Journal of Structural Biology 171 (2010) 353–360

Contents lists available at ScienceDirect

Journal of Structural Biology

journal homepage: www.elsevier.com/locate/yjsbi

Crystallographic analysis reveals a unique lidocaine binding site on human

serum albumin
Kim Langmach Hein a, Ulrich Kragh-Hansen b,*, J. Preben Morth a, Martin D. Jeppesen c, Daniel Otzen c,
Jesper V. Møller d, Poul Nissen a,**
a
Department of Molecular Biology, University of Aarhus, Gustav Wieds Vej 10C, DK-8000 Århus C, Denmark
b
Department of Medical Biochemistry, University of Aarhus, Ole Worms Alle, Building 1170, DK-8000 Århus C, Denmark
c
Interdisciplinary Nanoscience Center, University of Aarhus, Gustav Wieds Vej 10C, DK-8000 Århus C, Denmark
d
Department of Physiology and Biophysics, University of Aarhus, Ole Worms Alle, Building 1160, DK-8000 Århus C, Denmark

a r t i c l e i n f o a b s t r a c t

Article history: Human serum albumin (HSA), the major protein component in blood plasma and in extravascular spaces,
Received 17 March 2010 is known to participate in the binding and transport of a variety of endogenous and exogenous organic
Accepted 23 March 2010 compounds with anionic or electronegative features. We here report on the 3.3 Å resolution crystal struc-
Available online 27 March 2010
ture of HSA complexed with the cationic, and widely used, anesthetic lidocaine. We find that lidocaine
and HSA co-crystallise as a dimer in the unusual space group I41. The dimer consists of one HSA molecule
Keywords: without ligand and one HSA molecule with a single, bound lidocaine. HSA is a heart-shaped protein com-
Human serum albumin
posed of three homologous helical domains (I–III), which can be subdivided into two subdomains (A and
Lidocaine binding
Sulphate ion binding
B), and lidocaine binds to a unique site formed by residues from subdomain IB facing the central, inter-
Crystal structure domain crevice. In the crystal, binding seems to introduce only local conformational changes in the pro-
Conformational changes tein. According to intrinsic fluorescence experiments with aqueous HSA binding results in widespread
Intrinsic fluorescence conformational changes involving Trp214 in subdomain IIA. Results obtained with equilibrium dialysis
and isothermal titration calorimetry show that lidocaine binding is of a low affinity and occurs at one dis-
crete binding site in accordance with the X-ray data. Another crystal form of ligand-free HSA obtained in
the presence of ammonium sulphate was determined at 2.3 Å resolution revealing a sulphate ion accept-
ing cavity at the surface of subdomain IIIA. The present results contribute to a further characterisation of
the exceptional binding properties of HSA.
Ó 2010 Elsevier Inc. All rights reserved.

1. Introduction to its high concentration. In addition, structural information of

Human serum albumin (HSA) is the most abundant plasma pro-
tein (0.6 mM) and comprises 50–60% of the total plasma protein
in humans. An even larger pool of HSA is found in the extravascular
spaces although at a lower concentration (Peters, 1996). As one of
its main functions, HSA serves as a depot and transport protein for
numerous endogenous and exogenous compounds. Among the
exogenous compounds is a wide range of drugs, and binding to
HSA affects their absorption, distribution, metabolism and elimina-
tion. Because HSA has a limited number of high-affinity binding
sites, detailed molecular information about these sites is very help-
ful in the assessment of cooperative effects of binding of other
drugs or endogenous ligands. However, the distribution of com-
pounds will also be affected by low-affinity binding to HSA due
* Corresponding author. Fax: +45 8613 1160.
** Corresponding author.
E-mail addresses: ukh@biokemi.au.dk (U. Kragh-Hansen), pn@mb.au.dk
(P. Nissen).
both high and low-affinity binding sites is useful when designing
new drugs whether the aim is to avoid binding to HSA, or to make
use of its depot function.
HSA is an a-helical protein (67% a-helix) consisting of a single
polypeptide chain of 585 amino acids that form a heart-shaped pro-
tein with three homologous domains (I–III) (Carter and Ho, 1994;
Sugio et al., 1999). Each of the domains is composed of two subdo-
mains (A and B) with distinct helical folding patterns connected by
flexible loops. Small-angle X-ray scattering studies of HSA in solu-
tion show general agreement with the crystal structure (Olivieri
and Craievich, 1995). Also, a combined phosphorescence depolar-
isation–hydrodynamic modelling study has proposed that the over-
all conformation of HSA in neutral solution is very similar to that
observed in crystal structures (Ferrer et al., 2001).
The drugs binding to HSA are mainly lipophilic compounds with
anionic or electronegative features, and crystal structures of many
HSA complexes have been determined (Carter and Ho, 1994; Bhat-
tacharya et al., 2000b; Petitpas et al., 2001; Ghuman et al., 2005). In
plasma, cationic drugs usually bind to a1-acid glycoprotein (oroso-

