Binding of fatty acids facilitates oxidation of cysteine-34 and converts copper–albumin complexes...

Binding of fatty acids facilitates oxidation of cysteine-34 andconverts copper–albumin complexes from antioxidants

to prooxidants

Y.A. Gryzunov,a,c A. Arroyo,a J.-L. Vigne,d Q. Zhao,a V.A. Tyurin,a C.A. Hubel,e,f

R.E. Gandley,a,e,f Y.A. Vladimirov,c R.N. Taylor,d and V.E. Kagana,b,*

a Department of Environmental and Occupational Health, University of Pittsburgh, 3343 Forbes Avenue, Pittsburgh, PA 15260, USAb Department of Pharmacology, University of Pittsburgh, 3343 Forbes Avenue, Pittsburgh, PA 15260, USA

c Department of Biophysics, Russian Medical University, Moscow, Russiad Department of Obstetrics/Gynecology and Reproductive Sciences, University of California, San Francisco, CA 94143, USA

e Department of Obstetrics and Gynecology, University of Pittsburghf Magee-Womens Research Institute, Pittsburgh, PA 15213, USA

Received 29 October 2002, and in revised form 3 February 2003

Abstract

As a transition metal capable of undergoing one-electron oxidation–reduction conversions, copper (Cu) is essential for life and

fulfills important catalytic functions. Paradoxically, the same redox properties of copper can make it extremely dangerous because it

can catalyze production of free radical intermediates from molecular oxygen. Factors involved in regulation of redox activity of

albumin-bound copper have not been well characterized. In the present study, effects of modification of the albumin cysteine-34

(Cys-34) and binding of nonesterified fatty acids on the redox-cycling activity of the complex of copper with human serum albumin

(Cu/HSA) were studied. Because ascorbate is the most abundant natural reductant/scavenger of free radicals in blood plasma, the

electron paramagnetic resonance assay of ascorbate radical formation was used as a method to monitor Cu/HSA redox-cycling

activity. At Cu/HSA ratios below 1:1, the bound Cu was virtually redox inactive, as long as Cys-34 was in reduced state (Cu/HSA–

SH). Alkylation, nitrosylation, or oxidation of Cu/HSA resulted in the appearance of redox-cycling activity. Experiments with

ultrafiltration of Cu/HSA alkylated with N-ethylmaleimide (Cu/HSA–NEM) showed that at Cu/HSA–NEM ratios below 1:1, the

ascorbate radicals were produced by Cu tightly bound to HSA rather than by Cu released in solution. The rate of ascorbate radical

production in HSA–NEM and S-nitrosylated HSA (HSA–NO) was, however, more than one order of magnitude lower than that in

a solution containing equivalent concentration of free copper ions. While Cu/HSA–SH was redox inactive, binding of oleic or li-

noleic acids induced Cu-dependent redox-cycling with maximal activity reached at a fatty acid to protein molar ratio of 3:1 for oleic

acid and 2:1 for linoleic acid. Binding of fatty acids caused profound conformational changes and facilitated oxidation of the Cys-34

SH-group at essentially the same ratios as those that caused redox-cycling activity of Cu/HSA. We conclude that fatty acids regulate

anti-/prooxidant properties of Cu–albumin via controlling redox status of Cys-34.

� 2003 Elsevier Science (USA). All rights reserved.

Keywords: Human serum albumin; Nitrosylation; Fatty acids; Copper; Redox-cycling; Ascorbate radical; Cysteine-34

As a transition metal capable of undergoing one-

electron oxidation–reduction conversions, copper is es-

sential for life and fulfills important catalytic functions

in a number of enzymes such as Cu,Zn–superoxide

dismutase, cytochrome oxidase, and ceruloplasmin. In

these enzymatic reactions, copper tightly binds to pro-

teins such that redox activity of the resulting chelate

formed is strictly regulated. Paradoxically, the same

redox properties of copper can make it extremely dan-

gerous for life: it can catalyze production of free radical

intermediates from molecular oxygen, particularly when

copper is released from or mishandled by proteins.

Disruption of the fine structure of copper-binding

Archives of Biochemistry and Biophysics 413 (2003) 53–66

www.elsevier.com/locate/yabbi

ABB

* Corresponding author. Fax: 1-412-383-2123.

E-mail address: [email protected] (V.E. Kagan).

0003-9861/03/$ - see front matter � 2003 Elsevier Science (USA). All rights reserved.

doi:10.1016/S0003-9861(03)00091-2

mail to: [email protected]

domains of mutated proteins enhances the redox activityof copper and results in cell damage and death. Not

surprisingly, delivery of copper to target enzymes is

rigorously controlled in cells and biological fluids. In

plasma, ceruloplasmin and albumin are the two major

proteins (along with transcuprein) responsible for

binding and safe-guarded transport of copper [1] and

prevention of its detrimental redox activity.

The major plasma pool of copper bound to cerulo-plasmin [2] remains apparently inactive with respect to

free radical formation—at least if the stoichiometry of

binding does not exceed 6:1 and/or unless copper is re-

duced by superoxide radicals [3]. The second largest

pool of plasma copper is associated with albumin. Hu-

man serum albumin (HSA)1 contains one high-affinity

site for copper, the N-terminal tripeptide (Cu2þ=Ni2þ

binding motif) Asp-Ala-His [4]. Although under usualcircumstances, less than 1% of total albumin (about

110mg=dl � 1:6lM) contains copper [2,5], HSA still

represents a large pool of bound copper as the human

body contains approximately 400 g of albumin (approx.

6mmol). In some pathologic conditions, e.g., Wilson�sdisease or arthritis, the level of albumin-bound copper

may be significantly increased (two- to fivefold) [6–8].

Human serum albumin contains one reduced cysteineresidue (Cys-34) that, due to the large amount of albu-

min in plasma, constitutes the largest pool of reactive

thiols in plasma [9,10]. The albumin Cys-34 SH-group is

believed to be important in protection against oxidative

stress [11,12], and its concentration has been reported

to decrease under disease conditions and with aging

[13–17].

While both Cu-binding and nucleophilic properties ofCys-34 are important for albumin�s antioxidant func-

tion, specific mechanisms involved in regulation of its

redox activity in human plasma are not fully under-

stood. Recent work has established that chemical

modification of albumin (such as its glycation in

diabetes) yields Cu–albumin molecular species with

elevated redox-cycling activity of Cu, a likely source

of oxidative stress in plasma [18–20]. In addition toposttranslational chemical modifications, albumin may

experience profound conformational changes upon

binding with its natural ligands and different hydro-phobic chemicals and drugs [9,21]. Albumin acts as the

major plasma carrier of nonesterified fatty acids (NE-

FAs), which are accommodated within its two to three

high-affinity binding sites for fatty acids [9]. Notably, as

a result of dyslipidemia typical of several cardiovascular

diseases such as diabetes and preeclampsia, the number

of fatty acid molecules bound to albumin increases [22].

Binding of NEFAs to albumin may be associated withdramatic conformational changes in the albumin mole-

cule that can affect its functional properties [9,23–25].

However, relationships between NEFA binding and

antioxidant properties of albumin (including redox sta-

tus of Cys-34 and redox-cycling activity of Cu-contain-

ing HSA molecular species) have not been quantitatively

characterized.

In the present work, we determined how binding ofNEFAs and/or modification of Cys-34 affect redox-cy-

cling activity of the Cu/HSA complex. Because ascor-

bate is the most abundant natural reductant and

scavenger of free radicals in plasma [26,27], we used

EPR spectroscopy of ascorbate radicals to monitor re-

dox-cycling activity of the Cu/HSA complex. We found

that binding of two to three NEFAs to albumin results

in profound conformational rearrangements that aredetectable by changes in its pI (a redistribution to the 4.8

isoform) and by changes in fluorescence of site-specific

probes. These conformational transitions are accompa-

nied by accelerated oxidation of Cys-34 and increased

redox-cycling activity of bound Cu; consequently,

NEFA-induced rearrangements convert HSA from an

effective antioxidant to a prooxidant, which, in the

presence of Cu, may act as a source of oxidative stress inplasma.

Methods

Reagents

Ascorbic acid (AA), 5,50-dithiobis(2-nitrobenzoicacid) (DTNB); CuSO4 � 5H2O, sodium nitrite (NaNO2),

human serum albumin (essentially fatty acid free), re-

duced glutathione (GSH), dihydrolipoic acid (DHLA),

dithiothreitol (DTT), 2,9-dimethyl-4,7-diphenyl-1,10-

phenanthrolinedisulfonate (bathocuproinedisulfonate,

BCS), ethylenediaminetetraacetic acid (EDTA), diethy-

lenetriaminepentaacetic acid (DTPA), N-ethylmaleimide

(NEM), oleic acid, linoleic acid, ethanol and methanol(HPLC grade), and trichloroacetic acid (TCA) were

obtained from Sigma (St. Louis, MO). Fluorescent

probes, dansyl amide (DNSA) and dansyl sarcosine

(DNSS), were purchased from Sigma. 4,5-Diaminoflu-

orescein (DAF-2) was purchased from Calbiochem (San

Diego, CA). A fluorogenic maleimide-based SH-reagent,

ThioGlo-1, was obtained from Calbiochem (LaJolla,

1 Abbreviations used: AA, ascorbic acid; BCS, 2,9-dimethyl-4,7-

diphenyl-1,10-phenanthrolinedisulfonate (bathocuproinedisulfonate);

Cu/HSA, complex of Cu2þ with HSA; Cu/HSA–NEM, complex of

Cu2þ with HSA–NEM; DAF-2, 4,5-diaminofluorescein; DHLA,

dihydrolipoic acid; DNSA, dansyl amide; DNSS, dansyl sarcosine;

DTNB, 5,50-dithiobis(2-nitrobenzoic acid); DTPA, diethylenetriamine-

pentaacetic acid; DTT, dithiothreitol; EDTA, ethylenediaminetetra-

acetic acid; GSH, reduced glutathione; HSA, human serum albumin;

HSA–NEM, human serum albumin with Cys-34 alkylated with N-

ethylmaleimide; HSA–NO, S-nitrosylated human serum albumin;

HSA–SH, human serum albumin with reduced Cys-34; NEFA,

nonesterified fatty acid; NEM, N-ethylmaleimide; TCA, trichloroacetic

acid; PBS, phosphate-buffered saline; GSNO, S-nitroso-glutathione;

ROS, reactive oxygen species.

