Báo cáo khoa học: Identification and characterization of oxidized human serum albumin A slight structural change impairs its ligand-binding and antioxidant functions pptx

Identification and characterization of oxidized human
serum albumin
A slight structural change impairs its ligand-binding and antioxidant
functions
Asami Kawakami
1,
*, Kazuyuki Kubota
2,
*, Naoyuki Yamada
2
, Uno Tagami
2
, Kenji Takehana
1
,
Ichiro Sonaka
1
, Eiichiro Suzuki
2
and Kazuo Hirayama
2
1 Pharmaceutical Research Laboratories, Ajinomoto Co. Inc., Kawasaki, Japan
2 Institute of Life Science, Ajinomoto Co. Inc., Kawasaki, Japan
Human serum albumin (HSA) is the most abundant
protein in plasma ($40 mgÆmL
)1
or 0.6 mm), and
accounts for 50–60% of total plasma protein (75–
80 mgÆmL
)1
) [1]. HSA (66 kDa) is a single-chain

polypeptide of 585 residues, which has heterogeneity
as a result of post-translational nonenzymatic modifi-
cations such as oxidation and glycation. Plasma HSA
is divided into two types depending on its redox
state: reduced HSA (HMA; human mercaptoalbumin)
and oxidized HSA (HNA; human nonmercaptoalbu-
min). Reduced HSA contains 17 disulfide bonds and
one free thiol group at Cys34 [2]. Oxidized HSA is a
generic name for those proteins that have various
modifications at Cys34. HSA is a mixture of reversi-
bly and irreversibly oxidized HSA. Reversibly oxid-
ized HSA has mixed disulfide bonds with a thiol
compound such as cysteine, homocysteine [3,4] or
glutathione. In irreversibly oxidized HSA, Cys34 is
more highly oxidized to sulfenic acid (-SOH), sulfinic
acid (-SO
2
H), sulfonic acid (-SO
3
H), or S-nitroso
thiol (-SNO) [5,6]. As the major oxidized form is
reversibly oxidized HSA, the proportion of reduced
HSA [HSA(red)%] changes according to surrounding
conditions:
Keywords
human serum albumin; mercaptoalbumin;
nonmercaptoalbumin; ESI-TOFMS; oxidation
Correspondence
K. Takehana, Pharmaceutical Research
Laboratories, Ajinomoto Co. Inc., 1–1

Suzuki-cho, Kawasaki-ku, Kawasaki,
210–8681, Japan
Fax: +81 44 2105871
Tel: +81 44 2105822
E-mail:
*These authors contributed equally to this
work
(Received 26 April 2006, accepted 25 May
2006)
doi:10.1111/j.1742-4658.2006.05341.x
Human serum albumin (HSA) exists in both reduced and oxidized forms,
and the percentage of oxidized albumin increases in several diseases. How-
ever, little is known regarding the pathophysiological significance of oxida-
tion due to poor characterization of the precise structural and functional
properties of oxidized HSA. Here, we characterize both the structural and
functional differences between reduced and oxidized HSA. Using LC-ESI-
TOFMS and FTMS analysis, we determined that the major structural
change in oxidized HSA in healthy human plasma is a disulfide-bonded
cysteine at the thiol of Cys34 of reduced HSA. Based on this structural
information, we prepared standard samples of purified HSA, e.g. nonoxi-
dized (intact purified HSA which mainly exists in reduced form), mildly
oxidized and highly oxidized HSA. Using these standards, we demonstrated
several differences in functional properties of HSA including protease
susceptibility, ligand-binding affinity and antioxidant activity. From these
observations, we conclude that an increased level of oxidized HSA may
impair HSA function in a number of pathological conditions.
Abbreviations
HNA, nonmercarptoalbumin; HMA, mercarptoalbumin; HSA, human serum albumin; LC-ESI-TOFMS, liquid chromatography-electron spray
ionization-time of flight.
3346 FEBS Journal 273 (2006) 3346–3357 ª 2006 The Authors Journal compilation ª 2006 FEBS

fHSA(red)% ¼½reduced HSA=ðreduced HSA
þ reversibly oxidized HSAÞ Â 100g
HSA(red)% tends to be lower in patients with various
diseases or conditions such as hepatic disease [7], dia-
betes [8], renal disease [9], temporomandibular joint
disorders [10], aging [11], and tiredness or fatigue [12].
Although a large number of clinical studies have
reported changes of HSA(red)% in various clinical
conditions, little is known regarding its pathophysio-
logical significance.
HSA has various functions, such as: (a) maintenance
of colloid osmotic pressure; (b) binding and transport
of a wide variety of metabolites including steroids,
fatty acids, bilirubin, tryptophan and hemin; (c) sup-
plying an amino acid source during times of malnutri-
tion; and (iv) acting as an antioxidant by radical
scavenging [13–19].
The goal of this study is to clarify the structure–
function relationship between reduced and oxidized
HSA in healthy human plasma, and to assess the path-
ophysiological significance of change in HSA(red)%.
In this report, we determined an exact adduct bound
to Cys34 residue of oxidized HSA using mass spectro-
metry in order to prepare both reduced and oxidized
HSA as standard samples for functional studies. Using
these, we evaluated several functional differences
between purified HSA samples with various states of
oxidation.
Results
Analysis of the structure of oxidized albumin

purified from human plasma using
LC-ESI-TOFMS
Structural heterogeneity exists in plasma HSA, which
is a mixture of reduced, oxidized and glycated albu-
min. We analyzed the purified HSA from healthy
human plasma by LC-ESI-TOFMS in order to deter-
mine structure based on mass information.
The positive ionized albumin was observed with ions
distributed from [M + 66H]
66+
to [M + 36H]
36+
.
Figure 1A shows the ESI mass spectrum of albumin
purified by affinity chromatography. Here, the eluted
fraction by affinity chromatography was defined as
purified albumin. In the mass range (m ⁄ z 1290–1320),
all observed peaks were [M + 51H]
51+
ions. These
peaks were named from lower mass in alphabetical
order: m ⁄ z 1300.09 (peak a), 1301.55 (peak b), 1303.75
(peak c), 1306.10 (peak d) and 1306.94 (peak e),
respectively. Because all observed peaks were
[M + 51H]
51+
ions, the deconvoluted molecular
weights were 66 253.6 D a (peak a), 66 328.1 Da (peak b),
66 440.3 Da (peak c), 66 560.1 Da (peak d), and
66 603.2 Da (peak e), respectively. In particular, peak

c was closest to the theoretical mass (66 437.2) calcula-
ted from the known primary amino acid sequence of
HSA after subtracting 34 Da due to 17 pairs of disul-
fide bonds. Therefore, we regarded peak c as reduced
HSA, and the others were due to post-translational
modification resulting in mass differences compared
with peak c. When plasma was incubated at 37 °C
to promote aerobic oxidation, we observed gradual
increase in the intensity of peak d, while the intensity
of peak c was reciprocally decreased. Thus, we regar-
ded peak d as oxidized HSA. From peak heights, we
estimated that HSA(red)% of healthy human plasma
pool is 78.5. Peak d was 119.8 Da heavier than
reduced HSA (peak c), corresponding to being the
Cys-adduct of HSA via an S–S bond. We speculated
peak a to be the N-terminal Asp-Ala truncated form.
We also suggested peak b to be the C-terminal Leu
truncated form and peak e to be glycated HSA.
Subsequently, we prepared highly oxidized HSA
with Cys (HSA-Cys) as a standard. Figure 1B shows
the ESI mass spectrum of HSA-Cys. Under the reac-
tion conditions, excess Cys ⁄ cystine solution was added
to purified HSA of healthy human plasma pool, whose
Intensity
a
b
c
d
e
d’

