Uroporphyrinogen III Synthase

Uroporphyrinogen III synthase (URO-synthase, EC4.2.1.75) is the fourth enzyme of the heme biosynthetic pathway and is the defective enzyme in congenital erythropoietic porphyria. To investigate the erythroid-specific expression of murine URO-synthase, the cDNA and ∼24-kilobase genomic sequences were isolated and characterized. Three alternative transcripts were identified containing different 5′-untranslated regions (5′-UTRs), but identical coding exons 2B through 10. Transcripts with 5′-UTR exon 1A alone or fused to exon 1B were ubiquitously expressed (housekeeping), whereas transcripts with 5′-UTR exon 2A were only present in erythroid cells (erythroid-specific). Analysis of the TATA-less housekeeping promoter upstream of exon 1A revealed binding sites for ubiquitously expressed transcription factors Sp1, NF1, AP1, Oct1, and NRF2. The TATA-less erythroid-specific promoter upstream of exon 2A had nine putative GATA1 erythroid enhancer binding sites. Luciferase promoter/reporter constructs transfected into NIH 3T3 and mouse erythroleukemia cells indicated that the housekeeping promoter was active in both cell lines, while the erythroid promoter was active only in erythroid cells. Site-specific mutagenesis of the first GATA1 binding site markedly reduced luciferase activity in K562 cells (<5% of wild type). Thus, housekeeping and erythroid-specific transcripts are expressed from alternative promoters of a single mouse URO-synthase gene.

The genomic sequences encoding the eight mammalian (human and/or murine) heme biosynthetic enzymes have been isolated and characterized (1)(2)(3)(4)(5)(6)(7), with the notable exception of URO-synthase, 1 the fourth enzyme in the pathway (URO-syn-thase; hydroxymethylbilane hydrolyase (cyclizing), EC 4.2.1.75). The genes encoding the first three enzymes have erythroid-specific transcripts generated by a separate gene (5aminolevulinate synthase) (8) or by alternative promoters of a single gene (5-aminolevulinate dehydratase and hydroxymethylbilane synthase) (2,9). Their erythroid-specific promoters and enhancers all have GATA1, NF-E2, and CACCC elements that bind transcriptional factors to facilitate the high level expression of these transcripts in erythroid cells. Of the remaining five genes in the pathway, each has been shown to be a single gene with a single promoter containing both erythroidspecific binding elements for enhanced erythroid expression and housekeeping elements for constitutive expression in all cells (6, 10 -12), with the notable exception of the URO-synthase gene, which has not been isolated or characterized; nor has its erythroid-specific expression been investigated.
URO-synthase is responsible for the conversion of the linear tetrapyrrole, hydroxymethylbilane, to uroporphyrinogen III, the first cyclic tetrapyrrole and physiologic precursor of heme (13). The enzyme functions as both an isomerase and a cyclase as it catalyzes the intramolecular rearrangement of the pyrrole ring D and tetrapyrrole ring closure of the linear tetrapyrrole, respectively (14,15). In the absence of URO-synthase activity, hydroxymethylbilane is nonenzymatically cyclized to form the nonphysiologic uroporphyrinogen I isomer, which is then oxidized to uroporphyrin I, a nonmetabolizable and pathogenic compound. URO-synthase has been purified to homogeneity from human erythrocytes, rat liver, Euglena gracilis, and Escherichia coli (16 -20). The human enzyme has been shown to be a monomeric protein with an apparent molecular mass of 29.5 kDa. Mammalian cDNAs encoding human and mouse UROsynthase have been isolated and expressed in E. coli by our laboratory (21,22). Both the mouse and human URO-synthase cDNAs had open reading frames of 798 bp that encoded polypeptides of 265 amino acids. The murine cDNA and predicted amino acid sequences were 81 and 78% identical with the corresponding human sequences, respectively (22). In addition, URO-synthase cDNAs have been isolated from various bacteria, yeast, and rat (for a review, see Ref. 23), and their significant homologies provided further evidence for a single URO-synthase gene in prokaryotes as well as in the animal kingdom.
The markedly deficient, but not absent, activity of UROsynthase is the enzymatic defect in congenital erythropoietic porphyria (CEP), an inborn error of heme biosynthesis that occurs in humans and cattle (24 -26). The enzymatic defect results in the erythroid accumulation of uroporphyrinogen I, which leads to the clinical manifestations. In humans, the disease is clinically heterogeneous and ranges from nonimmune hydrops fetalis to a severe transfusion-dependent anemia of childhood to a milder disorder characterized primarily by light-induced cutaneous lesions (for a review, see Ref. 27). A variety of mutations that cause CEP have been identified in the human URO-synthase gene providing genotype/phenotype correlations for the marked clinical variability (e.g. see Refs. 28 -30). However, about 15% of the mutant alleles causing this erythropoietic porphyria have not been identified (27).
