Talanta 51 (2000) 1001 1007

www.elsevier.com/locate/talanta

Comparison of tryptophan interactions to free and grafted

BSA protein
F. Garnier, J. Randon *, J.L. Rocca
Laboratoire des Sciences Analytiques (CNRS UMR 5619), Uni6ersite Claude Bernard, Lyon 1, Bat 308,
43 Bd du 11 No6embre 1918, F-69622 Villeurbanne Cedex, France
Received 27 August 1999; received in revised form 9 December 1999; accepted 13 December 1999

Abstract

The binding of D- and L-tryptophan molecules to bovine serum albumin (BSA) protein has been studied using
liquid chromatography and ultrafiltration in the pH range from 7 to 11. A hydrophobic interaction between
tryptophan and BSA has been observed at pH 7.0 on BSA grafted chromatographic column. However, this
interaction is negligible at higher pH for which the interaction to the stereospecific site was predominant. For both
grafted and free proteins, the complexation mechanism was a competitive binding of D- and L-enantiomers on a single
site. The apparent complexation constants for both D- and L-tryptophan show a maximum in the pH range 9 10. The
variations of the apparent complexation constants versus pH were the result of the protonation of both the amino
acid and a single site of the protein assuming that the complexation occurs between the zwitter-ionic amino acid form
and the unprotonated BSA site. The apparent pKBSA is slightly shifted from 8.3 for grafted BSA protein to 9.4 for
free BSA protein. This shift is presumably as a result of the different protein conformation. 2000 Elsevier Science
B.V. All rights reserved.

Keywords: Enantioselective; Separation; Tryptophan; BSA

1. Introduction tion. However, scale-up could also be achieved

The separation of optical isomers from the
racemate is increasingly needed in biology, phar-
macology and food industry. Separations of
racemic mixtures are mainly performed by chro-
matography and diastereoisomeric salt crystalliza-
* Corresponding author. Tel.: +33-4-72431079; fax: + 33-
4-72431078.
E-mail address: randon@univ-lyon1.fr (J. Randon)
using enantioselective membrane [15] or by a
solution system using a membrane separation pro-
cess [68]. In the solution system, the addition of
a chiral compound to a racemic mixture allows
the formation of diastereoisomers in solution. The
choice of the chiral selector has to be made fol-
lowing two criteria: (a) the complexation criterion
(one enantiomer of the racemic mixture more
easily form a complex as compared to the other
enantiomer); and (b) the size criterion (the size of

