Cloning, sequencing, and heterologous expression of the murine peroxisomal flavoprotein, N1-acetylated polyamine oxidase.

The aminoacyl sequences of three regions of pure bovine N1-acetylated polyamine oxidase (PAO) were obtained and used to search GenBankTM. This led to the cloning and sequencing of a complete coding cDNA for murine PAO (mPAO) and the 5'-truncated coding region of the bovine pao (bpao) gene. A search of GenBankTM indicated that mpao maps to murine chromosome 7 as seven exons. The translated amino acid sequences of mpao and bpao have a -Pro-Arg-Leu peroxisomal targeting signal at the extreme C termini. A beta-alpha-beta FAD-binding motif is present in the N-terminal portion of mPAO. This and several other regions of mPAO and bPAO are highly similar to corresponding sections of other flavoprotein amine oxidases, although the overall identity of aligned sequences indicates that PAO represents a new subfamily of flavoproteins. A fragment of mpao was used as a probe to establish the relative transcription levels of this gene in various mature murine tissues and murine embryonic and breast tissues at different developmental stages. An Escherichia coli expression system has been developed for manufacturing mPAO at a reasonable level. The mPAO so produced was purified to homogeneity and characterized. It was demonstrated definitively that PAO oxidizes N1-acetylspermine to spermidine and 3-acetamidopropanal and that it also oxidizes N1-acetylspermidine to putrescine and 3-acetamidopropanal. Thus, this is the classical polyamine oxidase (EC 1.5.3.11) that is defined as the enzyme that oxidizes these N1-acetylated polyamines on the exo-side of their N4-amino groups. This enzyme is distinguishable from the plant polyamine oxidase that oxidizes spermine on the endo-side of the N4-nitrogen. It differs also from mammalian spermine oxidase that oxidizes spermine (but not N1-acetylspermine or N1-acetylspermidine) at the exo-carbon of its N4-amino group. This report provides details of the biochemical, spectral, oxidation-reduction, and steady-state kinetic properties of pure mPAO.

Although PAO has significant clinical and pharmacological relevance pertaining to cancer, ischemic tissue damage, apoptosis, etc., there is a paucity of solid data regarding the biochemical properties, mechanism of substrate oxidation, mechanism of inhibition by highly selective compounds such as N 1 ,N 4 -bis(butadienyl)-1,4-diaminobutane (MDL 72527) and N 1 -butadienyl-1,4-diaminobutane (MDL 72521) (1), or structural information for any mammalian PAO. This situation prompted us to initiate a program to create a system to produce PAO heterologously. This paper reports the cloning and the complete sequencing of murine pao (mpao), the sequencing of all but a small portion of bovine pao (bpao), and the production of active mPAO by Escherichia coli at a reasonable level. Various aspects of UV-visible biochemical, spectral, redox, and steady-state kinetic properties of the heterologously produced pure mPAO are presented below.
Analytical Procedures-Electrospray ionization mass spectrometry (ESI-MS; positive-ion mode and negative-ion mode) and elemental analyses of organic chemicals were done by HT Laboratories, San Diego, CA. One-dimensional 300 MHz (Nicolet/GE NT 300) 1 H NMR spectra were obtained from Acorn NMR, Inc., Livermore, CA. The liquid chromatography-electrospray ionization mass spectral analysis of pure mPAO was done using a Finnigan LCQ Deca XP (ion trap mass spectrometer). The work was performed at the State University of New York Health Science Center Mass Spectrometry Facility (Brooklyn, NY). Uncorrected capillary tube melting points were determined using a Misco aluminum block device.
Synthesis of the 2,4-Dinitrophenylhydrazone of 3-Acetamidopropanal-1-Acetamido-3,3-diethoxypropane, 250 mg (1.3 mmol), was mixed with 1.0 ml of 1.5 N HCl. After 2-3 min, this was added to a boiling solution of 5 ml of ethanol/0.5 ml of concentrated HCl containing 295 mg of 2,4-dintrophenylhydrazine (1.5 mmol). Heating was stopped immediately, and 10 ml of room temperature ethanol were added. In a few minutes, the 2,4-dinitrophenylhydrazone began to crystallize. After 1 h at room temperature and 1 h on ice, the solid was filtered and washed with a small volume of ice-cold ethanol. The yield of the 2,4dinitrophenylhydrazone of 3-acetamidopropanal was 335 mg (1.13 mmol, 86% yield); m.p. 157.5-158°C. 1  Purification of bPAO and Amino Acid Sequencing of Segments of This Enzyme-Bovine livers were covered with ice and transported to the laboratory within 1 h after their removal from live animals at a local slaughterhouse. Following a published procedure (34), bPAO was purified from fresh liver, or tissue that had been cut into 1-inch cubes from fresh liver and immediately frozen and stored at Ϫ80°C. About 1 mg of nearly pure bPAO was obtained from 1 kg of tissue. The enzyme was 2 SMO, referred to as PAO in Refs. 10 and 12, oxidizes SPM but not the N 1 -acetylated polyamines. By definition, PAO, designated EC 1.5.3.11, is a flavoprotein that oxidizes specifically N 1 -acetyl-SPM and N 1 -acetyl-SPD. To date, an EC number for SMO has not been assigned. To avoid further confusion concerning the identity of these two polyamine-oxidizing enzymes, the acronyms SMO and PAO taken from Ref. 11 will be used. purified further on a 10% Tris-HCl SDS "Ready Gel" (Bio-Rad), then electro-transferred onto an Immobilon-P SQ membrane (Millipore), and Coomassie Blue-stained. The membrane was submitted to the Biomolecular Resource Center (University of California, San Francisco, CA) for the N-terminal sequence analysis.
Purified bPAO was electrophoresed as before, and the Coomassie Blue-stained bPAO band was excised and subjected to an in-gel tryptic digest at the Protein Sequencing Center at the State University of New York, Brooklyn, NY. Two major internal peptides (Peptide I and Peptide II; see Fig. 2) were purified and sequenced.
Cloning and Sequencing of bpao and mpao-GenBank TM EST data bases were searched using three bPAO peptide sequences (see above). Two murine ESTs (GenBank TM accession numbers AA437705 and AI098814) were found to code for aminoacyl sequences that were ϳ85% identical to that of bPAO Peptide I (see Fig. 2). Both clones (IMAGE numbers 819909 and 1482295, respectively) were purchased from Genome Systems, Inc. (St. Louis, MO), and plasmids were isolated using a Qiagen kit and sequenced. (All DNA sequencing work was done at the Biomolecular Resource Center, University of California, San Francisco, CA). The AA437705 cDNA fragment (mpao1) was released from its plasmid by an XbaI/SalI digestion. This fragment was the template for generating a mixture of [ 32 P]dATP-labeled probes, by using a randomprimed DNA labeling kit (Roche Applied Science) (35). The 32 P-labeled probes were used to screen a gt 10-mouse 17-day embryo cDNA library (Clontech) and a Uni-ZAP XR bovine liver cDNA library (Stratagene). Seventeen positive phage plaques from the bovine cDNA library and 16 positive plaques from the mouse cDNA library were obtained and rescreened. The final positive clones, containing different length mpao and bpao cDNA inserts, were isolated and confirmed by standard Southern blotting (35) using the same 32 P-labeled probes. The largest mpao fragments from these clones were sequenced. A fragment of bpao cDNA was excised from the Uni-ZP XR vector of the pBluescript phagemid. This 1.6-kb fragment (bpao1) was sequenced using flanking primers (T3/T7), and it coded for all but a small section of the Nterminal portion of bPAO. The 5Ј-end of the fragment coded for a sequence that matched exactly the C-terminal portion of Peptide I (see Fig. 2). A region near the 3Ј-end coded for a protein sequence identical to that of Peptide II, which was determined later to be the C terminus of the enzyme (see Fig. 2). The high similarity between the sequences of bpao1 and mpao1 confirmed that the mpao1 screening probe codes for a portion of mPAO.
