Báo cáo khoa học: Mechanisms of cholinesterase inhibition by inorganic mercury potx

When the free sulfhydryl group is not sensitive to mercury Drosophila melanogaster acetylcholinesterase and human butyrylcholinesterase or is otherwise absent Electrophorus electri-cus a

Manuela F Frasco1,2, Jacques-Philippe Colletier3, Martin Weik3, Fe´lix Carvalho4,

Lu´cia Guilhermino1,2, Jure Stojan5and Didier Fournier6

1 ICBAS, Instituto de Cieˆncias Biome´dicas de Abel Salazar, Universidade do Porto, Portugal

2 CIMAR-LA ⁄ CIIMAR, Universidade do Porto, Portugal

3 IBS-UMR 5075, CEA-CNRS-UJF, Laboratoire de Biophysique Mole´culaire, Grenoble, France

4 REQUIMTE, Servic¸o de Toxicologia da Faculdade de Farma´cia da Universidade do Porto, Portugal

5 Institute of Biochemistry, University of Ljubljana, Slovenia

6 IPBS-UMR 5089, CNRS-UPS, Groupe de Biotechnologie des Prote´ines, Toulouse, France

Human activities have continuously contaminated the

environment with mercury, which has been used for

centuries in agriculture, industry, and medicine [1]

Nowadays, inorganic mercury is used in, for example,

thermometers, batteries, and fluorescent light-bulbs In

addition, large quantities of metallic mercury are

employed in the fabrication of electrodes for the

elec-trolytic production of chlorine and sodium hydroxide

from salt, as well as in gold mining [2,3] Although it presents unique properties that make it useful for many human purposes, mercury has no role in life processes and is highly toxic Nephrotoxicity [4] and genotoxicity [5] have been demonstrated Other adverse effects occur in neural tissues, where the targeting of enzymes and receptors involved in nerve impulse trans-mission is probably involved [6], as well as in the

Keywords

aggregation; cholinesterase; inhibition;

mercury; metals

Correspondence

D Fournier, IPBS-UMR 5089, 205 Route de

Narbonne, F-31077 Toulouse, France

Fax: +33 5 61 17 59 94

Tel: +33 5 61 55 54 45

E-mail: fournier@ipbs.fr

Database

The coordinates and structure factor

ampli-tudes of the complex structure of human

butyrylcholinesterase with HgCl2have been

deposited in the Protein Data Bank under

accession code 2J4C

(Received 18 December 2006, revised 1

February 2007, accepted 7 February 2007)

doi:10.1111/j.1742-4658.2007.05732.x

The poorly known mechanism of inhibition of cholinesterases by inorganic mercury (HgCl2) has been studied with a view to using these enzymes as biomarkers or as biological components of biosensors to survey polluted areas The inhibition of a variety of cholinesterases by HgCl2 was investi-gated by kinetic studies, X-ray crystallography, and dynamic light scatter-ing Our results show that when a free sensitive sulfhydryl group is present

in the enzyme, as in Torpedo californica acetylcholinesterase, inhibition is irreversible and follows pseudo-first-order kinetics that are completed within 1 h in the micromolar range When the free sulfhydryl group is not sensitive to mercury (Drosophila melanogaster acetylcholinesterase and human butyrylcholinesterase) or is otherwise absent (Electrophorus electri-cus acetylcholinesterase), then inhibition occurs in the millimolar range Inhibition follows a slow binding model, with successive binding of two mercury ions to the enzyme surface Binding of mercury ions has several consequences: reversible inhibition, enzyme denaturation, and protein aggregation, protecting the enzyme from denaturation Mercury-induced inactivation of cholinesterases is thus a rather complex process Our results indicate that among the various cholinesterases that we have studied, only Torpedo californica acetylcholinesterase is suitable for mercury detection using biosensors, and that a careful study of cholinesterase inhibition in a species is a prerequisite before using it as a biomarker to survey mercury in the environment

Abbreviations

DLS, dynamic light scattering.

immune system, for which both autoimmunity and

immune suppression have been reported [7–9]