f
g
h
i
j
1290 1295 1300 1305 1310 13201315
m/
z
A
B
C
Fig. 1. [M + 51H]
51+
ion from ESI-TOFMS spectra of HSA. (A)
Spectrum of HSA from fresh plasma, purified using a HiTrap Blue
HP column. (B) Spectrum of excess Cys ⁄ cystine solution added to
purified HSA. (C) Spectrum of excess Hcy ⁄ homocystine solution
added to purified HSA. The ions correspond to the following: (a)
Asp-Ala truncation from N-terminal of HSA, (b) Leu truncation from
C-terminal of HSA, (c) HMA, (d) HSA-Cys, (d¢) the identical mass to
peak d, (e) glycated HMA, (f) sulfonation after the cleavage of a
disulfide bond in HSA-Cys, (g) glycated HSA-Cys, (h) HSA-Hcy, (i)
sulfonation after the cleavage of a disulfide bond in HSA-Hcy, and
(j) glycated HSA-Hcy.
A. Kawakami et al. Characterization of oxidized human serum albumin
FEBS Journal 273 (2006) 3346–3357 ª 2006 The Authors Journal compilation ª 2006 FEBS 3347
HSA(red)% value was originally 78.5. After removing
excess Cys ⁄ cystine with a low molecular weight cut-off
ultrafilter membrane, the sample was applied to
LC-ESI-TOFMS in order to determine the structure

and purity of the HSA-Cys. In this mass spectrum,
although the molecular-related ion of reduced HSA
was hardly observed, peak d¢ showed the most signifi-
cant intensity in the range of m ⁄ z 1290–1320. The m ⁄ z-
value of peak d¢ (Fig. 1B) was identical to peak d
(Fig. 1A). The HSA(red)% of HSA-Cys was only 5%,
therefore HSA-Cys accounted for 95% with the excep-
tion of the other peaks in the cysteinylated HSA sam-
ple solution. The difference of relative molecular mass
of the peak d and the peak g (162.2 Da) and that of
peak c and peak e (162.9 Da) was consistent within
experimental error. Therefore, peak g was probably
glycated HSA-Cys.
Although the structure of peak f was unknown, it
could be due to a partially cleaved and irreversibly
oxidized S–S bond resulting in sulfenic acid (-SO
3
H).
This is supported by the mass difference between peak
d¢ and peak f (98.0 Da) corresponding to the mass of
six oxygen atoms.
Figure 1C shows the ESI mass spectrum of the pre-
pared highly oxidized HSA with Hcy (HSA-Hcy), where
excess Hcy ⁄ homocystine had been added to purified
HSA from healthy human plasma pool (HSA(red)% ¼
78.5). After removing excess Hcy ⁄ homocystine, the sam-
ple was analyzed by LC-ESI-TOFMS, as noted above.
In this mass spectrum, the molecular-related ion of
reduced HSA was again hardly observed, and peak h
showed the most significant intensity in the range of m ⁄ z

1290–1320. The molecular weight difference of peaks c
and h was 132.3 Da, corresponding to Hcy being incor-
porated by an S–S bond. The HSA(red)% was only 9%,
therefore HSA-Hcy accounted for 91% with the excep-
tion of the other peaks in the homocysteinylated HSA
sample solution.
The difference of relative molecular masses of peaks
h and j (161.7 Da) and those of peaks c and e
(162.9 Da) was again within experimental error, sug-
gesting that peak j was glycated HSA-Hcy. Peak i cor-
responded to peak f, possibly due to sulfenic acid
formation, as described above. If the thiol group at
Cys34 of reduced HSA was sulfenized (-SO
3
H), the
difference in relative molecular mass against reduced
HSA will be 49.0 Da due to the addition of three oxy-
gen atoms. However, peaks with a relative molecular
mass difference of 49 Da were hardly observed on the
baseline level. Calculated molecular weight and the
predicted structure of HSA corresponding to each
peak observed in the LC-ESI-TOFMS measurements
were listed in Table 1.
The results of our ESI-TOFMS measurements
showed that oxidized HSA in healthy human plasma
showed mainly cysteine, and not homocysteine,
adducts.
The comparative FTMS measurement of peptide
mixture derived from highly oxidized HSA
standard (HSA-Cys) and purified HSA of healthy

human plasma
From the result of ESI-TOFMS, we deduced that the
main form of oxidized albumin in healthy human
plasma is a cysteine adduct on reduced HSA. In order
to prove exactly where cysteine bonds to reduced
HSA, we digested the standard highly oxidized HSA
[HSA-Cys; HSA(red)% ¼ 5% and HSA-Hcy;
HSA(red)% ¼ 9%] and the purified nonoxidized HSA
[HSA(red)% ¼ 78.5] from healthy plasma with Lys-C.
The digested peptides were analyzed using FTMS.
FTMS has high performance in high mass resolution
and accuracy. If the precise mass is known, the exact
composition formula of a low molecular weight com-
pound can be determined.
The peptide containing Cys34 generated by Lys-C
enzyme reaction is from Ala21 to Lys40 (ALVLIA-
FAQYLQQCPFEDHVK). When the peptide was
cysteinylated through a S–S bond binding at Cys34,
the mono-isotopic mass of the multiply-protonated
Table 1. Estimation of the various HSA structure based on LC-
ESI-TOFMS information. All observed masses between m ⁄ z 1290
and 1320 were [M + 51H]
51+
ions in LC-ESI-TOFMS spectrum.
Peak
Observed
mass
[M + 51H]
51+
Molecular

weight
Difference in
mass from
peak c
Estimated
structure
of HSA
a 1300.09 66253.6 )186.7 Deficient form
(N-terminal Asp-Ala)
b 1301.55 66328.1 )112.2 Deficient form
(C-terminal Leu)
c 1303.75 66440.3 0 Reduced HSA
d 1306.10 66560.1 119.8 HSA-Cys
e 1306.95 66603.2 162.9 Glycated HSA
f 1308.02 66658.1 217.8 A disulfide bond
cleavage of
HSA-Cys fi -SO
3
H
+HO
3
S-
g 1309.28 66722.3 282.0 Glycated HSA-Cys
h 1306.34 66572.3 132.0 HSA-Hcy
i 1308.26 66670.3 230.0 A disulfide bond
cleavage of
HSA-Hcy fi -SO
3
H
+HO