DNA isolated from one plaque of the mouse cDNA library was cloned into the SalI sites of pUC19 to give the plasmid, pUC19_MPAO1, that was used for double-stranded sequencing. The sequence of this fragment (mpao2) was missing the 5Ј-end of the complete mpao sequence (see under "Results"). A 5Ј-extension (see Fig. 2) was obtained using the 5Ј-rapid amplification of cDNA ends PCR method with mouse 17-day embryo Marathon Ready cDNA (Clontech) as the template and using a SMART TM PCR cDNA Synthesis kit (Clontech) for the PCRs. The mpao gene-specific antisense primer, mpao1R (5Ј-GTTCTCTTCCGATAAT-TCTTTCTCC-3Ј), spans nucleotides 333 to 309 of mpao (see Fig. 2), and the Clontech AP1 universal sense adaptor primer was specific for the Marathon Ready cDNA; 5 cycles for 30 s at 94°C and 3 min at 72°C, 5 cycles for 30 s at 94°C and 3 min at 70°C, and 30 cycles for 20 s at 94°C and 3 min at 68°C. By using a 50-fold dilution of the resulting PCR product as template, and AP2 (an AP1-nested primer) and mpao1R as primers, a second PCR was carried out under the same conditions. The resulting cDNA fragment, about 400 bp long, was isolated from a 1% agarose gel using a Geneclean kit (Bio 101). Melded with mpao2, the sequenced PCR product provided a 1.7-kb section of mpao that coded for full-length mPAO.
Measuring the Relative Levels of mpao mRNA in Different Murine Tissues-Murine Rapid-Scan Gene Expression Panels were purchased from OriGene Technologies, Inc. Two gene-specific primers were designed according to the manufacturer's instruction. The sense primer, mpao2F (5Ј-TCGGAAGAGAACCAGCTTGTGG-3Ј, 22-mer), and the antisense primer, mpao2R (5Ј-CAATGACATGATGTGCAGGCA-3Ј, 22mer), generated a 570-bp-long mpao cDNA fragment by PCR. The 24 mouse cDNA samples, serially diluted over a 4-log range (ϫ1000, 100, 100, and 1) by the manufacturer, were arrayed into a 96-well PCR plate. The first step of the PCRs were carried out at 94°C for 3 min, which was followed by 35 cycles: 94°C for 30 s, 55°C for 1 min, and 72°C for 2 min. The control-primer pair for detection of ␤-actin cDNA, provided by the manufacturer, was used for a PCR that was carried out as just described with 25 cycles. The amplified fragments were electrophoresed on a 1% agarose gel and ethidium bromide-stained to provided a measure of mpao mRNA in each tissue.
Expression of mpao in E. coli-The pET 29 c(ϩ) vector (Novagen) was used to construct a mpao prokaryotic expression system, and E. coli DH5␣ was used for plasmid subcloning. First, a 5Ј-end fragment was generated by PCR using mpao1 as the template for the gene-specific antisense primer mpao1R (see above), and a sense primer, mpao1F, which contains SacI and NdeI sites and an ATG start codon (5Ј-GCG-AGCTCATACATATGGCGTTCCCTGGCCCGCGG-3Ј). The underlined regions indicate SacI and NdeI sites, respectively. A SacI/BamHI fragment of the PCR product was subcloned into pUC19-MPAO1 to give pUC19-MPAO. This construct contained the entire mpao gene. Next, the full-length mpao cDNA was ligated into NdeI and HindIII sites of pET 29c to give a plasmid denoted pET-MPAO. E. coli BL21 GOLD (DE3) (Invitrogen) was transformed with this plasmid for mPAO production.
Growth of Transformed Bacteria-A culture of the E. coli transformant was grown on Luria-Bertani (LB) agar plates containing 30 g/ml kanamycin. A single positive colony was inoculated into 3 ml of LB broth containing 30 g/ml kanamycin (LB-kan) for overnight growth at 37°C. This culture (500 l) was transferred to 80 ml of fresh LB-kan medium for overnight growth. Five milliliters of this culture were transferred to each of five 2-liter flasks containing 1 liter of fresh LB-kan medium for overnight growth at 37°C, with shaking. Each flask was added to one of five 14-liter New Brunswick FS-614 fermentors containing 12 liters of LB-kan media. The cell culture was incubated at 30°C with rapid stirring and vigorous aeration. When the A 600 of the culture reached 0.6 -0.7, isopropyl thio-␤-D-galactoside was added (final concentration, 50 M). Growth was continued until the A 600 reached 1.5-2.0. About 260 g of centrifuged cell paste were obtained from the 60 liters of growth media. The paste was stored at Ϫ80°C.
Purification of Heterologously Produced mPAO-Selected fractions for the various steps in the purification were assayed for N 1 -acetyl-SPM oxidase activities by a published method (36). This assay measured the time-dependent formation of H 2 O 2 (Fig. 1). The assay stock solutions are as follows: (A) 100 mM vanillic acid, pH 7.0 with KOH; (B) 50 mM 4-aminopyrine; (C) 400 units/ml horseradish peroxidase; (D) 50 mM N 1 -acetyl-SPM; (E) 100 mM glycine/KOH, pH 9.5, the pH for maximal activity (37). Thirty microliters each of solutions A-D were mixed with 2.88 ml of solution E, and 50 l of the resulting solution were pipetted into individual wells of a 96-well plate. Anywhere from 1 to 50 l of a particular fraction was added to a well. The relative activities of different fractions were assessed visually from the time-dependent intensity change of the developing pink color. The purity of various fractions were assessed by SDS-PAGE using pre-cast 10 -20% Tris-HCl "Ready Gels" (Bio-Rad), following the manufacturer's instructions.
