Free Radical Biology & Medicine, Vol. 37, No. 4, pp. 549 –556, 2004Copyright D 2004 Elsevier Inc.

Printed in the USA. All rights reserved0891-5849/$-see front matter

doi:10.1016/j.freeradbiomed.2004.05.012

Original Contribution

KINETICS OF THE REACTION BETWEEN NITRIC OXIDE AND GLUTATHIONE:

IMPLICATIONS FOR THIOL DEPLETION IN CELLS

LISA K. FOLKES and PETER WARDMAN

Gray Cancer Institute, Mount Vernon Hospital, Northwood, Middlesex HA6 2JR, UK

(Received 12 February 2004; Revised 13 May 2004; Accepted 14 May 2004)

Available online 31 May 2004

Add

Box 10

UK; Fa

Abstract—Nitric oxide in the absence of oxygen was suggested to react with 5–50 mM glutathione (GSH) over many

minutes when [NOS] b [GSH] (N. Hogg et al., FEBS Lett. 382:223–228; 1996). However, Aravindakumar et al. (J.

Chem. Soc. Perkin Trans. 2:663–669; 2002) provided data suggestingf200-fold higher reactivity under conditions of

[NOS] H [GSH]. To help resolve these differences, the rate of loss of NO

S(f9 AM) in aqueous solutions of GSH (2.5–

20 mM) was measured by chemiluminescence. An apparent second-order rate constant of 0.080 F 0.008 M�1 s�1 at pH

7.4, 37jC, was calculated based on the total [GSH] and ‘‘pseudo-first-order’’ kinetics; thiolate anion was much more

reactive than undissociated thiol. These data imply a half-life off30 min for low concentrations of NOSwith 5 mM

GSH, 37jC, pH 7.4, in the absence of oxygen. Possible kinetic schemes that can partially explain the divergent literature

reports are discussed, notably an equilibrium in the reaction between NOSand GSH. Human breast carcinoma MCF-7

cells were exposed to NOS(initiallyf18 AM) in alidded six well plate in an anaerobic chamber in vitro; intracellular

GSH levels decreased by half inf60 min. Aerobic exposure depletes GSH in cells in vitro much faster because of

autoxidation of NOSto NO2

S, >108 times more reactive toward GSH. D 2004 Elsevier Inc. All rights reserved.

Keywords—Nitric oxide, Glutathione, Kinetics, Thiols, Nitrogen dioxide, Free radicals

INTRODUCTION

Nitric oxide (NOS) is highly reactive toward heme

proteins, and its lifetime in vivo—thought to be perhaps

as low asf0.1 s or at most a few seconds [1]—seems

largely attributable to the time for diffusion of NOSfrom

production sites through a number of significant diffu-

sional barriers to hemes in the vasculature [2–4]. By

measuring of one of the reaction products, nitrous oxide,

reaction of NOSwith thiols such as glutathione (GSH)

was suggested to occur on time scales of tens of minutes

at physiological GSH levels, pH, and temperature [5],

but the rate of the reaction was not proportional to GSH

concentration. Direct reaction of NOSwith thiols in the

absence of oxygen, i.e., not involving the formation of

ress correspondence to: P. Wardman, Gray Cancer Institute, P.O.

0, Mount Vernon Hospital, Northwood, Middlesex HA6 2JR,

x: +44 1923 835 210; E-mail: wardman@gci.ac.uk.

549

higher oxides of nitrogen, thus seemed unlikely to be of

physiological importance in vivo. However, a subse-

quent study [6], in which the loss of thiol moiety was

measured in the presence of a large excess of NOS,

provided data from which it can be extrapolated that the

half-life of NOSin the presence of typical physiological

levels of GSH (f5 mM) at pH 7.4, 25jC, would be

around 10 s, at least 100-fold shorter than that suggested

by the data of Hogg et al. [5] and suggesting the

possibility of some physiological relevance of this reac-

tion. A third approach, using electrode measurements,

estimated a rate constant for reaction of NOSwith GSH

indicating a half-life of NOSoff40 s with 5 mM GSH,

37jC [7].

In the present work, an attempt has been made to

resolve these apparent disagreements concerning the

reactivity of NOS

toward GSH. Furthermore, predic-

tions from the kinetic data obtained were tested by

measuring the rate of thiol loss in human tumor cells in

vitro after treatment with NOS

in the absence of

L. K. Folkes and P. Wardman550

oxygen. It is stressed that unless otherwise indicated,

all measurements and discussion refer to reactions in

anoxia. The implications of the presence of atmospheric

oxygen in experiments with nitric oxide, when more

rapid thiol depletion could occur via the formation of

highly reactive nitrogen dioxide [8], are discussed

briefly later.

MATERIALS AND METHODS

Chemicals and media

Glutathione, diethylenetriamine pentaacetic acid

(DTPA), ethylenediamine tetraacetic acid (EDTA), 2,2V-azino-bis-(3-ethylbenzothiazoline-6-sulfonic acid)

(ABTS2�) diammonium salt, 2-mercaptoethanol, Eagle’s

modified medium (EMEM), fetal calf serum (FCS). and

nonessential amino acids (NEAA) were obtained from

Sigma–Aldrich (Gillingham, Dorset, UK). Phosphate-

buffered saline solution (PBS) was made up from tablets

(Oxoid, Basingstoke, UK). Metaphosphoric acid (MPA)

and NaOH were obtained from VWR International

(Poole, UK), and monobromobimane was from Molecu-

lar Probes (Cambridge Bioscience, Cambridge, UK).

