cDNA Cloning, Sequencing, and Characterization of Male and Female Rat Liver Aldehyde Oxidase (rAOX1)

Molecular characterization of male and female rat liver aldehyde oxidase is reported. As described for the mouse liver, male and female rat liver expressed kinetically distinct forms of aldehyde oxidase. Our data suggest that the two forms arise as a result of differences in redox state and are most simply explained by expression of a single gene encoding aldehyde oxidase in rats. In support of this argument we have sequenced cDNAs from male and female rat liver. We examined mRNA expression by Northern blot analysis with RNA from males and females, from several tissues, and following androgen induction. Purified rat liver enzyme from males or females revealed a single 150-kDa species consistent with cDNA sequence analysis. Both male and female forms were reactive to the same carboxyl-terminal directed antisera.K m (app) values obtained in crude extracts of male or female rat liver and post-benzamidine-purified aldehyde oxidase differed substantially from each other but could be interconverted by chemical reduction with dithiothreitol or oxidation with 4,4′-dithiodipyridine. Our data indicate that a single gene is most likely expressed in male or female rat liver and that the kinetic differences between male and female rat liver aldehyde oxidases are sensitive to redox manipulation.

and numerous pharmacologic agents. While substrate specificity for AOX is very broad, and wide species variation in substrate specificity exists, the general catalytic reaction takes the form of hydroxyl transfer from water to an aldehyde creating the cognate acid. For example, conversion of benzaldehyde to benzoic acid is a very efficient reaction for AOX from most species. Conversion of N-1-methyl nicotinamide (NMN) to the 2-or -4-pyridone has been used as a standard definition of AOXs, although it is usually a kinetically less efficient reaction than benzaldehyde oxidation.
AOX is of interest both for its role in drug metabolism and as a source of the reactive oxygen species (ROS), hydrogen peroxide, and the superoxide anion, that have been related to numerous human pathologies. The human gene encoding AOX has been linked to a rare form of amyotrophic lateral sclerosis, although it is unknown if AOX encodes the amyotrophic lateral sclerosis locus itself (4 -6). ROS are generated from AOX in an oxidative half-reaction following reduction of the enzyme by substrate. The FeS, FAD, and MoCo cofactors comprise an internal electron transfer chain in which electrons are passed from the active site molybdenum center to the FeS centers and finally to FAD. Partial reduction of oxygen at the the reduced FAD site produces ROS. Unlike the related enzyme xanthine dehydrogenase (XDH), AOX does not utilize NAD ϩ as a cofactor and therefore AOX does not undergo the classical "D-form" to "O-form" conversion characteristic of XDH (7).
Two distinct AOX activities have been reported for mouse liver (8). Hepatic AOX from male and female mice differed in K m and V max for use of the substrate, benzaldehyde. The male enzyme exhibited a K m of 40 M while the female enzyme had a K m of 115.4 M. The male V max was 423.7 nmol/min/mg of protein and the female enzyme was 203.0 nmol/min/mg of protein. Furthermore, male and female mouse AOX enzymes were dramatically regulated by testosterone. Treatment of female mice with testosterone proprionate (TP) converted both K m and V max values to the male pattern (8). Previous work had shown that castration resulted in the loss of a male pattern with conversion to a female pattern and TP supplementation restored the male-specific pattern of AOX expression (9). This observation was consistent with early reports of androgenic regulation of mouse hepatic AOX (10,11). More recently, regulation of mouse hepatic AOX by testosterone was found to be mediated by growth hormone (12) where, again, growth hormone supplementation was found to convert K m and V max of the female-specific pattern to a male-specific pattern. Nonetheless, when male and female hepatic AOXs were purified to homogeneity they revealed a single band of 150 kDa on SDS-PAGE analysis and a single active band on native PAGE (8). How the differences between male and female hepatic AOXs were generated remained unclear.
Two distinct AOX activities were also purified from rat liver and identified as NMN oxidases I and II (13,14). The two activities exhibited different K m values for the oxidation of NMN to either its 2-or 4-pyridone. These two activities also differed by pH optima, heat stability, and inhibitor sensitivity. NMN oxidases I and II were found to possess distinct kinetic parameters, K m and V max , for oxidation of several different substrates, including benzaldehyde and NMN. Furthermore, wide variation in K m has been observed between different species and between individual rat strains (15)(16)(17). As was found for mice, rat hepatic AOX could be purified to homogeneity to yield a native enzyme of 300 kDa that resolved into two 150-kDa subunits by SDS-PAGE (13,14).