1047-8477/$ - see front matter Ó 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.jsb.2010.03.014
354 K.L. Hein et al. / Journal of Structural Biology 171 (2010) 353–360

mucoid) (Schönfeld et al., 2008) however, some of them bind to
HSA (Kragh-Hansen, 1981). No structural information about such
binding sites is available. Therefore, in the present work we have
performed a crystallographic analysis of HSA complexed with lido-
caine, a Na+ channel blocker widely used as a local anesthetic and
antiarrhythmic drug. The structural work was supplemented with
intrinsic fluorescence data, equilibrium dialysis and isothermal
titration calorimetry. The study revealed a new crystal form of a
HSA-ligand that has not been observed before and the existence
of a unique binding site located between domains I and III. Further-
more we compare our results to a 2.3 Å resolution crystal structure
of HSA that we have obtained with bound sulphate.
2. Materials and methods
2.1. Materials
2.3. Data collection and processing
Initial testing of the crystals was performed in-house with a
copper rotating-anode X-ray generator, whereas further testing
using synchrotron radiation was performed at European Molecular
Biology Laboratory outstation at Deutsches Elektronen Synchro-
tron (EMBL/DESY, Hamburg) or at Berliner Elektronenspeicher-
ring-Gesellschaft für Synchrotronstrahlung (BESSY). X-ray
diffraction data for the HSA–lidocaine complex and HSA were col-
lected under cryo-conditions (110 K) using synchrotron radiation
at the Swiss Light Source (SLS, Zurich) and BESSY facilities, respec-
tively. The data were integrated and scaled with XDS (Kabsch,
1993). The 3.3 Å data set obtained on the HSA–lidocaine complex
was processed in space group I41, and the 2.3 Å data set obtained
for HSA was processed in space group P1; data collection details
and unit cell parameters are given in Table 1. The Wilson plot
and a diffraction image from the 3.3 Å data set are shown in Sup-
plementary materials.

HSA (A-7223, 95% pure), expressed in Pichia pastoris, was ob-
tained from Sigma–Aldrich (St. Louis, MO, USA). For removing
hydrophobic and hydrophilic ligands and additives, the protein
was treated with charcoal at pH 3 as described by Chen (1967), dia-
lysed extensively against deionised water, freeze-dried and then
stored at 20 °C until use. HSA (A-1887, >96% pure and essentially
fatty acid free) was also supplied by Sigma–Aldrich.
Lidocaine was from Sigma–Aldrich, and [carbonyl-14C]lidocaine
hydrochloride (99% pure according to the manufacturer) was from
American Radiolabeled Chemicals Inc. (St. Louis, MO, USA). Other
chemicals were also of the best grades commercially available.
2.4. Structure determination
The structures were solved by molecular replacement using
PHASER (McCoy et al., 2005) and employing an array of HSA struc-
tures to identify a suitable search model. Albumin complexed with
diflunisal (PDB 2BXE) and consisting of residues 5–78 and 88–582
provided a successful molecular replacement solution for the HSA–
lidocaine complex as probed by Z-scores, electron density maps of
unmodelled regions and subsequent refinement statistics. Rigid
body refinement of HSA–lidocaine domains revealed significant

Table 1
Data collection details and unit cell parameters.
2.2. Preparation and crystallisation of HSA with and without lidocaine HSA–lidocaine HSA–SO4
X-ray source Swiss Light Source–X06SA BESSY–PX2
To isolate monomeric HSA for crystallisation with lidocaine, the Wavelength (Å) 1.008 0.915
protein preparation was subjected to further purification on a Space group Tetragonal I41 Triclinic P1
Superdex S75 gel filtration column (GE Healthcare) using 50 mM Cell parameters
a (Å) 168.5 55.6
potassium phosphate with 150 mM sodium chloride, pH 7.5, as a
b (Å) 168.5 55.8
running buffer (Curry et al., 1998). Peak fractions were collected c (Å) 97.0 121.3
and analysed by SDS–PAGE. Fractions with HSA monomer were a (°) 90 80.5
pooled and concentrated in 2 mM potassium phosphate, pH 7.5, b (°) 90 90.7
to a concentration greater than 100 mg/ml. c (°) 90 63.5
Resolution range (Å) 120–3.3 (3.5–3.3) 120–2.3 (2.4–2.3)
Monomeric HSA was mixed with a 20-fold molar excess of lido- Completeness (%) 99.9 (100) 91.1 (85.6)
caine and incubated for 1–2 h. The free drug concentration was Rmrgd  F (%) 16.9 (144.7) 10.4 (50.5)
then fixed by repeated cycles of concentration and dilution in Rmeas (%) 10 (171.8) 5.8 (33.6)
20 mM potassium phosphate buffer, pH 7.5, containing 0.1 mM I/r(I) 13.04 (1.2) 13.46 (3.4)
Unique reflections 20,507 51,888
lidocaine using a 10,000 MWCO centrifugal concentrator (Viva-
Redundancy 6.3 1.8
spin); the final protein concentration was 100 mg/ml or higher.
Model refinement
For crystallisation without lidocaine, defatted, freeze-dried HSA
Resolution range (Å) 45–3.3 (3.45–3.3) 28–2.3 (2.34–2.3)
was used at a concentration of 100 mg/ml. Model amino acids 1164 1150
The crystallisation experiments were carried out by vapour dif- Rwork (%)a 22.0 (32.9) 20.9 (23.5)
fusion at 19 °C using the sitting or hanging drop method. In most Rfree (%)b 26.8 (36.0) 27.0 (31.0)
cases, a 2 ll protein sample was mixed with an equal volume of Overall B-factor (Å) 142.4 44.05
Average ligand B-factor (Å) 113.2 28.7
reservoir solution in 24-well plates (Hampton Research). Crystals RMSDbonds (Å)c 0.003 0.009
of HSA–lidocaine generally appeared after 3–4 days with crystalli- RMSDangles (Å)c 0.715 1.045
sation conditions consisting of 20–21% (w/v) polyethylene glycol Ramachandran (%)d 83.9 95.1
6000, 9–10% ethylene glycol, 50/100 mM ammonium acetate, 2%
The values for the highest resolution shell are given in parentheses.
(v/v) 2-methyl-2,4-pentanediol, and 50 mM Tris–HCl, pH 7.5. Crys- Rmeas = redundancy independent R-factor (intensities). Rmrgd  F = quality of
tals of HSA without lidocaine were prepared in crystallisation solu- amplitudes (F) in the scaled data set.
tions consisting of 22–25% polyethylene glycol 3350, 100–200 mM For definitions, see Diederichs and Karplus (1997).
a P P
Rwork (%) = 100  hkl|Fobs  Fcalc|/ hkl Fobs, where Fobs and Fcalc are the
ammonium sulphate and 50 mM Tris–HCl, pH 7.5. For use in X-ray
observed and calculated structure factors, respectively.
diffraction data collection, crystals were cryoprotected by adding b
Rfree(%) is calculated using a randomly selected 5% sample of reflection data
reservoir buffer containing glycerol to the drop, yielding a final omitted from the refinement.
c
glycerol concentration of approximately 20%. Afterwards, the crys- Root mean square deviation from ideal bond lengths and angles.
d
tals were frozen in liquid nitrogen. Residues in most favoured regions.
 K.L. Hein et al. / Journal of Structural Biology 171 (2010) 353–360 355

