Báo cáo khoa học: Role of Tyr84 in controlling the reactivity of Cys34 of human albumin potx

Role of Tyr84 in controlling the reactivity of Cys34
of human albumin
Alan J. Stewart
1
, Claudia A. Blindauer
1
, Stephen Berezenko
2
, Darrell Sleep
2
, David Tooth
2
and
Peter J. Sadler
1
1 School of Chemistry, University of Edinburgh, Edinburgh, UK
2 Delta Biotechnology Ltd, Nottingham, UK
Human albumin (66.5 kDa), a single-chain of 585
amino acids, is the most abundant protein in
blood plasma, typically present at concentrations of
 0.6 mm [1]. It consists of three structurally homolog-
ous, largely helical (67%) domains (I, II and III), each
consisting of two subdomains, A and B [2–4]. Like
other mammalian albumins, human albumin contains
17 disulfide bridges and a free thiol at Cys34, which
provides the largest fraction of free thiol in blood
serum. Cys34 is completely conserved within mamma-
lian albumins. In plasma, about 30% of the Cys34
thiol is blocked by disulfide bond formation with cys-
teine, homocysteine, or glutathione. Moreover, prepa-
rations of albumin usually contain 5–10% of dimeric

species, with Cys34 being a possible site of dimeriza-
tion [1]. Thus, the state of Cys34 is an important ori-
gin for heterogeneity in albumin. There has been much
interest in the Cys34 site, because not only does it act
as a physiological antioxidant [5,6], but also as a bind-
ing site for a wide variety of biologically and clinically
important small molecules, such as gold(I) antiarthritic
drugs [7,8], platinum(II) anticancer drugs [9,10], mer-
curials [11], as well as a variety of drugs which bind as
mixed disulfides, including captopril (an antihyperten-
sive) [12] and disulfiram (an alcohol-abuse drug) [13].
Importantly, 82% of nitric oxide in blood ( 7 lm)is
transported as an S-nitrosothiol at Cys34 [14]. Modifi-
cations of Cys34 are known to have allosteric effects
upon the reactivity and binding properties of other
sites on albumin. For example, nitrosylation of Cys34
decreases the binding affinity of albumin toward Cu
2+
ions, phenolsulfophthalein and palmitic acid [15,16],
whilst oxidation of Cys34 with cystine results in faster
N-homocysteinylation at Lys525 [17].
In order to maintain Cys34 in a reduced state in
the majority of albumin molecules in blood, and yet
Keywords
Cys34; disulfide interchange; human
albumin; NMR; thiol
Correspondence
P. J. Sadler, School of Chemistry, University
of Edinburgh, West Mains Road, Edinburgh,
EH9 3JJ, UK

Fax: +44 131650 6453
Tel: +44 131650 4729
E-mail:
(Received 21 September 2004, revised 5
November 2004, accepted 10 November
2004)
doi:10.1111/j.1742-4658.2004.04474.x
Cys34 in domain I of the three-domain serum protein albumin is the bind-
ing site for a wide variety of biologically and clinically important small
molecules, provides antioxidant activity, and constitutes the largest portion
of free thiol in blood. Analysis of X-ray structures of albumin reveals that
the loop containing Tyr84 occurs in multiple conformations. In structures
where the loop is well defined, there appears to be an H-bond between the
OH of Tyr84 and the sulfur of Cys34. We show that the reaction of 5,5¢-di-
thiobis(2-nitrobenzoic acid) (DTNB) with Tyr84Phe mutant albumin is
approximately four times faster than with the wild-type protein between
pH 6 and pH 8. In contrast, the His39Leu mutant reacts with DTNB more
slowly than the wild-type protein at pH < 8, but at a similar rate at pH 8.
Above pH 8 there is a dramatic increase in reactivity for the Tyr84Phe
mutant. We also report
1
H NMR studies of disulfide interchange reac-
tions with cysteine. The tethering of the two loops containing Tyr84 and
Cys34 not only appears to control the redox potential and accessibility of
Cys34, but also triggers the transmission of information about the state
of Cys34 throughout domain I, and to the domainI ⁄ II interface.
Abbreviations
DTNB, 5,5¢-dithiobis(2-nitrobenzoic acid); rHA, recombinant human albumin; HSA, human serum albumin; TNB, 5-thio(2-nitrobenzoic acid);
amu, atomic mass units.
FEBS Journal 272 (2005) 353–362 ª 2004 FEBS 353

simultaneously allow albumin to act as an antioxidant,
the reactivity of Cys34 must be finely controlled by the
protein environment. Crystallographic studies of albu-
min show that Cys34 is buried in a shallow crevice,
 9.5 A
˚
deep (Fig. 1A,B). The sulfur atom of Cys34 is
close to three ionisable groups: the carboxylate group of
Asp38, the imidazole ring from His39, and the hydroxyl
group of Tyr84. This environment is likely to be respon-
sible for the unusual properties of Cys34. In order to
elucidate the factors that control the reactivity of Cys34,
we compared the redox activity of native recombinant
human albumin with that of two mutants, His39Leu
and Tyr84Phe. The latter mutation in particular has a
dramatic effect on the reactivity of Cys34. We also dem-
onstrate that disulfide bond formation with cysteine
at Cys34 leads to structural changes around distant
residues.
Results and Discussion
Reaction of recombinant human albumin
with DTNB
Disulfide interchange reactions are of importance for
the physiological function of albumin and can be
conveniently monitored by studies of reactions with
AB
C
D
Fig. 1. Structural features of the Cys34 site in human serum albumin. (A) Domain structure of human serum albumin and location of Cys34.
The coordinates used are those of pdb 1AO6 (albumin isolated from blood plasma). (B) Cys34 and Tyr84 are located in juxtaposed loops.

