al. [21] have suggested that exon 0 as detected by K atayama et al. [20] and by t hemselves ( named e xon I i i n [ 21]) i n f act is b ased on an artifact, and could be the result of an accidental link-up of two unrelated sequences. Independently, we have isolated from porcine endothelial cells a t ranscript, 5¢-F, that does contain exon 0 in conjunction with additional, downstream a1,3GalT exons 1 and 3–9. The occurrence o f t his transcript in porcine cells seems to indicate that exon 0 as d escribed previously is indeed an authentic portion of the a1,3GalT gene. This would bring the total number of 5¢ noncoding exons up to four. At least four promoters, here called A –D, are involved in the initiation of transcription of porcine a1,3GalT. Alter- native start site usage together with alternative splicing in the 5¢ region generates at least six different transcripts. Moreover, it i s possible t hat two m ore transcripts a s described by Koike et al. [21] are controlled by still another regulatory sequence, as they initiate several hundreds of bp upstream of the start of transcript 5¢-B (Fig. 3B). Alternat- ive splicing of the porcine gene is not limited to the 5¢ flanking sequences, i t also occurs in the coding region. Previously, i t w as reported that the murine a1,3GalT gene is alternatively spliced in the sequence that encodes the Ôstem regionÕ of the protein [13]. A similar observation has b een made by Vanhove et al. [27, 28] for t he porcine gene, which further increases the number of transcripts that can be obtained from this single gene. As yet, it is unclear if the occurrence of multiple transcript isoforms has a physiolo- gical relevance. Heterogeneity at the 5¢ end o f t he mRNA does not affect the protein encoded. In contrast, splicing in the stem region will result in the production of a shortened enzyme molecule, w hich may be l ess sensitive to intracellular proteolysis, or differ in its ability to transfer galactose to Galb1, 4GlcNAc structures [20]. Various 5¢ flanking regions have been tested for t heir ability to drive a1,3GalT gene transcription, and a deletion analysis has been carried out to identify minimal promoter regions. E ach o f t he promoters A, B , and C w as found to be active in porcine endothelial cells. Sequence analysis of promoter region A, the 479-bp region directly upstream of exon-3, revealed the presence of several putative transcrip- tion factor binding sites (Fig. 2B). The region contains five GATA(like) sites. One of these, GATA nucleotides 1621– 1624, is in close proximity to an AP-1 motif . Cooperative interactions between AP-1 and GATA were reported to regulate transcription driven b y the human P-selectin promoter [29, 30]. For porcine a1,3GalT, the AP-1/GATA motif is located just upstream of the start site (at nucleotide 1633) of transcript 5¢-A. This start site is part of an octanucleotide t hat i s h ighly s imilar t o t he consensus transcriptional initiator sequence [31, 32]. The initiator sequence, together with the AP-1/GATA motif, is probably important for the production of transcript 5¢-A in porc ine endothelial cells. However, additional upstream sequence is essential for transcriptional activity. Construct A.2 that contains the nucleotides 1388–1786 region was found to be five times more active in PEC-A cells (luciferase assays, Fig. 4A) than construct A.3 (nucleotides 1483–1786). This suggests that a transcriptional activator binds to the region nucleotides 1388–1483. The segment needs to be linked to the transcriptional start site via StyI-fragment nucleotides 1483–1590, as deletion of the latter fragment results in zero activity (Fig. 4A). Apart from a single G ATA-box, no known transcription factor binding site is present in region