Unless noted otherwise, all purification steps were carried out at 4°C. Frozen E. coli cell paste (260 g) was thawed in a beaker with 10 mM MOPS buffer, pH 7.25. (The pH was adjusted at 21°C; the estimated pH at 4°C is 7.35). The 800-ml suspension was homogenized with a large glass/Teflon piston (Potter/Elvehjem) tissue grinder, and then passed twice through an Avestin Emulsiflex C5 Homogenizer at 15-20,000 pounds/square inch. Next, 15 mg of solid FAD were dissolved in the suspension, and it was centrifuged (50,000 ϫ g, for 30 min). The supernatant was dialyzed in the dark against 13 liters of 10 mM MOPS buffer, pH 7.25, for 4 h and then overnight against 13 liters of fresh buffer. The resulting solution was diluted to 2 liters with this buffer and applied to a 14 ϫ 25-cm DEAE-cellulose (Whatman, DE53) column with a flow rate of 20 ml/min. The column was then washed with 2 liters of the buffer and eluted with an 8-liter gradient from 0 to 400 mM KCl in the 10 mM MOPS, pH 7.25 buffer. Activity eluted from 4.7-7.8 liters after the gradient was initiated. The 3.1-liter volume was reduced to ϳ500 ml using 350 ml of Amicon pressure concentrators fitted with Amicon YM-10 membranes. After dissolving 15 mg of FAD, the resulting solution was dialyzed in the dark for 4 h against 13 liters of 10 mM HEPES buffer, pH 7.8 (pH adjusted at 21°C; estimated pH at 4°C was 8.05), and then overnight against 13 liters of fresh buffer. The resulting sample was applied to a 5 ϫ 39-cm DEAE-Spherodex LS column (100 -300 m bead size; Ciphergen) already equilibrated with the 10 mM HEPES buffer. The column was washed with 500 ml of this buffer before starting a 2.4-liter gradient from 0 to 500 mM KCl in the same buffer. The column, with a 7-foot pressure head, was run at the maximum flow rate. Once the gradient was started, 26-ml fractions were collected. The majority of the activity eluted in tubes 82-108, which were combined (ϳ700 ml) and concentrated to ϳ50 ml as described earlier. This solution was dialyzed for 4 h, against 7 liters of 10 mM KH 2 PO 4 /KOH buffer, pH 7.2, and then overnight against 7 liters of fresh buffer.
The sample was chromatographed on a Mono P HR 5/20 column (Amersham Biosciences) at room temperature. After injecting 2 ml of the sample at a flow rate of 1 ml/min with solution I (H 2 O), proteins were eluted with the following gradient: 0 -1% II (1 M KH 2 PO 4 /KOH, pH 7.2) in 4 min; 1-30% II in 125 min. mPAO, eluting from 38 to 41 min, was collected as a single fraction and immediately put on ice. This procedure was repeated until the entire sample had been processed. The mPAO fractions from all of the Mono P runs were combined, then concentrated, and washed into 1 mM KH 2 PO 4 /KOH buffer, pH 7.2, using 2-ml Centricon-10 centrifuge concentrators (Amicon). The final volume was 2 ml in the 1 mM buffer.
This sample was chromatographed on a 1 ϫ 10-cm ceramic hydroxyapatite (Bio-Rad, type II) column (Amersham Biosciences HR 10/10 column), at room temperature. The mPAO sample (100 l), diluted to 1 ml with H 2 O, was injected immediately onto the hydroxyapatite column with a flow rate of 2 ml/min. The elution was carried out as follows: 0% solution II for 7 min; 0 -1% II in 2 min; hold at 1% II for 10 min. mPAO eluted as a broad peak from 14 to 17 min. This step was repeated until the entire sample had been processed. The combined fractions were concentrated as for the Mono P fraction. The solution was washed into 10 mM KH 2 PO 4 /KOH buffer, pH 7.2, to give a solution that was 3.68 mg/ml mPAO (based on an ⑀ 458 ϭ 10,400 M Ϫ1 cm Ϫ1 and a molecular mass ϭ 56,101 Da for the enzyme; see under "Results"). The enzyme was judged pure by SDS-PAGE, by ion-exchange chromatography on an analytical TSK DEAE 2SW column (0.4 ϫ 25 cm; a 0.75 ml/min flow rate, with a gradient from 1 to 50% solution II in 30 min; a single sharp peak eluted at 23 min), and gel filtration chromatography on a TSK 3000SW column (0.7 ϫ 30 cm; 0.5 ml/min flow rate; 250 mM KH 2 PO 4 / KOH buffer, pH 7.2). The purity and integrity of the protein was confirmed also by the electrospray ionization (ESI) mass spectral analysis, which provided a peak of 55,311 Ϯ 6 mass units (the mass of apo-mPAO based on the sequence is 55,316 mass units). The yield of pure mPAO was 36.8 mg.
By using the conditions for the steady-state kinetic assay described below, it was found that the enzyme, at 2-4 mg/ml, was stable when frozen at Ϫ20 or Ϫ80°C and thawed through several cycles. However, at a concentration of 30 g/ml, activity was lost quickly after several freeze/thaw cycles, with more rapid loss occurring at Ϫ80°C. When 33% (v/v) ethylene glycol was added, mPAO was stable for several of freeze/ thaw cycles for solutions containing 20 g/ml to 4 mg/ml, regardless of the storage temperature. It was decided to store the enzyme at Ϫ20°C in the presence of 33% (v/v) ethylene glycol. The enzyme maintains full activity and measured biochemical, redox, M r , and kinetic properties after 16 months of storage under these conditions. Ethylene glycol removal and buffer exchange were accomplished easily by several concentration/dilution cycles using Centricon-10 centrifuge concentrators.
Binding of FAD to mPAO-The spectrum of a 0.1 mg/ml (0.8 ml) solution of mPAO in 10 mM KH 2 PO 4 /KOH buffer, pH 7.2, indicated that the sample contained 7.4 nmol of FAD. This solution was treated with 80 l of 55% trichloroacetic acid (38) and centrifuged to give a clear yellow supernatant and a white pellet. Overnight incubation of the isolated solution at room temperature in the dark resulted in the conversion of the liberated FAD to FMN. Fluorescence analysis with a Hatachi F-4010 fluorescence spectrophotometer (450 nm excitation, 525 nm emission; reference-authentic FMN) indicated that FAD was released quantitatively from mPAO. Thus, FAD is noncovalently bound to mPAO.
Spectral Characterization and Redox Properties of mPAO-All UVvisible spectra were recorded with a Hewlett-Packard 8452A diode array spectrophotometer. mPAO, in 50 mM KH 2 PO 4 /KOH buffer, pH 7.6, at 25°C, was titrated anaerobically with a solution of sodium dithionite. This solution was standardized by using it to titrate anaerobically a FAD solution of known concentration. The anaerobic cuvette and other details of this procedure are described elsewhere (39 -41). The anaerobic mPAO solution contained also 50 mM D-glucose, 3 g of catalase, and 50 g of glucose oxidase to scavenge trace dissolved O 2 . The spectral data were subjected to "Factor Analysis" using the SPECFIT program (Spectrum Software Associates, Chapel Hill, NC) as described earlier (40,41).
A 1.20 M solution of mPAO was titrated anaerobically with a solution of 0.5 mM N 1 -acetyl-SPD in 50 mM KH 2 PO 4 /KOH buffer, pH 7.6, at 25°C. The enzyme and substrate solutions contained 50 mM D-glucose, 3 g of catalase, and 50 g of glucose oxidase.
Steady-state Kinetic Experiments-Spectrophotometric assays were done at 30°C in 50 mM KH 2 PO 4 /KOH buffer, pH 7.6, following a published procedure (36). This method provided a continuous monitor of the H 2 O 2 produced in the reactions. The assays were done in 1-ml, 1 cm-path length cuvette with 0.8 ml of solution containing varying amounts of substrate, 0.1-0.2 g of mPAO, 1 mM vanillic acid, 0.5 mM 4-aminopyrine, and 4 units of horseradish peroxidase. By varying each of the last three components, it was found that none were inhibitory. The reactions were monitored at 498 nm with a UVIKON 840 spectrophotometer (Kontron Instruments) for the formation of the quinoneimine dye (⑀ ϭ 4650 M Ϫ1 cm Ϫ1 at pH 7.6; see Ref. 36), the condensation product of vanillic acid and oxidized 4-aminopyrine. The latter was produced from 4-aminopyrine by its interaction with horseradish peroxidase that had been oxidized by H 2 O 2 (36). Assays were done by varying the concentration of the amine substrate, whereas the dissolved [O 2 ] was constant at the air-saturated level (237 M) in the buffer at 30°C. The data were fit by nonlinear regression (42) to the appropriate steady-state kinetic equations.
