Primary Structure Deduced from Complementary DNA Sequence and Expression in Cultured Cells of Mammalian 4-Hydroxyphenylpyruvic Acid Dioxygenase THAT THE ENZYME IS A HOMODIMER OF IDENTICAL SUBUNITS HOMOLOGOUS TO RAT LIVER-SPECIFIC ALLOANTIGEN F*

EVIDENCE 4-Hydroxyphenylpyruvic acid dioxygenase is an im- portant enzyme in tyrosine catabolism in most orga-nisms. From porcine and human liver cDNA libraries we isolated complementary DNA inserts for the en- zyme. Protein sequence analysis of the porcine enzyme revealed a block of the amino terminus of the mature enzyme. Comparison of the amino acid sequence deter- mined by Edman degradation of peptides derived from porcine liver 4-hydroxyphenylpyruvic acid dioxygen- ase with the nucleotide sequences revealed the primary structure of the porcine and human enzymes. The ma- ture human and porcine enzymes have an 89% amino acid sequence identity in amino acid residues and are composed of 392 amino acid residues. A computer- assisted homology search revealed that the enzyme is 88% identical in amino acid sequence to rat liver- specific alloantigen F. A monoclonal antibody (mob 51), which can immu- noprecipitate both the human and porcine enzymes, was developed. Cultured BMT-10 cells transfected with the cDNA insert of the human enzyme, using the expression vector pCAGGSneodE, produced a polypeptide with an M, of 43,000, which was immunoprecipi- tated with mob 51. Enzymic activity of the enzyme was detected in the transfected cells but not in the mock transfected cells. These findings suggest that the hu- man 4-hydroxyphenylpyruvic acid dioxygenase is a homodimer of two identical subunits with an M, of 43,000. Liver-specific alloantigen F seems to be closely related to the enzyme or possibly to the subunit of the enzyme itself. Elucidation of the complete amino acid sequence of the enzyme is expected to reveal structure- function relationships of this metabolically important on

4-Hydroxyphenylpyruvic acid dioxygenase is an important enzyme in tyrosine catabolism in most organisms. From porcine and human liver cDNA libraries we isolated complementary DNA inserts for the enzyme. Protein sequence analysis of the porcine enzyme revealed a block of the amino terminus of the mature enzyme. Comparison of the amino acid sequence determined by Edman degradation of peptides derived from porcine liver 4-hydroxyphenylpyruvic acid dioxygenase with the nucleotide sequences revealed the primary structure of the porcine and human enzymes. The mature human and porcine enzymes have an 89% amino acid sequence identity in amino acid residues and are composed of 392 amino acid residues. A computerassisted homology search revealed that the enzyme is 88% identical in amino acid sequence to rat liverspecific alloantigen F.
A monoclonal antibody (mob 51), which can immunoprecipitate both the human and porcine enzymes, was developed. Cultured BMT-10 cells transfected with the cDNA insert of the human enzyme, using the expression vector pCAGGSneodE, produced a polypeptide with an M, of 43,000, which was immunoprecipitated with mob 51. Enzymic activity of the enzyme was detected in the transfected cells but not in the mock transfected cells. These findings suggest that the human 4-hydroxyphenylpyruvic acid dioxygenase is a homodimer of two identical subunits with an M, of 43,000. Liver-specific alloantigen F seems to be closely related to the enzyme or possibly to the subunit of the enzyme itself. Elucidation of the complete amino acid sequence of the enzyme is expected to reveal structurefunction relationships of this metabolically important *This work was supported in part by a grant-in-aid from the National Center of Neurology and Psychiatry (NCNP) of the Ministry of Health and Welfare, Japan and from Fujita Health University. Some data were presented at the annual meeting of the Japanese Society for Inborn Errors of Metabolism, Takamatsu, Japan, November, 1990. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence($ reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number ($ 1113390. To whom correspondence should be addressed Dept. of Pediatrics, Kumamoto University Medical School, Kumamoto 860, Japan. Tel.: 096-344-2111 (ext. 5654); Fax: 096-366-3471. enzyme and to shed light on inherited disorders related to tyrosine metabolism, especially tyrosinemia types 1 and 3.

