Báo cáo khoa học: Endogenous tetrahydroisoquinolines associated with Parkinson’s disease mimic the feedback inhibition of tyrosine hydroxylase by catecholamines doc

Protein kinase A also fully restored enzyme activity after incubation with N-methyl-norsalsolinol, demonstrating that tyrosine hydroxylase inhibition by 6,7-dihydroxylated tetrahydroisoq

Parkinson’s disease mimic the feedback inhibition

of tyrosine hydroxylase by catecholamines

Joachim Scholz1,2,*, Karen Toska3,*, Alexander Luborzewski1, Astrid Maass4, Volker Schu¨nemann5, Jan Haavik3and Andreas Moser1

1 Neurochemistry Research Group, Department of Neurology, University of Lu¨beck, Germany

2 Neural Plasticity Research Group, Department of Anesthesia and Critical Care, Massachusetts General Hospital and Harvard Medical School, Charlestown, MA, USA

3 Department of Biomedicine, Section of Biochemistry and Molecular Biology, University of Bergen, Norway

4 Fraunhofer-Institute for Algorithms and Scientific Computing (SCAI), Sankt Augustin, Germany

5 Department of Physics, Technical University Kaiserslautern, Germany

Keywords

enzyme stability; feedback inhibition;

Parkinson’s disease; tetrahydroisoquinolines;

tyrosine hydroxylase

Correspondence

J Scholz, Neural Plasticity Research Group,

Department of Anesthesia and Critical Care,

Massachusetts General Hospital and

Harvard Medical School, 149 13th Street,

Room 4309, Charlestown, MA 02129, USA

Fax: +1 617 7243632

Tel: +1 617 7243623

E-mail: scholz.joachim@mgh.harvard.edu

*These authors contributed equally to this

work

(Received 14 November 2007, revised 23

January 2008, accepted 28 February 2008)

doi:10.1111/j.1742-4658.2008.06365.x

N-methyl-norsalsolinol and related tetrahydroisoquinolines accumulate in the nigrostriatal system of the human brain and are increased in the cere-brospinal fluid of patients with Parkinson’s disease We show here that 6,7-dihydroxylated tetrahydroisoquinolines such as N-methyl-norsalsolinol inhibit tyrosine hydroxylase, the key enzyme in dopamine synthesis, by imitating the mechanisms of catecholamine feedback regulation Docked into a model of the enzyme’s active site, 6,7-dihydroxylated tetrahydroiso-quinolines were ligated directly to the iron in the catalytic center, occupy-ing the same position as the catecholamine inhibitor dopamine In this position, the ligands competed with the essential tetrahydropterin cofactor for access to the active site Electron paramagnetic resonance spectros-copy revealed that, like dopamine, 6,7-dihydroxylated tetrahydroisoquino-lines rapidly convert the catalytic iron to a ferric (inactive) state Catecholamine binding increases the thermal stability of tyrosine hydroxy-lase and improves its resistance to proteolysis We observed a similar effect after incubation with N-methyl-norsalsolinol or norsalsolinol Fol-lowing an initial rapid decline in tyrosine hydroxylation, the residual activity remained stable for 5 h at 37C Phosphorylation by protein kinase A facilitates the release of bound catecholamines and is the most prominent mechanism of tyrosine hydroxylase reactivation Protein kinase A also fully restored enzyme activity after incubation with N-methyl-norsalsolinol, demonstrating that tyrosine hydroxylase inhibition

by 6,7-dihydroxylated tetrahydroisoquinolines mimics all essential aspects

of catecholamine end-product regulation Increased levels of N-methyl-norsalsolinol and related tetrahydroisoquinolines are therefore likely to accelerate dopamine depletion in Parkinson’s disease

Abbreviations

CSF, cerebrospinal fluid; DA, dopamine; hTH, human tyrosine hydroxylase; L -DOPA, L-3,4-dihydroxyphenylalanine; MPTP, 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine; NMNorsal, N-methyl-norsalsolinol; NMSal, N-methyl-salsolinol; NMTIQ, N-methyl-1,2,3,4-tetrahydroisoquinoline; Norsal, norsalsolinol; PD, Parkinson’s disease; PKA, protein kinase A; ROS, reactive oxygen species; TH, tyrosine hydroxylase;

TIQ, tetrahydroisoquinoline.

N-methyl-norsalsolinol, salsolinol and

N-methyl-salso-linol are endogenous tetrahydroisoquinolines (TIQs)

formed through non-enzymatic condensation of

dopa-mine (DA) with aldehydes or pyruvic acid Increased

concentrations of these TIQs are found in the

cerebro-spinal fluid (CSF) of patients with Parkinson’s disease

(PD) [1–3] Accumulation of N-methylated TIQs in the

substantia nigra and the corpus striatum of the human

brain [2] and their structural similarity to

1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) (Fig 1)

