Binding of a chiral drug to a protein: An investigation of the 2-(3-benzoylphenyl)propionic acid/bovine serum albumin system by circular dichroism and fluorescence

Article abstract of DOI:10.1039/b509911k

A combined approach using global analysis of circular dichroism multiwavelength data and time resolved fluorescence was applied to investigate the interaction of R-(-)- and S-(+)-ketoprofen with bovine serum albumin in buffer solution at neutral pH. A characterization of the most stable drug : protein adducts of 1:1 and 2:1 stoichiometry, as individual chemical species, was obtained. The stability constants and the absolute circular dichroism spectra of the diastereomeric complexes were determined. The spectra of the 1:1 conjugates are opposite in sign, those of the 2:1 complexes are both negative, but different in shape from each other (peaks at 358 and 342 nm for S-(+)- and R-(-)-ketoprofen, respectively). A tryptophan residue was shown to be involved in the binding of the drug, in the primary site for the R-(-) and in the secondary site for the S-(+) enantiomer, thereby showing that chiral recognition by the protein causes the site of highest affinity being not the same for both optical antipodes. The Owner Societies 2005.

Full text of DOI:10.1039/b509911k

View Article Online / Journal Homepage / Table of Contents for this issue
R E S E A R C H P A P E R
Binding of a chiral drug to a protein: an investigation of the
2-(3-benzoylphenyl)propionic acid/bovine serum albumin system
by circular dichroism and fluorescence
Sandra Monti,*a Francesco Manoli,a Salvatore Sortino,b Raffaele Morronec and
Giovanni Nicolosic
a
`
Istituto per la Sintesi Organica e la Fotoreattivita-ISOF, CNR Area della Ricerca, Via P.
Gobetti 101, 40129 Bologna, Italy. E-mail: monti@isof.cnr.it; Fax: þþ39 051 639 9844;
Tel: þþ39 051 639 9813
b
`
Dipartimento di Scienze Chimiche, Universita di Catania, Viale A. Doria 8, I-95125 Catania,
Italy
c Istituto di Chimica Biomolecolare-Sezione di Catania, Via del Santuario 110,
I-95028 Valverde (CT), Italy
Received 12th July 2005, Accepted 20th September 2005
First published as an Advance Article on the web 4th October 2005
A combined approach using global analysis of circular dichroism multiwavelength data and time resolved
fluorescence was applied to investigate the interaction of R-(ꢀ)- and S-(þ)-ketoprofen with bovine serum albumin
in buffer solution at neutral pH. A characterization of the most stable drug : protein adducts of 1 : 1 and 2 : 1
stoichiometry, as individual chemical species, was obtained. The stability constants and the absolute circular
dichroism spectra of the diastereomeric complexes were determined. The spectra of the 1 : 1 conjugates are
opposite in sign, those of the 2 : 1 complexes are both negative, but different in shape from each other (peaks at
358 and 342 nm for S-(þ)- and R-(ꢀ)-ketoprofen, respectively). A tryptophan residue was shown to be involved
in the binding of the drug, in the primary site for the R-(ꢀ) and in the secondary site for the S-(þ) enantiomer,
thereby showing that chiral recognition by the protein causes the site of highest affinity being not the same for
both optical antipodes.
ques probing at solid-liquid interfaces, due to the complexity of
the sample organization and difficult interpretation of data.
Introduction
The nature and the strength of non-covalent interactions
Although using spectroscopic methods to determine the
between molecules is at the basis of a large number of
number of binding sites and binding constants can be proble-
phenomena of contemporary interest in chemistry, material
matic in general, in the present contribution we show how the
science and biology. For control of selectivity and maximiza-
tion of ligand-substrate binding energies understanding of the
analysis and time resolved fluorescence to the system 2-(3-
combined application of circular dichroism (CD)w with global
factors which govern the intermolecular interactions is a major
issue.1
In biosystems, where the complex structural organization of
benzoyl-phenyl)propionic acid (ketoprofen, KP, see Chart 1)
and bovine serum albumin (BSA) at KP/BSA ‘‘low’’ molar
ratios (see below) allowed to reach a description of the chiral
biomacromolecules in helices or sheets display ‘‘handedness’’,
recognition ability of the protein and a characterization of the
ligand chirality plays a crucial role. This holds true in parti-
binding sites of highest affinity.
cular for drugs which display their action as pure optical
The partner molecules of this study were chosen because of
isomers. In these systems knowledge about stereospecific inter-
their favourable features with respect to both binding and
chemical reactivity. KP is a photosensitive drug,3,4 pharmaco-
actions with biomolecules is a prerequisite for design, simpli-
fication of pharmacological profiles and elimination of the
so-called ‘‘isomeric ballast’’.2
logically active almost exclusively in the S-(þ) form. Its photo-
reactivity within a protein matrix, due to the presence of the
benzophenone moiety, has strong relevance to photoaffinity
labelling applications5,6 and to therapeutical use of the drug
itself (photoallergic side-reactions are known to be mediated by
A step to gain insight into the biochemical consequences of
drug optical asymmetry can be taken through the study of their
interaction with proteins, in the aim to obtain a characteriza-
tion of the diastereomeric adducts, as individual chemical
species in solution. In this respect, gaining detailed spectro-
scopic information, specific of the ligand-receptor site, is a key
point. Such information cannot be achieved in the context of
binding studies using conventional separation techniques.
Moreover it also is hardly provided by spectroscopic techni-
w Abbreviations: CD, circular dichroism; KP, 2-(3-benzoyl-phenyl)-
propionic acid (ketoprofen); NSAID, non-steroidal anti-inflammatory
drug; BSA, bovine serum albumin; HSA, human serum albumin; Trp,
tryptophan; Tyr, tyrosine.
