Aspartyl β-Hydroxylase (Asph) and an Evolutionarily Conserved Isoform of Asph Missing the Catalytic Domain Share Exons with Junctin*

The mouse aspartyl β-hydroxylase gene (Asph, BAH) has been cloned and characterized. The mouseBAH gene spans 200 kilobase pairs of genomic DNA and contains 24 exons. Of three major BAH-related transcripts, the two largest (6,629 and 4,419 base pairs) encode full-length protein and differ only in the use of alternative polyadenylation signals. The smallest BAH-related transcript (2,789 base pairs) uses an alternative 3′ terminal exon, resulting in a protein lacking a catalytic domain. Evolutionary conservation of this noncatalytic isoform of BAH (humbug) is demonstrated in mouse, man, andDrosophila. Monoclonal antibody reagents were generated, epitope-mapped, and used to definitively correlate RNA bands on Northern blots with protein species on Western blots. The gene for mouse junctin, a calsequestrin-binding protein, was cloned and characterized and shown to be encoded from the same locus. When expressed in heart tissue, BAH/humbug preferably use the first exon and often the fourth exon of junctin while preserving the reading frame. Thus, three individual genes share common exons and open reading frames and use separate promoters to achieve differential expression, splicing, and function in a variety of tissues. This unusual form of exon sharing suggests that the functions of junctin, BAH, and humbug may be linked.

In biological systems, protein diversity is enhanced by coand/or post-translational protein processing. Examples include phosphorylation, methylation, glycosylation, limited proteolysis, acetylation, and prenylation. In contrast to the growing understanding of the functional and regulatory roles these modifications play, a functional role for aspartyl ␤-hydroxylation of proteins has not been defined (1)(2)(3). The aspartyl ␤-hydroxylase (BAH) 1 hydroxylation consensus sequence is con-tained within calcium-binding epithelial growth factor domains that are found in proteins of diverse function. Consensus sequence domains contain the amino acids Asp, Asp/Asn, Asp/ Asn, and Tyr/Phe at defined positions. The alignment of these latter four residues are thought to signal post-translational hydroxylation of the third site in the consensus by BAH (4). The consensus sequence for aspartyl ␤-hydroxylation has been identified in a diverse group of proteins including clotting factors (5), Notch receptors and ligands (6 -8), in structural proteins of the extracellular matrix (9), and in ligands of the tyro-3/Axl family of receptor tyrosine kinases (10).
Two catalytically active forms of BAH (52-and 56-kDa polypeptides differing by a 22-amino acid amino-terminal extension) were purified from bovine liver microsomes (11). Because of proteolysis during isolation, the putative NH 2 terminus of bovine BAH was not identified until an apparent fulllength cDNA for bovine BAH was cloned (11,12). Based on the cDNA sequence and biochemical analyses, BAH can be divided into four distinct regions (12)(13)(14). BAH appears to be a type 2 integral membrane protein containing a short amino-terminal domain that projects into the cytoplasm. The amino-terminal domain is followed by a predicted transmembrane domain and a highly charged region that projects into the lumen of the endoplasmic reticulum. The COOH-terminal region of BAH, which corresponds to the 52-and 56-kDa polypeptides isolated from bovine liver, contains the aspartyl hydroxylase catalytic domain. This domain contains dibasic glycine and His 2 motifs that have been shown to be critical for catalytic activity (13,15).
Northern blotting of bovine liver mRNA with a 5Ј bovine BAH cDNA probe revealed two mRNA species of ϳ2.6 and 6.6 kb (12). More recently, a 4.3-kb cDNA isolated by immunoscreening against highly expressed antigens in a human osteosarcoma cDNA expression library led to observations of two transcripts (2.6 and 4.3 kb) in a Northern analysis of eight human tissues (14). At the time, the identity of the small (ϳ2.6-kb) transcript was unknown and was thought to represent an alternative splicing event of the BAH gene (14). The larger transcript encodes a 757-amino acid protein with a predicted mass of 85 kDa. Interestingly, in vitro translation of the full-length cDNA yielded a band on Western blots migrating at 107 kDa (14). Western analysis of human BAH from a variety of human cancer cell lines identified protein bands with masses ranging from 50 to 120 kDa (16), suggesting that smaller forms * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF289199, AF289200, AF289205-AF289215, and AF289486 -AF289494.
In order to better understand the organization and regulation of BAH, we have cloned and characterized the BAH genomic locus. Three major RNA forms of BAH-related sequences were cloned and analyzed. In addition to the fulllength murine BAH cDNA, a novel, alternatively spliced, transcript was obtained. This transcript (humbug) encodes a protein identical to BAH through the NH 2 -terminal half of the protein, but completely lacks the catalytic domain of BAH. The genomic organization of BAH is described, and the transcription start site is mapped. Monoclonal antibody reagents that allow specific recognition of the catalytically active version of human BAH were produced, characterized, and used to definitively assign bands observed on Northern blots to those observed on Western blots from human tumor cells.
A short region of homology between canine junctin, a 28-kDa calsequestrin-binding protein found in skeletal and cardiac muscle, and bovine BAH has been previously described (17). To understand the relationship between BAH, humbug, and junctin, a mouse junctin cDNA was cloned from a mouse heart library and the sequence compared with that for BAH and humbug. Mouse BAH, humbug, and junctin were found to be identical in sequence over the 220 nucleotides spanning BAH exons 2 and 3. Further genomic analysis revealed three additional junctin exons contained within the BAH genomic locus. Analysis of mouse heart RNA definitively demonstrated that BAH/humbug is expressed as a chimera that results in fusion of the amino-terminal half of junctin to the lumenal domain of BAH/humbug. A possible function of this unusual chimera is discussed.

