Phosphorylation and Mutations of Ser16 in Human Phenylalanine Hydroxylase

Phosphorylation of phenylalanine hydroxylase (PAH) at Ser16 by cyclic AMP-dependent protein kinase is a post-translational modification that increases its basal activity and facilitates its activation by the substrate l-Phe. So far there is no structural information on the flexible N-terminal tail (residues 1–18), including the phosphorylation site. To get further insight into the molecular basis for the effects of phosphorylation on the catalytic efficiency and enzyme stability, molecular modeling was performed using the crystal structure of the recombinant rat enzyme. The most probable conformation and orientation of the N-terminal tail thus obtained indicates that phosphorylation of Ser16 induces a local conformational change as a result of an electrostatic interaction between the phosphate group and Arg13 as well as a repulsion by Glu280 in the loop at the entrance of the active site crevice structure. The modeled reorientation of the N-terminal tail residues (Met1–Leu15) on phosphorylation is in agreement with the observed conformational change and increased accessibility of the substrate to the active site, as indicated by circular dichroism spectroscopy and the enzyme kinetic data for the full-length phosphorylated and nonphosphorylated human PAH. To further validate the model we have prepared and characterized mutants substituting Ser16 with a negatively charged residue and found that S16E largely mimics the effects of phosphorylation of human PAH. Both the phosphorylated enzyme and the mutants with acidic side chains instead of Ser16 revealed an increased resistance toward limited tryptic proteolysis and, as indicated by circular dichroism spectroscopy, an increased content of α-helical structure. In agreement with the modeled structure, the formation of an Arg13 to Ser16 phosphate salt bridge and the conformational change of the N-terminal tail also explain the higher stability toward limited tryptic proteolysis of the phosphorylated enzyme. The results obtained with the mutant R13A and E381A further support the model proposed for the molecular mechanism for the activation of the enzyme by phosphorylation.

Phosphorylation of phenylalanine hydroxylase (PAH) at Ser 16 by cyclic AMP-dependent protein kinase is a post-translational modification that increases its basal activity and facilitates its activation by the substrate L-Phe. So far there is no structural information on the flexible N-terminal tail (residues 1-18), including the phosphorylation site. To get further insight into the molecular basis for the effects of phosphorylation on the catalytic efficiency and enzyme stability, molecular modeling was performed using the crystal structure of the recombinant rat enzyme. The most probable conformation and orientation of the N-terminal tail thus obtained indicates that phosphorylation of Ser 16 induces a local conformational change as a result of an electrostatic interaction between the phosphate group and Arg 13 as well as a repulsion by Glu 280 in the loop at the entrance of the active site crevice structure. The modeled reorientation of the N-terminal tail residues (Met 1 -Leu 15 ) on phosphorylation is in agreement with the observed conformational change and increased accessibility of the substrate to the active site, as indicated by circular dichroism spectroscopy and the enzyme kinetic data for the full-length phosphorylated and nonphosphorylated human PAH. To further validate the model we have prepared and characterized mutants substituting Ser 16 with a negatively charged residue and found that S16E largely mimics the effects of phosphorylation of human PAH. Both the phosphorylated enzyme and the mutants with acidic side chains instead of Ser 16 revealed an increased resistance toward limited tryptic proteolysis and, as indicated by circular dichroism spectroscopy, an increased content of ␣-helical structure. In agreement with the modeled structure, the formation of an Arg 13 to Ser 16 phosphate salt bridge and the conformational change of the N-terminal tail also explain the higher stability toward limited tryptic proteolysis of the phosphorylated enzyme. The results obtained with the mutant R13A and E381A further support the model proposed for the molecular mechanism for the activation of the enzyme by phosphorylation.
