EPR and 1H-NMR spectroscopic studies on the paramagnetic iron at the active site of phenylalanine hydroxylase and its interaction with substrates and inhibitors

Eur. J. Biochem. 198,675-682 (1991)

0014295691 00387Q

FEBS 1991

EPR and 'H-NMR spectroscopic studies on the paramagnetic iron at the active site of phenylalanine hydroxylase and its interaction with substrates and inhibitors Aurora MARTiNEZ', Kristoffer K . ANDERSON', Jan HAAVIK ' and Torgeir FLATMARK'

Department of Biochemistry, University of Bergen, Norway Dcpartment of Biophysics, University of Stockholm, Sweden

(Received January 4/March 8, 1991) ~ EJB 91 0028

The paramagnetic iron at the active site of highly purified, catalytically active phenylalanine hydroxylase was studied by EPR at 3.6 K and one-dimensional 'H-NMR spectroscopy at 293 K. The EPR-detectable iron of the bovine enzyme was found to be present as a high-spin form (S = 5j2) in different ligand field symmetries depending on medium conditions (buffer ions) and the presence of ligands known to bind at the active site. At 3.6 K and in phosphate buffer, the paramagnetic iron is coordinated in an environment of rhombic symmetry (g = 4.3), whereas Tris buffer favours an environment of axial ligand field symmetry (g = 6.7, 5.3 and 2.0). The latter axial type of signals resembles those observed at g = 7.0, 5.2 and 1.9 for the enzyme in phosphate buffer when L-noradrenaline is added as an active-site ligand (inhibitor). The same proportion of iron that coordinates to L-noradrenaline seems to be reduced by the pterin cofactor and participate in catalysis. Experimental evidence is presented that Tris inhibits the enzyme by interacting with the enzyme-bound ferric iron and decreases its rate of reduction by the tetrahydropterin cofactor. Preincubation with dithiothreitol also inhibits the enzyme activity and prevents the reduction of its catalytically active ferric iron by pterin cofactors as well as binding of catecholamines to the enzyme.

'H-NM R spectroscopy revealed that the substrate (L-phenylalanine) and L-noradrenaline bind close to the paramagnetic iron, and that the catecholamine displaces the substrate from its binding at the active site. The results support our recently proposed model for the cooperative binding of inhibitor and substrate at the active site [Martinez, A. et al. (1990) Eur. J . Biochem. 193, 211 -2191.

Phenylalanine 4-monooxygenase (phenylalanine hydroxylase) is a tetrahydropterin-dependent enzyme that catalyses the formation of L-tyrosine from L-phenylalanine [l]. This 200-kDa tetrameric enzyme has non-heme iron coordinated at the active site, with a stoichiometry of one iron atom/ subunit in the fully reconstituted, active form [2, 31. In the original study on the native rat enzyme by EPR, only a signal at g = 4.3 was observed, corresponding to a high-spin Fe(I1I) ( S = 5/2) in an environment of predominant rhombic symmetry [4]. This signal was reported to disappear almost completely on addition of substrate and cofactor, indicating that the iron was involved in the catalytic process [4]. Also in later studies, the ferric iron has been found to be reduced to the ferrous state by the tetrahydropterin cofactor, which is oxidized to a quinonoid form [5,6]. By contrast, more recently it has been reported that the EPR spectrum of the resting rat enzyme exhibits signals at geff = 9.4-8.7, 4.3 (proposed to represent catalytically inactive iron) and g,,, = 6.7, 5.4 and 2 (catalytically active iron), consistent with the presence of two different populations of high-spin ferric iron in the enzyme 17,

Correspondence to A. Martinez, Department of Biochemistry,

Abbreviation. 6-MPH4, 6-methyl-5,6,7,8-tetrahydropterin. Enzymes. Phenylalanine 4-monooxygenase or phenylalanine hy-

droxylase (EC 1.14.16.1); tyrosine 3-monooxygenase or tyrosine hydroxylase (EC 1.14.16.2).

Arstadveien 19, N-5009 Bergen, Norway

81. A major discrepancy therefore exists in the literature with respect to the EPR spectrum of the catalytically active form of phenylalanine hydroxylase. In the present study we have addressed this question using the enzyme isolated from rat and bovine liver [9, 101. The resulting data provide evidence that the EPR spectrum of the enzyme as isolated is dependent on the buffer ions present. Furthermore, we have studied the interaction of the substrate (L-phenylalanine) and a potent active-site inhibitor (L-noradrenaline) by EPR and 'H-NMR spectroscopic methods in order to derive further information on the catalytically active/inactive forms of this enzyme.

MATERIALS AND METHODS Mu teriuls

L-Noradrendline, L-phenylalanine, glucose oxidase, catalase and dithiothreitol were purchased from the Sigma Chemi- cal Co. (St Louis, MO). ~-[~H]Noradrenaline was obtained from Amersham International (Amersham, UK). Phenyl- Sepharose CL-4B and DEAE-Sepharose fast flow were obtained from Pharmacia (Uppsala, Sweden). 6-Methyl- 5,6,7&tetrahydropterin (6-MPH4) was from Calbiochem (San Diego, CA) and 'HzO (99.8%) from Hydro (Norway).

Purification and assay of phenylulunine hydroxyluse Phenylalanine hydroxylase was isolated from rat and

bovine liver by the method (procedure I1 D) of Shiman et al.

676

[9] with certain modifications [2, lo], giving homogeneous enzyme preparations with typical specific activities of 10 and 4 pmol L-tyrosine/(min x mg protein) when assayed at pH 6.8 and 20"C, for the rat and the bovine enzyme, respectively. The enzyme activity was measured as described [9]. The enzyme was preincubated for 3 min at 20°C in 0.1 M potassium phosphate pH 6.8, containing 60 pg catalase and 1 mM L-phenylalanine. Then, 1 mM dithiothreitol and 0.2 mM of the synthetic cofactor 6-MPH4 were added simultaneously (final volume 1 ml) and the formation of L-tyrosine was followed spectrophotometrically at 275 nm ( E = 1700 M - l cm-') at 20°C.