To date, no mammalian genomic sequence encoding UROsynthase has been reported. Chromosome mapping studies of human and murine URO-synthase localized the respective genes to a single locus at human chromosome 10q25.2 3 q26.3 (31) and to a single locus on murine chromosome 7, in a region of conserved synteny with human chromosome 10 (22). These findings were consistent with a single gene coding for UROsynthase in mice and humans. Thus, an investigation was undertaken to isolate and characterize the murine URO-synthase genomic sequence and to determine if mouse URO-synthase had erythroid-specific regulation. In this paper, we report that the mouse URO-synthase gene has distinct erythroid and housekeeping promoters that generate three different transcripts, one being erythroid-specific. In addition, the genomic organization, promoter sequences, and 5Ј transcription start sites of the alternative transcripts were determined, and erythroid-specific expression was demonstrated in MEL and human K562 cells.

EXPERIMENTAL PROCEDURES
Rapid Amplification of 5Ј cDNA Ends (5Ј-RACE)-5Ј-RACE was performed using Marathon-Ready mouse spleen cDNA (CLONTECH, Palo Alto, CA) according to the manufacturer's procedure. For the PCR, the sense primer was the Marathon nested adaptor primer 2 (AP2; CLONTECH, Palo Alto, CA) and the mouse URO-synthase antisense primer 17 (Table I; cDNA bp 503-529, numbered according to the full-length housekeeping URO-synthase cDNA sequence; GenBank TM accession no. U18867.2). The reaction conditions were as follows: 95°C for 12 min followed by 35 cycles of amplification (94°C for 30 s and 70°C for 2 min) and then a final extension step at 70°C for 7 min. The 5Ј-RACE products were then subcloned into the pGem-T vector (Promega, Madison, WI). Positive clones were sequenced with a fluorescent automated DNA sequencer (ABI Prism TM model 377-XL; Perkin-Elmer Applied Biosystems, Inc, Foster City, CA.) using dRhodamine dye terminator chemistry.
RT-PCR Analysis of Mouse URO-synthase mRNAs-Total RNAs from different mouse tissues were isolated using the "Ultraspec RNA Isolation System" (Biotecx Laboratories, Inc., Houston, TX). RT-PCR was performed in a final volume of 50 l using the Titan TM One Tube RT-PCR System (Roche Molecular Biochemicals) according to the manufacturer's protocol, and the amplification was carried out in a PTC-150 MiniCycler (MJ Research, Inc., Watertown, MA). RT-PCR was initiated at 50°C for 30 min and then 94°C for 2 min, followed by 10 cycles of denaturation at 94°C for 30 s, annealing at 65°C for 30 s, and extension at 68°C for 1 min and then 30 cycles of denaturation at 94°C for 30 s, annealing at 65°C for 30 s, and extension at 68°C for 1 min (with each elongation lengthened by an additional 5 s/cycle). The sequences of the sense primers for amplification of exon 1A (primer 2), 1B (primer 4), and 2A (primer 7) are indicated in Table I. Each primer set was designed to produce a PCR product that included at least one intronexon boundary, thereby eliminating the possibility that DNA contamination was responsible for the resulting products amplified from the isolated tissue mRNAs. The RT-PCR products (10 l) were subjected to electrophoresis in 1.5% agarose gels and stained with EtBr. The RT-PCR products were purified using the QIAquick PCR Purification Kit (Qiagen, Valencia, CA) and then sequenced using the antisense primer 13 (Table I) to ensure that they corresponded to the targeted UROsynthase RNAs.
Isolation and Characterization of the Mouse URO-synthase Gene-A bacteriophage P1 DNA library of genomic DNA isolated from mouse strain 129/OLA ES cells (Genome Systems, St. Louis, MO) was screened with oligonucleotide primers 4 and 5 (Table I), which amplified an 88-bp fragment of exon 1A/1B, and primers 10 and 13, which amplified ϳ600 bp between exons 2 and 3. The conditions for PCR were as follows: 30 cycles of denaturation at 94°C for 30 s, annealing at 58°C for 30 s, and extension at 68°C for 1 min. Three positive clones containing exon 1 and/or exons 2 and 3 were identified and purified. To determine the coding region intron/exon boundaries, the genomic clones were amplified and sequenced using primers corresponding to the mouse UROsynthase cDNA as indicated in Table I. The intron sizes were estimated from the PCR products.
Sequencing and Analysis of the Housekeeping and Erythroid-specific Promoters-The mouse URO-synthase genomic P1 clone (mgUROS-1) was directly sequenced using dRhodamine dye terminator chemistry in an automated fluorescent DNA sequencer as described above. To characterize the housekeeping and erythroid promoter regions, the sequences upstream of exons 1A and 2A were initially sequenced using the respective exonic primers (5 and 9, Table I), and then subsequent sequencing reactions were carried out using oligonucleotide primers based on the most clearly readable 5Ј sequence obtained. All sequences were confirmed in both orientations.