EXPERIMENTAL PROCEDURES
cDNA Cloning and Characterization-A 580-bp fragment representing a partial murine BAH cDNA was obtained by RT-PCR using murine liver cDNA and primers designed from bovine BAH (12). The resulting clone was used to screen a murine liver cDNA library (18). The 5Ј end of mouse BAH was amplified from a liver cDNA library using PCR primers designed from murine BAH and the library vector. 5Ј-and 3Ј-RACE were performed using commercially available systems as described by the manufacturers (LifeTechnologies, Inc.; CLONTECH, Palo Alto, CA). Sequences were analyzed using Sequencher (Gene Codes Corp., Ann Arbor, MI) and the GCG sequence analysis package (Genetics Computer Group, Inc., Madison, WI).
Cloning Drosophila Aspartyl (Asparaginyl) ␤-Hydroxylase-Drosophila ␤-aspartyl hydroxylase sequences were obtained by EST data base searching and degenerate PCR using cDNA from embryo. Primers were designed from the human, bovine, and Caenorhabditis elegans sequences; optimum PCR amplification was seen with a degenerate sense primer (CARAGRTCNCTNTAYAA) and an antisense primer designed using Drosophila codon frequencies (CTCGTGCTCGAAGGAAT-CATC). A full-length cDNA was assembled by 5Ј-and 3Ј-RACE utilizing primers designed from the degenerate PCR products and ESTs. Splice variants were assessed by RT-PCR.
Genomic Mapping of Murine BAH Gene-A 129/SvJ ES cell BAC library was screened by both hybridization and PCR (InCyte Genomics, St. Louis, MO). cDNA sequences encoding the catalytic domain were used as a hybridization probe. PCR screening relied on primers from the 5Ј-UTR of the mouse cDNA. Three overlapping BAC clones containing the entire murine BAH gene were shotgun-subcloned essentially as described (19) using nebulized BAC DNA. Shotgun clones containing coding regions were identified by hybridization using radiolabeled fulllength cDNA. Vector primers were used to sequence coding region clones from both directions, with the sequence being assembled and analyzed using Sequencher (Gene Codes). In addition, specific restriction fragments from the BACs were subcloned and sequenced using the Primer Island transposition kit (PerkinElmer Life Sciences).
Cloning and Genomic Mapping Mouse junctin-Murine junctin sequences were obtained from heart cDNA by RT-PCR using a sense primer designed from the 5Ј end of the dog junctin open reading frame (17) and an antisense primer containing the junctin-like sequences from murine BAH. Primers were designed from this initial RT-PCR product for use in RACE, with the full-length sequence for murine junctin assembled using overlapping 5Ј-and 3Ј-RACE products. Southern analysis was used to map the mouse junctin sequences on the murine BACs containing the BAH locus. Appropriate fragments were subcloned and sequenced as described above.
Northern Blotting-Northern blots containing total RNA and mRNA were hybridized with radiolabeled cDNA probes following standard protocols (18).
Real Time PCR-Real time PCR was performed essentially as described (20,21). Primers and probes designed from the mouse BAH and junctin sequences were synthesized and purified by Biosearch Technologies, Inc. (Novato, CA). All probes, with the exception of the 18 S rRNA probe, were modified at the 5Ј end with the reporter dye 6-carboxyfluorescein-aminohexylamidite, and at the 3Ј end with the quencher dye 6-carboxytetramethyl rhodamine (Biosearch Technologies Inc.). The 5Ј end of the 18 S rRNA probe was modified with VIC (Applied Biosystems, Foster City, CA). For detection of BAH exon 19/20, primers ACC-CTGGCACGGATGATG and AAGTTCATACCACTTATATGCCTCT-TTG and probe 5Ј-TTCCCGACCCTCTGCATGGCA were used. For BAH exon 14a, primers GGAATTCAGGGTGTATGAGAAACAG and CCAGTGTATAAAGGAAGAGGCTCATC and probe CCCAGAGTTTG-CTGCTGGGTCCAA were used. For mouse junctin, primers ACCCAT-CAAAGAAGAGCTGAAGA and TCCTCCCCTTCCCTCTATCC and probe CCCTGCCTTCGCTCTTCATTCTTGCT were used. For 18 S rRNA, primers CGGCTACCACATCCAAGGAA and GCTGGAATTAC-CGCGGCT and probe TGCTGGCACCAGACTTGCCCTC were used. Total RNA was prepared from murine tissues using the RNeasy purification system (Qiagen) according to the manufacturer's instructions. cDNA synthesis was performed with the Advantage RT-PCR kit (CLONTECH) according to manufacturer's instructions using random hexamers and 1 g of DNase I-treated total RNA. Taqman-based real time PCR expression profiling was performed using 25 ng of each cDNA according to the manufacturer's instructions (Applied Biosystems) with real-time fluorescence being monitored with an ABI Prism 7700 (Applied Biosystems). Relative expression levels of the various transcripts were determined essentially as described (20). Briefly, standard curves were generated for each transcript using a serial dilution of mouse heart cDNA. Relative abundance was then determined by comparing the cycle threshold values for each reaction with this standard curve. Abundance levels calculated from negative control reactions performed in the absence of reverse transcriptase were then subtracted from experimental sample abundance. Input levels of cDNA were normalized to 18 S rRNA levels. All expression measurements were performed in duplicate using two independently generated cDNA samples.