The yellow ball is the S of Cys34, a potential H-bond from Tyr84 is shown in green. Further residues mentioned in the text are also shown.
(C) Superposition of the Cys34 region in 21 published X-ray crystal structures of human albumin, showing Cys34, Asp38, His39 and Tyr84.
The overlay has been generated by aligning the backbone atoms of residues Cys34 and His39 in each of the structures, using Swiss pdb
viewer (v. 3.7). Structures in blue are fatty acid-free, and structures in green contain bound fatty acid. The structure in magenta (pdb 1UOR;
[2]) refers to wild-type fatty-acid-free albumin, but differs substantially from all other structures with respect to the position of the loop con-
taining Tyr84, which is more than 12 A
˚
away from Cys34. In three fatty acid-free structures (1E78 (no ligands), 1E7A (with bound propofol),
1E7B (with bound halothane)), the side chain of Tyr84 is not resolved, and in three further fatty acid-free structures (1HK1, 1HK2, 3HK3, with
bound thyroxine), the entire loop from Val77 to Ala88 is not resolved. (D) Schematic showing secondary structure elements and disulfide
connectivities around Cys34 and Tyr84. The location of His residues likely to be perturbed by reactions at Cys34 is also shown. Residues
are labelled in one-letter code, and lower-case ‘h’ signifies helices.
Redox activity of albumin Cys34 A. J. Stewart et al.
354 FEBS Journal 272 (2005) 353–362 ª 2004 FEBS
5,5¢-dithiobis(2-nitrobenzoic acid) (DTNB) or 2,2¢-
dithiodipyridine, which are known to react specifically
with free sulfhydryl groups [18]. We determined appar-
ent rate constants for the reactions of wild-type recom-
binant human albumin (rHA) and His39Leu and
Tyr84Phe rHA mutants with DTNB and their depen-
dences on pH. Mass spectrometric analysis of the
product from the reaction of rHA [experimentally
determined mass, 66 438 atomic mass units (amu); the-
oretical mass, 66 438 amu] with DTNB gave a peak
corresponding to 66 643 amu. This is consistent with
the formation of a 1 : 1 albumin ⁄ TNB mixed disulfide,
which has a theoretical mass of 66 645 amu. In order
to allow the determination of rates using conventional
absorption spectroscopy, the kinetic studies were car-
ried out at 283 K, and over the pH range 6.0–10.2,

under pseudo-first order conditions. As expected, the
rates are pH-dependent (Fig. 2). The reactivity of
Cys34 appears to rise sharply between pH 6 and 8 for
all these albumins. The variation of rate constant with
pH for wild-type albumin follows a similar trend to
that reported by Pedersen and Jacobsen [19], who
studied reactions of human serum albumin (HSA) with
2,2¢-dithiodipyridine over the pH range 3–9.
Between pH 6 and pH 8, the reaction is approxi-
mately four times faster for the Tyr84Phe mutant than
for the wild-type protein. In contrast, the His39Leu
mutant reacts with DTNB much more slowly than the
wild-type protein below pH 8, but at a similar rate at
pH 8. Between pH 8 and 8.5 there is a pronounced
increase in reactivity for each of the albumins; how-
ever, the effect is much more dramatic for the
Tyr84Phe mutant. Cys34 reactivity for wild-type albu-
min appears to plateau at around pH 8.4, whilst signi-
ficant increases in Cys34 reactivity were observed for
the His39Leu and Tyr84Phe mutants until the pH was
increased to > 9. At pH 10.2, the reaction with DTNB
was approximately three times faster for the His39Leu
mutant, and 170 times faster for the Tyr84Phe mutant
than for wild-type rHA.
It is generally thought that disulfide interchange
reactions proceed via a nucleophilic attack of the
deprotonated thiolate on the disulfide [20]. Therefore,
an increase in reaction rate with increasing pH is
expected. In simple cases, the pH dependence of the
reaction rate represents the titration curve for the pro-

tein thiol, but in the case of albumin other factors,
most notably conformational transitions, also play a
role. Nevertheless, from the point at which the plateau
in the pH profile is reached, an upper limit for the
thiol pK
a
can be estimated by subtracting 1.7 pH units,
based on the assumption that only deprotonated thio-
late is present in the plateau region. This suggests that
the thiol of Cys34 has a pK
a
lower than 6.7 for the
wild-type protein, a finding consistent with previously
suggested literature values of  7 [20]. Thus, the pK
a
of Cys34 in recombinant HSA is substantially lower
than the average value of 8–9, which is typical for a
free cysteine side chain. Abnormally low pK
a
values
have been reported for a number of proteins. In most
cases, electrostatic interactions with positively charged
side chains or the dipole of helices have been identified
as the basis for low thiol pK
a
values. In several cases
(e.g. papain [20], protein tyrosine phosphatases [21],
O
6
-alkylguanine-DNA alkyltransferases [22] and some