changes with relative displacement of domain I and III. A model
consisting of HSA residues 5–582 was refined by simulated anneal-
ing using torsion angle dynamics and group B-factors using CNS
(Brünger et al., 1998). Inspection of electron density maps revealed
a ligand binding site in one of the two protein molecules of the
asymmetric unit; comparable in size and overall shape to lidocaine
(Fig. 2d). A single lidocaine molecule was then included in the
structure followed by further refinement cycles using PHENIX
(Adams et al., 2002) and applying TLS refinement. The model yields
Rwork and Rfree values of 22.0% and 26.8%, respectively (Table 1),
and shows favourable Ramachandran plot statistics (most favoured
regions: 83.9%). A successful molecular replacement solution for
HSA with sulphate was obtained using PDB 1BM0 as model. Initial
refinement was carried out with PHENIX and poorly resolved re-
gions, mainly domain IIIB, were manually rebuilt using COOT
(Emsley and Cowtan, 2004), with help from Arp/wArp (Perrakis
et al., 2001) and BUCCANEER (Cowtan, 2006). The final model
shows Rwork and Rfree values of 20.9% and 27.0%, respectively (Ta-
ble 1) with 95.1% of residues in the most favourable regions of
the Ramachandran plot.

2.5. Intrinsic fluorescence

The effect of lidocaine binding on the intrinsic fluorescence of

HSA was recorded at 300–425 nm with a Shimadzu RF 5301 PC
spectrofluorometer after excitation at 290 nm as previously de-
scribed (Kragh-Hansen et al., 2001). The protein (13 lM) without
or with ligand (0.71–7.1 mM) was solubilised in 50 mM Tris–HCl
(pH 7.4), and the spectra were recorded at 25 °C.

2.6. Equilibrium dialysis

Fig. 1. Chemical structure of lidocaine on its protonated form. The pKa of the drug is Ligand binding was quantified by use of Dianorm equilibrium
7.80–7.86 (Truant and Takman, 1959), and its molecular weight is 235 Da. dialysers (Dianorm Geräte, München, Germany). The dialysis cells,

