Evidence for a Self-catalytic Mechanism of 2,4,5=Trihydroxyphenylalanine Quinone Biogenesis in Yeast Copper Amine Oxidase*

Copper amine oxidases are representative of a new class of redox enzymes that contain a peptide-bound quinone cofactor, generated by posttranslational modification of amino acid side chain(& We have investi- gated the mechanism for the biogenesis of 2,4,5-trihy-droxyphenylalanine quinone (TPQ) in amine oxidase with two site-specific mutants of the yeast methylamine oxidase. Our results show that the capacity for TPQ for- mation in vivo is abolished when a putative ligand to copper, His-456, is changed to Asp; this H456D mutant binds copper at a low level (-4.1%), relative to the wild-type protein. In contrast, altering the active site consensus sequence that contains the precursor tyrosine does not affect TPQ production. The data implicate a self- catalysis mechanism for TPQ biogenesis, in which the protein-bound copper plays a key role. We propose that the minimal information required for TPQ biogenesis lies in a structural motif consisting of the copper site and the precursor tyrosine.

The self-catalysis mechanism has been suggested from a recent study in which a yeast amine oxidase gene that was cloned from Hansenula polymorpha (7) was expressed to produce an active and fully processed enzyme in Saccharomyces cereuisiae, an organism that does not appear to contain any endogenous copper amine oxidases (6). However, the enzymatic mechanism could not be unambiguously excluded in the prior study. It is clear that the modification of tyrosine to form TPQ is a highly specific event in which a single tyrosine residue per polypeptide chain of -80 kDa is converted to TPQ. Using our expression system for H. polyrnorpha methylamine oxidase (61, we have now investigated the determinants of specificity for this posttranslational modification process by examining the roles of the protein-bound copper and the tyrosine-containing active site consensus sequence in TPQ biogenesis.

EXPERIMENTAL PROCEDURES
Site-directed Mutagenesis-The site-directed mutagenesis was performed based on the method described by Kunkel et al. (8). Reagents and bacterial strains were provided by a Muta-Gene phagemid in vitro mutagenesis kit (Bio-Rad). The wild-type gene for methylamine oxidase from H. polymorpha (7) was inserted in plasmid pTZ19R as a HindIII were 5'-GCTGCCAATTACmTACTGTCTGTACTGGG-3' and 5'-fragment (6). The sequences of the synthetic mutagenic oligonucleotides CGCTCACAACWCAGCACC-3' for mutants E406N and H456D, respectively. The mutated nucleotides are shown in boldface and the mutated codons underlined. The amino acid residues were numbered according to Bruinenberg et al. (7). T7 RNApolymerase was used for the synthesis of the complementary strand. Clones were screened for the desired mutation by DNA sequencing using the dideoxy DNA sequencing method with Sequenase version 2.0 (U. S. Biochemical Corp.). The sequence of the synthetic sequencing primer was 5'-GCTTTTCAAG-CACTCTGACTTCAGAG-3', which hybridizes to a sequence 74 bases upstream from Glu-406 codon and 224 bases upstream from His-456 codon in the gene of the yeast methylamine oxidase (7). The mutated gene was isolated as a HindIII fragment following digestion of the selected plasmid with HindIII and subcloned into the expression vector pDB2O at the HindIII site. The plasmid with the mutant gene inserted at the correct orientation was selected, purified, and used to transform S. cereuisiae CG379 as described previously for the WT gene (6).
Yeast Cell Culture and Protein Purifzation-The expression, yeast cell culture, and protein purification of mutant proteins were performed according to the procedures described for the WT protein (6) with the following modifications in the purification procedure. The DEAE resin was DEAF,-Sepharose CL-GB from Pharmacia Biotech Inc. Following DEAE chromatography, the concentrated sample was separated further on a gel filtration column (75 cm long, 1.5 cm in diameter) packed with Sephacryl S-300 HR (Pharmacia Biotech Inc.). The column was preequilibrated with 10 m M potassium phosphate, pH 6.7, and washed with the same buffer to elute proteins. The most active fractions were pooled, concentrated, and stored at -20 "C. This pool of protein sample generally appeared >90% pure, as judged on SDS-polyacrylamide gels. Less active fractions were concentrated and reloaded onto the gel filtration column. The fast protein liquid chromatography step described in the original procedure (6) was not included for the purification of E406N protein because we found that it did not increase the enzyme specific activity in the final purified protein preparations. In the case of H456D mutant where no enzyme activity was detected, fractions from each steps and the protein purity were analyzed by SDS-polyacrylamide gel electrophoresis.