Nitric oxide was obtained from BOC Special Gases

(Guildford, UK) at 99.5% purity. Water was purified by

a Milli-RO 8/Milli-Q RG reverse osmosis/deionization

system (Millipore, Watford, UK).

Nitric oxide stock solutions

NOS

was bubbled in a stainless steel/glass system

through a scrubbing vessel with a sintered glass gas

distribution base containing 500 ml deaerated (N2) 1 M

NaOH to remove NO2S, followed by a similar vessel

containing 500 ml water. The efficiency of removal of

NO2S

was checked by bubbling through a deaerated

solution containing ABTS2�; the green color of the

oxidized dye (ABTSS�) was not detectable. For experi-

ments involving thiol depletion in cells, the purified NOS

was then passed through 25 ml water in a sealed vessel

with very little head space. For other experiments,

purified NOSwas bubbled into 60 ml of deaerated 10

mM NaOH in a 100 ml glass syringe with a 5 cm

capillary shank and A5 cone, eliminating the head space

and capping after degassing, to give a saturated solution

of NOSof approximately 1.8 mM. Samples were with-

drawn using a gas-tight microliter syringe via the capil-

lary shank.

Reaction of nitric oxide with glutathione

GSH (5 ml, 2.5–20 mM) in PBS containing 0.1 mM

DTPA at pH 7.4 was placed in a 7 ml thermostatted glass

reaction vessel equipped with a stirrer and a Perspex

(Lucite) plunger sealed with an O ring, with a capillary

bore for syringe sampling/adding reagents. The vessel

was an oxygen electrode chamber (Rank Brothers Ltd.,

Cambridge, UK) modified by replacing the electrode

base with a Pyrex disk, fixed by cyanoacrylate adhesive.

The plunger was inserted partially into the vessel and the

solution deaerated with N2 for 15 min. After deoxygen-

ation, the plunger was fully inserted to eliminate the

head space. NOSstock solution (25 Al, 10 mM NaOH),

which did not affect the pH of the GSH/PBS sample

significantly, was injected into the vessel to start the

reaction, with an initial NOSconcentration of f9 AM.

Samples (50 Al) were removed with a gas-tight syringe,

sampling from the main chamber, and injected into the

scrubbing vessel of a chemiluminescence detector

(Sievers Model 280 nitric oxide analyzer; Analytix

Ltd., Peterlee, UK). The scrubber contained water (5

ml) and the sampling syringe was rinsed with the sample

to be analyzed immediately before taking the sample.

The plunger was screwed down to force excess sample

into the capillary before each sampling, to reduce diffu-

sion of oxygen into the main reaction vessel; N2 was also

passed over the top of the capillary. The chemilumines-

cence signal was integrated and peak areas were plotted

against sampling time.

Cell culture

Human breast carcinoma MCF-7 cells (European

Collection of Cell Cultures, Salisbury, UK) were main-

tained in EMEM containing 10% FCS and 1% NEAA.

Cells were grown to a confluent monolayer before

removal with trypsin.

Measurement of GSH in MCF-7 cells

Experiments were performed in an anaerobic glove

cabinet (Don Whitley Scientific Ltd., Shipley, UK) in

an atmosphere of 5% CO2, 5% H2, 90% N2 with

palladium catalyst. All media, deaerated buffers, and

plastic ware were left in the chamber for at least 48 h to

ensure traces of oxygen had been removed. MCF-7

cells were plated in six well plates with lids (1 � 106

cells/well) and left for 5 h to attach. The medium was

then removed, the cells were washed with 2 ml PBS,

and NOSsolution (2 ml, f18 AM, PBS) was added to

the cells. After incubation for varying times, the NOS

was removed, the cells were washed with 2 ml PBS and

then scraped in 1 ml 5% MPA/1 mM EDTA. The

samples were stored at �20jC before analysis. GSH

was measured by HPLC after derivatization with mono-

bromobimane (MBB) as previously described [9], with

modifications. Samples were spun down and the super-

natant (0.4 ml) was taken and mixed with 2-mercaptoe-

thanol (25 Al, 0.1 mM), MBB (25 Al, 10 mM), Tris

HCl/EDTA (0.25 ml, 2 M/1 mM) for 15 min in the

dark. The samples were acidified with HCl (50 Al, 6 M)

to stop the reaction and the samples were cleaned up by

Fig. 2. (A and B) Dependence upon GSH concentration of the first-order rate constants obtained from exponential fits of thechemiluminescence signals from solutions initially containing f9AM NO

Sand GSH at pH 7.4. (A) 37jC; (B) 25jC. Each data point

is derived from six or seven measurements. (C and D) Dependenceupon GSH concentration of the second-order rate constants obtainedfrom second-order fits of the chemiluminescence signals, convertedto absolute NO

Sconcentrations using an average value for detector

sensitivity.