AOX genes are widely expressed phylogenetically and in some organisms appear to arise from multigene families. Even the Archae express MoCo enzymes related to AOX (18). Two different AOX cDNA sequences were reported for corn plants that were themselves 83% identical (19). Three cDNA sequences were reported for Arabidopsis thaliana (20), and tomato plants may also possess several AOX genes (21). Two AOX genes have been identified in Drosophila melanogaster, one encoding AOX and the other encoding the highly related pyridoxal oxidase (PO) which was identified as an AOX (22)(23)(24). Sequences for AOX or PO in Drosophila have not been reported. Multiplicity in AOX genes was also reported for the mouse where independently segregating loci appeared to encode AOX enzymes with distinct electrophoretic mobilities and these were identified as AOX1 and AOX2 (9). Importantly, the different isozymes appeared to segregate as different AOX genetic loci under differential developmental and androgenic regulation. Human and bovine AOX sequences have been published that are approximately 82% identical, and a small fragment of a mouse AOX sequence was published (5,(25)(26)(27). However, second copies of the vertebrate AOX genes have not been cloned or sequenced. Therefore, while at least two or three AOX genes appear to exist in plants and flys, they have not been confirmed by sequence analysis in higher organisms. Furthermore, Southern blot analysis of chromosomal DNA could be interpreted to suggest that only a single AOX gene was present in humans (28).
Molecular characterization of AOX from the rat has not been reported. Because both forms of AOX appeared to be expressed in rat liver, we have examined expression of AOX genes from both male and female rat livers. We have confirmed the existence of two kinetically different forms of AOX in male and female rats. Sequence analysis of the corresponding cDNAs indicates that a single AOX gene is most likely activated in the liver in male and female rats. By Northern blot analysis, male and female rat liver RNA contained a single mRNA species that did not exhibit induction by TP. Purified rat liver AOX from males and females revealed a single 150-kDa band on SDS-PAGE. Present experiments suggest that a primary difference between male and female forms of AOX may lie in their respective redox states.

MATERIALS AND METHODS
RNA Purification and cDNA Synthesis-RNA was prepared from organs of freshly killed Sprague-Dawley rats by quick freezing the tissue in liquid nitrogen followed by extraction in guanidine isothionate and phenol:chloroform:isoamyl alcohol (24:24:1) (29). Frozen tissues were stored at Ϫ70°C until use. Poly(A) ϩ RNA was prepared by fractionation on oligo(dT)-cellulose (Stratagene, La Jolla, CA). cDNA was prepared by reverse transcription in a final volume of 20 l as follows. 1.0 g of poly(A) ϩ RNA was mixed with diethyl pyrocarbonate-treated water, 1.0 l of primer oligonucleotide at 20 M, 1.0 l of 10 mM deoxyribonucleoside triphosphates (ACGT), 0.5 l of RNase inhibitor, 1.0 l of recombinant Moloney murine leukemia virus reverse transcriptase (CLONTECH Laboratories, Palo Alto, CA), and 4.0 l of 5 ϫ buffer (final conditions: 50 mM Tris-HCl, pH 8.3, 75 mM KCl, 3 mM MgCl 2 ). Reactions were incubated at 42°C for 60 min and then heated to 94°C for 5 min to inactivate reverse transcriptase. Prior to use, reactions were diluted to 100 l and 5 l was used for PCR amplification. Reverse transcriptase reactions were stored at Ϫ70°C.
3Ј and 5Ј RACE-A region from the middle of AOX1 was obtained by PCR amplification of reverse transcribed male rat liver poly(A) ϩ RNA using synthetic oligonucleotides (Life Technologies, Inc., Gaitherburg, MD) derived from a fragment of the mouse liver AOX1 sequence (27). Nucleotides 1,682 through 2,217 were amplified with the oligonucleotides MAO4RAT1 and MAO4RAT2 (Table I) to produce a single 535base pair fragment that was gel purified, sequenced in its entirety from two directions, and cloned as pMID. The resulting sequence was used to derive the unique sequence oligonucleotides for rat AOX1, FORINRAO, and REVINRAO. 3Ј RACE was performed as follows. Male rat liver poly(A) ϩ RNA was reverse transcribed using the oligonucleotide 3Ј RATRACE as a primer for reverse transcriptase. The resulting single strand DNA was amplified by PCR using the oligonucleotides MAO4RAT1 and 3Ј RATRACE. A second round of PCR amplification was performed using the 5Ј nested oligonucleotide, REVINRAO, and 3Ј RATRACE. A single product of 2,180 nt was obtained and sequenced entirely from both directions. This fragment was cloned as p3Ј RA-TRACE and showed 100% identity with the overlap region of pMID. 5Ј RACE was performed as follows. Male rat liver poly(A) ϩ RNA was reverse transcribed using random hexamer oligonucleotides. The resulting single strand DNA was amplified by PCR using the oligonucleotides MAO4RAT2 and IVS22. A second round of PCR amplification was performed using IVS22 and the