3
S-
j 1309.51 66734.0 293.7 Glycated HSA-Hcy
Characterization of oxidized human serum albumin A. Kawakami et al.
3348 FEBS Journal 273 (2006) 3346–3357 ª 2006 The Authors Journal compilation ª 2006 FEBS
molecule [M + nH]
n+
was theoretically calculated to
be 2552.2682 (1+), 1286.6380 (2+) and 851.4279
(3+), respectively, from peptide sequence information.
For the homocysteinylated peptide, these masses were
calculated to be 2566.2838 (1+), 1283.6458 (2+) and
856.0998 (3+).
Figure 2A,B shows FTMS spectra in the mass range
m ⁄ z 851–858, which includes the peptides following
Lys-C digestion of the HSA-Hcy and HSA-Cys stand-
ards, respectively. Both of the multiply-charged ions
were 3+. The mono-isotopic ions were observed at
m ⁄ z 856.1109 (Fig. 2A) and 851.4382 (Fig. 2B),
respectively. These values were within experimental
error of the theoretical mono-isotopic values (m ⁄ z
856.0998, 851.4279). Accordingly, both HSA-Cys and
HSA-Hcy standards would be derived from a Cys and
a Hcy being incorporated into reduced HSA at Cys34
via an S–S bond, respectively.
Figure 2C shows the FTMS mass spectrum (m ⁄ z
851–858) of purified nonoxidized HSA using affinity
chromatography from healthy human plasma. A 3+-
charged mono-isotopic ion was observed at m ⁄ z
851.4365. This is consistent with that of the digested

peptide from HSA-Cys standard. Therefore, HSA-Cys
with cysteine incorporated at Cys34 via a disulfide
bond is the main form of purified HSA in healthy
human plasma.
S-Cysteinylation affects susceptibility of albumin
to trypsin digestion
After we had determined the modification on Cys34
residue of oxidized albumin, we next examined
whether this affects its susceptibility to proteolysis. We
compared the proteolytic sensitivity of purified healthy
plasma HSA [nonoxidized HSA, HSA(red)% ¼
73.0%] and highly oxidized HSA [HSA-Cys,
HSA(red)% ¼ 8%]. As shown in Fig. 3A, both nonox-
idized HSA and highly oxidized HSA (HSA-Cys)
degraded in a time-dependent manner, but showed dif-
ferent susceptibility to digestion by trypsin, with highly
oxidized HSA being degraded far faster than nonoxi-
dized HSA. Figure 3B is the quantified results of each
HSA band shown in Fig. 3A. After 8-h digestion, the
remaining highly oxidized HSA was approximately
one-half that of nonoxodized HSA. As proteolysis pro-
ceeded, new peptide bands appeared with smaller
molecular weights below HSA (indicated by an open
arrow in Fig. 3A). This suggests that HSA is degraded
specifically into certain large fragments.
To identify the specific enzymatic cleavage site in
HSA, we analyzed the N-terminal sequence of the
ALVLIAFQYLQQ
34
CPFEGHFEDVK

Hcy
ALVLIAFQYLQQ
34
CPFEGHFEDVK
Cys
A
856 . 4452
856 . 1109
856 . 7793
857 . 1134
856 . 4443
856 . 7782
856 . 1099
856 . 1083
856 . 4427
857 . 1117
851 . 7714
852 . 1052
851 . 4365
852 . 4404
852 . 1066
851 . 7724
851 . 4382
B
C
Intensity
854852 856 858 m/
z
Fig. 2. ESI-FTMS spectrum, identification of binding site of adduct
to albumin. Mass range displayed from m ⁄ z 850.5–858.5. (A) Spec-

trum of the Lys-C digested peptide containing Cys34 from
HSA-Hcy conjugate. (B) Spectrum of the Lys-C digested peptide
contains Cys34 from HSA-Cys conjugate. (C) Spectrum of the
Lys-C digested peptides from purified HSA.
A
N H N H N H N H N H
0
1
2
48
(h)
64
(kDa)
B
Non-oxidized HSA
Highly oxidized HSA
Undigested HSA (%)
0
20
40
60
80
100
120
02468
Non-oxidized HSA
Highly oxidized HSA
Trypsin treated time (h)
Fig. 3. Susceptibility of reduced HSA and S-cysteinylated HSA to
tryptic proteolysis. (A) SDS ⁄ PAGE of HSA after tryptic digestion.

HSA samples were treated as described in Experimental proce-
dures for the indicated times. Twelve micrograms of each protein
were loaded into each well and electrophoresis was performed
using a 12.5% polyacrylamide gel. The filled arrow indicates the un-
digested HSA and the open arrow indicates the major tryptic frag-
ment of HSA. N, nonoxidized HSA; H, highly oxidized HSA. (B)
Densitometric analysis of intensities of undigested HSA. Changes
in intensities of HSA bands relative to the band of starting point of
digestion are shown.
A. Kawakami et al. Characterization of oxidized human serum albumin
FEBS Journal 273 (2006) 3346–3357 ª 2006 The Authors Journal compilation ª 2006 FEBS 3349
digested peptide by Edman degradation. The N-ter-
minal sequence read Glu-Thr-Tyr-Gly, and concluded
that one of the target sites of tryptic digestion was
located between Arg81 and Glu82 (data not shown).
The binding properties of reduced HSA and
oxidized HSA are different
One significant functional role of serum albumin is lig-
and binding. HSA binds many endogenous and exo-
genous small molecular compounds, including l-trp,
fatty acids, bilirubin, and drugs; HSA also plays an
important role in delivering these compounds to target
tissues. Several specific binding sites of these ligands
on HSA have been identified, and the two major bind-
ing sites are designated as sites I and II [20]. The bind-
ing properties of these sites are strongly correlated to
the structure of HSA. As the structural change caused
by S-cysteinylation affected proteolytic susceptibility,
there is a possibility that the binding properties of
reduced and oxidized HSA may also be different.

Therefore, we investigated the relative binding proper-
ties of these two types of HSA. We investigated the
binding of L(small)-Trp as an endogenous ligand
which binds to site II, and of cefazolin (site I-ligand)
and verapamil (site I and II-ligand) as exogenous lig-
ands. All these ligands are known for their high bind-
ing efficiencies to HSA.
The binding affinity of each compound to purified
HSA was evaluated by ultrafiltration. The results are
expressed as unbound fraction (%) in Table 2. All the
values tended to be relatively high in our experiments
compared with their binding capacities to human
plasma in the literature for unknown reasons. When
we compared the unbound fractions of nonoxodized
HSA [HSA(red)% ¼ 73.0] with mildly oxidized HSA
[HSA(red)% ¼ 55.4], l-Trp bound less strongly to
mildly oxidized HSA. The same result was obtained
when cefazolin was used as a ligand. While l-Trp and
cefazolin showed decreased affinity to mildly oxidized
HSA, verapamil binding to mildly oxidized HSA was
found to be slightly increased. These results suggest
that reduced HSA and oxidized HSA have different
ligand-binding properties.
The antioxidant property of albumin is impaired
in oxidized HSA
HSA is the major antioxidant in blood due to its free
thiol at Cys34. In this study, we investigated the poten-
tial effect of oxidation on the antioxidant capacity of
HSA by comparing the radical scavenging activities of
HSA in various states of oxidation. The hydroxyl