0039-9140/00/$ - see front matter 2000 Elsevier Science B.V. All rights reserved.
PII: S 0 0 3 9 - 9 1 4 0 ( 9 9 ) 0 0 3 6 5 - 3
1002 F. Garnier et al. / Talanta 51 (2000) 10011007
the chiral selector, and so the size of the
diastereoisomer complex, were big enough to be
retained by a membrane). In a filtration stage, the
free molecules (mainly those of the less complexed
enantiomer) went through the membrane. In the
following stage, the other enantiomer was col-
lected by modification of the feed solution
parameters to release the previously complexed
molecules. For such a method, the knowledge of
the complexation mechanisms is required to opti-
mize the separation (selectivity, recovery and
flux).
The separation of racemic tryptophan and its
analogs can be performed in a solution system
using bovine serum albumin (BSA) as a complex-
ing agent but the complexation mechanism is still
unclear [8 14]. Several studies of the binding
mode of tryptophan molecules to serum albumin
have been reported but opposite conclusions have
been obtained. They differ by the complexation
mode of D- and L-enantiomers: number of sites
and competition.
McMenamy and Oncley [9] analyzed the com-
plexation mechanism of tryptophan to albumins
from human and bovine serum. L-tryptophan
binding was found to be principally at one site
with an association constant much higher than at
any secondary site. A small amount of tryptophan
is bound to a secondary site with a very low order
of magnitude. The observed binding for D-tryp-
tophan was of the same order of magnitude as the
binding of L-tryptophan to its secondary site. The
pH dependence and the magnitude of binding of
tryptophan to bovine serum albumin were found
to be similar to those aspects of binding to human
albumin.
King and Spencer [10] reported that D-tryp-
tophan competitively displaced L-tryptophan
from bovine serum albumin suggesting that both
ligands bind at the same unique site. (0.1 M NaCl,
Tris buffer pH 7.95, T = 24C).
Yang and Hage [11,12] studied the separation
and binding of D- and L-tryptophan on an immo-
bilized HSA column. They showed that each
enantiomer was interacting at a single type of site
and there is no competition, either direct or indi-
rect, in their binding (0.05 M phosphate buffer,
pH 7.4, T= 25C). They concluded that D- and
L-enantiomers were binding to different regions
on HSA.
Gilpin et al. [13,14] examined principally the
effect of temperature on the chromatographic re-
tention of tryptophan for various silica-immobi-
lized albumins (phosphate buffer 50 mM, pH 7.4).
They observed a linear dependence of ln k (k:
retention factor) against 1/T for D-isomer
whereas, in the case of the L-isomer, the plots
were curved with a maximum at 2024C. They
assume that this changing binding strength results
from thermally induced minor conformational
changes in the protein. They concluded that the
D-isomer does not have a specific binding site
whereas the L-isomer exhibits a high specific bind-
ing site.
In this study, we analyze how D- and L-tryp-
tophan molecules bind to BSA protein using a
chromatographic silica column grafted with BSA
protein, and compare these results to free BSA
molecules in a solution system.
2. Experimental
2.1. Reagents
Fatty-acid-free bovine serum albumin (A-6003),
D-tryptophan (T-9753), L-tryptophan (T-8659)
were purchased from Sigma Chemicals. Other
chemicals were of reagent grade and were used
without further purification.
2.2. Chromatography
A 50 mm column (purchased from Hypersil
40207-065) filled with a 7 mm silica support
grafted by BSA protein was used. The mobile
phase was made of 100 mM KH2PO4/K2HPO4
buffer in the pH range 68, and 25 mM Borax
solution above pH 8.
The chromatographic system was a Shimadzu
LC-6A pump, a Rheodyne 20 ml injection loop, a
Shimadzu SPD-6A UV detector and a Spectra-
Physics SP 4270 integrator. The tryptophan detec-
tion was performed at 278 nm.
 F. Garnier et al. / Talanta 51 (2000) 10011007
Fig. 1. Effect of the sample concentration for D- and L-tryp-
tophan on the retention factor on a BSA grafted silica column.
Sample size: 20 ml, phosphate buffer 100 mM pH 7.0.
2.3. Filtration de6ice
All filtration experiments were carried out using
a dead-end stirred filtration cell (Sigma Aldrich
S2278) holding a flat sheet membrane having a
diameter of 25 mm. The effective membrane area
was 4 cm2. The system was set up at 20C with a
transmembrane pressure of 1.5 105 Pa and a
rotation speed of 600 rpm. All the feed solutions
were prepared just before filtration experiments,
and distilled water was prefiltrated through a 0.2
Fig. 2. Effect of L-tryptophan concentration in the mobile phase on the retention factor of
silica column, pH 7.0 (symbol: experimental results, line: model).
1003
mm membrane. A polysulfone membrane (Mil-
lipore PLGC 10 000) with a molecular weight
cutoff of 10 000 g mol 1 was used. Such a mem-
brane has a BSA rejection coefficient higher than
99.5%.
3. Results
Fig. 1 shows the retention factor of both D- and
L-tryptophan as a function of the concentration
of each isomer in the injection loop. For a con-
centration above 1.10 4 mol l 1, the BSA grafted
silica column is overloaded and the retention fac-
tors of tryptophan decrease. In the following part
of this study the injected tryptophan concentra-
tion was kept below 1.10 4 mol l 1.
3.1. Retention factor dependency on L
concentration in the mobile phase
Fig. 2 shows the variation of the retention
factor of tryptophan on a BSA grafted silica
column for various concentrations of L-tryp-
tophan in the mobile phase. When L-tryptophan
is added to the mobile phase, both L and D
retention factor decrease. Such variation for L-iso-
mer was expected whatever the retention mecha-
D- and L-tryptophan on a BSA grafted
1004 F. Garnier et al. / Talanta 51 (2000) 10011007