The URO-synthase housekeeping and erythroid-specific 5Ј-flanking promoter/enhancer regions were analyzed using the MatInspector program (32) with the Transfac 3.4 data base of vertebrate transcription factor consensus matrices (available on the World Wide Web) with core similarity of 1.0 and matrix similarity of 0.85.
Construction of the Promoter/Reporter Plasmids-To construct the housekeeping and erythroid promoter/reporter plasmids, the regions upstream of exon 1B and upstream of exon 2B were individually amplified and subcloned into the pGL3-Firefly Luciferase Vector (Promega). For amplification of the housekeeping promoter, sense primer 1 (corresponding to mouse genomic sequence nt Ϫ1772 to Ϫ1753; Fig. 5) and antisense primer 3 (corresponding to mouse genomic sequence nt 24 -43; Fig. 5) were employed. To amplify the erythroid promoter, sense primer 6 (corresponding to mouse genomic nt Ϫ1534 to Ϫ1516; Fig. 6) and antisense primer 8 (corresponding to mouse genomic nt 59 -77; Fig.  6) were used. For cloning purposes, 5Ј tails containing an SpeI or SalI restriction site were included in the sense and antisense primers, respectively. After amplification, the PCR products were double digested with SalI and SpeI and cloned into the XhoI-NheI sites of the pGL3-Luciferase Vector (Promega). The plasmids were purified using the Qiagen Plasmid Maxi Kit (Qiagen, Valencia, CA) and sequenced to confirm their authenticity. The housekeeping and erythroid promoter/ reporter constructs were designated mHPr and mEPr, respectively.
To test the functionality of the putative erythroid-specific promoter region, a point mutation at T Ϫ65 was introduced into the first putative GATA1 site (5Ј-TTATCA-3Ј 3 5Ј-TTACCA-3Ј) in the erythroid promoter/reporter construct mEPr using the QuickChange TM site-directed mutagenesis kit (Stratagene, La Jolla, CA), according to the manufacturer's instructions. Amplification was performed with sense primer 30 and antisense primer 31 (both corresponding to mouse erythroid promoter coordinates nt Ϫ79 to Ϫ51; Fig. 6). The mutant construct was designated mEPr-65C. Sequence analysis confirmed the authenticity of the mutant construct.
Cell Culture, Transfections, and Luciferase Assays-Mouse MEL cells (clone 745; a gift from Dr. George Atweh, Mount Sinai School of Medicine, New York) and human erythroleukemia K562 cells were grown in RPMI 1640 medium. Mouse fibroblast NIH 3T3 cells were grown in Dulbecco's modified Eagle's medium. Both media were supplemented with 10% fetal bovine serum, 2 mM L-glutamine, 100 units/ml penicillin, and 100 mg/ml streptomycin (Life Technologies, Inc., Gaithersburg, MD).
All transfections were performed in 12-well plates. For NIH 3T3 cell transfections, 1 ϫ 10 5 cells were plated 10 -16 h prior to transfection. The respective reporter construct (2 g) and 50 ng of the internal control, Renilla Luciferase-TK (pRL-TK) vector were transfected using 3.2 l of Lipofectamine and 5 l of Plus Reagent (Life Technologies). After transfection (24 h), the cells were washed twice with phosphatebuffered saline, harvested, and lysed with 100 l of Passive Lysis Buffer (Promega Corp.). For MEL and K562 cells, 8 ϫ 10 5 cells were transfected as above with 2 g of the reporter construct and 50 ng of the pRL-TK vector, using 4 l of DMRIE-C reagent (Life Technologies). After transfection (48 h), the cells were collected by centrifugation, washed with phosphate-buffered saline, and lysed with 100 l of Passive Lysis Buffer. When the MEL cells were induced to differentiate, 2% Me 2 SO was added to the medium 48 h prior to transfection. To induce the K562 cells, 25 M hemin was added to the medium 24 h prior to transfection. The Renilla and firefly luciferase activities were determined in 20 l of the cell lysates using the Dual-Luciferase TM Reporter Assay System (Promega) according to the manufacturer's procedure, with a Microtiter® Plate Luminometer (DYNEX Technologies, Inc., Chantilly, VA). The firefly luciferase activities were normalized for transfection efficiency using the Renilla luciferase activity as an internal control, and the data were expressed as relative light units. A negative control using the pGL-3 basic vector was included in all experiments. Data represent the means Ϯ S.D. of at least five independent transfection experiments.