have led to the hypothesis that TIQs are directly

involved in the degeneration of dopaminergic neurons

Like MPTP, TIQs inhibit mitochondrial respiration

However, the toxicity of TIQs is low because of their

limited ability to cross the mitochondrial membrane

[4] High concentrations of N-methyl-salsolinol

(NMSal) are required to induce apoptosis of

dopami-nergic cells in vitro [5], and NMSal causes a loss of

tyrosine hydroxylase-immunoreactive neurons in the

rat substantia nigra in vivo only after repeated

stereo-taxic injections [6] Some TIQs even have

neuroprotec-tive effects [7,8] Rather than provoking neuronal

degeneration, endogenous TIQs may interfere with DA

synthesis N-methyl-norsalsolinol (NMNorsal) [9] and

salsolinol [10,11] inhibit tyrosine hydroxylase (TH;

tyrosine 3-monooxygenase, EC 1.14.16.2), the key

enzyme in DA synthesis, in vitro, and a single injection

of NMSal into the rat corpus striatum markedly

reduces TH activity in vivo, leading to an almost

com-plete loss of DA in the absence of neuronal degenera-tion [6]

The CSF levels of TIQs increase in early PD and decrease as the disease progresses [12] TH inhibition

by endogenous TIQs may therefore be most prominent

at a critical time, when surviving substantia nigra neu-rons are challenged by the necessity to increase DA synthesis and release in order to uphold the functional integrity of the nigrostriatal pathway [13–15] Such adaptive neurochemical changes are likely to delay the appearance of clinical signs in PD, which lags several years behind the onset of dopaminergic neuron degen-eration in the substantia nigra [16] Animal models of

PD have demonstrated the plasticity of the nigrostria-tal system For example, near-complete recovery of motor function is achieved when striatal DA levels are restored by concomitant virus-mediated transfer of the genes encoding TH and GTP cyclohydrolase, the rate-limiting synthetic enzyme for the essential TH cofactor 6(R)-l-erythro-5,6,7,8-tetrahydrobiopterin (BH4) [17] Thus, understanding how TIQs block TH activity is important in order to develop treatment strategies that help to sustain dopaminergic nigrostriatal signaling in early PD

TH catalyzes the hydroxylation of tyrosine to l-3,4-dihydroxyphenylalanine (l-DOPA), which is the rate-limiting step in synthesis of the catecholamines DA, norepinephrine and epinephrine TH consists of four identical subunits that contain a C-terminal catalytic domain (residues 156–498) and an N-terminal regula-tory domain The active site of the enzyme is a 17 A˚ deep crevice with a ferrous iron atom located in its center [18] Alternative mRNA splicing of a single pri-mary transcript generates at least four isoforms of human TH (hTH) that are differentially expressed in tissues; the most prominent isoforms in the brain are hTH1 and hTH2 The isoforms differ only in the N-terminal regulatory region; the C-terminal domain with the active site is identical in all hTH isoforms, and the catalytic domain is also highly conserved across animal species and in other aromatic amino acid hydroxylases [19] TH activity is subject to intri-cate regulation Transcriptional control, modulation of RNA stability, translational regulation and enzyme stability establish a steady-state level of TH protein [20,21] Short-term regulation of TH activity includes feedback inhibition by catecholamine end products, allosteric modulation and phosphorylation-dependent activation by various kinases [22,23]

Using recombinant hTH1 and hTH4, we have exam-ined the inhibitory effect of NMNorsal and structur-ally related TIQs on TH Molecular docking revealed that 6,7-dihydroxylated TIQs associated with PD

Fig 1 Chemical structures of DA and the TIQs examined in this

study The 6,7-dihydroxylated TIQs Norsal, NMNorsal and NMSal

have an intact catechol moiety NMNorsal and NMSal are

endoge-nous compounds with structural similarity to MPTP.

compete with the essential tetrahydropterin cofactor of

hTH for access to the enzyme’s active site, whereas

MPTP has low affinity for the amino acid binding site

By binding directly to the ferrous iron atom in the

cat-alytic center and converting the iron to a ferric state,

6,7-dihydroxylated TIQs block hTH activity through a

mechanism that mimics the physiological feedback

inhibition by catecholamines Unlike DA, which

caused a near complete loss of hTH activity over time,

NMNorsal stabilized hTH, but with a reduced level of

activity

Results

6,7-Dihydroxylated TIQs inhibit human TH

The recombinant isoforms hTH1 and hTH4 produced

539 ± 41 nmolÆmin)1Æmg)1 and 564 ± 40 nmolÆ

min)1Æmg)1 l-DOPA, respectively Human TH activity

decreased in the presence of NMNorsal and NMSal,

two 6,7-dihydroxylated TIQs (Fig 1) that have

previ-ously been identified in the CSF of patients with PD

[1,3,12] NMNorsal inhibited hTH almost as strongly

(IC50= 0.3 lm) as the catecholamine end product DA

(IC50= 0.2 lm), whereas higher concentrations of

NMSal (IC50= 4.0 lm) were required to reduce hTH

activity (Fig 2A) A kinetic analysis indicated that

NMNorsal reduced hTH activity by competing with

the essential pterin cofactor (Fig 2B); hTH inhibition

by NMNorsal was noncompetitive with respect to

tyrosine (data not shown)