54 Y.A. Gryzunov et al. / Archives of Biochemistry and Biophysics 413 (2003) 53–66

CA). Phosphate-buffered saline (PBS; Ca2þ, Mg2þ free)was from Gibco BRL (Grand Island, NY). Chelex-100

resin, analytical grade 100–200 mesh, sodium salt was

from Bio-Rad (Hercules, CA). Cut-off filters were pur-

chased from Millipore Corp. (Bedford, MA). Double

distilled and deionized water was used in all experi-

ments.

Analytical methods

Concentration of copper. In the stock solution this

was quantified using the Cuþ chelator BCS. The BCS–

Cuþ complex exhibits a maximum absorbance at 480 nm

(molar absorbency is 13,500M�1 cm�1) [28]. Absorption

spectra were recorded immediately after addition of 5 llof CuSO4 solution to 1.0ml of PBS containing BCS

(400 lmol/L) and ascorbate (800 lmol/L). Ascorbateand BCS were added in 10- to 12-fold excess of Cu2þ.

Concentration of albumin. This was measured using

absorption at 279 nm. The absorbance at 279 nm for the

concentration of 1mg/ml in a 1-cm cuvette was taken as

E280 ¼ 0:609 [29].

Content of free SH-groups. In reduced HSA (HSA–

SH) and in HSA–NEM this was estimated with Ellman�sreagent, DTNB, using a molar extinction coefficient of13,600M�1 cm�1 at 412 nm [30].

Concentration of NO. This was determined fluoro-

metrically using DAF-2 as a specific probe [31]. To re-

lease NO from S-nitrosothiols, the samples were

irradiated with UV (Oriel UV light source, Model 66002)

using a cutoff filter (Balzers, Pittsburgh, PA) of >330 nm

for 10min (80 lW=cm2). Prior experiments showed that

under these conditions all NO was released from HSA[32]. Fluorescence intensity of DAF-2 was measured

under excitation at 495 nm and emission at 515 nm (slits

1.5 and 5.0 nm, respectively) using a Shimadzu RF-5301

spectrofluorimeter (Kyoto, Japan). A solution of S-nitr-

oso-glutathione (GSNO) was used for calibration of

DAF-2 assay as described in [33]. The concentration of

GSNO was determined photometrically using a molar

extinction coefficient of 900M�1 cm�1 at 336 nm [34].

Fluorescence assay of conformational changes in HSA

(binding sites I and II)

To assess fatty acid-induced conformational changes

of albumin, fluorescent probes DNSA and DNSS were

used as described [35]. DNSA and DNSS are specific

markers for principal albumin binding sites I and II,respectively [35]. Briefly, the fluorescent probes were

dissolved in 1:1 water/methanol mixture. Stock solutions

(10mM) were added to albumin samples (4.2 lM) at a

final concentration of 20 lM. The fluorescence emission

intensity of probes was determined at 483 nm, with ex-

citation at 370 nm (slit 5 nm) using a Shimadzu RF-5301

PC spectrofluorimeter. The data obtained were exported

and processed using Shimadzu RF-5301 PC personalsoftware.

Preparation of HSA derivatives

Several HSA derivatives were used in this study: (1)

HSA with fully reduced Cys-34 (HSA–SH), (2) alkyl-

ated HSA with Cys-34 blocked by N-ethylmaleimide

(HSA–NEM), and (3) HSA with S-nitrosylated Cys-34(HSA–NO). In a series of separate experiments, all

these HSA derivatives were modified with oleic or

linoleic acids.

HSA–SH. Commercial human albumin was 71–77%

oxidized (contained 0.23–0.29 mole of SH-groups/mole

protein). In all our experiments, HSA was, therefore,

reduced with dithiothreitol (1.5- to 2.0-fold excess) for

1.5 h at room temperature in PBS and the protein wasextensively separated (washed) from other solutes with

PBS (pH 7.4) using a 30,000-Da cutoff membrane filter.

After this treatment, approximately 92–98% HSA was

recovered in the reduced form. These results are in good

agreement with previously reported data [36]. This re-

duced HSA was used for further modifications (alkyl-

ation, nitrosylation, binding of Cu).

HSA–NEM. HSA–NEM was prepared by incubationof HSA–SH with N-ethylmaleimide (NEM=HSA–SH ¼5mole=mole), dissolved in PBS (pH 7.4) for 60min at

35 �C in the dark. The excess NEM was eliminated by

ultrafiltration as previously described.

HSA–NO. HSA–NO was prepared using a trans-

nitrosylation reaction with GSNO as a donor [37]. The

GSNO was prepared in a reaction of acidified NaNO2

with GSH. Briefly, 100mM GSH was mixed with100mM sodium nitrite in 20mM HCl at room temper-

ature [37]. The pH was adjusted to 7.4 with NaOH, and

the solution was used for S-trans-nitrosylation of HSA–

SH as rapidly as possible. (The solution was not stored

for more than 30min at 0 �C.) For S-nitrosylation of

HSA–SH, the protein (40mg/ml) was mixed with GSNO

and the solution was incubated for 12–15min at room

temperature. To prevent S-glutathiolation of HSA, thereaction was stopped with NEM, and the samples were

immediately and extensively washed to remove low-

molecular-weight components [37–40]. The resultant

HSA–NEM preparation was checked for the level of

residual HSA–SH and HSA–NO using DTNB and

DAF-2 assays, respectively. In typically used prepara-

tions of HSA–NO, the molar ratio of HSA–NO:HSA–

SH was 96:1.

Formation of complexes of Cu2+ with HSA–NEM (Cu/

HSA–NEM)

An aliquot of freshly prepared CuSO4 solution

in water was added to HSA–NEM and incubated for

45–60min at room temperature in the dark. Because

Y.A. Gryzunov et al. / Archives of Biochemistry and Biophysics 413 (2003) 53–66 55

HSA–SH undergoes rapid oxidation in the presence ofCu, complexes of HSA–SH with Cu were prepared by

preincubation for 40min at 4 �C in the dark (an aliquot

of CuSO4 solution in water was added to freshly pre-

pared HSA–SH). The redox-cycling activity of the

complex was determined after subsequent addition of

ascorbate.

Incorporation of nonesterified fatty acids into HSA

Oleic or linoleic acids were dissolved in methanol–

H2O (1:1, v/v) or pure ethanol, respectively, and incu-

bated with HSA for 2–3min at 25 �C. Solutions of fatty

acids were prepared and used immediately before incu-

bation with HSA. In experiments to determine whether

binding of fatty acids to albumin accelerates copper-

mediated oxidation of HSA–SH, fatty acids (oleic orlinoleic) were dissolved in 1:1 methanol/ethanol mixture.

An appropriate amount of NEFA solution was placed

in an Eppendorf tube, and the solvent was evaporated at

room temperature. Then 100 lM HSA–SH in PBS (pH

7.4) was added and incubated at room temperature for

15min to allow fatty acid binding to albumin; 30 lMCu2þ (as CuSO4, 10mM in ddH2O) was slowly added

and the sample was gently stirred. After 15min of in-cubation with copper, an aliquot was diluted with PBS

(pH 7.4) to 3 lM HSA–SH; then 10 lM fluorescent

probe ThioGlo-1 and 3.3mM SDS were added, and the

kinetics of the fluorescence intensity of the sample were

immediately measured.

Removal of low-molecular-weight components

To remove low-molecular-weight components, sam-

ples were washed five or six times with a four- to fivefold

excess of PBS (pH 7.4) using 30,000-Da cutoff mem-

brane filters and centrifugation at 14,000g for 20–30min

at 4 �C.

Removal of trace metals from solutions

To remove adventitious divalent metals from solu-

tions, water and PBS were treated with Chelex-100 Re-

sin as described in the manufacturer�s instructions. After

decanting the buffer from the resin, the pH was adjusted

to 7.4.

EPR assay of ascorbate radicals (redox-cycling activity)

For these experiments, HSA (100 lM in PBS, pH 7.4)

was mixed with ascorbic acid (18 lM), and EPR signals

of ascorbate radical were scanned during a 30-min time

period. The measurements were performed in gas-per-

meable Teflon tubing (0.8mm internal diameter,

0.013mm thickness) obtained from Alpha Wire Corp.