Detec-tion of mercury in the environment is thus of high

rele-vance for public health and in the framework of

sustainable development In this context,

cholinesteras-es have been suggcholinesteras-ested as potential biomarkers and as

the biological component of protein-based biosensors

The concept of biomarkers implies that in vivo

choli-nesterase inhibition is measured after exposure of an

animal to mercury, whereas in biosensors, inhibition

takes place in vitro For both applications, there is a

need for high sensitivity of cholinesterases to mercury

Thus, it is a prerequisite to investigate the

mechan-ism(s) of inhibition and to ascertain the reliability of

using cholinesterases

Cholinesterases are believed to be sensitive to

mer-cury; indeed, exposure of different organisms to

suble-thal concentrations of mercury was shown to induce a

significant decrease in cholinesterase activities in

sev-eral organs [10–15] However, uncertainties remain, as

mercury-induced stimulation of cholinesterase activity

has also been reported [16,17] Several authors have

described instantaneous reversible inhibition of

choli-nesterases in vitro [18–20] However, these findings

could be artefactual, as activity measurements were

often performed using Ellman’s reaction [21], the

prod-ucts of which react with mercury and thus interfere

with the measurement [22] In addition, irreversible

inhibition was described for Torpedo californica

acetyl-cholinesterase, which leads in a first step to the

forma-tion of a metastable state [23] that latter converts to a

partially unfolded one [24]

Depending on their locations, two main functions

have been ascribed to cholinesterases At cholinergic

synapses, cholinesterases are responsible for the

ter-mination of nerve impulse transmission, by rapid

hydrolysis of the neurotransmitter acetylcholine This

role is vital, as it allows restoration of neuronal

excita-bility in cholinergic neuron networks In

noncholiner-gic tissues, cholinesterases belong to the group of

‘scavenger proteins’, which are responsible for the

deg-radation of xenobiotics, e.g succinylcholine or cocaine

[25] Playing these important roles, cholinesterases are

among the most efficient enzymes in nature, with a

substrate turnover of 103)104s)1, depending on

spe-cies [26] Two types of cholinesterase are found in

mammals, acetylcholinesterase and

butyrylcholinest-erase, which are enzymatically distinguished by their

substrate specificity From the structural point of view,

these enzymes are very similar, and only a few critical

differences in the active site amino acid composition

account for their differential behavior towards

sub-strates [27–31]

Cholinesterases are  60 kDa globular proteins, and are found in various oligomeric states With regard to their cysteine content, three intrachain disulfide bonds are conserved, as well as another cysteine involved in intersubunit association Although no free cysteine

is found in some species, e.g Electrophorus electricus acetylcholinesterase, there is one accessible to the bulk solution in most of them Its position, however, is not conserved: 66 in human butyrylcholinesterase, 290 in Drosophila melanogaster acetylcholinesterase, and 231

in T californica acetylcholinesterase In the last case, the free cysteine has been shown to react with sulfhyd-ryl agents, resulting in irreversible inactivation of the enzyme [32,33]

In the present study, the kinetic mechanism of mer-cury-induced inactivation of cholinesterases was inves-tigated using four cholinesterases from various species

to probe the potential variability in sensitivity that may exist in biomarkers used in ecotoxicologic studies

T californica acetylcholinesterase, E electricus acetyl-cholinesterase, D melanogaster acetylcholinesterase and human butyrylcholinesterase were chosen because they are available in large amounts Kinetic studies were complemented by X-ray crystallographic experi-ments on human butyrylcholinesterase and dynamic light scattering (DLS) studies on D melanogaster acetylcholinesterase Two inhibition mechanisms are proposed, depending on the presence or absence of a sensitive free cysteine

Results

Inhibition of T californica acetylcholinesterase Figure 1 shows the kinetics of irreversible inactivation

of T californica acetylcholinesterase by 1–10 lm HgCl2 Inhibition follows a pseudo-first-order kinetics (Scheme 1A; ki¼ 9200 ± 480 m)1Æmin)1), suggesting that inactivation involves only one site Probably, this site is the same that has been shown to react with other thiols and organomercurial compounds [24,33], i.e Cys231

Inhibition of human butyrylcholinesterase Incubation of 15 nm enzyme with HgCl2 (1–10 mm HgCl2) leads to rapid inhibition until a plateau is reached (Fig 2A) Increasing the enzyme concentration diminishes the maximum inhibition (Fig 2B) The inhibition appears to be slowly reversible; indeed, the 10-fold dilution of a sample incubated with 10 mm HgCl2 leads to a slow increase of activity until a plat-eau is reached corresponding to the activity observed