5Ј-RACE of Mouse Splenic cDNA Demonstrates Multiple
URO-synthase 5Ј-UTRs-To assess the possible existence of separate housekeeping and erythroid-specific transcripts for the mouse URO-synthase, 5Ј-RACE was performed using Marathon-Ready cDNA made from mouse splenic mRNA (Fig.  1A). The 5Ј-RACE PCR product, a single broad band of ϳ450 bp, was subcloned into the pGem-T vector, and the 5Ј sequences of the subclones were determined. Two groups of clones (designated group 1A and 2A) were identified that had different 5Ј-UTRs followed by the same 3Ј sequence, designated 2B (exon 2B in Fig. 2). The lengths of the unique 5Ј-UTRs varied from 22 to 40 bp for all seven group 1A clones and from 5 to 41 bp for 17 representative group 2A clones (Fig. 1B). Of the over 200 5Ј-RACE clones identified, there were almost 30 times more group 2A than group 1A clones. Both group 1A and 2A clones had 5Ј-UTRs that differed from 5Ј-UTR (designated group 1B) of the previously reported mouse URO-synthase cDNA (pmUROS-1) (22), while the 3Ј sequence adjacent to the different 5Ј-UTRs was common in all three groups (GenBank TM accession no. U18867; pmUROS-1 sequence, bp 68 -450) and had the same translation start site. Since none of the group 1A or 2B transcripts found in spleen contained the group 1B 5Ј-UTR, efforts were directed to determine the genomic structure of mouse URO-synthase and to localize these exonic 5Ј-UTRs.
The Mouse URO-synthase Gene Has 10 Exons and Three Alternative 5Ј-UTRs-Screening of a mouse genomic P1 library produced three clones positive for URO-synthase primers am-plifying both the 5Ј-UTR 1B region and coding exons 2 and 3. Restriction analysis of the three P1 clones revealed similar band patterns on agarose gel electrophoresis, indicating that the three clones contained common regions of genomic DNA. Therefore, the genomic organization of the clone with the largest insert, mgUROS-1, was determined. Amplification and sequencing of the mgUROS-1 P1 DNA using primers designed from the 5Ј-UTR sequences 1A, 1B, and 2A and the first region of common sequence (designated 2B) in the 1A, 1B, and 2Acontaining transcripts facilitated determination of the 5Ј region of the gene ( Fig. 2A). As shown in Fig. 2, the 5Ј-UTR sequence previously reported for the mouse liver URO-synthase cDNA (22) was 19 bp downstream from the group 1A 5Ј-UTR sequences. The group 2A 5Ј-UTR sequences were located ϳ6 kb downstream of the group 1A 5Ј-UTR gene sequences and were directly upstream of the common exonic region containing the translation start site, which was designated exon 2B. Therefore, the group 1A 5Ј-UTR sequence was designated exon 1A and the directly adjacent downstream 1B 5Ј-UTR sequence was designated exon 1B. Analysis of the longest URO-synthase cDNA clone (GenBank TM accession no. U16216) revealed that it contained the entire 85-bp exon 1B, 21 bp of exon 1A, and 41 bp at the 5Ј-end, which was an artifact derived from the murine mitochondrial 16 S rRNA. As shown in Fig. 2B, the mitochondrial 16 S rRNA sequence shared a 7-bp overlapping region with the 5Ј-end of the URO-synthase cDNA, which contains a BamHI restriction site. Most likely, this artifactual sequence resulted from concatamerization of the Sau3A1 partial digests used to construct the library. Computer-assisted searches of GenBank TM revealed additional examples of mouse URO-synthase cDNAs including alternatively spliced sequences for exons 1A and 1B as well as a fetal mouse EST containing the exon 2A extension of exon 2B (Fig. 2B).
Amplification and sequencing of the exonic regions in the mouse URO-synthase gene using primers based on the cDNA sequence (Table I) demonstrated that the P1 clone contained the entire ϳ24-kb URO-synthase gene with 10 exons. The exon/intron boundaries were sequenced, and the sizes of the introns were estimated from PCR products (Table II). The introns ranged in length from approximately 0.5 to 7.5 kb, while the exons ranged from 75 to 761 bp, and their sequences matched that of the previously reported cDNA sequence (22). The sequences at all of the intron-exon boundaries conformed to the GT-AG rule (33).