Kinetics of reaction between NO and GSH 551

extraction of potentially interfering contaminants with

dichloromethane (0.5 ml). Chromatography (HPLC)

involved a Waters Model 616 pump and 717 autosam-

pler (Waters, Elstree, UK) using a 250 � 4.6 mm

Hypersil 5ODS column (Hichrom, Reading, UK). Sep-

aration involved a linear gradient of A, 40 mM

NH4PO4/10 mM H3PO4/5 mM octane sulfonic acid,

and B, 75% acetonitrile, with a gradient of 10–40% B

over 10 min, flow rate 2 ml/min. GSH was detected by

fluorescence emission, using an LS40 fluorescence

detector (Perkin–Elmer, Beaconsfield, UK) with exci-

tation at 398 nm and emission at 476 nm, calibrated

with authentic standards.

Numerical integration of kinetic models

The FACSIMILE code (AEA Technology, Winfrith,

Dorchester, UK) was used.

RESULTS

Reaction of nitric oxide with glutathione

Samples withdrawn at varying intervals from solu-

tions containing initiallyf9 AM NOSand up to 20 mM

GSH showed loss of NOSover tens of minutes (Fig. 1).

Typically, six or seven samples were analyzed for each

run. Figure 1A shows loss of NOSat a fixed concentra-

tion of GSH (10 mM) and varying pH at 27jC, demon-

strating faster reaction as the pH was increased between

pHf6.4 and 8.3; other runs at pHf7.0 and 7.9 were

intermediate in rate between the data points shown.

Figure 1B shows similar data at pH 7.4, 37jC, demon-

strating faster reaction as the GSH concentration was

increased from 5 to 20 mM. However, some decay of

Fig. 1. (A) Variation with time of the chemiluminescence signal fromNOS

in solutions initially containing f9 AM NOS, 10 mM GSH,

phosphate buffer, at 27 F 1jC. (.) pH 6.45; (o) pH 7.39; (n) pH8.28. (B) Chemiluminescence signal from solutions initially contain-ing f9 AM NO

S, phosphate buffer, pH 7.4, at 37jC. (.) No GSH;

(o) 5 mM GSH; (n) 10 mM GSH; (5) 20 mM GSH. (A and B)Dashed lines, exponential fit; solid lines, second-order fit.

NOSoccurred even in the absence of GSH, presumed to

arise from unavoidable ingress of some atmospheric

oxygen via the O ring, Lucite plunger, or sampling

procedure/syringe.

Initially, it was assumed that reaction between GSH

and NOS

involved a simple bimolecular, irreversible

reaction as a rate-limiting step which, under conditions

under which [GSH] H [NOS], would exhibit ‘‘pseudo-

first order’’ kinetics for loss of NOS. The data of [NO

S]

vs. time were therefore initially fitted to an exponential

function (Fig. 1, dashed lines) using nonlinear least

squares (NLLS) to yield first-order rate constants, k1st.

Plots of these rate constants vs. GSH concentration

(Figs. 2A and 2B) indicated first-order dependence on

[GSH] ( p < .0001), although considerable random

variation in independent experiments over many days

is apparent. Assuming a simple bimolecular reaction

(see below), the slopes of the plots of k1st vs. [GSH]

yielded estimates of second-order rate constants at pH

7.4 of 0.080 F 0.008 and 0.044 F 0.008 M�1 s�1 at 37

L. K. Folkes and P. Wardman552

and 25jC, respectively (all uncertainties are standard

errors, SEM).

On examination, it was apparent that in almost every

experiment, decay was not strictly exponential, but either

to an apparent equilibrium or of algebraic form similar to

that found with second-order kinetics. There is some

basis for the applicability of the latter function (see

below), and NLLS second-order fits are shown as solid

lines in Fig. 1. Both examination by eye and the

magnitudes of the uncertainties demonstrated unequivo-

cally the superiority of the second-order fits compared to

simple exponential decay. Extraction of absolute second-

order rate constants from the analysis of chemilumines-

cence signals required a sensitivity factor relating signal

to concentration. Because calibration of the detector was

not performed daily, an average value derived from the

intercepts of the signal at zero time was taken. This may

introduce additional error into the estimate of second-

order rate constants shown in Figs. 2C and 2D, perhaps

F10–20%. The slopes of the linear plots of estimated

second-order rate constants, k2nd vs. [GSH] (Figs. 2C and

2D) were (2.9 F 0.3) � 104 M�2 s�1 and (2.4 F 0.4) �104 M�2 s�1 at 37 and 25jC, respectively.

Depletion of GSH in MCF-7 cells on exposure to nitric

oxide

Human breast tumor cells (MCF-7) were exposed to

nitric oxide in an anoxic chamber and cells analyzed for

GSH after varying times. The experiments involved

exposing cells attached to the surface of standard six

well plastic plates to 2 ml of PBS initially containing

f18 AM NOS. The dishes were not agitated but some

diffusion of NOS

into the chamber atmosphere must

Fig. 3. Effect of NOS(f18 AM initially) on intracellular GSH levels in

MCF-7 cells in open six well plates at 37jC in anoxia. Values are meansof four independent experiments F SEM.

occur during exposure. This was not measured as the

intention was primarily to simulate typical in vitro

experiments in which chemical sources of NOSare used

in common protocols. Figure 3 shows depletion of GSH

after this procedure as a function of time. Control cells

were assayed as having 14.0 F 3.1 fmol GSH/cell.