The concentration of protein was estimated on the basis of amino acid analyses (stable amino acids), after acid hy- drolysis for 48 h and 96 h, and the published sequence of the rat enzyme [l 11 and the amino acid composition of the bovine enzyme [lo]. A subunit mass of 50 kDa was assumed in the calculations.

Inhibition of phenylulanine hydroxylase activity by L-noradrenuline and Tris

Samples of the rat enzyme were preincubated for 3 min at 20°C with 0.1 M potassium phosphate pH 7.25,60 pg catalase and 1 mM L-phenylalanine. Then, variable concentrations of L-noradrenaline (1 - 50 pM) or Tris (10-200 mM) were added and the samples incubated for an additional period of 3 min at 20°C. The reaction was initiated by the addition of 6-MPH4 (0.01 -0.4 mM) and 1 mM dithiothreitol (final volume 1 ml), and the formation of tyrosine was followed at 275 nm).

Metal analyses The metal content of the enzyme preparations was deter-

mined by a Perkin-Elmer model 402 atomic absorption spectrophotometer equipped with a graphite furnace (type HGA-76B from Perkin Elmer). The bovine and rat enzyme contained 0.94 0.06 and 1.19 0.067 atom ironlenzyme subunit, respectively. The bovine enzyme also contained about 0.2 mol zinc/mol enzyme subunit.

EPR spectroscopy X-band EPR spectra were obtained with a Bruker ESP

300 EPR system with an Oxford Instruments helium flow cryostat (ESR-9). The spectra were obtained at 3.6 K with 0.25 mW microwave power, 10 G (1 mT) field modulation, 9.235 GHz microwave frequency and 1.3 s time constant. The samples, either in 15 mM Tris/HCl pH 7.25 (at 20°C) containing 0.2 M KCI, or in 125 mM potassium phosphate pH 7.4 (at 20°C) containing 0.1 M KCl, or in the latter buffer to which the glucose oxidase system [12] had been added (see below), were kept in EPR tubes (inner diameter = 4 mm) and frozen in liquid nitrogen. Additions and incubations with L- noradrenaline, 6-MPH4, dithiothreitol and sodium dithionite were performed in the EPR tubes. Addition of sodium dithionite was carried out under argon atmosphere. The measurement of the relative intensity of the signal at g = 4.3 and the interconversion of rhombic and axial type of signals have been carried out by double integration of the spectra.

Generation of anuerobic conditions by the glucose oxidase system

All the solutions were made anaerobic by the glucose oxidase/catalase/glucose system as described [12] : 2 p1 glucose

oxidase stock solution (20 mg/ml) and catalase (2 mg/ml) were added/ml buffer (125 mM potassium phosphate pH 7.4 containing 0.1 M KCl for EPR measurements, or 20 mM potassium Hepes/phosphate pH 7.25 containing 0.2 M KCI for equilibrium dialysis experiments), containing 17 mM glucose and incubated for 30 min at 20"C, before phenylalanine hydroxylase was added. Oxygen was found to be effectively eliminated from the solutions within 20 min (tested by an oxygen electrode) and no change in pH was observed. Phenyl- alanine hydroxylase (56 pM), which had been incubated in the glucose oxidase system for 30 min at 20°C, did not show any loss of activity as compared to the control enzyme, preincubated under aerobic conditions.

Binding of' L-noradrenuline

The binding of L-noradrenaline to the enzyme was carried out by an equilibrium microdialysis procedure as described [3], utilizing a Hoefer instrument and Visking 30/32 dialysis membranes.

400-MHz H-NMR spectroscopy

Prior to 'H-NMR analysis, the enzyme samples were re- peatedly exchanged with 'HzO (99.8%). A typical sample (3 ml) of rat phenylalanine hydroxylase (3 mg/ml, originally in 15 mM Tris/HCl pH 7.25 containing 0.2 M KC1) was first dialysed against 50 ml 'HzO containing 20 mM potassium phosphate pH * 7.2- 7.4 (pH * = uncorrected value in 'H20) and 0.2 M KC1 for 30 min at 4°C and then subjected to three or four cycles of 20-fold concentration (Amicon Centricon 30 microconcentrator) and resuspension in the dialysis buffer, all operations at 4°C. The sample, with a final volume of approximately 150 p1 (0.8 - 1 mM enzyme subunit), was cen- trifuged in an Eppendorf microfuge and the supernatant trans- ferred to cylindrical semi-micro cell assemblies of 1 00-p1 sample size (WilmaCGEss Co.) (outer diameter = 5 mm). L-Phenylalanine and L-noradrenahne were dissolved in 'HzO (99.8%), containing 20 mM potassium phosphate pH * 7.2 - 7.4 and 0.2 M KC1. The assignments of resonance lines from L-phenylalanine and L-noradrenaline protons were obtained from the coupling patterns and the nuclear Overhauser effects.

The spectra were recorded at a probe temperature of 293 K on a Bruker AM-400 WB spectrometer using internal deu- terium lock and standard Fourier-transformation techniques (see figure legends for parameters). The chemical shifts were referenced to 2,2-dimethyl-2-silapentane-5-sulfonate.

The relaxation times, T1, were measured at 293 K by using a standard inversion-recovery sequence, with a composite 90" pulse for inversion (64 transients, recycle delay 5 x TI), and exponential non-linear fit of the data, using the programs available on the Bruker-400 FT-NMR data system. No apodization was applied in experiments aimed to determine T1 values in binding experiments.