radical scavenging activity of nonoxidized HSA and
highly oxidized HSA-Cys (10 mgÆmL
)1
) was studied
using ESR. The typical 1 : 2 : 2 : 1 four-peak ESR
spectrum of the hydroxyl radical was observed and is
shown in Fig. 4A. Addition of HSA caused a decrease
in the ESR signal intensities. While nonoxidized HSA
[HSA(red)% ¼ 73.0%] quenched up to 68.7% of the
hydroxyl radical signal, HSA-Cys [HSA(red)% ¼ 8%]
reduced it by only 54.4% compared with the control.
This suggests that the radical scavenging activity of
reduced HSA is greater than that of HSA-Cys. When
we used mildly oxidized HSA [HSA(red)% ¼ 54.4%],
the signal decreased by 62.3%. As shown in Fig. 4B,
the HSA(red)% and hydroxyl radical scavenging activ-
ities for each sample show a high positive correlation.
To eliminate the possibility that varying iron binding
affinity of HSA decreased radical generation, we gener-
ated the hydroxyl radical by a different reaction, UV
photolysis of H
2
O
2
. The same results were obtained,
showing that oxidized HSA had a decreased radical
scavenging activity (Fig. 4C). Therefore, we concluded
that oxidation of HSA reduced its radical scavenging
activity.
Discussion

Oxidation of HSA has been reported in numerous dis-
eases. Although oxidation has been suggested to be of
particular pathophysiological relevance for various
conditions, there is no direct proof that oxidation of
HSA leads to aberrant alterations in its structural con-
formation and its functional properties.
In this study, in order to clarify the pathological
consequences of oxidation, we identified an exact
adduct and position of the modification of oxidized
HSA from human plasma and characterized its speci-
fic functional properties by comparing among the
purified HSA samples which had distinct HSA(red)%
values.
To distinguish the different functional properties of
oxidized HSA, it was first necessary to prepare clearly
Table 2. Binding of L-Trp, cefazolin and verapamil to purified HSA
and oxidized HSA. Values are expressed in unbound fraction (%).
All experiments were performed in duplicate and each CV% was
less than 1%.
HSA sample
HSA
(red)%
Unbound fraction (%)
L-Trp Cefazolin Verapamil
Purified non-oxidized HSA 78.5 50.4 17.9 63.3
Mildly oxidized HSA 54.4 65.9 62.3 51.1
Characterization of oxidized human serum albumin A. Kawakami et al.
3350 FEBS Journal 273 (2006) 3346–3357 ª 2006 The Authors Journal compilation ª 2006 FEBS
defined and highly purified HSA samples. Initially, we
attempted to analyze the structure of native oxidized

HSA from healthy human plasma. Oxidized HSA has
generally been regarded to have various modifications
on free thiol group at Cys34. There are a number of
previous reports in which diverse attempts to identify
the actual structures of oxidized albumin in detail have
been made. Sugio et al. [21] determined the X-ray
structures of HSA, derived from human pooled plasma
or from a Pichia pastoris expression system, at a reso-
lution of 2.5 A
˚
. However, they were not able to
investigate in detail the final differences in the electron
density map around the sulfhydryl side chain of Cys34.
Therefore, X-ray analysis has not been able to precisely
observe the structure of oxidized HSA. Bar-Or et al.
[22] showed profiles of both typical commercial albu-
min preparations and normal healthy volunteer human
serum albumin by LC-ESI-TOFMS measurement.
Sengupta et al. [23] also showed that thiols (Cys and
Hcy) disulfide-bonded to albumin-Cys34 could be
removed by treatment with dithiothreitol to form albu-
min-Cys34-SH by ESI-TOFMS. In these studies, the
structure of oxidized albumin was determined by MS.
However, it is impossible to determine the exact bind-
ing site for thiols by measurement of the mass of the
whole albumin molecule. Additionally, Kleinova et al.
[24] showed that the structure of pharmaceutical-grade
HSA were mainly oxidized HSA. However, as intact
albumin from human plasma was not used in their
study, the true structure of physiologically oxidized

HSA was never proven. Moreover, reduced HSA was
indirectly measured after alkylation with 4-vinylpyri-
dine and subsequent tryptic digestion. Therefore, for
definitive analysis of albumin oxidation it is necessary
to show that the oxidized albumin has no thiol group
and has conjugated to cysteine, homocysteine or gluta-
thione through a disulfide bond at Cys34.
In this study, to solve the structure of physiological
oxidized HSA, we purified HSA from healthy human
plasma. We employed ESI-TOF-MS analysis of HSA
as a whole molecule and FTMS analysis of proteolytic
digests. By combining both results, we succeeded in
revealing that the majority of oxidized HSA in healthy
human plasma has only a single modification at
Cys34, which is due to a disulfide bond to cysteine.
After the determination of the molecular structure
of oxidized HSA, we subsequently prepared standard
forms of nonoxidized HSA and oxidized HSA, using
purified HSA from healthy human plasma. The identi-
ties of prepared HSA samples were confirmed by
LC-ESI-TOFMS (Fig. 1) and FTMS analysis (Fig. 2).
Therefore, well characterized and highly purified oxid-
ized HSA samples were used for functional analysis.
330.5 332.5 334.5 336.5 338.5 340.5
Field (mT)
A
NaCl/P
i
Nonoxidized HSA
radical scavenging activity (% of control)

0
10
20
30
40
50
60
70
Non-
oxidized
HSA
Highly
oxidized
HSA
C
B
20
30
40
50
60
70
80
0 20 40 60 80 100
radical scavenging activity (% of control)
HSA(red)%
Fig. 4. Scavenging effects of purified HSA with various HSA(red)%
on hydroxyl radical. (A) ESR spectra of the spin adducts of DMPO-
OH. Grey spectrum corresponds to the radical generated without
HSA and dotted spectrum corresponds to the one generated when

nonoxidized HSA was added to the reaction. (B) Radical scavenging
activities are plotted against HSA(red)% of the samples. (C) Radical
scavenging activities of nonoxidized and highly oxidized HSA
against the hydroxyl radical generated by UV photolysis. Values are
the mean ± SD (n ¼ 3, P < 0.005).
A. Kawakami et al. Characterization of oxidized human serum albumin
FEBS Journal 273 (2006) 3346–3357 ª 2006 The Authors Journal compilation ª 2006 FEBS 3351
In order to examine if the localized modification at
Cys34 of oxidized HSA causes an overall conforma-
tional changes in albumin, we investigated the differ-
ence in the susceptibility of the two types of HSA to
tryptic proteolysis. As shown in Fig. 3, highly S-
cysteinylated oxidized HSA showed increased suscepti-
bility to tryptic proteolysis. Consistent results were
reported by Glowacki et al. [25] in their study that
compared the proteolytic susceptibility of HSA-Cys
with dithiothreitol-treated, highly reduced HSA. These
findings suggest that HSA undergoes a conformational
change upon S-cysteinylation, thereby making the clea-
vage sites more accessible. We identified the cleavage
site of tryptic digested HSA by Edman sequencing and
revealed that the region encompassing Glu82 in
domain I becomes more susceptible to proteolysis. To
gain further insight into the molecular basis for the
effects of oxidation on conformational change, molecu-
lar modeling of HSA-Cys was performed using the
crystal structure of human serum albumin (RSC PDB
ID: 1BM0). The expected modeled structures of HSA-
Cys were subjected to molecular dynamics simulations
using insightii (Accelrys Inc., San Diego, CA, USA).