Chart 1
4002
P h y s . C h e m . C h e m . P h y s . , 2 0 0 5 , 7 , 4 0 0 2 – 4 0 0 8
T h i s j o u r n a l i s & T h e O w n e r S o c i e t i e s 2 0 0 5

View Article Online
protein-drug covalent photoadducts).7,8 BSA is an effective
physiological carrier for a large variety of xenobiotics and a
potent chiral selector, able to discriminate between enantiomers
of many drugs (ref. 9 and citations therein). Previous studies of
stereospecific recognition of KP enantiomers by human serum
albumin (HSA, ref. 10 and citations therein) and BSA (refs.
10–12 and citations therein), have shown that the latter has the
best discriminating properties.
By CD titration experiments and application of a method of
global analysis of multiwavelength data, the binding constants
and the absolute CD spectra of the KP : BSA diastereomeric
adducts of 1 : 1 and 2 : 1 stoichiometry were determined by
exploring the molar ratio range 0.4–2.2. A parallel time
resolved study of the intrinsic protein fluorescence, carried
out in the same experimental conditions used in the CD
experiments, allowed to further chacterize the binding sites of
the KP enantiomers and to gain insight into the chiral recogni-
tion ability of BSA at the molecular level.
column using n-hexane/dichloromethane (20 : 80, v : v) as
eluent. The compound was fully characterized by mass spectro-
1
metry (ESI-MS) and H-NMR.
Phosphate buffer 0.01 M at pH 7.4 was used as solvent for
spectroscopic and fluorescence measurements. Water was
purified by passage through a Millipore MilliQ system.
3-Ethylbenzophenone (3-EB) was obtained by photoirradia-
tion of a 10ꢀ3 M KP solution in 0.1 M phosphate buffer pH
7.4. A volume of 330 ml of solution was exposed to light from a
high pressure 150 W Xe lamp for ca. 2 h. Light components
below 324 nm were cut off. Magnetic stirring and argon
bubbling were maintained during the irradiation. After addi-
tion of NaOH 0.01 M and extraction with CH2Cl2 the main
photoproduct, 3-EB, was separated by column chromato-
1
graphy and fully characterized by H NMR.
Sample preparation
A series of identical aliquots of BSA were weighted and
dissolved each in identical volumes of either pure buffer or
KP solutions at the required concentration. The samples were
gently stirred up to complete protein dissolution and were
carefully protected from ambient light during the manipula-
tions in order to avoid photodegradation of ketoprofen, occur-
ring in aqueous solutions at neutral pH with very high
quantum yields.3,4 All the experiments described below were
performed at a constant BSA content of 3 mg mlꢀ1 (4.5 ꢁ 10ꢀ5
mol dmꢀ3). This value was chosen because it was in the linear
region of the dependence of the BSA ellipticity on the protein
concentration. The analysis of experiments performed at con-
stant drug but varying BSA concentration proved to be much
more difficult because of the dependence of the drug binding
constants on the protein concentration (see below).
Materials and methods
Materials
S-(þ)-2-(3-Benzoylphenyl)propionic acid (ketoprofen, KP),
racemic KP (rac-KP), bovine serum albumin (BSA, 99%,
essentially globulin and fatty acid free) were purchased from
Sigma.
Preparation of R-(ꢀ)-2-(3-Benzoylphenyl)propionic acid
The preparation of R-(ꢀ)-KP was accomplished by a biocata-
lytic procedure consisting of three sequential steps and exploi-
ting lipase from Candida antarctica, of known R stereo-
preference, adsorbed on acrylic resin (Novozyme 435).
Step I (esterification). KP (3 g, 11.8 mmol) was dissolved in
toluene (300 ml) containing triethyl orthoformate (5.8 ml, 35.4
mmol) and lipase (15 g). The mixture was placed into a
thermostated shaker (45 1C, 300 rpm) and the progress of the
reaction was monitored by chiral HPLC. After 2 h the con-
version reached 45% and the reaction was stopped and the
enzyme filtered off. The KP ethyl ester and the unreacted KP
were separated by selective partition with a sodium bicarbo-
nate solution. The organic phase, evaporated to dryness furn-
ished R-(ꢀ)-KP ethyl ester with 60% ee (1.4 g).
Spectroscopy and time resolved fluorescence
All the measurements were performed at 25 1C. Ultraviolet
absorption spectra were recorded on a Perkin-Elmer l49
spectrophotometer using 0.2 cm path cells.
Circular dichroism spectra were obtained with a Jasco J-715
spectropolarimeter in cells of 1 cm pathlength. The spectra were
registered in the 290–400 nm range, by using accumulation and
time integration for improvement of signal to noise ratios.
Fluorescence emission spectra were obtained on a Spex
Fluorolog 111A spectrofluorimeter. Right angle detection geo-
metry was used. Experiments were carried out in cells of 1 cm
pathlength. Fluorescence lifetimes were determined by means
of a time-correlated single photon counting system (IBH
Consultants Ltd.) in air-equilibrated solutions. The nanose-
cond flashlamp, filled with deuterium, was thyratron-driven at
40 kHz; the instrumental response function had a full width at
half maximum of 2 ns. The absorbance of the solutions was ca.
0.1–0.3 in the usual 1 cm cell at the excitation wavelength of
295 nm. Emission was collected at right angles. For optically
dense solutions a 0.5 cm path cell was used; the emitted light
was collected from the first millimeter of the excited solution.
Fluorescence decay profiles were analysed by the least-squares
method, assuming multiexponential decay functions and re-
convolution of the instrumental response. The software pack-
age was provided by IBH Consultants Ltd. The resolution limit
after deconvolution was B0.2 ns.