RESULTS

EPR spectra of bovine phenylulunine hydroxyluse and the efiect of oxygen, L --henv(a/anzhe, cofictor andL-noradrenaline

Bovine phenylalanine hydroxylase as isolated contained 0.94 0.06 atom iron bound/subunit. In 125 mM potassium phosphate pH 7.4, 0.1 M KC1, the enzyme revealed a low- temperature (3.6 K) EPR spectrum typical of high-spin

677

g= 9.2 7.0 5.2 4.3 t t t t

al > m .- - ._ b U c .- I a 0, n

a LT w

200 G - Fig. 1. EPR spectra (low-field region) of bovine phenylalanine hydroxylase and the formution of a L-noradrenaline - enzyme complex. The spectra were recorded in 125 mM potassium phosphate pH 7.4 (at 20':C), containing 0.1 M KCI. Spectrum A, enzyme as isolated (54 pM subunit); spectra B, C and D, sample A after incubation for 5 min at 4°C with 7, 50 and 150 pM L-noradrenaline, respectively; spectrum E, same as D after incubation with 5 mM sodium dithionite for 5 rnin at 432. All spectra are corrected for dilutions and were recorded at the following conditions: 3.6 K, 0.25 mW microwave power, 10 G (1 mT) field modulation, 9.235 GHz microwave frequency, sweep range 1600 GHz, scan time 5.5 rnin and 1 accumulation. Receiver gain 1 x lo6. 200 G = 20 mT

(S = 5/2) Fe(II1) (Fig. 1, spectrum A). The spectrum is dominated by a split resonance centered around g = 4.3, indicating that the high-spin iron is in a ligand field of rhombic symmetry, and with an IE/iol value of approximately 0.33-0.31. The relatively intense and nearly isotropic resonance around g = 4.3 originates from the middle Kramer's doublet, while the very weak resonances from the lower Kramer's doublets are spread out in the region g = 9.7-9.2 (Fig. 1, spectrum A). Also minor species with g-values at 7.0, 5.2 and 1.9 and lE/ DI z 0.04 are present. On the other hand, when the enzyme solution is prepared in Tris buffer, the EPR spectrum also revealed resonances at g-values 6.7, 5.3 and 2.0 (see below).

In order to determine the effect of dioxygen on the EPR spectrum of bovine phenylalanine hydroxylase, the enzyme sample was prepared anaerobically by the glucose oxidase system [12] in 125 mM potassium phosphate pH 7.4 containing 0.1 M KCl. Under these conditions, an EPR spectrum similar to that of Fig. 1, spectrum A was observed, indicating that dioxygen does not change the redox state or the coordination of the iron in the enzyme. A similar spectrum to that of Fig. 1, spectrum A, was also obtained when the enzyme was incubated for 5 rnin at 20°C with 1 mM L-phenylalanine. Subsequent addition of 2 mM of the synthetic cofactor 6- MPH4 and 0.1 mg/ml catalase and incubation for 5 rnin at 20 "C (i. e. turnover conditions) resulted in an approximate 30% reduction of the integrated intensity of the g = 4.3 signal (computer subtraction of the EPR-spectra, results not shown).

In our recent studies [3] on the binding of the inhibitor L-noradrenaline to phenylalanine hydroxylase we have found that about 0.5 mol L-noradrenaline binds/mol subunit of the rat enzyme with relatively high affinity (half-maximal binding, S50, at 0.25 pM L-noradrenaline) and only about 0.3 mol/mol subunit of the bovine enzyme (S50 = 0.43 pM L-noradrenaline), in both cases with positive cooperativity. Spectroscopic studies have revealed that catecholamines (noradrenaline, adrenaline and dopamine) bind to the enzyme by a direct

coordination to the high spin (S = 5/2) Fe(II1) at the catalytic site [13, 141 and reduction of this iron to the ferrous state eliminates their binding [3,5]. Catecholamine binding changes the coordination geometry of the Fe(II1) at the active site and the formation of the catecholamine-iron complex in the enzyme can be followed by EPR spectroscopy [15]. The nearly axial type of EPR spectrum with g-values at 7.0 and 5.2 increased with increasing concentration of L-noradrenaline (Fig. 1, spectra B-D). The g-values and line widths were identical to those observed for the catecholamine-Fe(II1) complex in bovine adrenal tyrosine hydroxylase as isolated [ 16, 171. Concomitant to the appearance of the catecholamine- Fe(II1) complex, a decrease (= 30%) of the high-spin Fe(1II) with g-values around 4.3 was observed at saturating concentrations of L-noradrenaline (Fig. 1, spectrum D). Similar results were obtained when the complex was generated at anaerobic conditions (data not shown). At the highest concentration of L-noradrenaline (0.150 mM), the enzyme revealed no residual catalytic activity, indicating that all the enzymatically active iron was in the form of a catecholamine complex. Fur- thermore, neither the complexed iron (g-values at 7.0 and 5.2), nor the remaining iron (g = 4.3) could be reduced by the pterin cofactor, and a spectrum similar to that shown in Fig. 1, spectrum D, was obtained after incubation for 5 rnin at 20°C with 1 mM of the synthetic cofactor 6-MPH4. As seen in Fig. 1, spectrum E, the remaining iron with g-value around 4.3 could only be effectively reduced after a 5-min incubation of the enzyme with 5 m M sodium dithionite under argon flushing, while the iron complexed with L-noradrenaline was only partially reduced by dithionite.

Effect of Tris on the EPR spectrum of bovine phenylalanine hydroxylase

Tris has been reported to inhibit phenylalanine hydroxylase activity in a competitive manner with respect to the pteridine cofactor [5]. An apparent Ki of about 80 mM was estimated for the rat enzyme at pH 7.8, but no inhibition was found at pH values lower than 6.8 [5]. Our studies on the bovine enzyme have shown that both Tris and L-noradrenaline inhibit the hydroxylase activity, competitive to the cofactor 6-MPH4, and the apparent Ki was estimated to be 74 mM for Tris and 0.18 pM for L-noradrenaline at pH 7.25 and 20°C.