nested oligonucleotide FORINRAO. A single band of 1,930 nt was obtained, sequenced in its entirety from both directions, and cloned as p5ЈRATRACE. This sequence revealed 100% identity in the region of overlap with pMID. The extreme 5Ј end and upstream region of the male rat AOX1 cDNA was obtained by a modified 5Ј RACE as follows. Male rat liver poly(A) ϩ RNA was reverse transcribed using random hexamer oligonucleotides. RNA was hydrolyzed in sodium hydroxide and the resulting single strand DNA was subjected to two cycles of nested PCR, the first using an adapter ligated oligonucleotide at the 5Ј end. Oligonucleotide 3ЈIVS11 was phosphorylated with polynucleotide kinase in the presence of ATP. Following extraction with phenol:chloroform:isoamyl alcohol and ethanol precipitation the phosphorylated oligonucleotide was treated with dideoxyadenosine triphosphate and terminal transferase from bacteriophage T4 to block elongation from the 3Ј end. Blocked, phosphorylated oligonucleotide was then ligated to single strand DNA using bacteriophage T4 RNA ligase in the presence of hexamine cobalt chloride to produce adapter-ligated single strand DNA. Adapter-ligated single strand-DNA was subjected to first round amplification using the oligonucleotides 3Ј IVS11 and RAT5. The resulting PCR products were subjected to a second round of amplification using 3Ј IVS11 and the nested primer, RAT6. The resulting 160-base pair DNA was cloned (p5ЈEND-male), sequenced, and showed 100% identity with the overlap region of p5ЈRATRACE. This sequence was inferred to contain the translation initiation site and 49 nucleotides of 5Ј-untranslated region because it showed excellent deduced amino acid sequence homology with human and bovine AOX sequences, a single ATG was found to be in-frame with the downstream sequence, and translational termination sequences were observed upstream of this ATG and in the same reading frame.
DNA Sequence Analysis-Direct fluorescence sequence analysis was performed on plasmid DNA or PCR products using oligonucleotide primers designed to yield approximately 400 nt between primers and approximately 100 nt of overlap. A list of sequencing oligonucleotides and their sequences is available upon request. Prior to sequence determination, plasmids were prepared by alkaline lysis (29). PCR products were purified from low melting point agarose by phenol extraction and precipitation in ethanol. Sequences were determined using dideoxynucleotide chain terminating system from Perkin-Elmer Applied Biosystems (Foster City, CA). Reactions used the ABI PRISM TM Dye Terminator cycle sequencing ready reaction kits (Perkin-Elmer). Sequence reactions were fractionated on an ABI PRISM 310 DNA sequencer equipped with a 43-cm microcapillary (Perkin-Elmer). All sequences were determined from both directions and sequence data were compiled manually.
Northern Blot Analysis and Quantitation-Northern blots were run using formaldehyde-agarose gels and 5 g of poly(A) ϩ RNA (6,29). Hybridization probes were isolated from the clones pMID, p3ЈRATRACE, and p5ЈRATRACE by PCR amplification and agarose gel electrophoresis. Following isolation in phenol, DNA fragments were labeled by random primed synthesis in the presence of [ 32 P]dATP for use as hybridization probes. High stringency hybridization and washing were conducted as described (6,29). Following hybridization and autoradiography, each 4,500-nt band was cut from the hybridization filter and counted by liquid scintillation counting for quantitation. The remaining filter was dissociated from residual 32 P and rehybridized with a ␤-actin specific probe. Actin hybridization was also quantitated by excising the bands from the filter and liquid scintillation counting. Each AOX1 hybridization signal was normalized to the corresponding signal for ␤-actin after correction for background hybridization.
Castration and Hormone Supplementation-Castrated or sham castrated male Sprague-Dawley rats were obtained from Charles Rivers Laboratory (Wilmington, MA) following 1 week of recovery from surgery. Surgery was performed when rats were 4 weeks of age. Rats were maintained at Webb-Waring facilities for 1 additional week of equili-bration. Testosterone proprionate was administered at a dose of 50 mg/kg body weight in corn oil by daily subcutaneous injection. Growth hormone was administered at a dose rate of 0.05 IU/100 g of body weight in a buffer composed of 30 mM NaHCO 3 , 150 mM NaCl, pH 8.25, by subcutaneous injection twice daily. Sham castrated and castrated controls received corn oil injection. Following 10 days of treatment, rats were killed by sodium pentabarbitol administration. Organs were harvested immediately and dropped into liquid nitrogen for subsequent RNA preparation. RNA was analyzed from individual organs with 4 rats in the sham controls, 4 rats in the castrated control group, 5 rats in the castrated and testosterone supplemented group, and 5 rats in the castrated and growth hormone supplemented group.
