Monooxygenase X, a Member of the Copper-dependent Monooxygenase Family Localized to the Endoplasmic Reticulum*

Based on sequence comparisons, MOX (monooxy-genase X), is a member of the copper monooxygenase family that includes dopamine (cid:1) -monooxygenase (DBM) and peptidylglycine (cid:2) -hydroxylating monooxygenase (PHM). MOX has all of the residues expected to be critical for copper binding, and its cysteine residues can yield the intramolecular disulfide bond pattern observed in DBM. Although DBM and PHM function within the lumen of the secretory pathway, the published sequence for human MOX lacks a signal sequence, suggesting that it does not enter this compart-ment. We identified an upstream exon that encodes the signal sequence of human MOX. A retained intron yields minor amounts of transcript encoding MOX without a signal sequence. MOX transcripts are widely expressed, with the highest levels in the salivary gland and ovary and moderate levels in brain, pituitary, and heart. Despite the presence of a signal sequence, exogenous MOX is not secreted, and it localizes throughout the endoplasmic reticulum in both endocrine or non-endocrine cells. Neither appending green fluorescent protein to its C terminus nor deleting the hydrophobic domain near its C terminus facilitates secretion of MOX. MOX is N -glycosylated, is tightly membrane-associated, and forms oligomers that are not disulfide-linked. Based on its sequence and localization, MOX is predicted to hydroxylate a hydrophobic substrate in the endoplasmic reticulum. C mMOX (MOX-GFP). C-terminal hydrophobic (Leu 597 -Leu mMOX Arg 596 GFP (MOX-Arg 596 -GFP) pEGF-P-N2 (Clontech). MOX-GFP from the mMOX pEAK10 a forward primer -TCTCAAGCCTCAGACAGTGGTTCA. KpnI site: MOX-GFP, (cid:1) -CGGGGTACCGCAAGCCCTGG- CTGCTCAGGA; MOX596s-GFP, 5 (cid:1) -CGGGGTACCACCGGAGGGAG-AAAATGCCGTG. TA-cloning

The copper/ascorbate-dependent monooxygenases constitute a small, but essential family of enzymes that use molecular oxygen and ascorbate to catalyze the hydroxylation of their substrates (EC 1.14.17.x) (1). The defining member of this family of enzymes was dopamine ␤-monooxygenase (DBM, 1 also known as dopamine ␤-hydroxylase or DBH; EC 1.14.17.1) (2, 3) (see Fig. 1). Hydroxylation of the ␤-carbon of dopamine consumes 1 mol of oxygen and 2 mol of ascorbate, yielding norepinephrine plus 2 mol of semidehydroascorbate. DBM was purified, sequenced, and studied in detail before it was cloned (4,5). The second member of this family, peptidylglycine ␣-hydroxylating monooxygenase (PHM; EC 1.14.17.3), catalyzes the ␣-hydroxylation of the C-terminal Gly residue in many different secreted peptides (6 -8). The 315-amino acid catalytic core of PHM, defined by truncation mutagenesis, is 28% identical to the corresponding region of DBM (9). Structural studies on the catalytic core of PHM defined the active site and revealed key roles for six copper-binding ligands (10 -12). Enzymes homologous to DBM and PHM are not found in yeast or bacteria.
Although the reactions catalyzed by DBM and PHM use very different substrates and produce very different products, the chemistry involved is similar and both enzymes produce products stored in regulated secretory granules and used for intercellular communication. Both enzymes are essential for survival. Genetically engineered mice lacking DBM are not viable (13). Mice lacking PHM develop massive edema at about embryonic day 14, with no live progeny produced (14). Drosophila lacking a functioning PHM gene generally die as late embryos (15). Both enzymes function in the lumen of the secretory pathway, and both require adequate supplies of substrate, reduced ascorbate, and copper. The vesicular monoamine transporters (VMAT1 and VMAT2) couple uphill transport of dopamine to efflux of protons, to deliver dopamine to the lumen of the secretory pathway (16,17). The peptidylglycine substrates of PHM are produced from prepropeptides synthesized in the endoplasmic reticulum and subject to endo-and exoproteolytic processing as they progress through the secretory pathway (6,18).
A third family member, monooxygenase X (MOX), was identified in a search for genes whose expression was altered in senescent human fibroblasts (19). Although key active site residues are conserved, the published sequence for human MOX has no signal sequence, making it difficult to see how it could function in a manner similar to DBM or PHM. A screen for genes involved in neural crest development yielded chicken MOX (also known as dopamine ␤-hydroxylase-related, DBHR) (20), which includes an N-terminal signal sequence. A potential homolog identified in the mouse genome (BAA95089), also includes an N-terminal signal sequence. No substrate has been identified for human or chicken MOX.
Because both DBM and PHM produce essential signaling molecules, it was tempting to speculate that MOX plays a similarly important role in signaling. Although very little is known about the MOX protein, transcripts encoding hMOX are prevalent in brain, kidney, and lung, with levels increasing in some lines of senescent fibroblasts (19,20). In the developing chick embryo, MOX transcripts are expressed in newly differentiating neural crest cells, most migrating neural crest cells, and in non-neuronal tissues such as the myotome. The properties of the MOX protein have not been explored.