Phenylalanine hydroxylase (PAH 1 ; EC 1.14.16.1, phenylalanine 4-monooxygenase) belongs to the family of aromatic amino acid hydroxylases. PAH catalyzes the hydroxylation of L-Phe to L-Tyr, the rate-limiting step in the catabolism of L-Phe (1,2), and it requires a nonheme iron, molecular oxygen, and a pterin cofactor for catalysis (3). Genetic defects in the human enzyme (hPAH) cause phenylketonuria, with a broad range of metabolic and clinical phenotypes (4) as well as enzymatic phenotypes (5,6).
hPAH is a tetrameric/dimeric enzyme, and crystal structure analyses (7)(8)(9) have shown that each of the chains folds into three domains, i.e. an N-terminal regulatory domain (residues 1-110) that includes the single phosphorylation site Ser 16 , a middle catalytic domain, and a C-terminal oligomerization domain. Mammalian PAH is activated severalfold by preincubation with its substrate L-Phe, which represents the most important mechanism for its regulation in hepatocytes (1). Phosphorylation of PAH at Ser 16 by cAMP-dependent protein kinase (PKA) represents an additional post-transcriptional regulation of the enzyme (10 -14). The two mechanisms of activation are interdependent, i.e. L-Phe enhances its rate of phosphorylation by PKA, and the phosphorylated enzyme requires a lower concentration of substrate for its activation (15,16). It appears that these two mechanisms act synergistically also in vivo and that L-Phe promotes the phosphorylation and activation of PAH in rat liver (13,17). However, the molecular mechanism of this interdependence is not well understood (18,19) and has not been explained by the crystal structure analyses of the phosphorylated (at Ser 16 ) ⌬C24-truncated dimeric form of rat PAH (rPAH) (9). Phosphorylation of the human enzyme results in a mobility shift on SDS-PAGE (20) that is also observed when hPAH is expressed in Escherichia coli (16,21) and in the in vitro transcription-translation system (5,22). The enzyme expressed in the latter system is recovered as a double band on SDS-PAGE, corresponding to the phosphorylated (ϳ51 kDa) and nonphosphorylated (ϳ50 kDa) forms. Furthermore, we have previously shown by Fourier transform infrared spectroscopy that phosphorylation of the isolated recombinant Nterminal regulatory domain (residues 2-110) results in an apparent increase in the content of ␣-helical structure (23). In the present work we have extended these studies, including molecular modeling by the anchor grow method using the program DOCK 4.0 (24), and the possible conformations of the 18-residue N-terminal tail have been estimated for the nonphospho-rylated and phosphorylated states. The modeled conformations have given a structural explanation of the changes in enzyme kinetic properties and conformation observed as a result of phosphorylation. To validate this modeled structure we substituted Ser 16 by residues with negatively charged (Glu and Asp) and neutral (Ala) side chains. In addition, we have studied the effect of charged-to-alanine scanning mutagenesis of residues in the regulatory (Arg 13 ) and catalytic (Glu 353 and Glu 381 ) domains considered to be involved in electrostatic interactions and thus important for determining the conformation and orientation of the N-terminal tail in the nonphosphorylated and phosphorylated states. The biochemical data support the structures obtained for the full-length enzymes by molecular modeling.

EXPERIMENTAL PROCEDURES
Molecular Modeling-Docking and preparation of structures was performed on a SGI Octane workstation. The DOCK 4.0.1 suite of programs (University of California at San Francisco) (24) was used to fit the conformers of the N-terminal peptide sequence Met 1 -Gly 19 (MAAV-VLENGVLSRKLSDFG) into the crystal structure of the dimeric rPAH (9), including the regulatory domain (from Gly 19 ), the catalytic domain, and the dimerization motif (up to Thr 427 ). The solvent-accessible surface of this domain was calculated by the DMS program under Midas-Plus (University of California at San Francisco) (25), and a grid of 45 Å ϫ 29 Å ϫ 36 Å was constructed in which iron was included as a sphere of the correct radius and without including crystallographic water molecules. The grid is used by DOCK to evaluate the steric boundary of the protein as well as electrostatic and van der Waal's interactions between the protein and the ligand atoms during the docking procedure, using the energy scoring function. The atom potential types and the partial charges of the 19-residue N-terminal peptide were assigned for the AMBER forcefield using InsightII (Accelrys). A grid space of 0.3 Å, a dielectric factor of 4, and a sampling size of 50 configurations/cycle were used. The anchor grow method (26) was used for docking, including energy minimization of the bound peptide structure during docking. The anchor grow method is based on the partition of the ligand (the N-terminal sequence) into rigid segments containing the largest set of adjacent atoms separated by nonrotatable bonds, whose position is optimized during each step. The coordinates for Gly 19 were used as anchor for growing the remaining 18 N-terminal residues. The programs InsightII and WebLab Viewer (Accelrys) were used to prepare the figures of the modeled full-length conformers of phosphorylated and nonphosphorylated PAH. Snapshots of the structure of each of the 30 lowest energy conformers of the docked peptides with the rest of the protein were prepared using the VMD software (27,28). Then the MediaConvert software for IRIX 6.5 (SGI) was used to prepare a video showing the ensemble of the 30 snapshots, with increasing docking score (see the supplemental material).
