Novel affinity ligands for chromatography using combinatorial chemistry

Article abstract of DOI:10.2174/138620711795222482

Spatially addressable combinatorial libraries were synthesized by solution phase chemistry and screened for binding to human serum albumin. Members of arylidene diamide libraries were among the best hits found, having submicromolar binding affinities. The results were analyzed by the frequency with which particular substituents appeared among the most potent compounds. After immobilization of the ligands either through the oxazolone or the amine substituent, characterization by surface plasmon resonance showed that ibuprofen affected the binding kinetics, but phenylbutazone did not. It is therefore likely that these compounds bind to Site 2 in sub domain IIIA of human serum albumin (HSA).

Full text of DOI:10.2174/138620711795222482

Combinatorial Chemistry & High Throughput Screening, 2011, 14, 267-278
267
Novel Affinity Ligands for Chromatography Using Combinatorial
Chemistry§
Tor Regberg1, Charlotta Lindquist*,2, Åke Pilotti2, Christel Ellström3, Lars Fägerstam4, Ann
Eckersten5, Yasuro Shinohara6, Steven L. Gallion7 and Joseph C. Hogan Jr.8
1Karolinska Institutet, 171 77 Stockholm, Sweden
2Stockholm University, 106 91 Stockholm, Sweden
3Biotage AB, 753 18 Uppsala, Sweden
4Vattholmav. 71, 754 40 Uppsala, Sweden
5Gyros, Uppsala, Sweden
6CGI Pharmaceuticals Inc, 36 East Industrial Road, Branford, USA
7Anthill Technologies Inc., 3-G Gill st, Woburn, MA 01801, USA
8Graduate School of Advanced Life Science, Hokkaido University, Sapporo 001-0021, Japan
Abstract: Spatially addressable combinatorial libraries were synthesized by solution phase chemistry and screened for
binding to human serum albumin. Members of arylidene diamide libraries were among the best hits found, having
submicromolar binding affinities. The results were analyzed by the frequency with which particular substituents appeared
among the most potent compounds. After immobilization of the ligands either through the oxazolone or the amine
substituent, characterization by surface plasmon resonance showed that ibuprofen affected the binding kinetics, but
phenylbutazone did not. It is therefore likely that these compounds bind to Site 2 in sub domain IIIA of human serum
albumin (HSA).
Keywords: Affinity chromatography, combinatorial chemistry, high-throughput screening, arylidene diamide library.
1. INTRODUCTION
synthesis and screening technologies. Small designed
combinatorial libraries have been employed to facilitate the
identification of affinity ligands [7-9]. In this paper, we
describe the integrated and iterative application of design,
automated high-throughput synthesis and screening
approaches for the discovery of small molecule affinity
ligands specific for human serum albumin [10, 11].
Affinity chromatography is an established technology for
the large-scale industrial purification of macromolecules [1].
The method has had particular success with the production
of biopharmaceutical agents, such as monoclonal antibodies.
With the burgeoning pipeline of protein drugs in
development, there is an increasing need for new affinity
media that demonstrate highly selective binding
characteristics. Many of the current affinity ligands are
proteinaceous and exhibit limited stability in the harsh basic
conditions used for cleaning-in-place (CIP) procedures that
are a necessary part of biopharmaceutical manufacturing.
Ligands of non-biological origin may achieve greater
chemical stability without sacrifice of specificity. Examples
of such ligands are Cibacron Blue F3GA that has group
selectivity for a large number of nucleotide-binding proteins
and a number of different media types having specificity for
antibody adsorption [2-6].
The general strategy for this project is outlined in Fig. (1).
The strategy begins by compiling available information about
known intermolecular interactions. This information is then
translated into reagents compatible with available combinatorial
synthetic methods. Libraries are produced using parallel
solution phase chemistry and binding is measured through the
use of solution phase and chromatographic evaluation. Analysis
of the resulting structure-activity relationships (SAR) is used for
iterative optimization and to select compounds for further
evaluation. The binding characteristics of the ligands were also
studied using the BIACORE system. The covalent
immobilization on chromatographic media requires a reactive
group in the final ligand suitable for attachment to the support
material. After immobilization, the binding characteristics of the
media are examined by affinity chromatography.
Historically, most affinity ligands have been discovered
in a serendipitous manner, though rational design approaches
have been used [5]. This is similar to the drug discovery
process before the advent of high-throughput parallel
2. EXPERIMENTAL
2.1. Materials
*Address correspondence to this author at the Stockholm University, 106 91
Stockholm, Sweden; Tel: + 46 8 16 24 84; Fax: +46 8 15 49 08;
E-mail: lotta@organ.su.se
All solvents and chemicals for synthesis were purchased
from Aldrich or Sigma and solvents were of the highest
§The work was carried out at ArQule, Inc., Woburn, MA, USA and
Amersham Pharmacia Biotech, Uppsala, Sweden
1386-2073/11 $58.00+.00
© 2011 Bentham Science Publishers Ltd.