(Elizabeth, NJ) on a JEOL-RE1X spectrometer at 25 �C.

The Teflon tube (approximately 8 cm in length) was fil-led with 65 ll of the reaction mixture, folded into halves,

and placed into an open EPR quartz tube (inner diam-

eter of 3.0mm) in such a way that the sample was en-

tirely within the microwave radiation area. In a typical

experiment, the spectra of ascorbate radicals were re-

corded under the following conditions: center field

3354G, power 20mW, field modulation 0.79G, sweep

times 10 and 20 s, sweep width 2.5G, receiver gain 4000,time constant 0.1 s. Spectra were collected using EPR-

ware software (Scientific Software Services, Blooming-

ton, IL).

Fresh human plasma

Samples of pooled human plasma were obtained

from Central Blood Bank (Pittsburgh, PA). Sodiumcitrate was used as an anticoagulant (3.15%, final con-

centration).

Isoelectric focusing

This was performed using a Multiphor II electro-

phoresis unit (LKB) connected to a MultiTemp II

thermostatic circulator set at 10 �C. Electrophoresis wasconducted according to the manufacturer�s protocol

using ampholine PAGplates cast on a plastic support

(pH range 4.0–6.5). The anode solution consists of 0.1M

glutamic acid in 0.5M phosphoric acid and the cathode

solution was 0.1M b-alanine. The running conditions

were as recommended: voltage ¼ 2000V, current ¼25mA, time ¼ 2:5 h. Immediately after isoelectric fo-

cusing, the gel was fixed in 10% TCA for 1 h, stained for10min in 0.2% Coomassie brilliant blue dissolved in

25% ethanol/8% acetic acid, and destained in 25% eth-

anol/8% acetic acid. Gels were scanned and the distri-

bution of pI ¼ 5:6 and pI ¼ 4:8 isoforms of albumin

was quantified, assigned arbitrary units, and plotted.

Gels were scanned using a Bio-Rad Fluor-S Multiim-

ager (Richmond, CA). Image capture and subsequent

densitometric analysis of the intensity of bands wasperformed using Bio-Rad Multi-Analyst Software ac-

cording to the manufacturer�s directions.

Results

Redox-cycling activity of Cu/HSA–SH complex is stim-

ulated by Cys-34 oxidation

To determine the extent to which catalytic activity of

Cu/HSA is dependent on the redox status (e.g., oxida-

tion or alkylation) of its Cys-34 residue, we comparedreactions of fatty acid-free HSA with completely re-

duced Cys-34 (HSA–SH) and HSA obtained by a


complete blockade of the cysteine SH-groups with NEM(HSA–NEM).

Fig. 1A shows continuous repetitive recordings of

EPR signals obtained from mixtures of different con-

centrations of Cu (5, 18, and 28 lM) with ascorbate

(18 lM) in the absence of HSA (Fig. 1A, traces 1–3).

Expectedly, Cu exhibited a remarkable catalytic redox-

cycling activity toward ascorbate as evidenced by the

formation of its one-electron oxidation products,ascorbate radicals, in the reaction

Ascorbateþ Cu2þ ! Ascorbate� þ Cuþ:

The radicals were immediately detectable as a charac-

teristic doublet signal with aH 1.79G in the EPR spectra.

The magnitude of the signal decayed over time due toconsumption of ascorbate in the course of reaction (1).

The magnitude of the signal and the life span of ascor-

bate radical signal were dependent on the amount of Cu

present (Fig. 1A, traces 1–3). At 28 lM Cu, a robust and

rapidly decaying response was observable for 40 s (note

that a significant part of the initial signal could not be

observed due to a very rapid decay during the initial

�20 s required to tune the instrument with the sample).The decay time was longer for 18 lM Cu. A smaller

signal and relatively slow decay was observable in the

presence of 5 lM Cu. When Cu was preincubated

with HSA–SH to form a complex of Cu with the

reduced albumin (Cu/total albumin¼ 0.05mole/mole;

Fig. 1. Effect of HSA Cys-34 modification on the time course of ascorbate radical EPR signal formation catalyzed by Cu/HSA complex. (A) Typical

time course of EPR signals from ascorbate radicals produced by different concentrations of Cu in the absence of HSA. 1, Sample contained a mixture

of 5 lM Cu+18 lM ascorbic acid in PBS, pH 7.4; 2, sample contained a mixture of 18 lM Cu+18lM ascorbic acid in PBS, pH 7.4; 3, sample

contained a mixture of 28 lM Cu+18lM ascorbic acid in PBS, pH 7.4. (B) Typical time course of ascorbate radical EPR signals in different samples

of Cu/HSA in the presence of low concentrations of Cu. 1, Sample contained a mixture of Cu/HSA–SH [(100 lM HSA–SH+5 lM Cu2þ) + 18 lMascorbic acid in PBS, pH 7.4]; 2, sample contained a mixture of Cu/HSA–NEM [(100 lM HSA–NEM+5 lM Cu2þ) + 18lM ascorbic acid in PBS,

pH 7.4]; 3, formation of EPR signal of ascorbate radicals as a function of Cu/HSA–NEM concentration. (C) Typical time course of ascorbate radical

EPR signals in different samples of Cu/HSA in the presence of high concentrations of Cu. 1, Sample contained a mixture of Cu/HSA–SH [(100lMHSA–SH+28lM Cu2þ) + 18lM ascorbic acid in PBS, pH 7.4]; 2, sample contained a mixture of Cu/HSA–SH and Cu/HSA–NEM [72lM HSA–

SH+ (28lM HSA–NEM+28lM Cu2þ) + 18lM ascorbic acid in PBS, pH 7.4]; 3, sample contained a mixture of Cu/HSA–NEM [(100lMHSA–NEM+28lM Cu2þ) + 18 lM ascorbic acid in PBS, pH 7.4]; 4, sample contained a mixture of HSA–NO and Cu/HSA–NEM [72lM HSA–

NO+ (28lM HSA–NEM+28 lM Cu2þ) + 8lM ascorbic acid in PBS, pH 7.4]. (D) Duration of lag period for the appearance of ascorbate radical

EPR signals catalyzed by Cu/HSA in the presence of different amounts of Cu. 1, Sample contained a mixture of HSA–SH and Cu/HSA–NEM [72 lMHSA–SH+ (28lM HSA–NEM+72 lM Cu2þ) + 18 lM ascorbic acid in PBS, pH 7.4]; 2, sample contained a mixture of HSA–SH and Cu/HSA–

NEM [72 lM HSA–SH+ (28lM HSA–NEM+28 lM Cu2þ) + 18lM ascorbic acid in PBS, pH 7.4]. In most experiments Cu was added into samples

as Cu/HSA–NEM complex.


HSA–SH=Cu ¼ 100lM=5lM), addition of ascorbatedid not reveal any signals of ascorbate radicals within at

least 20–25min of incubation. The same complex of Cu

prepared with alkylated albumin, HSA–NEM, gener-

ated the signal of ascorbate radicals immediately after

the addition of ascorbate (Fig. 1B, trace 2). The mag-

nitude of the EPR signal was significantly lower than

that obtained from Cu in the absence of albumin (Fig.

1A, trace 1); the magnitude of the EPR signal did notsignificantly change over the initial several minutes of

recording. Varying molar ratios of Cu in Cu/HSA–

NEM complex from 1 to 5 mole% caused proportional

changes in the magnitude of the EPR signal of ascorbate

radicals (Fig. 1B). Thus, binding of Cu by albumin de-

creases its redox-cycling activity. Fully reduced albumin,

HSA–SH, was remarkably more effective in quenching

redox-cycling activity of Cu than alkylated albumin(HSA–NEM).

Because Cys-34 of HSA–SH undergoes rapid oxida-

tion in the presence of Cu (see below), we further utilized

mixtures of Cu/HSA–NEM complex with HSA–SH to

study redox-cycling activity of albumin-bound Cu.

Again, HSA–SH coincubated with HSA–NEM complex

(Cu/total albumin¼ 0.28mole/mole; HSA–SH=HSA–

NEM=Cu ¼ 72lM=28lM=28lM) showed virtually noredox-cycling activity (no detectable EPR signals of

ascorbate radicals) during the initial incubation with Cu

and ascorbate and out to �30min (Fig. 1C, trace 2).

Note that the same amount of Cu bound by HSA–SH

(Cu/total albumin¼ 0.28mole/mole; HSA–SH=Cu ¼100lM=28lM) also did not exert any redox-cycling ac-

tivity within the initial several minutes of incubation with

ascorbate (Fig. 1C, trace 1). In contrast, HSA–NEMwas not able to prevent redox-cycling activity in the pres-

ence of Cu/HSA–NEM complex (Cu/total albumin¼0.28mole/mole; HSA–NEM=Cu ¼ 100lM=28lM) as

evidenced by an immediate appearance of typical

doublet signals of ascorbate radicals in the EPR spectra

(Fig. 1C, trace 3).

Overall, the data show that ascorbate radicals do not

appear if Cys-34 of fatty acid-free HSA is in the reducedstate. These results suggest that the SH-group of Cys-34

in albumin is critically involved in regulation of copper

redox-cycling activity.