Fig. 2. Structure of HSA–lidocaine complex. (a) The asymmetric unit showing two HSA molecules of which one has bound a lidocaine molecule (green). (b) Overview of HSA
with lidocaine bound to subdomain IB in the cleft between domains I and III. The subdivision of HSA into domains (I–III) and subdomains (A and B) is shown. (c) Surface view
of the lidocaine binding site. The domains are colour-coded as follows: I, deep blue; II, marine blue; III, light blue. (d) Yellow; positive Fo  Fc sa-omit difference electron
density map, with the lidocaine molecule omitted from the phasing model, contoured at 2.5r and drawn in a radius of 4 Å around lidocaine. Red; positive Fo  Fc difference
electron density map before including lidocaine, contoured at 2.5r (e) 2Fo  Fc electron density map contoured at 1r. (f) HSA with lidocaine (blue) superposed with HSA
(orange, PDB 1AO6) using the residues from position 100 to 200. All illustrations in this and the following structural figures were prepared using PyMol (Delano, 2002).
356 K.L. Hein et al. / Journal of Structural Biology 171 (2010) 353–360
with half-cell volumes of 250 ll carved into cylindrical disks, were
assembled pair-wise, together with flat pieces of separating cellu-
lose tubing. For this we used Visking size 9, / = 4.0 cm, with a
molecular mass cut off of 12,000–14,000 from Medicell Interna-
tional Ltd., London, UK.
The experiments were carried out at 25 °C with media buffered
with 50 mM Tris–HCl, pH 7.4. Solutions of different concentrations
of labelled plus unlabelled ligand (25–250 lM) were prepared with
a constant albumin concentration (150 lM). Sample aliquots of
175 ll containing protein and ligand were pipetted into the left-
side of the cells, and sample of 175 ll of buffer alone were pipetted
into the right-side of the cells. After closure of the cells, the equi-
librium dialysers were placed in a temperature-controlled water
bath, and the cells were rotated for ca. 20 h about their horizontal
axes. Then the half-cells were emptied, and 100 ll portions from
the left and right compartments were used for counting of radioac-
tivity. Control experiments revealed that equilibrium was estab-
lished within the period of time used. The recovery of
radioactivity was 95.6% (±0.4%) (n = 10). The loss must be supposed
to be caused by adsorption to the Teflon of the cell compartments
and/or to the cellophane membrane.
The radioactivity of lidocaine solutions which had not been dia-
lysed was used to represent the known concentration of total li-
gand. After dialysis, the radioactivity of the samples from left-
side and right-side compartments represent the concentration of
unbound plus albumin-bound ligand and unbound ligand, respec-
tively. The association constant (K) for lidocaine binding was calcu-
lated by using the following equation:
m ¼ ðK  ½LÞ=ð1 þ K  ½LÞ
In this equation, t is the average number of ligand molecules bound
per protein molecule, and [L] is the concentration of unbound
ligand.
2.7. Isothermal titration calorimetry
Lidocaine binding was examined using a Valerian Plotnikov (VP)
ITC MicroCalorimeter from MicroCal (Northampton, USA). De-
gassed solutions of 0.1 mM albumin in 50 mM Tris–HCl buffer,
pH 7.4 was transferred to the ITC cell, and the syringe was filled
with 10 mM lidocaine in the same buffer. The ITC titration was
done at 25 °C with an initial injection of 1 ll followed by 59 injec-
tions of 5 ll, all with a 10-min equilibration period between the
injections. The heat of dilution from lidocaine alone was measured
in a similar way by having only buffer in the ITC cell. The heat of
dilution signal was subtracted from the binding data using ITC
Data Analysis in MicroCal Origin 7.0.
2.8. Protein Data Bank accession numbers
Coordinates and structure factors have been deposited in the
Protein Data Bank with accession numbers 3JQZ (HSA–lidocaine)
and 3JRY (HSA–sulphate).
3. Results
3.1. Structure of the HSA–lidocaine complex and of HSA with sulphate
Lidocaine is a tertiary amine compound, and its protonated
form is shown in Fig. 1. The drug and HSA were dissolved in phos-
phate buffer, pH 7.5, at a 20:1 M ratio of ligand to protein and co-
crystallised as a dimer in the unusual space group I41 (Fig. 2a). The
dimer consists of one HSA molecule without ligand and one HSA
molecule with a single, bound lidocaine (an asymmetric dimer).
A combination of crystal packing and low-affinity binding of
lidocaine may explain why an asymmetric 2:1 HSA:lidocaine com-
plex has occurred. An examination of the literature (Carter and Ho,
1994; Sugio et al., 1999; Ghuman et al., 2005; Lejon et al., 2008;
Zunszain et al., 2008; Zhu et al., 2008) revealed that so far no other
HSA crystal has been reported to belong to this type of space group.
The structure was determined by molecular replacement and re-
fined at 3.3 Å resolution. The refined model has an Rfree value of
26.8% and valid stereochemistry (Table 1).
The fact that lidocaine binds to only one of the two HSA mole-
cules in the asymmetric dimer is illustrated in Fig. 2a. The binding
site is situated in subdomain IB in the upper part of the interdo-
main cleft (Fig. 2b and c). The difference electron density (Fo  Fc
simulated annealing omit map) shows clear density for a molecule
the size of lidocaine at that position (Fig. 2d). Fig. 2e gives the
molecular details for binding of lidocaine (and as further indicated
by final 2Fo  Fc map). As seen from the illustration, the aromatic
part of lidocaine (Fig. 1) is involved in cation–p interactions with
Arg114, and the carbonyl group interacts with Lys190. Further sta-
bilization is provided by electrostatic interaction between the pos-
itively charged, tertiary amine nitrogen and the carboxylate group
of Asp187. Apparently, the interactions between the non-aromatic
part of lidocaine and Asp187 and Lys190 do not cause any signifi-
cant displacement of the two residues (Fig. 2f). By contrast, the
aromatic part of the ligand attracts Arg114; the residue has been
moved about 4 Å. Furthermore, Arg186 is pushed aside making
room for the incoming ligand.
HSA without lidocaine was crystallized in space group P1 in the
presence of sulphate ions, and its crystal structure was determined
by molecular replacement and refined at 2.3 Å resolution yielding a
ð1Þ
final Rfree of 27.1% (Table 1). Although HSA is known to bind inor-
ganic anions such as Cl, F, SCN and SO2 4 apparently no informa-
tion exists on the exact location and construction of these sites
(Peters, 1996; Kragh-Hansen, 1981). As seen in Fig. 3a, we were
able to detect a sulphate ion accepting cavity at the surface of sub-
domain IIIA. The ion interacts with Ser480 and the backbone nitro-
gens of Leu481 and Val482 (Fig. 3b). Additional stabilization
consists of van der Waals interactions with the Lys351 side chain
and coordination to water molecules in the cavity. Finally,
Lys560 from a symmetry related HSA molecule also takes part in
binding, and it is likely that this interaction seals the binding site
and at the same time improves the crystal stability. The positions
of the great majority of residues are not affected by binding, but
Ser480 and part of the peptide chain in which it is placed have
been slightly displaced (Fig. 3c).
Fig. 4a compares the structures of the two HSA molecules in the
asymmetric dimer. It is seen that domain I of HSA without ligand
(yellow) is tilted towards the middle of the molecule when com-
pared to HSA binding lidocaine (green). Furthermore, domain III
of the ligand-free form is slightly more outer placed. The tilting
of domain I and/or local conformational changes in subdomain
IB apparently hinder lidocaine binding. Fig. 4b–d compare our
structures with a HSA structure found in the literature. In a review
(Curry, 2004), Curry compared the ligand-free structure of HSA
published by Sugio et al. (1999) with those published by He and
Carter (1992) and his own group (Bhattacharya et al., 2000b). He
found only some small variations between the three structures,
particularly in the disposition of domain III. In the present context
we have used the 2.8 Å structure published by Sugio et al. (1999)
as the reference structure. Fig. 4b shows that the position of do-
main III of HSA–lidocaine (green) is shifted outwards as compared
with that of the same domain in the structure of Sugio et al. (red).
By contrast, the structure of the two other domains is very similar.
Fig. 4c shows a more pronounced outward disposition of domain
III and a tilting of domain I when comparing our ligand-free form
(yellow) with the structure of Sugio et al. (red). Finally, our 2.3 Å
structure with sulphate (blue) resembles the ligand-free form
 K.L. Hein et al. / Journal of Structural Biology 171 (2010) 353–360 357

Fig. 3. Structure of HSA–sulphate ion complex. (a) Overview of HSA with sulphate bound at the surface of subdomain IIIA. (b) A 2Fo  Fc electron density map contoured at 1r
showing the binding site for sulphate (yellow–red) with interacting residues. Lys560(sym) shown in grey is from a symmetry related HSA molecule. (c) HSA with bound
sulphate (yellow–red) superposed with HSA (orange, PDB 1AO6) using domain II (residues 208–366). The red spheres in panels (b) and (c) are water molecules.