268 Combinatorial Chemistry & High Throughput Screening, 2011, Vol. 14, No. 4
Regberg et al.
with a HiTrap Desalting Column. The second step was
performed on a Gilson 306 HPLC equipped with a HiSep 2.0
cm pre-column. Chromatographic evaluations were
performed on a SMARTꢀ System or on an ÄKTAꢀ
Explorer 10 (Amersham Pharmacia Biotech). Kinetic and
competitive studies were performed on a BIACORE 2000.
Total organic carbon (TOC) was determined on a TOC 5000
analyzer from Shimadzu. Elemental analyses performed by
Mikro Kemi AB, Uppsala, Sweden.
Target/Ligand Binding Motifs
Electronically Screen for Reagents
Solution Phase Parallel Synthesis
2.3. Methods
2.3.1. Syntheses
Solution Phase Parallel Screening
Over 20,000 compounds were synthesized and tested
using variations of Mannich [12], triazine [13],
hydroxyaminimide [14, 15] and oxazolone-based arylidene
diamide [16] chemistries. The generic structure of these
libraries is represented in Fig. (2). The arylidene diamide
libraries, produced the highest hit rate and thus the primary
chemistry used in this work involved the formation of
arylidene diamides from sets of oxazolones, aldehydes and
amines.
Structure-Performance Characterization
(SAR)
Conversion to Sepharose - Immobilized Compound
Chromatographic Characterization
OH
R2
R1
H
N
O
N
R3
R2
R3
N
H
R1
Fig. (1). Strategy for the high throughput discovery of small
organic affinity ligands.
O
purity available. ECH Sepharoseꢀ 4B, EAH Sepharose 4B,
EAH Sepharose HP, EAH Sepharose 6FF, NHS Activated
Sepharose 4FF, Cibacron Blue Sepharose, Sephasilꢀ C8 (4
x 10 mm) and HiTrapꢀ Desalting Column were obtained
from Amersham Pharmacia Biotech. HiSep 2.0 cm
precolumn was obtained from Supelco. Acetonitrile for
HPLC (LiChrosolv, gradient grade) was obtained from
Merck. Human serum albumin, (A-3782), Bovine serum
albumin (A-7511), lysozyme (L-6876), transferrin (T-4515),
ꢀ-lactoglobulin (L-2506) and human serum (S-7023) were
all purchased from Sigma. Polyclonal immunoglobulin G
(IgG) (Gammanorm 165 mg/ml) was obtained from
Pharmacia Corporation and ovalbumin from Amersham
Pharmacia Biotech. HEPES (N-[2-Hydroxyethyl]piperazine-
N’-[2-ethane sulfonic acid]) buffer was obtained from
BIACORE AB. The syntheses were followed by LC on silica
gel Merck 60 F254. Flash chromatography was performed on
silica gel 60 (Merck 230-400 mesh).
Arylidene diamides (R2 = Ar)
Mannich type compounds
R1
N
R2
+
OH
O
N
N
N
R2
R1
N
H
R3
-
Triazines
Hydroxyaminimides
Fig. (2). Structures of the screening library compounds.
Syntheses of starting oxazolone and arylidene diamide
compounds were performed as exemplified below in Fig. (3).
For the library productions a pilot library was first
synthesized and fully characterized by LCMS and NMR. In
the library production the condensation and the opening of
the oxazolone were done without any intermediate
purification or work-up. The final libraries were
characterized by LCMS. Final yields in the libraries were
above 80% (HPLC).
2.2. Instrumentation
The synthetic reactions were followed by HPLC
(Shimadzu LC10) at 254 nm with Sephasil C8 (4 x 10 mm)
using acetonitrile (with 0.035% trifluoroacetic acid (TFA)),
water (with 0.05% TFA) and 30-100% acetonitrile as
solvents (8 min). Characterizations were performed on a Jeol
Eclipse-270 MHz NMR. The samples were run in a 5 mm
probe and the substances were dissolved in CDCl3 or
DMSO-d6. Trimethylsilane (TMS) was used as an internal
standard.
Synthesis of 2-Thienylglycinate (I)
In a 1 L 3-necked reaction flask glycine (77.0 g, 1.03
mol) was dissolved in 600 ml water with a mechanical
stirrer. NaOH (12.0 g) was added to form sodium glycinate.
The reaction mixture was cooled to 5°C. 2-
Thiophenecarboxylic acid chloride (100 g, 0.68 mol) was
added drop wise and NaOH-solution (50%) was periodically
added to keep pH around 10 during 1.5 hrs. The temperature
rose to 12°C during the addition and the solution became
The first step in the screening method was performed on
a FPLCꢀ system (Amersham Pharmacia Biotech) equipped

Novel Affinity Ligands for Chromatography
Combinatorial Chemistry & High Throughput Screening, 2011, Vol. 14, No. 4 269
O
H
OH
1. NaOH
2. HCl
R1
Cl
R1
N
H2N
OH
O
O
O
(I)
R2
N
R2-CHO
(R2 = Ar)
N
O
O
R1
O
R1
O
(II)
(III)
R2
O
R3-NH3
H
N
R1
N
H
R3
O
(IV)
Fig. (3). Synthesis of arylidene diamides.
homogeneous. After another hour the mixture was acidified
to pH 2 using conc. HCl (70 ml) and the stirring was
continued for two hours. The precipitated crystals were
filtered off and washed with water. The product was
confirmed with NMR after drying in a vacuum oven at 60°C.
1H NMR shifts: ꢀ 4.05 (s, 2H), 7.12 (dd, 1H), 7.65 (dd, 1H),
7.71 (dd, 1H). Yield: 123 g (98%), mp 172-173°C.
General Synthesis of Arylidene Diamides (IV)
The substituted oxazolones synthesized in the last step
(0.57 mmol) were mixed with different amines (0.57 mmol)
in THF in screw-cap tubes. The tubes were placed on a
heating block at 60oC over night (18 hrs) followed by
evaporation of the solvent with heat and/or nitrogen. The
products were not purified further. The raw products were
analyzed with TLC, LCMS and 1H NMR. Yields were
typically 90-95%.
Synthesis of 2-(2-Thienyl)-Oxazolone (II)
In a 2 L 3-necked flask equipped with a mechanical
stirrer dicyclohexyl carbodiimide (66.8 g, 324 mmol) was
dissolved in 700 ml anhydrous THF. 2-Thienylglycinate (60
g, 324 mmol) dissolved in 600 ml anhydrous THF was added
drop wise during 30 min. The reaction mixture was then
allowed to stir for 24 hours at ambient temperature. The
mixture was cooled to 5°C and the dicyclohexylurea was
filtered off. After evaporation of THF the solid product was
dissolved in boiling dichloromethane and then cooled so that
more dicyclohexylurea could be filtered off. The solution
was evaporated and chromatographed on 400 g silica
through a 15 cm wide column with dichloromethane. The
2.3.2. Immobilization
ECH Sepharose 4B, EAH Sepharose 4B, EAH Sepharose
HP, NHS Activated Sepharose 4FF and EAH Sepharose
6FF, were supplied pre-swollen in ethanol (20 %). For the
amino-substituted media (here exemplified with EAH
Sepharose 6FF, 25 ꢀmol amino-groups/ml) 10 ml gel was
washed on a glass filter with water followed by THF until no
traces of water was found in the eluate. The ligands (0.20
mmol) and dicyclohexyl carbodiimide (2.0 mmol) were
dissolved in THF (10 ml) and then mixed with the gel. The
suspension was rotated over night (18 hrs) at ambient
temperature. The gel was washed with 100 ml water, 100 ml
isopropanol, 100 ml acetonitrile and finally 100 ml water.
Remaining amino-groups were blocked with a mixture of 1.7 M
acetic acid and 1 M dicyclohexyl carbodiimide in dioxane.
The gel was washed with 100 ml 40°C isopropanol, 100 ml
THF, 100 ml acetonitrile, 100 ml 40°C isopropanol, 100 ml
ethanol (95%), and then three cycles of alternating pH with
100 ml 0.1 M acetate buffer (pH 4.1) containing 0.5 M NaCl
and 100 ml 0.1 M Tris-HCl buffer (pH 8) containing 0.5 M
NaCl and finally 100 ml water. Substitution levels,
determined by elemental analysis, were typically in the range
of 3-10 ꢀmol/ml wet gel.
1
first 3 L was collected and evaporated. H NMR shifts: ꢀ
7.14 (dd, 1H), 7.59 (dd, 1H), 7.7ppm (dd, 1H). Yield: 70%
(38 g), mp 106-107°C.
Synthesis of 2-(2-Thienyl-)4-(N-Methyl-3-Indolyl)-5(H)-
Oxazolone (III)
2-(2-Thienyl)-oxazolone (3.0 g, 18 mmol) was mixed
with 1-methyl-indole-3-carboxaldehyde, (2.0 g, 12.6 mmol)
in 12 ml toluene in a screw-cap tube. Triethylamine (0.8 ml)
was added and the closed tube placed on a heating block at
70°C over night. The reaction mixture was diluted with 100
ml toluene and a small amount of acetone and extracted with
3x100 ml water. The toluene layer was dried with MgSO4,
evaporated to 20 ml. Crystallization occurred over night.
Crystals were collected and dried in a vacuum oven at 60°C.
1H NMR shifts: ꢀ 3.93 (s, 3H), 7.17 (dd, 1H), 7.3-7.4 (m,
3H), 7.60 (dd, 1H), 7.62 (s, 1H), 7.81 (dd, 1H), 7.95 (d, 1H),
8.42 (s, 1H). Yield: 50% (2.8 g), mp 178-180°C.
For the carboxyl substituted media (here exemplified
with NHS Activated Sepharose, 16-23 ꢀmol carboxyl-
groups/ml) 10 ml gel was washed on a glass filter with water
followed by THF until no traces of water was found in the

