Purification and Properties of Avian Liver p-Hydroxyphenylpyruvate Hydroxylase

Avian liver p-hydroxyphenylpyruvate hydroxylase (EC 1.13.11.27) was purified to a 1000-fold increase in specific activity over crude supernatant, utilizing a substrate analogue, o-hydroxyphenylpyruvate, to stabilize the enzyme. The preparation was homogeneous with respect to sedimentation with a sedimentation velocity (s20,w) of 5.3 S. The molecular weight of the enzyme was determined to be 97,000 +/- 5,000 by sedimentation equilibrium, and the molecular weight of the subunits was determined to be 49,000 +/- 3,000 by sodium dodecyl sulfate polyacrylamide gel electrophoresis. Polyacrylamide gel electrophoresis revealed heterogeneity of the purified enzyme. The multiple molecular forms were separable by isoelectric focusing, and their isoelectric points ranged from pH 6.8 to 6.0. The amino acid compositions and tryptic peptide maps of the three forms isolated by isoelectric focusing were very similar. The forms of the enzyme had the same relative activity toward p-hydroxyphenylpyruvate and phenylpyruvate. Conditions which are known to accelerate nonenzymic deamidation of proteins caused interconversion of the multiple molecular forms. Iron was the only transition metal found to be associated with the purified enzyme at significant levels. The amount of enzyme-bound iron present in equilibrium-dialyzed samples was equivalent to 1 atom of iron per enzyme subunit. Purification of the enzyme activity correlated with the purification of the enzyme-bound iron. An EPR scan of the purified enzyme gave a signal at g equal 4.33, which is characteristic of ferric iron in a rhombic ligand field.


SUMMARY
Avian liver #-hydroxyphenylpyruvate hydroxylase (EC 1.13.11.27) was purified to a lOOO-fold increase in specific activity over crude supernatant, utilizing a substrate analogue, o-hydroxyphenylpyruvate, to stabilize the enzyme. The preparation was homogeneous with respect tn sedimentation with a sedimentation velocity (s~,,,~) of 5.3 S. The molecular weight of the enzyme was determined to be 97,000 f 5,000 by sedimentation equilibrium, and the molecular weight of the subunits was determined to be 49,000 + 3,000 by sodium dodecyl sulfate polyacrylamide gel electrophoresis.
Polyacrylamide gel electrophoresis revealed heterogeneity of the purified enzyme. The multiple molecular forms were separable by isoelectric focusing, and their isoelectric points ranged from pH 6.8 to 6.0. The amino acid compositions and tryptic peptide maps of the three forms isolated by isoelectric focusing were very similar. The forms of the enzyme had the same relative activity toward p-hydroxyphenylpyruvate and phenylpyruvate.
Conditions which are known to accelerate nonenzymic deamidation of proteins caused interconversion of the multiple molecular forms.
Iron was the only transition metal found to be associated with the purified enzyme at significant levels. The amount of enzyme-bound iron present in equilibrium-dialyzed samples was equivalent to 1 atom of iron per enzyme subunit. Purification of the enzyme activity correlated with the purification of the enzyme-bound iron. An EPR scan of the purified enzyme gave a signal at g = 4.33, which is characteristic of ferric iron in a rhombic ligand field.
The early studies on p-hydroxyphenylpyruvate hydroxylase' by La Du and Zannoni (2) revcalcd that ascorbate was required for the oxidation of p-hydrosyphenylpyruvate.
This requirement for ascorbate was nonspecific, being fulfilled by nonstoichiometric amounts of reduced 2,6-dichlorophenolindophenol as well as several other reducing agents including hydroquinone, isoascorbate, reduced coenzyme QlO, and a variety of hydroquinoidal 1 The name suggested for p-hydroxyphenylpyruvate hydroxylase by the 1972 Commission on Biochemical Nomenclature is 4-hydroxyphenylpyruvate dioxygenase (1).