nucleotides 1388–1483 (Fig. 2 B). The GATA box shows only imperfect homology with the consensus sequence. Therefore, activation by t he nucleotides 1388–1483 segment may r esult from the binding of a s till unknown transcription factor. The 538-bp promoter A fragment can also drive transcription in COS7 cells, s o does not appear to confer cell type specificity. A second porcine genomic DNA fragment was isolated that contains sequences u pstream of exon 2 (nucleotides 1–1219 in Fig. 2B). This putative promoter C contains multiple transcription factor binding sites, including five NF-jB sites. The latter sites could be important in endothelial cell-specific expression and i n the cytokine response to TNFa [33, 34]. It has been reported that TNFa can induce t he expression o f a1,3GalT [27], a nd NF-jBsites are possibly involved in mediating this effect. Other transcription f actors such as Sp1, G ATA or ets-1 as detected in promoter C (Fig. 2B) could also be important for endothelial cell-specific expression [35–38]. Reporter gene assays have indicated t hat regions important for promoter activity are mostly located in the 3¢ portion of the fragment (nucleotides 719–1219). This region contains only asingleNF-jB site. The presence of a TATA box at nucleotide 1196 may help to direct initiation specifically to the position nucleotide 1220. An enhancer may b e present in the region nucleotides 1–164 of promoter C b ecause deletion of that region results in a fourfold reduction of luciferase activity in transfected PEC-A cells. As shown by reporter luciferase assays, similar activities are observed in both COS7 and PEC-A cells, suggesting that promoter C does not contain species-specific regulatory sequences. The analysis of promoter region B confi rmed that the region directly upstream of exon 1 contains a GC-rich sequence o f  1.5 kb, as reported earlier by Koike et al. [21]. Consequently, numerous Sp1 binding sites h ave been predicted in t hat region (Fig. 3B) . The lack of TATA o r CAAT-boxes together with the presence of many Sp1 binding sites, as observed f or promoter B, is a c haracteristic of ÔhousekeepingÕ genes. For most of these genes, transcrip- tional s tart is likely to b e imprecise, and indeed a set of transcripts differing in their 5¢ ends is produced from promoter B [20, 21]. S everal glycosyltransferase p romoters present similar structure and characteristics [39–42]. For example, the promoter of the long form of b1,4-gala ctosyl- transferase contains 12 Sp1 binding sites, and is active in a variety of cell types [42, 43]. Two putative NF-jB b inding sites have been found in promoter B sequence (nucleotides Ó FEBS 2002 Regulation of a1,3GalT expression in pig endothelial cells (Eur. J. Biochem. 269) 1471 898–907 and nucleotides 2398–2407) suggesting that this promoter may respond to endothelial c ell activation. Interestingly, a1,3GalT transcripts generated from pro- moter B contain a GC-rich 5¢ untranslated region, which is predicted to form stable h airpin loops. This m ay interfere with the efficiency of translation of the mRNA as has been shown fo r b1,4-galactosyltransferase [44]. I n that way, increased transcription from promoter B could be compen- sated for by low t ranslation efficiency. We have established w hich promoter is used preferen- tially in porcine endothelial cells, and how a1,3GalT gene transcription is affected by TNFa-activation of endothelial cells. Results obtained for nonactivated p PAEC and P EC-A cells were similar. In both cell types the 5 ¢-B transcript is the most highly expressed i soform (Fig. 5 A,B), and corresponds to  80–90% of the total amount of a1,3GalT transcript. This indicates that promoter B is the main a1,3GalT promoter implicated in endothelial cell