DISCUSSION

Previous studies have shown that the anaerobic reac-

tion between NOSand GSH results in the formation of

the disulfide GSSG and nitrous oxide. S-Nitrosothiol is

generally not observed in the absence of oxygen (except

when iron impurities are present [10,11], see below) and

reaction is accelerated as the pH is increased [5,6,12,13].

It seems likely that the initial reaction involves reaction

between the thiolate nucleophile and NOS

to form a

radical anion, possibly protonated at oxygen in the pH

range of present interest:

GSH W GS� þ Hþ; ð1Þ

GS� þNOSðþHþÞ W GSNSOH: ð2Þ

Two possible routes to the products GSSG and N2O have

been suggested [6,12,13]. In the first, the radical formed

in Eq. (2) disproportionates to form GSSG and hyponi-

trous acid (Eq. (3)), which then breaks down to N2O:

2 GSNSOH ! GSSGþHONNOH; ð3Þ

HONNOH ! N2OþH2O: ð4Þ

The second possibility is that the initial radical product

reacts further with NOSto form a sulfenic acid and N2O

(Eq. (5)), the sulfenic acid reacting with GSH to yield the

disulfide:

GSNSOHþNOS ! GSOHþ N2O; ð5Þ

GSOHþGSH ! GSSGþH2O: ð6Þ

We consider first the possible mechanism involving

reactions (1)–(4). Study of the kinetics of reaction by

measuring N2O raises the possibility of complications

involving reaction (4) becoming at least partially rate

limiting if reaction conditions are such that observations

are on the time scale of tens of minutes. This is because

the decomposition of hyponitrous acid/hyponitrite is on

this time scale at pH 7.4. Reaction (4) is pH dependent

because of the two prototropic equilibria of hyponitrous

acid (pKaf7 and 11). Reanalysis of the data of Hughes

and Stedman [14] using NLLS fit of the appropriate

function [14] confirms their conclusion that reaction (4)

proceeds via the HONNO� intermediate. This reanalysis

yields estimates, for 25 and 45jC, respectively, of pKa1 =

Kinetics of reaction between NO and GSH 553

7.22 F 0.10 and 6.86 F 0.04 and pKa2 = 11.05 F 0.04

and 10.65 F 0.04 and rate constants for HONNO�

decomposition of (7.22 F 0.10) and (86.5 F 1.1) �10�4 s�1. Arrhenius parameters at pH 6.98 of log10(A/

s�1) = 15.26 F 0.36 and Ea = 109 F 2 kJ mol�1 were

derived. Interpolating the data provides an estimate of

the effective rate constant of reaction (4) of f2.4 �10�3 s�1 at pH 7.4, 37jC (half-life f5 min; note that

the rate constant would be sensitive to ionic strength

because of the effect on pKa). Hogg et al. estimated

first-order rate constants for the formation of N2O under

their conditions of 4.8 � 10�4 and 8.3 �10�4 s�1 at

[GSH] = 5 and 50 mM, respectively (half-livesf24 and

14 min, respectively).

If the reaction involves two, partially rate-limiting

sequential steps involving first-order reactions (Eqs. (2)

and (4)) then the expected buildup of N2O with time is

sigmoidal in nature rather than exponential [15], but

within the random error of typical data [5] this might

not be evident. Simulation of the appropriate function

[15] for a 1–2 h run with k4 = 2.4 � 10�3 s�1 (see above)

and k2obs = 4 � 10�4 or 4 � 10�3 s�1 (corresponding to

k2 = 0.08 M�1 s�1 from our data and [GSH] = 5 or 50

mM, respectively), and NLLS fit to apparent exponential

formation of the N2O product, yields estimates of appar-

ent first-order rate constants off2.7 � 10�4 or 1.3 �10�3 s�1. These values are within a factor of 2 of the

experimental observations of Hogg et al. under these

conditions (kobs = 4.8 � 10�4 and 8.3 � 10�4 s�1,

respectively). The simulations also illustrate, if the model

is correct, that decomposition of hyponitrite will contrib-

ute significantly to the observed rate of formation of N2O

even with [GSH] as low as 5 mM if k2 = 0.08 M�1 s�1

(real and apparent kobsf4 � 10�4 and 2.7 � 10�4 s�1,

respectively).

In very limited separate experiments (data not shown)

we confirmed using gas chromatography and mass

spectrometric head-space analysis that N2O was indeed

formed over tens of minutes with [NOS]0 f9 AM and

[GSH]0 = 1 or 5 mM. Overall, then, our data are

therefore reasonably consistent with the observations

of Hogg et al.