When the EPR spectrum of the bovine enzyme was recorded at 3.6 K in 15 mM Tris pH 7.25 (at 20"C), an axial type of EPR signal with gcff at 6.7, 5.3 and 2.0 was very pronounced (Fig. 2, spectrum B) as compared with the intensity of these type of signals in the EPR spectrum of the native enzyme in phosphate buffer (Fig. 2, spectrum A), which only has an intense g = 4.3 signal. Hence, the EPR spectrum of the bovine enzyme in Tris buffer, with both axial and rhombic Fe(III), is similar to that previously reported for the rat enzyme in the same buffer [7,8], and also resembles that of the catecholamine enzyme complex (Fig. 1, spectrum D).

Dithiothreitol inhibits phenylalanine hydroxylase activity and the binding of L-noradrenaline to the enzyme

In the assay of the activity of aromatic amino acid hy- droxylases, dithiothreitol is frequently used to regenerate the tetrahydro form of the pteridine cofactor [18,19]. Quite unexpected, however, was our finding that preincubation of both the bovine and the rat enzyme for 30 min and 20°C with 1 mM dithiothreitol, in the absence of added substrate and cofactor, almost completely abolished the catalytic activity. Preincu-

678

.:1 t t

V C

200 G - Fig. 2. Effect of'the buffer on the E P R spectrum of bovinephenylulanine hydroxylase. (A) Enzyme as isolated (50 pM subunit) in 125 mM potassium phosphate pH 7.4 (at 20"C), containing 0.1 M KCI. (B) Enzyme as isolated (70 pM subunit) in 15 mM Tris/HCl pH 7.25 (at 20°C), containing 0.2 M KCI. (C) Buffer blank, similar for both 125 mM potassium phosphate pH 7.4, containing 0.1 M KCI, and 15 mM Tris/HCl pH 7.25, containing0.2 M KCI. The EPR conditions were as in Fig. 1. 200 G = 20 mT

g= 4.3

+

J l C

200 G - Fig. 3. EPR spectra (low-field region) of bovine phenylalanine hydroxylase and the effect of dithiothreitol. Spectrum A, enzyme as isolated (54 pM subunit) in 125 mM potassium phosphate pH 7.4 (at 20"C), containing 0.1 M KCI. Spectrum B, same as A after incubation with 1 mM dithiothreitol for 40 min at 4°C. Spectrum C, same as B after incubation for 5 rnin with 150 pM L-noradrenaline. Spectrum D, same as C after additional incubation with 5 mM sodium dithionite for 5 rnin at 4'C under argon flushing. All spectra have been corrected for dilutions. EPR conditions as in Fig. 1. 200 G 3 20 mT

bation with 1 mM mercaptoethanol (for 30 rnin and 20°C) also reduced the activity of both enzymes by about 60%.

The effect of dithiothreitol on the EPR spectrum of the bovine enzyme is shown in Fig. 3. No reduction of its ferric iron was observed on incubation with 1 mM dithiothreitol under aerobic conditions (Fig. 3, spectra A and B); only a small change in the degree of splitting of the signal centered at g = 4.3 was found. On the other hand, under anaerobic

conditions (glucose oxidase system), the amplitude of the signal at 4.3 decreased slightly (about 10%) upon incubation with 1 mM dithiothreitol for 30 min, indicating some reduction of Fe(II1) to the Fe(I1) form (data not shown). In contrast to the spectra shown in Fig. 1, spectra B-D, preincubation of the enzyme with 1 mM dithiothreitol for 30 rnin at 4°C prevented the formation of a catecholamine-enzyme complex (Fig. 3, spectrum C). The finding was confirmed by equilibrium microdialysis experiments, in which radiolabeled L- noradrenaline was found not to bind to the bovine enzyme under these conditions. Furthermore, the Fe(II1) in the dithiothreitol-treated enzyme could not be reduced by 1 mM 6-MPH4 cofactor (results not shown), but only by incubation with 5 mM sodium dithionite (Fig. 3, spectrum D).

400-MHz H - N M R spectra of rat phenylalanine hydroxylase and the interactions of L-phenylalanine and L-noradrenaline with its paramagnetic iron

The 400-MHz 'H-NMR spectrum of native, resting rat phenylalanine hydroxylase between 5.5 - 8.0 ppm (i. e. the aromatic region) revealed, as expected, little fine structure due to the slow tumbling rate of this large highly folded protein molecule (200 kDa). The relatively broad line widths of the proton resonances from the many aromatic residues (i.e. 27 Phe, 22 Tyr, 11 His and 3 Trp residues [l 11) overlap, and only a broad spectrum between 7.5-6.75 pprn was observed. The primary goal of our study was to probe the interactions of the substrate (L-phenylalanine) and an inhibitory analogue (L-noradrenaline) with the paramagnetic iron at the active site. The short relaxation time of these paramagnetically enhanced resonances permits the use of a rapid pulse sequence (see legend to Fig. 4), which secondarily reduces the relative intensity of other resonances ( T , > 1 s).

When the enzyme was incubated for 20 rnin at 20°C in the presence of a small excess of the substrate L-phenylalanine (the molar ratio of L-Phe/enzyme subunit = 1.25: l), a new small and rather broad signal centered at 6 = 7.38 ppm was observed (Fig. 4, spectrum B). The fine structure of the spectrum obtained for free L-phenylalanine (without any enzyme present), at the same acquisition conditions, is shown in Fig. 5, spectrum B, for comparison. Upon further incubation of the enzyme with additional substrate (final molar ratio of L-Phe/ enzyme subunit = 3.75: l) , broad signals with 6 7.33 and 7.42 pprn increased in amplitude and appeared as clear peaks out of the aromatic proton envelope from the protein (Fig. 4, spectrum C). Separate signals for the bound and free substrate were not observed, indicating a fast exchange (on the NMR time scale) between the bound and free L-phenylalanine. This conclusion was further supported by a separate series of experiments in which the line width of the aromatic protons of the substrate was found to increase linearly with the concentration of added enzyme (results not shown). Thus, the L-phenylaia- nine which is added to the enzyme solution exchanges rapidly between a free, rapidly tumbling form, not affected by paramagnetic relaxation, and a form or forms with signals that are broadened due to interaction with the paramagnetic iron center of the enzyme (shortened T1 and T2), as well as by the slow tumbling rate of the enzyme. The ratio between the different forms is changed in favour of the free form of the substrate following incubation with 1 mM L-noradrenaline (the ratio of L-noradrenaline/subunit = 1.25: 1) (Fig. 4, spectrum D), as indicated by more narrow line widths of the signals at 6 = 7.33 and 7.42 ppm. Thus, the changes in the NMR spectrum show that L-noradrenaline, which binds with high