The value for the apparent dissociation constant, K D (ϭ K I ), for each inhibitor was estimated by measuring the rates of N 1 -acetyl-SPM oxidation as its concentrations and that of the inhibitor were varied. It was assumed that these substances, which are either very poor or nonsubstrates, are competitive inhibitors for the oxidation of the substrate. These data were analyzed also by nonlinear regression using the appropriate equation.
Steady-state kinetic assays were done also by varying the [N 1 -acetyl-SPM] in buffer saturated with pure O 2 (1.12 mM). After several minutes of bubbling a cuvette solution with a stream of pure O 2 gas, a small aliquot of substrate was added followed by the enzyme. Once the enzyme was added, the bubbling was terminated, and the absorbance change was recorded.
Some ( -acetyl-SPM and N 1 -acetyl-SPD were determined by progress curve analyses of reactions that were allowed to go to completion (dissolved [O 2 ] ϭ 0 at t ϭ ϱ). By using a published procedure (43), the data were fitted to the integrated Michaelis-Menten equation. The analyses were done using Maple VI (Windows 2000) software (Waterloo Maple, Inc.) running on a PC computer. By using K I values (see Table  I) as a gauge, the saturating concentrations of N 1 -acetyl-SPM and N 1 -acetyl-SPD were made high enough (3.7 mM) so that product inhibition by SPD or PUT, respectively, was insignificant at all times during the reaction (see Table I). Inhibition by the H 2 O 2 , formed as a product of polyamine oxidation by mPAO, was assessed by addition of 2 l of 30 mg/ml (30,000 units/mg) solutions of catalase before a reaction was started. The catalase converted immediately each mole of H 2 O 2 formed to 0.5 mol of dissolved O 2 . After correcting the data by a factor of 2, the rate of O 2 consumption was the same as in the absence of catalase. This indicated a lack of inhibition by the H 2 O 2 formed in the assays lacking catalase.
Analyses of the Aldehyde Produced When N 1 -Acetyl-SPM and N 1 -Acetyl-SPD Are Oxidized by mPAO-mPAO, 29 g, was dissolved in 2 ml of 50 mM KH 2 PO 4 /KOH buffer, pH 7.6, containing 0.8 mM N 1 -acetyl-SPM and 30 g of catalase (30 units). An identical solution containing 0.8 mM N 1 -acetyl-SPD in the place of N 1 -acetyl-SPM was also prepared. Each solution was stirred at room temperature for 2 h. After dansylating a small aliquot, HPLC analysis (see below) indicated that the substrates had been oxidized completely for both solutions. Each solution (100 l) was mixed separately with 100 l of a 2,4-dinitrophenylhydrazine solution (100 mg in 94 ml of ethanol/6 ml of concentrated HCl). Each of the resulting solutions (25 l) was injected onto a Prodigy octadecylsilyl silica gel HPLC column (5 m particle size, 0.46 ϫ 5.0 cm; Phenomenex): flow rate, 1 ml/min; gradient elution 0% B for 0.5 min, 0 -35% B from 0.5 to 1.5 min, hold at 35% B from 1.5 to 5.0 min, 35 to 100% B from 5.0 to 9.0 min; solutions A and B were H 2 O and acetonitrile, respectively, both containing 0.5% (v/v) trifluoroacetic acid. The HPLC system used SpetraSYSTEM P2000 gradient pumps, a UV6000LP Diode Array Detector, and the ThermoQuest ChromQuest Chromatography Data System (Thermo Separation Products). The 368-nm chromatograms were used for quantitative analyses.
Authentic 3-acetamidopropanal, the expected product of mPAO oxidation of N 1 -acetyl-SPM and N 1 -acetyl-SPD, was generated by treating 10 l (ϳ10 mg, 53 mol) of 1-acetamido-3,3-diethoxypropane with 100 l of 1.5 N HCl for 1 min and then diluting immediately to 0.8 mM with 50 mM KH 2 PO 4 /KOH buffer, pH 7.6. A 0.8 mM solution of 3-aminopropanal was produced similarly after treating 10 l (9.1 mg, 62 mol) of 1-amino-3,3-diethoxypropane with 100 l of 1.5 N HCl for 1 min. A 0.8 mM solution of commercial acrolein was also prepared. Each of these solutions was mixed immediately, 1:1 (v/v), with the 2,4-dintrophenylhydrazine reagent and analyzed by HPLC as described in the previous paragraph. Another stock solution containing 0.8 mM of all three alde-hydes was treated and analyzed in the same manner. These analyses provided the retention times and reference peak areas for unreacted 2,4-dinitrophenylhydrazine and the 2,4-dinitrophenylhydazones of each aldehyde (see Fig. 6).
The following experiment was carried out in order to obtain an analytical amount of the aldehyde produced by mPAO oxidation of N 1 -acetyl-SPM. A 10-ml solution of 50 mM KH 2 PO 4 /KOH buffer, pH 7.5, containing 5.0 mM N 1 -acetyl-SPM (50 mol or 17.7 mg total), 7.25 g/ml mPAO, and 7.5 g/ml catalase was stirred at room temperature. The reaction was monitored by treating 10 l of the solution with dansyl chloride (see below) and analyzing by TLC; silica gel plates using cyclohexane/ethyl acetate, 2:3 (v/v); R f values were 0.04 for N 1 -acetyl-SPM and 0.77 for SPD. After 23 h, there remained no N 1 -acetyl-SPM. The solution was filtered using Centricon-10 centrifuge concentrators (Millipore), and 1.25 ml of concentrated HCl was added with stirring. Next, 23 mg (115 mol) of 2,4-dinitrophenylhydrazine in 0.625 ml of tetrahydrofuran were added to the filtrate. After stirring for 1-2 min, the solution became slightly cloudy. The mixture was put on ice for 2 h and filtered, and the solid material was washed with 1-2 ml of ice-cold 1. A literature procedure for dansylation with dansyl chloride and sample preparation was used with minor modifications (44). To 50 l of an unknown or reference solution, in a 1.5-ml screw-cap plastic vial, was added 200 l of a saturated Na 2 CO 3 solution and 200 l of a 10 mg/ml dansyl chloride solution in acetone. Each sample was vortex mixed for 20 s before incubation at 65°C for 10 min. After cooling on ice for several minutes, 100 l of a proline solution (250 mg/ml) was added, and the sample was vortex mixed for 10 s. The phases were separated by centrifugation, and the upper organic phase (ϳ350 l) was removed. Ten microliters of this phase were injected onto a Prodigy HPLC column (octadecylsilyl silica gel, 5-m particle size, 0.46 ϫ 5.0 cm), using a flow rate of 1 ml/min, and the following elution gradient: 0 -45% B from 0 to 0.1 min, 45-80% B 0.1 to 8 min, hold at 80% B from 8 to 11 min, 80 -90% B from 11 to 12 min. Detection was accomplished with a Gilson Spectra/ Glo fluorescence detector using a 7-51X excitation filter (330 -400 nm) and a 3-72M emission filter (460 -600 nm). The retention times (min) are (data not shown) as follows:

Cloning and Sequencing of mpao and Sequence Analysis-
With the hope of studying the properties of a mammalian PAO, this enzyme was purified from bovine liver following a published protocol (34,45). Although it was reported that ϳ20 mg of PAO could be obtained from 1 kg of bovine liver, we obtained ϳ1 mg from this mass of tissue in two separate attempts. In order to carry out careful biochemical and kinetic studies, larger amounts of PAO were required. Therefore, we decided to clone and sequence the gene for a mammalian PAO and attempt to produce the enzyme in a heterologous system. Because a mammalian peroxisomal pao gene was unavailable, in order to clone such a gene, amino acid sequence information was needed. The amino acid analysis of purified bPAO provided the N-terminal sequence, EAEAPGRGPRVLVVGGGIAGL. The underlined segment identifies bPAO as a member of family of FAD-containing oxidases (31) that includes human MAO-A, human MAO-B, corn (maize) polyamine oxidase (cPAO), and an N 1 -acetyl-SPM oxidase from C. boidinii (Cb Ac-SMO) (22,23,26). The GXGXXG sequence is a flavoprotein fingerprint. This motif is present in members the GR 1 (glutathione reductase) and GR 2 family of flavoproteins, which have known structures (32). MAO-A, MAO-B, cPAO, and several non-amine oxidizing enzymes are known to be of the GR 2 structural type.