Table 1 BSAstat + DmobUBSA Dstat

Summary of the parameters used to describe the retention of
D- and L-tryptophan on BSA column at pH 7.0 according to [BSA D]stat
relations (56) KD,pH 7 = (2)
[BSA]stat [D]mob
pKL,pH 7 pKD,pH 7 Vstat qBSA,0 hydrophobic interaction without stereo specific
KH
Vmob Vmob recognition (both D- and L-isomers have the
same partition constant KH)
4.94 3.72 0.67 1.37104
[L]stat
LmobULstat KH = (3)
[L]mob
nism. But the retention factor of D-isomer de-
[D]stat
pends on the complexation mode of D compare to DmobUDstat KH = (4)
L. [D]mob
If D- and L-tryptophan bind on different sites, Based on relations (14), the retention factors of
the retention factor of D-isomer should be inde- D- and L-tryptophan can be expressed as a function
pendent on L concentration in the mobile phase of the complexation constants on the competitive
(this phenomenon has never been observed).
 
site, KL and KD, and the partition constant, KH:
If D- and L-tryptophan bind competitively on
KD,pH 7 qBSA,0 V
the same unique site, the retention factor of D- kD = + KH stat (5)

tryptophan should decrease when L concentration
increases in the mobile phase: some of the BSA
grafted molecules are already engaged with L-
enantiomer when D molecules arrive and so, the D
molecules continue downstream and come out
earlier than without L-enantiomer in the mobile
phase. A very large concentration of L-isomer in
the mobile phase should lead to a saturation of
the BSA sites and so the retention of D-isomer
should disappear, e.g. the retention factor de-
crease to zero. The observed decrease of the D-iso-
mer retention factor does not agree with this
mechanism because the retention factor for high L
concentration never decreased below 0.7 suggest-
ing another interaction between D-tryptophan and
the protein. This secondary interaction is not
dependant on L-tryptophan concentration in the
mobile phase.
Based on these remarks, the retention data
obtained at pH 7.0 have been analyzed assuming
the existence of two kinds of interaction between
tryptophan and BSA protein:
competitive binding on one site for D- and
L-isomers, each isomer having its own complex-
ation constant:
Vmob 1+ KL,pH 7[L] Vmob
KL,pH 7 qBSA,0 V
kL = + KH stat (6)
Vmob 1+ KL,pH 7[L] Vmob
where, qBSA,0 is the amount of BSA in the column,
Vmob the volume of the mobile phase and Vstat the
volume of the stationary phase involved in the
partition process.
The first term in relations (5) and (6) corre-
sponds to the competitive binding of D- and L-tryp-
tophan. The amount (qBSA,0/(1+ KL,pH 7[L])) is
related to the free protein accessible for the sample
which is different from the initial amount of
protein in the column. This initial amount, qBSA,0,
is reduced by the L molecules in the mobile phase
based on a Langmuir isotherm.
Four parameters have been used to fit the exper-
imental data according to the relations (5) and (6):
the complexation constants KL and KD, the parti-
tion term KH(Vstat)/Vmob, the amount of protein
per volume of mobile phase (qBSA,0)/Vmob. These
parameters have been adjusted to minimize the
relative difference between experimental and calcu-
lated values (the values are summarized in Table
1).

