Conformational properties and stability of tyrosine hydroxylase studied by infrared spectroscopy. Effect of iron/catecholamine binding and phosphorylation.

The conformation and stability of recombinant tetrameric human tyrosine hydroxylase isoenzyme 1 (hTH1) was studied by infrared spectroscopy and by limited tryptic proteolysis. Its secondary structure was estimated to be 42% α-helix, 35% β-extended structures (including β-sheet), 14% β-turns, and 10% nonstructured conformations. Addition of Fe(II) or Fe(II) plus dopamine to the apoenzyme did not significantly modify its secondary structure. However, an increased thermal stability and resistance to proteolysis, as well as a decreased cooperativity in the thermal denaturation transition, was observed for the ligand-bound forms. Thus, as compared with the apoenzyme, the ligand-bound subunits of hTH1 showed a more compact tertiary structure but weaker intersubunit contacts within the protein tetramer. Phosphorylation of the apoenzyme by cyclic AMP-dependent protein kinase did not change its overall conformation but allowed on iron binding a conformational change characterized by an increase (about 10%) in α-helix and protein stability. Our results suggest that the conformational events involved in TH inhibition by catecholamines are mainly related to modifications of tertiary and quaternary structural features. However, the combined effect of iron binding and phosphorylation, which activates the enzyme, also involves modifications of the protein secondary structure.

The conformation and stability of recombinant tetrameric human tyrosine hydroxylase isoenzyme 1 (hTH1) was studied by infrared spectroscopy and by limited tryptic proteolysis. Its secondary structure was estimated to be 42% ␣-helix, 35% ␤-extended structures (including ␤-sheet), 14% ␤-turns, and 10% nonstructured conformations. Addition of Fe(II) or Fe(II) plus dopamine to the apoenzyme did not significantly modify its secondary structure. However, an increased thermal stability and resistance to proteolysis, as well as a decreased cooperativity in the thermal denaturation transition, was observed for the ligand-bound forms. Thus, as compared with the apoenzyme, the ligand-bound subunits of hTH1 showed a more compact tertiary structure but weaker intersubunit contacts within the protein tetramer. Phosphorylation of the apoenzyme by cyclic AMP-dependent protein kinase did not change its overall conformation but allowed on iron binding a conformational change characterized by an increase (about 10%) in ␣-helix and protein stability. Our results suggest that the conformational events involved in TH inhibition by catecholamines are mainly related to modifications of tertiary and quaternary structural features. However, the combined effect of iron binding and phosphorylation, which activates the enzyme, also involves modifications of the protein secondary structure.
Tyrosine hydroxylase (TH, 1 EC 1.14.16.2) is a non-heme iron and tetrahydrobiopterin-dependent enzyme that catalyzes the conversion of L-tyrosine to L-dihydroxyphenylalanine (L-DO-PA), the first and rate-limiting step in the biosynthesis of catecholamines (1,2). The catalytic activity of TH seems to be short-term regulated by feedback inhibition by catecholamines and by reversal of this inhibition by phosphorylation (1,3).
Until the recent cloning and overexpression of TH from different species (4 -6), the source of the enzyme has mostly been the bovine adrenal medulla or pheochromocytoma (PC12) cells. However, these enzymes are isolated partially inhibited by catecholamines bound at the active site and are activated severalfold by phosphorylation of Ser-40 by the catalytic subunit of cyclic AMP-dependent protein kinase (cAPK) (7)(8)(9). This activation has been related to the increased dissociation rate of catecholamines from the active site on phosphorylation (3).
Human TH exists as several different isoforms generated by alternative splicing of pre-mRNA (10,11); isoform 1 (hTH1) is the most abundant species in the adrenal medulla and in the substantia nigra of the brain (10). When expressed in Escherichia coli, human TH isoforms are isolated as homogeneous tetrameric apoenzymes composed of identical 56-59-kDa subunits (12) that are rapidly activated (up to 40-fold) by binding of 1 atom of Fe(II) per subunit (4). The recombinant human enzymes are inhibited by catecholamines that chelate iron at the active site, and this inhibition is partially reversed by phosphorylation of Ser-40 by cAPK (3,12,13).