The sequencing of two tryptic bPAO peptides resulted in the internal protein sequences SEHSFGGVVEVGAHWIHGPS (Peptide I) and LMTLWDPQAQWPEPR (Peptide II). The underlined segment of Peptide I aligns with the end of the FADcontaining enzyme superfamily motif near the N termini of these proteins (31).
Translated EST GenBank TM sequences were screened using these bPAO sequences, and two mouse GenBank TM EST sequences (accession numbers AA437705 and AI098814) containing a translated sequence very similar to Peptide I (85% identity) were identified. The plasmid DNA for AA437705 and AI098814 were purified and sequenced. Both clones were truncated at the 5Ј-end. The DNA from EST clone AA437705, with a longer mpao insert, was subjected to restriction enzyme digestions and purified. A 968-bp segment (mpao1) was excised from this clone and sequenced. The mpao1 cDNA fragment was used as a library screening probe ("Experimental Procedures"). This process allowed us to clone a nearly complete mpao cDNA (mpao2; 1710 bp) from the mouse embryo library. The sequence of the full-length mpao (1770 bp; Fig. 2) was obtained by combining the sequence information obtained for mpao2 and from a 5Ј-rapid amplification of cDNA ends PCR procedure. However, we failed to obtain the missing 5Ј-end of bpao. A search of GenBank TM revealed that mpao maps to murine chromosome 7 (cytogenic position 7F4) as 7 exons (GenBank TM accession number NW_000335).
During the review of the current work, a paper by Vujcic et al. (46) appeared that reported the cDNA sequences for mPAO and hPAO. These sequences were found by BLAST searching GenBank TM using the SMO cDNA sequence. Although the details are not provided herein, we have also cloned and sequenced the hpao gene and submitted its sequence in Gen-Bank TM (accession number AF312698) on October 11, 2000. Vujcic et al. (46) made no attempt to sequence the mpao and hpao cDNA from the clones that they obtained commercially. Compared with our cDNA sequences, their sequences differ at numerous positions. We are confident that our bPAO, mPAO, and hPAO sequences are correct because we sequenced each at least twice; ambiguous regions were sequenced three or four times.
Vujcic et al. (46) reported also the transient transfection of HEK-293 cultured human kidney cells with mpao and hpao cDNA. Although these cells expressed the enzymes from the transfected genes, no effort was made to purify the proteins for biochemical characterization (46).

A Comparison of mPAO and bPAO with Each Other and with Other Known Amine Oxidases, and the Identification of These as Peroxisomal Proteins-
The full-length mpao cDNA (1770 bp) (GenBank TM accession number AF226656) and deduced mPAO amino acid sequences are presented in Fig. 2. Upstream from the ATG start codon, there is a single TAA stop codon. No other possible translational start sites were found upstream from this ATG codon. The mpao coding region terminates with a single TGA stop codon. The gene contains a 1512-bp open reading frame that encodes for 504 amino acids. The mature apoprotein (minus the N-terminal Met) has a mass of 55,316 Da. The incompleted bpao nucleotide (1625 bp) and deduced amino acid (452 amino acids) sequences can be found at Gen-Bank TM (accession number AF226658). Its 5Ј-end nucleotide sequence is missing. The coding region of bpao terminates with a single TGA stop codon. Both mpao and bpao have an ATAAA sequence as polyadenylation signals that are near to the poly(A) tails.
An inspection of the bPAO and mPAO sequences (Fig. 3) indicates the presence of the peroxisomal targeting signal sequence, -PRL, at the C termini of these proteins. This consensus sequence -(S/A/C/P)-(K/H/R)-(I/L/M) (47), which is not cleaved after protein import into the peroxisome, is seen also in the Cb Ac-SMO and MAO-N sequences (i.e. -SKL and -ARL, respectively) (Fig. 3), indicating that these enzymes reside also in the peroxisomes of the respective host yeast cells. Like many oxidases, PAO is localized in the peroxisomes where its oxidation product H 2 O 2 can be degraded by catalase.
A ClustalW (version 1.8) alignment of many (but not all) known flavoprotein amine oxidase amino acid sequences is provided in Fig. 3. Among these sequences, there are two regions of high similarity. One, near the N termini, is clearly a ␤␣␤ consensus domain that interacts with the ADP moiety of FAD (32,33,49). The second conserved region, near the C termini, is involved also in FAD binding. The C-terminal region contains a conserved region that harbors the Cys residues that are covalently linked to FAD in the human monoamine oxidases: Cys 406 (MAO-A) and Cys 397 (MAO-B) (24,25). For bPAO and mPAO, a Ser (Ser 429 of mPAO) aligns with these Cys residues. We have determined that mPAO bind FAD noncovalently (see under "Experimental Procedures").
The Distribution of mpao mRNA in Murine Tissues-The availability of mpao cDNA allows, for the first time, the determination of the transcription level of this gene in mammalian tissues. We probed PCR-amplified mRNA of murine tissue from numerous organs and murine tissues at different developmental stages (see under "Experimental Procedures"). mpao mRNA is detected in all the murine tissues tested (Fig. 4), with the liver and stomach having the highest levels. This is in accord with an earlier finding of large levels of mPAO in the liver of various mammals (34,37,50,51). Lesser but significant levels of mpao were detected in heart, spleen, thymus, small intestine, muscle, pancreas, uterus, and breast at various developmental stages. Relatively lower levels of mpao mRNA are expressed in brain, kidney, lung, testis, skin, adrenal gland, and prostate gland. The 100ϫ panel (Fig. 4) clearly shows that the mRNA level increases during embryonic development; there is a gradual increase in the tissues on going from 8.5-to 19-day embryos. mRNA levels change also with breast development. Fig. 4 shows that level of mpao mRNA is very low in the virgin breasts, is quite high in the pregnant breasts, but is decreased in lactating and involuting breasts. These findings for breast and embryo were confirmed by repeating this analysis with a murine multiple tissue panel from a different lot (OriGene). Apparently, high mpao transcription and presumably translation is important in tissue growth and development (1). These data suggest that mpao expressions are regulated by growth hormones. It has been proposed that PAO, via its participation in the polyamine interconversion pathway, is an important regulator for maintaining cellular polyamine and tissue homeostasis.