proteins of the thioredoxin family [23]) the low pK
a
is
due to the formation of an ion pair with protonated
histidine. The proximity to a His residue can lower the
pK
a
by up to 5 units. Our data show that the lowering
of the pK
a
of Cys34 can be ascribed partly to its proxi-
mity to His39 and Tyr84, as both mutants exhibit pK
a
values that are  0.5 pH units higher. Both Tyr84Phe
and His39Leu mutations involve replacement of a
polar residue by a nonpolar one. This is expected to
Fig. 2. Pseudo-first order rate constants for the reaction of DTNB
with rHA and mutant albumins over the pH range 6.0–10.2. The
acid dissociation constants of the carboxyl groups in DTNB are 1.57
and 2.15 [39] and the pK
a
of the thiol of TNB is 4.50 [36]. Therefore
the carboxyl groups of DTNB will be deprotonated and only the S
–
form of the TNB product is present over the pH range studied
(pH 6.0–10.2). Note that at pH 6.0 and 6.2 no detectable activity
was observed with the His39Leu mutant, these points are there-
fore not present on the graph due to the logarithmic scale.
A. J. Stewart et al. Redox activity of albumin Cys34
FEBS Journal 272 (2005) 353–362 ª 2004 FEBS 355

lead to an increase in thiol pK
a
, as charged species are
stabilized by a polar environment, and destabilized by
nonpolar surroundings.
Albumin is known to undergo a structural transition
above pH 8 from the N (neutral) to B (basic) form of
the protein [24]. It is thought that the N-to-B-trans-
ition involves a ‘loosening-up’ of the entire protein
structure [24], and that this transition might have a
role in the delivery of metabolites, e.g. to the liver.
The formation of the B-conformer is likely to account
for the steep rise in reactivity between pH 8 and 8.5.
The enhanced antioxidant activity of HSA at alkaline
pH has previously been attributed to the B conforma-
tion [25], and Cornell et al. [26] concluded that the
Cys34 crevice opens in the B conformation.
Our results regarding the Tyr84Phe mutant provide
a clue as to how this can be achieved. Apparently, the
removal of the hydroxyl group dramatically enhances
the redox activity of Cys34. A survey of the environ-
ment of Cys34 in all published X-ray crystal structures
of albumin (Fig. 1C) reveals that there is a consider-
able variability with respect to the position of Tyr84.
Remarkably, in a number of crystal structures of fatty
acid-free albumin (pdb codes 1E78, 1E7A, 1E7B,
1HK1, 1HK2, 1HK3), Tyr84 does not seem to be
resolved. Tyr84 is situated in a loop, and in the former
three structures, no electron density for the side chain
of Tyr84 is resolved, whereas in the latter three, the

entire loop from residue 77 to residue 88 is absent,
indicating conformational flexibility in this region. In
all but two of the structures in which Tyr84 is
resolved, however, the hydroxyl oxygen is within
hydrogen-bonding distance to the Cys34 sulfur (O–S
distances vary between 2.7 and 3.4 A
˚
). Tyr84 is thus
located on a flexible loop between two short helices,
and is tethered to the Cys34 site by a hydrogen bond
(Fig. 1B,D). This appears to be the only H-bond that
stabilizes the relative positions of the Tyr84 loop and
the Cys34 loop, although the Tyr84 loop itself is con-
nected via two disulfide bonds (C75 ⁄ 91 and C90 ⁄ 101)
to helices 4 and 5 (Fig. 1D).
Such hydrogen bonding is expected to stabilize an
anionic thiolate side chain of Cys34, similar to the way
in which glutathione S-transferases acidify the SH
group of glutathione by the formation of a hydrogen
bond with an active site Tyr OH group [27]. Also such
an environment is likely to stabilize the reduced thiol
form [28]. The tethering of the loop makes Cys34 less
accessible than it would otherwise be. Thus, the role of
Tyr84 is threefold: it enhances the nucleophilicity of
the sulfur of Cys34, contributes to its relatively high
redox potential, and simultaneously restricts access.
The stabilizing action of the Tyr84-OHÆÆÆS-Cys34
H-bond appears to be especially important when other
stabilizing H-bonds are cleaved, as is thought to occur
during the N-to-B transition.

Tyr84 is known to be the cleavage site for chymase,
a serine peptidase enzyme secreted by mast cells [29].
It is likely that chymase-cleaved albumin molecules in
the blood have altered Cys34 reactivity in vivo. At pre-
sent, the extent and consequences of chymase-cleavage
on albumin in the body is unknown. Several studies
suggest that chymase is also expressed in the liver,
where albumin is synthesized. Chymase expression in
the liver is increased in patients who suffer from auto-
immune disease [30].
The lowered reactivity of DTNB toward the
His39Leu mutant compared to wild-type rHA at pH
values < 8 indicates that His39 also has a role in con-
trolling the reactivity of Cys34. At low pH, His39
might increase reactivity by formation of an ion pair,
as noted for other proteins with reactive Cys residues
[28].
Reaction of recombinant serum albumins
with cystine
Previous
1
H NMR studies have shown that the His He1
resonance of His3 is sensitive to reactions at Cys34 [31].
In most albumin preparations, whether isolated from
plasma or produced by recombinant techniques, two
sets of peaks are observed for His3, with the lower inten-
sity set ( 30%) being assignable to albumin containing
oxidized (or blocked) Cys34 (Fig. 3, peak 11b; Fig. 4A,
His3¢).
The nature of the conformational change which

occurs at His3 on oxidizing or blocking Cys34 is not
clear, but in the case of reactions with antiarthritic
gold compounds or disulfide-bond forming drugs, it
has been suggested that Cys34 becomes more exposed
[31].
We acquired 1D and 2D NMR spectra of the
His39Leu, Tyr84Phe, and Cys34Ala mutants, to enable
identification of His39 and assess the folding behaviour
of the mutant proteins. Despite recent advances in pro-
tein NMR, the size of albumin and its dynamic prop-
erties do not allow a complete sequential assignment
of its resonances. However, by employing resolution
enhancement during data processing, it is possible to
resolve sharp resonances for slowly relaxing protons
[31]. We focus here on the He1 resonances of His side
chains, which can be conveniently observed in the low-
field region of spectra for samples in which backbone
NH protons have been exchanged with deuterium.
The low-field region of resolution-enhanced 1D
1
H
and 2D[
1
H,
1
H] TOCSY spectra of wild-type rHA, and
Redox activity of albumin Cys34 A. J. Stewart et al.
356 FEBS Journal 272 (2005) 353–362 ª 2004 FEBS
the mutants Cys34Ala rHA, His39Leu rHA, and
Tyr84Phe rHA are shown in supplementary Figs S1–