679

3-5 2,6

I 1 A

8.0 7.5 7.0 6.5 6.0 Chemical shift (ppm)

Fig. 4. The 400-MHz ' H - N M R spectra (aromatic region) of rat phe- nvlulanine hytiroxyfuse und its interactions with L-phenyfaiunine und L-norudrenuline. Recorded in 'H20 (99.8%), containing 20 mM potassium phosphate pH*7.2 and 0.2 M KC1 at 293 K ; 50" radiofrequency pulse (5 ps); acquisition time 0.197 s; no relaxation delay; line broadening 0.5 Hz; 1000 transients. Spectrum A, enzyme as isolated (0.8 mM subunit). Spectrum B, same as A after addition of 1 mM L-phenylalanine and incubation for approximately 20 min at 293 K. Spectrum C, same as B after further addition of 2 mM L-phenylalanine and incubation for 10 min at 293 K. Spectrum D, same as C after further addition of 1 mM L-noradrenaline and incubation for approximately 5 min. Spectrum E, enzyme as isolated (0.3 mM subunit) and 0.6 rnM 1.-noradrenaline

affinity exclusively at the active site [3, 131, displaces the substrate from its binding at this site. Furthermore, the binding of the catecholamine is characterized by a broad signal at 6 = 6.94 ppm (Fig. 4, spectra D and E), corresponding to the aromatic protons of L-noradrenaline (Fig. 5, spectrum D), a feature characteristic of a form of the catecholamine bound to the slowly tumbling enzyme and close to the paramagnetic center. These data support the conclusion that L-phenylalanine and L-noradrenaline bind at the same or overlapping sites (i. e. close to the paramagnetic iron) of phenylalanine hydroxylase, and the L-noradrenaline displaces the substrate due to its higher affinity for the enzyme.

The longitudinal relaxation time ( T I ) for the signals from the aromatic protons of L-phenylalanine and the aromatic protons of L-noradrenaline, as well as the changes in T1 of the residual HDO signal at 6 = 4.8 ppm were measured in the presence (Table 1) or the absence of the enzyme (Table 2). The residual water signal was found to have a very low T1 value (0.55 s) in the presence of 0.8 M rat phenylalanine hydroxylase. It was found that the rate of relaxation (T; ') of the residual water signal was proportional to the concentration of the enzyme, i.e. T1(HDO) x [enzyme subunit] z 0.43 5 0.018 mM . s, indicating that the Fe(Ii1) site in the enzyme is accessible to the solvent. Thus H 2 0 should be considered a putative ligand to the iron in the resting enzyme. When this Fe(II1) was reduced to low-spin Fe(I1) by a strong reductant such as dithionite, the relaxation time of HDO

v 5

2.5

1 6

8.0 7.5 7.0 6.5 6.0 Chemical shift (ppm)

Fig. 5. The 400-MHz ' H NMR-spectra (aromatic region) of L-phenylalanine (spectra A and B ) and L-noradrenaline (spectra C and 0). Recorded in 'H20 (99.8%), containing 20 mM potassium phosphate pH * 7.2 and 0.2 M KCI at 293 K, at aconcentration of 5 mM for both compounds. Spectra B and D were recorded at the same conditions as in Fig. 4 and for spectra B and D the acquisition time is 1.01 s and the line broadening 0.1 Hz

increased to a value characteristic of a diamagnetic system ( T I = 14 s, Tables 1 and 2), which clearly demonstrates that the high-spin Fe(II1) at the active site is mainly responsible for the rapid relaxation of resonances close to it. The increase in Tl of the residual water signal upon addition of the substrate (Table 1) points to a binding of the substrate very close to the paramagnetic iron at the active site, with the displacement of H 2 0 from its proximity to the metal.The short TI of the aromatic protons of L-phenylalanine interacting with the enzyme (6 = 7.33 ppm), also demonstrates paramagnetic relaxation and confirms the proximity of the aromatic region of the amino acid to the Fe(II1) (Tables 1 and 2). On further addition of L-noradrenaline, TI increased for both the residual water signal (6 = 4.8 ppm) and the aromatic protons of L- phenylalanine, suggesting a displacement of the substrate from the active site by the aromatic moiety of the high-affinity binding inhibitor, which revealed a reduction in T I from 2.49 s (free form) to 0.14 s. After the addition of L-noradrenaline the T1 values of the aromatic part of the substrate and the residual water signal are still shorter than in diamagnetic systems (Tables 1 and 2). This may be explained by the heterogeneity of the iron population with respect to L-noradrenaline binding (Fig. 1) [3].

Since the inhibition of the enzyme by L-noradrenaline is kinetically competitive to the pterin cofactor (see above), the

680

Table 1. Summary of the eflects uf L-phenylalanine and L-noradrenaline binding to rat phenylalanine hydroxylasi. on the longitudinal relaxation times (T,) of their aromatic protons Experimental details were as stated in Materials and Methods and in the legend to Fig. 4. The TI of the resonances in the aromatic region from L-phenylalanine at 6 = 7.33 (or 7.38 in sample B) ppm and L-noradrenaline (NA) at 6 = 6.94 ppm, as well as the residual HDO signal at 6 = 4.8 ppm, was measured at 293 K. Measurements were made after preincubation of the sample A (0.8 mM rat phenylalanine hydroxylase, prepared in 99.8% 'H20, pH * 7.4, containing 20 mM potassium phosphate 0.2 M KCI) at 20°C for 20 rnin with 1 mM Phe (sample B), with 3 mM Phe for 30 rnin (sample C) and with 1 mM sodium dithionite for 5 min (sample E). Sample D was prepared by preincubation of sample C with 1 mM NA for 5 min at 20°C. n.s., no signal present. The error in T1 values is < 6%

Sample TI for

HDO Phe NA

S

A: 0.8 mM enzyme subunit 0.55 n. s. n.s. B: A + 1 rnM Phe 0.58 0.1 1 n.s. C : B + 2 mM Phe 0.77 0.14 n.s. D : C + l m M N A 0.93 0.31 0.144 E: A + 1 mM dithionite 14.13 n. s. n.s.