4-Hydroxyphenylpyruvic acid
dioxygenase (HPD)' (EC 1. 13.11.27), an enzyme that participates in the catabolism of tyrosine in most organisms, is present in liver and kidney. Homogentisic acid is produced from 4-hydroxyphenylpyruvic acid by HPD, and the reaction involves decarboxylation, oxidation, and rearrangement. The enzyme has been isolated from porcine (l), human (2), and avian livers (3), and the properties of the enzyme have been extensively characterized ( 2 , [4][5][6][7][8][9][10][11][12]. Purification studies suggested that the human enzyme is a homodimer of identical subunits with an M, of 43,000 ( 2 ) , whereas the porcine liver enzyme was found to be composed of two nonidentical subunits with a similar M , of 44,000 (1). The enzyme from these sources contains iron; part of the enzymic activity is restored by the addition of Fez', and the amino terminus of the enzyme is apparently blocked. The primary structure of the enzyme, from any source, has not been reported. Molecular cloning of the mammalian HPD and elucidation of the primary structure of the enzyme is expected to provide a basis for elucidating molecular mechanisms involved in the oxidation of 4-hydroxyphenylpyruvic acid.

Isolation of Porcine Liver HPD and Development of Antibodies-
For antibody production and sequence analysis, porcine liver HPD was purified as previously described (5) but with some modifications (13). The enzyme activity was eluted from a Mono Q column as three peaks of isozymes, the same as for the human liver enzyme (2). The enzyme in the first peak was additionally purified on a Superose column (13).
Antiserum and specific IgG directed to the enzyme protein were prepared as described (13). For preparation of mouse monoclonal antibody directed against the enzyme protein, mice were immunized a t 2-week intervals with 25-50 pg of the purified protein, using Freund's adjuvant, and hybrid cells were isolated, as described (14). These hybrid cells were monitored for production of IgG that crossreacted with the porcine liver enzyme. Mouse IgG was purified from ' The abbreviations used are: HPD, hydroxyphenylpyruvic acid dioxygenase; HPLC, high performance liquid chromatography; PE, pyridylethyl. ascites fluid, as described (15) and was used for additional experiments.
Protein Sequence Analysis-Porcine H P D was reduced with tri-nbutylphosphine (Wako Pure Chemical Co., Osaka, Japan) (16) and pyridylethylated with 4-vinylpyridine (Tokyo Kasei, Kogyo, Japan) (17) in the presence of 7 M guanidine hydrochloride containing 0.1 M Tris-HCI (pH 8.5) and 1 mM EDTA and kept overnight in a dark room at 25 "C. The reaction mixture was separated by gel permeation HPLC on tandem columns of TSK G3000 S W X L (7.8 X 300 mm each, Tosoh, Tokyo, Japan) in 6 M guanidine hydrochloride containing 10 mM sodium phosphate (pH 6.0), and then S-pyridylethyl (PE)protein was desalted by reversed-phase HPLC using a Bakerbond WP-C4 column (4.6 X 50 mm, J. T. Baker Inc., Phillipsburg, NJ). Methionyl bonds of the PE-protein were cleaved with cyanogen bromide in 70% formic acid, as described by Gross (18). Peptides were primarily separated by gel permeation HPLC on a TSK G2000 SW XL (7.8 X 300 mm, Tosoh, Tokyo, Japan). Peptide fractions were desalted and further separated by reversed-phase HPLC on a column of Senshu Pak VP304 (2.1 X 125 mm, Merck, Germany) with gradients of acetonitrile into 0.1% aqueous trifluoroacetic acid (19).
Amino acid compositions of the intact S-PE-protein and cyanogen bromide peptides were determined in a Hitachi L-8500 amino acid analyzer or with a Waters Pico-tag system (20,21). Sequence analyses were carried out in an Applied Biosystems 470A protein sequenator (22) connected to a 120A phenylthiohydantoin analyzer.
Oligonucleotide Probes-Mixtures of oligonucleotide probes corresponding to the appropriate portions of the peptides were synthesized and radiolabeled at the 5' ends, with [-y-:"P]ATP (specific activity 3,000 Ci/mmol) and T4 polynucleotide kinase (23). The sequences of the parts of the peptides and corresponding sense or antisense oligonucleotides probes (mixtures) were as follows: probe 1, C(C/G)AA(A/

A/T/C)GC(G/A/T)AT for Ile-Ala-Leu-Lys-Thr-Glu-Asp.