The studies of Aravindakumar et al. [6] involved

quite different reaction conditions and analytical meth-

odology. They used a vacuum-line technique at 25jCto expose solutions of 0.1 mM GSH at pH 6.1 to an

initial concentration of NOS

of 1.52 mM, the latter

almost 200-fold higher than in our experiments. De-

pletion of thiol, measured by Ellman’s reagent after

removal of NOS, was exponential as expected because

[NOS]0 H [thiol]0. The time scale of reaction was over

tens of minutes at pH 6.1; at this pH the half-life

was f10.7 min, corresponding to kobs f1.1 � 10�3

s�1 and effective k2 f 0.71 M�1 s�1 at pH 6.1

because [NOS] = 1.52 mM (defining k2 in terms of

total [GSH], not [GS�]). These authors also measured

similar reactivity of NOSwith cysteine, demonstrating

first-order dependence of the rate constant on the

degree of ionization of cysteine. For GSH, a value of

pK1 f8.8 was suggested and so reaction at pH 7.4

would be expected to be f19-fold faster than at pH

6.1. Thus from these data a value of k2 f13 M�1 s�1

at pH 7.4 is extrapolated, more than 2 orders of

magnitude higher than our estimates at either 37 or

25jC.A potential problem with the method used by Ara-

vindakumar et al. is the necessity to remove all traces of

NOSfrom the solution before exposure to air for analysis

of thiol, because autoxidation of any remaining NOSto

form NO2Swould result in very rapid oxidation either of

remaining thiol [8] or, when Ellman’s reagent is used, of

the measured thionitrobenzoic acid [16]. However, Ara-

vindakumar et al. checked for the absence of S-nitro-

sothiol, and random error which might have arisen from

variations in the efficiency of removal of NOSis not

evident in their data. The demonstrated dependence on

the degree of thiol ionization (for cysteine) is consistent

with prior observations [12] and our data (Fig. 1A). It

seems unlikely that extrapolation of the data obtained by

Aravindakumar et al. from pH 6.1 to pH 7.4 on this basis

is inappropriate.

It is possible that these differences in apparent

reactivity of NOStoward GSH arise from a change in

mechanism when [NOS] is increased from a few micro-

molar to 1–2 mM. Reaction (5) might become more

important than reaction (3), for example, or reaction (2)

might actually be an equilibrium, as suggested by Gow

et al. [17] (see below). A study involving measurement

of GSH loss and N2O formation from 0.1 mM GSH

with up to 2 mM NOSfrom a chemical source (DEA/

NO, see below) reported complete thiol loss and sub-

stantial N2O formation after 30 min treatment at 37jC,pH 7.4 [13]. Unfortunately, the time course was not

studied, but complete thiol loss after 30 min is consis-

tent with the kinetic data of Aravindakumar et al. but

not with our data nor those of Hogg et al., the latter two

studies both involving much lower concentrations of

NOS.

However, the suggestion that the rapid kinetics

reported for millimolar levels of NOSmight not be

appropriate to extrapolate to micromolar levels of NOS

with GSH in large excess (the cellular/physiological

situation) must be approached with caution. Another study

[7] reported a rate constant for reaction of NOSwith GSH

at 37jC, pH 7.4, of 3.31F 0.33M�1 s�1,f40-fold higher

than our estimate and derived from measurements in

which [NOS]0f0.2 AM and [GSH]0 = 0.2–0.8 mM. An

electrode was used tomonitor NOSbut the data showed the

L. K. Folkes and P. Wardman554

signal ascribed to [NOS] decreased by about half inf6

min even in the absence of GSH. This seems too high to

ascribe to autoxidation by oxygen contamination, because

the rate is proportional to the square of [NOS] and the half-

life of NOSwould probably be rather longer than 6 min at

submicromolar [NOS] and likely levels of oxygen con-

tamination [18]. We have also attempted to use a similar

electrode to make such measurements but found a similar

lack of stability of the signal (data not shown).

In addition to concentrations of reagents, a further

possible variable is the role of trace metal contamina-

tion, particularly iron and copper. In both the present

work and the study of Hogg et al. [5], DTPA was

included to minimize such possible effects. Significant

enhancement of the nucleophilic reactivity of thiols

when complexed to such metals is known [19], and

effects of iron on the anaerobic reaction between thiols

and NOS

have been reported [10,11]. In the latter

studies, added iron chelators blocked the formation of

S-nitrosothiols. Micromolar levels of iron are found in

millimolar solutions of commercial cysteine and GSH

[20]. Trace metals might act as electron acceptors from

the radical formed in reaction (2), the redox properties

of which are unknown (cf. the suggested reaction with

NAD+ [17]). Release of NOS

from S-nitrosothiols,

catalyzed by copper, is a further possible complication

[21,22], especially after aerobic reaction of NOSwith

thiols.

Notwithstanding the apparently conflicting data, our

observations of possible second-order decay of [NOS] with

time (Fig. 1) prompted us to consider in more detail a

mechanism in which reaction (2) is an equilibrium (k-2 >

0). For simplification, we ignore the possible reactions (3)

and (4) and consider a scheme in which reaction (2) is an

equilibrium and reaction (6) is sufficiently fast so as not to

be rate limiting. (It is likely that k6 > 700 M�1 s�1 [23].) It

can be shown, using the steady-state approximation for

[GSNSOH] [24], that the rate law for loss of NO

Sindicates

second-order dependence on [NOS] and first-order depen-

dence on [GSH]; further, under conditions under which

k-2 H k5 [NOS], the overall third-order rate constant for

loss of NOS(estimated by the slopes of the lines in Figs. 2C

and 2D) approximates to 2K2k5. Our data thus suggest that

K2k5f104 M�2 s�1. By comparison with other radical/

radical reactions of NOS[25] it is possible that k5 > 108

M�1 s�1 and thus K2 < 10�4 M�1. For the rate of the back

reaction (�2) to be significant compared to reaction (5)

with [NOS] as low asf10�5 M and k5 > 108 M�1 s�1,

k-2 > 103 s�1.