Antibodies to the COOH-terminal Catalytic Region of Human BAH-Recombinant human aspartyl ␤-hydroxylase 52-kDa catalytic fragment (11,16) was purified to greater than 90% purity, as judged by SDSpolyacrylamide gel electrophoresis. Purified protein was used to prepare monoclonal antibodies in the mouse by conventional techniques (22). Two of these antibodies (HBOH-1 and HBOH-2) were characterized and used for Western analysis of human cell lines.
Epitope Mapping of HBOH-1 and HBOH-2-Monoclonal antibodies HBOH-1 and HBOH-2 were biotin-labeled using NHS-LC-biotin (Pierce) as follows. Approximately 50 g of antibody were incubated with 2 l of 3.9 mM NHS-LC-biotin in PBS buffer. After 2 h, 1 ml of 1 M ethanolamine was added to block unreacted biotin reagent. To minimize losses during washing, 20 l of 50 mg/ml bovine serum albumin were added and the labeled antibody washed three times with 2 ml of PBS using a Centricon concentrator (Millipore, Bedford, MA) with a 30-kDa cut-off.
Phage displayed peptide libraries encoding seven or eight random amino acids (23, New England Biolabs, Beverly, MA) were screened to identify peptide-binding phage as follows. One microgram of biotinylated antibody was incubated with approximately 10 11 plaque-forming units of the library at 4°C overnight. The phage antibody mixture was incubated for 10 min on streptavidin-coated dishes (0.5 mg/ml in 0.1 M NaHCO 3 , pH 8.0) that had been blocked with bovine serum albumin (29 mg/ml in 0.1 M NaHCO 3 ). Plates were washed 10 times with 3 ml of Tris-buffered saline containing 0.5% Tween 20 over 30 min. Bound phage were eluted with glycine buffer, pH 2.2, neutralized with 2 M Tris, and titered for plaque-forming units on XL-1 Blue. Neutralized phage were amplified by infection of XL-1 Blue E. coli and used for additional rounds of selection. After three rounds of selection, individual plaques were picked and amplified to recover DNA for sequencing.
Epitope Mapping FB50 Monoclonal-FB-50 was the kind gift of Dr. Jack R. Wands (16). A set of 90 peptides comprising the region between residues 82 and 354 of human BAH (16) were synthesized by Chiron Mimotopes (Raleigh, NC). The fifteen residue peptides were synthe-sized covalently attached to pins to facilitate antibody screening in a 96-well format. The pins were used according to the manufacturer's instructions to identify peptide fragment(s) recognized by the antibodies in question.
Antisense Studies and Western Analysis-Oligonucleotides for antisense experiments were synthesized and purified using protocols previously described (24). A549 cells were plated at a density of 5.5 ϫ 10 5 cells/6-cm plate and cultured in high glucose Dulbecco's modified Eagle's medium containing 10% fetal calf serum, 2 mM glutamine, 100 units/ml penicillin, and 10 g/ml streptomycin, for 24 h. The following day, the oligonucleotides (0.4 M final concentration) were mixed with lipofectin (20 g/ml final concentration, Life Technologies, Inc.) and the cells were incubated with this complex for 4 h. The oligonucleotidelipofectin complex was washed off, and fresh medium containing 10% fetal calf serum was added to the cells. A second addition of the complex was carried out 48 h later. Cells were harvested for Western analysis 24 h after this second addition and resuspended in SDS lysis/gelloading buffer to a final concentration of 2 ϫ 10 6 cells/ml buffer. Five l of lysate solution was fractionated on a 7.5% SDS-polyacrylamide gel (Bio-Rad) for 1 h at 100 V in Tris-glycine buffer. Protein was transferred onto a nitrocellulose membrane by electrotransfer and probed with primary antibody (FB-50 or HBOH-1 at 1:5000 and 1:500 dilutions, respectively) at 4°C overnight. Secondary goat anti-mouse antibodies conjugated to horseradish peroxidase (Vectastain ABC kit, Vector Laboratories, Burlingame, CA) were applied according to manufacturer's instructions. Detection was accomplished using the Renaissance ® Western blot chemiluminescence reagent (PerkinElmer Life Sciences).

Murine BAH mRNA Is Present as Three Major Bands in Multiple Tissue Northern Blots-Previous Northern analyses of human and bovine
RNAs have demonstrated that multiple RNA species contain BAH-related sequences (12,14,16). To clarify the relationship of BAH to these transcripts, a detailed RNA, cDNA, and genomic analysis of the murine BAH gene was undertaken. Northern analysis of poly(A) ϩ RNAs derived from multiple mouse tissues probed with a murine BAH cDNA, which includes the BAH intralumenal domain (probe A, Fig.  1D), demonstrated the presence of three different size transcripts (2.8, 4.5 and 6.6 kb). These transcripts are expressed in a wide variety of normal murine tissues (Fig. 1A).