Tissue-specific Expression of the Alternative URO-synthase Transcripts-To study expression of the URO-synthase alternative transcripts in different mouse tissues, RT-PCR was performed with total RNA from whole fetus (embryonic day 10 -12) and adult brain, kidney, liver, and spleen using sense primers designed to bind specifically to exon 1A (primer 2), the alternative splice junction region of exon 1A and 1B (primer 4), or exon 2A (primer 7; Fig. 3A). Amplification with exon 1Aspecific primer 2 generated two products, which were sequenced (Fig. 3A). The smaller RT-PCR product was 484 bp and contained exon 1A spliced to exon 2B, while the longer product was 569 bp and had exons 1A and 1B (designated exon 1A/1B) fused to exon 2B. RT-PCR products with alternatively spliced exon 1A or exon 1A/1B were detected in all tissues examined, but the relative amount of the exon 1A/1B product was reduced in brain and fetal tissue (Fig. 3C). Primer 4 yielded a 528-bp RT-PCR product, which contained exon 1A/1B and was present in all tissues studied but again was reduced in brain and fetal tissue (Fig. 3B), indicating that the exon 1A/ 1B-containing transcripts were ubiquitously expressed, presumably from a housekeeping promoter.
In contrast, a markedly different expression pattern was observed for exon 2A-specific transcripts. A single 660-bp RT-PCR product (Fig. 3D) was detected only in spleen, whole fetus, and G1/ER cell RNA, the latter RNA from a murine erythroid proerythroblast cell line that was induced by GATA1 to un-

5Ј-GCCCTCTGTCACTGGTAAGGCTGAGGGAG-3Ј
a Coordinates are from the full-length housekeeping cDNA (GenBank™ U18867.2) which contains the exon 1A sequence (Fig. 5). b Underlined nucleotides are not part of the URO-synthase sequence and were added at the 5Ј-end of those primers for cloning purposes. c Coordinates in the URO-synthase housekeeping promoter sequence were Ϫ1753 to Ϫ1772 (Fig. 5). d Coordinates for primers 6 and 7 in the URO-synthase erythroid promoter sequence were Ϫ1515 to Ϫ1534 and 10 to 27, respectively (Fig. 6), while coordinates for primers 30 and 31 were Ϫ51 to Ϫ79 (Fig. 6).  rare transcripts, supporting the finding that the alternative exon 2A-containing transcript was expressed only in erythroid cells. Thus, the URO-synthase gene contains alternatively spliced exons (1A, 1A/1B, and 2A), which encode different 5Ј-UTRs that are expressed in the housekeeping or erythroidspecific transcripts (Fig. 4). Sequence Analysis of the Housekeeping Promoter/Enhancer Region-To characterize the "housekeeping" promoter/enhancer region, a 1772-bp region immediately upstream of exon 1A/1B was sequenced from the URO-synthase P1 genomic clone ( Fig. 5; GenBank TM accession no. AF133257). The proximal promoter region lacked a TATA box; however, there was a multiplicity of transcription start sites as indicated by the multiple 5Ј-RACE ends and cDNA clones (Figs. 1B, 2B, and 5). According to convention (e.g. see Ref. 35), the cap site (position ϩ1) was designated the most 5Ј base in the longest murine URO-synthase cDNA clone (GenBank TM accession no. AA980856), which was only 21 bp longer than the longest 5Ј-RACE product (Fig. 5). The full-length URO-synthase cDNA containing the exon 1A/1B 5Ј-UTR is available (GenBank TM accession no. U18867.2).
Sequence Analysis of the Erythroid-specific Promoter/Enhancer Region-The transcription initiation site of the erythroid-specific URO-synthase transcript was based on the 5Ј-end of the longest cloned cDNA, which was consistent with the sites found by 5Ј-RACE analysis. Among the erythroid-specific clones, the clone with the longest exon 2A (31 bp) was from a mouse fetal liver library (GenBank TM accession no. AA272198). 5Ј-RACE analysis revealed 10 different exon 2A initiation sites, ranging in size from 5 to 41 bp, with a median length of 16 bp.
Computer-assisted analysis of the 1534-bp region upstream of the exon 2A cap site revealed a series of erythroid-specific transcription factor binding sites, no consensus TATA box, and three Inr initiation sequences ( Fig. 6; GenBank TM no. AF133258). The Inr sequence is a powerful promoter core initiation element that is functionally analogous to the TATA box (36,37). Transcription initiation typically occurs within the YYAN(T/A)YY consensus sequence for Inr, most often at the first A (38). A consensus Inr motif (TCACTCT) was located at Ϫ3 to ϩ4 in the erythroid promoter region, while two additional tandem Inr motifs were in exon 2 (Inr2 (nt 7-13) and Inr3 (nt 13-19)), each with a single mismatch to the consensus sequence. Of interest, the most frequent 5Ј-RACE ends were within the Inr2 and Inr3 regions. The assignment of the CAP site was consistent with the following findings: 1) the first nucleotide of the longest cloned cDNA; 2) the predicted transcription start site for the observed Inr element was only 1 bp longer than the URO-synthase clone from the mouse fetal liver library (GenBank TM accession no. AA272198); and 3) all but one RACE end were shorter than the assigned CAP site. In the proximal promoter region, perfect matches to the GATA1 consensus elements were at nt Ϫ63, Ϫ200, and Ϫ383. More distally, one NF-E2, one CACCC, and six GATA1 elements were present between Ϫ280 and Ϫ1535 bp.