The three-dimensional structure of TH remains to be determined, and the potential conformational changes associated with the activation by phosphorylation and inhibition by catecholamines are not characterized. It has been postulated that phosphorylation induces a conformational change that decreases the thermostability of the enzyme (14,15). However, the influence of the enzyme-bound catecholamines on these effects was not considered. In the present work, we have studied the secondary structure and stability of recombinant hTH1, with special reference to the effects of its activation by Fe(II) and phosphorylation by cAPK and its inhibition by dopamine. As experimental techniques, we have used infrared (IR) spectroscopy and limited tryptic proteolysis.

MATERIALS AND METHODS
Dopamine and deuterium oxide were from Sigma. Human TH isoform 1 (hTH1) was expressed in E. coli and purified to homogeneity as described (4,12). Buffers were passed through a Chelex-100 ion-exchange resin to avoid iron contamination. The enzyme preparations used in this study contained 0.02 Ϯ 0.01 atoms of iron/subunit, as determined by atomic absorption spectrometry, and was considered to represent the apoenzyme (apo-hTH1). The holoenzyme, Fe(II)-hTH1, was prepared by incubation of apo-hTH1 with equimolar concentrations of ferrous ammonium sulfate for 5 min at 20°C (4,16), and the dopamine-Fe(III)-hTH1 complex was obtained by the incubation of Fe(II)-hTH1 with equimolar concentrations of dopamine (16). The concentration of purified recombinant enzyme was determined by the absorbance at 280 nm (⑀ 1% ϭ 10.4 cm Ϫ1 ) at neutral pH (7). The catalytic subunit of cAPK was purified from bovine heart as described (17). Phosphorylated Fe(II)-hTH1 at Ser-40 was prepared from the phosphorylated apo-hTH1 as described for the nonphosphorylated form of the enzyme (see above).
Limited Proteolysis by Trypsin and SDS Gel Electrophoresis-Stock solutions of trypsin were prepared daily. Tryptic digestion of nonphosphorylated and phosphorylated forms of hTH1 (1.5 mg/ml) was performed at 30°C in 20 mM Na-Hepes, 150 mM NaCl, pH 7.4, with a variable trypsin:protein ratio. The reaction was stopped after 20 min by the addition of 0.5 mM phenylmethylsulfonyl fluoride and 0.5 mM L-1tosylamido-2-phenylethyl chloromethyl ketone. The products of proteolysis were analyzed by SDS-polyacrylamide gel electrophoresis performed at 180 V (2 h) in 10% (w/v) polyacrylamide slab gels (20), stained with Coomassie Brilliant Blue, and scanned at 633 nm using a LKB Ultroscan XL laser densitometer.
Infrared Spectroscopy-Samples for IR spectroscopy were made up by dissolving the lyophilized protein in 20 mM Na-Hepes, 150 mM NaCl, pH or pD (H 2 O-or D 2 O-based buffers) 7.4. Before use, samples were centrifuged 10 min at 14,000 ϫ g to remove any undissolved protein aggregates. Infrared spectra were recorded in Nicolet 520 (D 2 O samples) or Magna 550 (H 2 O samples) spectrometers equipped with MCT detectors. Samples, containing 12-14 mg protein/ml (0.20 -0.28 mM enzyme subunit), were placed in a thermostated cell with CaF 2 windows; 50-m spacers were used for D 2 O and 6-m for H 2 O solutions. Samples in deuterated buffer were left 3 h at room temperature to allow isotopic substitution of the protein exchangeable NH groups; longer exposure to the solvent did not induce further spectral changes. A total of 200 (D 2 O) and 1000 scans (H 2 O) were accumulated for each spectrum, using a shuttle device. Thermal stability studies were carried out by heating the samples in steps of about 3°C, in the temperature interval 25-85°C. After every heating step, the sample was left to stabilize for 5 min, and the spectrum was recorded. Spectra were analyzed in a personal computer where solvent subtraction, deconvolution, band-position determination, and curve fitting of the original amide I band were performed as reported previously (21). Briefly, band component