The conditions of calculation were as follows: force
field ¼ discover3, run time ¼ 1000 fs, temperature ¼
298 K. These calculations showed that the conforma-
tional changes induced by S-cysteinylation of HSA
were not large scale, but were localized to five regions
of the molecule (indicated as orange ribbons in Fig. 5).
One of the five conformationally altered regions, a
loop between Thr79 and Leu85, which is located near
to the Cys34 residue in the three-dimensional structure,
may be the reason for increased susceptibility to pro-
teolysis, as it contains the specific site for tryptic diges-
tion, Glu82. Stewart et al. [2] recently analyzed X-ray
structures of recombinant HSA and showed that the
sulfur of Cys34 is tethered to the hydroxyl oxygen of
Tyr84 by a hydrogen bond. They speculated that the
formation of disulfide at Cys34 would lead to the loss
of the H-bond between Cys34 and Tyr84, thereby
resulting in a conformational change. In fact, using
1
H NMR study, they demonstrated that S-cysteinyla-
tion altered the conformation and dynamics of the
entire domain I, and also the domain I ⁄ II interface.
All these results suggest that oxidation, a single modifi-
cation at Cys34, could result in a number of regional
conformational changes of HSA resulting in increased
susceptibility to proteolysis.
The structure of the protein should be highly associ-
ated to its specific functional activities. We attempted
to investigate whether the conformational change of
oxidized HSA affects certain functional properties of

albumin. To elucidate the structure–function relation-
ship between reduced and oxidized HSA, we examined
two functional properties, ligand-binding properties
and antioxidant activities, using both purified nonoxi-
dized and oxidized HSA samples.
Ligand-binding properties are the most significant
functions of HSA [26]. HSA binds and transports
numerous endogenous and exogenous compounds, and
controls their solubility and toxicity in vivo. Among the
several endogenous ligands of albumin, we focused on
l-Trp, which circulates in plasma mostly bound to site
II on HSA [27]. Because l -Trp is the precursor of sero-
tonin, it is hypothesized that increased levels of free
l-Trp in plasma enhance serotonin synthesis and
release at the brain. Excessive serotonin secretion is
observed in many clinical conditions and modulates
numerous physiological and psychiatric systems. In this
study, highly oxidized HSA showed decreased l-Trp
binding. The binding site of l-Trp on HSA is reported
to be located at site II [28], which is distant from Cys34
in the three-dimensional structure. However, a spatial
correlation between these two regions has been implica-
ted in the study by Muscaritoli et al. [29]. In an in vitro
study, they reported that the level of free unbound
l-Trp increased in the presence of cisplatin, an anti-
neoplastic drug that binds to HSA at Cys34. They
suggested that cisplatin administration caused l-Trp
Fig. 5. Molecular dynamics simulation of reduced HSA and HSA-
Cys. Blue ribbon represents reduced form of HSA and green ribbon
represents S-cysteinylated oxidized form. Five regions of HSA-Cys,

highlighted by orange, were conformationally changed from
reduced HSA. The Cys34 residue is shown in pink.
Characterization of oxidized human serum albumin A. Kawakami et al.
3352 FEBS Journal 273 (2006) 3346–3357 ª 2006 The Authors Journal compilation ª 2006 FEBS
displacement from HSA and enhanced precursor avail-
ability for serotonin synthesis and release at the brain,
and that might contribute to the pathogenesis of cisp-
latin-induced emesis. These findings indicate that the
state of Cys34 of HSA affects the l-Trp binding affinity
on the other side of the molecule. As shown in Table 2,
our study also showed that S-cysteinylation at Cys34
decreased binding affinity towards l-Trp of HSA. This
might also be related to increased levels of serotonin
and the frequent occurrence of some adverse complica-
tions of diseases which have an increased level of oxid-
ized HSA. On the other hand, exogenous ligands of
albumin such as cefazolin and verapamil also showed
different binding affinities between nonoxidized HSA
and mildly oxidized HSA. Mera et al. [30] recently
demonstrated that purified albumin from hemodialysis
patients with decreased HSA(red)% showed reduced
drug-binding properties to warfarin and ketoprofen. In
our experiment, mildly oxidized HSA displayed bidirec-
tional changes in binding properties dependent on dif-
ferent types of ligands. These observations indicate that
structural changes of HSA caused by oxidative modifi-
cation on the thiol moiety of Cys34 affect the drug-
binding properties of this protein. These phenomena
are important from a therapeutic point of view, as the
concentration of unbound free drugs in plasma has an

impact on pharmacokinetics, monitoring efficacy and
adverse effects. These factors are essential in specifying
the patient’s therapeutic regimen. Altered steady-state
plasma concentration of drugs is a clinical therapeutic
problem for the treatment of various inflammatory dis-
eases, especially in elderly patients. It is assumed that
reduced HSA(red)%, together with hypoalbuminemia,
often found in these patients is responsible for this
problem.
Antioxidant activity is also an important function of
albumin. The function is believed to be ascribed to its
single exposed thiol group at Cys34 [31–33]. Because
albumin accounts for most of the total plasma thiol
content (about 80%), it can act as a major antioxidant
in plasma or extracellular fluids where the amounts of
antioxidant enzymes are relatively small [34,35]. In the
former article, Mera et al. [30] reported that purified
albumin from hemodyalysis patients showed a
decreased ability to scavenge chemical synthetic DPPH
radicals. In this study, we also demonstrated that
oxidized HSA has reduced scavenging ability against
highly reactive oxygen species, in this case, hydroxyl
radicals (Fig. 4C). It is found that the degree of
hydroxyl radical scavenging activity of HSA is highly
correlated with HSA(red)% (Fig. 4B)
.
This observa-
tion confirms that the antioxidant activity of HSA, at
least in part, depends on the state of the thiol at
Cys34. As oxidized HSA has decreased antioxidant

activity, decreased HSA(red)% not only reflects the
oxidative shift of the redox state of the human body,
but also may be a factor influencing the redox state of
a number of diseases.
In conclusion, the present study demonstrated that
oxidized HSA, primarily cysteinylated via a disulfide
bond at Cys34, exhibits various differences in its biolo-
gical properties relative to reduced HSA. Although it
is still unknown whether highly oxidized HSAs from
patients with a number of diseases have a similar
structure, we suggest that reduced HSA(red)% may
result in impaired function of HSA. We suggest that
there may be potential diagnostic and therapeutic
benefits of measuring HSA(red)% in a variety of dis-
ease conditions.
Experimental procedures
Plasma collection
Blood (160 mL) was collected using a vacuum tube collec-
tion system, using heparin as an anticoagulant, from two
healthy volunteers. Plasma fractions of the two volunteers
were isolated by centrifugation at 4 °C for 20 min (2000 g),
and were mixed together. The pooled plasma, which is used
as healthy human plasma in this study, was immediately
frozen in liquid nitrogen prior to long-term storage at
)80 °C.
Preparation of purified HSA with various states
of oxidation
We prepared purified albumin samples with various
HSA(red)% from healthy human plasma.
Non-oxidized HSA