Table 2. Summary of the longitudinal relaxation times ( T I ) of the aromatic protons of free L-phenylalanine and L-noradrenaline Experimental details were as stated in the legend to Table 1. The T1 of the resonances in the aromatic region from L-phenylalanine at 6 = 7.42/7.33 ppm and L-noradrenaline (NA) at 6 = 6.96/6.88 ppm, as well as the residual HDO signals at 6 = 4.8 ppm, were measured at 293 K. The buffer was 20 mM deuterated potassium phosphate, prepared in 'H20 (99.8%), pH* 7.4, containing 0.2 M KCI. The solutions of Phe and NA were prepared in this deuterated buffer. n. d., not determined, n.s., no signal present. The error in TI values is 2%

Sample T1

HDO Phe NA

4.8 ppm 7.42 pprn 7.33 ppm 6.96 ppm 6.88 ppm

S

Buffer 13.47 n. s. n.s. n. s. n. s. 5 m M P h e 12.50 2.36 2.24 n.s. n.s. 5 m M N A n.d. n. s. n. s. 2.49 1.87

study of the binding of cofactor to the enzyme by paramagnetic relaxation enhancement could provide further information about the active site. However, the ferric iron in the enzyme is reduced to the ferrous state by the cofactor [5] and the paramagnetism specific of high-spin Fe(II1) is lost. Furthermore, the cofactor is simultaneously oxidized to a quinoid dihydropterin [5, 61 which makes it impossible to use the same approach as we have employed to study the competitive interaction of L-phenylalanine and L-noradren- dine with the enzyme.

DISCUSSION

Effect of buffer ions on EPR spectra of phenylalanine hydroxylase

EPR spectroscopy is a valuable method to determine the oxidation state and the electronic environment of transition metal complexes containing unpaired electrons, e. g. in metallo enzymes. In agreement with the EPR spectra of rat phenylalanine hydroxylase [4, 7, 81, the bovine enzyme also contains high-spin d5 ferric ions in the ground state. There is, however, a discrepancy in the literature with respect to the main features of the spectrum of the isolated native enzymes and the g- values of the iron populations [4, 7, 15, 201 (and this work). In the present study, experimental evidence is presented that these differences are related to variations in buffer ions present in the enzyme solution. Thus, in phosphate buffer (pH 7.4 at 20"C), the EPR spectrum is dominated by a signal at g = 4.3, associated with an S = 5/2 system with lE/D] = 0.33-0.31, which corresponds to high-spin iron in an environment of largely rhombic symmetry. By contrast, when the EPR spectrum is recorded in Tris buffer (pH 7.25 at 20"C), the bovine enzyme as isolated shows an axial type of signals (g = 6.7 and 5.3) in addition to a g = 4.3 signal, similar to that reported for the rat enzyme in the same buffer [7,8]. A possible explanation for the effect of the buffer in the EPR spectra is discussed below.

The binding of substrate and the inhibitor L-noradrenaline to phenylalanine hydroxylase

We have recently studied the equilibrium binding of L-noradrenaline to rat and bovine phenylalanine hydroxylase [3]. In the case of the bovine enzyme (0.94 & 0.06 atom iron/ subunit), only 0.3 mol L-noradrenaline was found to be bound/mol subunit with high affinity (half-maximal binding, So.s = 0.43 pM) [3]. As seen by EPR spectroscopy, the formation of the catecholamine - phenylalanine-hydroxylase complex (in phosphate buffer) gives rise to EPR signals with g = 7.0, 5.2 and 1.9 ]E/Dl = 0.04 (Fig. 1). This type of signals, ascribed to high-spin ferric iron in an essentially axial environment, has been observed in many heme proteins, in bovine tyrosine hydroxylase as isolated [ 161, in protocatechuate 3,4- dioxygenase after addition of substrates [21, 221 and in synthetic Fe(1igand)catechol complexes [23]. A strong reductant, such as dithionite, was unable to reduce completely the catecholamine-iron complex (Fig. 1). This finding suggests that, in addition to a steric hindrance at the active site, the L-noradrenaline binding may inhibit the catalytic activity by decreasing the redox potential of the iron, and thus prevent its reduction by the pterin cofactor.

By comparing the integrated intensities of the EPR signals at g = 4.3 of the native resting enzyme (in phosphate buffer) and when the enzyme is saturated with L-noradrenaline, it was estimated that approximately 30% of the ferric iron in the bovine enzyme can form a Fe(1II)-catecholamine complex (g = 7.0, 5.2 and 1.9), which agrees with our previous calculations based on equilibrium binding studies [3]. If we consider that the same fraction of the iron is reduced by the cofactor under turnover conditions (this work), it seems that only the iron which participates in catalysis can bind catecholamines. Therefore, the stoichiometry of catecholamine binding ap- pears to give a useful estimate of the fraction of catalytically active iron in phenylalanine hydroxylase [3] and thus represent active-site probes which are very useful in studies on the bind-

68 1

ing of the substrate L-phenylalanine to the enzyme by NMR spectroscopy.