after incubating the enzyme with 1 mm HgCl2(see

sup-plementary Fig S1A) To investigate this slow

reacti-vation of the enzyme, human butyrylcholinesterase at

a concentration of 15 nm was incubated with 1 or

10 mm HgCl2 for 30 min Samples were then dialyzed

for 5 h (with a dilution factor of 1000), and activity

was recorded as a function of time A time-dependent

reactivation of the enzyme was observed, suggesting

that inhibition is reversible (Fig 2C) Equilibrium is

reached after 15–20 min (Fig 2A), so the analysis of

equilibrium between human butyrylcholinesterase and

mercury was performed by incubating 15 nm enzyme

for 30 min at different HgCl2 concentrations

Subse-quently, the substrate o-nitrophenyl acetate was added,

and the activity was measured Data are best fitted by

a model accounting for noncompetitive inhibition, with

an apparent Kiapp¼ 0.4 ± 0.06 mm (Fig 2D) Inhibi-tion of human butyrylcholinesterase by mercury thus occurs at a concentration 1000 times higher than that inhibiting T californica acetylcholinesterase Addition-ally, human butyrylcholinesterase inhibition by HgCl2 can be described as an apparent, slow, noncompetitive reversible inhibition, which depends on the enzyme concentration, whereas the data for T californica acetylcholinesterase can only be described as irrever-sible inhibition

Inhibition of D melanogaster acetylcholinest-erase and E electricus acetylcholinestacetylcholinest-erase

D melanogaster acetylcholinesterase and E electricus acetylcholinesterase are inhibited by mercury in the same range of concentrations as human butyrylcholi-nesterase (i.e 1000-fold higher than the concentration necessary to inhibit the Torpedo enzyme) In contrast

to what was observed for human butyrylcholinesterase, inhibition of these two enzymes by HgCl2 did not lead

to a plateau, but rather showed a double exponential decay, as shown in Fig 3A for D melanogaster acetyl-cholinesterase (see supplementary Fig S2 for E electri-cus acetylcholinesterase) In addition, reactivation upon dilution is partial, restoring less than 10% of the initial activity (see supplementary Fig S1B,C) As for human butyrylcholinesterase, however, inhibition decreases with enzyme concentration (Fig 3B) Hence, inhibition of D melanogaster acetylcholinesterase and

E electricus acetylcholinesterase by mercury can be described as a slow, noncompetitive, reversible process that depends on the enzyme concentration However,

an irreversible inhibition also takes place, which was not evident for human butyrylcholinesterase within 1 h

of incubation with mercury

The D melanogaster acetylcholinesterase used herein

is recombinant; thus, it was possible to introduce sequence modifications by site-directed mutagenesis

To check whether this inactivation pattern was due to the free cysteine (C290) present on the surface of

D melanogaster acetylcholinesterase, this residue was mutated into an alanine The resulting inhibition pat-tern was virtually identical to that of the wild-type enzyme, suggesting that this cysteine residue is not involved in the inhibition mechanism Analogously, the alanine residue (A269) equivalent to the free cys-teine in position 231 of T californica acetylcholinest-erase was mutated into a cysteine In the micromolar range of HgCl2, this replacement also did not change the inhibition pattern of D melanogaster acetylcholi-nesterase, strongly suggesting the involvement of the residues surrounding C231 in T californica

acetylcholi-B

A

ki

M

k3

K0

k2 k1

M

M

D

M

M

E 2 M 2 E

E

k6 k7

k4

k8

D

Scheme 1 (A) Scheme proposed to describe the inhibition of

T californica acetylcholinesterase E and M represent enzyme and

mercury molecules, respectively, and the form EM is inactive (B)

Scheme proposed to describe the inhibition of D melanogaster

acetylcholinesterase All forms are as active as the native enzyme,

except for EM 2 and D, which are inactive upon mercury removal.

Fig 1 Remaining activity of T californica acetylcholinesterase

(7 n M ) following incubation with mercury (1–10 l M ) Curves

corres-pond to a single multi-nonlinear fit of data for all concentrations of

mercury in the equation derived from Scheme 1A [55].

nesterase in its high sensitivity to mercurial agents To

check whether the introduction of a free cysteine inside

the active site of D melanogaster acetylcholinesterase

would change the inhibition pattern, mutations F330C and Y370C were analyzed These replacements did not change the inhibition pattern

Fig 3 Remaining activity of D melanogaster acetylcholinesterase following incubation with mercury (A) Effect of mercury concentration with 500 n M enzyme (B) Effect of protein concentration with 5 m M mercury.

Fig 2 Remaining activity of human butyrylcholinesterase following incubation with mercury (A) Effect of mercury concentration with 15 n M enzyme (B) Effect of protein concentration with 2.5 m M mercury (C) Slow reactivation of inhibited enzyme following dialysis after 30 min of incubation with mercury (D) Steady-state analysis of inhibition: enzyme and mercury were incubated for 30 min, after which the substrate o-nitrophenyl acetate was added to the cuvette at different concentrations, without significant dilution of the sample.