expression. Prefer- ence for promoter B is not affected by TNFa stimulation. Organ transplantation from pig to man results in an activation of the donor organ vascular endothelium, con- comitant with changes in cell surface structures. Activation of endothelial cells by TNFa treatment was reported to enhance a1,3GalT expression [28]. Using Q-PCR we found that, initially, q uantities of a1,3GalT transcripts indeed increased slightly (25–50%, F ig. 5). However, this increase was followed over time by a strong decrease (fivefold) in primary endothelial cells, whereas no down-regulation was observed in PEC-A. This indicates that the response of a1,3GalT to TNFa stimulation in PEC-A is different from that inprimary cells. The same holds forE-selectin expression (about 300-fold increase in pPAECs vs. 3 -fold in PEC-A). It remains to be investigated to what extent changes in endothelial levels of a1,3GalT and other t erminal glycosyl- transferases under inflammatory conditions will affect cell surface c arbohydrate structure, and thus influen ce t he outcome of organ transplantation. The results obtained in this study will help us to manipulate t he expression of a1,3GalT in porcine cells and tissues in a precise way. It is hoped t hat, ultimately, this approach will facilitate clinical xenotransplantation. ACKNOWLEDGEMENTS This work was s upported b y the EU biotechnology project on xenotransplantation N° BIO4-CT97-2242. REFERENCES 1. Bu ¨ hler, L., Friedman, T., Lacomini, J. & Cooper, D.K.C. (1999) Xenotransplantation-state of the art-Update. Front. Biosci. 4, d416–d432. 2. Soin, B., Vial, C.M. & Friend, P.J. (2000) Xenotransplantation, Br. J. Surg. 87, 138–148. 3. Saadi, S. & P latt, J.L. (1997) Immunology of xe notransplantation. Life Sci. 62, 365–387. 4. Bach, F.H. (1998) Xenotransplantation: problems and prospects. Annu. Rev. Med. 49, 301–310. 5. Daniels, L.J. & Platt, J.L. (1997) Hyperacute xenograft r ejection as an immunologic b arrier to xenotransplantation. Kidney Int. 51, S28–S35. 6. Sandrin, M.S. & McKenzie, I.F. (1994) Gal alpha(1,3)Gal, t he major xenoantigen (s) recognised in pigs by human natural anti- bodies. Immunol. Rev. 141, 169–190. 7. Samuelsson, B.E., Rydberg, L., Breimer, M.E., Backer, A., Gustavsson, M., Holgersson, J., Karlsson, E., Uyterwaal, A C., Cairns, T. & Welsh, K. (1994) Natural antibodies and human xenotransplan tation. Immunol. Rev. 141, 151–168. 8. Cooper, D .K., Koren, E. & Oriol, R . (1994) Oligosaccharides and discordant xenotransplantation. Immunol. Rev. 141, 31–58. 9. Galili, U., C lark, M .R., Shohet, S.B., Buehler, J . & Macher, B .A. (1987) Evolutionary relationship between the natural anti-Gal antibody and the Gala1 fi 3Gal epitope in primates. Proc. Natl Acad. Sci. USA 84, 1369–1373. 10. Galili, U., Shohet, S.B., Kobrin, E., Stults, C.L. & Macher, B.A. (1988) Man, apes, and Old World monkeys differ from other mammals in the expression of alpha-galactosyl epitopes on nucleated cells. J. Biol. Chem. 263, 17755–17762. 11. Oriol,R.,Barthod,F.,Bergemer,A M.,Ye,Y.,Koren,E.& Cooper, D .K.C. (1994) Monomorphic and polymorphic carbo- hydrate antigens on pig tissues: implication for organ xeno- transplantation i n the pig-to-human mo del. Tranplant. Int. 7, 405–413. 12. Joziasse, D.H., Shaper, J.H., van den Eijnden, D.H., Van Tunen, A.J . & S haper, N.L. (1989) Bovine alpha 1–3-galacto- syltransferase: isolation a nd c haracterization of a cDNA clone. Identification of homologous sequences in human ge no mic DNA. J. Biol. Chem. 264, 14290–14297. 13. Joziasse, D.H., Shaper, N.L., Kim, D., van den Eijnden, D.H. & Shaper, J.H. (1992) Murine alpha 1,3-galactosyltransferase. A single gene lo cus specifies four i soforms of the enzyme by alter- native splicing. J. Biol. C hem . 