Step II (hydrolysis). R-(ꢀ)-KP ethyl ester (1.3 g, 4.6 mmol, ee
60%) was dissolved in acetonitrile (130 ml) containing water
(2 ml) and Candida antarctica lipase (6.5 g). The suspension
was incubated at 45 1C under shaking (300 rpm) and the
reaction continued for 45 h up to 65% conversion. Removal
of the enzyme, followed by solvent evaporation under reduced
pressure, left a residue that was partitioned between tert-
butylmethyl ether and sodium bicarbonate solution. The aqu-
eous phase, after acidification with H2SO4 N, was extracted
three times with tert-butylmethyl ether to give R-(ꢀ)-KP
(730 mg, ee 75%).
Step III (esterification). R-(ꢀ)-KP (700 mg, 2.7 mmol, ee
75%) was subjected to esterification for 80 min, using the same
conditions as described in the above step I, to give the
corresponding R-(ꢀ)-KP ethyl ester with 92% ee. After con-
ventional chemical hydrolysis, followed by crystallization using
benzene/hexane (3 : 10), R-(ꢀ)-KP was obtained (360 mg,
ee 497%; [a]D – 54.4, (c ¼ 1 CH2Cl2),13 þ54.4 (c ¼ 2.71
CH2Cl2)].
KP methyl ester was obtained by stirring overnight at 80 1C
a methanol solution of KP (5 ꢁ 10ꢀ3 M) in the presence of 1%
sulfuric acid. The solution was then evaporated and the
product was isolated by liquid chromatographic on silica gel
Calculations
The assessment of the best complexation model and the
determination of the association constants of the complexes
were performed by global analysis of CD spectra obtained at
10–12 different ketoprofen concentrations, using the whole
290–400 nm wavelength range for the calculations. The com-
mercially available computer program SPECFIT/32 (Spectrum
T h i s j o u r n a l i s & T h e O w n e r S o c i e t i e s 2 0 0 5
P h y s . C h e m . C h e m . P h y s . , 2 0 0 5 , 7 , 4 0 0 2 – 4 0 0 8
4003

View Article Online
Software Associates) was used for this purpose. The program is
based on singular value decomposition and non-linear regres-
sion modeling by the Levenberg–Marquardt method. No
equations were used, the procedure being completely numer-
ical. The starting hypothesis was that the system can be
described by contemporary presence of 1 : 1 and 2 : 1 KP :
BSA complexes and free components. In some case the further
presence of 3 : 1 complexes was considered. The CD spectra of
the free KP enantiomers were introduced in the calculation as
known data. The CD signal of the protein alone, dissolved in
the buffer, was subtracted from the total CD signal of each of
the BSA-KP mixtures, before analysis. No assumptions on the
binding mechanism, i.e. about interdependency of binding sites
or cooperativity in multiple binding, were introduced. The
goodness of fit was judged on the basis of the sum of squared
residuals, residual distribution and standard deviation. The
output consisted of the best association constants and the
absolute spectra of the complexes.
Results
Circular dichroism of KP enantiomers and BSA
The CD spectra of S-(þ)- and R-(ꢀ)-KP in phosphate buffer
0.01 M, pH 7.4, at 25 1C, in the region 290–400 nm are
reported in the inset of Fig. 1A. The De(l) is characterized by
a weak band at ca. 340–350 nm, corresponding to the aromatic
carbonyl n,p* S1 ’ S0 transition, positive for S-(þ)- and
negative for R-(ꢀ)-KP, and a weak negative signal around
300 nm, corresponding to the onset of the p,p* S2 ’ S0 band
(3). The origin of such a small intrinsic CD in the lowest energy
band has been attributed to an interconversion of chiral
conformers promoted by the flexibility of the benzophenone
moiety.11 The ionization of the carboxyl group of the propionic
substituent, completely dissociated at this pH (pKa ¼ 4.7, ref.
14), is not expected to influence significantly the CD in this
spectral region, due to the peripheral location of the propionic
acid functionality, containing the asymmetric carbon center,
with respect to the benzoylphenyl moiety.
BSA in the same wavelength range, exhibits a negative CD,
corresponding to the onset of the strong negative protein bands
at 209/222 nm.15 The ellipticity follows a linear relationship vs.
the BSA concentration up to 5 ꢁ 10ꢀ5 mol dmꢀ3. Above this
limit a deviation from linearity was observed and was attributed
to the occurrence of molecular aggregation.15 This process affects
the binding ability of the BSA molecules, making unfeasible the
determination of the binding constants by means of titration
experiments with constant KP and variable BSA concentration.
Indeed concentration dependent binding constants were reported
for HSA.10,16 For this reason all the experiments described below
were performed at a constant BSA concentration.
Fig. 1 A. Ellipticity changes (Dy) obtained by titration of BSA 4.5 ꢁ
10ꢀ5 mol dmꢀ3 with S-(þ)-KP in the range 2 ꢁ 10ꢀ5 mol dmꢀ3–1.2 ꢁ
10ꢀ4 mol dmꢀ3. Phosphate buffer 0.01 mol dmꢀ3 pH 7.4, 25 1C, cell 1
cm. The signal of BSA alone was subtracted. The spectrum of BSA was
taken at protein 4.5 ꢁ 10ꢀ5 mol dmꢀ3. B. Comparison between
experimental and calculated ellipticity changes (Dy) for the system
BSA-S-(þ)-KP. C. Circular dichroism spectra (DDe) of S-(þ)-KP :
BSA associates. Inset of part A: circular dichroism (De of R-(ꢀ)- and S-
(þ)-KP and of BSA (right y-axis) in phosphate buffer 0.01 mol dmꢀ3
pH 7.4, 25 1C.