Structure of the HgCl2–human

butyrylcholinesterase complex

In this complex, four mercury-binding sites were

char-acterized (Fig 4A) These were attributed on the

basis of very clear anomalous signals of mercury ions

(Fig 4B–E) at the employed wavelength (i.e 1.54 A˚)

An isomorphous difference map, computed using the

structure factors of the complex and those obtained

from a native crystal (data not shown), confirmed

these positions, displaying four very strong positive

peaks overlapping with the anomalous peaks In

addi-tion, a pair of positive and negative densities was

found in the active site, next to the catalytic serine

residue (Ser198), on the atypical bond with the bound

butyrate [31] This feature was interpreted as a

dis-placement of the butyrate upon complexation with

mercury No mercury ion was found in the active

site

The first mercury-binding site (occupancy: 75%) was

localized behind the ammonium-binding loci of the

active site (Fig 4B) At this site, mercury mainly

inter-acts with His77Ne2 and Met81Sd (distances: 2.75 and

3.6 A˚, respectively), as well as with surrounding water

molecules (distances: 2.96, 3.09, and 3.81 A˚,

respect-ively) The second mercury ion (occupancy: 50%)

binds to His423Nd1, Asn504Od1, and Thr505Oc1

(dis-tances: 2.33, 2.95, and 3.09 A˚, respectively; Fig 4C)

The third mercury ion (occupancy: 50%) is in close

proximity to, and undergoes electrostatic interaction

with, a sulfate anion (Fig 4D) from the mother liquor

solution (distance to the closest oxygen atom: 2.3 A˚)

The sulfate ion also interacts with His372Nd1 (distance

to the closest oxygen atom: 2.5 A˚), as has already been

reported for the native structure At these three

previ-ously described loci, mercury binding occurs on the

surface of the enzyme but does not involve crystal

con-tacts At the last binding site (occupancy: 25%),

how-ever, a mercury ion was found at a special position in

the crystal, in close proximity to the two Met511Sd

(distance: 2.6 A˚) residues of two symmetry-related

molecules in the crystal (Fig 4E) Hence, it is involved

in crystal packing interactions

The structure did not show any mercury ion bound

to a sulfhydryl group, as was observed in the case of

T californica acetylcholinesterase However, the only

free and accessible cysteine residue, Cys66, was

per-sulfured (Cys-S-SH) in this batch of enzyme Soaking

of human butyrylcholinesterase crystals with mother

liquor containing EDTA, dithiothreitol or l-cysteine

did not allow reduction of the per-sulfur Therefore,

the potential binding of mercury to this cysteine ‘in

solution’ remains an open issue

DLS assays

At all mercury concentrations, data were fitted as only one species with a low polydispersity (polydispersity index¼ 0.3) An increase in the hydrodynamic radius

of D melanogaster acetylcholinesterase was observed with increasing HgCl2 concentrations (Fig 5) Under the experimental conditions used (5 lm enzyme), the hydrodynamic radius increased linearly with mercury concentration Physical changes and protein aggrega-tion occurred in the first minute after mercury addi-tion, and the size of the aggregate remained stable for

at least 1 h

Kinetic model for D melanogaster acetylcholinesterase inhibition by mercury

D melanogaster acetylcholinesterase was incubated with HgCl2 for various times, and remaining activities were measured for 10 s following dilution (10-fold or 100-fold) of the sample The incubation time varied from 30 s to 1 h, enzyme concentrations were 50, 100,

300, 500, 700 and 900 nm, and HgCl2 concentrations were 1, 2.5, 5 and 10 mm The 22 experimental curves were simultaneously analyzed with concurrent models, taking into account the information obtained from other experiments: (a) inhibition appears to be noncom-petitive, binding sites of mercury are located on the pro-tein surface, and inhibition does not involve residues in the active site; and (b) mercury binding promotes aggre-gation, and hence indirectly diminishes the enzyme sen-sitivity to mercury, most likely because aggregation reduces accessibility to the second mercury-binding site Among all tested possibilities, Scheme 1B appears as the most simple and appropriate model to describe the irreversible and slow reversible inhibition of D melano-gaster acetylcholinesterase by mercury (see kinetic constants in Table 1 and curve fitting in supplementary Fig S3) According to this model, one mercury ion binds to the enzyme to form the complex EM (E and