267, 5534–5541. 14. Joziasse, D.H., Shaper, J.H. & S haper, N.L. (1999) The a1,3-Galactosyltransferase gene. I n a-Gal an d Anti-Gal ( Galili, U. & Avila, J.L., eds), p p. 25–47. K luwer Academic/Plenum Pub- lishers, New York. 15. Sandrin, M.S., Dabkowski, P.L., H enning, M.M., Mouhtouris, E . & McKenzie, I.F.C. (1 994) Ch aracterization of cDNA clones for pig a(1,3)galactosyl transferase: the enzyme generatin g the Gala(1,3)G al epitope. Xenotra nsplantation 1, 81–88. 16. Larsen, R.D., Rajan, V.P., Ruff, M.M., Kukowska-Latallo, J., Cummings,R.D.&Lowe,J.B.(1989)IsolationofacDNA encoding a murine UDPgalactose: beta- D -galactosyl-1,4-N-acetyl- D -glucosaminide alpha-1,3-galactosyltransferase: expression clon- ingbygenetransfer.Proc. Natl Acad. Sci. USA 86, 8227–8231. 17. Larsen, R.D., Rivera-Marrero, C.A., Erns t, L.K., Cummings, R.D. & Lowe, J.B. (1990) Frameshift and nonsense mutations in a human g enomic sequence ho mologous to a murine UDP-Gal:beta- D -Gal(1,4)- D -GlcNAc a lpha(1,3)–galactosyltrans- ferase cDNA. J. Biol. Chem. 265, 7055–7061. 18. Joziasse, D.H., Shaper, J .H., Jabs, E.W. & Sha per, N.L. (1991) Characterization of an al pha 1–3-galactosyltransferase h omologue on human chromosome 12 that is organized as a processed pseudogene. J. Biol. Chem. 266, 6991–6998. 19. Shaper, N.L., Lin, S .P., Joziasse, D.H., Kim, D.Y. & Y ang-Feng, T.L. (1992) A ssignment of two human alpha-1,3-galactosyl- transferase gene sequences (GGTA1 and GGTA1P) to chromo- somes 9q33-q34 and 12q14-q15. Genomics. 12, 613–615. 20. Katayama, A., Ogawa, H., Kadomatsu, K., Kurosawa, N., Kobayashi, T., Kaneda, N., Uchimura, K., Yokoyama, I., Muramatsu, T. & Takagi, H . (1998) Porcine alpha-1,3-galactosyl- transferase: full length cDNA cloning, genomic organization, and analysis of sp lici ng var iants. Glycoconj J. 15, 583–589. 21. Koike, C., Friday, R.P., Nakashima, I., Luppi, P., Fung, J.J., Rao, A.S., Starzl, T.E. & Trucco, M. (2000) Isolation o f t he re gulatory regions and genomic organization of the porcine alpha1,3- galactosyltransferase gene. Transplantation. 70, 1275–1283. 22. d’Apice, A.J., Tange, M.J., Chen, G.C., Cowan, P.J., Shinkel, T.A. & Pearse, M.J. (1996) Two genetic approaches to the galactose alpha 1,3 galactose xenoantigen. Transplant Proc. 28 , 540. 1472 D. Mercier et al. (Eur. J. Biochem. 269) Ó FEBS 2002 23. Strahan, K., Preece, A. & Gustafsson, K. (1996) Pig alpha1,3 galactosyltransferase: a major target for genetic manipulation in xenotransplantation. Front. Biosci. 1, e34–e41. 24. McKenzie, I.F., Osman, N ., Co hn ey, S ., Vau ghan, H .A., Patton, K., Mouhtouris, E., Atkin, J.D., Elliott, E., Fodor, W .L., Squinto, S.P., Burton, D., Gallop, M.A., Oldenb urg, K.R. & Sandrin, M.S. (1996) Strategies to overcome th e anti-Gal a lpha (1–3) G al reac- tion in xenotransplantation. Transplant Proc. 28, 537. 25. Khodadoust, M.M., Candal, F.J., Maher, S .E., Murray, A.G., Ades, E .W. & Bothwell, A.L.M. (1995) PEC-A: An immort alized porcine endothelial cell. Xenotransplantation 2, 79–87. 26. Sambrook, J., Fritch, E.F. & Maniatis, T. (1989) Molecular Cloning: a Laboratory Manual, 2 nd edn. Cold Sp ring Harbor Laboratory Press, New York. 27. Vanhove, B., Goret, F., Soulillou, J.P. & Pourcel, C. (1997) Por- cine alpha1,3-galactosyltransferase: tissue-specific and regulated expression of splicing i soforms. Biochim. Biophys. Acta. 1356, 1–11. 28. Vanhove, B., Sebille, F., Cassard, A., Charreau, B. & Soulillou, J.P. (1998) Transcriptional and posttranscriptional regulation of alpha 1,3-galactosyltransferase in activated endothelial cells results in de creased expression of Gal alpha 1,3Gal. Glycobiology 8, 481– 487. 