Functional Analysis of the Housekeeping and Erythroid-specific Promoters-Reporter gene constructs were designed to evaluate the function of the putative housekeeping and erythroid-specific promoters in cultured nonerythroid NIH 3T3 cells and erythroid MEL cells. The housekeeping promoter construct, mHPr, had 1772 bp of the genomic DNA upstream of exon 1A/1B as well as the first 21 bp of exon 1A/1B (Fig. 5), fused in front of the firefly luciferase cDNA in the pGL3 vector. The erythroid-specific promoter, mEPr, construct contained 1534 bp upstream of exon 2A as well as the entire exon 2A sequence and the first 46 bp of exon 2B ( Fig. 6; all noncoding) in front of the firefly luciferase cDNA in the pGL3 vector.
NIH 3T3 and MEL cells were co-transfected individually with each of the luciferase expression constructs and with an internal control pRL-TK vector containing the HSV-TK promoter upstream of the Renilla luciferase cDNA. Twenty-four h after transfection, the housekeeping promoter was active in both cell lines, while the erythroid promoter was detected only in uninduced and induced MEL cells, as shown in Fig. 7. In NIH 3T3 cells, the activity of the housekeeping promoter was over 80 times higher than that detected for the erythroid promoter. It should be noted that the luminescence measured for the erythroid construct in the NIH 3T3 cells was only 2 times higher than the background level of the pGL3 vector alone (data not shown). In contrast, in uninduced MEL cells, the level of housekeeping promoter expression was similar to that of the erythroid promoter. When the cells were induced with Me 2 SO to undergo erythroid differentiation, both activities increased; however, the erythroid promoter activity increased over 4 times higher than that of the uninduced cells, while the housekeeping promoter activity was increased only 1.7 times. It is important to note that a markedly lower transfection efficiency was achieved for MEL cells (0.7%) compared with that (13%) for the NIH 3T3 cells as determined by transfections with a green fluorescence protein vector (data not shown). This fact partially explains the lower luminescence values obtained when MEL cells were transfected with the promoter constructs.
To demonstrate the functionality of the erythroid promoter, a new construct, mEPr-65C, was prepared by introducing a T Ϫ65 3 C point mutation in the first putative GATA1 binding site in the erythroid promoter/reporter construct. The mEPr and mEPr-65C constructs were expressed in human K562 cells, and their luciferase activities were determined. Compared with the high level of luciferase activity expressed by the mEPr construct, the mutant EPr-65C construct expressed only 4.3 and 2.8% of wild type activity in uninduced and hemin-induced K562 cells, respectively (Table III). DISCUSSION Heme is the essential pigment of life and is the prosthetic group in hemoglobin, myoglobin, the cytochromes, and many other hemoproteins. In mammals, approximately 85% of heme synthesis occurs in erythroid cells, indicating the importance of the erythroid-specific production. The deficient activity of the individual enzymes in the heme biosynthetic pathway results in a group of disorders known as the porphyrias, which have been classified as hepatic or erythroid based on their clinical manifestations (39). Previous studies in this (2, 3, 40) and other laboratories (9,41) demonstrated that the first three enzymes of the heme biosynthetic pathway have unique erythroid-specific and housekeeping transcripts encoded by a single gene or by different genes. However, recent evidence suggested that each of the remaining five genes in the pathway had a single transcript under the control of a single promoter that contained the canonical binding sites for erythroid expression including GATA1, the CACCC-binding protein, and, usually, NE-F2 (6, 10 -12, 31). Of these genes, only URO-synthase had not been isolated; nor had its erythroid-specific regulation been investigated.
In this paper, we report that the murine URO-synthase gene has distinct housekeeping and erythroid-specific promoters, the latter controlling its erythroid-specific expression. The existence of the mouse URO-synthase erythroid-specific promoter was suggested by 5Ј-RACE experiments with splenic total RNA, which revealed two 5Ј-UTRs (1A and 2A) that differed from the previously reported mouse URO-synthase cDNA sequence (GenBank TM accession no. U16216; Ref. 22). A computer-assisted BLAST search of the mouse EST data base with the three 5Ј sequences identified an exon 1A-containing URO-synthase sequence from mouse mammary gland (GenBank TM accession no. AA980856) and an exon 2A URO-synthase cDNA from mouse fetal tissue (GenBank TM accession no. AA272198). The presence of an exon 1A transcript from mammary gland and an exon 2A transcript from fetal tissue supported the notion of housekeeping and erythroid-specific transcripts encoded by a single URO-synthase gene with alternative 5Ј-UTRs and promoters.