270 Combinatorial Chemistry & High Throughput Screening, 2011, Vol. 14, No. 4
Regberg et al.
eluate. The ligands (0.10 mmol) were dissolved in THF (10
ml) and then mixed with the gel. The suspension was rotated
over night (18 hrs) at ambient temperature. The gel was
washed with 100 ml THF, 100 ml acetone, 100 ml water,
100 ml isopropanol, 100 ml acetonitrile, 100 ml water and
then three cycles of alternating pH with 100 ml 0.1 M acetate
buffer (pH 4.1) containing 0.5 M NaCl and 100 ml 0.1 M
Tris-HCl buffer (pH 8) containing 0.5 M NaCl and finally
with 200 ml water.
dissociation rate constant calculated. In this case the best fit
was obtained by using a model in which there are two
independent binding sites on HSA for the ligand using the
formula xAT + yAT ꢀ (x+y)A + T where x and y represents
the two different binding sites (Fig. 4C). The two
dissociation rate constants for phenylbutazone were
calculated to be 1.1 x 10-1 s-1 and 7.3 x 10-3 s-1 respectively,
which are in accordance with those previously determined
100
0.25
0.2
0.15
0.1
0.05
0
A280
% B
2.3.3. Solution Phase Screening
A method for screening libraries of small synthetic
molecules for their ability to bind to target proteins was
developed. The method consists of three steps: 1) mixing of
target and ligand, 2) removal of free ligand and 3) analysis of
the target fraction for presence of bound ligand (Fig. 4). A
similar approach was used by Zuckermann, et al. to identify
peptide binding to a specific antibody [17].
5
0.63 ml/min
1.25 ml/min
2.5 ml/min
5 ml/min
10 ml/min
0
From a stock solution of the ligand (DMSO, 5 mM), 10
ꢀl was diluted with 890 ꢀl PBS (50 mM sodium phosphate,
150 mM sodium chloride, pH 7.0). To this solution 100 ꢀl
HSA (100 ꢀM) in PBS buffer was added at ambient
temperature (Mixture A). Excess ligand was removed by
separating 500 ꢀl of the mixture on a HiTrap Desalting
Column, equilibrated in PBS buffer, at a flow rate of 10
ml/min. The column was washed with PBS buffer containing
20 % acetonitrile after each run. Analysis of the void fraction
was performed on a HiSep 2 ml column equilibrated with
95:5 v/v 180 mM ammonium acetate and acetonitrile. The
bound ligand was eluted with a gradient of 95:5 v/v 180 mM
ammonium acetate and acetonitrile (A-buffer) and 10:90 v/v
ammonium acetate and acetonitrile (B-buffer). The elution
was registered at 254 and 280 nm.
Acetonitrile
10 12
Conductivity
Collected
2
0
4
6
8
14
16
18
Volume ml
Fig. (4A). Gel filtration on HiTrap Desalting Column of
HSA/phenylbutazone at different flow rates. The target fraction was
collected from 2.3 ml. A small molecule showing no adsorption to
the matrix would elute at ꢁ 4.5 ml.
A280
HSA
0.05 AU
Gel filtration
flow rate
PB
2.3.4 Binding Studies Using the Solution Phase Screening
Method
10 ml/min
5 ml/min
2.5 ml/min
1.25 ml/min
0.63 ml/min
Blank
By using the method described above and varying the
flow rate in the desalting step the dissociation rate constant
for a target-ligand complex could be estimated from the time
dependent dissociation of the complex. Five samples of
Mixture A containing the model ligand phenylbutazone (PB)
were sequentially separated on the HiTrap Desalting Column
at five different flow rates respectively (0.63, 1.25, 2.5, 5 and
10 ml/min). The time between sample application and
collection of the target fraction varies from 10.8 seconds at a
0
5
10
15
20
25
30
35
Time (minutes)
Fig. (4B). RPC analysis on the HiSep column of the HSA fraction
from the gel filtration step in Fig. (5a).
Peak area
2.5
flow rate of 10 ml/minute to 171 seconds at a flow rate of
0.63 ml/minute. Chromatograms for the reference ligand
phenylbutazone are shown in Fig. (4a).
2
1.5
1
The analysis of ligand content in the collected protein-
ligand fraction was performed by reversed phase
chromatography on a HiSep column in which large
molecules such as proteins are shielded from adsorption.
This eliminates interference from the target protein in the
ligand analysis. Fig. (4B) shows the analyses of the collected
fractions from Fig. (4A). All steps in the method were
automated using FPLC and Gilson HPLC Systems and a
Gilson 215 liquid handler.
0.5
0
Dissociation time (sec)
60 80
0
20
40
100
120
140
16
From the residence times (dissociation times) on the
column in Fig. (4A) and the phenylbutazone peak areas in
Fig. (4B), a dissociation plot was constructed and the
Fig. (4C). Plot of the ligand peak area as a function of the
dissociation time.