M represent enzyme and mercury molecules, respect-ively) with an equilibrium constant around 0.2 mm; this binding is instantaneously reversible and does not affect enzyme activity The binding of a second mer-cury ion to the same enzyme molecule (EM) leads to

an inactive form (EM2) This inactivation is slowly reversible In addition, this form is not stable and may result in irreversible enzyme denaturation (D) This part of the scheme describes the two phases of inhibi-tion To describe the effect of protection by enzyme concentration, we introduced into the model the form

E2M2, resulting from reversible aggregation of the form EM without any alteration of the enzymatic

B C

E D

Active-site gorge entrance

A

1

2

3

4

activity This aggregate form (E2M2) may either

dena-turate (form D) or dissociate, thereby giving the

reversible inactivated form (EM2)

Discussion

Mercury inhibits cholinesterases

The four cholinesterases analyzed in this study are

inhibited by mercury, but through different

mecha-nisms Mercury in micromolar concentrations inhibits

T californica acetylcholinesterase, but the other tested cholinesterases are sensitive only in the millimolar range Millimolar concentrations of mercury are irre-levant both under physiologic conditions and in the environment; indeed, concentrations up to 300 mgÆL)1 ( 1 mm) have never been reported, even after bio-accumulation in highly polluted areas

The initial objective of this study was to evaluate the inhibition of cholinesterases by mercury, with a view to using them as biomarkers to survey polluted areas or for incorporation in biosensors With regard

to the utilization of cholinesterases as biomarkers, our work obviously demonstrates that the type and effect-iveness of inhibition of a cholinesterase by mercury strongly depend on the species Therefore, the kinetic characterization of cholinesterase inhibition in the selected species would be a prerequisite for field stud-ies A biosensor is an alternative to a biomarker, in that the enzyme is linked to a surface, deep in the sur-veyed solution, and inhibition of cholinesterase is recorded Inhibition occurs in vitro, whereas it occurs

in vivo in biomarkers Cholinesterases as biological components were first developed to detect low levels of insecticides in the environment [34] Numerous subse-quent studies have been performed to develop trans-ducers and to increase enzyme sensitivity and stability [35]; the biosensor technology for cholinesterases is therefore available, and permits consideration of their use for surveying mercury in the environment Of the studied enzymes, it appears that only T californica acetylcholinesterase could be considered a good candi-date, as it is the only one that is sensitive enough

Cholinesterase inhibition is not related

to the active site Mercury inhibits a large number of enzymes with func-tional sulfhydryl group(s) in the active site [36–38] This does not apply to cholinesterases, as: (a) there is no free cysteine in the active site; (b) the introduction of a free cysteine inside the active site of recombinant D mel-anogaster acetylcholinesterase (F330C and Y370C) did not increase the sensitivity to HgCl2; and (c) the com-plex structures of T californica acetylcholinesterase

Fig 5 Aggregation of D melanogaster acetylcholinesterase (5 l M )

as a function of mercury concentration revealed by an increase in

the hydrodynamic radius estimated by DLS.

Fig 4 Binding of mercury ions in the HgCl2–human butyrylcholinesterase complex (A) Overview of mercury-binding sites on the surface of the enzyme (B) First mercury-binding site (numbered ‘1’), next to His77 and Met81 (i.e on the W-loop, behind the ammonium-binding loci

of the active site ) Trp82) (C) Second mercury-binding site (numbered ‘2’), next to His423, Asn504 and Thr505 (D) Third mercury-binding site (numbered ‘3¢), next to a sulfate ion bound to His372 (E) Fourth mercury-binding site (numbered ‘4’), at a special position in the crystal,

in proximity to Met511 of two symmetry-related molecules The omit 2F o ) F c electron density map (contour level 1.5r), as well as the anomalous map (contour level 4r), are superimposed on the model in (B), (C), (D), and (E).