We undertook these studies to determine whether a human MOX signal sequence could be identified. Monooxygenase family members present in Caenorhabditis elegans and Drosophila were identified. We next determined the sites of expression of MOX in adult mouse tissues and compared the pattern of MOX expression to that of DBM and PHM. Using several antisera to MOX, we characterized the expression of exogenous MOX in endocrine and non-endocrine cells. The unique features of MOX suggest that it functions in the endoplasmic reticulum.
To study the tissue-specific expression of human MOX mRNA and to verify the N-terminal splice variants, we designed three forward primers: 5Ј-ACCTATCCGCACCGGACCCT for N-terminal hMOX long form; 5Ј-GGACCTGATTCCCCAGTTGGA for N-terminal hMOX short form; 5Ј-GCACTGTGAGAGTGATCTGGG for the common region. A reverse primer (5Ј-GCCTCTCTGTATCACTGGCTC) in the common region was paired with the three forward primers to generate PCR products specific for long form (620 bp), the short form (600 bp), and both forms (250 bp). Human tissue total RNA was purchased from Clontech, and cDNA was prepared as described above. Amplification with ␤-actin-specific primers was used as an internal standard (21).
Construction of Expression Vectors-To generate an expression vector encoding full-length mMOX, we used a pair of primers with appended restriction enzyme sites (underlined): the forward primer included a Pci1 site, 5Ј-TACATGTGCGGCTGGCCACTGCT; the reverse primer included a NotI site, 5Ј-AGCGGCCGCGTACCCACCCATCTCTCCA. Mouse brain first strand cDNA was used as the PCR template, and the product was subcloned into the PCR II vector (TA-cloning vector, Invitrogen) and digested with Pci1 and NotI (New England Biolabs) to release the 1.8-kb mMOX insert. The pEAK10 expression vector (Edge Biosystems, Gaithersburg, MD) was digested with NcoI and NotI to receive the mMOX insert. The insert was sequenced in its entirety by the Molecular Core at the University of Connecticut Health Center. To visualize mMOX in live cells, we fused GFP to the C terminus of mMOX (MOX-GFP). To evaluate the role of the C-terminal hydrophobic region (Leu 597 -Leu 613 ), mMOX truncated at Arg 596 was fused to GFP (MOX-Arg 596 -GFP) in the pEGF-P-N2 vector (Clontech). Vectors encoding these MOX-GFP fusion proteins were prepared from the mMOX pEAK10 vector using a forward primer that contained the pEAK10 HindIII site just upstream of the translational start site, 5Ј-TCTCAAGCCTCAGACAGTGGTTCA. Both reverse primers included a KpnI site: for MOX-GFP, 5Ј-CGGGGTACCGCAAGCCCTGG-CTGCTCAGGA; for MOX596s-GFP, 5Ј-CGGGGTACCACCGGAGGGAG-AAAATGCCGTG. The PCR products were subcloned into the TA-cloning vector; inserts released with HindIII and KpnI were inserted into pEGFP-N2 vector digested with these same enzymes and sequenced.
MOX Antibodies-Four MOX peptides were synthesized, and each was linked to keyhole limpet hemocyanin using glutaraldehyde: CT45 was raised to mMOX( 243 SSNFNDSVLDFGHE 256 ); the mMOX( 448 KDR-AVMTWGGLSTR 461 ) and mMOX( 346 SLFHTIPPGMPEF 358 ) peptides yielded no useable antibodies; CT164 was raised to mMOX( 90 YFTNA-DRELEKDAQQDY 106 ). Production of antisera was carried out by Covance (Denver, PA). pEAK RAPID cells transiently transfected with vectors encoding GFP or mMOX were used to characterize the antisera. CT45 recognized native, but not SDS-denatured MOX; no signal was detected on Western blots of transiently transfected cells, but MOX could be visualized in paraformaldehyde-fixed transfected cells or immunoprecipitated from cell extracts. CT164 visualized a band of the expected mass only in cells expressing mMOX; samples prepared by denaturation at 37°C or 50°C exhibited substantially less aggregation than samples prepared by denaturation at 100°C.
Transfection of Cells-AtT-20 corticotrope tumor cells were cultured in Dulbecco's modified Eagle's medium containing 10% fetal calf serum, 10% NuSerum (Collaborative Research, Waltham MA). pEAK RAPID cells were cultured in Dulbecco's modified Eagle's medium containing 10% fetal calf serum. For transfection, cells were grown to 80% confluence and rinsed with complete serum-free medium (CSFM) (22,23). Expression vectors were incubated with LipofectAMINE 2000 (4 g of DNA/10 l of LipofectAMINE 2000; Invitrogen) for 20 min at room temperature, mixed with Opti-MEM (Invitrogen) and incubated with cells for 4 -6 h. Two days after transfection, cells were harvested for analysis.