Just after plasma collection, the intact purified albumin pre-
pared by affinity chromatography using HiTrap
TM
Blue HP
Column (Amersham Bioscience) was designated as ‘nonoxi-
dized albumin’. HSA(red)% of this sample was high as
78.5%. For functional assays (ligand-binding and antioxid-
ant assay), the purified nonoxidized HSA was concentrated
by ultrafiltration, followed by incubation at 37 °C for 48 h.
After the treatments, HSA(red)% of the purified nonoxi-
dized HSA slightly changed to 73.0.
Mildly oxidized HSA
Mildly oxidized HSA was purified from the plasma which
was incubated at 37 °C for 18 h to promote aerobic oxida-
tion. The HSA(red)% value of this sample was 54.4%.
A. Kawakami et al. Characterization of oxidized human serum albumin
FEBS Journal 273 (2006) 3346–3357 ª 2006 The Authors Journal compilation ª 2006 FEBS 3353
Highly oxidized HSA
Highly oxidized albumin was prepared by artificial incor-
poration of cysteine or homocysteine into reduced albumin.
Cysteinylated HSA (HSA-Cys) and homocysteinylated
HSA (HSA-Hcy) were prepared as follows. Purified reduced
HSA at a concentration of 4 mgÆmL
)1
(0.06 mm) was trea-
ted with a 50-fold molar excess of l-cysteine ⁄ cystine by
mixing 80 mL of 4 mgÆmL
)1
HSA, 72 mL of 3 mm cys-
teine, and 8 mL of 3 mm cystine. All solutions were in

0.1 m calcium carbonate ⁄ hydrogen carbonate buffer
(pH 10.0), and the mixture was incubated at 37 °C for
48 h. Next, low molecular weight compounds were removed
by ultrafiltration through an Ultrafree-3000 Da membrane
(Millipore, Billerica, MA, USA) at 4 °C. The HSA(red)%
value of this sample was 5–9%. HSA-Hcy was prepared by
exactly the same procedure except that dl-homocysteine
was used in place of l-cysteine, and homocystine was used
in place of cystine. LC-ESI-TOFMS was used to determine
the purity of HSA-Cys and HSA-Hcy. The additive reac-
tion resulted in high yield and HSA(red)% value of each
sample was <10%. We also confirmed the remaining l-cys-
teine ⁄ cystine or homocysteine ⁄ homocystine in the standard
solutions were minute using LC-MS ⁄ MS measurement.
LC-ESI-TOFMS measurement
Each plasma and albumin preparation was analyzed by
HPLC (LC-Packings, Amsterdam, Netherlands) coupled to
ESI-TOFMS (Bruker Daltonics Inc., Billerica, MA, USA).
Plasma or purified HSA samples was diluted to
0.4 mgÆmL
)1
(0.06 mm). The diluted sample was filtered
with a 0.1 lm cut-off membrane filter microfilter tube. Sub-
sequently, aliquots of 100 lL of filtered sample were trans-
ferred to a sample vial. Two microliters of each sample was
injected into the pre-column (100 mm · i.d. 200 lm, length
packed with monolith C18; GL Science Inc., Tokyo,
Japan). Albumin was eluted using a 20-min linear gradient
method by 75 : 25 water ⁄ acetonitrile containing 0.1% for-
mic acid (solution A) to 10 : 90 water ⁄ acetonitrile contain-

ing 0.1% formic acid (solution B). All albumin samples
were desalted and concentrated by a column switching
method.
The eluted albumin from the main column was intro-
duced into ESI-TOFMS (microTOFÒ; Bruker Daltonics
Inc.). In all experiments, spectra were acquired over the
range m ⁄ z 50–3000. The observed mass spectrums were
averaged from 30 scans of the albumin ion. The averaged
data were smoothed using a Gauss algorithm and baseline
subtracted using the microtofÒ software.
To determine HSA(red)% of the samples, we focused on
the mass range of m ⁄ z 1250–1480 to identify certain signa-
ture peaks. Eight pairs (reduced and oxidized) of albumin
ions charged from 47+ to 54+ were obtained in this
range. HSA(red)% was calculated independently for each
identically charged ion using the following formula:
HSA(red)% ¼ {(peak height of reduced HSA) ⁄ [(peak height
of reduced HSA) + (peak height of oxidized HSA)]} · 100
The average of the eight values was assumed as
HSA(red)% of the sample.
Enzymatic digestion for FTMS measurement
Lyophilized lysylendopeptidase (20 lg per vial; mass spectr-
ometry grade; Wako Pure Chemical Industries, Osaka,
Japan) was dissolved in 2 mL of 0.1 m ammonium acetate
buffer (pH 6.0). This lysylendopeptidase solution was
diluted 2.4 times with 0.1 m ammonium acetate buffer
(pH 6.0). Purified fresh nonoxidized HSA and highly oxid-
ized HSA-Cys solution buffers were changed to 0.1 m
ammonium acetate by membrane filtration using 3000-Da
cut-off filters. One hundred microliters of 30 lm purified

nonoxidized HSA or HSA-Cys standards were mixed with
100 lL of 0.15 mm lysylendopeptidase, and incubated at
37 °C for 8 h. The enzymatic digestion was inactivated with
the 200 lL of 0.1% formic acid.
Digested peptides measurement by FTMS
We used a Bruker Daltonics Apex II FT-MS equipped with
a Bruker-Magnex actively shielded superconducting magnet
operating at 7.0 T and a nano ESI Sources with Microme-
talTip stainless emitter (Eisho-metal Co. Ltd, Tokyo Japan).
The digested sample was diluted 10-fold with 0.1% formic
acid in 50% acetonitrile ⁄ water (v ⁄ v). For syringe infusion
experiments, the digested albumin samples were individually
100-fold diluted and infused into the ESI source at a flow
rate of 200 nLÆmin
)1
. MS data were acquired in the positive
ionized mode over an m ⁄ z range of 400–2000. Data acquisi-
tion was performed using with the xmass software (Bruker
Daltonics). We focused on the peptide containing Cys34 in
the peptide mixture produced by enzyme digestion.
Tryptic digestion of HSA
Purified nonoxodized HSA (300 lg) and highly oxidized
HSA (HSA-Cys) (2 mgÆmL
)1
) were submitted to trypsin
digestion by incubating the samples at the final substrate to
trypsin (86 units, Wako) ratio of 25 : 1 in NaCl ⁄ P
i
at 37 °C
for 8 h. Aliquots from each digested sample were collected