We further tested different combinations of Cu/

HSA–SH and Cu/HSA–NEM to determine redox-cy-

cling activity of Cu at different ratios (combinations) of

fully reduced and fully alkylated Cu/HSA. We ob-

served a lag period in the appearance of ascorbateradical signal whose duration was dependent on the

amounts of HSA–SH and on Cu content in the com-

plex. Following a lag period, the signal of ascorbate

radical grew and reached a steady state level. A typical

time course of ascorbate radical for a relatively high

copper concentration (Cu/total albumin¼ 0.7mole/mole;

HSA–SH=HSA–NEM=Cu ¼ 72lM=28lM=72lM) is

shown in Fig. 1D (line 1). In this case, the steady statelevel of ascorbate radical was reached after a short lag

period of about 2–3min. At lower levels of Cu in the

Cu/total albumin complex (0.28mole/mole, HSA–SH=HSA–NEM=Cu ¼ 72lM=28lM=28lM), the steady

state levels of ascorbate radicals were achieved after

more prolonged incubations (�20min) (Fig. 1D, line

2). Decreases of the lag period at higher concentrations

of Cu and increases of the lag period at higher HSA–SH/HSA–NEM ratios suggest that Cu-catalyzed oxi-

dation of Cys-34 SH-groups in HSA, occurring during

the course of incubation, facilitate redox-cycling ac-

tivity of Cu/HSA, decreasing the lag period for ascor-

bate radical production.

To experimentally test this hypothesis we performed

measurements of the time course of oxidation of SH-

groups in HSA using a fluorogenic maleimide SH re-agent, ThioGlo-1. We found that, indeed, HSA–SH was

oxidized both upon addition of Cu and ascorbate (Fig.

2) and in the presence of Cu/HSA–NEM (data not

shown). Thus, time-dependent increases in redox activ-

ity of the Cu/HSA complex are likely to result from Cys-

34 oxidation during the course of incubation. Because

addition of Cu to HSA–SH caused its oxidation, in

subsequent experiments we chose to standardly addnonoxidizable Cu/HSA–NEM complex to HSA–SH or

HSA–NO in tests of the redox-cycling activity.

Fig. 2. Oxidation of Cys-34 SH-groups of HSA–SH catalyzed by Cu/

HSA. 1, HSA–SH alone; 2, HSA–SH+ascorbate; 3, HSA–SH+Cu2þ;4, HSA–SH+Cu2þ +ascorbate. Oxidation of HSA–SH Cys-34 cata-

lyzed by Cu/HSA was measured by ThioGlo-1 fluorescence. The in-

cubation medium contained 3 lM HSA–SH, 0.9lM CuSO4, 50lMascorbate in 100mM PBS, pH 7.4. At indicated times, 200 ll aliquotswere taken and incubated for 5min with 10lM ThioGlo-1 in the

presence of 3.3mM SDS. Fluorescence was recorded at kex ¼ 388 nm,

kem ¼ 483nm. Mean results from four independent experiments are

shown; on average, SE for each time point varied between 4 and 8% of

the initial mean value. Note that a pronounced Cu-dependent oxida-

tion of HSA–SH occurs in the presence of ascorbate.


Nitrosylation of Cys-34–SH confers redox-cycling activ-

ity to HSA

Since HSA–NO represents the major reservoir of

bound nitric oxide in plasma [34], we further determined

whether S-nitrosylation of Cys-34 in HSA–SH confers

redox-cycling activity to its complex with copper. Upon

addition of ascorbate to HSA–NO and Cu/HSA–NEM

mixture, the ascorbate radical signal appeared immedi-ately without any lag period (Fig. 1C, trace 4); however,

in the mixture of HSA–SH and Cu/HSA–NEM, the

signal could be detected only following a lag period of

�20min (Fig. 1D). Thus, in the presence of HSA–NO,

albumin-bound Cu was catalytically as reactive in re-

dox-cycling of ascorbate as it was in Cu/HSA–NEM

complexes.

Redox-cycling activity of Cu/HSA–NEM is due to

albumin-bound (but not free) Cu

To test whether redox activity of HSA–NEM may be

attributed to HSA-bound Cu or to small amounts of

unbound Cu, HSA–NEM (100 lM, 45min) was incu-

bated with different amounts of CuSO4, and the mixture

was filtered through 30,000-Da cutoff ultrafilters usingcentrifugation at 14,000g for 20min. Ascorbate (18 lM)

was then added to either the initial solution or the fil-

trate, and the ascorbate radical EPR sweep signals were

monitored. In Fig. 3A, the magnitude of ascorbate

radical EPR signal is plotted against the Cu/HSA–NEMratio for both initial unfiltered solutions (curve 1) and

ultrafiltrates (curve 2). No EPR signals were detectable

in the ultrafiltrate at Cu/HSA–NEM ratios less than 1:1.

At higher concentrations of Cu, a relatively low-inten-

sity ascorbate radical signal was apparent in EPR

spectra of ultrafiltrates; the magnitude of the signal in-

creased in a manner dependent on the Cu/HSA–NEM

ratio. These results demonstrate that no detectable freeCu was present at a Cu/HSA–NEM ratio less than 1.0.

In unfiltered solutions, the magnitude of the EPR signal

of ascorbate radical increased linearly with increases in

the molar ratios of Cu/HSA–NEM from 0 to 1 (Fig. 3A,

curve 1). At Cu/HSA–NEM ratios exceeding 1, the re-

dox activity of copper sharply increased in unfiltered

samples (Fig. 3A, curve 1). Comparison of the two

slopes of curve 1 obtained at Cu/HSA–NEM ratios <1and ratios >1, respectively, shows that the redox-cycling

activity of secondary low-affinity binding sites of HSA–

NEM (loosely bound copper) is approximately three

times as high as that of tightly bound Cu.

Thus, at Cu/HSA–NEM ratios <1, the redox-cycling

activity toward ascorbate may be strictly attributed to

Cu tightly bound to the high-affinity binding site on the

protein. At higher Cu concentrations, some Cu was ableto diffuse through the membrane of 30,000-Da cutoff

filter as evidenced by ascorbate radical formation in the

ultrafiltrate. In a separate series of experiments, we

found that Cu tightly bound to HSA–NEM produced

Fig. 3. Formation of ascorbate radicals (A) and oxidation of ascorbate (B,C) catalyzed by Cu/HSA–NEM in the presence of different amounts of Cu.

(A) EPR assay of ascorbate radical formation. 1, EPR signals were recorded from a mixture Cu/HSA–NEM+ascorbate; 2, EPR signals were re-

corded in the samples where ascorbate was added after ultrafiltration of Cu/HSA–NEM complex. Dashed line indicates 1:1 Cu/HSA–NEM ratio.

Dotted line continues the slope detected at Cu/HSA–NEM ratios <1. (B) Spectrophotometric assay of Cu/HSA–NEM-catalyzed ascorbate oxidation.

Ascorbate (38lM) was incubated with samples of Cu/HSA–NEM (containing different amounts of Cu) for 25min at room temperature and then

samples were filtered through 30,000-Da cutoff filters. The absorbance of the filtrate was measured at 265 nm and nonoxidized ascorbic acid con-

centration was calculated using a molar absorbance of 13,600M�1 cm�1. Note that the angles between dotted lines show that the rate of ascorbate

oxidation is more than three times higher for loosely bound copper (at Cu=HSA–NEM > 1) than for tightly bound copper (at Cu=HSA–NEM < 1).

(C) DHLA recovers ascorbate after its oxidation by Cu/HSA–NEM (5-min incubation at room temperature). Control, HSA–NEM (100 lM HSA–

NEM)+18 lM ascorbate; Cu/HSA–NEM, Cu/HSA–NEM (100lM HSA–NEM+28 lM Cu2þ) + 18 lM ascorbate; Cu/HSA–NEM+DHLA, [Cu/

HSA–NEM (100lM HSA–NEM+28 lM Cu2þ) + 18 lM ascorbate] +DHLA (6mM).


ascorbate radical signals at least one order of magnitudelower than signals from equivalent amounts of free Cu

(Figs. 1A and B).

To further assess redox activities of tightly bound and

loosely bound copper (at Cu/HSA–NEM ratios >1),

spectrophotometric measurements of ascorbate oxida-

tion were performed. Fig. 3B shows that the amounts of

ascorbate oxidized during a 25-min incubation were

linearly dependent on the Cu/HSA–NEM ratio. Theslope of the curve was, however, approximately three

times greater for Cu/HSA–NEM ratios >1 than for Cu/

HSA–NEM ratios <1—in agreement with our EPR data

on enhanced ascorbate radical production by loosely vs

tightly bound Cu in HSA–NEM.

To determine the formation of dehydroascorbate

during Cu/HSA–NEM-catalyzed oxidation of ascor-

bate, we utilized dihydrolipoic acid known to reducedehydroascorbate back to ascorbate [41,42]. We found

that after 5-min incubation of ascorbate with Cu/HSA–

NEM (Cu/total albumin¼ 0.28mole/mole; HSA–NEM=Cu ¼ 100lM=28lM) 58.9% of ascorbate was oxidized

(Fig. 3C). After addition of dihydrolipoate, 72.0% of

oxidized ascorbate was reduced back to ascorbate, such

that 83.3% of reduced ascorbate was recovered. This

suggests that Cu/HSA–NEM-catalyzed oxidation ofascorbate via intermediate formation of ascorbate rad-

icals generates dehydroascorbate as the major reaction

product.