We compared the inhibitory effects of NMNorsal

and NMSal with those of two other TIQs,

norsalsolin-ol (Norsal) and

N-methyl-1,2,3,4-tetrahydroisoquino-line (NMTIQ) Norsal has an intact catechol moiety

like NMNorsal and NMSal, but its piperidine nitrogen

is unmethylated; NMTIQ lacks the two hydroxyl resi-dues at positions 6 and 7 of its benzene ring (Fig 1) Norsal decreased enzymatic l-DOPA synthesis with an

IC50 of 10.0 lm (Fig 2A) In contrast, hTH activity remained unchanged in the presence of the non-hydroxylated NMTIQ (Fig 2A) Consequently, hTH inhibition depends critically on the catechol moiety of 6,7-dihydroxylated TIQs Methylation of the piperidine nitrogen or a neighboring carbon modulates the effi-cacy of 6,7-dihydroxylated TIQs in reducing hTH activity, but is not responsible for the overall inhibi-tory effect

Molecular docking

TH belongs to a family of tetrahydropterin-dependent amino acid hydroxylases that also includes phenylala-nine hydroxylase and tryptophan hydroxylase These enzymes are composed of four identical subunits, each containing a divalent iron atom in its catalytic domain that is required for activity [24] To explore the mecha-nism of TH inhibition by 6,7-dihydroxylated TIQs, we identified potential binding sites of NMNorsal, NMSal and Norsal in the crystal structure of the enzyme’s cat-alytic domain (Protein Data Bank identification code 2toh) [18] using molecular docking We also deter-mined the energetically favored docking sites for NMTIQ and MPTP, and compared all conformations with the binding site of the physiological feedback inhibitor DA

The most favorable placements of NMNorsal, NMSal and Norsal overlapped almost completely and were identical to that of DA (Fig 3) This common binding mode for ligands with a catechol moiety was characterized by a tight bidentate bonding of the cate-chol oxygen atoms to the catalytic iron The mean

Fig 2 TH inhibition by NMNorsal and structurally related TIQs (A) Activity of recombinant hTH in the presence of DA and TIQs Data are shown as the percentage of the activity level in the absence of inhibitors (n = 4) (B) A Lineweaver–Burk plot of hTH activity in the presence

of NMNorsal (1.0 l M ) at various concentrations of the pterin cofactor DPH4(n = 3).

distance between the catechol oxygen atoms and the

iron was 1.74 A˚ for both NMNorsal and Norsal and

1.72 A˚ for NMSal, compared to 1.71 A˚ for DA

(Table 1) The oxygen atoms were placed opposite the

e nitrogen atoms of His331 and His336, creating a

plane perpendicular to the benzene ring of Phe300

The two oxygen atoms thus formed an intrinsic part of

the iron coordination sphere, with the piperidine rings

of the TIQs and the aminoethyl moiety of DA

project-ing from the bindproject-ing pocket (Fig 3) A potential

inter-action between the DA nitrogen and the backbone oxygen of Leu294 was outweighed by loss of rotational entropy of the DA side chain Table 1 summarizes the energy components that characterize the most favor-able conformations of NMNorsal, NMSal, Norsal and

DA Electrostatic (Coulomb) interactions with the active site iron and the surrounding TH amino acids were the largest energy contribution in all conforma-tions Separate docking runs for the (R) and (S) enantiomers of protonated NMNorsal and NMSal, respectively, revealed no differences in their binding sites or conformational energy components

The binding site of the 6,7-dihydroxylated TIQs and

DA interfered with that of the essential pterin cofactor [18,25], preventing the cofactor from gaining access to the active site In contrast, the energetically favored positions of the non-catechol compounds NMTIQ and MPTP (Fig 3) indicated a placement corresponding to the binding site of the amino acid substrate in the crys-tal structure of phenylalanine hydroxylase [26] In these conformations, hydrogen bonds formed between the positively charged nitrogen atoms of NMTIQ and MPTP and the backbone oxygen of Ser324 The dis-tances between the nitrogen atoms and the oxygen of Ser324 were 2.01 A˚ for NMTIQ and 2.26 A˚ for MPTP, respectively Substantially greater distances (5.49 A˚ for NMTIQ and 4.92 A˚ for MPTP) hindered

an alternative formation of hydrogen bonds between the nitrogen atoms of the ligands and the backbone oxygen of Pro325 Although we did not directly com-pare the molecular interaction energies of NMTIQ and MPTP with those of the physiological substrate tyro-sine, we hypothesize that the binding affinity of both ligands will be much lower because NMTIQ and MPTP lack the carboxylate group of the amino acid, which is likely to interact electrostatically with Arg316 Therefore, competition between NMTIQ or MPTP and tyrosine for the common binding site seems improbable

6,7-Dihydroxylated TIQs oxidize the catalytic iron The divalent state of the iron atom in the center of the catalytic site is an essential requirement for TH activity [24] Catecholamine inhibitors trap the iron in a ferric state, leading to inactivation of the enzyme [27] Using low-temperature electron paramagnetic resonance (EPR) spectroscopy, we examined the oxidation status and spin of the iron in hTH in the presence of DA and four structurally distinct TIQs