Characterization of the mouse URO-synthase gene from a P1 library revealed an approximately 24-kb sequence containing 10 exons. The exon 1A sequence or the entire exon 1 (1A/1B) was ubiquitously expressed in all tissues studied, and one or the other was alternatively spliced directly to the coding exons, beginning with exon 2B. In contrast, exon 2A was expressed only in erythroid tissues with transcription initiated from a 31-bp 5Ј extension of exon 2B, such that the entire erythroid message encoded the same polypeptide as the housekeeping transcript. Thus, the expression of the mouse URO-synthase gene, with unique transcripts resulting from housekeeping and erythroid-specific promoters, was similar to the expression of the human and mouse ⌬-aminolevulinate dehydratase (2, 42) and hydroxymethylbilane synthase genes (9, 41, 43). Two pro- moters generate transcripts in all three genes that differ in their 5Ј-UTRs; their transcripts encode the identical or essentially identical polypeptides, the housekeeping promoters are located upstream of the first exon, and the erythroid-specific promoters are in intron 1, immediately upstream from exon 2 in each gene. Unique to the URO-synthase gene, exon 1 undergoes alternative splicing with either exon 1A or the entire exon 1 (1A/1B), as shown in Figs. 2A and 4. The functional basis of the two housekeeping transcripts is not known.
The functional analysis of the erythroid and housekeeping promoter sequences revealed that the erythroid promoter was active only in erythroid cells, indicating its erythroid specificity (Fig. 7), while the housekeeping promoter was active in erythroid (MEL) and nonerythroid (NIH 3T3) cells. Me 2 SO-induced erythroid differentiation of the MEL cells increased the activity of both promoters (Fig. 7B). Nevertheless, the erythroid promoter was more responsive to erythroid differentiation, as the ratio of erythroid to housekeeping activity increased from 0.6 to 1.6 (Fig. 7B). Moreover, the murine erythroid promoter/reporter construct was highly active in uninduced and induced human K562 cells (Table III). Of particular relevance, sitespecific mutagenesis of the first GATA1 binding element in the murine erythroid promoter markedly reduced promoter/reporter activity in both uninduced and induced K562 cells. The fact that the wild type construct was highly expressed in these human erythroleukemia cells and the fact that the GATA1 site point mutation essentially eliminated promoter/reporter construct activity indicated that the function of the murine erythroid promoter was highly conserved and that the first GATA1 site was functional in both murine and human erythroid cells.
Transcriptional activities of the housekeeping and erythroid promoters during erythroid differentiation also were similar to each other for the 5-aminolevulinate dehydratase and the hydroxymethylbilane synthase genes (43,44). Of note, however, the ratio of the URO-synthase erythroid to housekeeping 5Ј-UTRs in the mouse spleen was 31:1 in the RACE experiments. Since only ϳ1.7and ϳ1.6-kb promoter sequences for the housekeeping and erythroid promoter, respectively, were analyzed, and since the in vitro expression of a reporter gene does not necessarily reproduce the in vivo situation, it is likely that other regulatory sequences in the URO-synthase gene regulate the activity of these promoters by either enhancing the activity of the erythroid promoter or inhibiting the activity of the housekeeping promoter. For example, while transcription initiation from the murine hydroxymethylbilane synthase housekeeping promoter was high in MEL cells, it was blocked from elongation (43,45).