Binding of fatty acids to HSA–SH enhances its Cu-

dependent redox-cycling activity

Albumin is the major carrier of NEFA in plasma

that, upon binding, changes the conformation of albu-

min and its ability to bind other substrates [9,21].

Therefore, we compared ascorbate radical production

by Cu/HSA–SH in the presence of different amounts of

fatty acids (oleic acid and linoleic acid) with that pro-duced by fatty acid-free Cu/HSA–SH. In the absence of

fatty acids, virtually no ascorbate radical signals were

detected during the initial 2min of incubation of a

mixture of 72% HSA–SH and 28% Cu/HSA–NEM in

the presence of ascorbate (Fig. 4A, trace 1). When

HSA–SH containing oleic acid (at a ratio of 3:1) was

coincubated with Cu/HSA–NEM, pronounced ascor-

bate radical formation was detected without any lagperiod (Fig. 4A, trace 2). At lower oleic acid/HSA ratios,

lag periods were observed, and their durations were in-

versely related to the amounts of oleic acid bound to

albumin (Fig. 4B). Similarly, ascorbate radical forma-

tion was immediately observed in HSA–SH and Cu/

HSA–NEM mixtures upon addition of linoleic acid (at

molar concentrations of >2:1) but only after a lag period

at lower concentrations of linoleic acid (Fig. 4B).To avoid complications due to potential exchange of

free fatty acids between different species of albumin we

determined effects of oleic acid on Cu-induced redox-

Fig. 4. Effect of free fatty acids on the ascorbate radical EPR signals (A) and on the duration of lag period for the appearance of ascorbate radical

EPR signals (B) catalyzed by Cu/HSA–NEM in the presence of HSA–SH. (A) Typical time course of ascorbate radical EPR signal formation

catalyzed by Cu/HSA–NEM in the absence and in the presence of oleic acid and linoleic acid. 1, In the absence of free fatty acids; 2, in the presence of

linoleic acid:HSA–SH (3:1, mole/mole); 3, in the presence of oleic acid:HSA–SH (3:1, mole/mole). (B) Duration of lag period for the appearance of

ascorbate radical EPR signal. 1, In the presence of oleic acid; 2, in the presence of linoleic acid. Each sample contained 72lM HSA–SH+28lMHSA–NEM+28lM Cu in 100mM PBS, pH 7.4. Ascorbate concentration was 18lM. (C) Duration of lag period for the appearance of ascorbate

radical EPR signals catalyzed by Cu/HSA in the absence and in the presence of oleic acid. 1, Sample contained a mixture of Cu/HSA–NEM

[(100lM HSA–NEM+28lM Cu2þ +300lM oleic acid) + 18lM ascorbic acid in PBS, pH 7.4]; 2, sample contained a mixture of Cu/HSA–NEM

[(100lM HSA–NEM+28 lM Cu2) + 18 lM ascorbic acid in PBS, pH 7.4]; 3, sample contained a mixture of Cu/HSA–SH [(100lM HSA–

SH+28lM Cu2þ +300lM oleic acid) + 18lM ascorbic acid in PBS, pH 7.4]; 4, sample contained a mixture of Cu/HSA–SH [(100lM HSA–

SH+28lM Cu2þ) + 18lM ascorbic acid in PBS, pH 7.4]; 5, sample contained a mixture of HSA–NEM [(100lM HSA–NEM)+18 lM ascorbic acid

in PBS, pH 7.4].


cycling of ascorbate in simplified systems using HSA–SH alone or HSA–NEM alone. We found that Cu/

HSA–SH (28 lM Cu, 100 lM HSA–SH preincubated at

4 �C for 40min) did not catalyze one-electron oxidation

of ascorbate during the initial 20min as evidenced by the

absence of any detectable signals of ascorbate radicals in

the EPR spectra. After a lag period of approximately

20min, the EPR signals of ascorbate radicals were de-

tectable in the spectra. In contrast, preincubation of theCu/HSA–SH complex with oleic acid (Fig. 4C, trace 3)

resulted in an immediate appearance of typical doublet

signals of ascorbate radicals that slightly increased over

time. This suggests that oleic acid modifies redox-cycling

activity of Cu–albumin. To assess the role of Cys-34 in

this effect, we further determined to what extent this

could be observed in alkylated albumin. We found that

both in the absence and in the presence of oleic acid (at aratio of 3:1 NEFA/HSA), Cu/HSA–NEM complex was

able to generate ascorbate radicals immediately upon

addition of ascorbate without any lag period (Fig. 4C,

traces 1 and 2). Thus, in already modified (alkylated)

albumin (HSA–NEM), oleic acid did not significantly

change the appearance of redox-cycling response.

Comparison of the results indicates that the effect of

oleic acid is likely realized through its ability to modifyCys-34 in albumin.

Binding of fatty acids induces conformational changes of

HSA

Fatty acids are known to cause conformational

changes in serum albumin affecting the properties of

principal binding sites I and II (see Scheme 1) [9,35]. Weused DNSA and DNSS as fluorescent probes for HSA

binding sites I and II, respectively, to characterize con-formational transitions induced by fatty acids in chem-

ically modified HSAs with reduced, alkylated, and

S-nitrosylated Cys-34. There were similar fluorescence

responses for all three Cys-34-modified HSAs (Fig. 5).

DNSA fluorescence intensity optima were observed at

oleic acid/HSA ratios between 2 and 4. Fluorescence

intensity of DNSS did not change appreciably, but

slightly decreased at oleic acid:HSA ratios >3:1. Thefluorescence responses observed are in line with previ-

ously reported data on fatty acid-induced transitions of

HSA [35]. Redox status of Cys-34 did not affect char-

acteristic fatty acid-induced conformational changes

detected with fluorescent probes.

We further assessed fatty acid-induced changes in

HSA using isoelectric focusing (Fig. 6). Albumin prep-

arations were incubated in the presence of increasingamounts of fatty acids at molar ratios of 0, 1, 2, 3, and 4

relative to HSA. Fig. 6A shows typical profiles of HSA–

SH, HSA–NEM, and HSA–NO incubated in the ab-

sence or in the presence of oleic acid (at a ratio of oleic

acid:HSA of 3:1). One can see that bands corresponding

to albumin isoforms with pI ¼ 5:6 and pI 4.8 were most

prominent although additional bands were also detect-

able. Gels were scanned and the distribution of pI ¼ 5:6and pI ¼ 4:8 isoforms of albumin were quantified, as-

signed arbitrary units, and plotted (Fig. 6B). The iso-

electric profiles of HSA–SH, HSA–NO, and, to a lesser

extent, HSA–NEM demonstrated shifts toward the

more acidic isoform (pI 4.8) as the molar ratio (oleic

acid:HSA) increased. The EC50 of the overall isoelectric

shift was at approximately 3moles of fatty acid/mole of

albumin. Similar data were obtained for linoleic acid(data not shown).

Scheme 1. Locations of domains I, II, and III in human serum albumin

(HSA); high-affinity Cu-binding site and NEFA binding sites are also

shown. Cys-34 that undergoes oxidation, S-nitrosolation, and S-al-

kylation is also indicated. Fluorescent probes DNSA and DNSS bind

to principal site I (domain II) and principal site II (domain III), re-

spectively.

Fig. 5. Effects of oleic acid on fluorescence intensity of DNSA and

DNSS in the presence of HSA–NO (A), HSA–SH (B), and HSA–NEM

(C). Note that the redox status or modification of Cys-34 in HSA did

not affect conformational changes induced by oleic acid as revealed by

DNSA binding to principal binding site I (curve 1) and DNSS binding

to principal binding site II (curve 2). Fluorescence intensity is pre-

sented in arbitrary units (as a ratio of intensities in the presence and

absence of fatty acid).


Fatty acid-induced conformational changes of HSA lead

to increased accessibility and accelerated oxidation of its

Cys-34 SH-group

Fatty acid-induced conformational changes of HSA

may cause increased accessibility of Cys-34 SH-group to

modifying agents. We tested this by measuring the re-

action of non-Cu-containing HSA–SH with an SH re-

agent, DTNB. As shown in Fig. 7, the time course of thereaction of DTNB with HSA–SH Cys-34 was dependent

on the presence of fatty acids. Significantly increased

reaction rates were observed for HSA–SH species whichcontained 3moles of oleic acid or 2moles of linoleic acid

per mole of HSA (Table 1).

Copper-catalyzed oxidation of Cys-34 in HSA–SH

may be affected also by binding of fatty acids to albu-

min. We evaluated the rate of Cys-34 oxidation in Cu/

HSA–SH based on the reactivity of its SH-group with a

fluorogenic maleimide reagent, ThioGlo-1 (Table 1).

Both fatty acids tested (oleic acid and linoleic acid) in-creased the rate of Cys-34 oxidation in Cu/HSA–SH in a

dose-dependent manner. The effect of fatty acids was

maximal at a NEFA:Cu/HSA molar ratio of 3 for oleate

and 2 for linoleate. Further increase of the molar ratio

of fatty acid per albumin molecule did not substantially

increase the rate of HSA–SH oxidation. This suggests

that discrete conformational changes of HSA induced

by fatty acid binding increase accessibility of the Cys-34SH-group to Cu-catalyzed oxidation.