compounds (2)(3)(4). Subsequently, Goswami (5) had shown that p-hydrosyphenylpyruvate hydroxylase can be reversibly inactivated by oxidation and can be reactivated by reducing agents. The reversible inactivation of the enzyme may reflect changes in the oxidation state of a transition metal prosthetic group. The existence of a metal prosthetic group had been suspected since certain metal chelators, o( ,oc'-dipyridyl, sodium azide, and diethyldithiocarbamate inhibited the enzyme (2, 6). The inhibition of the enzyme by diethyldithiocarbamate, a strong cupric copper chelator, suggested the participation of copper as the prosthetic group. La Du and Zannoni (2) noted that copper-deficient dogs were also deficient in liver p-hydroxyphenylpyruvate hydroxylase; however, the low enzyme level did not increase with in vitro addition of cupric copper. Evidence supporting ferrous iron as the metal prosthetic group has been presented by Goswami and Knox (7) and Goswami (5). Ferrous iron was reportedly removed from crude rat liver p-hydroxyphenylpyruvate hydroxylase by dialysis against o-phenanthroline and exhaustive dialysis against water, which resulted in a loss of activity. Contradictory evidence was obtained by Taniguchi and Armstrong (8), who noted that most of the enzyme activity was returned by the removal of the metal chelator by passing the enzyme through a bed of Sephadex G-50. Our own investigation of this question indicated that the enzyme activity was not significantly reduced by dialysis against chelators. Furthermore, o-phenanthroline and diethyldithiocarbamate both inhibit the enzyme through a strictly competitive mechanism, as evidenced by enzyme kinetics studies (9,10). Thus, if a metal prosthetic group is required for p-hydroxyphenylpyruvate hydroxylase activity, it must be tightly associated to the enzyme protein.
The purification of the avian liver enzyme was undertaken with the purpose of analyzing the purified enzyme for associated transition metals. Up until recently, this enzyme has resisted purification and exhibited extreme instability.
We wish to report a successful purification of avian liver p-hydroxyphenylpyruvate hydroxylase utilizing a substrate analogue, o-hydroxyphenylpyruvate, to stabilize the enzyme during purification. Some of the chemical and physical properties of this highly active, purified enzyme preparation are described, including the transition metal content of the enzyme.

AND DISCUSSION
The study of p-hydroxyphenylpyruvate hydroxylase has been hindered by the inability to obtain highly purified enzyme preparations which retain their enzymic activity. Lindblad et al. (12) purified the enzyme to a high degree of purity, a single band in polyacrylamide gel electrophoresis; yet, the increase in the specific activity was no better than loo-fold, an achievement previously reported by Edwards et al. (26) and Zannoni and La Du (27). Testing various methods of enzyme stabilization, we found that storage under nitrogen gas in the presence of a substrate analogue, o-hydroxyphenylpyruvate, significantly increased t'he stability of the enzyme. Furthermore, inclusion of 0.4 mM o-hydroxyphenylpyruvate in the buffer systems of the purification process resulted in increased yields and a 16-fold increase in the specific activity of the final preparation (Table I). o-Hydroxyphenylpyruvate was also found to be a required constituent of buffers used in electrophoretic procedures which depend upon the preservation of enzyme activity to mark the position of the enzyme. Without the inclusion of the substrate analogue, electrophoretic procedures such as polyacrylamide gel electrophoresis and isoelectric focusing resulted in proteins which were distributed in the same patterns but without activity. Our final preparation of avian liver p-hydroxyphenylpyruvate hydroxylase, the QAE fraction, was found to be homogeneous with respect to molecular weight and size. During gel filtration in Sephadex G-150, the enzyme activity eluted with the protein peak at a constant ratio, with no other protein peaks observed by optical density at 280 nm. During the sedimentation velocity experiment, a symmetrical boundary was observed to sediment at a velocity corresponding to an ~20,~ value of 5.3 S at 59,780 rpm. The equilibrium sedimentation data, a linear plot of log AY versus ~2, also indicated a homogeneous solute with a molecular weight of 97,000 f 5,000. The method of sodium dodecyl sulfate polyacrylamide gel electro-  (16), with 5 ccg of purified enzyme.
phoresis and staining with Coomassie brilliant blue revealed a single principal protomer with a molecular weight of 49,000 f 3,000 and some minor contamination (Fig. 1). A more serious challenge to the homogeneity of the enzyme preparation was raised when polyacrylamide gel electrophoresis without sodium dodecyl sulfate demonstrated heterogeneity, denoted by multiple, close banding of protein is stained gels. The multiple banding of purified enzyme was observed consistently, in several different enzyme preparations.