Site-directed Mutagenesis-The mutations S16A, S16D, S16E, R13A, E353A, and E381A were introduced in the hPAH cDNA using the QuikChange TM site-directed mutagenesis kit (Stratagene). The primers for mutagenesis (provided by Eurogentec and M.W.G. Biotech) are shown in Table I. The authenticity of the mutagenesis was verified by DNA sequencing using the primers described in Ref. 29.
Expression and Purification of the Enzymes-Growth of E. coli transformed with the pMAL vectors, purification of the fusion proteins by affinity chromatography on amylose resin (New England Biolabs), cleavage by enterokinase (EKMax™ from Invitrogen) using 20 units EKMax™/mg of fusion protein at 4°C for 4 h, and isolation of the tetrameric enzyme forms by size exclusion chromatography were performed as described (21).
Phosphorylation by PKA-The phosphorylation of wt-hPAH and its mutant forms (50 M subunit) was performed at 30°C for 1 h, in 15 mM Na-HEPES, pH 7.0, 3 mM dithiothreitol, 0.03 mM EDTA, 0.1 mM EGTA, 10 mM magnesium acetate, and 200 nM of PKA catalytic subunit (Upstate Biotechnology Inc.). The reaction was initiated by adding 500 M MgATP, and the extent of phosphorylation was measured by SDS-PAGE (mobility shift) and/or by 32 P autoradiography using [␥-32 P]ATP (Amersham Biosciences). At these conditions ϳ1 mol of phosphate was incorporated per mol isolated hPAH subunit (16). The samples of nonphosphorylated wt-hPAH and mutants utilized in this study were treated at identical conditions as the phosphorylated enzymes, except for the absence of kinase during assay.
SDS-PAGE and Ferguson Plot Analysis-SDS-PAGE was performed at 15 mA/gel on 10% (w/v) polyacrylamide gels (except as otherwise indicated; see below), containing 0.1% (w/v) SDS. The gels were stained by Coomassie Brilliant Blue, dried, scanned, and further analyzed using the software Deskscan II (Hewlett-Packard Co.) and Phoretix 1D Plus (Nonlinear Dynamics Ltd). Ferguson plot analyses were performed by running SDS-PAGE at various concentrations of acrylamide (from 8 to 13%, w/v) (30,31). The relative electrophoretic mobility (R m ) was calculated using standardized gels with molecular mass standards of known radii (Bio-Rad), and the log(R m ) was plotted versus the acrylamide concentration (which determines the pore size). Denatured proteins differing only in the effective Stokes radius (hydrated radius of the protein-detergent complex) have different slopes and a common intercept at the y axis, whereas proteins differing only in the electrical charge have identical slopes with different y axis intercepts (30,32).
Assay of PAH Activity-PAH activity was assayed at 25°C with a standard reaction mixture (final volume, 50 l) of 10 M ferrous ammonium sulfate, 1 mM L-Phe, 5 mM dithiothreitol, and 75 M BH 4 , 0.04 g/l catalase, 0.05% (w/v) bovine serum albumin, and 1 g of the various hPAH forms in 100 mM Na-HEPES, pH 7.0, as described (21). Two types of reactions were performed: (i) the enzyme was preincubated for 4 min at 25°C in a mixture containing buffer, L-Phe, and catalase, then Fe(II) was added, and after another 1-min incubation, the reaction was started by the addition of BH 4 (L-Phe activated enzyme) and (ii) the reaction was performed as above, except that L-Phe is added together with BH 4 (nonactivated enzyme). The reaction was stopped after 30 s with 1% (v/v) acetic acid in ethanol. The amount of L-Tyr was measured by high pressure liquid chromatography and fluorimetric detection (12,21). The kinetic parameters were determined using variable concentrations of the substrate (0 -5 mM L-Phe with 75 M BH 4 ) and of the cofactor (0 -500 M BH 4 with 1 mM L-Phe) and were calculated by nonlinear regression analysis of the experimental data using the program Enzfitter (Elsevier-Biosoft).