Novel Affinity Ligands for Chromatography
Combinatorial Chemistry & High Throughput Screening, 2011, Vol. 14, No. 4 271
with other techniques [18]. The peak areas were normalized
using the HSA peak area as an internal standard. The dotted line
represents the fitted values using a two-site binding model.
and 14 for the free ligands and at pH 2, 3, 11, 12, and 13 for
the immobilized ligands respectively. 0.5 ml water at the
respective pH was added to 100 ꢀl of the wet gel. The
mixture was stored in a capped tube for a week. The
supernatant (0.2 ml) was diluted with water (4 ml) to
determine any leakage products in the interval 0.5-5 ppm.
The carbon leakage from the gels was analyzed using a TOC
analyzer [20]. The analyses for the free ligands were
2.3.5. Binding Studies by Surface Plasmon Resonance
Interaction analyses based on surface plasmon resonance
(SPR) were performed using a BIACORE 2000 (BIACORE
AB). SPR correlates with the changes in mass on the surface of
the gold film involving association and dissociation between
proteins and ligands [19]. The effect is expressed in resonance
units (RU). After subtraction of the background RU due to bulk
refractive index change (determined by injection over a blank
surface), Req/ligand concentration (C) was plotted versus Req
where Req is the equilibrium response (in RU) of analyte at the
injected analyte concentration. The Ka was calculated from the
slope by linear least-square curve fitting using the equation:
Req/C = Ka x Rmax – Ka x Req, where Rmax is the maximum
amount of bound ligand.
1
performed by H- and 13C-NMR to determine degradation
products. 5 mg of each substance were weighed into two
NMR-tubes and dissolved in 0.01 M DCl (DMSO) and 1 M
KOD (DMSO) respectively and were kept in a thermo stated
1
water bath for a week. H NMR spectra were recorded after
2, 4 and 7 days respectively.
3. RESULTS AND DISCUSSION
3.1. Screening Results of Mapping Arrays and Design of
Directed Libraries
For these studies each target protein was covalently
immobilized onto the BIACORE CM5 sensor chip by NHS
activation of carboxymethylated dextran on the surface. The
immobilized levels of HSA and BSA were 8756 and 8687 RU,
respectively. All the affinity measurements were done in HBS
buffer (comprised of 10 mM HEPES, pH 7.4, 150 mM NaCl,
3.4 mM EDTA and 0.005% BIACORE surfactant P20) with
1.1% DMSO. A ligand solution (<100 ꢀM, in the same running
buffer) was injected across the chip surface at a flow rate of 30
ꢀl/min. The interaction was monitored at 25°C as the change in
the SPR response [19]. After monitoring for 3 min, the running
buffer was introduced onto the sensor chip in place of the ligand
solution to start the dissociation. The affinity constant, Ka, was
determined from the equilibrium binding level (Req) by
Scatchard plot analysis.
Although hits were identified from every chemotype, the
arylidene diamide libraries produced the largest number of
strongest binding compounds. All subsequent directed
libraries were produced using this chemistry (Fig. 3).
Frequency tables were constructed from the primary
screening binding data, recording how often the various R1,
R2 and R3 fragments were present in the hits. This is
exemplified in Fig. (5). The frequency was normalized for
the number of times the component appeared among all
compounds in the library. It is clear from these plots that
certain components occur in hits with significantly greater
frequency than others. When correlated with structure,
certain trends emerge. In general, the same 3-4 oxazalones
consistently appear but this position does not appear as
sensitive to structural variation as the others. This may be a
function of the limited structural diversity of available
oxazolones. However, variation of amine and aldehyde
component lead to pronounced differences in hit rates.
Large, aromatic aldehydes appear to work well (typically
found in 10% or more of the hits). 2-Phenyl or 3-alkyl
branched amines are also frequently found.
2.3.6. Chromatographic Evaluation
The prototype gels and Cibacron Blue Sepharose were
packed in HR 5/5 columns (1 ml) in distilled water and
equilibrated with PBS buffer for protein separations and with 50
mM citric acid, 0.1 M Na2HPO4, pH 7 for the serum separations
on the SMART system before use.
Using this initial SAR information, a 1600 member
directed library was constructed using four oxazolones (R1),
10 heterocyclic aldehydes (R2), and 40 amines (R3) as
shown in Fig. (3). to increase the selectivity of the ligands
heterocyclic aldehydes at R2 were included since the SAR
indicated that large aromatic ring systems increased the
binding properties. Only the 4 top scoring oxazalones were
included from the parent libraries. To thoroughly test the
SAR at R3 with a focused R1, R2 combination a diverse set
of 40 pre-qualified amines was used (the building blocks are
listed in Fig. (6)).
500 ꢀl of the target molecule (HSA, 3 mg/ml) and the test
proteins (3 mg/ml of human IgG, lyzosyme, ovalbumin, ꢀ-
laktoglobulin and transferrin) were applied at a flow rate of 500
ꢀl/min (150 cm/hrs). The elution was performed using 20 mM
citrate buffer, pH 3. The chromatograms were registered at 280
nm.
500 ꢀl Human serum, diluted 4 times with PBS buffer,
(corresponding to 5 mg HSA) was applied to the Cibacron Blue
Sepharose column (reference column) and the prototype column
at a flow rate of 500 ꢀl/min (150 cm/hrs). Both columns were
eluted with 50 mM KH2PO4 with 1.5 M KCl, pH 7. The
prototype column was further eluted with 20 mM citrate buffer
at pH 3.0.
The screening of this directed library resulted in several
confirmed hits. Compound (1), Fig. (7) displayed
exceptionally strong binding towards HSA. A second series
of directed libraries comprising a total of more than 400
compounds were then designed to explore immobilization
possibilities. The SAR information is displayed by the
various R-groups in Fig. (5). Variation of the oxazolone did
not significantly affect binding. However, two aldehydes
2.3.7. Stability Studies
Samples of free ligands and ligand-substituted gels were
incubated for one week at 40°C in aqueous solutions at pH 2