Currently, we do not know how these mRNA levels relate to the amount of the mPAO protein in different murine tissues. However, a study measuring PAO enzyme activity in various rat tissues has been reported (1). High activity was found in the pancreas and liver. Lower but significant activity was seen in spleen, kidney, small intestines, testes, prostate tissue, thymus, brain, heart, and lung; and very low activity was observed in skeletal muscle. It was found that PAO activity increased from a low level at birth to quite high levels at 70 days postnatal and beyond in rat brain and liver.
It is important to note that Vujcic et al. (46) reported the relative levels of the transcripts for pao and smo in numerous normal and neoplastic human tissues. For the few normal tissues probed, the agreement with our findings for murine tissues is fair, except that no pao mRNA was detected in normal human spleen. The overlined segment represents the 5Ј-segment that was missing from the original truncated mpao1 clone. This sequence was obtained by using the 5Ј-rapid amplification of cDNA ends PCR method. The double underlined nucleotide sequence denotes the region corresponding to the antisense primer (bases 333 to 309) used for the PCR 5Јextension experiment. The underlined portions of the amino acid sequence correspond to the regions of bPAO that were sequenced by the Edman degradation method, and the asterisk denotes the stop codon. GenBank TM accession number L38858), and C. boidinii N 1 -acetyl-SPD oxidase (GenBank TM accession number AB018223) (22). The question marks for the bPAO sequence indicate a region of unknown composition. The composition of the segment preceding this region was obtained by protein sequencing, whereas the sequence following this region was deduced from the translated cDNA sequence. The cPAO sequence is the only one with a recognizable N terminus transport signal sequence, which is underlined. At the C termini of bPAO, mPAO, MAO-N, and Cb Ac-SMO, the tripeptide peroxisomal transport signals are indicated by asterisks. The position of Cys residues that are covalently linked to the FAD in MAO-A, MAO-B (24,25), and fMAO are indicated by the ‡ symbol. The ϭ symbols above the sequences indicate regions that are highly conserved in this alignment. For example, the regions labeled Beta-1, Alpha, and Beta-2 are components of the ␤ 1 ␣␤ 2 motif near the N termini that interacts with the ADP portion of FAD. The positive end of the ␣-helix of this motif interacts with the diphosphoryl group of the ADP moiety. The regions labeled Fl or Flx are in the flavin-binding domains of MAO-B and cPAO (26,33). These regions constitute elements of the Rossmann fold. The helix that has its positive end interacting with the N1/C2/C2-O locus of FAD is labeled Flx (near the C termini) (33). The regions labeled Sub are conserved regions in the substrate-binding domain, which are remote from the FAD and seemingly remote from the substrate/inhibitor-binding site (26,33). The extended C-terminal regions of MAO-A, MAO-B, and fMAO anchor these proteins to the outer surface of mitochondrion (33).
Heterologous Production of mPAO by E. coli and Characterization of the Pure Enzyme-The cloned mpao gene was heterologously expressed in E. coli, and active mPAO was purified to homogeneity and characterized. PAGE, gel filtration, ion exchange chromatography, and mass spectral analysis indicated that the highly purified, homogeneous enzyme is a monomer of the expected molecular mass. Fig. 5 displays the UV-visible spectrum of the pure oxidized enzyme. The max values (and relative absorbances) for the protein are 274 (1.0), 377 (0.09), and 456 (0.11). The calculated ⑀ 274 ϭ 66,000 M Ϫ1 cm Ϫ1 for the protein component, based on the amino acid composition (52), and an estimated ⑀ 274 ϭ 26,000 M Ϫ1 cm Ϫ1 for the bound FAD (assuming 1 mol of FAD/mol of protein) (53) provide an ⑀ 274 of 89,000 M Ϫ1 cm Ϫ1 . By using this value, and assuming that the ⑀ 458 is the same as free FAD, i.e. 11,300 M Ϫ1 cm Ϫ1 , the estimated molar ratio of protein to FAD is 1:0.90. This supports the contention of 1 mol of noncovalently bound FAD/mol of enzyme. The pI and the molecular mass values for apo-mPAO (minus the N-terminal Met), calculated from the amino acid composition, are 4.84 and 55,316 Da, respectively, whereas the calculated molecular mass of the holoenzyme (FAD-containing) is 56,101 Da. (Because of the negative charges of the phosphate groups of FAD, the pI value is expected to be somewhat lower than 4.84.) ESI mass spectral analysis of purified mPAO gave a molecular mass ϭ 55,311 Ϯ 6 mass units, in perfect agreement with that deduced from the cDNA-translated protein sequence.
mPAO was titrated anaerobically with a standardized solution of sodium dithionite (Fig. 5). A significant amount of the one-electron reduced flavin radical formed in the initial phase FIG. 4. Agarose electrophoresis of "Rapid-Scan Gene Expression Panel" PCR-amplified mpao cDNA samples for 24 major mouse tissues and developmental stages. The right frame presents the results for the PCR-amplified cDNA of a 540-bp portion of the ␤-actin gene for each tissue, which is the control. The middle frame presents the results for the PCR-amplified cDNA of a 570-bp portion of mpao for each tissue using a high level of first-strand cDNA for the PCR (the 100ϫ panel). The right frame displays the results for the panel using 100ϫ lower first-strand murine cDNA (the 1ϫ panel). The total mRNA of each tissue was subjected to oligo(dT) selection, and the first-strand cDNA used for the PCRs for each tissue were generated from the poly(A ϩ ) mRNA using oligo(dT) primers and Moloney murine leukemia virus-reverse transcriptase. The amplified fragments were electrophoresed on an agarose gel, and the intensity of the ethidium bromide-stained bands provided a measure of the level of mpao mRNA in each tissue.
FIG. 5. The anaerobic dithionite titration of pure mPAO. The titration was done in a 1-ml, 1-cm path anaerobic cuvette, in 50 mM KH 2 PO 4 /KOH buffer, pH 7.6, at 21°C. The concentration of the standardized sodium dithionite solution was 0.541 mM. A shows the spectrum of the oxidized enzyme (---), those obtained at the beginning of the titration (solid lines; 2.16 and 4.33 nmol of dithionite added), and that of fully reduced enzyme (---; 17.3 nmol dithionite added). The arrows indicate the direction of the absorbance changes that occurred as more dithionite was added. In the 380-nm region, the increase in absorbance indicates the formation of the red radical, whereas the small increase in the 550 -700 nm region indicates the formation of a small amount of the blue radical (54). B displays the spectral changes that occurred in the latter phase of the titration. The arrows indicate the direction of the absorbance changes that took place as progressively more dithionite was added: 4.33, 6.49, 8.66, 10.8, 13.0, 15.1, and 17.3 nmol. Although impossible to see in this reproduction, the absorbance in the 550 -700-nm region increased slightly and then decreased during this phase of the titration. The inset to B shows a graph of A 377 , A 458 , and A 590 versus the amount of dithionite added. From this plot, it was determined that 15.2 nmol of dithionite were required to fully reduce the enzyme sample. C displays the spectra of the fully oxidized (---), the radical (solid line), and the fully reduced (solid line) forms of FAD bound to mPAO that resulted from the factor analysis of the titration data presented in A and B.