,
the values of the binding constants (Table 1). A complexation
model with simultaneous presence of both 1 : 1 and 2 : 1 KP :
BSA complexes was found to apply quite well. The agreement
obtained between calculated and experimental ellipticity values
can be appreciated in Fig. 1B and 2B for key wavelengths.z
Circular dichroism of BSA-KP complexes
Fig. 1A and 2A show the effect of addition of S-(þ)- and R-
(ꢀ)-KP at various concentrations on the ellipticity of a 4.5 ꢁ
10ꢀ5 mol dmꢀ3 BSA solution. In these experiments molar
ratios KP/BSA were kept between B0.4 and 2.2 to limit the
number of protein sites occupied by the drug.11,12 The intrinsic
y signal of BSA was subtracted to better evidence the changes
due to the formation of the associates. The ellipticity of KP in
the 320–380 nm region is much stronger in the protein matrix
than in buffer and is characterized by a well defined vibrational
progression of ca. 1200 cmꢀ1. This structure is typical of a,b-
unsaturated asymmetric ketones in non-polar solvents,17 and
has been attributed to a strong rigidity imposed to the ben-
zoylphenyl moiety by the protein matrix and to a hydrophobic
character of the environment of the ketone.11
z A model involving the concomitant presence of 3 : 1 KP : BSA
complexes was also checked. For S-(þ)-KP the best fit to experimental
data did not improve in quality, clearly indicating that there was no
significant amount of such complex in the used molar ratio interval, in
agreement with the presence of a group of two binding sites with a
substantially higher affinity with respect to additional sites. Accord-
ingly, in HSA a factor ca. 100 in the mean affinity constants for racemic
KP was reported to exist between a group of two primary sites and a
group of six-seven secondary sites.18 In the case of R-(ꢀ)-KP an
acceptable fit was achieved under this hypothesis also, but it did not
correspond to actual convergence in the optimisation procedure.
Relevant binding parameters were characterized by large errors and
were therefore discarded. On the basis of these results it was concluded
that, in the conditions of the experiment, formation of a complex where
a BSA molecule is charged with three R-(ꢀ)-KP molecules is not
needed for the reproduction of the CD signals.
Each set of experiments were globally analysed using an
iterative numerical procedure (see Experimental). This analysis
afforded the stoichiometry of the most stable complexes and
4004
P h y s . C h e m . C h e m . P h y s . , 2 0 0 5 , 7 , 4 0 0 2 – 4 0 0 8
T h i s j o u r n a l i s & T h e O w n e r S o c i e t i e s 2 0 0 5

View Article Online
negative sign for S-(þ)- and R-(ꢀ)-KP, respectively; the 290–
300 nm region is characterized by a positive sign for both
isomers. In the 2 : 1 spectra the carbonyl region has negative
sign with absolute minima located at 358 and 342 nm for the
S-(þ)- and R-(ꢀ)-KP, respectively; a positive signal at 290 nm,
an isoelliptic point at 294 nm and a negative peak at 300 nm are
observed with S-(þ)-KP, whereas a positive band at 290 nm
and an isoelliptic point at 300 nm are observed with R-(ꢀ)-KP.
Association of a racemic KP was studied by CD titration at
4.5 ꢁ 10ꢀ5 mol dmꢀ3 BSA and total drug concentration
varying from 2 ꢁ 10ꢀ5 mol dmꢀ3 to 1.0 ꢁ 10ꢀ4 mol dmꢀ3
(10 concentrations, data not shown). The data were globally
analysed using a model based on the contemporary presence of
the 1 : 1 complexes of each enantiomer and a 1 : 1 : 1 BSA : R-
(ꢀ)-KP : S-(þ)-KP mixed complex. The association constants
of the 1 : 1 complexes were fixed to the values of Table 1 and
the relevant molar ellipticities were those of Fig. 1C and 2C.
The binding constant of the 1 : 1 : 1 complex was optimised as
parameter. An excellent agreement between calculated and
experimental ellipticity values was found with log(K111/molꢀ2
dm6) ¼ 9.92 ꢂ 0.15. The calculated DDe values for the 1 : 1 : 1
complex showed features of both the 1 : 1 and the 2 : 1 species,
i.e. negative sign in the n,p* region, overall intensity close to
those of the diastereomeric 2 : 1 associates, similar size for the
342 and 358 nm vibronic bands, a very weak negative band at
300–305 nm, a positive sign at 290 nm.
The interaction of KP-methyl ester with the protein was also
examined by CD titration in the same experimental conditions
used for KP in the carboxylate form. A racemic mixture of the
ester was used. The generated signal was unstructured and
characterized by a positive band in the 310–400 nm n,p* region
with maximum at 335 nm, in complete agreement with pre-
vious findings.20 These results were taken as an indication for
the involvement of the carboxylate group in the binding of KP
enantiomers to BSA. The same conclusion had been reached
with flavoprotein.21
Fluorescence of BSA-KP complexes
To check for a possible involvement of aromatic aminoacid
residues in the binding of KP to BSA, the protein fluorescence
was studied at increasing KP concentrations, with medium and
BSA concentration identical to those used in the CD experi-
ments. With excitation at 295 nm the fluorescence of BSA
(lmax ¼ 344 nm) is essentially due to the emission of trypto-
Fig. 2 A. Ellipticity changes (Dy) obtained by titration of BSA 4.5 ꢁ
10ꢀ5 mol dmꢀ3 with R-(ꢀ)-KP in the range 2 ꢁ 10ꢀ5 mol dmꢀ3–1.2 ꢁ
10ꢀ4 mol dmꢀ3. Phosphate buffer 0.01 mol dmꢀ3 pH 7.4, 25 1C, cell 1
cm. The signal of BSA alone was subtracted. B. Comparison between
experimental and calculated ellipticity changes (Dy) for the system R-
(ꢀ)-KP-BSA. C. Circular dichroism spectra (DDe) of R-(ꢀ)-KP : BSA
associates.