S4. Overall, the spectra of the wild-type and mutant
proteins are similar. Small differences in chemical
shifts are due mainly to small variations (< 0.1 pH
unit) in the pH* of the solutions. As the chosen pH* is
close to the pK
a
values of histidine residues, small
changes in pH have significant effects on chemical
shifts of histidine protons.
Human albumin contains 16 histidine residues. In
the spectra of the wild-type and Cys34Ala mutant pro-
tein, 11–12 major resonances from histidine He1 pro-
tons can be distinguished between about 7.8 and
8.5 p.p.m. With the aid of 2D TOCSY spectra (supple-
mentary Figs S1–S4), most of the corresponding Hd2
protons can be identified as well.
Comparison of the
1
H NMR spectra of the wild-
type and His39Leu rHA show that none of the sharp
He1 resonances has disappeared. However, the relat-
ively broad resonance at 8.115 p.p.m. (Fig. 3, 7) dis-
appears upon mutation of His39. Therefore resonance
7 can be assigned to the He1 proton of His39. Due to
relaxation phenomena, it is expected for a protein the
size of albumin that only flexible side chains give rise
to sharp resonances; protons of buried residues gener-
ally have broad lines. The appearance of the His39
peak suggests that it is buried to some extent, and
indeed, in the published X-ray crystal structures of

albumin [3,4] His39 lies slightly deeper in the crevice
that contains Cys34.
The 1D
1
H spectra of Tyr84Phe rHA contain
broader lines than those of wild-type rHA and the
other mutants, suggesting that this mutation changes
the dynamic behaviour of the protein. In contrast, the
His39 resonance in the spectrum of the Tyr84Phe
mutant appears slightly sharper than for wild-type
rHA, which is consistent with an increased mobility of
His39 in the mutant protein in the crevice holding
Cys34 and His39. This observation supports the idea
that the hydroxyl group of Tyr84 is essential for main-
taining a closed crevice.
The pair of peaks 11 and 11b (Fig. 3) has previously
been assigned to His3 [31]. In this context it is import-
ant to note that
1
H spectra of the Cys34Ala mutant
show only a single peak for He1 of His3 (see 2D
NMR data in supplementary Figs S1–S4).
The reaction of albumin with cystine, leading to
disulfide formation at Cys34, is of physiological
importance in blood. We monitored the reaction of
wild-type rHA with a 2.5-fold molar excess of l-cystine
by 1D
1
H NMR in 100 mm potassium phosphate buf-
fer (Fig. 4) at pH 7.0.

Due to the use of resolution enhancement, analysis
of the area of His peaks cannot be made in a quantita-
tive fashion, but qualitative observations have import-
ant implications. It appears that small local alterations
have a substantial effect on the entire domain I, and
even on the interface between domains I and II.
Our data reveal that not only the He1 resonance of
His3, but up to six other His He1 resonances, are
affected by the disulfide interchange reaction at Cys34.
Fig. 4A shows that the intensity of His3¢ resonance
increases during the reaction (any decrease in intensity
of peak H3 cannot directly be observed as it is over-
lapped by peak 12). The effects on the pair of peaks
6 ⁄ 6b are more readily seen: peak 6 decreases, and peak
6b increases in intensity. As for His3, there is only a
single He1 resonance (peak 6) for this histidine residue
in the Cys34Ala mutant (Fig. 3).
Notably, the low-field He1 peaks 1, 2, 3 and 4 are
also perturbed during the reaction. Peak 1 gradu-
ally decreased in intensity over time, with no new
resonance being detectable. Peak 4 also decreased, and
a new resonance appeared (4b) next to it. A minor
Fig. 3. 1D
1
H NMR spectra of recombinant albumins. Solutions of
respective albumins (1 m
M) were prepared in 50 mM Tris ⁄ DCl,
50 m
M NaCl, pH* 7.33–7.35. He1 proton resonances are labelled
with numbers from 1 to 12, f denotes formate, added as a chem-

ical shift reference. Peaks 7 and 11 are assigned to His39 and
His3, respectively. The arrows indicate resonances that are absent
in the respective spectra, e.g. the resonance 6b is not observed in
the Cys34Ala mutant, suggesting that for the wild-type protein it
pertains to a second conformation of His proton 6 in a species with
blocked or oxidized Cys34.
A. J. Stewart et al. Redox activity of albumin Cys34
FEBS Journal 272 (2005) 353–362 ª 2004 FEBS 357
decrease in intensity was also noticeable for peaks 2
and 3. Finally, the resonance for His39 He1 became
somewhat sharper during the reaction. Similar obser-
vations were made for the Tyr84Phe mutant (Fig. 4B).
Two of these peaks (1 and 4) can be assigned to His67
and His247 (CA Blindauer, KE Bunyan, AJ Stewart,
D Sleep, S Berezenko & PJ Sadler, unpublished results).
In a typical crystal structure of fatty acid-free albu-
min (pdb: 1AO6), the His39 He1 proton is within 4 A
˚
of the Cys34 sulfur; thus it is not surprising that a
reaction at Cys34 affects this proton. However, the dis-
tances between Cys34 and the next nearest six His resi-
dues in albumin are between 16 A
˚
(His67 and His105)
and 29 A
˚
(His9). His146 and His247 are about 20 and
22 A
˚
away, and His242 is 26 A