of the titration and disappeared in the final phase as it converted to the two-electron fully reduced form of bound FAD. The intermediate one electron-reduced species was predominantly the anionic (so-called "red") radical, but a trace of the neutral (so-called "blue") radical was evident also by the low absorbance in the 500 -650 nm region of the spectrum (Fig. 5) (54). From this titration, the ⑀ 458 for the bound FAD was found to be 10,600 M Ϫ1 cm Ϫ1 , whereas the ⑀ 274 was determined to be 99,200 M Ϫ1 cm Ϫ1 . mPAO was titrated anaerobically with N 1 -acetyl-SPD (data not shown). We chose N 1 -acetyl-SPD as the mPAO reductant rather than the better substrate N 1 -acetyl-SPM because we wanted to avoid the possible slow reduction of the high concentration enzyme by SPD, the N 1 -acetyl-SPM oxidation product. Because mPAO oxidizes N 1 -acetyl-SPM, N 1 -acetyl-SPD, or SPM at the exo-carbon of secondary amino groups (see below), there is no chance that the N 1 -acetyl-SPD oxidation product PUT (which does not have a secondary amino group) would be oxidized during the titration. Based on the A 458 , the concentration of enzyme was 1.20 M, whereas a concentration of 1.16 M was determined from the N 1 -acetyl-SPD titration; as expected 1 mol of substrate reduced 1 mol of FAD. Thus, all enzyme molecules in the preparation are capable of oxidizing the substrate. No trace of a flavin radical was detected during this titration. This indicates rapid transfer of 2 electrons from enzyme-bound substrate to enzyme-bound FAD.
The Steady-state Kinetic Properties of mPAO-It was assumed that steady-state mechanism for the oxidation of the various polyamine derivatives listed in Table I is of the pingpong type. This is supported by the fact that the apparent k cat /K S values for assay done in air-saturated (0.237 mM O 2 ) and pure O 2 -saturated (1.2 mM O 2 ) buffers were approximately equal. Further support for this contention is provided by the k cat /K O values for N 1 -acetyl-SPM and N 1 -acetyl-SPD. These values are equal (Table I), as expected for a ping-pong type mechanism (42).
The steady-state kinetic studies indicated that N 1 -acetyl-SPM is the best substrate for mPAO, although N 1 -acetyl-SPD is also a good substrate (Table I). The k cat /K S value (the socalled "specificity constant") for the former substrate is over an order of magnitude higher than for the latter. Whereas SPM can be oxidized by the enzyme, it is much less efficient than for the oxidation of N 1 -acetyl-SPM or N 1 -acetyl-SPD; the k cat /K SPM value is 4 orders of magnitude lower than that for N 1 -acetyl-SPM. It was found that SPD, PUT, N 8 -acetyl-SPD, and benzylamine were not oxidized by mPAO.
Vujcic et al. (46) expressed cloned hpao and mpao genes in a human kidney cell line and determined that these cells oxidized substrates with the following preference: N 1 -acetyl-SPM Ϸ N 1 -acetyl-SPD Ͼ N 1 ,N 12 -diacetyl-SPM Ͼ Ͼ spermine. These findings are basically the same as those reported herein.
Interestingly, both BESPM and BENSPM are fairly good mPAO substrates, both being better than SPM (Table I). This is an important finding because BENSPM has been used for phase II cancer clinical trials (55). These N-ethylated polyamines have been used widely also to study the physiological effects of polyamine-metabolizing enzymes. They down-regulate polyamine biosynthetic enzymes, but dramatically up-regulate SSAT synthesis (13,14), which results in mammalian cells becoming apoptotic.
It has been reported that terminally alkylated polyamine analogs like BESPM and BENSPM are oxidatively dealkylated by PAO; for BESPM and BENSPM this would result in the formation of N 1 -ethyl-SPM and N 1 -ethyl-nor-SPM, respectively, and acetaldehyde (56 -59). In contrast, it was reported recently that the lysates of cultured human cells transiently transfected with mpao or hpao cDNA did not dealkylate BESPM and BENSPM but converted them to N 1 -ethyl-SPD and N 1 -ethyl-nor-SPD ({3-[(3-aminopropyl)amino]propyl}ethylamine), respectively, and N-ethyl-3-aminopropanal. It was not possible for us to resolve this dilemma by analyzing the products formed when BESPM or BENSPM were oxidized by pure mPAO, as was done for N 1 -acetyl-SPM and N 1 -acetyl-SPD (see below). The appropriate reference compounds (i.e. N 1ethyl-SPM, N 1 -ethyl-nor-SPM, N 1 -ethyl-SPD, N 1 -ethyl-nor-SPD, or N-ethyl-3-aminopropanal) are not available commercially and are not conveniently synthesized.
Inspection of K I values (Table I) indicates that SPM and PUT (but not benzylamine) are weak inhibitors for the oxidation of N 1 -acetyl-SPM (Table I), whereas N 8 -acetyl-SPD and SPD are somewhat better inhibitors. All of these compounds are competitive inhibitors, because in assay with each inhibitor at levels that would produce significant inhibition, the inclusion of N 1 -acetyl-SPM at a saturating concentration eliminated the inhibition.
Ideally, we would like to compare the biochemical and kinetic properties of the E. coli-produced mPAO with that isolated from a natural source. However, this was not feasible for several reasons. First, it is not prudent to compare the E. coliproduced mouse enzyme with the enzyme purified from bovine liver because of the species difference. Furthermore, the specific activities obtained from two different purifications of the bPAO were not the same and were lower than expected for a fully functional enzyme. Because only 1 mg of PAO could be obtained from 1 kg of bovine liver, it would be difficult to obtain sufficient quantities of PAO from mouse liver for comparative studies. We attempted to obtain a full-length bpao coding cDNA fragment. Unfortunately, we could not identify the appropriate clones from any cDNA library that was screened. Therefore, we were not able to heterologously produce bPAO for comparison with the enzyme obtained from bovine liver. Furthermore, a k cat /K S value of 2.54 ϫ 10 6 M Ϫ1 s Ϫ1 for oxidation of N 1 -acetyl-SPM by mPAO is that expected for a native, fully active enzyme (60). Additionally, recombinant mPAO is a highly stable, monomeric protein that maintains its biochemical, redox, and kinetic properties even after prolonged storage. Finally, 1 mol of enzyme FAD is reduced efficiently by 1 mol of substrate in the reductive titration (see above). Thus, we are confident that this E. coli-produced mPAO is in its native, fully active form.
The Nature of the Product Resulting from the Oxidation of N 1 -Acetyl-SPM and N 1 -Acetyl-SPD by mPAO-By using the HPLC method described under "Experimental Procedures," it was found that complete oxidation of N 1 -acetyl-SPM and N 1acetyl-SPD by mPAO yielded 96 and 94% (based on 1 mol of aldehyde/mol of substrate), respectively, of the 2,4-dinitrophenylhydrazone of 3-acetamidopropanal (Fig. 6). There was no trace of any other 2,4-dinitrophenylhydrazones such as would be seen if the enzyme oxidized the substrates on the endo-side of their N 4 -nitrogens. We expect that the phenylhydrazone of this aldehyde, because of its positive charge, would have very short HPLC retention times; the retention time of the 2,4dinitritophenylhydrazone of 3-aminopropanal (a reference compound that is positively charged in the HPLC solutions containing trifluoroacetic acid) has a much shorter retention time than the same derivatives of 3-acetamidopropanal and acrolein, which are uncharged.