AOX Activity Assays-AOX activity and initial rate data were determined spectrophotometrically in a 1-ml reaction containing 50 mM potassium phosphate buffer, pH 8.0, NMN at 5 mM or as needed, 10% dimethyl sulfoxide, 250 international units of CAT, 5-50 M menadione as required, and appropriate levels of purified or partially purified enzyme. Initial rate data were obtained over a 5-min period. Formation of the pyridone of NMN was monitored at 300 nm.
AOX Enzyme Purification and Characterization-AOX enzyme activity was purified from male and female Spraque-Dawley rat livers. After removal of the liver, all procedures were performed at 4°C or on ice using ice-chilled buffers. Liver sections (10 -20 g) were diced, rinsed several times, homogenized, and dounced in 3 volumes of ice-cold 100 mM potassium phosphate, pH 7.5, containing 25 mM benzamidine hydrochloride, .2 mM phenylmethylsulfonyl fluoride, .1 mM EDTA. Homogenates were centrifuged for 1 h at 100,000 ϫ g. The supernatant was brought to 5 mM DTT and incubated for 1 h. MnCl 2 was then added to the supernatant to a final concentration of 10 mM. The solution was then centrifuged for 5 min at 17,000 ϫ g and the pellet was discarded. Dry ammonium sulfate was added with stirring to achieve a final concentration of 30%. The slurry was centrifuged and the pellet discarded. The resulting supernatant was brought to 50% saturation with ammonium sulfate, centrifuged, and the supernatant discarded. The resulting pellet (50% pellet) was resuspended in 1/20 the original volume of potassium phosphate buffer. Acetone fractionation was subsequently achieved using acetone chilled with dry ice. Suspensions were brought to 40% in chilled acetone, centrifuged, and the pellet discarded. Supernatants were brought to 50% in acetone, centrifuged, and the pellets collected. The 50% pellet was resuspended in the original volume of buffer and dry ammonium sulfate was added to achieve a 60% saturation. The pellet was recovered by centrifugation and resuspended in 1/20 of the original volume in 100 mM glycine, pH 9, containing 100 mM NaCl. After resuspension, insoluble debris was removed by centrifugation and the solution was desalted on a 5 ϫ 15-cm Sephadex G-25 column in the above glycine buffer. The desalted solution was loaded onto a 2.5 ϫ 10-cm benzamidine-Sepharose 6B (Pharmacia) column equilibrated in the same buffer. The column was washed with 3 column volumes of buffer and elution was achieved by flushing the column with 500 mM benzamidine hydrochloride. Elution was monitored at 436 nM and the single eluting peak was collected and precipitated with ammonium sulfate at 60% saturation. The pellet was stored at 4°C for up to 1 day or was resuspended in a minimum volume of 100 mM potassium phosphate, pH 7.5, and dialyzed against the same buffer overnight.
The OD 280/450 ratio was between 5 and 7 for male or female preparations. Specific activity for NMN hydroxylation to the pyridone was 100 -250 nmol/min/mg for the female or male enzymes. Both enzyme preparations were inhibited to greater than 95% by inclusion of 50 M menadione. We observed persistent aggregation of the soluble protein fraction when preparations did not include treatment with DTT early in the fractionation. Aggregation reduced the overall yield of AOX enzyme to less than 0.1% of the starting activity. Reduction of the crude lysate prior to MnCl 2 treatment improved the overall yield to 7% of the starting activity. SDS-PAGE analysis of the aggregated proteins suggested no obvious bias for specific aggregated proteins. Furthermore, reduction in 5 mM DTT was significantly more effective in preventing aggregation than was 10 mM cysteine. Reduction of male or female liver extracts permitted purification of both enzymes to homogeneity.
Preparation of Antibody to AOX-The amino-terminal decapeptide comprising the sequence NH 2 -DRASELLFYV-COOH and the carboxylterminal decapeptide comprising the sequence NH 2 -GSYVPWNIPV-COOH were synthesized by Dr. Hans-Richard Rackwitz (German Cancer Research Center, Heidelberg, Germany). Antibody to the synthetic oligo peptides was produced in rabbits by intravenous injection of 20 g of peptide. Rabbits were boosted with peptide every 2 weeks. The IgG fraction was prepared from serum by ammonium sulfate precipitation following coagulation of the blood and sedimentation. This produced two antisera preparations: AOX-NT (amino-terminal antibody) and FIG. 1. PCR amplification, cloning, and sequence strategy for male and female rat AOX1. The upper bar shows the deduced cDNA structure for both male and female rat liver AOX1. PCR amplified and cloned fragments are shown below. Note that the 3Ј RACE product for the female has been truncated within the untranslated region and does not comprise the entire untranslated region. Each cloned PCR fragment was subjected to DNA sequence analysis using a battery of oligonucleotides designed to encompass approximately 400 nucleotides between oligonucleotide primers. Each fragment was sequenced entirely from both directions. A list of sequence analysis primers and their sequences is available upon request. The assembled cDNA sequences for both male and female rat liver AOX have been deposited in the NCBI gene bank data base.