Metabolic Labeling, Extraction, Immunoprecipitation, and Western Blot Analysis-Replica wells of pEAK RAPID cells were transiently transfected with mMOX.pEAK10. After 48 h, cells were rinsed with CSFM, incubated in medium lacking Met for 10 min, and then incubated in medium containing [ 35 S]Met (0.7 mCi/ml) for 30 min (Pulse). After the pulse labeling, cells were rinsed with CSFM and either harvested immediately (Pulse) or further incubated in CSFM for the designed amount of time (Chase). Cells were extracted in 0.5 ml of TM buffer (20 mM sodium N-tris(hydroxymethy)methyl-2-aminoethanesulfonic acid, 10 mM mannitol, pH 7.4, containing protease inhibitors) (24). Following centrifugation at 150,000 ϫ g for 30 min, the supernatant (soluble fraction) was collected and the pellets (particulate fraction) were solubilized with 0.5 ml of TMT buffer (TM with 1% Triton X-100), followed by centrifugation at 150,000 ϫ g for 30 min, and the supernatant was kept in fresh tubes. Spent chase medium was collected, and protease inhibitors were added. For immunoprecipitation, samples were incubated with MOX antibody CT45 or CT164 for 2 h on ice; antigen/antibody complexes were isolated using Protein A-Sepharose beads. Bound proteins were dissociated by boiling the resin in 1ϫ Laemmli sample buffer and fractionated on 4 -15% SDS gels (Bio-Rad); gels were soaked in Enhance and Amplify (PerkinElmer Life Sciences/ Amersham Biosciences), dried, and exposed to film. For co-immunoprecipitation experiments, pEAK RAPID cells transiently expressing mMOX were extracted as above and MOX was isolated using CT164. Immunoprecipitates were fractionated by SDS-PAGE and visualized using antisera to MOX (CT164, 1:1000), calnexin (BD Biosciences, 1:1000), or calreticulin (BD Biosciences, 1:2500).
To assess glycosylation, immunoprecipitates of metabolically labeled soluble and particulate MOX were heated at 100°C for 5 min in 0.5% SDS, 1% ␤-mercaptoethanol, and then diluted 2-fold with 50 mM sodium citrate buffer (pH 5.5). After treatment with endoglycosidase H (500 units/l, New England Biolabs) at 37°C for 1 h, samples were analyzed using 4 -15% SDS-PAGE and fluorography. pEAK RAPID cells transiently transfected with the mMOX.pEAK10 vector for 48 h were pretreated with 2 g/ml of tunicamycin for 3.5 h. Cells were then incubated in medium containing [ 35 S]Met and processed as described above.
Membrane Fractionation and Sodium Carbonate Extraction-Isolation of intracellular membranes and extraction with sodium carbonate was described (25). pEAK RAPID cells were transfected with pEAK10-MOX, pCIS-PAM-3, or pCIS-PAM-1 expression vectors for 48 h and harvested into TM buffer. After a 5-min centrifugation at 5,000 ϫ g to remove debris, the supernatants were centrifuged at 355,000 ϫ g for 30 min to separate the soluble fractions and pellets. Pellets were resuspended in 100 mM sodium carbonate, pH 11.0, for 30 min on ice, followed by centrifugation at 355,000 ϫ g for 1 h to generate sodium carbonate insoluble fraction, which was then sonicated into TMT buffer, and a sodium carbonate soluble fraction. Equal amounts of each fraction were subjected to Western blot analysis with CT164 for MOX, GFP polyclonal antibody (AbCam, 1:2500) for MOX-GFP and MOX-Arg 596 -GFP, and PAM polyclonal antibody 1761 (1:1000) for PAM-3 and PAM-1.
Sucrose Gradient Fractionation-Continuous sucrose gradients were prepared using solutions of 5 and 20% sucrose in TMT buffer; a cushion of 50% sucrose in the same buffer was placed below the 1878 l of gradient. Samples of metabolically labeled MOX (250 l of a 30-min Pulse sample) and culture medium from a stably transfected DBM cell line were placed above the gradient. Before application to the gradients, samples were either treated with dithiothreitol (1 mM for DBM; 10 mM for MOX; 30 min at 37°C) or not. Gradients were centrifuged for 5 h at 214,000 ϫ g in a Ti-55 swinging bucket rotor (22). For calibration, a sample containing cytochrome c, ovalbumin, bovine serum albumin, catalase, and apoferritin (200 g each) was analyzed simultaneously. After centrifugation, samples were collected starting from the top of the gradient; 1ϫ Laemmli sample buffer was used to recover protein that reached the bottom of the gradient. MOX was immunoprecipitated from each fraction using CT45 and fractionated on a 10% SDS gel. For DBM, equal aliquots of each fraction were fractionated on a 10% SDS gel and detected with a DBM antibody (1:1000, Ab3125).

Distinguishing Features among Copper Monooxygenase Fam-
ily Members-Although their catalytic cores are highly homologous, the topologies predicted for MOX, DBM, and PAM are distinctly different (Fig. 1). Although mouse MOX is predicted to have a cleaved signal sequence, human MOX lacks a signal sequence (19). Both mouse and human MOX have a stretch of 17 hydrophobic residues near their C termini. DBM has a signal anchor, and its disulfide-linked dimers form tetramers (26). PAM is a Type 1 membrane protein with an N-terminal signal sequence and a transmembrane domain near its C terminus (27). Similar disulfide bonding patterns have been iden-tified in DBM and PAM, and these Cys residues are conserved in MOX (Fig. 1).
Both the human genome and the Drosophila genome contain three copper monooxygenase family members (Table I). Consistent with their size and predicted topologies, the catalytic core of MOX is more similar to DBM than to PHM. The mouse genome contains a fourth copper monooxygenase family member, DBHL, which is slightly more similar to MOX than to DBM. Although the intron/exon boundaries of mDBHL are similar to those of hMOX and hDBM, a gene resembling mD-BHL could not be identified in the human genome. The functions of two of the Drosophila copper monooxygenases have been studied: the catalytic core of dPHM is 39% identical to that of hPHM, and the catalytic core of Drosophila tyramine ␤-hydroxylase is 50% identical to that of hDBM (Table I). The third family member in Drosophila is almost equally identical to hMOX and hDBM and is designated dMOX. The C. elegans genome encodes only two family members. C. elegans PHM is 38% identical to hPHM. The protein encoded by C. elegans cosmid H13N06 (19021-24000 bp) is 48% identical to hDBM, identifying it as part of the catecholamine biosynthetic pathway. In C. elegans, as in Drosophila, the major function of this  enzyme is hydroxylating tyramine to produce octopamine (28).