Screening of cDNA Libraries-A porcine liver cDNA library, constructed by inserting cDNA copies of poly(A') RNA from porcine liver into the EcoRI site of bacteriophage vector hgtll, was purchased from Clontech Laboratories, Inc. (Palo Alto, CA). Approximately 6 X 10" recombinant phage plaques from the cDNA library were screened by hybridization with radiolabeled nucleotide probes. Prehybridization, hybridization, and washing of nitrocellulose filters were carried out a t 40 "C. Other conditions were as described elsewhere (24). The second screening of porcine and human cDNA libraries with the cDNA insert as a probe was carried out as described (25). A human liver cDNA library, constructed by inserting cDNA copies of poly(A+) RNA from human liver into the EcoRI site of bacteriophage vector Xgtll, was purchased from Clontech Laboratories, Inc.
DNA Sequence Analysis-Restriction fragments of the cDNA inserts were subcloned into appropriately digested pUC18. The DNA sequences were determined by the dideoxy chain termination method (26) (kit obtained from Takara, Kyoto, Japan) using alkali-denatured plasmids as templates (27).
Plasmid Construction and DNA Transfection-We used the plasmid vector pCAGGSneodE (28), which contained the neomycin resistant gene, the cytomegalovirus intermediate early enhancer, the chicken /%actin promoter, and a polyadenylation signal. Part of the cDNA insert of the human HPD, which contained entire coding, 11base pair 5' noncoding, and 23-base pair 3' noncoding sequences was inserted into the EcoRI site of the plasmid vector pCAGGSneodE. HMTlO cells derived from a monkey cell line (29) were grown in KPMI 1640 (GIBCO) supplemented with 10% fetal calf serum (GIBCO), maintained in a 5% CO, atmosphere at 37 "C, and used for the transfection experiments. Transfection of DNA was performed using lipofection (Bethesda Research Laboratories, Gaithersburg, MD) (30), according to instructions from the supplier. The transfected cells were treated as described (31) and then were cultured for 21 days in the presence of G418 and collected using a rubber policeman. Cell extracts were prepared (15,32) and used for protein assay, immunoprecipitation, immunoblot analysis, and enzyme assay. For mock transfection, the expression vector alone was used.
Enzyme Assay, Immunoprecipitation, and Immunoblots-The enzyme activity of HPD was measured as described (32, 33) using radiolabeled 4-hydroxyphenylpyruvic acid as a substrate. Immunoprecipitation of the HPD protein in the extract of liver or transfected cells was carried out using the monoclonal antibody as follows. Tissue and cells were homogenized in 50 mM potassium phosphate buffer (pH 7.4) and centrifuged a t 10,000 X g for 20 min a t 4 "C. The enzyme-IgG complex was precipitated with Protein A-agarose (Pharmacia LKB Biotechnology Inc.) as described (15), and then the proteins bound to the agarose were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (34). Immunoblot analyses with anti-HPD rabbit IgG were carried out by the method of Towbin et al. (35) as described (13).

RESULTS
The conventionally purified HPD from porcine liver contained a single species of polypeptide with an MI of 43,000 ( Fig. 1). The results of sodium dodecyl sulfate-polyacrylamide gel electrophoresis and gel filtration on a calibrated Superose column suggested that the purified porcine HPD was a homodimer of subunits with an M , of 43,000. The purified porcine liver HPD was used for antibody production.
An attempt at immunoaffinity isolation of porcine HPD was made using the mob 51, one of the monoclonal mouse IgGs that recognizes porcine HPD. In this experiment, a partially purified HPD preparation (DEAE-Sephadex step, Ref. 5) was applied on a Sepharose gel column immobilized with mob 51 (2 mg/ml gel). The column was washed with 50 mM Tris-HC1 (pH 7.4) containing 500 mM NaCl, and proteins were eluted with 100 mM NaPCOs (pH 11.0). A single species of protein with an M, of 43,000 was found in the eluate obtained by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (data not shown).