Circular Dichroism-CD analyses were performed at 25°C on a Jasco J-810 spectropolarimeter equipped with a Jasco 423S Peltier element. The samples (10 M subunit of hPAH forms) were prepared in 20 mM sodium phosphate, 50 mM KF, pH 7.0, with stoichiometric amounts of ferrous ammonium sulfate. Wavelength scans were taken between 185 and 300 nm using quartz cells with path lengths of 0.1 and 0.5 cm. The data were analyzed with the standard analysis program provided with the instrument. The amount of secondary structure elements was estimated by the CDNN program that applies a neural network procedure (33).

RESULTS
Modeling the Conformational Changes Induced by Phosphorylation-In the crystal structure of the dimeric C-terminal truncated form of rPAH (Protein Data Bank code 1PHZ) (9), no electron density was observed for the 18-residue N-terminal tail. Therefore, to obtain some information on the structure of the full-length enzyme and on the conformational changes induced by phosphorylation of Ser 16 , we have created a modeled structure of the N-terminal tail by docking the peptide sequence into the crystal structure of rPAH, using DOCK with the anchor grow method (24). Using the coordinates for Gly19 as an anchor for growing the remaining 18 N-terminal residues, the most probable conformations for the 19-residue Nterminal tail was estimated for the nonphosphorylated and phosphorylated forms. The top scoring structures resulting from the molecular docking are shown in Fig. 1 for the nonphosphorylated (Fig. 1A) and the phosphorylated forms ( Fig.  1B) (see also the supplemental material). Whereas the residues 16 -18 could be superimposed for the two forms, phosphorylation resulted in a significantly different conformation/orientation of residues 1-15. Furthermore, Arg 13 was found to change its electrostatic interactions from Glu 353 and Glu 381 in the nonphosphorylated enzyme (Fig. 1C) to the phosphate group of Ser 16 (Fig. 1D). Although the nonphosphorylated tail establishes interactions with residues on both sides of the entrance to the active site crevice structure, leading to a relatively "closed" conformation of this structure, the phosphorylated tail is reoriented along one side of the channel (Table II), resulting in a more open active site conformation. This change in orientation would allow for the proposed facilitated access of the substrate (9) to its binding site close to the iron (34).
The electrostatic interaction between the phosphate group at Ser 16 and Arg 13 thus seems to be a key determinant for the different conformation adopted by the phosphorylated enzyme. To further validate this structural model and to get information on the molecular basis for the kinetic and conformational changes that occur in wt-hPAH on phosphorylation, we substituted Ser 16 with acidic residues (using alanine as a control) and characterized the biochemical properties of the resulting enzyme forms.
Expression, Purification, and SDS-PAGE Analysis of Enzyme Forms of hPAH with Substitutions at Ser 16 -The mutants S16A, S16D, and S16E were expressed and purified at similar yields as those obtained for wt-hPAH, i.e. about 35 mg of fusion protein were obtained per liter of bacterial culture, and about 8 -12 mg of pure tetrameric enzyme forms were recovered after the cleavage of the fusion protein by EKMax TM . When wt-hPAH is phosphorylated on Ser 16 by PKA, a mobility shift is observed on SDS-PAGE for the phosphorylated form (which appears at an apparent M r of ϳ51 kDa) with respect to the nonphosphorylated form (ϳ50 kDa) (16,21) (Fig. 2A). A similar result was obtained when wt-hPAH expressed in the in vitro transcription-translation system was analyzed (Ref. 5 and data not shown). The relative mobilities (R m values) on SDS-PAGE of the S16D and S16E mutants are also lower than that of the nonphosphorylated wt-hPAH, although the mobility shift was less pronounced than that observed for phosphorylated wt-hPAH ( Fig. 2A). To get further information on the mobility shift observed for phosphorylated wt-hPAH, the shift was measured as a function of the pore size of the gel and analyzed by the Ferguson plot method (30, 31) using standardized gels with protein standards of known radii. From Fig. 2B it is seen that the difference in R m values between the phosphorylated and nonphosphorylated subunits increased with decreasing pore size. These data indicate that the mobility shift results from a change in the effective Stokes radius of the SDS-denatured enzyme, as reported for other proteins (32). Analysis of the mobility shift of the mutant forms S16D and S16E relative to nonphosphorylated wt-hPAH or to the S16A mutant showed that the differences in the slope of the log (R m ) versus the acrylamide concentration plots were not significant. Attempts to study the mobility shift following phosphorylation under nondenaturing conditions was hampered by the fact that the purified tetrameric enzyme does not show a unique band when subjected to native electrophoresis but rather a smear of multiple bands, 2 probably related to the microheterogeneity caused by deamidation of several labile Asn residues in the subunit (35).