AOX-CT (carboxyl-terminal antibody).
Western Immunoblot Analysis-Protein was electrophoresed on SDS-PAGE and transferred to polyvinylidine difluoride membranes (Bio-Rad). Filters were sliced for staining with Commassie Brilliant Blue or processed for immunoblot analysis. For reaction with antisera, filter strips were blocked with gelatin overnight prior to reaction with preimmune sera, AOX-NT, or AOX-CT antisera. Antigen-antibody complexes were detected by reaction with alkaline phosphatase streptavidin kit (Bio-Rad). male and female rat livers. Apparent K m (K m(app) ) values were determined from Lineweaver-Burk plots by measuring conversion of NMN to its pyridone. K m(app) for male rat liver AOX was 538.8 M and K m(app) for the female was 1062.3 M, consistent with previous reports showing two forms of AOX in livers from rats and different forms of AOX in livers from male and female mice. While no explanation for this difference has been produced, the two AOX genes identified in mice, AOX1 and AOX2, suggested the possibility that two different AOX genes may be expressed in rat liver.

Different Forms of AOX Exist in Male and
cDNA Sequence Analysis of Male and Female Rat Liver AOX1- Fig. 1 illustrates the PCR amplification strategy used to obtain segments of male rat liver AOX1. DNA sequence of the three PCR products was assembled to produce the male rAOX1 cDNA. RAOX1 comprised 4,304 nucleotides, including 30 nt of polyadenylation, 210 nt of 3Ј-untranslated region, and 49 nt of 5Ј-untranslated region. A single open reading frame was identified that encoded a protein of 1,333 amino acids and a deduced mass of 147,009 Da. The deduced male rAOX1 protein exhibited 82% sequence identity with human AOX1 and 81% sequence identity with bovine AOX1. Multiple sequence clustal analysis revealed conservation of co-factor domains corresponding to FeS I, FeS II, FAD, and five small domains within the MoCo binding segment (Fig. 2).
Female rat liver AOX1 cDNA sequence was obtained using a similar strategy with the exception that a unique sequence oligonucleotide derived from the male sequence, 3Ј RATUTR, was used to obtain the 3Ј RACE product. Thus, 47 nucleotides of 3Ј-untranslated region was obtained for the female and this does not include the polyadenylation site (Fig. 1). The assembled cDNA sequence for female rat liver AOX1 encoded a deduced protein of 1,333 amino acids and 146,919 Da. The female cDNA sequence was 99.8% identical to the male sequence and the deduced protein sequence was 99.6% identical to the male sequence. Of the 10 nucleotide differences detected between the male and female rat liver AOX1 cDNA sequences described here, five resulted in changes to the deduced amino acid sequences (Table II). However, while nucleotides 405 and 408 differ between male and female sequences reported here and to the GenBank data base, these variations were also found between individual male clones and therefore do not represent gender differences but differences between individual rats. The full extent of individual variation was not determined and it remains possible that all of the differences observed between the two clones described may be attributed this cause alone.
Expression of Rat AOX1 mRNA- Fig. 3A shows Northern blot analysis of poly(A) ϩ RNA from male rat liver. Hybridization probes were derived from each of the three male clones, pMID, p5ЈRATRACE, and p3ЈRATRACE. The region between nucleotides 1,682 and 2,217 of the male cDNA, corresponding to the pMID hybridization probe, produced hybridization signals at approximately 4,500 and 2,500 nt. Hybridization probes derived from both the p5ЈRATRACE and p3ЈRATRACE clones produced predominantly a single band at 4,500 nt with weak hybridization to the band at 2,500 nt. We conclude that the predominant mRNA for rAOX1 detected by Northern blot analysis in males is approximately 4,500 nt, consistent with the cDNA sequence assembled for rAOX1. The unexpected signal at 2,500 nt may represent a cross-reactive species largely localized to the pMID region.
Hybridization probes derived from the p5ЈRATRACE produced predominantly a single band from both male and female RNA (Fig. 3B). This RNA was also estimated to be 4,500 nucleotides in size, and no difference in size or number of hybridizing bands was detected between males and females.
Northern blot analysis of poly(A) ϩ RNA from several different tissues showed expression of a single 4,500-nt RNA for all tissues examined (Fig. 3C). Different tissues did not show variation in either the size or multiplicity of AOX RNAs. Var- FIG. 3. Northern blot analysis of rat AOX1 expression. A, each of the three clones used to assemble male rat liver AOX1 was used as a hybridization probe for independent Northern blots of male rat liver poly(A) ϩ RNA. The major band at 4,500 nt corresponds to the AOX1 mRNA. The band at 2,500 nt that has greater localization to the pMID region of AOX is assumed to represent a cross-reactive species. Although it has not been excluded that this may represent a breakdown product of the larger RNA, it is too small to encode a full-length AOX. B, poly(A) ϩ RNA from male and female rat liver has been analyzed by Northern blot using the p5ЈRATRACE clone as a hybridization probe. No attempt is made here to indicate a difference in abundance of the AOX mRNA between males and females. C, poly(A) ϩ RNA from several different male rat tissues has been analyzed by Northern blot using the AOX1 insert from p5ЈRATRACE as a hybridization probe.