Knowledge of the crystal structure of PHMcc and biochemical modification of DBM provide another means of analyzing MOX (Table II). The six ligands that interact with the two essential copper atoms in PHM are completely conserved among all family members. The active site residues that interact with the peptidylglycine substrate have been identified (1,12). In addition, molecular modeling has predicted residues in DBM that may interact with its substrate, dopamine (1). When the homologous residues in MOX are identified, it is striking that all are hydrophobic. For example, Arg 240 , which binds the peptide carboxylate in PHMcc, is Gln 395 in hDBM and Leu 319 in hMOX. Similarly, Asn 316 in PAM is another Leu in MOX. If the active site of MOX resembles that of PHM, its substrate is unlikely to be charged or highly polar.
Expression of MOX Is Widespread-We next set out to determine where MOX was expressed and to compare its expression pattern to those of PAM and DBM (Fig. 2). Total RNA prepared from the indicated adult mouse tissues was reverse-transcribed, and the cDNA used as a PCR template (Fig. 2). Genespecific primers were designed for PAM (420 bp), DBM (520 bp), MOX (420 bp), and DBHL (702 bp). MOX transcripts were readily identified in a wide variety of adult tissues. Levels were highest in the salivary gland and ovary, with lower levels in olfactory bulb, cerebellum/brain stem, parietal cortex, pituitary, atrium, ventricle, adrenal, thymus, testis, and kidney. Still lower levels of MOX transcript were detected in lung, lymph nodes, spleen, and liver (Fig. 2). Expression of DBM is much more restricted, with significant signal detected only in adrenal and detectable amounts in brainstem. As expected from similar studies in the rat, PAM transcripts are widely expressed, with highest levels in pituitary, atrium, ventricle, lung, and adrenal (Fig. 2). Expression of DBHL is very low and highly restricted. Signal was detected only in the thymus and the testis, and only after a nested PCR; the smaller PCR product detected in the testis could be nonspecific or a splice variant.
hMOX Can Include an N-terminal Signal Sequence-Following the identification of human MOX (hMOX; GenBank TM accession number AY007239), homologous genes were identified in mouse (mMOX; GenBank TM accession number: BAA95089), and chicken (GenBank TM accession number AF327450) (20). Based on SMART and Signal P analyses, both mouse MOX (mMOX) and Gallus MOX include N-terminal signal sequences (smart.embl-heidelberg.de; www.cbs.dtu.dk/services/SignalP-2.0); cleavage of the mMOX signal sequence is predicted to occur after Gly 19 (Fig. 3A, open arrow) In addition to lacking a signal sequence, the published sequence for hMOX (19) ap-peared to lack another 67 amino acid residues common to mouse (Fig. 3A) and chicken MOX. After this region of divergence at the N terminus, the amino acid sequences of hMOX and mMOX are 81% identical.
To search for an N-terminal signal sequence for hMOX, we used the nucleotide sequence unique to mMOX to search for homologous sequences in the human EST data base. A clone (GenBank TM accession number AI751100) encoding an amino acid sequence 76% identical to that of mMOX-(1-100) was identified (Fig. 3B). SMART and Signal P analyses of this sequence revealed that the first 19 residues should function as a signal sequence, with cleavage occurring after Gly 19 (Fig. 3B,  open arrow).
The gene encoding hMOX is situated on chromosome 6q (19) and, like DBM, includes multiple exons (Fig. 3C). Intron 3, the longest intron, spans 43.9 kb at the Pro 156 -Val 157 junction, whereas intron 2, the shortest, spans only 170 nucleotides. Blast analysis of the human genome using the unique sequence encoded by human EST AI751100 identified an additional exon upstream of those previously thought to define hMOX (Exon 0, Fig. 3C). We will refer to forms of hMOX that include this signal sequence as the long form of hMOX; the form originally identified by Chambers et al. (19), will be referred to as the short form of hMOX. The sequence unique to the N terminus of the short form of hMOX (S1a, Fig. 3C) is encoded by the intron that separates the first two exons of the human MOX gene.
The Major Form of hMOX Has a Signal Sequence-To compare expression of the long and short forms of the hMOX transcript, we paired three different sense primers with an antisense primer in the region common to both the short and long forms of hMOX (Fig. 4A). When paired with a sense primer in the common region, both forms of hMOX yield a 250-bp fragment (Common). When paired with a forward primer in exon 0, only the long form of hMOX yields a product (620 bp). When paired with a forward primer situated in the intron between exons 0 and 1, only the short form of hMOX yields a product (600 bp). Total RNA from adult human brain was reverse-transcribed, and PCR amplification was carried out using the three primer pairs for the same number of cycles. Very little product was observed using the primer pair specific for the short form of hMOX described originally (19) (Fig. 4B). Consistent with this observation, similar amounts of product  were observed using the primer pair specific for the long form of hMOX or the primer pair that amplifies both isoforms of hMOX.