Steady-state Kinetic Analysis-As seen from Table III, the V max values obtained for the three mutant forms S16A, S16D, and S16E were similar to that of wt-hPAH when the enzymes were activated by preincubation with L-Phe. As expected (16), phosphorylation of hPAH resulted in a 1.4-fold increase in the basal activity (i.e. without L-Phe preincubation) as well as a higher apparent affinity for L-Phe. No significant difference in the degree of substrate activation or the kinetic parameters was measured for the S16A mutant form, as compared with the nonphosphorylated wt-hPAH, whereas the mutants with the negatively charged residues, notably S16E, revealed a higher basal activity and a lower [S] 0.5 value for L-Phe, thus partially mimicking the phosphorylated form (Table III). The characteristic positive kinetic cooperativity of L-Phe binding to tetrameric PAH, with a Hill coefficient of h ϭ ϳ2 (1, 3, 21) was maintained both upon phosphorylation and substitution of Ser 16 . Neither phosphorylation of wt-hPAH (16) nor the mutations at Ser 16 had any significant effect on the K m value for the cofactor BH 4 (in the range of 29 -35 M). Limited Tryptic Proteolysis-We also studied the effect of phosphorylation and substitutions of Ser 16 on the susceptibility of the enzyme to limited tryptic proteolysis as a conformational probe. As seen from Fig. 3A, wt-hPAH (about 50 kDa for the subunit) generated several forms with molecular masses close to that of the full-length enzyme (about 49 kDa), in addition to truncated forms of molecular mass around 35 kDa, corresponding to the catalytic core structure (36). The rate of cleavage by limited tryptic proteolysis was decreased for the phosphorylated enzyme (Fig. 3, B and E). Although proteolysis of the mutant S16A gave results similar to those for the nonphosphorylated enzyme (Fig. 3, C and E), the substitution of Ser 16 with a negatively charged residue also protected against proteolysis by trypsin (Fig. 3, D and E, and data not shown for the S16D mutant).
Circular Dichroism-From the far UV CD spectra of nonphosphorylated and phosphorylated wt-hPAH and of the mutants S16A and S16E (Fig. 4), it is seen that the negative ellipticity in the 205-235-nm range increased on phosphorylation and on substitution of Ser 16 with negatively charged residues, compatible with an increase in the apparent ␣-helical content. Thus, using a standard analysis program, the ␣-helical content of nonphosphorylated hPAH and the mutant S16A was calculated to be 30.1 Ϯ 1.3%, as compared with 38.7 Ϯ 1.7% for the phosphorylated enzyme and 35.0 Ϯ 2.4% for S16E when the spectra (185-250 nm) were simulated (33). A 33.1 Ϯ 2.0% ␣-helical content was calculated for the S16D mutant form. The content of the ␣-helical structure for wt-hPAH estimated by CD is in agreement with that obtained by the same method for rPAH (31%) (37) and as determined in the crystal structure (ϳ 35%) (9).