1
phan (Trp), whose La excited state is located at this energy.22
Steady state emission measurements cannot give information
on the protein–drug interaction, because competitive light
absorption by the non-fluorescent ketone introduced a trivial
effect of quenching of the protein emission intensity. Therefore
time resolved measurements were needed.
Comparison of the Kij values of Table 1 with literature data
is not easily feasible, since the experimental conditions (in
particular the protein concentration and the medium) critically
influence the binding process;10 moreover, average association
constants for groups of sites were generally reported.18,19 We
notice that the values of K11 compare well with those for the
binding of the KP enantiomers to the highest affinity site of
HSA (at protein 1 mg mlꢀ1, log K/dm3 molꢀ1 was 6.3 and 5.8,
for the R-(ꢀ) and the S-(þ) enantiomers, respectively).10
In Fig. 1C and 2C we report the absolute induced CD in
DDe, reconstructed on the basis of the Kij values (Table 1). In
the 1 : 1 spectra the 310–400 nm region exhibits positive and
A 4.5 ꢁ 10ꢀ5 mol dmꢀ3 BSA solution was excited at 295 nm
and the fluorescence was collected at 330 and 400 nm, i.e. on
the blue and red side of the emission band maximum. The
native protein exhibited good biexponential emission decays,
P
with lifetimes ti and relative amplitudes (Aiti/ jAjtj, with Ai
preexponential factors) slightly different at the two detection
wavelengths: t1 ¼ 3.9 ns (20%) and t2 ¼ 6.7 ns (80%) (w2 ¼
1.09) at lem ¼ 330 nm, t1 ¼ 4.2 ns (27%) and t2 ¼ 7.5 ns (73%)
(w2 ¼ 1.19) at lem ¼ 400 nm. This effect is well known in
protein fluorescence and is attributed to heterogeneity in the
Trp environments.23,24 However we notice that this effect was
rather small in our conditions.
Table 1 Association constants of the two most stable diastereomeric
KP : BSAcomplexes in phosphate buffer 0.01 mol dmꢀ3 pH 7.4 at 25 1C
1 : 1
log(K11/dm3 molꢀ1
2 : 1
log(K21/dm6 molꢀ2
)
Addition of S-(þ)- and R-(ꢀ)-KP did not introduce further
components in the decays and a biexponential analysis remained
appropriate. Of the two lifetimes of the native protein, t2 was
slightly affected and t1 was reduced more and more at increasing
KP concentrations (Table 2). The dependence of the rate
constants k1 ¼ 1/t1 and k2 ¼ 1/t2, measured at lem ¼ 330 nm,
on the S-(þ)- and R-(ꢀ)-KP concentrations is reported in
)
R-(ꢀ)-KP : BSAa
5.1 ꢂ 0.5
5.3 ꢂ 0.3
9.0 ꢂ 0.4
9.6 ꢂ 0.3
S-(þ)-KP : BSAa
BSA, 4.5 ꢁ 10ꢀ5 mol dmꢀ3 (3 mg/ml); KP/BSA molar ratio range,
a
0.4–2.2.
T h i s j o u r n a l i s & T h e O w n e r S o c i e t i e s 2 0 0 5
P h y s . C h e m . C h e m . P h y s . , 2 0 0 5 , 7 , 4 0 0 2 – 4 0 0 8
4005

View Article Online
Table 2 Lifetimes from biexponential analysis of fluorescence emis-
sion decays of BSA-KP mixtures at lexc 295 nm and lem 330 nm (w2
were all better than 1.2). BSA 4.5 ꢁ 10ꢀ5 mol dmꢀ3 in phosphate buffer
0.01 mol dmꢀ3 pH 7.4, 25 1C
KP Conc/mol dmꢀ3
t1/nsa
t2/nsa
S-(þ)-KP
0
3.8
3.4
3.1
2.9
2.3
7.1
6.4
6.3
6.2
5.6
3.0 ꢁ 10ꢀ5
4.0 ꢁ 10ꢀ5
6.0 ꢁ 10ꢀ5
1.2 ꢁ 10ꢀ4
R-(ꢀ)-KP
0
3.8
3.0
2.5
2.7
1.8
2.1
1.7
1.8
1.8
1.9
1.8
7.1
6.7
6.5
6.5
6.1
6.1
5.8
5.6
5.5
5.5
5.5
5.0 ꢁ 10ꢀ6
1.0 ꢁ 10ꢀ5
2.0 ꢁ 10ꢀ5
3.0 ꢁ 10ꢀ5
4.0 ꢁ 10ꢀ5
6.0 ꢁ 10ꢀ5
9.0 ꢁ 10ꢀ5
1.1 ꢁ 10ꢀ4
1.2 ꢁ 10ꢀ4
1.4 ꢁ 10ꢀ4
Fig. 4 A. Dependence of the rate constants k1 and k2 extracted from
biexponential analysis of the BSA fluorescence decay at 330 nm (lexc
¼
295 nm) on the concentration of R-(ꢀ)-KP in the range 2 ꢁ 10ꢀ5 mol
dmꢀ3–1.2 ꢁ 10ꢀ4 mol dmꢀ3. BSA 4.5 ꢁ 10ꢀ5 mol dmꢀ3, phosphate
buffer 0.01 mol dmꢀ3 pH 7.4, 25 1C. B. Dependence of k1 on the
concentration of the 1 : 1 complex. C. Dependence of k1 on the
concentration of the 2 : 1 complex. Concentrations calculated accord-
ing to Kij of Table 1.
a
Uncertainty within 7%.
considered (Fig. 4C). It was concluded that most of the
quenching of Trp emission is produced in a specific site which
is the secondary binding site for S-(þ)-KP and the primary one
for R-(ꢀ)-KP.