Characterization of the Multiple BAH Transcripts by cDNA Cloning and Northern Analysis-Seven BAH cDNA clones were isolated from a murine liver cDNA library. The largest clone included a polyadenylation signal and extended 5Ј to include the transmembrane region of BAH, but stopped short of the translation start site. 5Ј-RACE was used to isolate a complete open reading frame for BAH and to define the transcription start site of BAH, 174 nucleotides upstream of the ATG codon (Fig. 1D, 1). This full-length clone is 4,419 bp in length (GenBank accession no. AF289486) and is consistent with the FIG. 1. Mouse and human BAH cDNAs and relationship to Northern blots. A, a normalized multiple tissue Northern blot from adult normal mouse using BAH probe A. B, a Northern blot of mouse liver hybridized with probe B (1), and then stripped and hybridized with probe C (2). C, Northern blot of mouse liver hybridized with probe B (1) and then stripped and hybridized with probe D (2). D, schematic diagrams of the four alternative spliced genes that are a part of the BAH genomic locus including the three isoforms of BAH seen in Northern blots (1-3) and mouse junctin (4). The four regions of BAH are indicated in color: green, NH 2 -terminal cytoplasmic; yellow, transmembrane region; purple, highly charged lumenal region; red, catalytic domain. Junctin-specific sequences are indicated in blue. nts, nucleotides. middle transcript band detected in the Northern analysis. To examine the 3Ј end of the BAH transcript, a 3Ј-RACE experiment was performed, priming from the 3Ј-UTR mouse BAH clone. Fragments isolated from this RACE experiment identified a second class of transcript which terminated after a second polyadenylation signal 6,629 bp downstream from the transcription start site (Fig. 1D, 2, accession no. AF289487). This transcript is consistent with the largest size transcript, 6.6 kb, identified by Northern analysis. Both the 4.5-and 6.6-kb clones are identical in sequence through the cytoplasmic domain, transmembrane region, highly charged region and catalytic domain. They share a common 3Ј-UTR up to the first polyadenylation signal at 4.5 kb. The longer clone extends the 3Ј-UTR for another 2.1 kb until it terminates at a second polyadenylation signal.
Characterization of additional murine BAH related cDNA clones identified a class whose homology with the full-length BAH clone terminated 1,094 bp downstream of the transcriptional start site (Fig. 1D, 3, accession no. AF289488). This cDNA is consistent in size with the smallest (2.8 kb) transcript identified on the mouse tissue Northern blots (Fig. 1, A and D (3)). Comparison of the sequence of this truncated clone to BAH revealed that they are identical in the cytoplasmic and transmembrane domains, and diverge near the beginning of the BAH catalytic domain. The coding sequence of the truncated clone terminates at a stop codon two amino acids downstream of the break in identity (Fig. 2). This truncated form of the BAH transcript (humbug) does not include the catalytic domain and therefore is predicted to be catalytically inactive as an aspartyl hydroxylase.
The three BAH-related transcripts observed in Northern blots were directly related to the three classes of mouse liver cDNAs by further Northern analysis. Northern blots of adult mouse liver RNA probed with a 5Ј fragment (probe B) that is contained within all three BAH-related transcripts (Fig. 1, B (lane 1) and C (lane 1)) identified all three bands (2.8, 4.5, and 6.6 kb) previously seen on Northern analysis (Fig. 1A). Rehybridization of this Northern with a BAH catalytic domain probe (probe C) recognized both of the larger transcripts (4.5 and 6.6 kb), but not the 2.8-kb band (Fig. 1B, lane 2). In contrast, when the Northern blot depicted in Fig. 1C (lane 1) was re-probed with a fragment from the 3Ј-UTR of the truncated clone (probe D), only the 2.8-kb transcript was identified (Fig. 1C, lane 2). These results are consistent with the 2.8-kb transcript representing humbug, the truncated catalytically inactive form of BAH, and the two larger transcripts representing the full-length catalytically active version of BAH, differing only in their use of alternative polyadenylation signals.
The humbug Transcript Exists in Drosophila, Mouse, and Man-To determine if the catalytically inactive form of BAH (humbug) exists in humans, 3Ј-RACE was performed on a human liver cDNA library using a primer chosen from the highly charged region of human BAH. A 2.8-kb human ortholog of mouse humbug was obtained (accession no. AF289489). The deduced amino acid sequences of the mouse and human humbug transcripts indicate that protein truncation occurs at the identical position, 2 amino acids downstream of the break in homology with the full-length BAH sequence (Fig. 2).
With the identification of a major alternatively spliced humbug transcript in the mouse and human, and the knowledge that BAH activity has previously been demonstrated in insect cells (25), the existence of humbug in Drosophila was explored. Drosophila BAH sequences were obtained by degenerate PCR, EST data base searching, and 5Ј-and 3Ј-RACE. Two different cDNAs were identified. The first contained a contiguous open reading frame of 2,358 bp (786 amino acids) that included a putative cytoplasmic domain, transmembrane domain, a highly charged region, and a catalytic domain. The second cDNA, however, encoded a truncated open reading frame of 1149 bp (383 amino acids) whose structure and sequence are consistent with it being the Drosophila ortholog of humbug (Figs. 4C and 2, accession nos. AF289493 and AF289494).
Mouse junctin Shares 100% Sequence Identity over a 220-bp Region with Mouse BAH and humbug-Previously, a crossspecies comparison of bovine BAH and dog junctin has shown that these genes share 95% amino acid identity over a 73residue region, which includes the putative transmembrane domain of BAH (17). To determine if this high degree of homology is derived from two closely related genes or whether junctin shares exons with BAH and humbug, a mouse junctin cDNA was cloned (accession no. AF289490) from a mouse heart cDNA library. When the sequences of mouse BAH/humbug derived from mouse liver and mouse cardiac junctin are compared, a region of 100% identity spanning 220 nucleotides is observed (Fig. 3). The region of identity includes the BAH/humbug transmembrane region and the first 42 amino acids of the protein on the endoplasmic reticulum lumenal side of the membrane. The sequence of the cytoplasmic domain of junctin and the downstream lumenal sequence diverge from BAH/humbug (Figs. 1D (4) and 3).