at 0, 0.5, 1, 2, 4, and 8 h after the beginning of proteolysis.
Reactants were mixed with equal volumes of
2 · SDS ⁄ PAGE sample buffer and denatured for 10 min at
95 °C. Twelve micrograms of HSA per lane were subjected
to SDS ⁄ PAGE on 12.5% polyacrylamide gels for 30 min at
constant voltage (200 V). Protein bands were visualized by
Coomasie brilliant blue staining. The gel image was captured
using ImageMaster LabScan 5.0 (Amersham Biosciences,
GE Healthcare, Piscataway, NJ, USA) and protein bands
Characterization of oxidized human serum albumin A. Kawakami et al.
3354 FEBS Journal 273 (2006) 3346–3357 ª 2006 The Authors Journal compilation ª 2006 FEBS
corresponding to HSA were quantified using imagequant
5.1 software (Molecular Dynamics, Sunnyvale, CA, USA).
Ligand binding experiments
L-Trp binding property
l-Trp was added to purified nonoxidized HSA
(30 mgÆmL
)1
) and highly oxidized HSA (HSA-Cys,
30 mgÆmL
)1
) in NaCl ⁄ P
i
(pH 7.3) to give a final concentra-
tion of 100 lm. The unbound ligand fractions were separ-
ated using ultrafiltration membranes in Centricon YM100
devices (Amicon ⁄ Millipore Inc., Bedford, MA, USA) by
centrifugation (3000 g, 5 min, 37 °C). Fifty-microliter aliqu-
ots of the initial (before ultrafiltration), filtrate, and apical
(not ultrafiltered fraction) fractions were mixed with the

same amount of 8% (w ⁄ v) trichloroacetic acid solution,
respectively. Protein-free supernatants were separated by
centrifugation at 3000 g for 5 min. l-Trp concentration of
each fraction was determined by LC-MS analysis.
Drug-binding properties
The binding of cefazolin (5 lm) and verapamil (5 lm)to
HSA (30 mgÆmL
)1
) with two oxidative states in NaCl ⁄ P
i
(pH 7.3, 37 °C) was examined. The ultrafiltration method
was the same as in the l-Trp binding experiment. Fifty-
microliter aliquots of each fraction was mixed with 300 lL
acetonitrile and left at room temperature for 5 min. Pro-
tein-free supernatants were obtained by centrifugation at
3000 g for 5 min and the solvents were removed by evapor-
ation. Samples were dissolved in the eluent for the follow-
ing LC-MS analysis. The concentration of each compound
was determined according to the standard curve made by
LC-MS analysis.
Calculation of the unbound fraction (%)
The unbound fraction (%) of each ligand was calculated as
follows:
Unbound fraction (%) ¼ [ligand concentration in filtrate ⁄
the initial ligand concentration (before ultrafiltration)] · 100
Recovery was validated by the following calculation:
Recovery (%) ¼ {[(ligand concentration in filtrate) ·
50+ (ligand concentration in apical fraction · 400)] ⁄ (the
initial ligand concentration · 450)} · 100
ESR spectroscopy

ESR spectra were obtained using a JES-FR80S spectro-
meter (JEOL, Tokyo, Japan) with a manganese marker at
room temperature. Hydroxyl radicals were generated by the
two following different methods and trapped by
5,5-dimethyl-1-pyrroline-N-oxide (DMPO) (Sigma-Aldrich,
St Louis, MO, USA).
Fenton–reaction
A mixture of 5 mm DMPO, 0.025 mm FeSO
4
, 0.1 mm
diethylenetriaminepentaacetic acid (DETAPAC), 0.5 mm
H
2
O
2
, and purified HSA samples with various states of oxi-
dation (10 ngÆmL
)1
) in NaCl ⁄ P
i
(pH 7.3) was transferred to
a quartz flat cell and placed in the cavity of the ESR spectro-
meter for the measurement. The conditions of ESR measure-
ment were as follows: center field 335.5 mT, microwave
power 1 mW, modulation frequency 100 Hz, modulation
width 0.1 mT, receiver gain 79, sweep width 10 mT, time
constant 0.03 s, and sweep time 1 min.
UV photolysis
A mixture of 88 mm DMPO and 0.3% H
2

O
2
in NaCl ⁄ P
i
(pH 7.3) with or without HSA samples (10 ngÆmL
)1
) was
transferred to a quartz flat cell and irradiated for 8 s under
UV lamp at 230–430 nm in the cavity of the ESR spectro-
meter. The following measurement was conducted under
the conditions as follows: center field 335.5 mT, microwave
power 3 mW, modulation frequency 100 Hz, modulation
width 0.063 mT, receiver gain 250, sweep width 5 mT, time
constant 0.03 s, and sweep time 1 min.
Calculation of the radical scavenging activity (%)
The scavenging effects of HSA on hydroxyl radicals were
determined by the following calculation:
Radical scavenging activity (%) ¼ [(h
PBS )
h
HSA
) ⁄
h
PBS
] · 100%
Here, h
PBS
and h
HSA
are the ESR signal intensities without

and with HSA, respectively. The intensities of the signals
were normalized to that of manganese as an internal control.
Acknowledgements
We would like to give our special thanks to Dr Seiichi
Era, Department of Physiology and Biophysics, Gifu
University Graduate School of Medicine, and Drs
Hisataka Moriwaki and Hideki Fukushima, the First
Department of Internal Medicine, Gifu University
Graduate School of Medicine, for numerous discus-
sions and helpful suggestions. We are grateful to Dr
Masaichi-Chang-il Lee, Clinical Care Medicine Divi-
sion of Pharmacology, Kanagawa Dental College, for
valuable advice on ESR analysis. We also thank Dr
Itsuya Tanabe and other researchers in Ajinomoto
Co., Inc., for technical advice and helpful discussions.
References
1 Peters TJ (1995) All about albumin: Biochemistry, Genet-
ics, and Medical Applications. Academic, New York.
A. Kawakami et al. Characterization of oxidized human serum albumin
FEBS Journal 273 (2006) 3346–3357 ª 2006 The Authors Journal compilation ª 2006 FEBS 3355
2 Stewart AJ, Blindauer CA, Berezenko S, Sleep D, Tooth
D & Sadler PJ (2005) Role of Tyr84 in controlling the
reactivity of Cys34 of human albumin. Febs J 272, 353–
362.
3 Sengupta S, Chen H, Togawa T, DiBello PM, Majors
AK, Budy B, Ketterer ME & Jacobsen DW (2001)
Albumin thiolate anion is an intermediate in the forma-
tion of albumin-S-S-homocysteine. J Biol Chem 276,
30111–30117.
4 Gabaldon M (2004) Oxidation of cysteine and homocy-

steine by bovine albumin. Arch Biochem Biophys 431,
178–188.
5 Carballal S, Radi R, Kirk MC, Barnes S, Freeman BA
& Alvarez B (2003) Sulfenic acid formation in human
serum albumin by hydrogen peroxide and peroxynitrite.
Biochemistry 42, 9906–9914.
6 Zhang H & Means GE (1996) S-nitrosation of serum
albumin: spectrophotometric determination of its nitro-
sation by simple S-nitrosothiols. Anal Biochem 237,
141–144.
7 Watanabe A, Matsuzaki S, Moriwaki H, Suzuki K &
Nishiguchi S (2004) Problems in serum albumin mea-
surement and clinical significance of albumin microhe-
terogeneity in cirrhotics. Nutrition 20, 351–357.
8 Suzuki E, Yasuda K, Takeda N, Sakata S, Era S,
Kuwata K, Sogami M & Miura K (1992) Increased
oxidized form of human serum albumin in patients
with diabetes mellitus. Diabetes Res Clin Pract 18,
153–158.
9 Soejima A, Matsuzawa N, Hayashi T, Kimura R,
Ootsuka T, Fukuoka K, Yamada A, Nagasawa T &
Era S (2004) Alteration of redox state of human serum
albumin before and after hemodialysis. Blood Purif 22,
525–529.
10 Tomida M, Ishimaru J, Hayashi T, Nakamura K,
Murayama K & Era S (2003) The redox states of serum
and synovial fluid of patients with temporomandibular
joint disorders. Jpn J Physiol 53, 351–355.
11 Era S, Kuwata K, Imai H, Nakamura K, Hayashi T &
Sogami M (1995) Age-related change in redox state