Fig. 3A and 4A. As to k1, a steep increase with a substantially
linear course for S-(þ)-KP and a saturating behavior for R-(ꢀ)-
KP were found. As to k2, the overall variations were r20%.
Similar results were obtained for lem ¼ 400 nm.
The slopes of the linear plots in Fig. 3C and 4B represent
apparent quenching constants for the Trp singlet excited state
two-three orders of magnitude larger than those expected for a
collisional process (1010 dm3 molꢀ1 ꢀ1) and therefore are
s
indicative of a ground state interaction of the drug with the
protein.22 The small changes observed on k2 are consistent with
an indirect effect.
A deeper insight into the meaning of these data was obtained
by examining the dependence of k1 on the concentrations of the
1 : 1 and 2 : 1 complexes, calculated on the basis of the
equilibrium constants of Table 1. For S-(þ)-KP k1 did not
correlate with the concentration of the 1 : 1 complex (Fig. 3B),
but was linearly dependent on that of the 2 : 1 complex (slope is
7.1 ꢁ 1012 dm3 molꢀ1 sꢀ1, Fig. 3C). On the contrary for R-(ꢀ)-
KP k1 increased linearly with the concentration of the 1 : 1
complex up to its maximum value (Fig. 4B, slope is (9.3 ꢂ 1.1)
Discussion
Serum albumins are known to contain two main sites, specia-
lized for binding of drugs of medium size, such as most of the
NSAIDs. The structure of these sites was determined by X-ray
crystallography in HSA25,26 and can be assumed to be similar
in BSA, whose X-ray structure is not available, because the two
proteins are 80% homologue as regards the primary se-
quence.15 These two sites, previously defined as site I and site
II by Sudlow et al.,27 are located in subdomains IIA and IIIA,
respectively.25,26
ꢁ 1012 dm3 molꢀ1
s
ꢀ1), whereas it showed a sharp increase
below 3 ꢁ 10ꢀ6 mol dmꢀ3 followed by a wide plateau up to 2 ꢁ
10ꢀ5 mol dmꢀ3, if the concentration of the 2 : 1 complex was
BSA is endowed by two Trp residues; one of them, namely
Trp 134, is located in subdomain IA, in a region not directly
involved in binding of NSAIDs, whereas the other one, i.e. Trp
212, is placed in site I (subdomain IIA).15,25,26 Although
multiexponential fluorescence decays in proteins containing a
few tryptophans generally preclude assignement of each ex-
ponential component to a specific Trp residue,23,24 the biexpo-
nential Trp emission decay in BSA (see Table 2 and Fig. 3A
and 4A) can be reasonably attributed to the substantially
different microenvironments relevant to the two tryptophan
residues above. This simple interpretation has been already
adopted in the case of other two-tryptophan proteins.24 All this
considered, the decay kinetics of BSA in presence of KP
supports a scenario where the rate constant k2, little perturbed
by drug binding, is assigned to Trp 134, whereas the most
affected rate constant k1 is attributed to Trp 212. Alternatively,
by admitting the possibility of substantial heterogeneity in
protein conformations, k2 could be assigned to protein con-
formations not significantly involved in drug binding at these
molar ratios. Whatever the rationale for the k2 behavior, the
linear dependence of k1 on the concentration of the complexes
(Fig. 3C and 4B) leads to identify site I in subdomain IIA as the
Fig. 3 A. Dependence of the rate constants k1 and k2 extracted from
biexponential analysis of the BSA fluorescence decay at 330 nm (lexc
¼
295 nm) on the concentration of S-(þ)-KP in the range 2 ꢁ 10ꢀ5 mol
dmꢀ3–1.2 ꢁ 10ꢀ4 mol dmꢀ3. BSA 4.5 ꢁ 10ꢀ5 mol dmꢀ3, phosphate
buffer 0.01 mol dmꢀ3 pH 7.4, 25 1C. B. Dependence of k1 on the
concentration of the 1 : 1 complex. C. Dependence of k1 on the
concentration of the 2 : 1 complex. Concentrations calculated accord-
ing to Kij of Table 1.
4006
P h y s . C h e m . C h e m . P h y s . , 2 0 0 5 , 7 , 4 0 0 2 – 4 0 0 8
T h i s j o u r n a l i s & T h e O w n e r S o c i e t i e s 2 0 0 5

View Article Online
site of highest affinity for R-(ꢀ)-KP and the secondary one for
S-(þ)-KP. It is reasonable to admit that site II in subdomain
IIIA represents the main binding site for S-(þ)-KP and the
secondary one for R-(ꢀ)-KP. The satisfactory reproduction of
the CD of the rac-KP : BSA system (data not shown, see results
section), within the model of coexistence of 1 : 1 S-(þ)-KP :
BSA, 1 : 1 R-(ꢀ)-KP : BSA plus 1 : 1 : 1 S-(þ)-KP : R-(ꢀ)-KP :
BSA conjugates, furtherly support these conclusions. The fact
that chiral recognition of KP by BSA involves location of the
‘‘first bound’’ molecule of each drug enantiomer in different
parts of the biomolecule is directly relevant to the interpreta-
tion of displacement data in competitive binding experiments
involving rac-KP.28
The binding between KP and the protein matrix appears to
be stereoselective in both sites, where specific interactions and
essentially hydrophobic conditions are capable of imposing a
high rigidity to the aromatic carbonyl chromophore. This is
supported by the shape of the CD spectra of the diastereomeric
complexes, well distinct and structured with occupation of
either one or both protein sites (Fig. 1C and 2C). Chiral
recognition manifested by the spectra was not reflected in the
association constants for both the 1 : 1 and the 2 : 1 associates.