The presence of paramagnetic iron [Fe(III), S = 5/21 in phenylalanine hydroxylase makes it possible to study the active site of this enzyme by 'H-NMR spectroscopy. High-spin Fe(II1) has isotropic values of the magnetic susceptibility and, therefore, relaxation effects rather than dipolar shifts are expected in the nuclear resonances close to it [24,25], as we have actually observed in the present study. Nuclei in the vicinity of the paramagnetic iron will experience a large local field which will fluctuate due to electron spin relaxation and molecular motions, giving rise to the phenomenon of paramagnetic relaxation enhancement [24-261. The magnitude of this enhancement can provide information about the distance of the nucleus from the paramagnetic ion. Our studies on phenylalanine hydroxylase by one-dimensional 400-MHz 'H-NMR spectroscopy add some important information to our present knowledge about the Fe(II1) binding site of the enzyme. First, this site is accessible to water, which is a putative ligand to the Fe(II1) in the native, resting state. Secondly, the binding of both L-phenylalanine and L-noradrenaline occurs close to the paramagnetic iron at the active site, and thus displacing water. Finally, L-noradrenaline and L-phenylalanine compete for the same or overlapping binding sites at the active site of the enzyme, L-noradrenaline having the highest affinity. As seen by resonance Raman spectroscopy [13], the binding of catecholamines to phenylalanine hydroxylase occurs by coordination of the catechol moiety to the ferric iron, most likely by a bidentate catecholate ligand-to-metal charge-transfer interaction [13], in a similar way as in the analogous enzyme tyrosine hydroxylase [14]. The EPR data also support the conclusion 131 that catecholamines coordinate only to the catalytically active iron (this work). The fact that the binding of L-noradrenaline at the active site of phenylalanine hydroxylase occurs with positive cooperativity [3] and that the catecholamine releases L-phenylalanine from binding at the same or overlapping sites, as seen by 'H-NMR, strongly support the conclusion that the positive cooperativity of substrate binding [19, 27, 281 reflects its binding at the active site, as we have recently proposed [3].

Tris us un inhibitor ofphenylulanine hydroxyluse

Both catecholamines and Tris are inhibitors of bovine phenylalanine hydroxylase, i. e. competitive to the pterin cofactor ( K , values of 0.18 pM and 74 mM for L-noradrenaline and Tris, respectively, at pH 7.25 and 20"C), in agreement with previous studies on the rat enzyme [5, 291. Moreover, in Tris buffer the enzyme shows an EPR spectrum at 3.6 K with axial type of signals (Fig. 2), which resembles that observed for the catecholamine-enzyme complex prepared in phosphate buffer (Fig. 1). In a previous study [8] only a slight downfield shift of the g = 6.7 signal (Fig. 2) was found on addition of catecholamines to the rat enzyme in Tris buffer (pH 7.25 at 20°C). Thus, EPR spectroscopy suggests that both Tris and L-noradrenaline interact with the paramagnetic iron in the enzyme, preventing its reduction by the pterin cofactor IS]. In spite of the low affinity of binding of Tris, as compared to L-noradrenaline (see above), and the low concentration used in the EPR studies, at 3.6 K Tris seems to form a complex with an equivalent fraction of the paramagnetic iron in the enzyme, as observed for the catecholamine (Figs 1 and 2). This finding might partly be explained by an additional effect on pH of freezing. Thus, it has been reported that freezing to 77 K of a Tris buffer with pH = 7.25 at 20°C results in a pH

of about 10.5, whereas a potassium phosphate buffer of pH 7.4 at 20°C gives a pH of about 6.3 at 77 K [30, 311. Since the inhibition of the enzyme by Tris is most pronounced at alkaline conditions [5], freezing of Tris buffer would favour its binding to the enzyme. Thus, the many similarities in the inhibition by Tris and catecholamines, i. e. competitive with the cofactor [5, 291, decreasing the rate of reduction of the Fe(II1) [5], the pH dependence [3, 51 and our EPR results, indicate a coordination of both inhibitors to the catalytically active Fe(II1) in the enzyme.

Related to this point, Wallick et al. [7] and Bloom et al. [8] found that the activity of the rat enzyme, in 30 mM Tris pH 7.25, containing about 1 atom iron/enzyme subunit, was proportional to the integrated intensity of the EPR signals with g,,, = 6.7, 5.4 and 2 (accounting for approx. 40% of the total iron), and inversely proportional to the remaining g = 4.3 signal (iron that could not be reduced by the cofactor). It seems that the iron interacting with Tris (as visualised by EPR at 5 K), accounts for the active form of iron in the enzyme. The iron that interacts with Tris during EPR measurements (alkaline pH) can still be effectively reduced by the cofactor and participate in catalysis when the sample is brought to 2 0 T and the pH returns to its original values, since Tris is only a weak inhibitor at neutral pH. By contrast, phosphate has no inhibitory effect even at alkaline pH [5].

Although the chemical basis of the pH changes obtained on freezing are not clear [31], they may partly explain the discrepancy which exists in the literature with respect to the EPR spectrum of the catalytically active form(s) of phenylalanine hydroxylase. Of particular interest is that the first reported EPR spectrum of the rat enzyme [4], obtained in Tris buffer (pH 6.8 at 20°C) at 77 K, did not show any axial type of signal (g = 7.0 -4.3), which is probably explained by a temperature effect. Thus, due to the rapid relaxation of resonances from the lowest Kramer's doublet, the low-field features of the EPR spectrum arising from it are in many cases not detected above 30 K [21, 321. Moreover, the spin population of the lowest doublet decreases at 77 K [32].