Copper and Protein Concentration Determination-Copper content was analyzed by the standard addition method according to the procedure described previously (6). The purified and dialyzed H456D protein was diluted with distilled and deionized water to a final concentration of 16.7 pg/ml with copper standard (Fisher) added to 0-28 ng/ml. Due to the instability of the mutant protein in water, the standard addition was also performed in the following way: For each reading, 0-42 ng of copper standard prepared in 16 pl of deionized water or 0.05 M HNO, was injected into the graphite tube, followed by 10.5 pg ofthe purified H456D protein sample in 4 pl of 10 m~ potassium phosphate, pH 6.7. The stoichiometry of protein-bound copper was calculated based on a molecular mass of 151.5 kDa for the dimer (see "Results and Discussion") and protein concentrations determined using the Bio-Rad protein assay with bovine serum albumin as the standard.
Titration with Phenylhydrazine and Kinetic Measurements-Methods for the titration with phenylhydrazine and for measuring ethylamine and benzylamine oxidation were as reported (6). The quinone stoichiometry of the E406N mutant protein was calculated based on a molecular mass of 152 kDa for the dimer and corrected for protein purity. The steady-state kinetic parameters were calculated by fitting data to the Michaelis-Menten equation using nonlinear regression.

RESULTS AND DISCUSSION
Mutation in the Copper Binding Site-Three histidines have been implicated as ligands to the copper in copper amine oxidases from spectroscopic investigations (9-11). We have found recently that 3 histidine residues are highly conserved in the primary sequences of copper amine oxidases at a fixed distance from the active site TPQ (5). Two such histidines are present as His-X-His, a motif well known to contain copper ligands in several copper proteins (12). In the protein sequence of the yeast methylamine oxidase from H. polymorpha, this motif is contained in the sequence His-Asn-His-Gln-His. The middle histidine His-456 is most certainly one of the copper ligands, although without a three-dimensional structure we cannot be certain which of the other 2 histidines performs a similar role (5).
We decided first to change the copper ligand field by replacing His-456 with an aspartic acid residue by site-directed mutagenesis. The mutated gene was expressed in S. cerevisiae and the protein purified. We found that the mutant protein was expressed to a level comparable with that of the WT protein based on analyses by SDS-polyacrylamide gel electrophoresis. The H456D mutant protein was eluted from the anion exchange and gel filtration columns in a manner virtually identical to that of the WT enzyme throughout the purification process; the yield for the purified mutant protein was also similar. These results indicate that the mutation does not affect the folding of the mutant polypeptide and that the protein remains as a dimer analogous to the WT enzyme.
We found that H456D protein does not exhibit any amine oxidase activity and, furthermore, contains no quinone moiety detectable by the redox-cycling staining method (13). Copper analyses indicate a low level (-4.1% of wild type), suggesting a greatly reduced affinity for Cu(I1) in this mutant protein. These results indicate that disruption of the native copper binding site generates a protein incapable of cofactor production. Consistent with the lack of quinone cofactor, the H456D protein lacks the characteristic pink color and absorption peak at 472 nm associated with the active WT enzyme (6) (Fig. 1).
To confirm the identity of the purified mutant protein, the N terminus of the protein was sequenced with a highly purified sample. The first 19 amino acid residues from the N terminus were identified as: N-Ala-Pro-Ala-Arg-Pro-Ala-His-Pro-Leu-Asp-Pro-Leu-Ser-Thr-Ala-Glu-Ile-Lys-Ala.