We next used the common primer pair to evaluate expression of MOX in other human tissues (Fig. 4C). MOX transcripts were most prevalent in adult and fetal brain and spinal cord, with lower levels in cerebellum and fetal liver and barely detectable levels in placenta. In all of these tissues, PCR using primer pairs specific to the long and short forms of hMOX revealed a pronounced preponderance of the long form (data not shown). The major MOX transcript encodes a protein with a predicted cleavable N-terminal signal sequence, meaning that MOX, like DBM and PHM, should function within the lumen of the secretory pathway.
Despite Having a Signal Sequence, MOX Is Not Secreted-To explore the properties of MOX, we expressed it transiently in pEAK RAPID cells, which lack regulated secretory granules, and in AtT-20 corticotrope tumor cells, which have secretory granules. We developed antisera to several synthetic mMOX peptides (Fig. 5A). Antiserum (CT164) to the more N-terminal peptide identified a 72-kDa band, approximately the mass predicted for mMOX (69-kDa) (Fig. 5B). The signal was not shown by preimmune serum, and was blocked when the antigenic peptide was included with the antibody (data not shown). Antiserum generated to the more C-terminal peptide (CT45) detected native MOX but failed to reveal a signal on Western blots.
When spent medium was harvested from transfected cells expressing MOX, no secreted MOX was detected (Fig. 5B). For comparison, cells transiently transfected with PAM-3, a soluble, secreted protein, were analyzed at the same time (Fig. 5B); secretion of PAM-3 was readily detected. To explore the reason for this unexpected result, we constructed two fusion proteins: MOX-GFP, with GFP appended to the C terminus of MOX, and MOX-Arg 596 -GFP, with MOX truncated just before the C-terminal hydrophobic stretch and GFP appended (Fig. 5A). Neither variant of MOX was secreted; removing the C-terminal hydrophobic stretch did not facilitate the secretion of MOX (Fig. 5B). Cells expressing MOX-GFP consistently contained a protein the size of MOX, suggesting that proteolysis separated MOX from the appended GFP.
To understand why MOX was not secreted, we prepared FIG. 4. The most prevalent isoform of hMox includes an Nterminal signal sequence. A, primer pairs used to amplify a region common to both hMOX transcripts (250 bp) or unique to either the short form (600 bp) or the long form (620 bp) are identified. B, cDNA generated from adult human brain was amplified (35 cycles) using the primer pairs indicated in panel A. Similar results were obtained for all of the other human tissues examined (data not shown). C, cDNA was prepared from total RNA isolated from the indicated human tissues. The hMOX common region primers were used for PCR; ␤-actin primers were used as an internal control.

FIG. 3. The human genome encodes an isoform of MOX with a signal sequence.
A, SMART analysis (www.cbs.dtu.dk/services/SignalP-2.0) predicts that mMOX (BAA95089) has a cleaved signal sequence at its N terminus, with signal peptidase cleaving after Gly 19 (open arrow). In contrast, hMOX (AY007239) lacks a signal sequence. The region unique to mMOX is underlined. After this unique region, mMOX and hMOX are 87% identical at the amino acid level. The locations of the exon/intron boundaries in the hMOX gene are indicated (black arrows); the exon/intron boundary in mouse MOX that does not align with a boundary in human MOX is marked by a filled diamond. B, a Blast search of the human EST data base with the mMOX cDNA sequence encoding its unique N-terminal region identified an EST (AI751100) encoding a sequence with 76% identity at the amino acid level. As for mMOX, SMART analysis identifies a signal sequence in this version of hMOX, with Gly 19 serving as the signal peptidase cleavage site (open arrow). The exon/intron boundary is indicated (black arrow). C, the genomic DNA encoding this alternate N terminus (Exon 0) is located 26.4 kb upstream of the genomic sequence encoding the published N terminus of hMOX. The transcript encoding hMOX without an N-terminal signal sequence (short form, S1a) is indicated by a thick line. Exons are indicated by boxes and introns by lines (dotted lines when not drawn to scale). AI751100 is the GenBank TM number for the hEST clone encoding hMOX with a signal sequence. crude soluble and particulate fractions from pEAK RAPID cells transiently expressing MOX, MOX-GFP, MOX-Arg 596 -GFP, an integral membrane protein (PAM-1) or a soluble, secreted protein (PAM-3). The crude particulate fraction was washed with 0.1 M Na 2 CO 3 , pH 11.0, to remove peripheral proteins (25), and the carbonate pellet and supernatant were examined along with the soluble fraction. Almost all of the MOX was recovered in the carbonate pellet (Fig. 5C). As expected, intact PAM-1 was recovered in the carbonate pellet, whereas cleavage products lacking a transmembrane domain were found in the soluble fraction. PAM-3, which is secreted efficiently, was present in the soluble fraction and in the carbonate pellet. Like PAM-3, MOX-GFP was present in the soluble fraction and the carbonate pellet (Fig. 5C). Removing the C-terminal hydrophobic region of MOX before appending GFP had little effect on the partitioning of MOX among these different fractions (Fig. 5C). Although appending GFP to MOX or to MOX-Arg 596 increased its solubility and diminished the amount of protein recovered in the carbonate pellet, neither MOX-GFP fusion protein was secreted (Fig. 5B).
mMOX Is Localized to the Endoplasmic Reticulum-To explore the localization of MOX, we used differential centrifugation to fractionate pEAK RAPID cells transiently expressing MOX (Fig. 6A). MOX was largely recovered in the low speed pellet (P1), the crude nuclear fraction, and P2, an ER-and plasma membrane-enriched fraction, with detectable amounts in the high speed P3 microsomal pellet. MOX was not detected in the cytosolic fraction (Fig. 6A, left). Western blot analysis for an ER marker, protein disulfide isomerase, confirmed that protein disulfide isomerase was concentrated in the P2 fraction (Fig. 6A, right).