of human serum albumin. Biochim Biophys Acta 1247,
12–16.
12 Imai H, Hayashi T, Negawa T, Nakamura K, Tomida
M, Koda K, Tajima T, Koda Y, Suda K & Era S
(2002) Strenuous exercise-induced change in redox state
of human serum albumin during intensive kendo train-
ing. Jpn J Physiol 52, 135–140.
13 Andre C, Jacquot Y, Truong TT, Thomassin M, Robert
JF & Guillaume YC (2003) Analysis of the progesterone
displacement of its human serum albumin binding site
by beta-estradiol using biochromatographic approaches:
effect of two salt modifiers. J Chromatogr B Analyt
Technol Biomed Life Sci 796, 267–281.
14 Manni A, Pardridge WM, Cefalu W, Nisula BC, Bardin
CW, Santner SJ & Santen RJ (1985) Bioavailability of
albumin-bound testosterone. J Clin Endocrinol Metab
61, 705–710.
15 Burczynski FJ, Wang GQ & Hnatowich M (1995) Effect
of nitric oxide on albumin-palmitate binding. Biochem
Pharmacol 49, 91–96.
16 Bhattacharya AA, Grune T & Curry S (2000) Crystallo-
graphic analysis reveals common modes of binding of
medium and long-chain fatty acids to human serum
albumin. J Mol Biol 303, 721–732.
17 Pascolo L, Del Vecchio S, Koehler RK, Bayon JE,
Webster CC, Mukerjee P, Ostrow JD & Tiribelli C
(1996) Albumin binding of unconjugated [3H]bilirubin
and its uptake by rat liver basolateral plasma membrane
vesicles. Biochem J 316, 999–1004.
18 Eckenhoff RG, Petersen CE, Ha CE & Bhagavan NV

(2000) Inhaled anesthetic binding sites in human serum
albumin. J Biol Chem 275, 30439–30444.
19 Zunszain PA, Ghuman J, Komatsu T, Tsuchida E &
Curry S (2003) Crystal structural analysis of human
serum albumin complexed with hemin and fatty acid.
BMC Struct Biol 3, 6–14.
20 Sudlow G, Birkett DJ & Wade DN (1975) The charac-
terization of two specific drug binding sites on human
serum albumin. Mol Pharmacol 11, 824–832.
21 Sugio S, Kashima A, Mochizuki S, Noda M & Kobaya-
shi K (1999) Crystal structure of human serum albumin
at 2.5 A
˚
resolution. Protein Eng 12, 439–446.
22 Bar-Or D, Bar-Or R, Rael LT, Gardner DK, Slone DS
& Craun ML (2005) Heterogeneity and oxidation status
of commercial human albumin preparations in clinical
use. Crit Care Med 33, 1638–1641.
23 Sengupta S, Wehbe C, Majors AK, Ketterer ME, Di-
Bello PM & Jacobsen DW (2001) Relative roles of albu-
min and ceruloplasmin in the formation of
homocystine, homocysteine-cysteine-mixed disulfide, and
cystine in circulation. J Biol Chem 276, 46896–46904.
24 Kleinova M, Belgacem O, Pock K, Rizzi A, Bucha-
cher A & Allmaier G (2005) Characterization of
cysteinylation of pharmaceutical-grade human serum
albumin by electrospray ionization mass spectrometry
and low-energy collision-induced dissociation tandem
mass spectrometry. Rapid Commun Mass Spectrom 19,
2965–2973.

25 Glowacki R & Jakubowski H (2004) Cross-talk between
Cys34 and lysine residues in human serum albumin
revealed by N-homocysteinylation. J Biol Chem 279,
10864–10871.
26 Kragh-Hansen U, Chuang VT & Otagiri M (2002) Prac-
tical aspects of the ligand-binding and enzymatic prop-
erties of human serum albumin. Biol Pharm Bull 25,
695–704.
27 Sasaki E, Saito K, Ohta Y, Ishiguro I, Nagamura Y,
Shinohara R, Takahashi H & Tagaya O (1991) Specific
binding of l-tryptophan to serum albumin and its func-
tion in vivo. Adv Exp Medical Biol 294, 611–614.
Characterization of oxidized human serum albumin A. Kawakami et al.
3356 FEBS Journal 273 (2006) 3346–3357 ª 2006 The Authors Journal compilation ª 2006 FEBS
28 Yamasaki K, Kuga N, Takamura N, Furuya Y, Hidaka
M, Iwakiri T, Nishii R, Okumura M, Kodama H,
Kawai K et al. (2005) Inhibitory effects of amino-acid
fluids on drug binding to site II of human serum albu-
min in vitro. Biol Pharm Bull 28, 549–552.
29 Muscaritoli M, Peverini P, Cascino A, Cangiano C,
Fanfarillo F, Russo M, Fava A & Rossi Fanelli F
(1999) Effect of cisplatin and paclitaxel on plasma free
tryptophan levels: an in vitro study. Adv Exp Medical
Biol 467, 275–278.
30 Mera K, Anraku M, Kitamura K, Nakajou K,
Maruyama T & Otagiri M (2005) The structure and
function of oxidized albumin in hemodialysis patients:
Its role in elevated oxidative stress via neutrophil burst.
Biochem Biophys Res Commun 334, 1322–1328.
31 Halliwell B (1988) Albumin – an important extracellular

antioxidant? Biochem Pharmacol 37, 569–571.
32 Halliwell B & Gutteridge JM (1990) The antioxidants of
human extracellular fluids. Arch Biochem Biophys 280,
1–8.
33 Cha MK & Kim IH (1996) Glutathione-linked thiol
peroxidase activity of human serum albumin: a possible
antioxidant role of serum albumin in blood plasma.
Biochem Biophys Res Commun 222, 619–625.
34 Soriani M, Pietraforte D & Minetti M (1994) Antioxi-
dant potential of anaerobic human plasma: role of
serum albumin and thiols as scavengers of carbon radi-
cals. Arch Biochem Biophys 312, 180–188.
35 Quinlan GJ, Mumby S, Martin GS, Bernard GR,
Gutteridge JM & Evans TW (2004) Albumin influences
total plasma antioxidant capacity favorably in patients
with acute lung injury. Crit Care Med 32, 755–759.
A. Kawakami et al. Characterization of oxidized human serum albumin
FEBS Journal 273 (2006) 3346–3357 ª 2006 The Authors Journal compilation ª 2006 FEBS 3357