˚
from Cys34. The dis-
tance between His3 and Cys34 in (fatty acid loaded)
albumin (pdb: 1BJ5; His3 is not resolved in structures
of fatty acid-free albumin) is over 30 A
˚
. Clearly, reac-
tions at Cys34 have far-reaching allosteric effects, but
how are these effects transmitted throughout domain I
and beyond? The largely helical structure of albumin
makes it an extremely flexible molecule; the interhelix
contacts mainly being stabilized by disulfide bonds and
weak interactions. We can speculate that the formation
of a Cys disulfide in the Cys34-containing crevice will
have a major effect on the Tyr84 containing loop, and
is likely to lead to the loss of the H-bond between
Cys34 and Tyr84. From Fig. 2D it is evident that
movement of the Tyr84 loop will affect the two helices
(4 and 5) to which it is connected. His67 is at the start
of helix 4, and Asn99 at the start of helix 5. His247 is
in close proximity to His67, as both residues are
connected via hydrogen bonds to Asp249; these three
residues form, together with Asn99, the interdomain
high-affinity zinc site on albumin [32]. In this way, the
interface between domains I and II is affected by reac-
tions at Cys34. His105, which is situated in the loop
following helix 5, is also likely to be influenced by
movements of the Tyr84 loop.
The three domains of mammalian albumins are
structurally homologous, including the arrangement of

disulfide pairs. Interestingly however, the disulfide
bonding pattern for the first 50 residues in domain I
differs from that of the other two domains. The first
two helices in the albumin sequence in domain I are
not tethered by disulfide bridges. Cys34 is located in
the loop after helix 2, thus a structural change in this
loop is also likely to alter the orientation of these two
helices with respect to the entire domain. This assump-
tion could account for the effects observed for His3,
but also implies that His9 might be affected in a sim-
ilar manner. In this context it is noteworthy that resi-
due 35 is proline; and it has previously been suggested
that the structural change occurring upon reactions at
Cys34 might involve a cis –trans isomerization of this
residue [31].
In conclusion, we have shown that both His39 and
Tyr84 influence the reactivity of Cys34, with Tyr84
Fig. 4.
1
H NMR studies of the reaction of wild-type and Tyr84Phe mutant rHA with L-cystine. Conditions: 1 mM respective albumin, pH*
7.0, 100 m
M potassium phosphate buffer, 2.5 molar excess of L-cystine, 310 K. (A) Recombinant wild-type albumin. (B) Tyr84Phe mutant
rHA.
Redox activity of albumin Cys34 A. J. Stewart et al.
358 FEBS Journal 272 (2005) 353–362 ª 2004 FEBS
playing a pivotal role in both lowering the pK
a
of the
thiol and contributing to its high redox potential. In
addition, the loop containing Tyr84 appears to be vital

in controlling accessibility to Cys34, and is tethered to
the Cys34-containing loop by a Tyr84-OHÆÆÆS-Cys34
hydrogen bond. We also find that physiologically rele-
vant reactions at Cys34 such as disulfide bond forma-
tion with half cystine have an impact on the
conformation and dynamics of the entire domain I,
and on the domain I ⁄ II interface, with the Tyr84 loop
again being involved in some of the observed struc-
tural changes. This has a number of implications.
Allosteric effects on Tyr84 caused by ligand binding
elsewhere on the molecule, or post-translational modi-
fications (such as cleavage by chymase) are likely to
alter Cys34 reactivity in vivo. Consequently, a signifi-
cant reduction or increase in Cys34 reactivity is likely
to affect circulatory processes such as the transport of
nitric oxide, removal of reactive oxygen species, or
indeed, drug binding and transport. Mutant albumins
with altered reactivity at this site could prove useful
for scavenging reactive oxygen species or could be
administered to increase the efficacy of Cys34-specific
therapeutics.
Experimental procedures
Mutagenesis, protein expression and purification
Oligonucleotide-directed mutagenesis was used to prepare
cDNAs encoding the mutated albumins. Mutagenesis was
performed by the method of Kunkel [33]. A clone contain-
ing the desired mutation was identified by nucleotide
sequence analysis across the mutation site by the dideoxy
chain termination sequencing. The mutated cDNA was
inserted into a pAYE316 based yeast expression plasmid

[34] and Saccharomyces cerevisiae strain DXY1 [35] was
transformed to leucine prototrophy by electroporation.
Albumin was expressed in S. cerevisiae DXY1 cells and
purified by cation-exchange chromatography on SP-Seph-
arose (Amersham Bioscience, Buckinghamshire, UK; final
elution 85 mm sodium acetate containing 5 mm octanoic
acid, pH 5.5), anion-exchange chromatography on DEAE-
Sepharose (Amersham Biosciences; elution with 110 mm
borate, pH 9.4) and by affinity chromatography on Delta
Blue Agarose (ProMetic Biosciences; elution with 50 mm
phosphate buffer containing 2 m NaCl, pH 6.9). Purity of
the proteins by SDS ⁄ PAGE was > 99%, and ESI-MS of
native and mutant proteins gave peaks within 4 amu of the
calculated masses. CD was carried out as described in [32].
CD spectra of the Cys34Ala, His39Leu and Tyr84Phe
mutants were similar to the native protein showing that the
mutations did not cause any significant changes in secon-
dary structure. Aliquots of the mutant proteins (10 mL in
50 mm phosphate buffer ⁄ 2 m NaCl) and native rHA
(3.8 mm, 145 mm NaCl, containing 15 mgÆL
)1
Tween-80
and 40 mm octanoic acid) were routinely dialysed twice in
100 mm ammonium carbonate (277 K, 5 L, 24 h) before
use. Dialysis reduced the amount of octanoate bound to
rHA (from  8to<4molÆmol protein
)1
).
Mass spectrometry
rHA (15 lm) was incubated with 40 molar equivalents of