Fig. 4. Superposition of HSA structures using the Ca-atoms of domain II (residues 208–366). (a) HSA with lidocaine (green) and HSA without lidocaine (yellow, 3.3 Å
structure) from the asymmetric dimer. (b) HSA with lidocaine (green) and HSA as determined by Sugio et al. (1999) (red, PDB 1AO6). (c) HSA without lidocaine (yellow, 3.3 Å
structure) and HSA (red, PDB 1AO6). (d) HSA with sulphate (blue, 2.3 Å structure) superposed with HSA (red, PDB 1AO6).

published by Sugio et al. (red) (Fig. 4d). The only clear difference is
a pronounced outward displacement of domain III in our structure,
which could be due to sulphate ion binding (Fig. 3c) and/or caused
by contact with the other albumin molecule in the crystal dimer;
see, e.g., Lys560 in Fig. 3b. We also found that the conformations of
HSA with bound sulphate ion or with bound lidocaine are alike;
the only difference is a small variation in the relative position of
domain III (not shown, but see Fig. 4). In summary, a tilting of do-
main I was observed for the ligand-free HSA molecule comple-
menting the asymmetric dimer of the lidocaine complex
obtained at 3.3 Å resolution, and both HSA-forms obtained
without lidocaine, at 3.3 and 2.3 Å, were broader than the form
found by Sugio et al. (1999) due to an outward displacement of
domain III.
The three HSA structures studied in this work have also been
compared with HSA saturated with myristate (fatty acid:HSA mole
ratio of 12:1) (Bhattacharya et al., 2000a) and with a HSA–hemin
complex partially saturated with myristate (fatty acid:hemin:HSA
mole ratio of 4:1:1) (Zunszain et al., 2003). Binding of high molar
ratios of fatty acid results in rotations of domains I and III relative
to domain II (Bhattacharya et al., 2000a). Partial saturation of HSA
with myristate results in a similar rotation of domain III but in a
less pronounced rotation of domain I (Zunszain et al., 2003). In
both fatty acid–HSA structures, domain III is placed further away
from the central crevice than in our structures. Although the posi-
tion of domain I, especially in the partially fatty acid loaded HSA,
resembles more our structures, it is placed closer to the central cre-
vice; this is especially so when comparing to HSA with lidocaine or
358 K.L. Hein et al. / Journal of Structural Biology 171 (2010) 353–360
sulphate ion. It should be noted that, in contrast to fatty acid bind-
ing, no rotations of domain I and III have taken place in our struc-
tures, rather it is a question of tilting of domain I and/or a pushing
aside of domain III.
3.2. Effect of lidocaine binding on intrinsic tryptophan fluorescence
HSA has a single tryptophan residue (Trp214) but the crystallo-
graphic studies showed that the residue is not directly involved in
lidocaine binding; the closest distance between the two structures
is about 24 Å (not shown). However, the structural features at the
residue could be modified as a result of lidocaine binding. This pos-
sibility was examined for by performing intrinsic fluorescence
studies with solubilised HSA. Excitation of the protein at 290 nm
resulted in an emission spectrum with a kmax of 342 nm (Fig. 5).
Addition of increasing concentrations of lidocaine caused a pro-
gressive reduction of the fluorescence intensity, accompanied by
a decrease of kmax to 338 nm. These findings suggest the existence
of lidocaine-induced conformational changes in HSA leading to a
decreased polarity of the environment of Trp214.
3.3. Binding experiments
Equilibrium dialysis revealed that binding of lidocaine was of a
low affinity, and that it could be described in terms of binding to
one site. The association constant (±SEM) for binding to HSA was
calculated to be 3.1(±0.2)  102 M1 (n = 5). As a control, we also
examined binding to native HSA isolated from human serum (un-
like the recombinant source otherwise used in this study). These
Fig. 5. Changes in HSA intrinsic fluorescence caused by lidocaine binding. The protein was excitated at 290 nm, and the effect of different ligand concentrations on the
emission spectrum is shown. The data in the inset are lidocaine-induced decreases in HSA fluorescence at kmax. The results are given as means ± SEM (n = 3). The broken curve
representing lidocaine binding is constructed by using Eq. (1) and K = 3.1  102 M1.
experiments resulted in a similar association constant,
K = 3.2(±0.3)  102 M1 (n = 5).
The inset to Fig. 5 shows that the changes in fluorescence inten-
sity, resulting from addition of increasing amounts of lidocaine, can
be described fairly well by the hyperbolic curve. The curve repre-
sents lidocaine binding and is constructed by using the association
constant determined by equilibrium dialysis. Thus, the results of
the two different methods agree on low-affinity binding of lido-
caine to one site on HSA.
Only sparse information can be found in the literature about
lidocaine binding to HSA; and only obtained by equilibrium dialysis
as here. Krauss et al. (1986) used [14C] ethyl–lidocaine and reported
binding to one site with a K-value, which depended somewhat on
the type of albumin preparation used, i.e. 0.6–2.3  102 M1. Bailey
and Briggs (2004) measured lidocaine by a homogeneous enzyme
immunoassay, and from their data it is possible to calculate a K-va-
lue of 5.1  102 M1, assuming binding to one site. In contrast to the
present results and to the studies just cited, Sawinski and Rapp
(1963) several years ago reported binding to two sites with the
same and much higher K-value, namely 1.3  105 M1. These
authors quantified unbound lidocaine in an indirect way. First,
the ligand was precipitated with ammonium reineckate, afterwards
it was dissolved in acetone and determined spectrophotometri-
cally. Whether this approach, other experimental differences, or
the fact that the study was performed at 4 °C is the reason for the
proposal of two binding sites with a high K-value is not clear.
We also tried to study lidocaine binding by isothermal titration
calorimetry, but reliable binding constants could not be deter-
mined due to insufficient release of heat in our experimental
 K.L. Hein et al. / Journal of Structural Biology 171 (2010) 353–360 359