To test this latter mechanism we therefore numerically

integrated the simplified reaction scheme (reactions (2),

(5), and (6) with k6 = 103 M�1 s�1), using the FACSIM-

ILE code. Guided by the above rationale, we found, for

example, that values of k2f 102 M�1 s�1, k-2f1.5 �

106 s�1, and k5 f108 M�1 s�1 yielded simulations not

dissimilar (within a factor of f2) to the experimental

data for both our study and that of Aravindakumar et al.

With these values, decay of NOS

is simulated to be

accurately second order with a first half-life off830 s

with [NOS]0 = 9 AM, [GSH]0 = 10 mM (experimental

data (Fig. 2) suggest a value off870 s at 37jC). With

1.52 mM NOSand [GSH]0 = 0.1 mM, decay of GSH was

close to exponential after slight initial deviation, with

half-lifef33 s; extrapolation of the data of Aravindaku-

mar et al. at pH 6.1 to pH 7.4 suggested a half-life of

f35 s. Noting the possible errors involved in extrapo-

lating well beyond the pH data space, the agreement is

excellent. Other combinations of parameters also fitted

the data, providing k-1 was adjusted to keep K1k2constant atf7 � 103 M�2 s�1. However, these param-

eters do not fit the data of Hogg et al. [5] nor those of

Friedman et al. [7]. We therefore stress that many

uncertainties remain.

Our main interest in this study was in predicting the

effects of low (micromolar) concentrations of NOSon

intracellular thiols in vitro. The experimental observa-

tions in Fig. 3 are consistent with our kinetic measure-

ments, bearing in mind that NOSwill be lost through

diffusion to the atmosphere, the effects varying with

experimental conditions [26,27]. As intracellular reac-

tion occurs, more NOS

diffuses into cells from the

reservoir of the medium (2 ml � 18 AM NOS= 36

nmol; 106 cells � 14 fmol/cell GSH = 14 nmol); the root

mean square three-dimensional diffusion distance of

NOSin water at 25jC isf0.4 mm in 10 s, so diffusion

is unlikely to be rate limiting. In other studies [28] an

intracellular probe for NOS

with a linewidth-sensitive

electron paramagnetic resonance signal was used to

demonstrate removal of NOS

(initially f75 AM) by

hypoxic Chinese hamster ovary cells in vitro at a rate

of 0.014 fmol/cell/s (f50 nmol/106 cells/h), but it is

noteworthy that inhibition of mitochondrial respiration

effectively inhibited NOSremoval.

All the above data and discussion relate to anaerobic

systems. We now consider briefly the effects of oxygen.

A convenient method to expose cells to nitric oxide is to

add a compound that spontaneously decomposes to yield

NOS: the diazeniumdiolates (‘‘NONOates’’) are particu-

larly useful [29]. It is easy to calculate, using the steady-

state approximation or more complex models [30], that

adding, e.g., 1 mM diethylamine analog (DEA/NO) to

pH 7.4 buffer at 37jC can lead to steady-state levels of

NOSof tens of micromolar if the solution is air-saturated,

hydrolysis of DEA/NO with k f5.9 � 10�3 s�1 [29]

(half-lifef2 min) defining NOSinput and autoxidation

controlling its removal [30]. With such a treatment,

depletion of intracellular GSH is much more rapid than

under anaerobic conditions because autoxidation of NOS

Kinetics of reaction between NO and GSH 555

followed by reaction of NO2Swith GSH [8] is much faster

than direct reaction of NOSwith GSH [18]. In earlier

studies (with Dr. K.A. Smith, data not shown) we

confirmed this prediction by observing depletion of

intracellular GSH in human leukemia HL-60 cells in

vitro by the aerobic addition of 1 mM DEA/NO to 8 �105 cells/ml. After 30 min, intracellular GSH was <2% of

the control value. Another study showed that aerobic

exposure of neutrophils to 0.1 mM NOSfor 2 min at

37jC, pH 7.4, decreased intracellular GSH from 1.8 F0.5 to 0.4 F 0.2 fmol/cell [31]. Trapping NO

Swith

ferrous diethyldithiocarbamate in aerobic solutions con-

taining initially 2 AM NOSand 2 mM GSH demonstrated

a rate of NOS

loss consistent with NOS

autoxidation

being rate limiting rather than a direct reaction between

NOSand GSH [20]. In addition to acceleration of thiol

depletion, even 1% oxygen is sufficient to result in S-

nitrosothiol (GSNO) formation [32], but GSNO is gen-

erally not observed in the absence of oxygen [13,33,34].

Formation of GSNO in solutions of GSH with NOS/O2

occurs at a rate limited by the kinetics of reaction

between NOSand O2 [33–35], as expected from the

kinetics of the reaction of NO2Swith GSH or cysteine

[8,36,37] (reaction of N2O3 with GSH is likely to

dominate only at nonphysiological levels of NOS[8,34]).