Immunoprecipitation of human HPD from crude human liver homogenates was carried out using the immobilized mob 51. When the immunoprecipitate was analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and immunoblots were analyzed using conventional antiserum, an immunostained protein migrated slightly faster than did the porcine HPD (Fig. 1). Among the monoclonal mouse IgGs, only mob 51 cross-reacted with human HPD in the liver homogenate. Thus, the immunochemical procedure provided evidence that porcine and human HPD are composed of a single subunit with an MI of 42,500-43,000.  2-4). For immunoprecipitation, crude extracts of human and porcine liver were incubated with mob 51 (3 pg) for 60 min a t room temperature, and then 50 p1 of protein A-agarose (Pharmacia) suspended in Tris-HCI buffer (pH 7.4, 1:1, v/v) was added to the mixture. The incubation was continued for another 60 min, and then the preparation was centrifuged a t low speed. The gel pellet was washed three times with ice-cold phosphate-buffered saline solution, and the protein associated with the gel was analyzed by sodium dodecyl sulfatepolyacrylamide gel electrophoresis, followed by immunoblotting, using anti-HPD rabbit IgG (22). Lane 36 Sequenator analysis of the porcine S-PE-protein (70 pmol) yielded no phenylthiohydantoin in three cycles of Edman degradation. Thus, the amino terminus of porcine HPD was blocked.
Amino acid sequences of seven peptides (denoted by the  prefix M: M2, M3, M4, M5, M6, M7, and M8) were determined partially or completely through the carboxyl terminus ( Table I). On the basis of the amino acid sequences, several oligonucleotide probes were synthesized and used to screen the porcine liver cDNA library. Among the oligonucleotide probes, the oligonucleotides named probe 4,5, and 6 produced three positive clones. Each of these recombinant phage clones produced positive signals with the three oligonucleotide probes.
The nucleotide sequences of the inserts from these clones and deduced amino acid sequences revealed that the cDNA inserts encode the porcine HPD precursor protein (Figs. 2  and 3). We repeated screening of the same library and obtained additional phage clones. The nucleotide sequences of the cDNA inserts of these phage clones were determined. The human liver cDNA library was then screened with the cDNA insert obtained from the porcine liver library to obtain phage clones carrying inserts that covered the entire coding region of human HPD (Fig. 2).
Two phage clones named XhpdPl from the porcine library and XhpdH31 from the human library were analyzed in detail.  Both the porcine and human cDNA inserts contained an open reading frame that could be translated to 393 amino acid residues (Figs. 3  and 4). The nucleotide sequence surrounding the putative initiation codon in the human cDNA insert was similar to the consensus sequence described by Kozak (36). A polyadenylation signal was present at 180 base pairs downstream of the stop codon of TAG in human cDNA. Northern blot analysis of mRNA indicated only a single species of mRNA was present in the human liver, and size of the mRNA for human HPD was estimated to be 1.7-1.8 kilobases long (data not shown).
The predicted primary structures of porcine and human HPD precursor proteins were compared with the partial amino acid sequences of peptides of the porcine enzyme. As shown Fig. 3, these sequences were the same in six regions, residues 83-118, 151-174, 194-224, 230-273, 341-365, and 341-365 of the predicted sequence, thereby confirming that the isolated cDNA insert codes for porcine HPD. Among eight peptides (denoted by the prefix M) isolated from a digest of the S-PE-protein (1.5 nmol) with cyanogen bromide, the amino acid composition of peptide M1 resembled that of residues 2-82 of the amino acid sequence predicted from cDNA inserts (Fig. 3). Sequenator analysis showed that the amino terminus of peptide M1 was also blocked (not shown).
After digestion of peptide M1 by Achromobacter protease I,    These results taken together show that the translation product of porcine HPD is processed at both the amino and carboxyl termini; the putative initiation methionyl residue is removed, and the newly formed amino-terminal threonyl residue seems to be blocked by an acetyl group (37). It is most likely that the mature porcine HPD is composed of 386 amino acid residues (residues 2-387) with an acetyl group at the amino terminus. Based on this, the relative M, of the subunit of the human enzyme was calculated to be 44,276.7, a value in a good agreement with the 42,500-43,000 estimated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis.