272 Combinatorial Chemistry & High Throughput Screening, 2011, Vol. 14, No. 4
Regberg et al.
binding activity. The only effective compound contained a
methyl amino group in the 4-position of the benzene ring of
the phenyloxazolone component. Compound (1) could not be
immobilized through the hydroxyl function of the ephedrine
part in good yields. Immobilization of the ligand through a
para- hydroxyl group in the phenyl ring of the ephedrine
component resulted in an inactive ligand.
Variation of Activity at R1
10,0
8,0
6,0
4,0
2,0
0,0
The introduction of handles for immobilizations in R2 of
the molecule was restricted to the indole structure by adding
N-allyl- and N-(4-chlorobutyl)-groups to the indole nitrogen.
The structures used were 2-methyl and 5-methoxy-indoles.
Both the types of N-substituted derivatives were found to be
weak HSA-binders when immobilized to various activated
Sepharose materials.
1
2
3
4
Oxazalone
Several of the possible ephedrine analogues were tested
as ephedrine mimetics but most of these compounds also
exhibit undesirable pharmacological activity. To avoid this
Variation in Activity at R2
12,0
10,0
8,0
problem
a
sub-library containing (S)- and (R)-N-
methylphenylalanine were combined with phenyl- and
thiophene-oxazolones that had been reacted with 1-methyl-3-
indolecarboxaldehyde and naphthaldehyde. Compounds (2)
and (3), Fig. (8), in this series showed binding affinity
comparable to compound (1). Compound (4), the methyl-
ester of corresponding compound (2), was also synthesized
to estimate potential changes in affinity upon
immobilization. A summary of the various replacement
amines used to explore SAR and immobilization at R3 is
shown in Fig. (9). Each of these amines was reacted with the
adduct between phenyl- or thienyloxazolones and 1-methyl-
3-indolecarboxaldehyde.
6,0
4,0
2,0
0,0
1
2
3
4
5
6
7
8
9
10
Aldehyde
Variation of Activity at R3
The binding characteristics of ephedrine, N-
methylphenylalanine and phenylalaninol containing ligands
did not appear to be dependent on the absolute configuration
around the chiral centers. Both (S)- and (R)-phenylalaninol
was found to bind less well than ephedrine and N-
methylphenylalanine compounds, so some sensitivity in the
position of the hydroxyl group can be inferred. In most
cases, N-methylation significantly enhanced binding. Both
enantiomers of the N-methylphenylalanine exhibited the
same binding towards HSA as compound (1), and thus all
immobilizations were performed using the (S)-isomer.
25,0
20,0
15,0
10,0
5,0
0,0
1
4
7
10 13 16 19 22 25 28 31 34 37 40
Amine
Sensitivity to changes in the geometry of the central R2
component was explored by performing a reduction of the
carbon-carbon double bond for compound (2). One of the
diastereomers formed was inactive and the other one showed
only weak activity. Due to the poor activity the absolute
configuration of the two different stereoisomers was not
determined. The presence of this double bond and the
concomitant conformational restrictions appear to confer
significant contributions to the total energetics of binding.
Fig. (5). Frequencies of occurrence of various R1, R2, R3 groups
among hits for the directed arylidene diamide library.
(1-methyl-3-indolecarboxaldehyde and 6-methyl-2-pyridine-
carboxaldehyde) were prevalent among the best compounds,
as was a single amine (N-methyl-ephedrine). It should also
be noted that separate tests of the various starting materials
and intermediates such as oxazolones condensed with
aldehydes or amine-opened oxazolones without the aldehyde
component did not yield any compounds that exhibited
measurable HSA binding.
3.2. Binding Stoichiometry and BIACORE Binding
Studies
Several attempts were made to introduce functional
groups such as –NH2, -CH2NH2 (both protected as
carbamates), -CN, -COOH, -CH2NO2, and -CH2Cl into the
R1 ring system for matrix immobilizations. The oxazolone
was then condensed with a set of aldehydes and the resulting
substituted oxazalone ring opened with ephedrine. Most of
the ligands carrying functional groups at R1 showed no
The HSA binding characteristics of compound (1) was
studied by determining the dissociation rate constant using
different flow-rates in the screening gel filtration step as
described in Methods. Compound (1) was found to bind at a
2 to 1 ratio (ligand:HSA) under saturating conditions. To

Novel Affinity Ligands for Chromatography
Combinatorial Chemistry & High Throughput Screening, 2011, Vol. 14, No. 4 273
a) Oxazolones (R1)
N
O
N
O
N
O
N
CF3O
S
O
O
O
O
O
CH3O
R1-4
R1-3
R1-2
R1-1
b) Aldehydes (R2)
O
O
O
O
O
S
S
S
O2N
S
S
Br
H
H
H
H
H
H
Br
R2-1
R2-3
R2-2
R2-4
R2-5
O
O
O
H
H
O
O
O2N
O
N
H
N
MeO
N
N
H
H
H
R2-7
R2-8
R2-10
R2-6
R2-9
c) Amines (R3)
NH2
O
NH2
NH2
NH2
NH2
O
R3-6
NH2
R3-4
R3-5
R3-3
R3-9
R3-2
R3-1
OH
S
HN
NH2
NH2
NH2
NH2
NH2
R3-11
R3-10
NH2
R3-12
R3-18
OH
R3-8
R3-7
NH2
O
N
O
N
NH
NH
EtO
N
NH2
N
H
H
R3-17
NH2
R3-16
R3-14
R3-15
R3-20
R3-13
NH2
NH2
NH2
NH
N
NH2
R3-22
R3-19
R3-21
R3-23
NH2
NH2
O
N
NH2
NH2
N
NH2
NH
O
R3-28
R3-29
R3-27
R3-25
R3-24
R3-26
NH2
N
N
N
NH
NH2
NH2
NH
NH2
R3-32
R3-31
R3-30
R3-35
R3-34
R3-33
NH2
NH2
N
N
NH
NH2
O
NH2
N
CF3
R3-37
R3-38
R3-36
R3-39
R3-40
Fig. (6). Composition of the 1600 member arylidene diamide array. a/ Oxazolone building blocks, corresponding to R1. b/Aldehyde building
blocks, corresponding to R2. c/Amine building blocks, corresponding to R3.