TABLE II
Sequence differences in male and female rat AOX1 clones described here The differences in cDNA sequence between the male and female rat AOX1 clones are shown. nt-site refers to the specific base changed from male to female using the A of the translational initiator as nucleotide ϩ1. The nature of the base pair changed is shown along with the change, if any, in the deduced amino acid sequence and the corresponding amino acid number. Note, these changes reflect differences between our clones and do not necessarily reflect consistent differences between genders. RNA from male rats that had been sham castrated, castrated, castrated and treated with TP, or castrated and treated with growth hormone was analyzed by Northern blot. Fig. 4 shows that steady state RNA levels were only slightly affected by any of these treatments. When hybridization signals were quantitated and normalized to either OD 280 or to ␤-actin hybridization signal (Table III), we found no statistically significant difference between groups in AOX1 mRNA abundance. These data do not support significant regulation of rat AOX1 mRNA abundance by TP.
AOX Enzyme Purification and Characterization-AOX enzyme activity was purified to homogeneity from male rat livers (Fig. 5A). These preparations produced a single band of 150 kDa by SDS-PAGE analysis. We were unable to sequence this protein by direct Edman degradation suggesting that its amino terminus was blocked, as was found for both the rabbit liver and bovine liver AOXs (5,26). Western immunoblot analysis of partially purified AOX from male and female rat livers revealed excellent reactivity of each enzyme preparation to this synthetic peptide-derived antibody (Fig. 5B). Furthermore, reactive proteins from male and female revealed a predominant polypeptide of approximately 150 kDa with no evident difference in size between genders. (Fig. 5C).
K m(app) values for oxidation of NMN to its pyridone were determined by analysis of Lineweaver-Burk plots. Fig. 6 and Table IV show these results. As noted above, crude liver ex-tracts from males produced K m(app) of 538.8 and 1062.3 M for females. Reduction of the crude liver extract with 5 mM DTT shifted K m(app) for males and females. Reduced male rat liver produced K m(app) of 359.9 M and for the reduced female enzyme K m(app) was 354.5 M (Fig. 6B). Thus, male and female forms of AOX can be converted to a form with indistinguishable K m(app) values by chemical reduction in a crude lysate. Oxidation of the reduced female preparation with 4,4Ј-DTDP converted K m(app) back to a form similar to that obtained from the untreated female preparations (Fig. 6B). Furthermore, AOX purified from female rat liver through the post-benzamidine stage could be reduced with DTT to yield an enzyme with K m(app) of 261 M. Reoxidation of the reduced enzyme with 4,4Ј-DTDP converted the K m(app) to 1673 M (Fig. 6C). Thus, the capacity to modulate K m(app) of rat liver AOX was maintained through purification of the enzyme and may therefore reflect an intrinsic property of the enzyme.

DISCUSSION
In the present work have assembled full-length sequences for AOX cDNAs from male and female rat liver. We examined expression of AOX by Northern blot analysis from males and females, from several tissues, and have examined the effect of androgen regulation. We purified AOX from rat livers of males and females and obtained a single 150-kDa band by SDS-PAGE analysis consistent with the deduced amino acid sequences from males or females of 1,333 amino acids and 147 kDa. Antisera raised against synthetic decapeptides reacted with AOX from males or females. Furthermore, K m(app) values for crude extracts of male or female rat liver and post-benzamidine purified AOX differed substantially but could be interconverted by chemical reduction with DTT or oxidation with 4,4Ј-DTDP.
The deduced amino acid sequences for male and female rat liver AOX are 81 and 82% identical to human and bovine AOXs and are themselves 99.6% identical. They show excellent conservation in the domains attributed to iron, FAD, and MoCo binding. By this criterion, the rat AOXs clearly belong to the molybdenum iron-sulfur flavoproteins that include AOX, XDH, and XO. Furthermore, the rat AOXs conform to the domain models proposed for XDHs (7,30,31). Amino acids critical for catalysis of this class of enzymes are conserved in the rat AOX sequences as they are in most other AOXs and XDHs. In particular, Glu-869 of the Mop enzyme is critical for catalysis and is conserved in nearly all XDHs and AOXs (see Ref. 21 for alignment of several sequences), including the two rat sequences reported here where it is found at amino acid 1265 in MoCo domain 5 (Fig. 2).