positions were obtained from deconvolution and derivation; bandwidth estimations were obtained from derivative spectra, and the Gaussian fraction was set to 90%. Since the results obtained after iterations may not be unique, the following restrictions were applied (22): i) the band position cannot diverge from initial guesses more than the distance between data points; and ii) the width of the bands should be less than one-half of the amide I bandwidth. The use of several spectra recorded at different temperatures below the thermal denaturation transition reduces the error of the quantification procedure to around 3% (21,23). Fig. 1 shows the 1800 -1500 cm Ϫ1 region of the original and the amide I band of the deconvoluted infrared spectra of apo-hTH1 and the dopamine-Fe(III)-hTH1 complex, recorded in D 2 O medium. The deconvoluted spectra of these samples revealed that the amide I band can be resolved into several components both in D 2 O (Fig. 1) and H 2 O (data not shown). Assignment of individual band components to specific types of secondary structure has been discussed recently (22). The intense and well resolved component bands appearing at around 1654 and 1635 cm Ϫ1 have been assigned to ␣-helical and ␤-sheet conformations, respectively. However, nonstructured conformations may also contribute to the intensity of the helical and ␤-sheet components in H 2 O and D 2 O, respectively (see below). Minor components are seen in both solvents at around 1668 and 1677 cm Ϫ1 , which mainly arise from ␤-turns. The rather weak component band at around 1625 cm Ϫ1 has been assigned to ␤-strands, extended chains not forming ␤-sheets, and have been related, in native oligomeric proteins, to protein-protein contacts (21,23,24).

Secondary Structure of Recombinant Human Tyrosine Hydroxylase (hTH1)-
Quantification of protein secondary structure was performed following a procedure described recently (21,23).  Table I. isotopic shift that is known to occur in certain band components when transferred to the latter medium (23). The values obtained for the band position and percentage area of each component for apo-hTH1 and Fe(II)-hTH1 are presented in Table I. The difference between the spectra recorded in H 2 O and D 2 O in percentage area of the band around 1655 cm Ϫ1 was 8 -10%, but a defined band did not appear at 1640 cm Ϫ1 in D 2 O, as expected from a shift of the band corresponding to unordered structures with backbone deuteration. Instead, the area of the band attributable to ␤-sheet structure, which appears at 1636 cm Ϫ1 (25), increased, and its position shifted upward upon isotopic substitution (see band positions in Table I), indicating that the bands corresponding to ␤ and nonregular structure overlap in D 2 O.
The effect of phosphorylation at Ser-40 by cAPK on the secondary structure of Fe(II)-hTH1 is also shown in Table I. The spectra of the phosphorylated apo-hTH1, Fe(II)-hTH1, and Fe(II)-hTH1 in the presence of equimolar amounts of dopamine look very similar to those described above for the nonphosphorylated samples, including the number and position of the amide I component bands (data not shown). It should be noted that binding of dopamine to the phosphorylated Fe(II)-hTH1 is almost abolished due to a dramatic decrease in the binding affinity (13). A quantitative estimate of the secondary structure of nonphosphorylated and phosphorylated apo-hTH1 and their ligand-bound forms is given in Table II. No significant differences (Ͼ3%) were detected between the nonphosphorylated and phosphorylated forms of apo-hTH1. By contrast, the addition of an equimolar concentration of Fe(II), which generates the catalytically active enzyme form, induced a conformational change in phosphorylated apo-hTH1, notably an increase of about 10% in the ␣-helix content, concomitant with a loss of unordered structure. These changes are maintained in the presence of equimolar concentrations of dopamine.