Indeed differences in Kij between S-(þ) and R-(ꢀ) did not
exceed the experimental uncertainty (Table 1).
It was previously proposed that in this system binding of KP
to a hydrophilic site of BSA would generate an unstructured
positive CD band with both enantiomers.11,12 This was not
confirmed by the present study. In our experimental conditions
an unstructured positive CD band developed only if the KP
solutions were not adequately protected from ambient light
during the manipulations and was attributed to sample photo-
degradation. Indeed KP in phosphate buffer at pH 7.4 under-
goes efficient photodecarboxylation with formation of 3-ethyl-
benzophenone (3-EB), as main photoproduct.3,4 CD of the 3-
EB : BSA system was actually characterized by an unstruc-
tured, positive band in the 300–380 nm region, with a maxi-
mum at 330 nm (data not shown).
An unstructured, positive CD, analogous to that with 3-EB,
was generated with the KP-ester. This indicates that the
association of KP enantiomers to BSA at neutral pH is assisted
by the ionic interaction between the carboxylate group of the
propionic substituent of the drug and a positively charged
NH31 group of the protein. Moreover, the DDe values at 290–
300 nm (Fig. 1C and 2C) support that aromatic aminoacids
(Trp or Tyr) are involved in the binding. A stacking interaction
via the p systems has already been proposed for the binding of
KP to flavoprotein.21
We notice that the correlation between CD changes and
specific interactions at each binding site is not straightforward
in the higher order complexes, because the overall protein
conformation could be modified upon sequential association
of drug molecules to subdomains IIA and IIIA, which share a
common interface.26 However, at least for the 1 : 1 BSA-KP
complexes, a theoretical rationalization of the stereospecificity
effects observed is possible. Indeed the optical activity induced
in a chromophore by a chiral environment can be described by
the sum of three terms, relying on one-electron, dipole–dipole
(d–d) and electric–magnetic interaction (m–m) mechanisms.
The first term derives from the interaction of the electric and
magnetic transition dipole moments relevant to different elec-
tronic states of the chomophore and represents a small con-
tribution in case of state mixing due to the electrostatic field of
the host cavity. The second term, d–d, is particularly suited to
describe the induced circular dichroism of guests with low-
lying, allowed p,p* states and has been widely utilised, in the
approximate form, using the polarizability of the host states.27
The third term, m–m, can be very important for symmetry-
forbidden or weakly allowed n,p* transitions, such as in the
present case, and indeed has been shown to adequately repro-
duce the circular dichroism of carbonyls29 and peptides.30
According to Tinoco,31 this term is expressed by:
ꢂ
ꢀ
ꢁꢃ
XX
mjb0na þ lj0bmi0anb
hðn2b ꢀ n2aÞ
R0aðmꢀm Þ ¼ ꢀ2
Im Via0;j0b mi0a
j¼i b¼a
ð1Þ
where
Via0,j0b ¼ lia0 ꢃ Tij ꢃ lj0b
(2)
and
"
#
3RijRij
R2ij
1
R3ij
Tij
¼
1 ꢀ
ð3Þ
In eqn (1) m and l stand for the magnetic transition dipole
moment of the guest and electric transition dipole moment of
the host, respectively, h for Planck’s constant and Im for
imaginary part. The sums extend over all the excited states a
and b, having frequencies of na and nb, and over the bonds i and
j of the chromophore and of the surrounding host residues,
respectively. The geometrical factor V of eqn (2) contains Tij,
the dipole interaction tensor, which is a direct function of the
distance Rij between the bonds i and j (eqn (3)), and is
responsible for the stereospecificity effects in host–guest bind-
ing. We notice that the structure of a variety of cyclodextrin
complexes, determined by this way, was accurate enough to
account for the spectroscopic, photophysical and photochemi-
cal behavior of such systems.32 A reliable calculation of this
term in the present case would require knowledge of the spatial
coordinates of the aminoacid residues surrounding the drug
chromophore in the protein pocket as well as optimisation of
the geometry of the protein–drug associated species. The X-ray
structure of HSA can be used to this purpose. This treatment,
which was outside the purpose of the present paper mainly
devoted to the experimental determination of the stereospeci-
ficity effects, is presently under consideration in our laboratory.
Conclusions
In this study, by an approach involving both spectroscopic and
photophysical methods, a specific characterization of the drug :
protein conjugates has been achieved. The knowledge gained
on the sites of binding and on the properties of the diaster-
eomeric complexes, can be used to gain insight into the
geometry of the association. Indeed the CD spectra of the
1 : 1 conjugates, remarkably different each other and, presum-
ably, not much influenced by conformational changes in the
protein moiety (which, on the contrary, could substantially affect
the spectral properties of the 2 : 1 species) can be combined with
suitable molecular mechanics and rotational strength calcula-
tions and give stereospecific structural information.
Finally, it worth noting that the photodegradation quantum
yield of R-(ꢀ)-KP had been reported to be ca. 40% higher than
that of S-(þ)-KP at molar KP : BSA ratio of 1.27.12 In the light
of the information gained in the present study, the photochem-
istry of KP enantiomers in the BSA matrix was reinvestigated
and found to be stereoselective and dependent on the stoichio-
metry of the association.33
Acknowledgements
We thank Dr Giancarlo Marconi (ISOF-CNR) for useful
discussions relevant to the theoretical interpretation of the
stereospecificity effects in circular dichroism. This work was
performed with financial support of the CNR of Italy within
the frame of the project ‘‘Ligand–receptor interactions in
model systems for cellular adhesion’’ and of MIUR (Rome,
Italy).