A~6 !~P r s A ! a P r~~!~l y p l _ y S L y~L y~~~~~!~! ! _ e G ! n G l u T y~V a I A s g T~! A s _ n G ! y G ! y A l~G !~V a !~!~H
Comparison of the predicted amino acid sequences of the porcine and human enzymes revealed that both the enzymes are highly homologous, with an 87.5% overall identity. The homology of nucleotide sequences between the porcine and human HPD cDNAs was 88.6% in the coding region.
Transfection and expression experiments of HPD were done to determine whether the cDNA of HPD was indeed coded for the mammalian HPD protein. When the cDNA insert from the human liver cDNA library was transfected into BMT-10 cells, a considerable amount of immunoreactive protein with an M, 43,000 to the mob 51 was synthesized (Fig.  5). On the other hand the mock transfected cells produced no detectable amount of the immunoreactive polypeptide. Thus, the transfected cells synthesized a polypeptide with an M , of 43,000, which was immunoprecipitated with mob 51. Measurement of the enzymic activity of HPD in these cells indicated that the extract from the transfected cells contained the activity, whereas the mock transfected cells showed little activity (Fig. 6).
These results show that the cDNA insert carried all of the sequence information necessary for full expression of the enzyme protein and activity. It is also evident that the mature human of protein was composed of two identical subunits with an M, 43,000. DISCUSSION We obtained evidence that mammalian HPD is a homodimer of the identical subunit composed of 386 amino acid residues. Purification of the human HPD by other workers (8,12) indicated that the relative molecular mass of the active human enzyme is 87,000 and the subunit mass is 43,000, determined using polyacrylamide gels containing SDS. These data are in good accord with our results. It was reported that the amino terminus of human and avian HPD was blocked (2, 6); and our present data suggest that the amino-terminal threonyl residue seems to be acetylated.
Microheterogeneity of purified enzymes from various sources revealed three major forms of avian enzyme that are enzymatically active with isoelectric points of I = 6.0, I1 = 6.2, and I11 = 6.4 (6). The porcine enzyme was eluted from the Mono Q column as three peaks (5). A similar heterogeneity was noted for the human enzyme, and these forms were immunologically identical (2). Processing of the carboxylterminal amino acids seen in the present study might relate to heterogeneity of the enzyme protein.
A computer-assisted search for homology of the amino acid sequences suggested that the sequences are not homologous to any known oxidases, including pyruvate dehydrogenase (38) or branched-chain keto acid dehydrogenase (39). The human and porcine HPD are highly homologous to rat liverspecific alloantigen F (40) (Fig. 4).
Liver-specific antigen F, first reported in 1968, was immunoprecipitated by sera from mice immunized with watersoluble liver extracts of allogenic mice (41). Although biological function of the antigen is not well understood, this antigen is widely distributed among species of mammals (42-46). Part of the primary structure of rat antigen F was deduced by molecular cloning of the cDNA obtained by immunoscreening of the rat liver cDNA library with allo-antisera to the antigen (40). The present study revealed that the amino acid sequence of rat antigen F is highly homologous to human and porcine HPD, with an approximately 90% identity. The relative molecular masses of antigen F from mouse and human liver were reported to be 43,000 (45) and 44,000 (42), respectively, which is similar to the M , of 43,000 noted for mouse and human liver HPD (8,13). In addition, both antigen F and HPD (47) are expressed in the liver and kidney. These data suggested that HPD is a protein closely related to antigen F or is antigen F itself. There are three known types of hereditary tyrosinemias. In a mouse model for the type I11 tyrosinemia, HPD activity is genetically defective, and the subunit protein of HPD is undetectable by immunoblot analysis of liver extracts (13). Our preliminary study on tyrosinemic mice' indicated that (i) tyrosinemic mice lack the F antigen, and (ii) the hypertyrosinemic gene in the mice is located on chromosome 5 as is the F antigen (48). Molecular analysis of tyrosinemic mice indicated that mRNA related to the HPD cDNA was absent in the liver? All of these data support our working hypothesis that HPD enzyme protein is F antigen itself.
Molecular cloning of the mammalian HPD and elucidation of the primary structure of the enzyme are expected to provide a basis for elucidating molecular mechanisms involved in the oxidation of 4-hydroxyphenylpyruvic acid and the molecular events related to disorders of tyrosine metabolism.