We observed that the male and female cDNA sequences were not identical. While several arguments could be advanced to explain the differences, we suggest that the least tenable ar-FIG. 4. Testosterone does not regulate steady state levels of rat hepatic AOX. Poly(A) ϩ RNA from individual male rat livers has been analyzed by Northern blot. Individuals from the following four groups have been used: castrated, sham castrated, castrated and testosterone supplemented, castrated and growth hormone supplemented. ␤-Actin was used as a control to reveal uniform loading of RNA. Blots were first probed with the AOX probe, the corresponding region cut from the filter for counting, and the remaining filter was dissociated of all 32 P and rehybridized with the ␤-actin probe.  Fig. 4 were cut from the filters and counted by liquid scintillation counting. Counts from randomly selected regions were averaged and subtracted from each signal to account for background radioactivity. Counts for each AOX signal were then normalized to the ␤-actin signal derived from the same lane, and the normalized, background subtracted counts were multiplied by 100. The number of animals in each group is shown by the n. The mean and standard error of the mean (SE) were calculated for each group. No statistical significance could be established between group means. gument is for the expression of two different AOX genes in the livers of male and female rats. Nucleotide differences between males and females could arise during PCR amplification, cloning, or sequence analysis since this is also PCR based. The small number of differences observed between male and female sequences could also arise from allelic differences between males and females or simply from individual differences between rats. Sequence analysis of only one additional male cDNA uncovered two of the changes (nt 405 and 408) found between the male and female clones. The minimal number of differences found between males and females coupled with the individual variation already observed suggests that the differences are unlikely to represent expression of alternative genes. Furthermore, sequence polymorphism is a well described phenomenon in Drosophila XDH (32)(33)(34). Multiplicity of AOX genes has been established for some organisms (19 -24), and in plants these data are supported by multiple cDNA sequences of approximately 80% identity (19,20). Thus, the observation that mice appear to express two AOX genes in the liver was not surprising (9). However, genetic data derived in the mouse have not been supported yet by corresponding molecular data and few efforts have been made to establish AOX gene copy number in vertebrates. Southern blot analysis of chromosomal DNA failed to reveal second AOX genes in several vertebrates under conditions in which at least 80% identity would have produced hybridization (28).
Northern blot analysis of AOX mRNA from male or female rats, and from several different tissues, demonstrated expression of a single 4,500 nt RNA. We did not observe variation in size or multiplicity of AOX mRNAs in the liver from males or females where kinetically distinct forms of AOX were identified. Furthermore, castration and/or testosterone supplementation resulted in no significant alteration in AOX mRNA abundance and we infer that testosterone does not appear to exert significant regulation of AOX mRNA abundance or form in the rat liver. We did observe surprisingly strong hybridization to an RNA of 2,500 nt that could be localized to the region from ϩ1,682 to ϩ2,217. While the identity of this RNA is unknown, it is too small to encode an AOX of 150 kDa. Thus, RNA analysis supports expression of a single AOX gene in the liver where post-translational events may be important for determining the differences between male and female kinetic variants.
Interestingly, AOX-3 from A. thaliana encoded a protein of only 568 amino acids truncated at the amino terminus (20). It must differ from a true AOX because it cannot encode a protein capable of binding the full set of co-factors. AOX-3 from A. thaliana showed striking homology to AOX-1 and AOX-2 in the MoCo-binding region, confirming that it is indeed a member of the MH family. Confirmation that vertebrates encode such a protein would be of great interest since it may represent the RNA detected at 2,500 nt.
Since our observations suggested that rats express a single AOX gene in the liver, we examined the possibility that redox status might underlie the differences in male and female variants. We found that partially purified crude extracts of male and female rat liver did indeed reveal different K m(app) variants of AOX. Reduction of crude extracts from males or females with DTT resulted in conversion to a single form. Subsequent reoxidation of the reduced AOX with 4,4Ј-DTDP resulted in conversion to a more female like K m(app) . Reduction or oxidation of post-benzamidine purified AOX also resulted in interconversion between the two extremes of K m(app) suggesting that the variants differed by their intrinsic oxidation state. Since DTT directly reduces protein disulfides to thiols while 4,4Ј-DTDP forms disulfides from thiols (35)(36)(37)(38), we infer that manipulation of thiol oxidation state can interconvert kinetic variants of AOX.