Effect of Ligand Binding on the Thermal Stability of hTH1-Further insight into the structural changes that occur on ligand binding and phosphorylation of hTH1 was obtained from thermal stability studies. The deconvoluted IR spectra of apo-hTH1 in D 2 O buffer revealed major changes in the amide I (1700 -1610 cm Ϫ1 ) and amide II (1550 cm Ϫ1 ) modes between 44 and 52°C (Fig. 3). These changes included a broadening of the overall amide I contour and the appearance of well defined components at 1618 and 1685 cm Ϫ1 , highly characteristic of thermally denatured proteins (26), which represent extended structures formed on aggregation of the unfolded proteins (21,27). Loss of the residual amide II band at around 1550 cm Ϫ1 , within the above mentioned temperature interval, indicated that thermal denaturation exposed the protein core to deuterium exchange.
empirical parameter, defined as the intensity ratio of the amide I band at 1618 cm Ϫ1 and 1650 cm Ϫ1 (I 1618 :I 1650 ) (Fig. 4). Both the midpoint denaturation temperature (T m ) and the temperature interval in which thermal denaturation took place (⌬T) increased on Fe(II) and Fe(II) plus dopamine binding to nonphosphorylated apo-hTH1 ( Fig. 4A; Table III). Phosphorylation of apo-hTH1 did not substantially modify its thermal stability (Fig. 4B), since both the T m (49°C) and the width of the thermal unfolding transition (9°C) were similar to those of the nonphosphorylated apoenzyme (Table III). As found for the nonphosphorylated hTH1, iron binding to phosphorylated apo-hTH1 stabilized the structure of the phosphorylated enzyme, as evidenced by the increase in T m , and broadened the unfolding transition (Table III). A slightly increased stability was observed for the phosphorylated holoenzyme Fe(II)-hTH1 (T m ϭ 57°C) compared to the nonphosphorylated form (T m ϭ 54°C). It is worthwhile to mention that the high affinities of apo-hTH1 for Fe(II) and nonphosphorylated Fe(II)-hTH1 for dopamine (estimated K D in the M range) (4, 13) ensured that under the experimental conditions used in this work, essentially all (more than 95%) of the apo/holoenzyme formed complexes with iron/dopamine. On the other hand, the low affinity of phosphorylated Fe(II)-hTH1 for dopamine explained the fact that equimolar amounts of inhibitor did not further stabilize its conformation ( Fig. 4B; Table III). Therefore, we measured the thermal denaturation of the nonphosphorylated and phosphorylated holoenzyme in the presence of a large excess of dopamine (50:1 molar ratio). The results demonstrated that although the thermal stability of the former was very similar to that of the 1:1 complex (Table III), the behavior against the thermal challenge of the phosphorylated holoenzyme resembled that of the phosphorylated apoprotein ( Fig. 4B; Table III). Thus, it seems that a large excess of free dopamine, which binds to both free and enzyme-bound iron, was able to effectively extract the iron from the phosphorylated protein.
Limited Proteolysis by Trypsin-The effect of ligand binding and phosphorylation on the global conformation of hTH1 was also examined by studying the susceptibility of the different enzyme forms to limited tryptic proteolysis. Such treatment of rat and bovine TH produces a 34-kDa core fragment with an increased catalytic activity (1,28). As expected, the activity of hTH1 also increased when apo-hTH1 was incubated for 20 min at 25°C with trypsin, with maximal activation (about 40%) at a trypsin:protein ratio of 0.005:1 (g/g). However, as shown by SDS-polyacrylamide gel electrophoresis (Fig. 5), the major product of the nonphosphorylated apo-hTH1 was a 47-kDa protein fragment. The same species were observed with the ligand-bound forms of the enzyme, regardless of its phosphorylation state. At higher protease concentrations, the 47-kDa fragment was further digested to species of lower electrophoretic mobility, and extensive proteolysis was observed at a trypsin:protein molar ratio of 0.1:1 (Fig. 5). Reconstitution of the nonphosphorylated (Fig. 6A) and phosphorylated apo-hTH1 ( Fig. 6B) with Fe(II) increased the resistance to proteolysis of the active 47-kDa enzyme form. A comparison of the fragmentation patterns corresponding to phosphorylated and nonphosphorylated holoprotein revealed that the 47-kDa fragment of the former was more resistant to proteolysis, in good agreement with the thermal stability studies. Dopamine binding to nonphosphorylated holoprotein further stabilized the 47-kDa species (Fig. 6A), whereas no effect on the degradation pattern was found on addition of an equimolar amount of dopamine to the phosphorylated holoenzyme prior to proteolysis (data not shown). DISCUSSION We have studied the secondary structure of human tyrosine hydroxylase (hTH1) and the conformational changes related to  its activation by Fe(II) and phosphorylation of Ser-40 as well as its inhibition by dopamine, using the conformational sensitivity of the infrared amide I band and the susceptibility to limited proteolysis.