274 Combinatorial Chemistry & High Throughput Screening, 2011, Vol. 14, No. 4
Regberg et al.
investigate the effect of the replacement of ephedrine with N-
methyl-phenylalanine in more detail, interactions of 1 and its
(S)-N-methyl-phenylalanine analogue 2 were analyzed by
BIACORE using immobilized HSA. As shown in Fig. (10A,
B), both compounds (1) and (2) clearly interacted with
immobilized HSA. Scatchard plot analysis shows marked
curvature for both compounds, which suggests the presence
of plural binding with different affinities. Because of this
curvature, two values for Ka were calculated using either
lower ligand concentrations (0.4-1.8 ꢀM for higher Ka) or
higher ligand concentrations (14-56 ꢀM for lower Ka).
Though both of these Ka values tend to decrease by the
replacement of ephedrine with N-methyl-phenylalanine, they
are still fairly similar (2-3x106 M-1 as higher Ka and 7-
12x104 M-1 as lower Ka).
R2
N
O
OH
N
N
H
R3
R1
O
S
Fig. (7). Compound (1), lead compound, having highest binding
capacity from the 1600 member directed library.
and its methyl ester, compound (4). As shown in Fig. (10A),
this protection causes significant decrease in resonance
signal when the ligand concentration is high. Interestingly,
the Scatchard plot obtained for compound (4) is fairly linear
(Fig. 10B). The lower Ka observed commonly for both
compounds (1) and (2) is decreased, while the higher Ka is
The effect of free carboxyl group of compound (2) was
also analyzed by comparing the interaction of compound (2)
(2) R = H, (S)-N-Methylphenylalanine
(3) R = H, (R)-N-Methylphenylalanine
(4) R = Me, (S)-N-Methylphenylalanine
N
O
N
N
H
O
S
O
R
O
Fig. (8). Compounds having N-methyl-phenylalanine as analogues to ephedrine at R3.
O
OH
OH
OH
(S)-phenylalanine
(S)-phenylalaninol
NH2
NH2
O
OH
N-Methyl-(R)-phenyl
alanine
(R)-phenylalaninol
NH2
OH
HN
O
H
N
N-Methyl-(S)-phenyl
alanine
OH
(1S, 2R)-(+)-Ephedrine
HN
OH
H
N
(2S, 3R)-3-Phenyl-
pyrrolidine-2-
carboxylic acid
NH
COOH
(1S, 2S)-(+)-Ephedrine
OH
H
N
(1S, 2S)-(-)-Ephedrine
OH
NH2
(1S, 2R)-(+)-Phenyl
propanolamine
Fig. (9). Structural analogues used for the variation at R3.

Novel Affinity Ligands for Chromatography
Combinatorial Chemistry & High Throughput Screening, 2011, Vol. 14, No. 4 275
100 RU
80
a) Cpd (1)
b) Cpd (2) c) Cpd (4)
Bound/Free (RU/M)
x107
5.0
4.0
3.0
2.0
1.0
0.0
60
40
20
0
-20
0
20
40
60
80
0
500
1000
Bound (RU)
Time
Fig. (10). Interaction analysis of ligand (1), its N-methyl-phenylalanine analogue (2) and methyl ester of (2), (4) with immobilized HSA by
SPR. (A) Sensorgrams showing the interaction of ligand (1) (a), analogue 2 (b) analogue 4 (c) with immobilized HSA. Each compound was
injected onto the HSA-immobilized surface at concentrations of 0.44, 0.88, 1.75, 3.5, 7, 12, 28 and 56ꢀM. (B) Scatchard plot obtained for
each interaction; i.e. ligand 1 (ꢀ), analogue 2 (ꢁ) and analogue 4 (ꢂ).
not significantly altered. It may therefore be possible that
presence of the free carboxylate contributes to the binding
energy of the lower affinity binding site. Since protection of
free carboxyl group did not affect the higher affinity
interaction, the usage of this functional group for the
attachment to a gel is justified.
3.4. Selectivity Studies
The selectivity of potential HSA ligands can be rapidly
estimated by measuring the differential binding to HSA and
the highly homologous protein BSA (Bovine Serum
Albumin). As shown in the sensorgrams in Fig. (12A) and
the Scatchard plots in Fig. (12B), compound (1) clearly
interacts with immobilized HSA, while it shows only slight
increase in resonance signal with immobilized BSA. Once
ephedrine is replaced with N-methyl-phenylalanine in
compound (2), it was found to bind both HSA and BSA (Fig.
12A, B). However, the compound (4) again showed good
selectivity toward HSA (Fig. 12A, B). This further supported
the validity of replacing ephedrine with N-methyl-
phenylalanine and immobilizing via the carboxyl group.
Although immobilization of protein might lead to loss or
reduction of binding activity, the Ka obtained for
phenylbutazone and ibuprofen by this method were
comparable with previously reported values. As shown in the
Scatchard plot in Fig. (12B) compound (4) displayed approx.
25 times higher selectivity versus HSA as compared to BSA.
3.3. Competitive Studies
SPR was used to establish the protein-ligand interaction
sites. The ligands were immobilized onto the sensor surface
using standard conditions. Phenylbutazone (4-butyl-1,2-
diphenyl-3,5-pyrazolidinedione) and ibuprofen ((R,S)-ꢀ-
methyl-4-(isobutyl) phenyl acetic acid) are known to bind to
binding sites 1 and 2 of HSA, respectively [21]. Essentially
fatty acid free (approx. 0.005 %) and globulin free (approx.
99 %) HSA (62.5 ꢀg/ml) was injected over the ligand
substituted surface at a constant concentration with the two
binders in series of increasing concentrations respectively.
The response will be affected by the known ligand
concentration if competition occurs between the immobilized
ligand and the known binder.
3.5. Affinity Chromatography
As can be seen in Fig. (11), only ibuprofen inhibited the
binding of HSA with immobilized compound (5) under the
conditions used. A weak response could be found even at
high concentrations of ibuprofen, which could be explained
by the fact that oxazolones were found to bind to two
binding sites on HSA in the screening studies and in other
BIACORE studies. It has been established that ibuprofen
binds at site 2, although it has been reported that it also
interacts with a further unknown binding site, which is
neither site 1 nor site 3 [18]. The same results were obtained
in competitive studies using immobilized compound (2).
The immobilizations of the ligands were performed on
the various amino-substituted gels to achieve substitution
degrees between 3 and 10 μmol ligand/ml wet gel. The
spacer arms used were between 6 and 12 atoms long. Several
different types of agarose matrices were also used.
Ligand (2) immobilized on EAH Sepharoseꢁ 4B (ligand
density 3 ꢀmole/ml wet gel) was used for selectivity tests by
applying the test proteins onto the column in PBS. A small
amount of binding was observed for ꢁ-lactoglobulin but not
for any of the other test proteins (polyclonal IgG, ovalbumin,
lysozyme, transferrin).