BSAstat + LmobUBSA Lstat 3.2. pH dependency of the retention factor

[BSA L]stat Fig. 3 shows the variation of the retention
KL,pH 7 = (1)
[BSA]stat [L]mob factor of D- and L-tryptophan as a function of the
 F. Garnier et al. / Talanta 51 (2000) 10011007 1005

pH of the mobile phase. The initial increase from [D 9 BSA]

KD = (9b)
pH 7 to 9 was attributed to a change in protein [D 9 ] [BSA]
conformation resulting from the dissociation of
protonated groups [9]. Above pH 10, the decrease As a pH function, the apparent complexation
in the retention factor has been related to the constant can be related to the acid basic constant
negatively charged amino-acid molecules above and to the complexation constant. For example,
pKa2 creating a strong electrostatic repulsion force the apparent complexation constant of the L form
between the amino-acid and the protein. is given below
In this paper, we analyzed the variations of the
apparent complexation constants versus pH as a [L 9 BSA]
KL,pH =
result of the protonation of both the amino-acid % free BSA % free L
and a single site of the protein assuming that the
complexation occurs between the zwitter-ionic [L 9 BSA]
=
amino acid form and the unprotonated BSA site: ([BSA]+ [BSA+])([L 9 ]+ [L])

   
[AA] [H+] [L 9 BSA]
AA 9 v AA +H+ Ka = (7) = KL,pH
[AA 9 ] [H+] 9 Ka
[BSA] 1+ [L ] 1+ +
KBSA [H ]
[BSA] [H+]
BSA+ v BSA+ H+ KBSA = (8)
  
[BSA+] 1
= KL (10)
[L 9 BSA] 10 pH Ka
1+ 1+ pH
AA 9 + BSA v AA 9 BSA KL = KBSA 10
[L 9 ] [BSA]
(9a) The pKa of tryptophan is 9.4. pKBSA, pKL and

Fig. 3. Variation of the retention factor of D- and L-tryptophan as a function of pH on a BSA grafted silica column (symbol:
experimental results, line: model).
1006 F. Garnier et al. / Talanta 51 (2000) 10011007

Table 2 interactions. However in this pH range, the contri-

Summary of the parameters used to describe the retention of
bution of the hydrophobic interaction to the reten-
D-and L-tryptophan on BSA column as a function of pH
according to relations (59) tion is negligible compared to the interaction of
both D- and L-isomers to the stereospecific site.
pKL pKD pKBSA Vstat Results are summarized on Table 2 and calculated
KH
(relation 9a) (relation 9b) (relation 8) Vmob apparent complexation constants are plotted in
Fig. 3.
5.9 4.6 8.3 0.62
3.3. Free BSA molecules