Furthermore, preincubation of the enzyme with 1 y. /I-mercaptoethanol and running the polyacrylamide gel electrophoresis with 0.1 y0 &mercaptoethanol in the upper buffer did not alter the multiple banding. Since each of these bands was demonstrated to possess enzyme activity by the elution of the activity from sliced gels and correlation of the peaks of activity with a scan of a stained gel from the same run, Fig. 2, the microheterogeneity was that of p-hydroxyphenylpyruvate hydroxylase and not that of contaminants.
It was also noted that, during the purification of the enzyme, QAE-Sephadex chromatography resolved the enzyme activity into multiple peaks, an observation that had been previously puzzling. These peaks may now be interpreted as representing different molecular forms of the enzyme. This observation of the microheterogeneity of avian liver p-hydroxyphenylpyruvate hydroxylase was confirmed by the separation of the purified enzyme into at least four active forms by isoelectric focusing. The isoelectric points of the four detectable forms were: I, 6.0; II, 6.2; III, 6.4; and small amounts of IV, 6.8 (Fig. 3). Since the ratio of p-hydroxyphenylpyruvate/phenylpyruvate activities was the same (approximately 45) for the four forms of the enzyme, the four forms of the p-hydroxyphenylpyruvate hydroxylase appear to be the same with respect to their substrate specificity. The fractions containing I, II, and III were separately pooled and dialyzed for 30 hours against three changes of 50 mM Tris-HCl, pH 7.6, to remove the ampholytes and buffer glycine for further structural analyses, The nature of the difference distinguishing each of the multiple molecular forms of p-hydroxyphenylpyruvate hydroxylase appeared to be that of charge, since this is the physical property by which electrophoretic and ion-exchange chromatographic methods separate proteins of the same molecular weight. The possibility of subunit polymerization through disulfide linkage was obviated by the observation that &mercaptoethanol did not influence the microheterogeneity and by the homogeneity of the enzyme with respect to molecular weight. The possibility of grossly different primary structure of the multiple molecular forms of the enzyme was ruled out by the observed similarity of the amino acid compositions and dansyl-peptide maps of the three forms isolated by isoelectric focusing. The amino acid compositions of the Forms I, II, and III were determined by analysis of the 20.hour hydrolysates in 6 N HCl at 110". The ratios, II/I and III/I, were approximately 1 for most of the amino acids indicating very similar compositions (Table II). The tryptic dansylpeptide maps of Forms I and II were almost identical, except for three spots which did not match indicated by arrows to a, 6, and c (Fig. 4). The dansylpeptide map of III was also similar to I and II in the pattern of the spots, but more than two unmatched spots were noted. NH&erminal analysis by the method of Gros and Labouesse (31) failed to detect a free NHzterminal amino acid for all three forms of the enzyme.
A total amino acid composition of the purified p-hydroxy- phenylpyruvate hydroxylase was determined by hydrolysis of the enzyme with 3 M p-toluenesulfonic acid, 0.2% 3-(2-aminoethyl)indole at 110" for 24,48, and 72 hours, and automatic amino acid analysis according to the method of Liu and Chang (28) (Table  II). Hexosamines, which can be determined by this method, were not found in the enzyme hydrolysates. Since almost all glycoproteins contain at least 1 residue of hexosamine (32), the possibility that the microheterogeneity of the enzyme was due to a heterogeneous carbohydrate moiety is unlikely.