276 Combinatorial Chemistry & High Throughput Screening, 2011, Vol. 14, No. 4
Regberg et al.
120
phenylbutazone
100
80
60
ibuprofen
OH
O
40
20
0
N
N
H
N
O
1
10
100
1000
Compound (5)
Concentration of inhibitor (µM)
Fig. (11). Inhibitory effect by ibuprofen and phenylbutazone on the interaction of HSA with immobilized compound 5. HSA was introduced
onto the compound 5-immobilized surface at a concentration of 62.5 ꢀg/ml with different concentrations of ibuprofen or phenylbutazone.
RU
100
100
a)
b)
c)
BSA
80
80
HSA
HSA
60
60
40
40
HSA
20
20
BSA
BSA
0
0
-20
-20
0
500
1000
2000
2500
0
500
1000
11550000
2000
2500
Time
s
Time (sec)
Fig. (12A). Comparison of the ability of immobilized ligands to differentiate between HSA and BSA on the BIACORE system. Sensorgrams
showing the interaction of ligand (1) (a), analogue (2) (b) analogue (4) (c) with immobilized HSA and BSA. The concentrations used for
each ligand are the same as shown in Fig. (10).
a)
b)
c)
5.0
4.0
3.0
2.0
1.0
0.0
5.0
4.0
3.0
2.0
1.0
0.0
5.0
4.0
3.0
2.0
1.0
0.0
0
20
40
60
80
100
0
20
40
60
80
100
0
20
40
60
80
100
Bound (RU)
Fig. (12B). Scatchard plots obtained for ligand (1) (a), analogue (2) (b) and analogue (4) (c) with immobilized HSA (ꢀ) and BSA (ꢁ).
It is evident from the chromatograms obtained with
ligand (2) that a small part of the bound serum protein is
eluted under high salt conditions; however, the major part is
eluted using acidic buffer. This indicates different binding
modes for the new ligand as compared to the dye ligand. In
the primary screening method we could also observe that the
interaction involved two binding sites on the protein (Fig.
4C). The dynamic binding capacity for HSA at 10% flow
To test the binding characteristics for this ligand the
separation of human serum was performed using the
standard conditions for serum separation on Cibacron Blue
Sepharose. Binding was performed with 50 mM citric acid
and 100 mM Na2HPO4, pH 7, and the elution was performed
using high salt conditions (50 mM KH2PO4, 1.5 M KCl, pH
7.0) in the first step. In a second step the ligand (2) column
was eluted using 20 mM citric acid buffer, pH 3.0.