Mouse BAH Genomic DNA Spans over 200 kb and Encodes Three Distinct Proteins: BAH, humbug, and junctin-To characterize the genomic organization of BAH, humbug, and junctin, a 129/SvJ mouse ES cell BAC genomic library was screened. Probes were used from both the 5Ј and 3Ј ends of mouse BAH cDNA to maximize the chances of obtaining the genomic regions spanning the entire open reading frame. Three separate BAC genomic clones were isolated, shotgun-cloned, and sequenced (Fig. 4). This analysis indicated that BAH is encoded by 24 exons and extends over 200 kb of genomic DNA (accession nos. AF289205-AF289215). Interestingly, exons 1 through 14 splice in a codon position that would permit alternative splicing to occur without disrupting the reading frame ( Table I).
The sequence homology between full-length BAH and humbug ends precisely at the end of BAH exon 13 (Table I, Fig. 2). Sequencing of the region between BAH exons 13 and 14 revealed the presence of the humbug alternatively spliced exon (exon 14a) downstream of BAH exon 13 (Fig. 4B, 7; accession no. AF289207). The distance between exon 14a and exon 14 was determined to be ϳ23 kb. Translation of an exon 13 to exon 14a alternative splice indicates that exon 14a encodes for only two additional amino acids followed by a stop codon (Fig. 2, Table I), in agreement with the sequence of the previously isolated humbug cDNA. This demonstrates that the humbug transcript is produced by alternative splicing of BAH exon 13 to humbug exon 14a.
A comparison of junctin sequence to the 24 BAH exons reveals that the 220 nucleotides of identity exactly match exons 2 and 3 of BAH. This demonstrates that the common region of junctin and BAH/humbug arise by exon sharing and suggests that the upstream and downstream sequences of junctin are encoded in other exons that are distinct from the 24 BAH exons. Sequence analysis of the genomic region between BAH exons 3 and 4 revealed two additional exons, 4a and 5a (42 and 1311 bp, respectively) that encode the remaining part of the lumenal domain of junctin (accession no. AF289200). Sequence analysis of the BAH genomic region between exons 1 and 2 identified one small exon, 1a, which encoded the cytoplasmic domain of junctin and includes the initiator methionine (accession no. AF289199). Exon 1a is located approximately 8 kb downstream of exon 1 and 4 kb upstream of exon 2. Thus, mouse junctin is encoded in 5 exons spread across ϳ25 kb, where exons 2 and 3 are shared with BAH/humbug (Fig. 4, A  and B (10)).
Alternative Splicing and Conservation of BAH Structure from Insect to Man-Northern analysis of BAH and humbug expression in mouse cardiac RNA suggested that the transcripts of these genes in cardiac tissues were slightly smaller than in other mouse tissue RNAs (Fig. 1A). To investigate this observation, 5Ј-RACE experiments utilizing BAH/humbug The putative TM region is overlined. Potential glycosylation sites are marked by asterisks above the human sequence. The epitopes of monoclonal antibodies FB-50, HBOH-1, and HBOH-2 are indicated. Arrowheads mark the putative NH 2 terminus of the processed 52-and 56-kDa forms of bovine BAH. The His 2 and dibasic Gly (DBG) motifs are indicated. Residues that are identical between all four species are shaded in black, and those with identity between only two or three species are shaded in gray. Arrows denote positions of the C termini in the truncated forms of the murine and human BAH (identical site and sequence) and Drosophila proteins, whereas the text following arrows denotes COOH-terminal sequences that are unique to the truncated forms. Small numbers above demarcated sequences indicate the ends of selected exons. exon 5 and 6 primers (chosen to be outside the exons shared with junctin) were performed on mouse heart tissue RNA. Twelve BAH/humbug 5Ј-RACE cDNAs were analyzed, and all were found to use the first exon of junctin (Fig. 4B, 3-6 and  8 -9). This suggests that in heart tissue, where junctin is highly expressed (see below), alternative splicing can occur such that use of the promoter associated with exon 1a leads to fusion of this exon into transcripts including BAH/humbug downstream exons. In fact, detailed analysis of the 12 isolated cDNAs revealed multiple alternative splicing patterns that included transcripts in which junctin exon 4a was fused to BAH/humbug exon 4. These changes do not alter the reading frame of downstream BAH sequences (Table I). In summary, the genomic loci of junctin (Ն25 kb, 5 exon gene), humbug (Ն60 kb, 14 exon gene), and BAH, (200 kb, 24 exon gene) overlap and share exons. Further, in heart tissue BAH/humbug utilizes junctin exon 1 (and sometimes exon 4) without altering its reading frame (accession nos. AF289491 and AF289492), while junctin shares exons 2 and 3 with BAH/humbug in all transcripts studied.