set-up (data not shown). This indicates a very low affinity, in accor-
dance with the above estimates.

4. Discussion

HSA binding of cationic organic ligands has not been studied

much, and only a few and fairly simple studies have dealt with
lidocaine binding. Thus, Sawinski and Rapp have reported binding
to two sites with the same K-value (Sawinski and Rapp, 1963),
whereas other studies have suggested low-affinity binding to only
one site (Krauss et al., 1986; Bailey and Briggs, 2004). The present
results are in accordance with the latter binding model. The equi-
librium dialysis results propose binding to one site with a K-value
of 3.1  102–3.2  102 M1. These results are supported by those
obtained by isothermal titration calorimetry and intrinsic
fluorescence.
Crystallographic analysis of HSA complexed with lidocaine re-
vealed binding of the drug to a superficially placed site in subdo-
main IB (Fig. 2). Binding is mainly the result of cation–p
interactions with Arg114, polar interaction with Lys190 and elec-
trostatic attraction to Asp187. A possible explanation for the weak
drug binding is the predominant presence of the positively charged
residues in the binding site. Hydrophobic interactions do not con-
tribute to binding. This finding is in contrast to binding of cationic
aliphatic detergents. That type of ligands binds with the aliphatic
chain inserted in a hydrophobic cavity in the protein and with
the polar headgroup placed at the surface of the protein. The bind-
ing affinity is due to hydrophobic forces supported by polar inter-
actions at the protein surface (Peyre et al., 2005). Binding of drugs
with anionic or electronegative features also usually takes place in
cavities or pockets in the protein molecule, where hydrophobic
interactions can take place. This is the case whether they bind in
subdomain IB (e.g., fusidic acid (Zunszain et al., 2008)), in drug site
1 in subdomain IIA (e.g., warfarin (Ghuman et al., 2005), phenylbu-
tazone (Ghuman et al., 2005), indomethacin (Ghuman et al., 2005),
and salicylic acid (Zhu et al., 2008)), in drug site 2 in subdomain
IIIA (e.g., ibuprofen (Ghuman et al., 2005) and diazepam (Ghuman
et al., 2005)) or in subdomain IIIB (e.g., propofol (Bhattacharya
et al., 2000b)) (Fig. 6a). Therefore, the low affinity of lidocaine
Fig. 6. Overview of HSA binding sites. (a) Lidocaine (green), drug (pink) and fatty
binding probably also reflects the open nature of the binding site acid (blue) binding sites. (b) Interdomain cleft with bound decanoate (yellow, PDB
and the absence of hydrophobic interactions between ligand and 1E7E), L-thyroxine (pink, PDB 1HK4) and lidocaine (green). (c) Close-up of
protein. interdomain cleft and subdomain IB with bound fusidic acid (pink, PDB 2VUF),
In the following, the position of the lidocaine binding site has myristic acid (yellow, PDB 1HK4) and lidocaine (green). The colour codes used in (b)
and (c) for the domains are as follows: I, deep blue; II, marine blue; III, light blue.
been related to those of other sites identified by X-ray crystallogra-
phy. The most studied type of ligand is aliphatic fatty acids, and ten
sites have been identified for decanoate (Fig. 6a) seven of which are
common to fatty acids having from 10 to 18 C-atoms (stearate)
(Bhattacharya et al., 2000a). The sites numbered 8 and 9 can bind the opening of the interdomain cleft (Fig. 6a), it involves only res-
decanoate but not fatty acids of a longer aliphatic chain. They are idues from subdomain IB (Fig. 6a), and it is located further towards
placed in the large crevice between domains I and III (Fig. 6a and the back of the protein molecule (Fig. 6b).
b). Molecule number 8 binds close to the base of the crevice, Fusidic acid (Fig. 6a and c), azapropazone and indomethacin can
whereas number 9 is located further up and interacts with, among all bind within subdomain IB without involving Arg114, Asp187 or
others, Lys190 and Lys436. Binding of these two molecules can Lys190 (Ghuman et al., 2005; Zunszain et al., 2008). Also binding of
take place because of small but significant movements of domains aliphatic fatty acids (site FA1 in Fig. 6a) takes place in the subdo-
I and III, which cause a narrowing of the crevice between them. main without interacting with the three residues mentioned
One of the common sites (FA2 in Fig. 6a) is formed by residues (Fig. 6c) (Bhattacharya et al., 2000a). By contrast, when triiodoben-
from both subdomain IA and IIA, and fatty acid binding to this site zoic acid (Curry et al., 1998) and the large bilirubin (Zunszain et al.,
results in global changes in the HSA molecule (Bhattacharya et al., 2008) and hemin molecules bind in the cavity Lys190 is involved;
2000a). The changes create a new binding site for L-thyroxine in the latter ligand also interacts with Arg114 (Zunszain et al., 2003).
the cleft between domains I and III (Fig. 6a and b) (Petitpas et al., However, triiodobenzoic acid, bilirubin and hemin bind in a cavity