Table 1 Kinetic constants describing D melanogaster

acetylcholi-nesterase inhibition by mercury according to Scheme 1B Binding

of the first mercury ion is treated as an instantaneous step (affinity

is 0.2 m M ) The fit was done simultaneously to all inactivation data

(supplementary Fig S3) and to the curve of reactivation data

(sup-plementary Fig S1B) To emphasize the reactivation data, they

were weighted to the same y as the inhibition curves.

a

k 6 was set identical to k 4 , and k 7 to k 5 , as k 7 ¼ k 2 · k 5 ⁄ (k 2 + k 5 )

and k2 k 5

([24]) Protein Data Bank accession code 2J4F) and

human butyrylcholinesterase (this work) Protein Data

Bank accession code 2J4C) soaked in HgCl2 failed to

show mercury binding in their active site, at least with

a dissociation constant lower than 10 mm This

obser-vation is surprising, considering the strong electrostatic

dipole aligned with the gorge axis, which should attract

positive charges within the active site [39] The absence

of mercury within the active site may be a consequence

of its large solvation shell (I Silman, unpublished

results), which could prevent it from entering the gorge

The quaternary nitrogen of the substrate, on the other

hand, is not hydrated [40], and can thus readily enter

the gorge The model we propose here for

cholinest-erase inhibition by mercury is a rather general model

for the effect of mercury on proteins; only in cases

where a sensitive free sulfhydryl group is available

should another model be considered, as, for example,

in T californica acetylcholinesterase

Mercury-binding sites on cholinesterases

Organic and inorganic mercurials are capable of

forming very tight bonds with functional groups such

as thiolates of cysteine [41,42] Sulfhydryl groups are

considered to be the main targets of mercury, as they

are the most reactive nucleophilic sites of protein

amino acid side chains HgCl2 binds to a single

resi-due to form R-S-Hg-Cl Results for cholinesterase

suggest that inactivation by mercury involves the free

thiol group of Cys231 This functional group was

shown to be the target of other sulfhydryl reagents,

including organomercurials Modifications of this

resi-due lead to an irreversible inhibition of the enzyme

that follows pseudo-first-order kinetics [24,33]

How-ever, introducing a cysteine at the same position in

D melanogaster acetylcholinesterase did not result in

increased sensitivity to mercury, suggesting an

import-ant role of the surrounding environment in the

Torpedoenzyme

Mercury may also react with S–S bonds, leading

to their disruption (R-S-Cl + Cl-Hg-S-R) [43] The

R-S-Cl moiety may later be oxidized by another

mer-cury ion to form the compound R-S-Hg-Cl The

abil-ity of HgCl2 to cleave S–S bridges enables it to

disturb the tertiary structure of proteins and hence to

lower their stability In cholinesterases, three

intra-chain disulfide bonds are conserved Both site-directed

mutagenesis of the cysteines involved in disulfide

bond formation and cholinesterase treatment with

reducing agents inactivate the enzyme [44], showing

that these disulfides are essential for the protein to

function However, neither the complex structure with

mercury of human butyrylcholinesterase, nor that

of T californica acetylcholinesterase [24], showed evidence of binding at these positions Most likely, S–S bridges are too deeply buried inside the protein and thus are not accessible to the highly hydrophilic mercury

Metals are also capable of forming very tight bonds with histidine and methionine side chains as, for exam-ple, in metalloenzymes In cholinesterases, mercurials were found to be linked to these residues, as evidenced

by the complex structures with mercury of human butyrylcholinesterase (Fig 4; Protein Data Bank entry 2J4C) and T californica acetylcholinesterase (Protein Data Bank entry 2J4F)

Mercury induces protein aggregation Aggregation induced by metal ions has been observed for other protein systems, particularly for proteins involved in protein deposition diseases [45–48] DLS experiments have shown that binding of mercury ions promotes the aggregation of D melanogaster acetyl-cholinesterase, perhaps as a consequence of the cross-linking of two coordinating residues present at the surface of the protein This aggregation depends on protein concentration, and does not affect the folding

of the protein; it may therefore protect the enzyme from unfolding, due to the binding of two mercury ions on the same enzyme molecule

Putative mechanism of inhibition

A goal of this study was to address the issue of a clearcut model for cholinesterase inhibition by mer-cury It appears that two different mechanisms for cholinesterase inhibition by mercury may be consid-ered The first one, illustrated by the Torpedo enzyme (Scheme 1A), results from the binding of a mercury ion to a sensitive free cysteine and leads to irreversible inactivation Similar mechanisms have been described for several proteins, e.g the Na+–K+)2Cl– cotrans-porter, cystic fibrosis transmembrane conductance reg-ulator or urease [49–51]

The three other cholinesterases studied herein illus-trate the second mechanism of mercury-induced inhibi-tion It would probably also be operative for