A comparison of the deduced protein coding sequences of mouse, bovine, human and Drosophila BAH is presented in Fig. 2. The putative transmembrane domain shows complete identity between mouse and human and is limited to a portion of exon 2 in the mouse. One of the major differences noted between the bovine and human sequence in the 5Ј end was the absence of a string of 15 amino acids in the human sequence (14). This additional sequence (Fig. 2), derived from a bovine brain cDNA, is inserted at the junction of exons 2 and 3, suggesting that an additional exon and splicing event accounts for this sequence. A search of the current EST data bases for human BAH sequences revealed one human EST clone (accession no. F07451) isolated from infant brain that contains this apparently missing exon and thus increases the overall homology between the orthologues. Because both clones that contain this additional sequence were derived from brain RNA, it is possible that this additional sequence represents a brain-specific alternatively spliced form. Other regions where evidence of alternatively spliced forms exists occurs at the ends of exons 10 and 13 where differences exist among several species (Fig. 2). Overall, the highest degree of homology shared between the various forms of BAH are sequences that encompass exons 2-3 and 14 -24 of mouse BAH, areas that encode the transmembrane and catalytic regions, respectively. Specifically, the catalytic domain of mouse BAH shares 44% identity with Drosophila BAH and greater than 95% identity with human or bovine BAH. The critical dibasic Gly and His 2 motifs of BAH are encoded on exons 22 and 23, respectively.
Drosophila BAH Genomic Structure-Comparison of the Drosophila BAH cDNA (accession no. AF289493) to sequences from genomic P1 clone DS03910 (accession no. AC004248) indicates that the gene contains 11 exons spanning 6.5 kb (Fig.  4C). The Drosophila humbug cDNA (accession no. AF289494) results from alternative splicing into exon 8, generating a protein that terminates 4 amino acids after the alternative splice (Fig. 2).
Expression Profiling of BAH, humbug, and junctin by Real Time PCR-To determine the tissue distribution of BAH/humbug/junctin transcription, quantitative real time RT-PCR profiling was performed across a panel of 14 murine tissues. Probe sets were generated to permit the specific detection of each of the three transcripts. These probe sets were derived from 1) the   catalytic domain region of BAH, 2) the 3Ј-UTR of humbug, and 3) junctin exon 5. The relative expression levels of these transcripts were normalized to the level of expression of each transcript in liver. Results from these experiments are shown in Fig. 5. The BAH catalytic domain is expressed at the highest levels in the heart, ovary, adrenal gland, lung, and fat (Fig. 5A).
The humbug expression profile was very similar to that of BAH with the same five tissues showing elevated levels of expression, but brain, stomach, and large intestine also showed significant levels of expression relative to liver (Fig. 5B). In contrast, mouse junctin is predominantly expressed in two tissues, heart and skeletal muscle, with relatively lower levels in ovary and adrenal gland (Fig. 5C). cDNA analysis (see above) has shown that transcription of BAH, humbug, and junctin in cardiac tissue is driven primarily from the promoter driving exon 1a (Fig. 4). Epitope Mapping Monoclonal Antibody FB-50 -A set of noncleavable 15 residue peptides attached to pins were screened to identify the region of human BAH recognized by mAb FB-50 (16). Strongly bound antibody was indicated for four of the immobilized peptides with signal ranging from 1.8 to 3.7 optical density at 405 nm. The peptides that bound FB-50, peptides 69 -72, comprise residues 277-300 of BAH. The common core peptide found in each of the 15-mers and, therefore, the epitope for FB-50 includes residues 286 -291 (NPVEDS) (Figs. 2 and  6A). Based on the position of this epitope, FB-50 mAb would be expected to bind to both full-length BAH and humbug, but not junctin. Further, the epitope recognized by FB-50 is amino-terminal to the start of the proteolytically processed 56-and 52-kDa forms of BAH isolated from bovine liver. This suggests that FB-50 should not bind to these proteolytically generated truncated catalytic forms.
Epitope Mapping mAbs HBOH-1 and HBOH-2-Phage displayed peptide libraries were screened to identify the BAH epitopes recognized by mAbs HBOH-1 and HBOH-2. During three rounds of selection for each antibody, the percentage of bound phage increased more than 3 orders of magnitude indicating enrichment of specific peptide binding phage. Sequencing of the random peptide inserts of selected phage indicated consensus sequences that correlate well with residues 573-579 (QPWWTPK) for HBOH-1 and a peptide comprising residues 613-620 (LPEDENLR) of human BAH for HBOH-2 (Figs. 2  and 6A). These antibodies are specific for the full-length, catalytically active form of BAH and would not be expected to recognize humbug or junctin.
Antisense Oligonucleotide Studies on A549 Cells Correlate Northern and Western Data for BAH and humbug-In the mouse, three transcripts corresponding to two deduced protein isoforms (BAH and humbug) have been identified. To determine whether humbug is expressed as a protein and to correlate the transcripts to protein species, Northern and Western analyses were performed on RNA and protein prepared from cultured human A549 cells. In contrast to the mRNA sizes observed in mouse Northerns (2.8, 4.5, and 6.6 kb, Fig. 1), the three BAH/humbug RNA transcripts in human A549 cells migrate at approximately 2.8, 4.5, and 5.2 kb (Fig. 6E). When FB-50 antibody (should recognize both BAH and humbug) is used to probe protein extracts on a Western gel from A549 cells, one band at 60 kDa and a doublet at 120 -130 kDa are identified (Fig. 6B). In contrast, monoclonal antibodies HBOH-1 and HBOH-2, which were epitope-mapped to the COOH-terminal region of human BAH, recognize only the upper (120 -130 kDa) bands observed on Western blots of A549 cells. These results demonstrate that the alternative splice into exon 14a identified in mouse and human humbug results in production of a humbug protein that migrates at 60 kDa in size while the fulllength version of BAH is represented by a doublet in the 120 -130-kDa range. A549 cells were treated with antisense oligos complementary to either the NH 2 -terminal (oligo 583) or COOH-terminal regions (oligo 1302), and cell extracts were subjected to Western analysis. Oligo 583 treatment, which is complementary to BAH and humbug, reduced all three transcripts observed in Northern blots of A549 cells (Fig. 6E) and reduces both the 60-and 120 -130-kDa bands by Western analysis (Fig. 6C). In contrast, oligo 1302, which is complementary only to the BAH transcript, resulted in a reduction of only the upper two transcripts on Northern blots of A549 cells (Fig. 6E) and the 120 -130-kDa protein doublet on Western analysis in A549 lysates (Fig. 6D). These results are consistent with the conclusion that the 2.8-kb transcript encodes the humbug protein which migrates at 60 kDa in Western blots of A549 cells, whereas the larger (4.5 and 5.2 kb) human BAH transcripts result in a protein with a relative molecular mass of approximately 120 -130 kDa.