The existence of multiple alleles for p-hydroxyphenylpyruvate hydroxylase was also deemed unlikely on the basis of the rigor of natural gene selection, especially in the high proportions found for the multiple molecular forms. An alternative explanation was that the progressive nonenzymic deamidation of the asparagine and glutamine residues of a single nascent enzyme resulted in the multiple molecular forms. This phenomenon has been observed for many proteins (33). Flatmark observed that the process of nonenzymic deamidation was accelerated by high pH and high ionic strength in the case of beef heart cytochrome c (34); therefore, the effect of these conditions on p-hydroxyphenylpyruvate hydroxylase was examined. Polyacrylamide gel electrophorcsis was utilized to characterize enzyme incubated for varying lengths of time in 0.5 M potassium phosphate, pH 9.0. A progressive loss of protein from the bands of lower mobility and a relative increase of protein in the bands of higher mobility were observed in the stained gels (Fig. 5). The over-all loss of total protein seemed to be due to denaturation and precipitation at the high pH From the enzyme incubated for a total of 70 hours, the gels revealed the appearance of an additional protein band of higher mobility than those observed in the unincubated control. This new band was labeled j in Fig. 5 enzyme with P-mercaptoethanol and inclusion of P-mercaptoethanol in buffers during electrophoresis did not alter these result's The conversion of lower to higher mobility corresponds to an increase in the net negative charge of the protein, neglecting possible changes in protein size or shape. Nonenzymic deamidation could account for the increase in negative charge and electrophoretic mobility. The increase in mobility could not be explained by the formation of intramolecular disulfide bridges since P-mercaptoethanol had no effect on the mobilit,y of treated enzyme forms. Thus, the available evidence indicates nonenzymic deamidation as t,he probable cause of the multiple molecular forms of p-hydroxyphenylpyruvate hydroxylase; the significance of this phenomenon is unclear, since it may be occurring both in vivo and as a purification artifact. Analysis of p-hydroxyphenylpyruvate hydroxylase for bound transition metals was performed by neutron activation analysis. Only copper was found to be present in significant amounts greater than the dialysate, and act,ivation analysis proved inadequate for the detection of iron at the microgram level. Therefore, the copper and iron content of the enzyme sample was quantitated by at,omic absorption spectroscopy. The results are presented in Table III, where they are expressed as the difference in metal content between t'he dialysate and enzyme samples. The amount of bound iron found in the enzyme fraction eluted from QAE-Sephadex was 12 times the amount of the bound copper found in the same fraction, and the iron to protein mole The increase in the iron to protein mole ratio, also, correlated well with the purification of the enzyme specific activity, going from a partially purified SP fraction to the Q,AE fraction. The copper to protein mole ratio, however, remained unchanged (Table  III). When special precautions were taken to eliminate nonspecifically bound metal ions from the enzyme samples, a decrease in the iron to protein mole ratio was noted with an accompanying decrease in the enzyme specific activity. The QAE fraction was dialyzed against buffer containing EDTA, and this QAE fraction and the SP fraction were run through separate beds of Sephadex G-25. These enzyme samples were then passed through Chelex 100 deionizing resin columns, and the samples were aspirated directly into the atomic absorption spectrophotometer.
The data presented in Table IV demonstrated that iron was again co-purified with the enzyme activity, even after the removal of nonspecifically bound metal ions. The @%-fold increase in the bound iron of the QAE fraction over the SP fraction corresponded to the go-fold increase in specific activity. The fact that a portion of iron was removed by the Chelex resin with a corresponding loss of specific activity could be explained by assuming denaturation of the enzyme and release of the tightly bound metal ion. An attempt at reactivation of Chelex-treated enzyme by the addition of ferrous ions was unsuccessful.
A confirmation of the atomic absorption data supporting the existence of an iron prostetic group for p-hydroxyphenylpyruvate was obtained by EPR spectroscopy. A single strong signal was obtained in the g = 4 region for scans of the purified enzyme (Fig. 6) integration of the signal and was found to be 4.33. This signal is a common ferric iron EPR characteristic of high spin ferric iron in a rhombic ligand field (35). A similar signal has been observed for iron associated to transferrin, g = 4.14 (36) and for pyrocatechase which is an iron activated oxygenase, g = 4.2 (37). Ferric EDTA is a model system which also has been noted to give the g = 4.3 EPR. The EPR of p-hydroxyphenylpyruvate hydrosylasc is probably due to the ferric form of the enzyme, since either the high or low spin configuration of ferrous iroil would be espccted to bc diamagnetic 3 giving IIO EPR. Activatioli of the elizyme with ascorbic acid or rotlucetl tlicl~loropl~c~~oli~~dophenol would result in reduction of the enzyme to the ferrous form which is, apparently, the active form of the enzyme; reduction of the iron would also lead to a loss of the EPR signal. Proof that iron plays a catalytic role in the enzymic reaction mechanism is yet to be accomplished, but the cvitlcncc for the specific association of iron to p-hydrosypheirylpyruvatf~ hydrosylase is now directly and unambiguously demonstrated by its co-purification with enzyme activity and the failure of EDTA and Cheles 100 to remove it without destruction of enzyme activity.