Structural Changes Related to hTH1 Activation by Fe(II) and Inhibition by Catecholamines-No significant changes were observed in the secondary structure of apo-hTH1 when reconstituted with Fe(II); the major effect was an increase in protein stability. Thus, our results support a role for iron in the stabilization of TH conformation, in addition to its catalytic function. The only remarkable difference detected in the IR spectrum of the protein on dopamine binding is the loss of the 1625 cm Ϫ1 band component. A band at this position has been associated with intersubunit contacts, and it has been described in dimeric (21) but not in monomeric (30) cytochrome c oxidase. Therefore, its disappearance may indicate a catecholaminemediated weakening of intersubunit interactions within the protein tetramer, as also suggested by the increase in the width of the thermal unfolding transition. A broad transition would indicate a more or less independent subunit unfolding, whereas a sharp transition would suggest that the tetramer behaves essentially as a cooperative unit (31). However, the stabilizing effect of iron on the native conformation of hTH1 (see above) is further strengthened on dopamine binding to the holoenzyme. This effect is most likely on the monomer of the tetrameric enzyme, since each subunit binds equimolar amounts of the ligands, as found with other soluble proteins (24,31,32). Thus, inhibition of the enzyme by catecholamines is associated with a stabilization of its conformation and could be related to the proposed existence of two forms of the bovine enzyme (33), i.e. a labile active form and a stable inactive form, generated by catecholamine binding to the labile form.
Effect of Phosphorylation on the Conformation of hTH1-The molecular and/or structural mechanism by which phosphorylation of Ser-40 results in an activation of TH, kinetically represented by a decrease in the apparent K m for the cofactor and an increase in K i for dopamine, is as yet unknown (1,12,13). Although phosphorylation of apo-hTH1 does not significantly modify its conformation, a significant finding in this work is that the secondary structure of the catalytically active holoenzyme does indeed depend on its phosphorylation state. When phosphorylated apo-hTH1 is reconstituted with iron, a signifi-cant (ϳ10%) increase in the ␣-helical content and a concomitant decrease in unordered structure were observed. The fact that this structural transition is triggered by iron binding suggests a close proximity and/or interaction of the phosphorylation site (Ser-40 at the regulatory N-terminal domain) and the iron binding site (at the catalytic core domain). A recent study based on circular dichroism has provided similar values (Ϸ50%) for the ␣-helix content of TH isolated from rat PC12 cells (15) as those found in this work for hTH. However, our results differ from their proposal that phosphorylation of TH destabilizes its conformation. The phosphorylated active holoenzyme (Fe(II)-hTH1) also shows a slightly increased thermal stability (Table III) and resistance to limited proteolysis (Fig. 6) as compared with the nonphosphorylated form. This apparent discrepancy may be due to either (i) different enzyme source and/or (ii) TH isolated from rat PC12 cells contains catecholamines bound at the active site (9) and, therefore, the decreased thermostability could be a consequence of a phosphorylation-induced catecholamine release from the active site.
Proteins that undergo regulatory serine phosphorylation may be classified into two groups, i.e. those that undergo a global conformational change upon phosphorylation and those that do not (34). An example of the first group is glycogen phosphorylase whose N-terminal segment, bearing the phosphorylation site (Ser-14), becomes helical upon phosphorylation (35). Our data indicate that the combined effect of iron binding and phosphorylation could induce a similar conformational change that may modulate hTH1 function.