The divalent iron of the unbound enzyme was EPR-silent Addition of DA produced a signal with g values

of 7.1 and 4.8 (Fig 4A) This signal originated from

Fig 3 6,7-Dihydroxylated TIQs and DA bind at identical sites in

the catalytic center of TH The ball and stick view shows the

ener-getically most favorable conformations of NMNorsal (red), NMSal

(yellow) and Norsal (green) in the crystal structure of the enzyme’s

catalytic domain (Protein Data Bank identification code 2toh) [18]

as determined by molecular docking; superposed is the binding

position of DA (dark blue) [25] In these bidentate conformations,

the immediate environment surrounding the active site iron (amber)

was a slightly distorted octahedral shape, formed by the two

cate-chol oxygens of the TIQs, a water molecule, and the TH residues

His331, His336 and Glu376 NMTIQ (light blue) and the exogenous

neurotoxin MPTP (orange) bound at a greater distance from the

catalytic iron The nitrogen atoms of TH residues are colored blue,

oxygen atoms are shown in red; hydrogen atoms are omitted for

clarity.

the ground state Kramers’ doublet of a trivalent

S= 5⁄ 2 spin system with a rhombicity parameter of

E⁄ D = 0.05, indicating that the iron had been

oxi-dized to Fe(III) The same characteristic signal was

detected in the presence of NMNorsal (Fig 4A),

NMSal and Norsal We determined the proportion of

oxidized iron after adding DA or these TIQs at a

con-centration equimolar to the hTH subunit concon-centration

(220 ± 7.5 lm) Equimolar DA concentrations led to

the generation of 80% high-spin Fe(III); NMNorsal

produced 64% high-spin Fe(III), NMSal 78% and

Norsal 76% (Fig 4B) Oxidation of the iron strongly

indicates that 6,7-dihydroxylated TIQs, like DA,

coor-dinate directly to the active site iron In contrast,

adding an equimolar concentration of the

non-hydrox-ylated NMTIQ caused only formation of nonspecific

high-spin ferric iron, which accounted for less than 4%

of the total hTH iron content (Fig 4B)

TH reactivation by protein kinase A

The primary mechanism of short-term TH regulation

is post-translational modification of the

catecholamine-bound enzyme by protein kinases, which

phosphory-late TH at serine residues of the N-terminal domain

[22,23,28] Phosphorylation by protein kinase A (PKA)

at Ser40, the most prominent of these regulatory sites,

increases the dissociation rate of bound catecholamine

inhibitors [29,30] Catecholamine removal facilitates

Fe(III) reduction by tetrahydropterin, leading to an

increase in Vmax of the enzyme reaction [31,32]

We compared the effects of PKA on TH activity

after inhibition with either NMNorsal or DA PKA

did not change the basal enzyme activity when hTH

was fully reconstituted with Fe(II) and concentrations

of the pterin cofactor DPH4 were saturating (Fig 4C)

However, tyrosine hydroxylation increased when PKA was added after hTH inhibition by 0.1 lm DA (Fig 4C) PKA likewise reactivated hTH after inhibi-tion by NMNorsal Incubainhibi-tion of the enzyme with 0.1 lm NMNorsal reduced its activity to approxi-mately 50% When PKA was added after the incuba-tion, hTH activity was fully restored (Fig 4C)

TIQs stabilize TH, albeit at a reduced level

of activity

DA and other catecholamines have a stabilizing effect

on the conformation of TH [33] We therefore com-pared the thermostability of hTH at 37C in the pres-ence of NMNorsal, Norsal and DA In the abspres-ence of these ligands, the specific activity of hTH decreased slowly but continuously After 20 min, the hTH activ-ity was 69 ± 4% compared to baseline; after 5 h, the remaining activity was reduced to 3% (Fig 5A) DA (20 lm) markedly accelerated the initial loss of hTH activity and decreased tyrosine hydroxylation to 23% within 10 min In contrast to the uninhibited enzyme, the activity remained steady at this level for 90 min (Fig 5A) Similar to DA, NMNorsal and Norsal (20 lm each) also provoked a fast initial decline of hTH activity but stabilized the activity at 59% and 48%, respectively, for 5 h Even after 20 h at 37C, the enzyme activity was not completely lost, with resid-ual activities of 28% and 23%, respectively, compared

to 2% after 20 h of incubation with DA Surprisingly,

a small transient increase in hTH activity occurred after 1 h incubation in the presence of NMNorsal and Norsal (Fig 5A)

DA is an unstable neurotransmitter (Fig 5B) Its autoxidation leads to the formation of DA quinone and is accompanied by the generation of hydrogen

Table 1 Molecular interaction energies of DA, TIQs and MPTP in the crystal structure of TH (Protein Data Bank identification code 2toh).

ND, not determined.

peroxide and reactive oxygen species (ROS) [34,35].