To determine whether our findings in model systems

hold true for physiologically relevant conditions, we

performed several additional experiments with undi-

luted plasma. In these assays, redox-cycling of endoge-

nous ascorbate was estimated. No detectable EPR

signals were observed from intact plasma samples (Fig.

8, trace 1). Addition of a low concentration of Cu withalkylated albumin (Cu/HSA–NEM, 5 lM Cu+100 lMHSA–NEM) resulted in appearance of a weak but dis-

tinctive ascorbate radical EPR signal (Fig. 8, trace 2).

The magnitude of the signal increased eight- to ninefold

when a significantly higher concentration of Cu/HSA

was used (Fig. 8, trace 4). Remarkably, when a low

concentration of Cu/HSA–NEM complex pretreated

with oleic acid was added to undiluted plasma, the EPRsignal of ascorbate radicals was approx twofold greater

than that obtained in the absence of fatty acid (Fig. 8,

traces 2 and 3).

Fig. 7. Effect of oleic acid on the accessibility of HSA Cys-34–SH to

DTNB. 1, Oleic acid:HSA–SH (0:1, mole/mole); 2, oleic acid:HSA–SH

(1:1, mole/mole); 3, oleic acid/HSA–SH (2:1, mole:mole); 4, oleic

acid:HSA–SH (3:1, mole/mole). Accessibility of HSA–SH–Cys-34 to

DTNB was assessed by changes in absorbance at 412 nm vs the blank

without HSA–SH. HSA–SH, 25 lM; DTNB, 250lM in 100mM PBS,

pH 7.4.

Fig. 6. Effect of oleic acid on the isoelectric focusing of HSA (A) and on the distribution of pI 4.8 and pI 5.6 isoforms of HSA (B). Samples of HSA–

SH (a), HSA–NEM (b), and HSA–NO (c) were incubated in the presence of increasing amounts of oleic acid (molar ratios of NEFA/HSA from 0 to 4

were used; results from 0 and 3 are shown in A) and then separated by isoelectric focusing. We observed, in all three albumin forms tested, a re-

distribution between the fractions with pI 5.6 and pI 4.8 with increasing amounts of NEFA. Similar data were obtained for linoleic acid. Data are

mean� SD, *p < 0:01, **p < 0:03, ***p < 0:05 vs control from corresponding HSA (pI 5.6 without free fatty acid). Open bars represent pI 4.8

isoform of HSA; closed bars represent pI 5.6 isoform of HSA.


Discussion

Dietary copper is required for normal functioning of

more than 30 mammalian enzyme systems [43] in which

its catalytic function is due to its ability to act as a

transition metal. Unfortunately, the same redox activity

of copper as both a donor and an acceptor of electrons

may cause oxidative stress and damage to mammalian

cells through catalysis of free-radical reactions. Free

copper readily catalyzes oxidation of the two mostabundant reductants in cells and biological fluids—

ascorbate and thiols—to generate their free radical in-

termediates, i.e., to catalyze their redox-cycling. For

ascorbate, the reaction scheme typically includes the

following reactions [44]:

Cu2þ þAscorbate ! Cuþ þAscorbate�; ð1Þ

O2 þ Cuþ ! Cu2þ þ �O�2 ; ð2Þ

2ð�O�2 Þ þ 2Hþ ! H2O2; and ð3Þ

H2O2 þ Cuþ ! Cu2þ þ �OHþOH�: ð4ÞOne-electron oxidation of ascorbate to its radical

triggers subsequent reactions in which ascorbate radicals

disproportionate to form ascorbate and dehydroascor-

bate, a two-electron oxidation product of ascorbate.Thus during Cu/HSA-catalyzed oxidation, there is a

balance between ascorbate, ascorbate radical, and de-

hydroascorbate. Within this balance, the concentration

of ascorbate radical that was produced from ascorbate

reflects the equilibrium between the fully reduced and

the fully oxidized forms of ascorbate. We further found

that addition of dihydrolipoic acid, a reducing agent

capable of two-electron reduction of dehydroascorbate[41,42], recovered most of the reduced ascorbate (83.3%)

after its oxidation by Cu/HSA–NEM. The unrecovered

part of ascorbate was, most likely, converted into sub-

sequent products of dehydroascorbate degradation such

as diketogulonic acid [45].

Redox-cycling of Cu by bioreductants is accompa-

nied by production of deleterious reactive oxygen spe-

cies (ROS). DNA and particularly its telomericsequences is susceptible to Cu-mediated ROS damage

attributed to hydroxyl radicals [46]. Therefore, it is not

surprising that in biological systems copper is normally

tightly bound to physiological chelators. The ‘‘redox-

cycling activity’’ of copper complexes, i.e., their ability

to undergo reduction of Cu2þ to Cuþ, depends, to a

large extent, on two factors: (i) properties/coordination

of copper in its microenvironment and (ii) accessibilityof bound copper to a reducing agent. Both intracellular

redox-catalysis by copper-containing enzymes and de-

livery of copper to target proteins by copper chaper-

ones are strictly regulated to prevent its uncontrolled

redox-cycling activity.

Table 1

Effects of oleic acid and linoleic acid on HSA–SH susceptibility to DTNB

FA/HSA SH susceptibility to DTNBa SH oxidation rate, k � 103b

Oleate Linoleate Oleate Linoleate

0 0:56� 0:04 0:56� 0:04 15:7� 2:0 13:7� 0:7

1 0:68� 0:07 0:72� 0:05 15:1� 0:6 15:1� 0:8

2 0:74� 0:02 0:81� 0:02 23:5� 1:8 23:3� 1:3

3 0:89� 0:03 0:84� 0:03 22:1� 0:4 27:5� 1:6

4 0:87� 0:05 0:86� 0:05 14:8� 1:8 20:2� 1:5

Data are mean�SD min�1 �103.a Susceptibility of HSA–SH was evaluated as A5=Amax, where A5 is the sample absorbance at 412 nm after 5min of incubation of HSA–SH with

DTNB and Amax is the sample absorbance corresponding to HSA–SH concentration as determined by Ellman�s method [28].b Fluorescent reagent ThioGlo-1 was used to measure SH-groups and the first order reaction rate constants k of HSA–SH oxidation were cal-

culated.

Fig. 8. Typical time course of EPR signals of ascorbate radicals gen-

erated from endogenous ascorbate in undiluted plasma samples in the

presence of Cu/HSA–NEM complex with different amounts of Cu in

the absence or in the presence of oleic acid. 1, Sample contained un-

diluted human plasma; 2, sample contained mixture of Cu/HSA–NEM

(100lM HSA–NEM+5lM Cu2þ) + undiluted human plasma; 3,

sample contained mixture of Cu/HSA–NEM/oleic acid (100 lM HSA–

NEM+5lM Cu2þ +300lM oleic acid)+ undiluted human plasma; 4,

sample contained a mixture of Cu/HSA–NEM (100 lM HSA–

NEM+100lM Cu2þ) + undiluted human plasma.


In plasma, specialized plasma proteins such as ceru-loplasmin, transcuprein, and albumin prevent redox-

cycling activity of copper and govern storage and

delivery of copper. In particular, albumin is believed to

play an important role in safeguarding copper in the

circulation. This is because of effective copper-chelating

properties of the peptide motif at the N terminus of

human albumin (see Scheme 1). In fact, Asp-Ala-His-

Lys, a specific copper-chelating tetrapeptide (d-analogof the N-terminus of human albumin), was shown to

attenuate DNA strand breaks in a dose-dependent

manner [46]. This important protective role of albumin

is determined not only by its ability to safely sequester

copper but also by its interactions with other copper-

binding proteins. Notably, albumin can limit redox-cy-

cling activity of a single copper atom at the surface of

ceruloplasmin, near His-426, which may cause oxidationof lipoproteins in the presence of superoxide [47]. Thus

albumin normally protects against redox-cycling activity

of copper and fulfills an important antioxidant function.

The antioxidant role of albumin in plasma is fortified by

its Cys-34 residue, which can directly participate in

radical scavenging [48,49]. Obviously, modifications of

Cys-34 resulting in loss of its electron-donating SH-

group preclude its direct radical-scavenging antioxidanteffects. In fact, it has been shown that the level of al-

bumin�s SH-groups may change during the course of

several diseases and in aging [13–17]. Several factors

may affect SH-groups in albumin, including free radi-

cals, glutathione, and nitric oxide [34,50]. For example,

enhanced S-nitrosylation of HSA–SH was demonstrated

in plasma from women with preeclampsia as compared

with plasma from women with normal pregnancies [32];elevated levels of albumin-bound fatty acids also are

noted in this condition [51].

Recent studies demonstrated that copper bound to

albumin can mediate the conversion of homocysteine to

its low-molecular-weight disulfide forms (homocystine

and homocysteine–cysteine mixed disulfide) by thiol/di-

sulfide exchange reactions [36]. Notably, albumin me-

diates the formation of disulfide forms of homocysteineby thiol/disulfide exchange, whereas ceruloplasmin

converts cysteine to cystine by copper-dependent au-

tooxidation. Moreover, Cys-34 of HSA–SH can be a

target of the disulfide exchange reactions itself resulting

in the formation of thiolated albumin. This modification

renders albumin–Cu complex highly redox-active, be-

cause removal of disulfide bonded to HSA–Cys-34 thiols

(cysteine and homocysteine) by treatment with dith-iothreitol to form albumin–Cys-34–SH (mercaptalbu-

min) completely blocks the conversion of homocysteine

to its disulfide forms. Since significant amounts of thi-

olated HSA may be present in plasma [61,62] this

modified albumin may function as a potent redox-cy-

cling source contributing to depletion of ascorbate and

induction of oxidative stress in plasma similarly to

S-nitrosylated or alkylated forms of albumin reportedhere.