T h i s j o u r n a l i s & T h e O w n e r S o c i e t i e s 2 0 0 5
P h y s . C h e m . C h e m . P h y s . , 2 0 0 5 , 7 , 4 0 0 2 – 4 0 0 8
4007

View Article Online
17 R. E. Ballard, S. F. Mason and G. W. Vane, Discuss. Faraday
Soc., 1963, 33, 43–47.
References
18 L. Deschamps-Labat, F. Pehourcq, M. Jagou and B. Bannwarth,
J. Pharm. Biomed. Anal., 1997, 16, 223–229.
1
K. N. Houk, A. G. Leach, S. P. Kim and X. Zhang, Angew.
Chem., Int. Ed., 2003, 42, 4872–4897.
19 V. Russava, R. Rakovska, N. Stavreva, D. Mihailova and N.
Berova, Pharmazie, 1994, 49, 519–522.
20 H. Nagakawa, K. Shimizu and K. Yamada, Chirality, 1999, 11,
516–519.
21 N. Mano, Y. Oda, Y. Ishihama, H. Katayama and N. Asakawa,
J. Liq. Chromatogr. Relat. Technol., 1998, 21, 1311–1332.
22 J. R. Lakowicz, Principles of Fluorescence Spectroscopy, Kluwer
Academic/Plenum Publisher, New York, 2nd edn., 1999.
23 A. Grinvald and I. Z. Steinberg, Biochim. Biophys. Acta, 1976,
427, 663–678.
2
3
D. Burke and D. J. Henderson, Br. J. Anaesth., 2002, 88, 563–576.
S. Monti, S. Sortino, G. De Guidi and G. Marconi, J. Chem. Soc.,
Faraday Trans., 1997, 93, 2269–2275.
S. Monti, S. Sortino, G. De Guidi and G. Marconi, New J. Chem.,
1998, 599–604.
´
G. Dorman and G. Prestwich, Biochemistry, 1994, 33, 5661–5673.
V. T. G. Chuang, A. Kuniyasu, H. Nakayama, Y. Matsushita,
S. Hirono and M. Otagiri, Biochim. Biophys. Acta, 1999, 1434,
18–30.
J. Moser, Z. Sarabia, H. Minter, W. W. Lovell and G. M. J.
Beijersbergen van Henegouwen, J. Photochem. Photobiol., B,
2000, 58, 37–45.
4
5
6
7
24 J. M. Beechem and L. Brand, Annu. Rev. Biochem., 1985, 54,
43–71.
25 X. M. He and D. C. Carter, Nature, 1992, 358, 209–215.
26 D. C. Carter and J. X. Ho, Adv. Protein Chem., 1994, 45, 153–203.
27 G. Sudlow, D. J. Birkett and D. N. Wade, Mol. Pharmacol., 1976,
12, 1052–1061.
8
9
A. Lahoz, D. Hernandez, M. A. Miranda, J. Perez-Prieto, I. M.
Morera and J. Castell, Chem. Res. Toxicol., 2001, 14, 1486–1491.
X. Zhu, Y. Ding, B. Lin, A. Jakob and B. Koppenhoefer,
Electrophoresis, 1999, 20, 1869–1877.
28 C. Bertucci, Chirality, 2001, 13, 372–378.
10 N. Dubois, F. Lapicque, M. Abiteboul and P. Netter, Chirality,
1993, 5, 126–134.
11 M. Zandomeneghi, Chirality, 1995, 7, 446–451.
12 C. Festa, N. Levi-Minzi and M. Zandomeneghi, Gazz. Chim. Ital.,
1996, 126, 599–603.
29 J. A. Schellman, Acc. Chem. Res., 1968, 1, 144–151.
30 R. Woody and I. Tinoco Jr., J. Chem. Phys., 1967, 46, 4927–4945.
31 I. Tinoco Jr., Adv. Chem. Phys., 1962, 4, 113–160.
32 (a) P. Bortolus, G. Marconi, S. Monti, B. Mayer, G. Kohler and
¨
G. Grabner, Chem.–Eur. J., 2000, 6, 1578–1591; (b) P. Bortolus,
G. Marconi, S. Monti and B. Mayer, J. Phys. Chem. A, 2002, 106,
1686–1694; (c) S. Monti, G. Marconi, F. Manoli, P. Bortolus, B.
13 D. P. G. Hammon, R. A. Massy-Westropp and J. L. Newton,
Tetrahedron, 1995, 51, 12645–12660.
14 F. Bosca, M. A. Miranda, G. Carganico and D. Mauleon,
Photochem. Photobiol., 1994, 60, 96–101.
¨
Mayer, G. Grabner, G. Kohler, W. Boszcyck and K. Rotkiewicz,
Phys. Chem. Chem. Phys., 2003, 5, 1019–1026.
33 S. Monti, I. Manet, F. Manoli, R. Morrone, G. Nicolosi and
S. Sortino, Photochem. Photobiol. DOI: 10.1562/2005-07-11-
RA-608.
15 For a review on serum albumins, see, T. Peters, Adv. Protein
Chem., 1985, 37, 161–245.
16 S. Rendic, T. Alebic-Kolbah, F. Kajfez and V. Sunjic, Farmaco,
´ ´ ´
ꢀ
1978, 35, 51–59.
4008
P h y s . C h e m . C h e m . P h y s . , 2 0 0 5 , 7 , 4 0 0 2 – 4 0 0 8
T h i s j o u r n a l i s & T h e O w n e r S o c i e t i e s 2 0 0 5