Expression, Purification, and Characterization of the Mutant Forms R13A, E353A, and E381A-In addition to the site of phosphorylation, several residues reveal themselves as significant in maintaining the correct conformation and orientation of the N-terminal tail in the nonphosphorylated and phosphorylated hPAH. In the modeled structure (Fig. 1) the residues Arg 13 , Glu 280 , Glu 353 , and Glu 381 are of particular interest, and to further asses their role we substituted Glu 353 , Glu 381 , and Arg 13 by charged-to-Ala scanning mutagenesis. Substitution of Glu 280 , which (in addition to Arg 13 ) participates in the reorientation of the N-terminal tail by repulsion with the phosphate group at Ser 16 (Fig. 1), was also considered. Glu 280 is important in stabilizing the active site structure by forming a salt bridge with Arg 158 (6), and Glu 280 mutations have been found to result in an unfolding and a low catalytic activity of the recombinant 2 M. Thórólfsson, unpublished results. Asp 145 (L), Pro 147 (L) b Glu 376 (R) Leu 6 Tyr 277 (L) Ser 378 (R) Glu 7 NC Ser 378 (R), Pro 384 (R) Asn 8 NC Thr 380 (R), Pro 384 (R) Gly 9 Thr 380 (R) NC Val 10 Thr 380 (R), Glu 381 (R) Ser 378 (R), Val 379 (R) Leu 11 Thr 380 (R) Val 379 (R) Ser 12 Pro 279 (L) Val 379 (R), Thr 380 (R), Glu 381 (R) Arg 13 Glu 353 (R), Glu 381 (R) Pro 279 (L), Asp 145 Lys 14 Thr 278 (L), Pro 281 (L), His 285 (L), Gly 346 (R) Glu 280 (L), Pro 281 (L), Gly 346 (R) Leu 15 NC NC Ser 16 NC NC Asp 17 NC NC Phe 18 NC NC Gly 19 NC NC a NC, no contacts. b The (R) and (L) correspond to location at the right and left side of the channel leading to the active site iron when looking towards the iron (same orientation as shown in Fig. 1), respectively. proteins (38,39). Both Glu 353 and Glu 381 are solvent exposed and are located in loops leading to the entrance of the active site, and in the modeled structure they form salt bridges with Arg 13 in the nonphosphorylated form, keeping the enzyme in a relatively "closed" low activity conformation. It was thus expected that the mutants E353A and E381A would show some of the kinetic properties of the more "open" phosphorylated enzyme. Elimination of the charge in Glu 353 seems to have deleterious effects on the conformation of the enzyme, and the purified mutant E353A shows a high degree of aggregation. The tetrameric form isolated by size exclusion chromatography represents less than 10% of the total E353A protein and is devoid of any catalytic activity. The mutant E381A, however, expressed well and revealed a basal activity and affinity for L-Phe characteristic of a partially activated enzyme, and the affinity for L-Phe increased on phosphorylation (Table III). However, its CD spectrum and susceptibility to proteolysis were not significantly different from that of the wild-type enzyme, both as nonphosphorylated (Figs. 3E and 4 and data not shown) and as phosphorylated enzyme (Figs. 3E and 4 and data not shown).
According to the modeled structure (Fig. 1, A and C) Arg 13 is involved in stabilizing the closed low activity conformation of the nonphosphorylated enzyme by forming salt bridges with Glu 353 and Glu 381 (see above). As found for the E381A mutant, the form R13A revealed a level of basal activity and an affinity for L-Phe that was higher than for wt-hPAH (Table III). Furthermore, the mutant R13A showed a higher resistance to limited tryptic proteolysis than the nonphosphorylated wildtype form of the enzyme (Fig. 3E), indicating that this residue is an important determinant for the susceptibility of hPAH to limited tryptic proteolysis.