DISCUSSION
Three major BAH-related transcripts have been identified by Northern analysis on RNA from multiple mammalian species (2.8, 4.5, and 6.6 kb in the mouse). cDNA analysis of these transcripts from mouse liver was used to define the structure of these transcripts. The two largest transcripts, 4,419 and 6,629 nucleotides, encode full-length BAH and have identical coding regions, which include cytoplasmic, transmembrane, highly charged lumenal region, and COOH-terminal catalytic domain. These transcripts differ only in the 3Ј-UTR, where the longer transcript uses a downstream poly(A) addition signal. cDNA clones (2,789 nucleotides) representing the shortest transcript revealed the existence of a truncated form that is identical to the BAH transcript from the NH 2 terminus through the highly charged lumenal region. This "humbug" transcript lacks the defining 56-kDa catalytic region of BAH (11,12). humbug is conserved in other species. For example, humans express humbug RNA in a wide range of tissues. Further, in studies presented above, a humbug ortholog exists in Drosophila. Conservation of this truncated transcript suggests a biologically significant function.
Using a combination of antisense and Western analyses with epitope mapped monoclonal antibodies, we have shown that the full-length human BAH protein runs as a doublet with an apparent mass of ϳ120 -130 kDa. In addition, this approach demonstrated that human humbug, the truncated transcript lacking the catalytic domain, is translated into protein with an apparent mass of ϳ60 kDa. It is interesting to note that the masses of BAH and humbug protein, predicted from transla-tion of the cDNA sequences, are 85 and 35 kDa, respectively. This differs significantly from the sizes seen in Western blots. The difference between the predicted mass and that seen on gels may be due to the highly charged nature of these two proteins (14) or glycosylation.
Previous Northern analysis (16) has been carried out to examine the expression of BAH in human tumor cell lines. It is now clear that the probe utilized in those experiments will hybridize with both BAH and humbug. Those studies revealed the presence of a major 2.8 kb transcript and a minor transcript of ϳ5.0 kb in a wide range of human tumor cell lines, but not normal human liver. Our results imply that the major transcript observed by Lavaissier et al. (16) corresponds to humbug and the minor transcript to BAH. Because the two genes share significant homology, use of selective and specific reagents will be required to identify the effects of BAH and humbug and their relationship to malignancy. For example, a recent report suggests that overexpression of BAH may be associated with malignant transformation (26). In one experiment presented in that study, immunoreactivity with the FB-50 antibody in human cholangiocarcinoma was convincingly demonstrated. Based on the results presented above, which demonstrate that FB-50 recognizes both BAH and humbug protein, this reactivity could be caused by binding to BAH, humbug or both.
Jones et al. (17) have previously shown that canine junctin shares significant homology with bovine BAH over a small region. To determine if the homology of junctin with BAH/ humbug is derived from two different genes or whether they share common exons, a mouse junctin cDNA was isolated and sequenced. A comparison of the mouse junctin sequence to mouse BAH/humbug demonstrated that the two sequences were identical for 220 nucleotides. This sequence was internal to the overall coding sequence of junctin and included the putative transmembrane domain. These results strongly suggest that junctin and BAH/humbug are encoded using common exons.
Clarification of the relationship of BAH to humbug and junctin was sought by the cloning and isolation of the BAH genomic locus from mouse. These studies demonstrated that BAH is encoded by 24 exons spanning over 200 kb of DNA. The humbug sequence was shown to be derived from the first 13 exons of the BAH locus with the break in homology arising precisely at the end of exon 13. Searching the DNA residing between BAH exon 13 and 14 revealed another exon, 14a, which completed the coding region of humbug and provided a stop codon and poly(A) addition signal. A blast search of GenBank with the 3Ј-UTR of human humbug revealed one sequence-tagged site (STS WT-11767) that confirms that humbug maps to the same chromosomal position as BAH (8q12) (16).
A comparison of mouse junctin sequence to BAH exon sequences revealed that the 220 nucleotides of identity between junctin and BAH were defined precisely by the coding boundaries of BAH exons 2 and 3. Characterization of the DNA between BAH exon 1 and 2 led to the identification of junctin exon 1a, which encoded the region of junctin upstream of the BAH exon 2 homology. Examination of the DNA downstream of BAH exon 3 permitted the localization of two additional exons, 4a and 5a, which completed the exon structure of junctin. Based on these studies, it is clear that the genomic coding DNA for junctin is contained in the BAH locus and, when expressed, junctin shares exons 2 and 3 with BAH and humbug.