2003). The site is rather open, and it is on the one side formed within the subdomain, whereas lidocaine binds to its outer surface
by the upper part of the long helix that connects domains I and II (Fig. 6c). Thus, the lidocaine binding site differs from those previ-
and on the other side by a pair of helices of subdomain IIIA. As seen ously reported. This finding implies that it is unlikely that even
from the illustrations, the lidocaine binding site differs from the high lidocaine concentrations competitively displace other ligands
decanoate and L-thyroxine sites, because it is placed closer towards simultaneously binding to HSA.
360 K.L. Hein et al. / Journal of Structural Biology 171 (2010) 353–360
The relatively large conformational changes observed for HSA
(Fig. 4) are most probably caused by differences in crystal packing
of the protein. By contrast, the changes accompanying binding of
lidocaine (Fig. 2f) and the sulphate ion (Fig. 3c) are minor and more
local. However, lidocaine binding also results in small conforma-
tional changes in the part of subdomain IIA which includes
Trp214. Intrinsic fluorescence measurements with solubilised
HSA support the view that these changes indeed are caused by
lidocaine binding (Fig. 5) and not by crystal packing. Apparently,
X-ray crystallographic analyses of HSA complexed with negatively
charged drugs (Ghuman et al., 2005), bilirubin (Zunszain et al.,
2008) or hemin (Zunszain et al., 2003) only revealed local confor-
mational changes at the respective binding sites. The only ligands
found so far which cause global changes in HSA are aliphatic fatty
acids (Curry et al., 1998). These changes are primarily caused by
fatty acid binding to site FA2 (Fig. 6a) and show as rotations of do-
mains I and III relative to domain II.
The characteristics of lidocaine binding to HSA differ from those
found for lidocaine binding to the a-subunit of Na+ channels. Kur-
oda et al. found, by using 1H NMR spectroscopy and mutagenesis,
p–p stacking of the aromatic rings of lidocaine and a phenylalanine
(Phe1489) and electrostatic attraction of the tertiary amine nitro-
gen to a glutamic acid (Glu1492) in the channel protein isolated
from rat brain (Kuroda et al., 2000). Using electrophysiology and
mutagenesis, Ahern et al. found that cation–p interaction between
lidocaine and a phenylalanine (Phe1579) is essential for drug bind-
ing to a cardiac Na+ channel expressed in Xenopus oocytes (Ahern
et al., 2008). Thus, in HSA the aromatic portion of lidocaine inter-
acts with an arginine instead of a hydrophobic phenylalanine,
and the electrostatic interaction is to an aspartate and not to a glu-
tamate. Furthermore, in HSA the binding is supplemented by a po-
lar interaction of the carbonyl group of lidocaine to a lysine residue
(Fig. 2).
The insight into the HSA–lidocaine interaction reported in this
study provides a preview into the structural basis of binding of cat-
ionic ligands and contributes to a greater understanding of the un-
ique binding properties of the protein. It will also facilitate efforts
to modify new therapeutics to control their interaction with HSA
and thereby optimise their distribution within the human body.
Acknowledgments
This work was supported by the A.P. Møller Foundation for the
Advancement of Medical Science and by a Hallas–Møller stipend of
the Novo-Nordisk Foundation.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.jsb.2010.03.014.
References
Adams, P.D., Grosse-Kunstleve, R.W., Hung, L.W., Ioerger, T.R., McCoy, A.J., Moriarty,
N.W., Read, R.J., Sacchettini, J.C., Sauter, N.K., Terwilliger, T.C., 2002. PHENIX:
building new software for automated crystallographic structure determination.
Acta Crystallogr. D Biol. Crystallogr. 58, 1948–1954.
Ahern, C.A., Eastwood, A.L., Dougherty, D.A., Horn, R., 2008. Electrostatic
contributions of aromatic residues in the local anesthetic receptor of voltage-
gated sodium channels. Circ. Res. 102, 86–94.
Bailey, D.N., Briggs, J.R., 2004. The binding of selected therapeutic drugs to human
serum a-1 acid glycoprotein and to human serum albumin in vitro. Ther. Drug
Monit. 26, 40–43.
Bhattacharya, A.A., Grüne, T., Curry, S., 2000a. Crystallographic analysis reveals
common modes of binding of medium and long-chain fatty acids to human
serum albumin. J. Mol. Biol. 303, 721–732.
Bhattacharya, A.A., Curry, S., Franks, N.P., 2000b. Binding of the general anesthetics
propofol and halothane to human serum albumin. High resolution crystal
structures. J. Biol. Chem. 275, 38731–38738.
Brünger, A.T., Adams, P.D., Clore, G.M., DeLano, W.L., Gros, P., Grosse-Kunstleve,
R.W., Jiang, J.S., Kuszewski, J., Nilges, M., Pannu, N.S., Read, R.J., Rice, L.M.,
Simonson, T., Warren, G.L., 1998. Crystallography & NMR system: a new
software suite for macromolecular structure determination. Acta Crystallogr. D
Biol. Crystallogr. 54, 905–921.
Carter, D.C., Ho, J.X., 1994. Structure of serum albumin. Adv. Protein Chem. 45, 153–
203.