We used several different expression vectors to examine the subcellular localization of exogenous MOX in AtT-20 mouse corticotrope tumor cells and in pEAK RAPID cells using immunofluorescence (Fig. 6B). We expressed MOX, MOX-GFP, and MOX-Arg 596 -GFP. Each MOX protein exhibited the same behavior when expressed in AtT-20 cells (Fig. 6B) or in pEAK RAPID cells (data not shown). We first compared the localization of MOX-GFP (Fig. 6B) to that of MOX (Fig. 6D). MOX-GFP visualized by GFP fluorescence or staining with MOX antibody after fixation and permeabilization yielded the same diffuse, reticular pattern (Fig. 6B, MOX versus MOX-GFP). MOX yielded a similar diffuse reticular pattern (Fig. 6D). MOX-GFP was partially co-localized with BiP, an ER marker (Fig. 6C); MOX-GFP was more concentrated in the perinuclear region of the cell, with BiP more evenly distributed throughout the ER. When cells expressing MOX were visualized for ACTH, a secretory granule marker (Fig. 6D), or syntaxin 6, a trans-Golgi network marker (Fig. 6E) (29), no co-localization was observed.  MOX (CT164, red). C, MOX-GFP was visualized directly, whereas BiP, an ER marker, was visualized simultaneously using a rabbit polyclonal antibody (red). D and E, MOX was visualized using CT164 (green), whereas ACTH, a secretory granule marker, and syntaxin 6, a trans-Golgi network marker, were visualized simultaneously using monoclonal antisera (red, syntaxin staining is indicated by the white arrowhead). Merged images are shown in the bottom panels. Images were taken with a 60ϫ oil objective and a deconvolved image through the middle of each cell is shown. Scale bar is 10 m. All of our data suggest that MOX is a protein of the endoplasmic reticulum. The uniform, reticular staining pattern observed for both MOX-GFP and MOX suggests that the protein does not form insoluble aggregates, a concern for any exogenous, overexpressed protein. DBM, the most homologous copper monooxygenase family member, exhibits a similar subcellular localization when expressed in AtT-20 cells (18). Expression of MOX in pEAK RAPID cells yielded similar results, indicating that localization of MOX to the ER is not cell type-specific.
MOX Is N-Glycosylated-We next utilized metabolic labeling and immunoprecipitation to determine whether MOX undergoes any post-translational modification. Multiple wells of pEAK RAPID cells transiently expressing exogenous MOX were incubated with medium containing [ 35 S]Met for 30 min and either harvested immediately or incubated with fresh medium lacking [ 35 S]Met for 30 min, 1 h, or 4 h (Fig. 7A). Medium was collected, and cell extracts were separated into a soluble fraction (Soluble) and a crude particulate fraction (Particulate). Following solubilization of the particulate fraction with detergent, MOX was immunoprecipitated and analyzed by SDS-PAGE and fluorography (Fig. 7A).
After the 30-min pulse, 69-and 72-kDa forms of MOX were identified in the soluble fraction. The molecular mass predicted for mMOX, 69 kDa, matches the smaller band in the soluble fraction. Even when the pulse-labeling period was shortened to 10 min, two forms of MOX were detected (data not shown). The particulate fraction contained only the 72-kDa form of MOX. The major change noted during the chase incubation was a decrease in the amount of labeled MOX. Although 72-kDa MOX in the particulate fraction had a half-life of ϳ4.6 h, both forms of soluble MOX had a slightly shorter half-life (3.8 h for soluble 72-kDa MOX and 2.9 h for soluble 69-kDa MOX). Medium collected from the transfected cells was analyzed together with cell extracts; consistent with the Western blot analysis shown in Fig. 5, [ 35 S]Met-labeled MOX was never detected in the spent medium (data not shown). Consistent with these different half-lives, Western blot analysis revealed a prevalence of 72-kDa MOX in the particulate fraction (Fig. 7B); the soluble fraction contained about one-third as much 72-kDa MOX and a barely detectable amount of 69-kDa MOX (Fig. 7B, Longer  Exposure, arrow).
The sequence of mMOX includes four potential N-glycosylation sites (Asn-Xaa-Ser/Thr): Asn 114 -Ser-Thr, Asn 247 -Asp-Ser, Asn 476 -Leu-Thr, and Asn 517 -Leu-Ser; all four sites are conserved in human MOX (Fig. 1). Only the first site is outside the region of the catalytic core; based on the structure of PHMcc (1), the other three sites would be located in loops connecting various ␤-strands and might be surface-accessible. To determine whether MOX is N-glycosylated, transfected pEAK RAPID cells were pre-treated with tunicamycin, a potent inhibitor of N-glycosylation (30). Pre-treatment with tunicamycin eliminated the 72-kDa form of MOX from both the soluble and particulate fractions, leaving only the 69-kDa form of MOX (Fig. 7C). More of the newly synthesized MOX was recovered from the soluble fraction following tunicamycin treatment.