The close coding relationship between BAH, humbug, and junctin led us to study the expression of these three transcripts in RNA derived from multiple tissues. These studies showed that both BAH and humbug were widely expressed with the highest levels occurring in heart, ovary, and adrenal gland. The distribution of junctin expression was more limited, with the highest levels of expression occurring in heart and skeletal muscle and lower levels in ovary and adrenal. The overlap of high relative levels of expression of junctin, BAH, and humbug in heart led to the examination of the structure of BAH/humbug transcripts in RNA derived from this tissue. Of 12 cDNA clones examined, all of the BAH/humbug transcripts in the heart began with exon 1a of junctin, not exon 1 of BAH/humbug. This suggests that the promoter associated with exon 1a provides strong tissue-specific transcription, which produces expression of BAH/humbug in this tissue as well as junctin.
Because the coding region of exon 1a of junctin is smaller than exon 1 of BAH/humbug, the cardiac forms of BAH/humbug should be at least 3.2 kDa smaller and perhaps smaller still since exon 1 of BAH/humbug contains a putative N-linked glycosylation site (Fig. 2). Analysis of the downstream sequences for these cDNAs revealed that, although most of the BAH/humbug clones examined spliced from exon 3 to exon 4, a significant number of clones spliced exon 4a of junctin to exon 4 of BAH/humbug. These data demonstrate that in heart, BAH/ humbug shares more similarity to junctin, which not only includes exon 2 and 3, but also frequently includes exon 1a and 4a of junctin. Thus, in heart, BAH/humbug is expressed as a transcript that includes 3 or 4 of the 5 exons of junctin. This leads to the hypothesis that transcription of BAH, humbug, and junctin may initiate with exon 1a in heart, and perhaps skeletal muscle, and exon 1 in most other tissues. According to this proposal, the level of expression of BAH, humbug, and junctin in a particular tissue would be controlled by both tissue-specific transcription initiation and tissue-specific RNA splicing. This idea is supported by the observation that, although junctin, BAH, and humbug are highly expressed in heart using exon 1a, only junctin appears to be highly expressed in skeletal muscle. This raises the possibility that, in skeletal muscle, both transcription initiation control and tissue-specific splicing of junctin, BAH, and humbug may be differentially regulated. Because junctin RNA is easily detected in ovary and adrenal gland, tissues where BAH and humbug are relatively highly expressed, junctin RNA can be studied in these tissues to determine if junctin downstream sequences are fused to BAH/ humbug exon 1. Detailed analysis of junctin, BAH, and humbug transcription initiation and splicing in other tissues is being investigated.
The fusion of junctin sequences to BAH/humbug in murine heart leads to a discussion of their functional role and relationship in this tissue. Junctin is known to be an important part of a complex of proteins in heart junctional sarcoplasmic reticulum that includes the ryanodine receptor, triadin, and calsequestrin. These proteins act coordinately to release Ca 2ϩ from intralumenal stores. Interestingly, examination of triadin, calsequestrin, and junctin reveals the presence of intralumenal domains that contain a large number of charged amino acids (rat triadin: 47% charged, 27% positive, 20% negative; mouse junctin: 49% charged, 29% positive, 20% negative). It has been suggested that the intralumenal charged regions of triadin and junctin are essential for forming this protein complex (27). The repeating KEKE motifs of the charged regions of calsequestrin appear critical for protein-protein interactions and may also play a role in binding of Ca 2ϩ (mouse calsequestrin: 38% charged, 11% positive, 27% negative). Interestingly, BAH/humbug also contain a region of highly charged amino acids on the putative lumenal side of the membrane, but this region is not derived from the junctin coding exons. Instead it is encoded in exons 4 -13 of BAH/humbug and is significantly more negatively charged (40% charged, 8% positive, 32% negative) than junctin. Exon sharing with junctin and the presence of a BAH/ humbug encoded highly charged domain proximal to the membrane on the lumenal side of the junctional sarcoplasmic reticulum leads to the proposal that, in heart, BAH/humbug may be associated with the ryanodine receptor complex in heart. Junctin overlay experiments done on canine cardiac junctional sarcoplasmic reticulum by Zhang et al. (27) revealed the presence of calsequestrin and triadin. In addition to these proteins there are bands in this overlay with sizes similar to humbug and BAH. Zhang et al. (27) also used calsequestrin affinity chromatography to examine this protein complex, and junctin and triadin were co-purified by this method. On this stained gel, proteins consistent with the sizes of humbug and BAH were identified. These results suggest that it is worth directly testing whether BAH and humbug are associated with this complex.
Alternative splicing produces humbug, which lacks the 56-kDa catalytic domain of BAH. There have been a number of published accounts of protein isoforms that result from alternative splicing interfering with the catalytic activity of their catalytically active partners either by competing for substrate molecules or by direct interaction/inhibition of the full-length isoforms (28 -31). It is possible that humbug acts as a modifier of BAH function by altering substrate-enzyme interactions, directly inhibiting enzyme activity or competing with BAH for assembly into larger protein complexes such as those seen associated with the ryanodine receptor.
The widespread expression of BAH and humbug in normal mouse tissue and human tumors and cancer cell lines, coupled with conservation of these genes back to Drosophila, suggest that BAH and humbug are likely to have an important functional roles in vivo. To approach the possible role of BAH and humbug in normal and disease processes in the mouse, the mouse BAH gene has been carefully characterized at the cDNA and genomic levels. This information is being used to develop transgenic and knockout models of these genes with the aim of carefully dissecting what is likely to be a highly complex process of protein modification, assembly, and signaling in vivo.