DTNB in 100 mm potassium phosphate pH 8.0 for 1 h at
298 K. All solutions were then desalted ⁄ concentrated using
reversed phase solid phase extraction with recovered protein
at concentrations of  20 lm. The solid phase extraction
protocol used mobile phase flow-through cartridges (Inter-
national Sorbent Technology Ltd, Hengoed, UK) under
vacuum. Protein samples were loaded in aqueous solutions
and eluted using 70% acetonitrile in 0.2% formic acid (v ⁄ v).
The protein solution was introduced into a triple quadru-
pole mass spectrometer (Micromass Quattro, Elstree,
Hertfordshire, UK), equipped with a conventional geometry
AP-ESI source in positive ion mode, using flow injection
analysis and 20 scans were typically averaged. For protein
analysis, the mass spectrometer was calibrated against the
protonated molecular ions of horse heart myoglobin (Sigma,
Poole, Dorset, UK) with resolution set similar to 2000.
Peptide samples were introduced using custom nanoelectro-
spray ion sources (both continuous flow ⁄ on-line and off-line
nanovial configurations) in positive ion mode. MS and tan-
dem MS product ion spectra were acquired with significant
scan averaging and the analysers were calibrated against
protonated and sodiated ions from a mixture of polyethy-
lene glycol.
Disulfide interchange reaction with DTNB
Reactions mixtures were set up as follows: 1.35 mL of
respective buffer, 1.35 mL of 50 mm KCl, 150 l Lof15mm
DTNB (dissolved in methanol). The following buffers were
used: 0.2 m potassium phosphate (pH range 6.0–8.0), 0.2 m
Tris ⁄ HCl (pH range 8.1–9.0) and 0.2 m, CAPS (pH range
9.1–10.2). It has been noted previously for the reaction of

BSA with DTNB that the rates are sensitive to ionic
strength but apparently independent of the buffer used [36].
The ionic strengths of the buffers were therefore kept con-
stant and were adjusted to that of 0.2 m phosphate
(0.466 m) with KCl when necessary.
The reaction mixture was equilibrated in the cuvette at
283 K whilst mixing. Following equilibration, the reaction
was initiated by addition of 150 lL of 150 lm albumin, at
283 K. The reactions were performed at 283 K so that
the rates were slow enough to measure by conventional
UV-visible spectroscopy. All reactions were carried out in
1-cm pathlength cells with stirring and were followed by the
A. J. Stewart et al. Redox activity of albumin Cys34
FEBS Journal 272 (2005) 353–362 ª 2004 FEBS 359
change in absorbance at 412 nm corresponding to the
formation of the TNB product (e
412
¼ 13.9 mm
)1
Æcm
)1
[37])
on a Cary 300 Scan spectrophotometer (Varian Ltd,
Walton-on-Thames, UK) fitted with a Peltier dual cell tem-
perature controller. The thiol contents, as determined from
the absorbance of TNB at 412 nm after completion of the
reaction, of rHA, His39Leu and Tyr84Phe albumins were
0.68, 0.63 and 0.63 molÆmol
)1
, respectively.

1
H NMR spectroscopy
To eliminate NH resonances, lyophilized samples of albu-
min were dissolved in D
2
O (99.9% isotopic purity; Aldrich,
Gillingham, Dorset, UK) at  50 mgÆmL
)1
, kept at 277 K
for 48 h, lyophilized, and then dissolved to give a 1 mm
protein solution in D
2
O containing either 50 mm NaCl,
50 mm Tris, or 100 mm potassium phosphate. Sodium for-
mate was added at a concentration of 1 mm as internal cal-
ibration standard [8.48 p.p.m. at pH* 7 relative to sodium
3-(trimethylsilyl) propionate]. The pH meter reading for
D
2
O solutions (pH*, calibrated with H
2
O buffers) is not
corrected for the effect of deuterium on the glass electrode;
the corresponding pD value can be obtained by adding 0.4
units to pH* [38]. pH* values were adjusted with 1 m DCl
or 2 m NaOD. 1D and 2D
1
H NMR experiments were car-
ried out at 310 K on a Bruker Avance 600 spectrometer
(Coventry, UK) operating at 599.82 MHz using a Z-gradi-

ent triple-resonance (
1
H,
13
C,
15
N) probe head. Typically,
512 transients were acquired for the 1D spectra (90° excita-
tion pulse, 9 kHz sweepwidth, 8 k time domain data points)
using a presaturation pulse sequence for residual water
suppression (1.5 s relaxation delay). The data were zero-
filled to 32 k, apodized with an optimized combination of
squared-sinebell and Gaussian functions for resolution
enhancement, and Fourier transformed.
For 2D TOCSY experiments (90° excitation pulse,
8.4 kHz frequency width, mixing time 65 ms, 1.3 s relaxation
delay), 48 or 56 transients for each of 2 · 512 t
1
increments
(hypercomplex acquisition using time-proportional phase
incrementation) were acquired into 4 k complex data points,
using a sensitivity-enhanced, double-pulsed field-gradient
spin-echo sequence for residual water suppression. The data
were apodized using squared-sinebell functions, and the real
Fourier transform was carried out on 2 k · 2 k data points.
Reaction with cystine followed by 1D
1
H NMR
l-Cystine (Sigma) was dissolved in 2 m NaOD in D
2