Accumulation of DA quinone and ROS may

contrib-ute to the rapid initial loss of hTH activity that we

observed during the incubation with DA [36]

Hydro-gen peroxide may also be Hydro-generated by partial

uncou-pling of the pterin oxidation from the tyrosine

hydroxylation [37] In the presence of iron, hydrogen

peroxide is converted to a hydroxyl radical and

hydroxide through the Fenton reaction [34,38] We

examined the possible involvement of ROS in hTH

inhibition by DA, NMNorsal and Norsal using

cata-lase (EC 1.11.1.6), which converts hydrogen peroxide

to oxygen and water Catalase (0.05 mgÆmL)1) slowed

the initial DA-induced decrease in hTH activity

with-out preventing the overall activity loss (Fig 5C)

Cata-lase did not alter the hTH inhibition by NMNorsal or Norsal (Fig 5C), nor did it have an effect on the decline of hTH activity in the absence of inhibitors (data not shown) We conclude that hydrogen peroxide

is formed and accelerates TH inhibition in the presence

of DA; in contrast, hydrogen peroxide appears not to

be involved in the TH inhibition by NMNorsal and Norsal, which are stable compounds compared to DA (Fig 5B)

Discussion

The catecholamines DA, norepinephrine and epineph-rine regulate TH activity through two types of inhibi-tion: reversible competition with the essential

Fig 4 Oxidation of the active site iron (A) Rapid freeze-quench EPR spectra of hTH in the presence of equimolar concentrations of

DA or NMNorsal exhibited a characteristic signal at g values of 7.1 and 4.8 (arrow-heads), indicating the formation of Fe(III) The signal at g = 4.3 stemmed from non-specific high-spin ferric iron; a Cu(II) impurity caused the signal at g = 2 The redox state

of the iron remained unchanged after addi-tion of NMTIQ (B) To determine the propor-tion of enzyme-bound iron converted to Fe(III), we compared the integrated absorp-tion spectra with a 1 m M Fe(III) cytochrome P450cam standard The assays contained

220 ± 7.5 l M hTH subunits fully reconsti-tuted with Fe(II) and equimolar concentra-tions of NMNorsal, NMSal and Norsal Nonspecific high-spin ferric iron formed in the presence of the non-hydroxylated NMTIQ accounted for less than 4% of the total iron **P < 0.01 compared to DA or any of the other TIQs in a one-way ANOVA

followed by Tukey’s test (C) Activity of hTH phosphorylated by PKA after inhibition with DA (0.1 l M ) or NMNorsal (0.1 l M ).

*P < 0.05 for the difference between hTH activities in the absence and presence of PKA (unpaired t test).

tetrahydropterin cofactor and an almost irreversible

blockade of TH activity by facilitating oxidation of the

catalytic iron [21,23] As catecholamine-bound TH is

thermally stable and resists proteolytic cleavage [33],

the enzyme becomes trapped in an inactive state

Using recombinant human TH, we show here that

endogenous TIQs associated with PD mimic the

mech-anisms of catecholamine feedback inhibition: TIQs

both compete with the tetrahydropterin for access to

the active site and form a tight bidentate ligation to

the iron atom in the center of the catalytic site, the

latter prompting oxidation of the iron and consequent

hTH inactivation TH inhibition by endogenous TIQs

depends critically on 6,7-dihydroxylation of the

ben-zene ring Only NMNorsal, NMSal and Norsal, which

possess an intact catechol moiety, are inhibitors of

hTH; hTH activity does not decrease in the presence

of the non-hydroxylated NMTIQ NMNorsal was the

strongest inhibitor among the 6,7-dihydroxylated TIQs

studied Its IC50 of 0.3 lm nearly equals that of DA, suggesting that even small intracellular concentrations

of NMNorsal are sufficient to produce a major effect

on neurotransmitter synthesis In comparison, up to

104-fold higher concentrations of TIQs are required to cause cytotoxic blockade of the mitochondrial respira-tory chain [4], indicating that 6,7-dihydroxylated TIQs primarily interfere with DA synthesis in PD rather than provoking neuronal degeneration Although the levels of 6,7-dihydroxylated TIQs in the substantia nigra and corpus striatum of patients with PD are unknown, a recent analysis indicated that the average concentration of NMSal in the substantia nigra, caudate nucleus and putamen of individuals without neurological or psychiatric disease is between 65 and

110 pmolÆg)1 [2] Salsolinol and NMNorsal are nor-mally not detected in the CSF, but elevated levels of these TIQs of up to 60 pmolÆmL)1 were found in patients with PD [3,12], and the concentration of

Fig 5 Thermostability of hTH increases after DA and TIQ chelation (A) Recombinant hTH was incubated with 20 l M DA, NMNorsal or Nor-sal for 20 h at 37 C Aliquots of the assay were removed at the indicated intervals to measure enzyme activity We carried out six indepen-dent measurements for the uninhibited enzyme and four for each inhibitor at every interval (B) Stability of DA, NMNorsal and Norsal during incubation at 37 C in Hepes buffer containing 1.5 l M Fe(II) sulfate (n = 4) (C) Activity of hTH incubated with DA, NMNorsal or Norsal (20 l M each) in the presence of catalase (50 lgÆmL)1) Catalase slowed the initial loss of hTH activity caused by DA but had no effect on hTH inhibition by NMNorsal or Norsal (n = 4).