T californica acetylcholinesterase if the sensitive Cys231, which causes irreversible inactivation, was absent (Scheme 1B) Binding of the first mercury ion is instantaneous, with an equilibrium constant around 0.2 mm, and does not affect enzyme activity The bind-ing of a second mercury ion, however, induces slow, reversible inactivation of the enzyme (EM2), which can

promote irreversible unfolding (D) It may be proposed

that mercury cross-links residues within the same

enzyme molecule, thereby inducing a conformational

change, which can lead to partial unfolding followed by

irreversible denaturation Mercury may also cross-link

residues belonging to different molecules, leading to

enzyme aggregation As the model is operative if the

resulting complex (E2M2) is fully active, one can

assume that these intermolecular cross-links do not

induce conformational changes Because inhibition

decreases as protein concentration increases (Fig 3B),

it may be argued that the intermolecular cross-links

protect the enzyme from the inactivating intramolecular

ones As residues involved in the binding of mercury at

the surface are not conserved in the cholinesterase

family, constants estimated for D melanogaster

acetyl-cholinesterase should vary for the other acetyl-cholinesterases;

for example, denaturation rate constants (k3and k8) are

anticipated to be lower for human butyrylcholinesterase

than for D melanogaster acetylcholinesterase, as

inhibi-tion reaches a plateau and total reactivainhibi-tion was found

following dilution

Experimental procedures

Enzymes

D melanogaster acetylcholinesterase was produced in the

baculovirus system and purified as previously described

[52] Mutants C290A, A269C, F330C and Y370C were

obtained by site-directed mutagenesis Mutations outside

the active site (C290A and A269C) did not affect the

enzyme activity, whereas an effect was observed with

muta-tions inside the active site (F330C and Y370C; see

supple-mentary Fig S4) T californica acetylcholinesterase was

generously provided by I Silman (Department of

Neuro-biology, Weizmann Institute of Science, Rehovot, Israel)

The native tetrameric E electricus acetylcholinesterase and

human butyrylcholinesterase used in the kinetic studies

were obtained from Sigma (St Louis, MO, USA) The

recombinant monomeric human butyrylcholinesterase used

in the crystallographic study was produced and purified as

previously described [53], and was generously provided by

F Nachon (Centre de Recherche du Service de Sante´ des

Arme´es, La Tronche, France)

Cholinesterase inhibition

Cholinesterases were diluted in Tris buffer (25 mm,

pH 7.0), and ions that could have interfered with the

solu-bility of mercury were removed by exclusion

chromatogra-phy on a Sephadex G25 column (PD10; Amersham, Saclay,

France) In order to ensure cholinesterase stability, BSA

was added to a final concentration of 0.1 mgÆmL)1, a

con-dition in which no loss of activity was observed after sev-eral hours at 25C Preliminary experiments showed that the impact of albumin on the toxic potency of mercury was negligible in the experimental conditions described herein,

in accordance with previous reports [54]

For the analysis of inhibition time-courses, enzymes were incubated with inorganic mercury (HgCl2) for different time periods, and residual activities were measured for 10 s, fol-lowing 10-fold or 100-fold dilutions A stock solution of the substrate o-nitrophenyl acetate (1 m) was prepared in dimethylsulfoxide, and then diluted to a final concentration

of 1 mm in the reaction buffer The release of the enzymatic product o-nitrophenol was monitored spectrophotometrically

by following its absorbance at 405 nm At the concentra-tions reported in this study, no significant interference of mercury occurred with either the substrate o-nitrophenyl acetate or the product o-nitrophenol [22]

Analysis of equilibrium between human butyrylcholinest-erase and mercury was performed by incubating the enzyme with HgCl2 for 30 min, after which the substrate o-nitro-phenyl acetate was added to the cuvette at different concen-trations without significant dilution of the sample Human butyrylcholinesterase reactivation experiments were per-formed by dialysis using Slide-A-Lyzer Dialysis Cassettes of

10 000 molecular weight cutoff, 0.5–3 mL capacity (Pierce, Rockford, IL, USA)

Experimental data were analyzed by multiple nonlinear regressions, using the fit program gosa (http://www bio-log.biz) Data for Scheme 1A were analyzed using the solved explicit equations [55] As integration of the differen-tial equations was too complex for equations corresponding

to Scheme 1B, numerically solved systems of differential equations were fitted to the data [56]

Crystallization of human butyrylcholinesterase, soaking procedure, and data collection for the HgCl2–human butyrylcholinesterase complex Tetragonal crystals (space group I422) of recombinant monomeric human butyrylcholinesterase were obtained at