On the other hand, the verification of the role of Arg 13 to preserve the correct orientation of the N-terminal tail by establishing electrostatic interactions with the phosphate group or the negative charged residues at position 16 in the phosphorylated form (or in the S16E or S16D mutants) is not a straightforward task. First, the mutant form R13A cannot be phosphorylated because this Arg is part of the specific recognition site of PKA (Arg 13 -Lys 14 -Leu 15 -Ser 16 ) (11). We have prepared the double mutant R13A/S16E to mimic the effect of this mutation in the phosphorylated enzyme. Secondly, the dual role of Arg 13 both in maintaining a closed conformation by salt bridging with Glu 353 and Glu 381 in the nonphosphorylated hPAH (see above) and in keeping the open conformation by interacting with the phosphate group in the phosphorylated enzyme would contribute to a mixed effect of the mutation R13A in the phosphorylated hPAH. In agreement with this assumption, the mutant R13A/S16E in fact revealed kinetic properties intermediate of

TABLE III
Kinetic properties of nonphosphorylated and phosphorylated wt-hPAH and of the mutants S16A, S16D, S16E, R13A, and E381A The catalytic activity of the isolated tetrameric forms was measured at pH 7.0 at 25°C. V max (Phe preincubation), the concentration of L-Phe at half maximal activity ([S] 0.5 ), and the Hill coefficient (h) for L-Phe were determined at 75 M BH 4 and variable concentrations of L-Phe (0 -5 mM); the enzymes were preincubated (5 min, 25°C, pH 7.0) with L-Phe at the same concentrations as in the assay. V max (no Phe preincubation) was determined without L-Phe during preincubation but at otherwise identical conditions. the nonphosphorylated and phosphorylated forms (data not shown).

DISCUSSION
In vitro phosphorylation of mammalian PAH by PKA at the single phosphorylation site Ser 16 results in an increased basal activity and an apparent increased affinity for the substrate L-Phe (12,14). Studies on both rPAH and hPAH have revealed an interplay between substrate activation and phosphorylation (12,15,16), which is considered to have a physiological relevance (1,13,17). Crystallographic analysis of the C-terminally truncated dimeric form of rPAH (residues 1-429) has revealed a similar overall structure of the phosphorylated and nonphosphorylated enzyme including residues 19 -427, whereas no electron density was observed for the 18 N-terminal residues (9). This finding supports a rather flexible structure of this N-terminal tail and supports the possibility that the proposed conformational changes triggered by Ser 16 phosphorylation (18) are localized to this sequence element. This conclusion is supported by the present modeled structure of residues 1-429 in rPAH, and by site-directed mutagenesis, steady-state enzyme kinetics, CD spectroscopy, and limited tryptic proteolysis (see below). The structures show that phosphorylation triggers a conformational change and reorientation of the N-terminal tail (residues 1-15) (Fig. 1, A and B) because of the formation of a new salt bridge between the phosphate group and the conserved Arg 13 (Fig. 1, C and D). Thus, the estimated average distance between the N 2 atom of Arg 13 and the three oxygen atoms of the phosphate group was 3.6 Å. In addition to Arg 13 , Glu 280 may also contribute to the reorientation of the N-terminal tail because this residue, located in a loop at the entrance to the active site (Fig. 1), would lead to a movement of the phosphorylated Ser 16 by electrostatic repulsion. Earlier studies on the effect of substitutions of Ser 16 in rPAH by negatively charged, positively charged, and uncharged amino acids have revealed that its activation by phosphorylation can be explained by the introduction of a negative charge at this position (40,41). In the present studies on hPAH, it is shown that the Ser 3 Glu substitution, and to a lesser extent the Ser 3 Asp substitution, also results in an increased basal activity (i.e. in the absence of preincubation with L-Phe) and an apparent increased affinity for L-Phe (Table III). Furthermore, these mutant forms are characterized by a decreased susceptibility to limited proteolysis by trypsin, similar to that observed on Ser 16 phosphorylation (Fig. 3), as well as by a mobility shift on SDS-PAGE (Fig. 2) and an apparent increase of the ␣-helical content as determined by CD spectroscopy (Fig. 4). Although the introduction of the negatively charged residues (Asp and Glu) mimic Ser 16 phosphorylation in terms of changes in enzyme kinetics and conformational properties, the effects are quantitatively smaller, notably for the difference in electrophoretic mobility in SDS-PAGE with respect to the nonphosphorylated wild-type form or the S16A mutant. The Ferguson plot analysis of the mobility shift between the SDS-denatured forms of the phosphorylated and nonphosphorylated forms indicates that the two forms (nonphosphorylated and phosphorylated) differ in their apparent Stokes radius, suggesting that the electrostatic interaction (between Arg 13 /Lys 14 and the phosphorylated Ser 16 ) results in a local conformational change at the N-terminal tail that is maintained in the partly denatured state. Because the maximum charge of the Glu side chain is Ϫ1 compared with Ϫ2 for the phosphate group, the electrostatic interaction in the mutant is weaker.