Because MOX is N-glycosylated, we used endoglycosidase H to assess the maturity of its N-linked oligosaccharide chains. If MOX remains in the endoplasmic reticulum, as suggested by our pulse/chase and immunofluorescence localization studies, its N-linked oligosaccharide chains might remain endoglycosidase H-sensitive. MOX was immunoprecipitated from the soluble and particulate fractions of transiently transfected cells labeled for 30 min (Pulse) or labeled and then chased for 3 h (Chase). The immunoprecipitated MOX was digested with endoglycosidase H or buffer control before analysis by SDS-PAGE and fluorography. In both the soluble and particulate fractions, endoglycosidase H treatment completely eliminated the 72-kDa form of MOX, converting it into a 69-kDa protein (Fig. 7D). Although MOX present after the 30-min Pulse might not have had time to mature, MOX present after the 3-h chase remained endoglycosidase H-sensitive.
Exogenous MOX Forms Oligomers-All three copper monooxygenase family members have a single Cys residue in their signal or signal/anchor sequence (Fig. 1). MOX shares the 12 Cys residues that form intrachain disulfide bonds in DBM, presumably yielding a similar intrachain disulfide bonding pattern. If so, Cys 600 , located within the C-terminal hydrophobic stretch, might be predicted to form an interchain disulfide bond. The pH-dependent association of disulfide-bonded DBM dimers yields tetramers (31). To explore the quaternary structure of MOX, extracts of transfected cells were applied to linear sucrose gradients (Fig. 8A). As for the metabolic labeling experiments, extracts were separated into crude soluble and particulate fractions; proteins in the particulate fraction were solubilized by incubation with 1% Triton X-100 (TMT buffer). Aliquots of both fractions were then applied to 5-20% sucrose gradients, and MOX was analyzed by immunoprecipitation. DBM secreted from stably transfected Chinese hamster ovary cells was analyzed for comparison. As expected, secreted DBM was largely tetrameric, and preincubation with dithiothreitol disrupted these tetramers (Fig. 8B). In contrast, the behavior of MOX solubilized from the particulate fraction was unaffected by thiol-reducing reagents. Most of the MOX was recovered in large aggregates, whereas the MOX remaining in the gradient fractionated as a monomer (Fig. 8A).

DISCUSSION
The Dominant Form of Human MOX Has a Signal Sequence-MOX (hMOX; short form, GenBank TM accession number AY007239) was first identified by differential screening of senescent human fibroblasts (19). Because the predicted protein lacked a signal peptide sequence, it seemed unlikely that MOX could function in a manner similar to the other members of this copper-dependent monooxygenase family, DBM and PAM. Following the identification of hMOX, homologous genes were identified in mouse (mMOX; GenBank TM accession number BAA95089) and chicken (GenBank TM accession number AF327450, also called DBHR) (20). Based on SMART and Signal P analyses, both mMOX and chicken MOX have cleavable N-terminal signal sequences. Using the N-terminal amino acid sequence of mMOX to search the human EST data base, we found an EST (GenBank TM accession number AI751100) that encoded an amino acid sequence, which was 76% identical to that of mMOX. SMART and Signal P analyses of this sequence identify the first 19 residues as a signal sequence. Human MOX cDNA with an N-terminal signal sequence was also identified by bioinformatic methods (GenBank TM accession number AY359094) (32). Using reverse transcription-PCR, we established that the form of hMOX with an N-terminal signal sequence is by far the most prevalent form. The hMOX signal sequence is eliminated when transcription starts in the intron between exons 0 and 1.
MOX Shows a Broad Tissue Distribution-MOX and DBHL, a gene identified in the course of sequencing the mouse T-cell receptor locus (33), are new members of the copper monooxygenase family. Mouse DBHL is the most divergent family member, even lacking some highly conserved residues, and the human genome contains no DBHL gene and no human EST was identified. MOX is more similar to DBM than to PHM. However, from the tissue distribution point of view, MOX is more like PAM, with expression in a number of different tissues in the adult mouse and in human. In the adult mouse, PAM is most prevalent in pituitary, atrium, ventricle, lung, and adrenal, whereas MOX is most prevalent in the salivary gland and ovary, which express low levels of PAM. In contrast, expression of DBM and DBHL is highly restricted, with DBM transcripts detectable only in the adrenal and cerebellum/brain stem, and DBHL only in mouse thymus and testis.
The expression pattern observed for MOX early in avian development suggests that MOX plays a role in the production of important signaling molecules (20). MOX transcripts are first detected at the neural plate border (stage 7), where neural crest induction occurs, and are expressed throughout the development of the neural crest (20). MOX transcripts are also detected in the myotome. The expression of MOX during development of the mouse embryo has not yet been studied. DBM, the only enzyme that can hydroxylate the ␤-carbon of dopamine to yield norepinephrine, serves as a marker for the noradrenergic system and is first detected in avian embryos at stage 18 (4,34). PHM, the only enzyme that can hydroxylate the Cterminal Gly residue in peptidylglycine intermediates to yield amidated product peptides, serves as a marker for peptidergic systems (1,35). PAM is expressed in the cardiogenic region of the developing mouse embryo at embryonic day 9 and appears in the developing nervous system, limb mesoderm, and mesenchyme (36).