NMSal in the CSF of patients in PD is twice as high

as in control individuals of a similar age [1] The

increased CSF levels probably reflect a rise in the

nigrostriatal concentration of 6,7-dihydroxylated TIQs

that is sufficient to provoke TH inhibition

Similar to the kinetics of TH inhibition by

catechol-amines [10,39], enzyme inhibition by NMNorsal is

competitive with respect to the tetrahydropterin

cofac-tor and noncompetitive with respect to tyrosine

Previ-ous molecular docking studies [25,40] and X-ray

crystallography [18] indicate that BH4 and analogue

pterins coordinate close to the active-site iron in TH,

forming an aromatic p-stacking interaction with

enzyme residue Phe300 In the modeled complex of the

enzyme’s catalytic domain, NMNorsal ligation

inter-fered with the docking site for BH4 By preventing the

pterin cofactor from binding, NMNorsal blocks

tyro-sine hydroxylation at a critical reaction step [41] The

natural pterin BH4 is considered to be the first

sub-strate to bind at the TH active site, followed by

mole-cular oxygen and tyrosine [42] Furthermore, electron

transfer from the BH4 carbonyl oxygen to the

mole-cular oxygen and the generation of a hydroxylating

intermediate are presumably rate-limiting for the

enzyme reaction [43,44]

Direct binding of NMNorsal, NMSal and Norsal to

the iron at the center of the catalytic site enabled

bid-entate ligation between the two hydroxyl residues of

their catechol moiety and the iron The unbound ends

of these molecules projected from the binding pocket

Potentially, they interact with the enzyme’s regulatory

domain, which has not been crystallized yet and was

not incorporated in our model The coordination mode

and actual binding site of the 6,7-dihydroxylated TIQs

are identical with those for the catecholamine feedback

inhibitor DA in a previous docking model [25] and an

X-ray absorption fine-structure study [27] In our

model, these TIQs had almost the same binding

affini-ties to the catalytic center as DA In contrast,

NMTIQ, which lacks a catechol moiety, occupied the

binding site of the amino-acid substrate tyrosine [26]

when docked into the catalytic center of TH The same

conformation was obtained with MPTP However,

nei-ther NMTIQ nor MPTP formed electrostatic

interac-tions with the surrounding TH residues, resulting in a

low binding affinity NMTIQ is therefore unlikely to

compete with tyrosine in vivo, which may explain why

NMTIQ does not have an inhibitory effect on TH

However, molecular docking in our model was limited

to the catalytic center of TH, and the lack of a strong

docking conformation here does not exclude the

existence of allosteric binding sites for NMTIQ or

MPTP

Tight ligation of DA and other catecholamine end products to the active-site iron of TH has two major consequences First, catecholamine binding increases the proportion of oxidized iron bound to the enzyme [10,27,39,45], leading to loss of TH activity [46,47] Second, thermal stability of the enzyme increases and its resistance to proteolysis improves [33] TH prepara-tions from animal tissues are inevitably contaminated with catecholamines and thus contain a sizable propor-tion of bound Fe(III) [45] In our study, we used recombinant hTH reconstituted with Fe(II), which allowed us to accurately quantify the formation of Fe(III) We found that equimolar concentrations of NMNorsal, NMSal or Norsal cause a rapid increase in Fe(III) In the presence of these TIQs, between 64 and 78% of the hTH iron was oxidized, compared to 80%

in the presence of DA The precise mechanism respon-sible for the oxidation of enzyme-bound Fe(II) is unclear Most likely, molecular oxygen is the actual oxidant [48] Formation of a TH–Fe(III)–catechol-amine complex induces an absorbance change at

700 nm that can be detected using visible spectroscopy Under anaerobic conditions, the absorbance change is only 50% of that observed in the presence of molecu-lar oxygen [27] An important effect of catechols appears to be a shift in the equilibrium of bound iron towards the ferric state and prevention of its reduction

to Fe(II) Consequently, catechols, which themselves are reducing agents, trap oxidized iron in the complex with the enzyme

To study the effect of DA and 6,7-dihydroxylated TIQs on TH stability, we incubated the enzyme with

DA, NMNorsal or Norsal at 37C for up to 20 h

In the absence of inhibitors, hTH activity gradually declined DA rapidly reduced hTH activity by 77%, but the residual activity remained stable for 90 min NMNorsal and Norsal also caused an initially rapid decrease in hTH activity; however, in the presence of these TIQs, residual activity levels of 59 and 48%, respectively, were sustained over several hours, exceed-ing the activity of the uninhibited enzyme DA is an unstable neurotransmitter and its metabolites are likely

to contribute to the more profound loss of hTH activ-ity during the incubation DA autoxidation produces

DA quinone, which inhibits TH through covalent modification of its cysteinyl residues [36,49] Both autoxidation and enzymatic DA metabolism lead to the generation of hydrogen peroxide In the iron-rich environment of the substantia nigra, hydrogen perox-ide is readily converted to ROS such as superoxperox-ide and hydroxyl radicals [34,38] Partial uncoupling of the hydroxylase reaction caused by ligands binding to the enzyme active site and changing its geometry may

provide another source of ROS [37] Catalase protects

against the formation of oxygen radicals by converting

hydrogen peroxide to oxygen and water Catalase

attenuated the loss of hTH activity during incubation

with DA, but had no effect on hTH activity in the

presence of NMNorsal or Norsal We conclude that

hydrogen peroxide formation accelerates TH inhibition

by DA, but is not involved in inhibition of the enzyme

by 6,7-dihydroxylated TIQs The similar residual

activ-ity levels of the enzyme after 20 h of incubation with

DA, NMNorsal or Norsal indicate that hydrogen

per-oxide is not required for the overall inhibitory effect of

catechols, including DA

Reactivation of catecholamine-bound TH is

medi-ated by phosphorylation at serine residues within the

N-terminal domain [22] Cyclic AMP-dependent

phos-phorylation at Ser40 by PKA does not directly

regu-late the reduction of Fe(III) [32], but strongly increases

the dissociation rate of catecholamines and decreases

the KMfor the pterin cofactor [29,30,50,51]