Otoa
final concentration of 100 mm. After the recording of an
initial spectrum of 1 mm rHA (wild-type or Tyr84Phe) in
100 mm potassium phosphate (in D
2
O; pH* 7.0), an aliquot
corresponding to a 2.5-fold molar excess of cystine was
added directly into the NMR tube. The pH* was readjusted
to 7.0 with 1 m DCl, and 1D spectra were recorded at
20-min intervals. At this pH* the reaction was slow enough
to be followed by NMR. As the observed reaction involves
(de)protonation steps, we used potassium phosphate in
these experiments instead of Tris, since it is a more effective
buffer in this pH* range. During these studies we noted
that the chemical shifts of the histidine He1 protons are
also dependent on the nature of the buffer. Generally, sig-
nals were shifted upfield in potassium phosphate buffer
compared to spectra taken at the same pH* in Tris buffer.
Consequently, the wild-type spectrum recorded in Tris at
pH* 7.37 (Fig. 3, bottom) closely resembles the spectrum
recorded in phosphate buffer at pH* 7.05 (Fig. 4A, bot-
tom). The overall quality of the spectra is similar, but some
resonances are sharper in Tris.
Acknowledgements
We thank the BBSRC and Delta Biotechnology Ltd.
(CASE award for AJS), EC (Marie Curie Fellowship
for CAB), and The Wellcome Trust (Edinburgh Pro-
tein Interaction Centre) for their support for this work.
We are grateful to Tony Greenfield and Lee Blackwell
(Delta Biotechnology Ltd) for help with expression

and purification of mutant proteins and Dr Sharon
Kelly (University of Glasgow) for CD studies.
References
1 Peters T Jr (1995) All About Albumin: Biochemistry,
Genetics, and Medical Applications. Academic Press,
New York.
2 Carter DC & Ho JX (1994) Structure of serum albumin.
Adv Prot Chem 45, 153–203.
3 Curry S, Mandelkow H, Brick P & Franks N (1998)
Crystal structure of human serum albumin complexed
with fatty acid reveals an asymmetric distribution of
binding sites. Nat Struct Biol 5, 827–835.
4 Sugio S, Kashima A, Mochizuki S, Noda N & Kobaya-
shi K (1999) Crystal structure of human serum albumin
at 2.5 A
˚
resolution. Protein Eng 12, 439–446.
5 Pirisino R, Disimplicio P, Ignesti G, Bianchi G &
Barbera P (1988) Sulfhydryl groups and peroxidase-like
activity of albumin as scavenger of organic peroxides.
Pharmacol Res Commun 20, 545–552.
6 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.
7 Malik NA, Otiko G & Sadler PJ (1980) Control of
intra- and extra-cellular sulphydryl-disulphide balances
with gold phosphine drugs:
31
P nuclear magnetic reso-

nance studies of human blood. J Inorg Biochem 12,
317–322.
8 Roberts JR, Xiao J, Schliesman B, Parsons DJ & Shaw
CF III (1996) Kinetics and mechanism of the reaction
Redox activity of albumin Cys34 A. J. Stewart et al.
360 FEBS Journal 272 (2005) 353–362 ª 2004 FEBS
between serum albumin and auranofin (and its isopropyl
analogue) in vitro. Inorg Chem 35, 425–433.
9 Ross SA, Carr CA, Briet JW & Lowe G (2000) Transfer
of 4¢-chloro-2,2¢:6¢,2¢¢-terpyridine platinum (II) between
human serum albumin, glutathione and other thiolate
ligands. A possible selective natural transport mechan-
ism for the delivery of platinum (II) drugs to tumour
cells. Anticancer Drug Des 15, 431–439.
10 Esposito BP & Najjar R (2002) Interactions of anti-
tumoral platinum-group metallodrugs with albumin.
Coord Chem Rev 232, 137–149.
11 Yasutake A, Hirayama K & Inoue M (1990) Interaction
of methylmercury compounds with albumin. Arch Toxi-
col 64, 639–643.
12 Keire DA, Mariappan SV, Peng J & Rabenstein DL
(1993) Nuclear magnetic resonance studies of the bind-
ing of captopril and penicillamine by serum albumin.
Biochem Pharmacol 46, 1059–1069.
13 Agarwal RP, McPherson RA & Phillips M (1983)
Rapid degradation of disulfiram by serum albumin. Res
Commun Chem Pathol Pharmacol 42, 293–310.
14 Stamler JS, Jaraki O, Osborne J, Simon DI, Keany J,
Vita J, Singel D, Valeri R & Loscalzo J (1992) Nitric
oxide circulates in mammalian plasma primarily as an