Dithiothreitol us an inhibitor ofphenylulanine hydroxylase

Quite unexpected was our finding that preincubation of both rat and bovine phenylalanine hydroxylase with dithiothreitol inhibits their catalytic activity and prevents the binding of L-noradrenaline to the enzymes. It has been shown by EPR spectroscopy that dithiothreitol reduces the catalytic Cu(I1) to Cu(1) in phenylalanine hydroxylase isolated from Chromobucterium violaceum [33] and it has been suggested that dithiothreitol also reduces the enzyme-bound ferric iron to the ferrous form in the rat enzyme [2]. However, our present EPR studies under aerobic conditions with the bovine enzyme (Fig. 3) do not support this proposal. Only a small change in the degree of splitting of the signal centered at g = 4.3 was observed on addition of dithiothreitol. Thus, dithiothreitol seems to perturb the conformation of the enzyme including the active site. Dithiothreitol is generally used in the assay of phenylalanine hydroxylase activity to regenerate chemically the fully reduced form of the cofactor [18, 191. On the time scale of this assay the negative effects of dithiothreitol on the enzyme are of less importance due to its positive effect on the regeneration of the cofactor, and inhibition has been found only when concentrations above 10 mM are used [18].

We are very grateful to Drs Sigridur Olafsdottir, Dagfinn Aksnes and Nils Age Frerystein for valuable discussions. This work was sup-

682

ported by grants from Nordisk Industrifond, the Norwegian Research Council for Science and Humanities and a long-term fellowship of the Federation of European Biochemical Societies (to A. Martinez).

REFERENCES 1. Kaufman, S. & Fisher, D. (1974) in Molecular mechanisms of

oxygen activation (Hayaishi, O., ed.) pp. 285- 369, Academic Press, New York.

2. Gottschall, D. W., Dietrich, R. F., Benkovic, S. J. & Shiman, R. (1982) J . Biol. Chem. 257, 845 - 849.

3. Martinez, A., Haavik, J. & Flatmark, T. (1990) Eur. J . Biochem.

4. Fisher, D. B., Kirkwood, R. & Kaufman, S. (1973) J . Biol. Chem.

5. Marota, J. J. A. & Shiman, R. (1984) Biochemistry 23, 1303-

6. Haavik, J., Dsskeland, A. P. & Flatmark, T. (1986) Eur. J . Bio-

7. Wallick, D. E., Bloom, L. M., Gaffney, B. J. & Benkovic, S. J.

8. Bloom, L. M., Benkovic, S. J. &Gaffney, B. J. (1986) Biochemistry

9. Shiman, R., Gray, D. W. & Pater, A. (1979) J. Biol. Chem. 254,

10. Dsskeland, A., Ljones, T., Skotland, T. & Flatmark, T. (1982) Neurochem. Res. 7.407 -421.

11. Dahl, H.-H. M. & Mercer, J. F. B. (1986) J . Biol. Chem. 261, 4148 -4153.

12. Englander, S. W., Calhoun, D. B. & Englander, J. J. (1987) Anal. Biochem. 161, 300-306.

13. Cox, D. D., Benkovic, S. J., Bloom, L. M., Bradley, F. C., Nelson, M. J., Que, L. Jr & Wallick, D. E. (1988) J . Am. Chem. Soc.

14. Anderson, K. K., Cox, D. D., Que, L. Jr, Flatmark, T. & Haavik,

193,211 -219.

25, 5161 -5167.

1311.

chem. 160, 1 - 8.

(1 984) Biochemistry 23, 1295 - 1302.

25,4204-4210.

11 300- 11 306.

106, 2026 - 2032.

J. (1988) J . Biol. Chem. 263,18621 -18626.

15. Martinez, A,, Anderson, K. K., Dahle, G., Flatmark, T. & Haavik, J. (1990) in Chemistry and biology ofpteridines 1989. Pteridines and folic acid derivatives (Curtis, H. Ch., Ghisla, S. & Blau, N., eds) pp. 644-647, Walter de Gruyter, Berlin.

16. Haavik, J., Anderson, K. K., Petersson, L. & Flatmark,T. (1988) Biochim. Biophys. Acta 953, 142 - 156.

17. Anderson, K. K., Haavik, J., Martinez, A,, Flatmark, T. & Petersson, L. (1989) FEBS Lett. 258, 9-12.

18. Bublitz, C. (1971) Biochim. Biophys. Acta 191, 249-256. 19. Shiman, R. (1985) in Folates andpterins, vol. 2 (Blakley, R. L. &

Benkovic, S. J., eds) pp. 179-249, John Wiley & Sons, New York.

20. Yang, A.3 . & Gaffney, B. J. (1987) Biophys. J . 51, 55-67. 21. Blumberg, W. E. & Peisach, J. (1973) Ann. N . Y . Acad. Sci. 222,

22. Whittaker, J. W., Lipscomb, J. D., Kent, T. A. & Miinck, E.

23. Cox, D. D. (1988) Ph.D. Thesis, University of Minnesota. 24. Morris, A. T. & Dwek, R. A. (1977) Q. Rev. Biophys., 421 -484. 25. Wiithrich, K. (1976) in N M R in biological research: peptides and

proteins, North-Holland Publishing Company, Amsterdam. 26. Bertini, I. & Luchinat, C. (1986) in N M R of paramagnetic mol-

ecules in biological systems, Ch. 3, Benjamin/Cummings Pub- lishing, California.

27. Shiman, R. & Gray, D. W. (1980) J . Biol. Chem. 255, 4793- 4800.

28. Parniak, M. A. & Kaufman, S. (1981) J . Biol. Chem. 256,6876- 6882.

29. Bublitz, C. (1971) Biochem. Pharmacol. 20, 2543-2553. 30. Orii, Y. & Morita, M. (1977) J . Biochem. (Tokyo) 81, 163-168. 31. Williams-Smith, D. L., Bray, R. C., Barber, M. J., Tsopanakis,

A. D. & Vincent, S. P. (1977) Biochem. J . 167, 593-600. 32. Orme-Johnson, N. R. & Orme-Johnson, W. H. (1978) Methods

Enzymol. 52,252 - 257. 33. Pember, S. O., Villafranca, J. J. & Benkovic, S. J. (1986) Biochem-

istry 25, 661 1 - 6619.

539 - 560.

(1984) J. Biol. Chem. 259,4466-4474.

Documents

EPR and 1H-NMR spectroscopic studies on the paramagnetic iron at the active site of phenylalanine hydroxylase and its interaction with substrates and inhibitors