The acetyl group of 3-acetamidopropanal did not hydrolyze during the enzymatic reaction or during the workup preceding the analyses. This hydrolysis would produce 3-aminopropanal, which can spontaneously convert to acrolein (18). However, the 2,4-dinitrophenylhydrazones of either 3-aminopropanal or acrolein were not detected in the HPLC analyses of the enzyme reaction solutions (Fig. 6).
To prove definitively that 3-acetamidopropanal was the true product of these enzymatic oxidations, a larger scale reaction between N 1 -acetyl-SPM and mPAO was carried out, and the 2,4-dintorphenylhydrazone of the product aldehyde was isolated in 89% yield. The chemical properties of this compound and the 2,4-ditrophenylhydrazone of 3-acetamidopropanal generated by organic synthesis were compared. The two substances were identical in all respects, and 1 H NMR and mass spectral analyses prove incontrovertibly that 3-acetamidopropanal is the enzymatic oxidation product.
By using an HPLC method to analyze dansylate polyamines, it was found that complete mPAO oxidation of N 1 -acetyl-SPM and N 1 -acetyl-SPD produced 95% SPD and 91% PUT (based on 1 mol/mol of substrate), respectively (data not shown). Another research group (46) exposed these substrates to lysates from HEK-293 cultured human kidney cells transiently transfected with the genes for hpao or mpao. By using an HPLC method similar to that described herein for analyzing dansylated polyamines, they found also that mPAO and hPAO converted these substrates to SPD and PUT, respectively.
These observations indicate that N 1 -acetyl-SPM and N 1acetyl-SPD are always oxidized at the carbon on the exo-side of their N 4 -nitrogens. Thus, there can be no doubt that mPAO is the classical polyamine oxidase that has been described and studied over the past few decades. DISCUSSION Peroxisomal PAO, an integral component of polyamine interconversion pathway, is an important player in regulating cellular polyamine levels. Thus, understanding the precise biochemical and structural properties of PAO are essential for a deeper understanding of its participation in many fundamental cellular processes. With this in mind, we set out to develop a system that would provide, in good yield, a highly purified preparation of a mammalian peroxisomal PAO. In the course of accomplishing this goal, we cloned and sequenced the entire mpao (Fig. 2) gene and most of the bpao gene (GenBank TM accession number AF226658). Based on a comparison of primary structures (Fig. 3), the sequence identity with other flavin-containing amine oxidases is less than 40%. This indicates that PAO represents a new subfamily of flavoproteins. Inspection of the translated mPAO and bPAO sequences indicated the presence of a ␤ 1 ␣␤ 2 FAD-binding fingerprint motif (Fig. 3) that interacts with the ADP moiety of the enzymebound FAD (26,33). This motif along with numerous other conserved regions are found in the "FAD-binding domain" and are elements of the Rossmann fold (Fig. 3). Two other conserved regions are located in the "substrate-binding domain" (26,33).
Cys 406 , Cys 397 , and Cys 399 , the residues that are covalently attached to the 8␣-carbon of the isoalloxazine ring of FAD in of MAO-B, MAO-A (24,25,33), and fMAO, respectively, are pointed out in the aligned sequences of Fig. 3. These Cys residues align with Ser residues in bPAO, mPAO (Ser 429 ), and the murine and human SMO (Ser 481 in both). However, except for the mitochondrial monoamine oxidases, FAD is noncovalently bound to all known amine oxidases of this family.
As with MAO-A, MAO-B (61, 62), and MAO-N (28), mPAO forms an intermediate anionic (red) radical when titrated with dithionite. This indicates that there is either a positively charged aminoacyl group (i.e. Arg or Lys) or the positive end of an ␣-helix dipole is near the N-1 position of the flavin's isoalloxazine ring. This positively charged environment stabilizes the negative charge of the red radical, which is localized at the N1/C2/C2O locus of the isoalloxazine ring of the flavin. In the MAO-B and cPAO structures, the positive end of an ␣-helix interacts with the N1/C2/C2-O locus of FAD (26,33). The helices span residues from Met 438 to Met 454 of MAO-B and residues from His 469 to Gln 487 of cPAO (Fig. 3). These segments of cPAO and MAO-B align with the highly conserved region near the C termini (25) of the other flavoprotein amine oxidases ( Fig. 3; for mPAO, Thr 475 -Gln 496 ), and the secondary structure prediction program "Psi-Pred" (version 2, at the web site, insulin.brunel.ac.uk/psipred/) indicated that this region of mPAO forms an ␣-helix.
Whereas SPM can be oxidized by peroxisomal mPAO, it is a poor substrate when compared with N 1 -acetyl-SPM and N 1acetyl-SPD. SPM and SPD are acetylated for transport from the cells and eventual excretion from the body (1). High PAO levels could prevent transport of N 1 -acetyl-SPM and N 1 -acetyl-SPD from the cell and increase the SPD and PUT levels of the polyamine pool of the cell (Fig. 1). In contrast, hSMO and mSMO oxidize SPM (K S ϭ 18 M) (10) but not N 1 -acetyl-SPM or N 1 -acetyl-SPD (10,11). In addition to the absence of a peroxisomal transport signal at the C termini of hSMO and MSMO, there are other significant sequence differences between these mPAO (Fig. 3); hSMO and mSMO also do not have N-terminal transport signals, suggesting that they are cytosolic enzymes.
Polyamine interconversion involving peroxisomal PAO helps maintains the intracellular balance of these substances. It has been proposed that in some cells under stress, polyamine oxidation generates the toxic byproducts H 2 O 2 and 3-aminopropanal (generated by enzymatic deacetylation of 3-acetamidopropanal, the product of N 1 -acetyl-SPM oxidation by PAO; Fig.  1), which can initiate cell death (15-17, 64 -69). In addition, 3-aminopropanal can spontaneously convert to the extreme cytotoxin acrolein (18). H 2 O 2 can be inactivated by catalase in peroxisomes, unless the levels of N 1 -acetyl-SPM and PAO are extremely high (or the catalase level low), which seems to be the case in some pre-apoptotic cells. It is believed also that the level of SSAT, which produces N 1 -acetyl-SPD and N 1 -acetyl-SPM from the polyamine pool, is elevated in these cells (15-17, 64 -68). However, in some cultured cancer cells, the levels of SSAT and thus the N 1 -acetylated polyamines are high, but PAO is low. In fact, the level of PAO activity decreases as the histological grade of breast cancer tumors increases (63). From these observations, it can be proposed that a PAO-dependent apoptosis initiation mechanism is intact in some precancerous cells. However, events take place whereby PAO activity is interrupted, shutting down cell death, and cellular proliferation ensues.
In contrast, for tissue damaged by ischemia/reperfusion, the level of SSAT, N 1 -acetylate polyamines, and PAO increase (19 -21). This results in the production of high levels of H 2 O 2 and 3-aminopropanal (and perhaps acrolein), which contributes to tissue damage.
With the work described herein, a program has been initiated to study the detailed chemical, biochemical, structural, kinetic, inhibition, and mechanistic properties of a mammalian peroxisomal PAO. Hopefully, this will lead to a richer appreciation of its involvement in apoptosis, cellular proliferation, cell signaling, tissue damage, wound healing, tissue development, and differentiation, etc. and aid in the development of clinically relevant approaches to treat cancer and ameliorate ischemic tissue damage.