MOX Is a Luminal Protein but Is Not Secreted-Because human MOX, like mouse and chicken MOX, has an N-terminal signal sequence, its secretion and localization in the secretory pathway were studied. None of the MOX antisera generated were sensitive enough to visualize endogenous MOX, so our conclusions are based on analyzing the expression of exogenous MOX. Attempts to express exogenous MOX in chick embryos were unsuccessful (20). Although exogenous MOX exhibited the same behavior in two very different cell types (non-endocrine and endocrine), data on the properties of endogenous MOX are needed. We never observed secretion of MOX. Consistent with this, both subcellular fractionation and immunocytochemistry indicate that MOX is localized to the ER, not to the trans-Golgi network or to secretory granules. Consistent with localization to the ER, the N-linked sugar attached to MOX remains sensitive to endoglycosidase H even 5 h after synthesis. Analyzing the mMOX sequence, we noticed a hydrophobic 17-amino acid stretch near the C terminus that might serve as a signal for addition of a glycosylphosphatidylinositol tail (37) (available at www.cbs.dtu.dk/services/SignalP/). Neither appending GFP to the C terminus nor removing this hydrophobic region altered the subcellular localization of MOX or resulted in MOX secretion. In addition, MOX remained membrane-associated after the particulate fraction was digested with phosphatidylinositol phospholipase C (1-phosphatidyl-D-myo-inositol phosphohydrolase; Sigma) to release glycosylphosphatidylinositol-anchored proteins (data not shown) (38). Neither calnexin nor calreticulin, ER chaperones that bind monoglucosylated glycans (39,40), were co-immunoprecipitated with MOX, suggesting that they are not responsible for retention of MOX in the ER.
MOX Associates with Membranes-MOX does not have a recognized ER retention signal (41). We used metabolic labeling to better understand its properties, revealing the presence of 69-and 72-kDa forms of MOX. Only the 72-kDa form is found in the particulate fraction. Because the 72-kDa form of MOX is absent from cells pre-treated with tunicamycin, and endoglycosidase treatment converts the 72-kDa form into a 69-kDa form, attachment of an N-linked oligosaccharide plays a role in FIG. 8. Exogenous MOX forms oligomers. Sucrose gradients were used to determine the molecular weight of MOX before and after exposure to dithiothreitol. A, pEAK RAPID cells expressing mMOX were incubated with [ 35 S]Met for 30 min and separated into soluble and particulate fractions. The particulate fraction was solubilized using 1% Triton X-100. Aliquots were analyzed directly (Con) or after treatment with 10 mM dithiothreitol at 37°C for 30 min. MOX was localized by immunoprecipitation, SDS-PAGE, and fluorography. Data shown are for the particulate fraction. B, as a control, medium from a Chinese hamster ovary cell line stably expressing rat DBM was collected and analyzed before (Con) or after exposure to 1 mM dithiothreitol for 30 min at 37°C. DBM was detected by Western blot analysis. Marker proteins analyzed simultaneously were localized following SDS-PAGE; peak positions are indicated, along with native molecular weights. its production. The 69-kDa form of MOX is found only in the soluble fraction, and its shorter half-life means that little of it is present at steady state (Fig. 7, A and B). We saw no evidence for conversion of 69-kDa MOX into 72-kDa MOX during chase incubations. Tunicamycin treatment results in the formation of membrane-associated 69-kDa MOX, indicating that N-glycosylation is not essential to the production of particulate MOX. The facts that newly synthesized MOX is not rapidly degraded and the distribution of MOX is uniform throughout the ER suggest that the behavior of the exogenous protein may accurately represent the behavior of the endogenous protein.
Particulate fractions were washed with carbonate to release peripherally associated proteins. The behavior of MOX and PAM-1, an integral membrane protein, were indistinguishable, with almost no full-length protein recovered in the soluble fraction and little removed by the carbonate wash. Appending GFP to the C terminus of MOX resulted in the recovery of some protein in the soluble fraction, making MOX mimic the behavior of PAM-3; the properties of MOX-GFP and MOX-Arg 596 -GFP were indistinguishable. DBM also occurs in membranebound and -soluble forms in neurosecretory vesicles (3,42,43); upon exocytosis, some DBM is released, but some remains membrane-associated and is internalized (44). The signal/anchor sequence at the N terminus of DBM plays a role in this process (45)(46)(47)(48).
The Substrate for MOX May Be Hydrophobic-If MOX functions in a manner analogous to DBM and PHM, it will require the presence of copper to function properly, and it will consume both dioxygen and ascorbate as it produces product. Although DBM and PHM are well characterized enzymes, MOX has not yet been shown to have catalytic activity. Detailed structural studies of PHM make it possible to predict the structure of the active site of MOX. The five His residues and single Met residue critical for binding two copper residues (His 107 , His 108 , and His 172 for Cu A and His 242 , His 244 , and Met 314 for Cu B ) in PHM are all conserved in MOX (1). The Arg residue that binds the peptidylglycine carboxylate in PHM is replaced by a Gln in DBM, and a neighboring Glu residue is thought to form hydrogen bonds with this Gln and with dopamine (1). In MOX, hydrophobic Leu residues replace the charged or hydrophilic residues (Arg 240 and Asn 316 ) that bind the peptide substrates of PHM (1), suggesting that the MOX substrate may be hydrophobic (Table II). We attempted to measure MOX-catalyzed consumption of ascorbate using subcellular (P2) fractions prepared from control or MOX-transfected cells. Although we could easily quantify ascorbate consumption catalyzed by secreted PAM-3, the need to add a crude particulate fraction as our source of MOX made it impossible to both stabilize low levels of ascorbate and provide copper to MOX.