Chen, R.F., 1967. Removal of fatty acids from serum albumin by charcoal treatment.
J. Biol. Chem. 242, 173–181.
Cowtan, K., 2006. The Buccaneer software for automated model building. 1. Tracing
protein chains. Acta Crystallogr. D Biol. Crystallogr. 62, 1002–1011.
Curry, S., 2004. Plasma albumin as a fatty acid carrier. Adv. Mol. Cell. Biol. 33, 29–46.
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.
Delano, W.L., 2002. The PyMOL Molecular Graphics System. DeLano Scientific, San
Carlos, CA, USA.
Diederichs, K., Karplus, P.A., 1997. Improved R-factors for diffraction data analysis in
macromolecular crystallography. Nat. Struct. Biol. 4, 269–275.
Emsley, P., Cowtan, K., 2004. Coot: model-building tools for molecular graphics.
Acta Crystallogr. D Biol. Crystallogr. 60, 2126–2132.
Ferrer, M.L., Duchowicz, R., Carrasco, B., de la Torre, J.G., Acuna, A.U., 2001. The
conformation of serum albumin in solution: a combined phosphorescence
depolarization–hydrodynamic modelling study. Biophys. J. 80, 2422–2430.
Ghuman, J., Zunszain, P.A., Petitpas, I., Bhattacharya, A.A., Otagiri, M., Curry, S., 2005.
Structural basis of the drug-binding specificity of human serum albumin. J. Mol.
Biol. 353, 38–52.
He, X.M., Carter, D.C., 1992. Atomic structure and chemistry of human serum
albumin. Nature 358, 209–215.
Kabsch, W., 1993. Automatic processing of rotation diffraction data from crystals of
initially unknown symmetry and cell constants. J. Appl. Crystallogr. 26, 795–800.
Kragh-Hansen, U., 1981. Molecular aspects of ligand binding to serum albumin.
Pharmacol. Rev. 33, 17–53.
Kragh-Hansen, U., Hellec, F., de Foresta, B., le Maire, M., Møller, J.V., 2001.
Detergents as probes of hydrophobic binding cavities in serum albumin and
other water-soluble proteins. Biophys. J. 80, 2898–2911.
Krauss, E., Polnaszek, C.F., Scheeler, D.A., Halsall, H.B., Eckfeldt, J.H., Holtzman, J.L.,
1986. Interaction between human serum albumin and a1-acid glycoprotein in
the binding of lidocaine to purified protein fractions and sera. J. Pharmacol. Exp.
Ther. 239, 754–759.
Kuroda, Y., Miyamoto, K., Tanaka, K., Maeda, Y., Ishikawa, J., Hinata, R., Otaka, A.,
Fujii, N., Nakagawa, T., 2000. Interactions between local anesthetics and Na+
channel inactivation gate peptides in phosphatidylserine suspensions as
studied by1H-NMR spectroscopy. Chem. Pharm. Bull. 48, 1293–1298.
Lejon, S., Cramer, J.F., Nordberg, P., 2008. Structural basis for the binding of
naproxen to human serum albumin in the presence of fatty acids and the GA
module. Acta Crystallogr. F64, 64–69.
McCoy, A.J., Grosse-Kunstleve, R.W., Storoni, L.C., Read, R.J., 2005. Likelihood-
enhanced fast translation functions. Acta Crystallogr. D Biol. Crystallogr. 61,
458–464.
Olivieri, J.R., Craievich, A.F., 1995. The subdomain structure of human serum
albumin in solution under different pH conditions studied by small angle X-ray
scattering. Eur. Biophys. J. 24, 77–84.
Perrakis, A., Harkiolaki, M., Wilson, K.S., Lamzin, V.S., 2001. ARP/wARP and
molecular replacement. Acta Crystallogr. D Biol. Crystallogr. 57, 1445–1450.
Peters Jr., T., 1996. All About Albumin: Biochemistry, Genetics, and Medical
Applications. Academic Press, San Diego, CA, USA.
Petitpas, I., Bhattacharya, A.A., Twine, S., East, M., Curry, S., 2001. Crystal structure
analysis of warfarin binding to human serum albumin. Anatomy of drug site I. J.
Biol. Chem. 276, 22804–22809.
Petitpas, I., Petersen, C.E., Ha, C.-E., Bhattacharya, A.A., Zunszain, P.A., Ghuman, J.,
Bhagavan, N.V., Curry, S., 2003. Structural basis of albumin–thyroxine
interactions and familial dysalbuminemic hyperthyroxinemia. Proc. Natl.
Acad. Sci. USA 100, 6440–6445.
Peyre, V., Lair, V., André, V., le Maire, G., Kragh-Hansen, U., le Maire, M., Møller, J.V.,
2005. Detergent binding as a sensor of hydrophobicity and polar interactions in
the binding cavities of proteins. Langmuir 21, 8865–8875.
Sawinski, V.J., Rapp, G.W., 1963. Interaction of human serum proteins with local
anesthetic agents. J. Dent. Res. 42, 1429–1438.
Schönfeld, D.L., Ravelli, R.B., Mueller, U., Skerra, A., 2008. The 1.8-A crystal structure
of alpha(1)-acid glycoprotein (orosomucoid) solved by UV RIP reveals the broad
drug-binding activity of this human plasma lipocalin. J. Mol. Biol. 384, 393–405.
Sugio, S., Kashima, A., Mochizuki, S., Noda, M., Kobayashi, K., 1999. Crystal structure
of human serum albumin at 2.5 Å resolution. Protein Eng. 12, 439–446.
Truant, A.P., Takman, B., 1959. Differential physical–chemical and
neuropharmacologic properties of local anesthetic agents. Anesth. Analg. 38,
478–484.
Zhu, L., Yang, F., Chen, L., Meehan, E.J., Huang, M., 2008. A new drug binding subsite
on human serum albumin and drug–drug interaction studied by X-ray
crystallography. J. Struct. Biol. 162, 40–49.
Zunszain, P.A., 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.
Zunszain, P.A., Ghuman, J., McDonagh, A.F., Curry, S., 2008. Crystallographic analysis
of human serum albumin complexed with 4Z,15E-bilirubin-IXa. J. Mol. Biol.
381, 394–406.