Phosphor-ylation at Ser40 also increases the affinity of TH for

14-3-3 proteins, which protect the enzyme from

dephosphorylation [51] Catecholamine removal allows

the pterin cofactor to regain access to the enzyme

active site and reduce the iron to its active ferrous

form [32,48] PKA also restored hTH activity after

inhibition by NMNorsal Because NMNorsal and DA

bind at identical sites in the catalytic center, we

hypothesize that hTH phosphorylation by PKA results

in a conformational change that facilitates the release

of NMNorsal in the same way as it promotes the

dissociation of catecholamine inhibitors However, the

precise conformational changes that are induced by

the phosphorylation of TH are unknown

Phosphory-lation at Ser40 may provide a negative charge that

interacts with the amino group of DA and the pyridine

moiety of NMNorsal in opposition to the bidentate

ligation of the catalytic iron, pulling the ligands away

from the iron and allowing them to leave the catalytic

center; alternatively, phosphorylation-induced

confor-mational changes may mimic protonation of an

N-ter-minal TH residue that interacts with positively charged

ligand groups, reducing their binding affinity [29,52]

The increase in endogenous TIQs is most prominent

in early PD [12,53], when approximately two-thirds of

the dopaminergic neurons in the substantia nigra have

been lost [54] Supported by other, non-dopaminergic

compensatory mechanisms [55], the remaining neurons

need to increase DA synthesis and release in order to

balance the shortfall caused by their degenerating

counterparts [14,15] We propose that blockade of

cat-echolamine synthesis by NMNorsal and related

endog-enous TIQs enhances DA depletion Dopaminergic

neurons of the substantia nigra are likely to be primar-ily affected, because endogenously formed 6,7-dihydr-oxylated TIQs accumulate in this region of the human midbrain [2] TH inhibition occurred at TIQ concen-trations substantially lower than those required for blockade of mitochondrial respiration and induction of neuronal cell death [4–6] Even though the remarkable stabilizing effect of TIQs on purified TH may translate into preservation of a reduced enzyme activity in vivo, the predominant effect of NMNorsal and other 6,7-dihydroxylated TIQs present in PD is likely to be a decrease in DA synthesis

Experimental procedures

Chemicals

Chemical compounds were purchased from Sigma-Aldrich (Munich, Germany) unless otherwise indicated NMNorsal was synthesized by demethylation of 2-methyl-6,7-dimeth-oxy-1,2,3,4-tetrahydroisoquinoline using 47% hydrogen bromide [56] The purity of the product was > 98% as determined by NMR spectroscopy

Recombinant human TH isozymes

Complementary DNAs of the coding sequences for hTH1 and hTH4 were inserted into a pET vector and transcribed

in a BL21(DE3) strain of Escherichia coli engineered to con-tain an isopropyl b-d-thiogalactopyranoside (IPTG)-induc-ible T7 RNA polymerase gene [47] After incubation in the presence of 0.4 mm IPTG for 2 h at 37C, the bacteria were harvested and stored at)20 C until use We lysed the bac-teria using a French press (Thermo Scientific, Waltham,

MA, USA), and, after centrifugation at 35 000 g for 1 h, purified the hTH isoforms from the supernatant as previ-ously described [47], using a combination of diethylamino-ethyl (DEAE)–Sepharose anion exchange chromatography, heparin–Sepharose affinity chromatography and size-exclu-sion chromatography on a Sephacryl S-300 gel column (GE Healthcare, Uppsala, Sweden) We verified by N-terminal amino acid sequence analysis that the isoforms were pure and had the predicted sequences [19] except for the N-termi-nal methionine residue, which was missing in 96% of hTH1 and 90% of hTH4 samples We concentrated the purified enzymes and stored them in liquid nitrogen The hTH iso-forms typically contained less than 0.1 iron atoms per sub-unit, had a high catalytic activity when reconstituted with Fe(II), and were stable at neutral pH [47]