S-nitroso adduct of serum albumin. Proc Natl Acad Sci
USA 89, 7674–7677.
15 Kashiba-Iwatsuki M, Miyamoto M & Inoue M (1997)
Effect of nitric oxide on the ligand-binding activity of
albumin. Arch Biochem Biophys 345, 237–242.
16 Burczynski FJ, Wang GQ & Hnatowich M (1995) Effect
of nitric oxide on albumin-palmitate binding. Biochem
Pharmacol 49, 91–96.
17 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.
18 Ashworth MRF (1976) The Determination of Sulphur-
Containing Groups. Analytical Methods for Thiol
Groups. Vol. 2. Academic Press, London.
19 Pedersen AO & Jacobsen J (1980) Reactivity of the thiol
group in human and bovine albumin at pH 3–9, as
measured by exchange with 2,2¢-dithiodipyridine. Eur J
Biochem 106, 291–295.
20 Shaked Z, Szajewski RP & Whitesides GM (1980) Rates
of thiol-disulfide interchange reactions involving pro-
teins and kinetic measurements of thiol pK
a
values. Bio-
chemistry 19, 4156–4166.
21 Zhang ZY & Dixon JE (1993) Active site labeling of the
Yersinia protein tyrosine phosphatase: the determination
of the pK
a
of the active site cysteine and the function of

the conserved histidine 402. Biochemistry 32, 9340–9345.
22 Guengerich FP, Fang Q, Liu L, Hachey DL & Pegg AE
(2003) O6-alkylguanine-DNA alkyltransferase: low pK
a
and high reactivity of cysteine 145. Biochemistry 42,
10965–10970.
23 Li H, Hanson C, Fuchs JA, Woodward C & Thomas
GJ Jr (1993) Determination of the pK
a
values of active-
center cysteines, cysteines-32 and -35, in Escherichia coli
thioredoxin by Raman spectroscopy. Biochemistry 32,
5800–5808.
24 Bos OJM, Labro JFA, Fischer MJE, Wilting J &
Janssen LHM (1989) The molecular mechanism of the
neutral-to-base transition of human serum albumin.
Acid ⁄ base titration and proton nuclear magnetic reson-
ance studies on a large peptic and a large tryptic frag-
ment of albumin. J Biol Chem 264, 953–959.
25 Lee H, Cha M-K & Kim I-H (2000) Activation of thiol-
dependent antioxidant activity of human serum albumin
by alkaline pH is due to the B-like conformational
change. Arch Biochem Biophys 380, 309–318.
26 Cornell CN & Kaplan LJ (1978) Spin-label studies on
the sulfhydryl environment in bovine plasma albumin.
2. The neutral transition and the A isomer. Biochemistry
17, 1755–1758.
27 Ibarra C, Nieslanik BS & Atkins WM (2001) Contribu-
tion of aromatic–aromatic interactions to the anomalous
pK

a
of tyrosine-9 and the C-terminal dynamics of gluta-
thione S-transferase A1–1. Biochemistry 40, 10614–
10624.
28 Guddat LW, Bardwell JCA & Martin JL (1998) Crystal
structures of reduced and oxidized DsbA: investigation
of domain motion and thiolate stabilization. Structure
6, 757–767.
29 Raymond WW, Ruggles SW, Craik CS & Caughey GH
(2003) Albumin is a substrate of human chymase. Pre-
diction by combinatorial peptide screening and develop-
ment of a selective inhibitor based on the albumin
cleavage site. J Biol Chem 278, 34517–34524.
30 Satomura K, Yin M, Shimizu S, Kato Y, Nagano T,
Komeichi H, Ohsuga M, Katsuta Y, Aramaki T &
Omoto Y (2003) Increased chymase in livers with auto-
immune disease: colocalization with fibrosis. J Nippon
Med Sch 70, 490–495.
31 Christodoulou J, Sadler PJ & Tucker A (1995)
1
H
NMR of albumin in human blood plasma: drug
binding and redox reactions at Cys34. FEBS Lett 376,
1–5.
32 Stewart AJ, Blindauer CA, Berezenko S, Sleep D &
Sadler PJ (2003) Interdomain zinc site on human
albumin. Proc Natl Acad Sci USA 100, 3701–3706.
33 Kunkel TA (1985) Rapid and efficient site-specific muta-
genesis without phenotypic selection. Proc Natl Acad
Sci USA 82, 488–492.

34 Sleep D, Belfield GP, Balance DJ, Steven J, Jones S,
Evans LR, Moir PD & Goodey AR (1991) Saccharo-
myces cerevisiae strains that overexpress heterologous
proteins. Biotechnology 9, 183–187.
35 Kerry-Williams SM, Gilbert SC, Evans LR & Ballance
DJ (1997) Disruption of the Saccharomyces cerevisiae
YAP3 gene reduces the proteolytic degradation of
A. J. Stewart et al. Redox activity of albumin Cys34
FEBS Journal 272 (2005) 353–362 ª 2004 FEBS 361
secreted recombinant human albumin. Yeast 14, 161–
169.
36 Wilson JM, Wu D, Motiu-DeGrood R & Hupe DJ
(1980) A spectroscopic method for studying the rates of
reaction of disulfides with protein thiol-groups applied
to bovine serum albumin. J Am Chem Soc 102, 359–
363.
37 Ellman GL (1959) Tissue sulfhydryl groups. Arch
Biochem Biophys 82, 70–77.
38 Glasoe PK & Long FA (1960) Use of glass electrodes
to measure acidities in deuterium oxide. J Phys Chem
64, 188–190.
39 Brocklehurst K & Little G (1973) Reactions of papain
and of low-molecular-weight thiols with some aromatic
disulphides. 2,2¢-Dipyridyl disulphide as a convenient
active-site titrant for papain even in the presence of
other thiols. Biochem J 133, 67–80.
Redox activity of albumin Cys34 A. J. Stewart et al.
362 FEBS Journal 272 (2005) 353–362 ª 2004 FEBS