29. Pan, J., Xia, L. & McEver, R.P. (199 8) Comparison of promoters for the murine and human P- selectin genes suggests species-specific and conserved mechanisms for transcriptional regulation in endothelial cells. J. Biol. Chem. 273, 10058–10067. 30. Foletta, V.C., Segal, D.H. & C ohen, D.R. ( 1998) Transcriptional regulation in the immune system: all roads lead to AP-1. J. Leu- kocyte Biol. 63, 139–152. 31. Weis, L. & Reinberg, D. (1997) Accurate positioning of RN A polymerase II on a n atu ral TATA-less promoter is independent of TATA-binding-protein-associated factors and initiator-binding proteins. Mol. Cell Biol. 17, 2973–2984. 32. Weis, L. & Reinberg, D. (1992) Transcription by RNA poly- merase II: initiator-directed formation of transcription-competent complexes. FASEB J. 6, 3300–3309. 33. Mantavoni, A., Bussolino, F. & Dejana, E. (1992) Cytokine reg- ulation of endothelial cell function. FASEB J. 6, 2591–2599. 34. Mantavoni, A ., Bussolino, F. & Introna, M. (1997) Cytokine regulation of endothelial cell f unction: fro m molecular level to the bedside. Immunol. Today 18, 231–240. 35. Karantzoulis-Fegaras, F., Antoniou, H., Lai, S L.M., Kulkarni, G., D’Abreo, C., Wong, G.K.T., Miller, T.L., Chan, Y., Atkins, J., Wang, Y. & Marsden, P.A. (1999) Characterization of the human endothelial n itric-oxyde synthase promoter. J. Biol. C hem. 274, 3076–3093. 36. Cowan, P .J., Tsang, D., Pedic, C .M., Abbott, L.R., Shinkel, T.A., d’Apice, A.J.F. & Pearse, M.J. (1998) The human ICAM-2 pro- moter is end othelial cell-spec ific in vitro and in vivo and contains critical Sp1 and GATA binding sites. J. Biol. Chem. 273 , 11737– 11744. 37. Iljin, K., Dube, A., Kontusaari, S., Korhonen, J., Lahtinen, I., Oettgen, P. & Alitalo, K. (1999) Role of Ets fac tors in t he activity and endothelial cell specificity of the mouse T ie gene promoter. FASEB J. 13, 377–386. 38. Neish, A.S., Khachigian, L.M., Park, A., Baichwal, V.R. & Collins, T. (1995 ) Sp1 is a component of the cytokine-inducible enhancer in the promoter of vascular cell adhesion molecule-1. J. Biol. Chem. 270, 28903–28909. 39. Takashima,S.,Kurosawa,N.,Tachida,Y.,Inoue,M.&Tsuji,S. (2000) Comparative analysis of the genomic st ructures and promoter activities of mouse S iaalpha2,3Galbeta1,3GalNAc GalNAcalpha2,6-sialyltransferase genes (ST6GalNAc III and IV): characterization of their Sp1 binding sites. J. Biochem. (Tokyo) 127, 399–409. 40. Soejima,M.,Koda,Y.,Wang,B.&Kimura,H.(1999)Functional analysis of the 5¢-flanking region of FTA f or expression of rat GDP- L -fucose: beta- D -galactoside 2-alpha- L -fucosyltransferase. Eur. J. Bio che m. 266, 274–281. 41. Yip, B ., Chen, S.H., Mulder, H., Hoppener, J.W. & Schachter, H. (1997) Organ ization of the human beta-1,2-N-acetylgluco- saminyltransferase I gene (MGAT1), which cont rols complex and hybrid N-glycan synthesis. Biochem. J. 321, 465–474. 42. Rajput, B ., Shaper, N.L. & Shaper, J.H. ( 1996) Transcriptional regulation of murine b1,4-galactosyltransferase in somatic cells. J. Biol. Chem. 271, 5131–5142. 43. Harduin-Lepers, A., Shaper, J.H. & Shaper, N.L. (1993) Char- acterization of two cis-regulatory regions in the murine b1,4-galactosyltransferase gene. J. Biol. Chem. 268, 14348–14359. 44. Charron, M., Shaper, J.H. & Shaper, N.L. (1998) The inc reased level of beta1,4-galactosyltransferase required for lactose bio- synthesis is achieved in part by translational control. Proc. Natl Acad.Sci.USA95, 14805–14810. Ó FEBS 2002 Regulation of a1,3GalT expression in pig endothelial cells (Eur. J. Biochem. 269) 1473 . Regulation of a1,3galactosyltransferase expression in pig endothelial cells Implications for xenotransplantation Dominique Mercier 1,2 ,. 2002 Regulation of a1,3GalT expression in pig endothelial cells (Eur. J. Biochem. 269) 1467 sites that could be important for promoter activity in pig cells,