The first residue corresponds to the residue at position 17 in the DNA-derived protein sequence of the yeast methylamine oxidase (7). This N-terminal sequence and the starting residue match those of the WT protein, also analyzed in a similar way. These results indicate that the mutant protein we purified was indeed encoded by the yeast amine oxidase gene and, furthermore, that the N terminus of the translated polypeptide chain was processed in the same way as the WT to yield the mature polypeptide. Thus, the calculated molecular weight for the mature polypeptide chain should be 75,731 (77,435 for the DNA-   (dashed line). The spectrum of the E406N protein was taken with a purified protein sample of 0.9 mg/ml in 100 m~ potassium phosphate, pH 7.2. The specific activity of the sample was 0.14 unit/mg, determined at 37 "C in 5 m M benzylamine and 100 m M potassium phosphate, pH 7.2. The phenylhydrazone derivative was generated by titrating small amounts (a fraction of 1 pl) of a freshly prepared phenylhydrazine hydrochloride stock solution (5.55 or 55.5 m M in water) into this E406N protein solution. The enzyme solution was incubated at 37 "C after each addition of phenylhydrazine and its spectrum monitored. The spectrum shown was recorded after 3 molar eq of phenylhydrazine (relative to the subunit) was added. derived polypeptide chain).
The observed properties of H456D mutant indicate that alteration of a histidine believed to be essential to copper binding leads to greatly reduced copper affinity. Whether the carboxyl group of the aspartate residue at the mutated site continues to serve as a ligand to copper in H456D protein is presently unknown. In any case, the most profound effect of the change in the copper ligand field is the absence of the redox cofactor TPQ in the protein. The fact that the redox-cycling staining assay did not detect any reactive quinone moiety on the H456D protein indicates that the modification of tyrosine did not occur at all, since the redox-cycling staining method could also detect 3,4-dihydroxyphenylalanine (dopa) (131, the first intermediate in the proposed scheme for TPQ biogenesis (4). It is clear that TPQ biogenesis requires the presence of a hnctional copper binding site. Mutation in the Consensus Sequence-The above conclusion is substantiated by findings with another mutant protein in which the active site consensus sequence is altered. The glutamic acid residue linked to the carboxyl group of TPQ in the active site of the yeast methylamine oxidase was mutated and replaced by an asparagine residue through site-directed mutagenesis. The mutant was expressed and the protein purified. The results show that the E406N protein behaved the same as the WT enzyme throughout the purification procedure, indicating that similar to the H456D mutant, the mutation does not

Steady-state kinetics of the wild-type and the mutant E406N yeast methylamine oxidase
All parameters were obtained in 0.1 M potassium phosphate, pH 7.2, at 25 "C. The concentrations of ethylamine were in the range of 59.3 p~ to of 0.249-13.0 m~ for the mutant enzyme. affect the overall structure of the protein. The expressed E406N protein was stained positive for quinoprotein using the redoxcycling staining method (13) and was active in catalyzing amine oxidation (see below). The purified E406N protein has an absorption spectrum (Fig. 2) similar, although not identical, to that of the WT enzyme (Fig. 1) and can be titrated with the carbonyl reagent phenylhydrazine to give rise to an adduct with a n absorption spectrum typical of the phenylhydrazone derivative of TPQ in copper amine oxidases (Fig. 2). 1.9 mol of quinoneldimer was found based on the titration with phenylhydrazine. The stoichiometry is higher than that found routinely with the WT enzyme (6). Stoichiometries of quinone less than 2ldimer in WT enzyme have been attributed to protein instability (16), suggesting that E406N may be somewhat less sensitive to oxidative inactivation. The estimated molar extinction coefficient for this phenylhydrazone derivative is 40.9 m " l cm-l (normalized to moles of phenylhydrazine per enzyme subunit), also very close to the value obtained with the WT enzyme (6). We note that the absorbance of the E406N mutant does not return to base-line level above 600 nm (Fig. 2), which may contribute to a higher apparent absorbance at 472 nm relative to WT enzyme (Fig. 1). The enzyme activity of the E406N mutant was compared with the WT enzyme. The results show that consistent with the WT enzyme, E406N mutant is more active in catalyzing the oxidation of ethylamine than benzylamine (Table I). The mutation appears to have no effect on the kinetics of ethylamine oxidation but increases the K,,, for benzylamine, resulting in a 5-fold reduction in kJKm for benzylamine. The most dramatic difference between these two enzymes is in the oxidation of methylamine. E406N enzyme is inactivated after 19-21 turnovers, although the initial velocity and K,,, for methylamine appear to be similar to those of the WT enzyme under the same conditions. We presently do not know the mechanism for such a n inactivation. Nonetheless, the kinetic data indicate that the glutamate residue in the consensus sequence is not required for the amine oxidase activity, ruling out the possibility that the carboxylate side chain of this residue functions as the active site base to catalyze proton abstraction from the substrate Schiff base to form a product Schiff base in amine oxidation (17). This conclusion most likely holds true for those amine oxidases that have an aspartate in the active site consensus sequence in place of glutamate in the yeast methylamine oxidase (18).