Modifications of other sites in the HSA molecule may

be further associated with significant changes in its ability

to bind and handle copper. Glycation of albumin has

been shown to impair its ability to regulate redox-cycling

activity of bound Cu [19,20], resulting in cytotoxic effects

on cultured smooth muscle cells due to ROS production

in its presence [52]. This elevated redox-cycling activity ofCu/HSA may contribute to enhanced oxidative stress in

plasma typical of diabetes [18]. In preeclampsia, in-

creased binding of NEFA dramatically changes the

properties of albumin such that its pI is shifted from 5.6

to a more acidic 4.8 conformation [51]. The consequences

of these posttranslational modifications for regulation of

Cu handling by albumin has not been well understood.

Curiously, the two- to threefold molar excess of NEFA inthis clinical setting appears to correspond to the en-

hanced oxidation and redox-cycling of Cys-34.

In the present study, we demonstrated that chemical

modifications of Cys-34 by its alkylation, S-nitrosyla-

tion, or oxidation result in a dramatic enhancement of

the redox-cycling activity of the Cu/HSA complex as

evidenced by one-electron oxidation of ascorbate to

acsorbate radical. We found that HSA–SH and HSAwith chemically modified SH groups are distinct entities

with dramatically different copper redox-cycling activi-

ties. Cu complexes of modified albumins are potent

catalysts of oxidation of ascorbate to its radical. Cu

complexes of HSA–SH, however, do not support one-

electron oxidation of ascorbate until the SH-group is

oxidized during the incubation with ascorbate. We fur-

ther established that the elevated redox-cycling activityof HSA with modified Cys-34 is not due to release of Cu

from albumin but rather is associated with redox-cycling

activity of HSA-bound copper. In line with our findings

on dramatically enhanced redox-cycling activity of Cu in

albumin with modified Cys-34, the Cu binding affinity of

the N-terminal protein site has been recently shown to

be quantitatively higher in HSA–SH than in HSA–NEM

or in albumin with oxidized Cys-34 [53].Thus, the presence of an intact Cys-34 SH-group in

albumin is a prerequisite for its role as an efficient regu-

lator of redox-cycling activity of bound copper. Sup-

pression of the redox-cycling activity of albumin-bound

copper by Cys-34 indicates its potential involvement in

an electron transfer process likely to be accompanied by

the formation of immobilized thiyl radical. In fact, im-

mobilized adducts of thiyl radicals of albumin with a spintrap, 5,5-dimethyl-1-pyrroline-N-oxide, have been de-

tected in the course of Cys-34 oxidation induced by

peroxynitrite [54] or UV light [55]. In this regard, it is

important to elucidate potential mechanisms of reduc-

tion (recycling) of albumin Cys-34 in human plasma.

Fatty acids are the main structural and energy sour-

ces of the human body. Within the organism, they are


presented to cells as fatty acid:albumin complexes [56].In our experiments with oleic and linoleic acids, we

showed that fatty acids are effective modulators of al-

bumin�s redox-cycling activity. This effect is apparent at

fatty acid:albumin ratios exceeding 2:1 for linoleic acid

and 3:1 for oleic acid. In keeping with this, we found

that maximal conformational changes in HSA (revealed

by fluorescence spectroscopy with site-specific probes

and by isoelectric focusing) and maximal changes in therate of Cys-34 oxidation during incubation of Cu/HSA–

SH complex were achieved at approximately the same

molar ratios of fatty acid:albumin. Furthermore, fatty

acid-induced conformational changes in HSA are re-

versible [9]. This implies that decreases of plasma levels

of fatty acids due to their increased uptake and utiliza-

tion in tissues (skeletal muscle, heart) can, potentially,

restore regulation of Cu/HSA redox-cycling activity.Previous work has established that a high-affinity

binding site for NEFA is located in domain III of al-

bumin; the binding capacity of this site is one to two

fatty acid molecules. Subdomain IIB is believed to

harbor the second binding site [9,24,57,58]. The third

binding site for fatty acids with lower affinity than the

other two is likely located in the center of domain I

[24,25], near the active site for copper binding and inproximity to Cys-34 (see Scheme 1). It has been shown

that binding of fatty acids can affect the accessibility of

fluorescent labels covalently bound to Cys-34 [59].

However, there was no prior evidence that this confor-

mational change can affect redox-cycling activity of Cu/

HSA complex. Our results indicate that binding oleic or

linoleic acids to HSA–SH (the most abundant form of

HSA in blood and in tissues [10,17]) dramatically affectsthe susceptibility of its Cys-34 to oxidation.

We found that fatty acid-induced conformational

changes in the Cu/HSA–NEM complex did not affect its

redox-cycling activity. Our experiments with DTNB and

ThioGlo-1, however, showed that NEFA-induced con-

formational changes in HSA–SH increased accessibility

of the albumin SH-group to the SH reagent (DTNB).

This is in agreement with previous experiments whereinbinding of fatty acids by albumin was shown to affect

the thiol accessibility for acrylodan, which covalently

binds to Cys-34 to form a fluorescent complex [59].

Binding of oleic acid to HSA led to a red shift of

acrylodan fluorescence that was interpreted as an in-

creased accessibility of the label to water. Also, other

previous work [60] showed that binding of oleic acid to

HSA increased the accessibility of albumin SH-group toDTNB.

Importantly, similar relationships between the redox

status of Cys-34 in albumin and its redox-cycling ac-

tivity and the effects of fatty acids are inherent to plas-

ma. This was demonstrated in our experiments in which

addition of increasing concentrations of a complex of

alkylated albumin with Cu to undiluted plasma pro-

portionally induced redox-cycling activity detectable byEPR of ascorbate radicals. Moreover, the redox-cycling

activity of Cu/HSA–NEM complex in plasma was sig-

nificantly enhanced by oleic acid.

In summary, we have established that binding of fatty

acids causes conformational changes, facilitates oxida-

tion of Cys-34 SH-groups, and induces redox-cycling

activity of Cu/HSA–SH, i.e., converts HSA from a

radical scavenger and an antioxidant to a Cu-redox-cycling prooxidant. Obviously, only Cu-containing HSA

species that simultaneously harbor either a modified SH-

group and/or are loaded with two to three fatty acids

may be involved in Cu-dependent redox-cycling activity.

One may assume that the concentrations of such Cu/

HSA molecular species in plasma are relatively low.

Given that approximately 1% of HSA in plasma nor-

mally contains Cu [2,5] and only about 0.2–0.5% ofHSA may be S-nitrosylated [32], one can estimate that

only a relatively small fraction of HSA (�0.1%) may

participate in redox-cycling reactions. However, the

high concentration of HSA (�600 lM) and perpetual

catalytic nature of the redox-cycling process can make

these relatively ‘‘rare’’ molecular species of HSA very

powerful sources of oxidative stress. Moreover, our re-

sults suggest that not only S-nitrosylated albumin butalso other HSA molecular species with modified Cys-34

can reveal redox-cycling activity. The fraction of such

modified HSA (e.g., S-glutathionylated HSA) may be

markedly higher (up to 25%) and may further increase in

disease and during aging [17]. Consequently, the fraction

of copper–albumin complexes involved in redox-cycling

activity can exceed 1.0% of total HSA.

Finally, our data clearly indicate that reducing an-tioxidants such as ascorbate cannot be effectively used to

prevent or ameliorate oxidative stress induced by Cu/

HSA. Our results demonstrate that the continuous re-

dox-cycling process depletes any exogenously added

ascorbate, but does not remove the source of redox-cy-

cling. This observation may be of interest for thera-

peutic trials of antioxidant vitamins in clinical

syndromes associated with oxidative stress (e.g., pre-eclampsia) [63]. Clearly, use of Cu chelators may rep-

resent a promising approach to protect against Cu/

HSA-induced oxidative stress. The specific Cu-chelating

tetrapeptide d-analog of the N terminus of human al-

bumin Asp-Ala-His-Lys may be an excellent candidate

for this role as it binds Cu with high affinity, inhibits its

redox-cycling activity [46], and is not sensitive to fatty

acids and/or Cys-34 modifiers that confer redox-cyclingactivity to HSA–SH.

Acknowledgment

This study was supported by NIH 1RO1HL64145-

01A1.


References

[1] M.C. Linder, L. Wooten, P. Cerveza, S. Cotton, R. Shulze,

N. Lomeli, Am. J. Clin. Nutr. 67 (1998) 965S–971S.

[2] K. Inagaki, N. Mikuriya, S. Morita, H. Haraguchi, Y. Nakahara,

M. Hattori, T. Kinosita, H. Saito, Analyst 125 (2000) 197–203.

[3] C.K. Mukhopadhyay, P.L. Fox, Biochemistry 37 (1998)

14222–14229.

[4] J.P. Laussac, B. Sarkar, Biochemistry 23 (1984) 2832–2838.

[5] C. Harford, B. Sarkar, Acc. Chem. Res. 30 (1997) 123–130.

[6] G.W. Rafter, Med. Hypotheses 22 (1987) 245–249.