Filtration experiments have been performed to

determine the variation of the complexation con-
stants for D- and L-tryptophan with BSA molecules
as a function of pH. A buffered solution of a pure
enantiomeric form of tryptophan ([trp]0 = 10 4
M), with a fixed ionic strength KCl 0.1 M, has been
mixed with BSA protein ([BSA]0 = 1.5 10 4 M)
and allowed to reach equilibrium. After filtration
over a 10 000 MWCO membrane, the free amino
acid concentration was determined in the filtrate by
UV spectrophotometry at 278 nm.
The apparent complexation constants are plot-
ted in Fig. 4 as a function of pH. The same profile
as for grafted BSA in HPLC column is observed.
However, the maximum of the curve is slightly
shifted to higher pH, this shift was modeled by a
higher apparent pKa for BSA (Table 3). This
increase is presumably due the free BSA molecules
which have a slightly different conformation than
grafted on silica surface. Using membrane separa-
Fig. 4. Variation of the apparent complexation constant of D-
and L-tryptophan as a function of pH in a solution system:
tion process, the hydrophobic interactions ob-
free BSA molecules (symbol: experimental results, line: model). served using liquid chromatography cannot be
quantified as a result of the poor accuracy in the
Table 3 measurement of the concentration at the equi-
Summary of the parameters used to describe the interaction of librium.
D-and L-tryptophan to free BSA molecules as a function of pH
Several filtration experiments were performed to
according to relations (79)
test the validity of the complexation model on the
Vstat same site (competitive effect). Solutions of BSA
pKL pKD pKBSA KH
Vmob 1.510 4 M, D-tryptophan 10 4 M and L-tryp-
5.7 4.5 9.4 No quantification tophan 10 4 M, with a fixed ionic strength KCl 0.1
M, at pH from 9 to 10 were filtrated and the
concentrations of both enantiomers were deter-
pKD have been estimated to fit all the experimental mined by liquid chromatography.
data obtained. The hydrophobic interaction con- In the pH range used, the L-tryptophan had a
stant KH was supposed constant over the entire pH larger complexation constant than D-enantiomer.
range. This assumption is certainly incorrect for So, the L form is mainly bonded to the BSA site.
pH values above 8 because the charge of the In the competitive site model, the interaction site
amino-acid is modified and so the hydrophobic became less available for the D form. The ex-
 F. Garnier et al. / Talanta 51 (2000) 10011007
Fig. 5. Prediction of D-tryptophan concentration based on a
independent sites model and on a competitive site model in a
solution system. Comparison to experimental results (D exp).
[BSA]0 = 1.5 10 4 M, [L]0 = 10 4 M, [D]0 = 10 4 M, KCl
0.1 M.
pected concentration of free D-tryptophan for
competitive site model at pH 9.5 was closed to
0.7510 4 M whereas with independent sites the
D concentration should be as low as 0.56 10 4
M. The calculated concentrations of D-tryptophan
obtained with then competitive model were al-
ways closer to the experimental results (Fig. 5). .
4. Conclusion
The binding of D- and L-tryptophan molecules
to BSA protein has been studied using liquid
chromatography and ultrafiltration in the pH
range from 7 to 11. For both grafted and free
protein, the complexation mechanism was a com-
petitive binding of D- and L-enantiomers on a
single site. The variations of the apparent com-
plexation constant versus pH were the result of
the protonation of both the amino acid and a
single site of the protein assuming that the com-
plexation occurs between the zwitter-ionic amino
1007
acid form and the unprotonated BSA site. A
hydrophobic interaction between tryptophan and
BSA has been observed at pH 7.0 on BSA grafted
chromatographic column. However, this interac-
tion is negligible at higher pH for which the
interaction to the stereospecific site was predomi-
nant.
The highest possible selectivity is required for
separation by ultrafiltration so the experimental
conditions for complexation (amino-acid concen-
tration, protein concentration, pH . . .) have to be
optimized. However, the mass transport through
a membrane is also governed by several parame-
ters such as protein concentration, pressure, rota-
tion speed which also require an optimization.
Both complexation and mass transport parame-
ters have to be simultaneously controlled to im-
prove recovery, purity or flux. This work is now
under investigation.
References
[1] T. Aoki, M. Kokai, Chem. Lett. 12 (1993) 2009.
[2] T. Aoki, S. Tomizawa, E. Oikawa, J. Membr. Sci. 99
(1995) 117.
[3] S. Tone, T. Masawaki, T. Hamada, J. Membr. Sci. 103
(1995) 57.
[4] T. Masawaki, M. Saisa, S. Tone, J. Chem. Eng. Jpn. 25
(1992) 33.
[5] W.H. Pirkle, W.E. Bowen, Tetrahedron: Asymmetry 5
(1994) 773.
[6] A. Higuchi, M. Hara, T. Horiuchi, T. Nakagawa, J.
Membr. Sci. 93 (1993) 157.
[7] A. Higuchi, Y. Ishida, T. Nakagawa, Desalination 90
(1993) 127.
[8] S. Poncet, J. Randon, J.L. Rocca, Sep. Sci. Technol. 32
(1997) 2029.
[9] R.H. McMenamy, J.L. Oncley, J. Biol. Chem. 233 (1958)
1436.
[10] T.P. King, M. Spencer, J. Biol. Chem. 245 (1970) 6134.
[11] J. Yang, D.S. Hage, J. Chromatogr. 645 (1993) 241.
[12] J. Yang, D.S. Hage, J. Chromatogr. A 725 (1996) 273.
[13] R.K. Gilpin, S.E. Ehtesham, R. Gregory, Anal. Chem. 63
(1991) 2825.
[14] V. Tittelbach, R.K. Gilpin, Anal. Chem. 67 (1995) 44.