The transcription factor consensus binding sites in the erythroid URO-synthase promoter region (Fig. 6) that are significant with respect to erythroid-specific control of transcription include vMYB (nt Ϫ27), GATA1 (nt Ϫ63, Ϫ200, and Ϫ383), E47 (nt Ϫ146 and Ϫ160), a CCAAT box (nt Ϫ324), and an NF-E2like site (nt Ϫ400). GATA1 is essential for erythroid development. In chimeric embryos or mice with a GATA1 knockout, erythroid differentiation fails to proceed beyond the proerythroblast stage (46). E47 has been shown to be a DNA binding partner of GATA1 and Tal1 in the formation of a transcriptional transactivating complex in erythroid cells (47) and may play a supporting role in the erythroid-specific control of UROsynthase expression. CCAAT is an enhancer element bound by a family of ubiquitous and tissue-specific factors (e.g. NF-Y (48)), which enhance the transcription of many eukaryotic genes. It occurs more frequently in TATA-less promoters (49) and presumably participates in the activation of erythroid URO-synthase transcription. The coincidence of the multiple Inr sites, also found in other TATA-less promoters (50), and the clusters of 5Ј-RACE ends suggest that these are the functional cap sites for the URO-synthase erythroid transcripts. MYB is a protein target for binding by the murine c-Maf protein, and this complex has been shown to inhibit gene expression in myeloid cells (51). Interestingly, c-Myb knockout mice die in utero from a severe anemia due to an inability to switch from yolk sac to

Effect of a GATA1 mutation (T Ϫ65 3 C) on erythroid promoter activity
The activities represent the mean values of three independent experiments and are expressed as relative light units (RLU) and percentage of wild type activity. In each experiment, the luciferase activity was corrected for transfection efficiency and background luminescence based on the pRL-TK and pGL3 basic vector activities, respectively (for details, see "Experimental Procedures"). fetal liver erythropoiesis (52). Finally, NF-E2 is important in the chromatin remodeling process required to prepare a promoter region for transcription. The NF-E2 element in the ␤-globin locus control region was shown to direct chromatin remodeling in the promoter region of the epsilon-globin gene in a functional minichromosome, activating it for erythroid-specific transcription (53). However, NF-E2 knockout mice were only mildly anemic and hypochromic but had no platelets and died of a bleeding diathesis (54), indicating the primary role of this transcription factor in megakaryocyte maturation. Functional analyses with promoter-reporter constructs in various cell types coupled with footprint and EMSA analyses would facilitate the delineation of transcription factor binding site occupancy for these putative binding sites and for those discussed below for the housekeeping promoter. Many of the transcription factor binding sites in the UROsynthase housekeeping promoter (Fig. 5) are used by ubiquitously expressed transcription factors (Sp1, NF1, AP1, Oct1, and NRF2), which would provide appropriate basal expression of the housekeeping URO-synthase transcripts in a variety of cell types. For example, NRF2 (nuclear respiratory factor-2) is a ubiquitously expressed activator of various nuclear encoded proteins involved in cellular respiration such as rat cytochrome C (55) and in the housekeeping 5-aminolevulinate synthase promoter (56). In addition, the URO-synthase housekeeping promoter had several binding sites for factors that might modulate expression for specific tissue needs, including Mzf1, a myeloid-specific transcription factor (57), and Ik-2, a member of a family of zinc finger transcription factors that regulate lymphocyte differentiation (58). Interestingly, there is an Sp1 site immediately 5Ј of the Ets site at Ϫ253 to Ϫ258 (Fig. 5), and this particular arrangement and orientation of the two elements has been associated with transcriptional regulation complexes in viral and cellular genes (59,60). The ability of Ets1 to form ternary complexes with various factors permitting gene activation or repression (e.g. see Ref. 61) and the multiple Ets factors present in the housekeeping URO-synthase gene suggest the potential to fine tune expression in different cellular environments.
This identification of the erythroid-specific regulation of URO-synthase expression has important implications for the human porphyrias. The use of an alternative erythroid-specific promoter is likely in the human URO-synthase gene, since the gene organization and regulation of the other murine and human heme biosynthetic genes are highly homologous. For example, the first and rate-limiting enzyme, 5-aminolevulinate synthase, is encoded by erythroid-specific and housekeeping transcripts of separate genes (62,63), the erythroid-specific gene being localized to the X chromosome in mice and humans and the housekeeping gene on chromosome 3p21 in humans but unmapped in mice (64 -66). Murine and human 5-aminolevulinate dehydratase and hydroxymethylbilane synthase (also known as porphobilinogen deaminase), the second and third enzymes in the pathway, have alternative promoters in both genes in both species controlling the synthesis of erythroidspecific and housekeeping transcripts (2,9,(41)(42)(43)(44). Similarly, both coproporphyrinogen oxidase and ferrochelatase have been shown to have a single gene and promoter in both species (5,12,67,68). Thus, it is likely that the human URO-synthase gene will be regulated by an alternative erythroid-specific promoter and that studies of its differential expression in erythroid and nonerythroid tissues may clarify why the defective enzyme in the human disorder produces primarily an erythroid porphyria as opposed to a hepatic or erythrohepatic porphyria. Furthermore, this mode of regulation may provide insight into why 15% of the CEP patients do not exhibit mutations in their coding sequences. Indeed, we recently identified a CEP patient with severe disease who was heteroallelic for a mutation (T Ϫ70 3 C) in the first GATA1 binding site in the human UROsynthase erythroid promoter and a coding region mutation (69).
In summary, the mouse URO-synthase gene has been cloned and characterized. Transcription generates three alternative mRNAs differing in their 5Ј-UTRs; two are expressed by a housekeeping promoter, while the third is expressed by an erythroid-specific promoter, which contains erythroid-specific transcription factor binding elements. Functional assays with promoter-reporter constructs demonstrated that the housekeeping promoter expressed ubiquitous transcripts, whereas the erythroid promoter expressed transcripts only in erythroid cells. Thus, these studies show that the expression of UROsynthase is differentially regulated by transcriptional activation in erythroid cells and potentially provide insight into the molecular pathogenesis of CEP.