In conclusion, the reactivity of nitric oxide toward

GSH in the absence of oxygen may not necessarily be

characterized by a single rate constant. The mechanism,

or certainly the apparent kinetic parameters, may change

according to reaction conditions. The reaction may also

be sensitive to metal catalysis. In the present study, we

have concentrated on measurements with low concentra-

tions of NOSto approach physiological relevance. There

are obvious deficiencies in our data, probably systematic

error from sampling techniques and significant random

error, both in our view within the bounds of acceptability

but nonetheless real problems. There is thus a need for

further studies with improved experimental techniques

that would enable a wider range of reaction conditions to

be investigated in a single study. Despite the remaining

uncertainties, it is important to be aware that exposure of

cells to low concentrations of nitric oxide in vitro may

result in significant thiol depletion even in the absence of

oxygen, on the time scale of common experimental

procedures. However, our data suggest that at the low

levels of nitric oxide that are likely to be achieved in vivo,

the reaction of nitric oxide with GSH is too slow to be of

physiological relevance. Even with the simplified pseudo-

first-order analysis, the half-life of low concentrations of

NOSis estimated to bef30 min with 5 mM GSH present,

orders of magnitude longer than the (at most) few

seconds’ ‘‘natural’’ lifetime of NOSin vivo. If the reaction

is second order in [NOS], the reaction might be even

slower in vivo than our data suggest.

Acknowledgment—This work was supported by Cancer Research UK.

REFERENCES

[1] Thomas, D. D.; Liu, X.; Kantrow, S. P.; Lancaster, J. R., Jr. Thebiological lifetime of nitric oxide: implications for the perivas-cular dynamics of NO and O2. Proc. Natl. Acad. Sci. USA 98:355–360; 2001.

[2] Lancaster, J. R., Jr. Simulation of the diffusion and reaction ofendogenously produced nitric oxide. Proc. Natl. Acad. Sci. USA91:8137–8141; 1994.

[3] Butler, A. R.; Megson, I. A.; Wright, P. G. Diffusion of nitricoxide and scavenging by blood in the vasculature. Biochim. Bio-phys. Acta 1425:168–176; 1998.

[4] Lancaster, J. R., Jr. The physical properties of nitric oxide: deter-minants of the dynamics of NO in tissue. In: Ignarro, L. J., ed.Nitric oxide: biology and chemistry. San Diego: Academic Press;2000: 209–224.

[5] Hogg, N.; Singh, R. J.; Kalyanaraman, B. The role of glutathionein the transport and catabolism of nitric oxide. FEBS Lett.382:223–228; 1996.

[6] Aravindakumar, C. T.; De Ley, M.; Ceulemans, J. Kinetics of theanaerobic reaction of nitric oxide with cysteine-containing pro-teins: implications for in vivo S-nitrosation. J. Chem. Soc. PerkinTrans. 2:663–669; 2002.

[7] Vriesman, M. F.; Haenen, G. R. M. M.; Westerveld, G.-J.; Paquay,J. B. G.; Voss, H.-P.; Bast, A. A method for measuring nitric oxideradical scavenging activity: scavenging properties of sulfur-con-taining compounds. Pharm. World Sci. 19:283–286; 1997.

[8] Ford, E.; Hughes, M. N.; Wardman, P. Kinetics of the reactions ofnitrogen dioxide with glutathione, cysteine, and uric acid atphysiological pH. Free Radic. Biol. Med. 32:1314–1323; 2002.

[9] Fahey, R. C.; Newton, G. L.; Dorian, R.; Kosower, E. M. Analysisof biological thiols: quantitative determination of thiols at thepicomole level based upon derivatization with monobromobi-manes and separation by cation-exchange chromatography. Anal.Biochem. 111:357–365; 1981.

[10] Vanin, A. F.; Malenkova, I. V. Iron is a catalyst of cysteine andglutathione S-nitrosation on contact with nitric oxide in aqueoussolutions at neutral pH. Biochemistry (Moscow) 61:374–379;1996.

[11] Vanin, A. F.; Malenkova, I. V.; Serezhenkov, V. A. Iron cata-lyzes both decomposition and synthesis of S-nitrosothiols: opti-cal and electron paramagnetic resonance studies. Nitric Oxide 1:191–203; 1997.

[12] Pryor, W. A.; Church, D. F.; Govindan, C. K.; Crank, G. Oxi-dation of thiols by nitric oxide and nitrogen dioxide: syntheticutility and toxicological implications. J. Org. Chem. 47:156–159;1982.

[13] DeMaster, E. G.; Quast, B. J.; Redfern, B.; Nagasawa, H. T.Reaction of nitric oxide with the free sulfhydryl group of humanserum albumin yields a sulfenic acid and nitrous oxide. Bioche-mistry 34:11494–11499; 1995.

[14] Hughes, M. N.; Stedman, G. Kinetics and mechanism of thedecomposition of hyponitrous acid. J. Chem. Soc. 1239–1243;1963.

[15] Capellos, C.; Bielski, B. H. J. Kinetic systems: mathematicaldescription of chemical kinetics in solution. New York: Wiley;1972.

[16] Gergel, D.; Cederbaum, A. I. Interaction of nitric oxide with 2-thio-5-nitrobenzoic acid: implications for the determination of freesulfhydryl groups by Ellman’s reagent. Arch. Biochem. Biophys.347:282–288; 1997.

[17] Gow, A. J.; Buerk, D. G.; Ischiropoulos, H. A novel reactionmechanism for the formation of S-nitrosothiol in vivo. J. Biol.Chem. 272:2841–2845; 1997.