The physical and kinetic properties of the E406N mutant indicate clearly that the replacement of the only charged residue in the active site consensus sequence to an uncharged one does not affect the capacity for TPQ biogenesis in vivo. Such a change would have had dramatic effects on this process, most likely leading to a n unprocessed and inactive mutant protein, if the active site consensus sequence were the recognition andlor signaling for the posttranslational modification of the precur- The sequence has been reported previously (6).
* The numbers are the residue number on the cDNA-derived protein e An unidentifiable phenylthiohydantoin-derivative. CmCys is a carboxymethylated cysteine residue. sequence (7). sor tyrosine. There is a possibility, however, that some Asn residues in the expressed E406N protein have undergone deamination to Asp since the occurrence of nonenzymatic deamination of Asn to Asp has been well documented (19). If this occurred, it would restore an active site sequence similar to one observed in most other amine oxidases (181, and the sequence Asn-Tyr-Asp could then serve as the recognition signal for the posttranslational modification of the precursor tyrosine. An active site peptide containing the phenylhydrazone derivative of TPQ was thus prepared and isolated from the E406N enzyme according to the procedure developed for the WT enzyme (6) in order to explore the possibility for deamination. The peptide was sequenced, and the result showed that it is an Asn that is present in the active site sequence as designed originally (Table 11). The possibility for deamination is therefore eliminated.
Conclusions-We have thus established that the preservation of the active site consensus sequence in the yeast methylamine oxidase is not necessary for TPQ biogenesis and have provided experimental evidence to support unequivocally our previous prediction that the specificity for TPQ biogenesis is not determined by the consensus sequence (6). If the tyrosine in the consensus sequence is the substrate of some posttranslational modification enzyme(s), the mutation of a conserved residue in the consensus sequence would most certainly interfere with the interaction of the enzyme(s) with the tyrosine residue and impair the formation of TPQ in the mutant protein to a certain extent. Furthermore, the total absence of TPQ in the H456D copper site mutant would not be expected because the mutation site is 51 amino acid residues from the precursor. These considerations argue strongly against the enzymatic mechanism in which a separate class of enzyme(s) catalyze the posttransla-tional modification of tyrosine in the copper amine oxidase.
The importance of the protein-bound copper for TPQ biogenesis is supported further by the demonstration that the H456D mutant, with its altered copper ligand field, is devoid of TPQ. This observation supports the previously proposed function for protein-bound copper in a self-catalytic mechanism for TPQ biogenesis (6). It is known that the protein-bound copper is absolutely required for the activity of copper amine oxidases (20,21). It has been proposed that copper is involved in the reoxidation of the reduced TPQ cofactor in the amine oxidase catalytic cycle, by mediating electron transfer between the reduced cofactor and molecular oxygen (22,231. Here, a new role is implicated for the protein-bound copper, Le. as a catalyst for tyrosine oxidation in the active site of the copper amine oxidases. In all of the known sequences of copper amine oxidases, the 3 histidines that are ligands to copper and the TPQ precursor tyrosine residue are located invariantly on a stretch of sequence of less than 100 amino acid residues in the carboxyl half of the protein sequence, and the spacing between these histidines and the tyrosine is highly conserved (3). These findings suggest that the structure of the copper site relative to the TPQ has been preserved through evolution. In the present study, we have shown that it is essential to maintain a functional copper binding site in order to form TPQ. We therefore propose that the sequence containing all of the histidine ligands to copper and the TPQ precursor comprises a three-dimensional structural motif, which is essential for TPQ formation in the active site of copper amine oxidases.