Novel Affinity Ligands for Chromatography
Combinatorial Chemistry & High Throughput Screening, 2011, Vol. 14, No. 4 277
through and a substitution level of 5 ꢀmole/ml gel was 10
mg/ml gel.
phase syntheses with active groups for post-synthesis
immobilization, or (b) direct solid phase syntheses.
3.6. Stability Studies and Stereochemistry
ACKNOWLEDGEMENTS
When any oxazolone is reacted with aldehyde, isomers
are formed around the double bond (22). The Z/E-ratio was
typically 9:1 as was confirmed by HPLC and 1H NMR. After
The authors would like to thank coworkers at ArQule Inc
and coworkers at Amersham Pharmacia Biotech for
contributions and support.
the opening of the oxazolone with the amine, the Z/E-ratio is
still 9:1. This is in accordance with previous work [22].
Initially some isomer pairs were separated by crystallization
and the double bond configurations were established using
1H- and 13C-NMR. Generally the major isomer was active in
the screening versus HSA, but the minor one was not. In the
libraries the 9:1 mixtures were used for screening without
separation.
REFERENCES
[1]
a) Scouten, W. H.; Affinity Chromatography, Bio-Selective
Adsorption in Inert Matrices, Wiley, New York, 1981. B)
Baumann, H.; Ohrman, S.; Shinohara, Y.; Ersoy, O.; Choudhury,
D.; Axén, A.; Tedebark, U.; Carredano. E. Rational design,
synthesis, and verification of affinity ligands to a protein surface
cleft. Protein Sci., 2003, 12, 784-793.
[2]
[3]
[4]
Travis, J.; Pannell, R. Selective removal of albumin from plasma
by affinity chromatography. Clin. Chem. Acta, 1973, 49, 49-52.
Porath, J.; Maisano, F.; Belew, M. Thiophilic adsorption a new
method for protein fractionation. FEBS Lett., 1985, 185, 306-310.
El-Kak, A.; Vijayalakshmi, M.A. Study of the separation of mouse
monoclonal antibodies by pseudobioaffinity chromatography using
matrix-linked histidine and histamine. J. Chromatogr. B., 1991,
570, 29-41.
Li, R.; Dowd, V.; Stewart, D.J.; Burton, S.J.; Lowe, C.R. Design,
synthesis and application of Protein A mimetic. Nat. Biotechnol.,
1998, 16, 190-195.
Ngo, T.; Khatter, N. Chemistry and preparation of affinity ligands
useful in immunoglobulin isolation and serum protein separation. J.
Chromatogr., 1990, 510, 281-291.
a) Teng, S.F.; Sproule, K.; Husain, A.; Lowe, C.R. Affinity
chromatography on immobilized “biomimetic” ligands. Synthesis,
immobilization and chromatographic assessment of an
immunoglobulin G-binding ligand. J. Chromatogr. B., 2000, 740,
1-15. b) Roque, A.C.A; Taipa, M.A.; Lowe, C.R. Synthesis and
screening of a rationally designed combinatorial library of affinity
ligands mimicking protein L from Peptostreptococcus magnus. J.
Mol. Recognit., 2005, 18, 213-224.
The stability of the ligands was studied using enforced
cleaning-in-place (CIP) conditions (pH 2 and 14, 40oC, 7
days). Samples were taken regularly and the possible
1
degradation was studied by H- and 13C-NMR. In general it
was found that all arylidene diamide compounds studied are
stable toward strongly basic conditions, whereas acidic
conditions are somewhat less tolerated.
[5]
[6]
[7]
Compound (1) is stable at pH 3-14 but is partly
hydrolyzed at pH 2. The hydrolytic products were detected
after 2 days and increased in concentration but not in
complexity. After 7 days 30% of the R3 component had been
hydrolyzed, but no further degradation products could be
established.
4. CONCLUSIONS
The results clearly demonstrate that combinatorial
chemistry is a promising technology for the identification of
affinity chromatography ligand candidates. The analysis of
the screening results was greatly facilitated by the use of
systematic reagent layout in a spatially addressable array
format. The ability to computationally screen structural
databases of commercially available reagents and virtual
libraries led to rapid identification of patentable lead
compounds for HSA affinity matrices [23]. The capability of
computationally screening virtual and physical combinatorial
libraries provided mechanisms for rapidly expanding the
SAR information in an efficient manner.
[8]
Saraswat, L.D.; Zhang, H.; Hardy, L.; Jones, S.; Bhikhabhai, R.;
Brink, C.; Bergenstråhle, A.; Haglund, R.; Gallion, S.L. Affinity
ligand selection from
a library of small molecules: Assay
development, screening, and application. Biotechnol., 2005, 21,
300-308.
[9]
Sproule, K.; Morill, P.; Pearson, J.C.; Buton, S.J.; Hejnaes, K.R.;
Valore, H.; Ludvigsen, S.; Lowe, C.R. New strategy for the design
of ligands for the purification of pharmaceutical proteins by affinity
chromatography. J. Chromatogr. B., 2000, 740, 17-23.
Curry, S.; Mandelkow, H.; Brick, P.; Franks, N. Crystal structure of
human serum albumin complexed with fatty acid reveals an
asymmetric distribution of binding sites Nat. Struct. Biol., 1998, 5,
827-835.
[10]
[11]
[12]
[13]
Carter, D.C.; Ho, J.X. Structure of serum albumin. Adv. Protein
Chem., 1994, 45, 153-203.
Compound (2) is clearly acceptable as an affinity ligand,
although further process optimization would be required to
Arend, M.; Westermann, B.; Risch, N. Angew. Chem. Modern
variants of the mannich reaction. Int. Ed., 1998, 37, 1044-1070.
Gustafson, G.R.; Baldino, C.M.; O’Donell, M.M.E.; Sheldon, A.;
Tarsa, R.J.; Verni, C.J.; Coffen, D. L. Incorporation of
carbohydrates and peptides into large triazine-based screening
libraries using automated parallel synthesis. Tetrahedron, 1998, 54,
4051-4065.
Rutenber, E.; McPhee, F.; Kaplan, A.P.; Gallion, S.L.; Hogan Jr.,
J.C., Stroud, R.; Craik, C. A new class of HIV-1 protease inhibitor:
The crystallographic structure, inhibition and chemical synthesis of
an aminimide peptide isostere. Bioorg. Med. Chem., 1996, 4, 9,
1545-1588.
produce
a commercially viable product. The ligand
demonstrated reasonable selectivity and dynamic binding
capacity, achieved reasonable purity from a complex
feedstock and showed low salt elution characteristics. The
ligand also worked with samples of whey, yeast and E. coli
extracts, which had been spiked with recombinant HSA.
[14]
However, in the course of this project, it became clear
that the solution phase screening of solution phase libraries
has limitations for affinity chromatography ligand discovery.
Hits found with this procedure must be converted to the
appropriate solid phase structure. Of course binding may be
diminished or eliminated as a result of the immobilization. A
more expeditious approach would involve the use of solid
phase screening technology together with either (a) solution
[15]
Peisach, E.; Casebier, D.; Gallion, S.L.; Furth, P.; Petsko, G.A.;
Hogan Jr, J.C.; Ringe, D. Interaction of
a peptidomimetic
aminimide inhibitor with elastase. Science, 1995, 269, 66-69.
Mukerjee, A.K. Azlactones: Retrospect and prospect. Heterocycles,
1987, 26, 1077-1097.
Zuckermann, R.N.; Kerr, J.M.; Siani, M.A.; Banville, S.C.; Santi,
D.V. Identification of highest-affinity ligands by affinity selection
[16]
[17]

278 Combinatorial Chemistry & High Throughput Screening, 2011, Vol. 14, No. 4
Regberg et al.
from equimolar peptide mixtures generated by robotic synthesis.
Proc. Natl. Acad. Sci. USA, 1992, 89, 4505-4509.
Aki, H.; Yamamoto, M. Thermodynamic characterization of drug
binding to human serum albumin by isothermal titration
microcalorimetry. J. Pharm. Sci., 1994, 83, 1712-1718.
a) Jönsson, U.; Fägerstam, L.; Ivarsson, B.; Johansson, B.;
Karlsson, R.; Lundh, K.; Löfås, S.; Persson, B.; Roos, H.;
Rönneberg, I.; Sjölander, S.; Stenberg, E.; Ståhlberg, R.;
Urbaniczky, C.; Östlin, H.; Malmqvist, M. Real-time biospecific
interaction analysis using surface plasmon resonance and a sensor
chip technology. Biotechniques, 1991, 11(5), 620-627. b) Ahmad,
A.; Ramakrishnan, A.; McLean, M.A.; Breau, A.P. Use of surface
plasmon resonance biosensor technology as a possible alternative
to detect differences in binding of enantiomeric drug compounds to
immobilized albumins. Biosens. Bioelectron., 2003, 18, 399-404.
[20]
Andersson, M.; Ramberg, M.; Johansson, B.L. The influence of the
degree of cross-linking, type of ligand and support on the chemical
stability of chromatography media intended for protein purification.
Process Biochem., 1998, 33, 47-55.
a/ Smith, G. D.; Ciszak, E. The structure of a complex of
hexameric insulin and 4'-hydroxyacetanilide. Proc. Nat. Acad .Sci.
USA, 1994, 91, 8851-8855. b/ Itoh, T.; Tsuda, Y.; Yamada, H.
Stereoselectivity and enantiomer-enantiomer interactions in the
binding of ibuprofen to human serum albumin. Chirality, 1997, 9,
643-649.
Rao, Y.S.; Filler, R. Geometric isomers of 2-Aryl(Aralkyl)-4-
arylidene(alkylidene)-5(4H)-oxazolones. Synthesis, 1975, 749-764.
Pilotti, Å.; Regberg, T.; Ellström, C.; Lindqvist, C.; Eckersten, A.;
Fägerstam, L. Method for binding albumin and means to be used in
the method. U.S. Patent, 5,994,507, November 30, 1999.
[18]
[19]
[21]
[22]
[23]
Received: August 27, 2010
Revised: October 25, 2010
Accepted: October 27, 2010