[7] M. Dastych, Vnitr. Lek. 45 (1999) 217–219.

[8] K.T. Suzuki, Y. Shiobara, A. Tachibana, Y. Ogra, K. Matsum-

oto, Res. Commun. Mol. Pathol. Pharmacol. 103 (1999) 189–194.

[9] D.C. Carter, J.X. Ho, Adv. Protein Chem. 45 (1994) 153–203.

[10] E. Schauenstein, F. Dachs, Z. Naturforsch. [C] 33 (1978) 803.

[11] B. Halliwell, J.M. Gutteridge, Arch. Biochem. Biophys. 280 (1990)

1–8.

[12] M.L. Hu, S. Louie, C.E. Cross, P. Motchnik, B. Halliwell, J. Lab.

Clin. Med. 121 (1993) 257–262.

[13] K. Kumano, S. Yokota, M. Go, K. Suyama, T. Sakai, S. Era,

M. Sogami, Adv. Perit. Dial. 8 (1992) 127–130.

[14] E. Suzuki, K. Yasuda, N. Takeda, S. Sakata, S. Era, K. Kuwata,

M. Sogami, K. Miura, Diabetes Res. Clin. Pract. 18 (1992)

153–158.

[15] M. Sogami, S. Era, S. Nagaoka, K. Kuwata, K. Kida, J. Shigemi,

K. Miura, E. Suzuki, Y. Muto, E. Tomita, et al., J. Chromatogr.

332 (1985) 19–27.

[16] A. Hayakawa, K. Kuwata, S. Era, M. Sogami, H. Shimonaka,

M. Yamamoto, S. Dohi, H. Hirose, J. Chromatogr. B Biomed.

Sci. Appl. 698 (1997) 27–33.

[17] S. Era, K. Kuwata, H. Imai, K. Nakamura, H. Tomoya,

M. Sogami, Biochim. Biophys. Acta 1247 (1995) 12–16.

[18] M. Qian, M. Liu, J.W. Eaton, Biochem. Biophys. Res. Commun.

250 (1998) 385–389.

[19] E. Bourdon, N. Loreau, D. Blache, FASEB J. 13 (2) (1999) 233–

244.

[20] J.W. Eaton, M. Qian, Mol. Cell. Biochem. 234–235 (1–2) (2002)

135–142.

[21] T.J. Peters, All About Albumin: Biochemistry, Genetics and

Medical Applications, Academic Press, San Diego, 1996.

[22] A.T. Hostmark, Med. Hypotheses 44 (1995) 539–541.

[23] I. Petitpas, T. Grune, A.A. Bhattacharya, S. Curry, J. Mol. Biol.

314 (2001) 955–960.

[24] A.A. Bhattacharya, T. Grune, S. Curry, J. Mol. Biol. 303 (2000)

721–732.

[25] S. Curry, P. Brick, N.P. Franks, Biochim. Biophys. Acta 1441

(1999) 131–140.

[26] B. Frei, L. England, B.N. Ames, Proc. Natl. Acad. Sci. USA 86

(1989) 6377–6381.

[27] B. Frei, in: B. Frei (Ed.), Free Radicals in Biology: Sources,

Reactivities, and Roles in the Etiology of Human Diseases,

Academic Press, New York, 1994.

[28] S.M. Lynch, B. Frei, J. Biol. Chem. 270 (1995) 5158–5163.

[29] G.D. Fasman, Practical Handbook of Biochemistry and Molec-

ular Biology, CRC Press Inc, Boston, 1990.

[30] G.L. Ellman, Arch. Biochem. Biophys. 82 (1959) 70.

[31] H. Kojima, N. Nakatsubo, K. Kikuchi, S. Kawahara, Y. Kirino,

H. Nagoshi, Y. Hirata, T. Nagano, Anal. Chem. 70 (1998) 2446–

2453.

[32] V.A. Tyurin, S.X. Liu, Y.Y. Tyurina, N.B. Sussman, C.A. Hubel,

J.M. Roberts, R.N. Taylor, V.E. Kagan, Circ. Res. 88 (2001)

1210–1215.

[33] V.A. Tyurin, Y.Y. Tyurina, S.X. Liu, H. Bayir, C.A. Hubel, V.E.

Kagan, Methods Enzymol. 352 (2002) 347–360.

[34] J.S. Stamler, O. Jaraki, J. Osborne, D.I. Simon, J. Keaney, J. Vita,

D. Singel, C.R. Valeri, J. Loscalzo, Proc. Natl. Acad. Sci. USA 89

(1992) 7674–7677.

[35] D.J. Birkett, S.P. Myers, G. Sudlow, Mol. Pharmacol. 13 (1977)

987–992.

[36] S. Sengupta, C. Wehbe, A.K. Majors, M.E. Ketterer, P.M.

DiBello, D.W. Jacobsen, J. Biol. Chem. 276 (50) (2001)

46896–46904.

[37] Y. Ji, T.P.M. Akeboom, H. Sies, J.A. Thomas, Arch. Biochem.

Biophys. 362 (1999) 67–78.

[38] J.W. Park, Biophys. Biochem. Res. Commun. 152 (1988) 916–920.

[39] E. Konorev, B. Kalyanaraman, H. Hogg, Free Radicals Biol.

Med. 28 (2000) 1671–1678.

[40] N. Hogg, Anal. Biochem. 272 (1999) 257–262.

[41] V.E.Kagan,A.A. Shvedova,E. Serbinova, S.Khan,C. Swansen,R.

Powell, L. Packer, Biochem. Pharmacol. 44 (8) (1991) 1637–1649.

[42] D.P. Xu, W.W. Wells, J. Bioenerg. Biomembr. 28 (1) (1996)

77–85.

[43] A.B. Lentsch, A. Kato, J.T. Saari, D.A. Schuschke, Am.

J. Physiol. Lung Cell. Mol. Physiol. 281 (2001) L387–L393.

[44] B. Halliwell, J.M. Gutteridge, Methods Enzymol. 186 (1990) 1–85.

[45] I. Koshiishi, Y. Mamura, T. Imaneri, Biochim. Biophys. Acta 379

(1998) 257–263.

[46] D. Bar-Or, G.W. Thomas, L.T. Rael, E.P. Lau, J.V. Winkler,

Biochem. Biophys. Res. Commun. 282 (2001) 356–360.

[47] P.L. Fox, B. Mazumder, E. Ehrenwald, C.K. Mukhopadhyay,

Free Radicals Biol. Med. 28 (2000) 1735–1744.

[48] H. Kondo, M. Takahashi, E. Niki, FEBS Lett. 413 (1997)

236–238.

[49] M. Soriani, D. Pietraforte, M. Minetti, Arch. Biochem. Biophys.

312 (1994) 180–188.

[50] D. Jourd�heuil, K. Hallen, M. Feelisch, M.B. Grisham, Free

Radicals Biol. Med. 28 (2000) 409–417.

[51] J.L. Vigne, J.T. Murai, B.W. Arbogast, W. Jia, S.J. Fisher, R.N.

Taylor, J. Clin. Endocrinol. Metab. 82 (1997) 3786–3792.

[52] N. Sakata, K. Miyamoto, J. Meng, Y. Tachikawa, Y. Imanaga,

S. Takebayashi, T. Furukawa, Atherosclerosis 136 (1998)

263–274.

[53] Y. Zhang, D.E. Wilcox, J. Biol. Inorg. Chem. 7 (2002) 327–337.

[54] J. Vasquez-Vivar, A.M. Santos, V.B.C. Junqurira, O. Augusto,

Biochem. J. 314 (1996) 869–876.

[55] J.A. Silvester, G.S. Timmins, M.J. Davies, Free Radicals Biol.

Med. 24 (5) (1998) 754–766.

[56] W. Stremmel, L. Pohl, A. Ring, T. Herrmann, Lipids 36 (2001)

981–989.

[57] H. Aki, M. Yamamoto, J. Pharm. Sci. 83 (1994) 1712–1716.

[58] C.B. Berde, B.S. Hudson, R.D. Simoni, L.A. Sklar, J. Biol. Chem.

254 (1979) 391–400.

[59] R. Narazaki, T. Maruyama, M. Otagiri, Biochim. Biophys. Acta

1338 (1997) 275–281.

[60] R. Narazaki, T. Maruyama, M. Otagiri, Biol. Pharm. Bull. 20

(1997) 452–454.

[61] A.K. Majors, S. Sengupta, B. Willard, M.T. Kinter, R.E. Pyeritz,

D.W. Jacobsen, Arterioscler. Thromb. Vasc. Biol. 22 (8) (2002)

1354–1359.

[62] S. Sengupta, H. Chen, T. Togawa, P.M. DiBello, A.K. Majors, B.

Budy, M.E. Ketterer, D.W. Jacobsen, J. Biol. Chem. 276 (32)

(2001) 30111–30117.

[63] L.C. Chappell, P.T. Seed, A.L. Briely, F.J. Kelly, R. Lee, B.J.

Hunt, K. Parmar, S.J. Bewley, A.H. Shennan, P.J. Steer, Lancet

354 (1999) 810–816.


Binding of fatty acids facilitates oxidation of cysteine-34 and converts copper–albumin complexes...

Documents

Transcript of Binding of fatty acids facilitates oxidation of cysteine-34 and converts copper–albumin complexes...