On the basis of the modeled structures ( Fig. 1) and mutagenesis at Ser 16 , Glu 381 , and Arg 13 , it is concluded that the positioning of the inhibitory autoregulatory N-terminal sequence in the nonphosphorylated form results in a relatively closed active site crevice structure. On phosphorylation of Ser 16 a more open active site is obtained, in agreement with the enzyme kinetic data with an increased affinity for the substrate (Table III), which binds close to the iron through interactions with several active site residues including the residues Arg 270 , Trp 326 , and Ser 349 (34). Thus, the modeled structure (Fig. 1, B and D) explains the observed interdependence of L-Phe activation and phosphorylation (12,14,18). Moreover, the bent backbone conformation around Ser(P) 16 may be related to the apparent increase in ␣-helical content as observed by CD spectroscopy. Interestingly, by Fourier-transform infrared spectroscopy we have previously observed an increase in the apparent ␣-helical content of the isolated 1-110-residue N-terminal regulatory domain of hPAH and of the iron reconstituted recombinant human tyrosine hydroxylase on phosphorylation of Ser 16 (23) and Ser 40 (42), respectively.
Our modeled structure of the full-length enzyme also explains the results obtained by limited proteolysis with trypsin ( Fig. 3). Thus, the higher content of noncovalent contacts, mostly van der Waal's interactions (Յ4 Å), between the Nterminal tail and the catalytic domain in the phosphorylated enzyme (Table II) is compatible with a more restrained configuration of the tail (supplemental material) and the decreased susceptibility to limited tryptic proteolysis shown by the phosphorylated wt-hPAH and the S16D and S16E mutant forms (Fig. 3). Furthermore, the R13A mutant also shows a reduced rate of tryptic proteolysis, indicating that this residue is an important determinant for trypsin binding and cleavage of hPAH. Accordingly, the protection against proteolysis observed in the phosphorylated enzyme and in the mutants with acidic residues at the 16 position also seems to be related to the lower accessibility of Arg 13 to trypsin in these enzymes, in which Arg 13 forms salt bridges with Asp 145 and the phosphate group on Ser 16 (or the carboxylate in the S16D and S16E forms) and thus becomes sandwiched between these residues (Fig. 1D). Arg 13 is essential for phosphorylation of Ser 16 (11), as part of the consensus sequence for PKA-catalyzed phosphorylation (Arg-(Arg/Lys)-Xaa-Ser), similar to that found in liver pyruvate kinase (43) and in the ␣ and ␤ regulatory subunits of phosphorylase kinase (for review see Ref. 44). Using model heptapeptides corresponding to the phosphorylation site of pyruvate kinase, it was shown that phosphorylation inhibits the rate of cleavage by trypsin-like enzymes (45), which has been related to an effect of phosphorylation in protein processing and turnover in vivo. Phosphorylation and/or introduction of negatively charged residues at the phosphorylation sites has also been associated with increased protein stability in HPr, a key regulatory protein in bacteria (46), in the ATF2 transcription factor (47), the p27(Kip1) protein (48) and the high mobility group B proteins (49), among others. Although it has been reported that the stabilization induced by phosphorylation of Ser 46 in the HPr protein is due to an electrostatic interaction between the negatively charged groups and the helix macrodipole (46,50), there are no direct insights into the specific mechanisms at the atomic level leading to this increase in stability for the other proteins. The structural model for PAH in Fig. 1 (B and D) is also in agreement with the finding that the phosphorylated form is better mimicked by the S16E mutant form than by S16D ( Ref. 41 and this work). Indeed, a stronger interaction is observed between the Arg 13 and the longer glutamate side chain than with the aspartate (data not shown).
Thus, the results from this study give further credit to the proposal that phosphorylation of PAH induces a local conformational change at the N-terminal autoregulatory sequence, brought about mainly by an electrostatic interaction between the phosphate group and Arg 13 . The conformation of the phosphorylated enzyme results in facilitated access of the substrate to the active site.