20C, using the hanging-drop vapor diffusion method The mother liquor solution was composed of 2.1 m ammonium sulfate and 0.1 m Mes buffer (pH 6.5), and the protein con-centration was 6 mgÆmL)1 As mercury induces protein aggregation and denaturation, we chose a crystal soaking procedure rather than cocrystallization to identify mercury-binding sites A native human butyrylcholinesterase crystal was soaked for 30 min at 20C, in a mother liquor solu-tion containing 10 mm HgCl2 Prior to the flash cooling, the crystal was soaked for 20 s in a cryoprotective solution composed of 2.3 m ammonium sulfate, 0.1 m Mes buffer (pH 6.5), 10 mm HgCl2, and 18% glycerol After flash-cool-ing of the crystal to 100 K in a nitrogen gas stream, X-ray diffraction data were collected on an in-house R-AXIS IV image plate detector installed on a Rigaku (Sevenoaks,

UK) rotating-anode generator At the employed wavelength

(k¼ 1.54 A˚), the anomalous signal of mercury ions

permit-ted their unequivocal identification and localization The

dataset was indexed, merged and scaled using xds⁄ xscale,

and the amplitude factors were generated using xdsconv

[57] For further details, see Table 2

Structure determination and refinement

The native structure of human butyrylcholinesterase

(Pro-tein Data Bank entry code 1POI) without ions, water and

sugar molecules was used as a starting model for rigid body refinement in the resolution range 20–4 A˚ Subsequently, the dataset underwent simulated annealing to 7500 K, with cooling steps of 10 K, followed by 250 steps of conjugate-gradient minimization Diffraction data from 20 to 2.75 A˚ were used for refinement, and maps were calculated using data between 15 and 2.75 A˚ All graphic operations, mode-ling and model building were performed with coot version 0.33 [58] Minimization and individual B-factor refinement followed each stage of manual rebuilding All refinements and map calculations were done using cns version 1.1 [59] Structure refinements were evaluated using the procheck module [60] of the CCP4 suite [61] Figure 4 was produced using pymol [62] A summary of refinement statistics is shown in Table 2

DLS assays Dynamic light scattering (DLS) measurements were performed to assess aggregate formation in samples of

D melanogaster acetylcholinesterase incubated with HgCl2 Samples contained 5 lm enzyme prepared in Tris buffer (25 mm, pH 7.0) and various concentrations of HgCl2 Prior to measurements, enzyme and mercury solutions were filtered through 0.2 lm polyethersulfone membrane dispo-sable filters to ensure elimination of dust particles whose signal would interfere with that of protein molecules Scat-tering data were collected for 60 min, at 20C, using a DynaPro MS⁄ X instrument (Wyatt Technology, Santa Barbara, CA, USA) Recorded data were analyzed using dynamics autocorrelation analysis software (ver-sion 6, Protein Solutions, Wyatt Technology), which allowed us to obtain the median hydrodynamic radius and

an estimate of the size distribution in the sample (polydis-perse index)

Acknowledgements

This work was partially supported by ‘Fundac¸a˜o para

a Cieˆncia e a Tecnologia’ and EU FEDER funds (M F Frasco PhD grant SFRH⁄ BD ⁄ 6826 ⁄ 2001; pro-ject ‘CHOLINEOMANIA’ POCI⁄ MAR ⁄ 58244 ⁄ 2004) and by bilateral cooperation projects Portugal⁄ Slovenia (GRICES⁄ Ministry of Education, Science and Sport, 2006) and Portugal⁄ France (GRICES ⁄ EGIDE, Pessoa program, 2006) We are grateful to Professor Israel Silman and Dr Florian Nachon for the generous gifts of Torpedo californica acetylcholinesterase and recombinant monomeric human butyrylcholinesterase, respectively Financial support by the CEA and the EMBO (ASTF230-2006) to M Weik and J P Colle-tier, respectively, is gratefully acknowledged We grate-fully acknowledge the ESRF for beamtime under long-term projects MX387 and MX498

Table 2 Data collection of the HgCl 2 –human butyrylcholinesterase

complex.

Human butyrylcholinesterase

in complex with

10 m M HgCl2 Protein Data Bank accession code 2J4C

Exposure time (min per frame) 20

Unit cell parameters (A ˚ )

Resolution range (A ˚ ) 20.00–2.75 (2.80–2.75) a

Observations ⁄ parameters ratio 1.90

rmsd with respect to native

structure (A ˚ )

(Protein Data Bank

accession code 1P0I)

0.1840

a

Values in paraentheses are for the highest resolution shell.

b R merge ¼Rhkl R i jI i ðhklÞ<I ðHKLÞ >j

R hkl R i I i ðhklÞ