[18] Wardman, P. Nitrogen dioxide in biology: correlating chemicalkinetics with biological effects. In: Alfassi, Z. B. ed. The chemistryof N-centered radicals. New York: Wiley; 1998, 155–179.

[19] Willson, R. L.; Searle, A. J. F. Metronidazole (Flagyl): iron cata-lysed reaction with sulphydryl groups and tumour radiosensitisa-tion. Nature 255:498–500; 1975.

L. K. Folkes and P. Wardman556

[20] Sheu, F.-S.; Zhu, W.; Fung, P. C. W. Direct observation of trap-ping and release of nitric oxide by glutathione and cysteine withelectron paramagnetic resonance spectroscopy. Biophys. J.78:1216–1226; 2000.

[21] Williams, D. L. H. The chemistry of S-nitrosothiols. Acc. Chem.Res. 32:869–876; 1999.

[22] Hogg, N. Biological chemistry and clinical potential of S-nitro-sothiols. Free Radic. Biol. Med. 28:1478–1486; 2000.

[23] Armstrong, D. A.; Buchanan, J. B. Reactions of O2S�, H2O2 and

other oxidants with sulfhydryl enzymes. Photochem. Photobiol.28:743–755; 1978.

[24] Espenson, J. H. Chemical kinetics and reaction mechanisms, 2nded. New York: McGraw-Hill; 1995.

[25] Czapski, G.; Holcman, J.; Bielski, B. H. J. Reactivity of nitricoxide with simple short-lived radicals in aqueous solutions. J. Am.Chem. Soc. 116:11465–11469; 1994.

[26] Chen, B.; Deen, W. M. Analysis of the effects of cell spacing andliquid depth on nitric oxide and its oxidation products in cellcultures. Chem. Res. Toxicol. 14:135–147; 2001.

[27] Chen, B.; Keshive, M.; Deen, W. M. Diffusion and reaction ofnitric oxide in suspension cell cultures. Biophys. J. 75:745–754;1998.

[28] Clarkson, R. B.; Norby, S. W.; Smirnov, A.; Boyer, S.; Vahidi, N.;Nims, R. W.; Wink, D. A. Direct measurement of the accumula-tion and mitochondrial conversion of nitric oxide within Chinesehamster ovary cells using an intracellular electron paramagneticresonance technique. Biochim. Biophys. Acta 1243:496–502;1995.

[29] Davies, K. M.; Wink, D. A.; Saavedra, J. E.; Keefer, L. K. Che-mistry of the diazeniumdiolates. 2. Kinetics and mechanism ofdissociation to nitric oxide in aqueous solution. J. Am. Chem.Soc. 123:5473–5481; 2001.

[30] Schmidt, K.; Desch, W.; Klatt, P.; Kukovetz, W. R.; Mayer, B.Release of nitric oxide from donors with known half-life: a mathe-

matical model for calculating nitric oxide concentrations in aero-bic solutions. Naunyn-Schmiedeberg’s Arch. Pharmacol. 355:457–462; 1997.

[31] Clancy, R. M.; Levartovsky, D.; Leszczynska-Piziak, J.; Yegudin,J.; Abramson, S. B. Nitric oxide reacts with intracellular gluta-thione and activates the hexose monophosphate shunt in humanneutrophils: evidence for S-nitrosoglutathione as a bioactive inter-mediary. Proc. Natl. Acad. Sci. USA 91:3680–3684; 1994.

[32] Jourd’heuil, D.; Jourdheuil, F. L.; Feelisch, M. Oxidation andnitrosation of thiols at low micromolar exposure to nitric oxide:evidence for a free radical mechanism. J. Biol. Chem. 278:15720–15726; 2002.

[33] Wink, D. A.; Nims, R. W.; Darbyshire, J. F.; Christodoulou, D.;Hanbauer, I.; Cox, G. W.; Laval, F.; Laval, J.; Cook, J. A.; Krishna,M. C.; DeGraff, W. G.; Mitchell, J. B. Reaction kinetics fornitrosation of cysteine and glutathione in aerobic nitric oxidesolutions at neutral pH: insights into the fate and physiologicaleffects of intermediates generated in the NO/O2 reaction. Chem.Res. Toxicol. 7:519–525; 1994.

[34] Goldstein, S.; Czapski, G. Mechanism of the nitrosation of thiolsand amines by oxygenated

SNO solutions: the nature of the nitro-

sating intermediates. J. Am. Chem. Soc. 118:3419–3425, 6806;1996.

[35] Kharitonov, V. G.; Sundquist, A. R.; Sharma, V. S. Kinetics ofnitrosation of thiols by nitric oxide in the presence of oxygen.J. Biol. Chem. 270:28158–28164; 1995.

[36] Prutz, W. A.; Monig, H.; Butler, J.; Land, E. J. Reactions ofnitrogen dioxide in aqueous model systems: oxidation of tyrosineunits in peptides and proteins. Arch. Biochem. Biophys. 243:125–134; 1985.

[37] Forni, L. G.; Mora-Arellano, V. O.; Packer, J. E.; Willson, R. L.Nitrogen dioxide and related free radicals: electron-transfer reac-tions with organic compounds in solutions containing nitrite ornitrate. J. Chem. Soc. Perkin Trans. 2:1–6; 1986.