Redox effects on XDH are well known. Conversion between the NAD ϩ dependent, D-form, and the oxygen dependent, Oform, is a redox-dependent process reversible by chemical reduction (7,39,40). Furthermore, redox effects on XDH have dramatic effects on the kinetic parameters of the enzyme (41). AOX does not have an NAD ϩ -dependent form of the enzyme, and therefore conversion between D-form and O-form is irrelevant. While cysteine residues critical for D-form to O-form conversion in XDH were not conserved in AOX (42), most of the 41 cysteine residues found in rat liver AOX are conserved with vertebrate XDHs. These cysteine residues would be expected to take part in the same biochemical reactivities as those found in XDH. Since we observed that redox effects on K m(app) were preserved from crude extracts through post-benzamidine-puri- FIG. 5. Analysis of rat liver AOX1 protein. A, male rat liver AOX was purified to homogeneity from initially reduced extracts as described under "Materials and Methods." Purified AOX was analyzed by SDS-PAGE and stained with Ponceau S prior to subjecting the enzyme to amino-terminal sequence analysis. B, antibody was raised against synthetic decapeptides from the amino (AOX-NT) and carboxy (AOX-CT) termini. AOX was partially purified to retain numerous unrelated proteins, analyzed by SDS-PAGE (Crude AOX), and reacted to each antibody as described under "Materials and Methods." Preimmune, AOX-NT, and AOX-CT antisera were used at a 1:500 dilution. C, male and female rat liver AOX was purified to homogeneity from initially reduced preparations. Enzymes were analyzed by SDS-PAGE and subjected to Western immunoblot analysis using the AOX-CT antisera. Both male and female preparations migrated with apparent size of 150 kDa and both preparations reacted to the AOX-CT antisera.
fied enzyme and that these effects alone were sufficient to explain the differences between male and female variants, we infer that the different K m(app) variants result not from expression of alternative genes but from redox effects that may act either on AOX itself or on a closely associated protein capable of modulating K m(app) for AOX.
Our observations can be explained by the activity of the hepatic microsomal monooxygenase. The cystamylating flavin monooxygenase catalyzes oxidation of cysteamine, a thiol, to cystamine, a disulfide, and this reaction provides a significant source of disulfide responsible for maintaining the intracellular thiol:disulfide potential (43,44). The thiol:disulfide potential is thought to reflect two ratios: the GSH:GSSG ratio and the cysteamine:cystamine ratio. Protein oxidation state depends on overall thiol:disulfide potential. Higher levels of monooxygenase lead to a more oxidizing environment, while reduced monooxygenase levels result in a more reducing environment. Significantly, monooxygenase from mice or rats is regulated by testosterone in precisely the fashion that kinetic variants of AOX are regulated (43). Male rat hepatic monooxygenase levels are lower than female levels. Castration of male rats elevates monooxygenase and leads to elevated cystamine and a more oxidizing cytosol. Testosterone reverses this effect and restores the reducing environment. Furthermore, testosterone treatment of females lowers monooxygenase and elevates cysteamine levels. These observations may explain both the effect of testosterone and the ability to purify kinetically distinct forms of AOX from male and female rats since neither thioltransferase nor glutathione reductase, the two other enzymes establishing thiol:disulfide potential, are regulated by testosterone (43). We posit that AOX is sensitive to the thiol:disulfide potential and that kinetic variants of AOX reflect the intracellular thiol:disulfide potential through thiol modification of AOX. Activity of the monooxygenase could thereby regulate AOX K m(app) by "setting" the thiol:disulfide potential leading to a more oxidized form of AOX in the female or a more reduced form of AOX in the male. This may have direct consequences for ROS generation from AOX since the kinetically less efficient enzyme may bias ROS generation for superoxide anion. This argument does not require expression of a second AOX gene and can account for the failure of testosterone to regulate AOX gene expression in rats despite finding kinetically distinct forms in males and females. AOX thiol modification could also explain both novel electrophoretic variants (9) and the widely variant kinetic characteristics of AOX from different organs and different species (16,(45)(46)(47).
FIG. 6. Chemical reduction or oxidation abolishes differences in K m(app) between male and female rat liver AOX. AOX activity was purified from unreduced male and female rat livers through (NH 4 ) 2 SO 4 fractionation and either treated with 5 mM DTT or not to produce the crude native or crude reduced preparations. A separate preparation was reduced initially and purified through benzamidine fractionation to produce the post-benzamidine preparation. A, Lineweaver-Burk plots were determined for male (squares) and female (diamonds) preparations using NMN as the oxidizing substrate. B, Lineweaver-Burk plots were determined for initially reduced male and female crude extracts. In addition, the reduced female extract was then treated with 1 mM 4,4Ј-DTDP to reoxidize it (open squares) and subjected to Lineweaver-Burk analysis. C, female rat liver AOX was purified through benzamidine fractionation from an initially reduced extract. Enzyme was treated with 5 mM DTT or 1 mM 4,4Ј-DTDP to reduce or oxidize the enzyme. Lineweaver-Burk analysis was then conducted on the reduced (diamonds) or oxidized (squares) preparations.

TABLE IV
Effect of reduction and oxidation on K m(app) for crude and purified rat liver AOX Lineweaver-Burk plots shown in Fig. 6 were fitted by least squares analysis. K m(app) were determined for each curve and the goodness of fit r 2 values were determined for each.