TH activity assay

We reconstituted recombinant hTH1 and hTH4 (subunit concentration 0.1 lm) using 0.1 mm Fe(II) sulfate before

we preincubated hTH with DA or TIQs for 15 min in

100 mm Hepes buffer (pH 7.0) containing 1 mgÆmL)1

bovine catalase The enzyme reaction was started by

addi-tion of 0.1 mm l-tyrosine (Merck, Darmstadt, Germany),

0.1 mm 6,7-dimethyl-5,6,7,8-tetrahydropterin (DPH4),

0.1 UÆmL)1 dihydropteridine reductase and 0.1 mm

NADH After incubation for 2 min at 30C under aerobic

conditions, we terminated the reaction by adding 1.1%

per-chloric acid For hTH activation by PKA, we reconstituted

the protein samples with Fe(II) as described above and

added 0.2 mgÆmL)1 bovine PKA, 0.4 mm MgCl2 and

0.1 mm ATP 5 min before starting the reaction [57] The

concentration of the reaction product l-DOPA was

mea-sured by HPLC with electrochemical detection We used

2-methyl-3-(3,4-dihydroxyphenyl)-dl-alanine (50 nm) as an

internal chromatography standard HPLC was performed

at 30C using a C18 column (Eurospher RP18, particle

size 5 lm, column size 250· 4.0 mm; Knauer, Berlin,

Germany) and pre-column (35· 4.0 mm; Knauer) The

mobile phase consisted of a degassed solution containing

0.3 mm Na2-EDTA, 0.52 mm 1-Na-octane sulfate, 11.5%

methanol and 0.1 m citrate buffer, pH 3.0 The detector cell

operated at 0.8 V Nonenzymatic l-DOPA formation was

determined using 0.1 mm d-tyrosine as substrate in the

presence of the TH inhibitor a-methyl-l-para-tyrosine

(0.1 mm) Enzymatic synthesis of l-DOPA was determined

by subtracting the concentration of nonenzymatically

formed l-DOPA from the total concentration [9,10]

Molecular docking

Ligand–protein complexes were based on the crystal

struc-ture of the enzyme’s catalytic domain (Protein Data Bank

identification code 2toh) [18] after removal of

co-crystal-lized 7,8-dihydrobiopterin and all water molecules except

for HOH601, which completes the iron coordination sphere

as a counterpart of Glu376 Residue 300 of TH was

reverted to phenylalanine [58] We employed the software

corina (version F; Molecular Networks, Erlangen,

Germany) to generate 3D structures of the ligands, flexx

(version 2.0.2; BioSolveIT, Sankt Augustin, Germany) for

the ligand docking, and amber (version 8; Department of

Pharmaceutical Chemistry, University of California, San

Francisco, CA, USA) to optimize complexes by force-field

energy minimization [25] General amber force-field atom

types and Gasteiger atomic charges were assigned to the

ligand atoms [59] We limited the output to a maximum of

200 placements per ligand; the (R) and (S) enantiomers of

NMNorsal and NMSal were treated separately

The docking runs included all active site atoms within a

radius of 12.0 A˚ around the catalytic iron The maximum

distance between the hydroxyl oxygen atoms of the ligands

and the active site iron was set at 5.0 A˚, allowing docking

modes that included monodentate and bidentate binding

[25] Ligand–protein complexes were subjected to 50 steps

of steepest-descent energy minimization, followed by 350 steps of conjugated-gradient energy minimization, applying

a distance-dependent dielectric constant of 2 r In order to focus on the most plausible placements, the resulting con-formations were clustered based on their rmsd values and force-field energies Starting from the energetically most favorable conformation as the reference placement, all con-formations with an rmsd of less than 1.4 A˚ with respect to the reference placement were considered identical and excluded from further analyses This continued with the next best conformation of the remaining placements until

no further placements were left We ranked alternative ligand placements according to the energy score of each conformation, and determined interaction energies for the

20 energetically most favorable conformations To account for aqueous solvation effects, we assessed electrostatic actions using the generalized Born method The linear inter-action energy with continuum electrostatics (LIECE) [60] was calculated as the sum of unweighted differences of van der Waals energies, electrostatic energies, electrostatic and nonpolar solvation energies, and an entropically reasoned penalty of 1.4 kJÆmol)1per rotatable bond in order to esti-mate the relative binding affinities For iron, the surface parameters for the generalized Born calculations were una-vailable and had to be estimated

Electron paramagnetic resonance spectroscopy

We reconstituted recombinant hTH samples with Fe(II) and incubated them with equimolar concentrations of DA or TIQs for 2 min under aerobic conditions at room tempera-ture We recorded rapid freeze-quench EPR spectra at a tem-perature of 10 K and a microwave frequency of 9.6456 GHz using a conventional X-Band spectrometer (Bruker 200D SRC, Karlsruhe, Germany) equipped with a helium-flow cryostat (ESR 910, Oxford Instruments, Witney, UK) [27] The microwave power was 80 mW The modulation ampli-tude was 0.5 mT and the modulation frequency was

100 kHz Spin quantifications were performed by integration

of the experimental absorption spectra and comparison with

a 1 mm Fe(III) cytochrome P450cam (camphor 5-monooxy-genase) standard from Pseudomonas putida The integrated areas were weighted using Aasa correction factors [27]

TH thermostability

Recombinant hTH (2 lm) was incubated with 20 lm DA, NMNorsal or Norsal at 37C in the presence of 1.5 lm Fe(II) sulfate Catalase was included in the assay as indi-cated At defined intervals, we removed aliquots of 5 lL to measure TH activity We incubated the aliquots for 2 min

at 30C in a reaction mixture containing 50 mm Hepes buffer (pH 7), 0.1 mm Fe(II) sulfate, 25 lm3H-tyrosine and

50 lgÆmL)1 catalase The enzyme reaction was started by the addition of 0.5 mm BH4 in 5 mm dithiothreitol, and