P1/P1' modified HIV protease inhibitors as tools in two new sensitive surface plasmon resonance biosensor screening assays.

Article abstract of DOI:10.1016/S0928-0987(01)00109-9

The commonly used HIV-1 protease assays rely on measurements of the effect of inhibitions on the hydrolysis rate of synthetic peptides. Recently an assay based on surface plasmon resonance (SPR) was introduced. We have taken advantage of the fact that the SPR signal is proportional to the mass of the analyte interacting with the immobilised molecule and developed two new improved efficient competition assay methods. Thus, high molecular weight binders were used as amplifiers of the surface plasmon resonance signal. Linkers were attached by a Heck reaction to the para-positions of the P1/P1' benzyloxy groups of a linear C2-symmetric C-terminal duplicated inhibitor to enable (a) biotin labelling or (b) direct immobilisation of the inhibitor to the biosensor surface matrix. The interaction properties of a series of 17 structurally diverse inhibitors was assessed and compared to previously reported data. The most sensitive assay was obtained by immobilising the enzyme and amplifying the signal with an antibody, giving a detection range between 0.1 nM and 10 microM. Immobilisation of the inhibitor resulted in a stable and durable surface but a narrower detection range (1-100 nM). The two competition assays are anticipated to be very suitable for fast screening of potential HIV inhibitors.

Full text of DOI:10.1016/S0928-0987(01)00109-9

European Journal of Pharmaceutical Sciences 13 (2001) 203–212
www.elsevier.nl/locate/ejps
P1/P19 modified HIV protease inhibitors as tools in two new sensitive
surface plasmon resonance biosensor screening assays
a
b
b
c
c
¨
¨
¨
Mathias Alterman , Hans Sjobom , Par Safsten , Per-Olof Markgren , U. Helena Danielson ,
b
b
a
d
d
¨ ¨ ¨
¨ ˚
´
¨
Markku Hamalainen , Stefan Lofas , Johan Hulten , Bjorn Classon , Bertil Samuelsson ,
Anders Hallberga ,
*
aDepartment of Organic Pharmaceutical Chemistry, Uppsala University, BMC, Box 574, SE-751 23 Uppsala, Sweden
bBiacore AB, Rapsgatan 7, SE-754 50 Uppsala, Sweden
cDepartment of Biochemistry, Uppsala University, BMC, Box 576, SE-751 23 Uppsala, Sweden
dMedivir AB, Lunastigen 7, SE-141 44 Huddinge, Sweden
Received 4 January 2001; accepted 8 January 2001
Abstract
The commonly used HIV-1 protease assays rely on measurements of the effect of inhibitions on the hydrolysis rate of synthetic
peptides. Recently an assay based on surface plasmon resonance (SPR) was introduced. We have taken advantage of the fact that the SPR
signal is proportional to the mass of the analyte interacting with the immobilised molecule and developed two new improved efficient
competition assay methods. Thus, high molecular weight binders were used as amplifiers of the surface plasmon resonance signal. Linkers
were attached by a Heck reaction to the para-positions of the P1/P19 benzyloxy groups of a linear C2-symmetric C-terminal duplicated
inhibitor to enable (a) biotin labelling or (b) direct immobilisation of the inhibitor to the biosensor surface matrix. The interaction
properties of a series of 17 structurally diverse inhibitors was assessed and compared to previously reported data. The most sensitive assay
was obtained by immobilising the enzyme and amplifying the signal with an antibody, giving a detection range between 0.1 nM and 10
mM. Immobilisation of the inhibitor resulted in a stable and durable surface but a narrower detection range (1–100 nM). The two
competition assays are anticipated to be very suitable for fast screening of potential HIV inhibitors.
rights reserved.
2001 Elsevier Science B.V. All
Keywords: HIV protease inhibitors; HIV protease; Screening; Surface plasmon resonance; Biosensor
1. Introduction
necessitates high capacity assay systems (Floyd et al.,
1999). The commonly used HIV-1 protease assay methods
rely on measurement of the hydrolysis rate of synthetic
peptides comprising the amino acid residues flanking the
scissile bonds of HIV PR55 gag and PR 160 gag-pol.
These assays are frequently complex, expensive, and
cumbersome, and the consumption of the protease is
usually high (Nashed et al., 1989; Matayoshi et al., 1990;
Tomaszek et al., 1990). Surface plasmon resonance (SPR)
HIV-1 protease has proved to be an attractive target in
the treatment of AIDS, due to its pivotal role in the
replication cycle of the virus. To date, there are six
FDA-approved HIV protease inhibitors, and several are
expected to enter the market. However, the cost of therapy
is high and the emergence of drug-resistant strains is
considered to be a significant problem. Thus, there is a
constant and urgent need for new inhibitors that also
possess activity against various mutants of the virus
(Tomasselli and Heinrikson, 2000). The development of
methods for fast synthesis of potential inhibitors, e.g.,
robotised parallel synthesis and combinatorial synthesis
¨ ˚
¨
(Lofas and Johnsson, 1990; Jonsson et al., 1991; Stenberg
et al., 1991; Malmquist, 1993; Karlsson, 1994; Stolz,
1998), an optical technique that measures the refractive
index of a medium near a thin film of metal on a glass
substrate, constitutes a rapid and alternative method for
identification and characterisation of inhibitors. The re-
ported SPR method (Markgren et al., 1998), based on the
monitoring of an inhibitor as a ligand to immobilised
HIV-1 protease, does permit characterisation of inhibitors
*Corresponding author. Tel.: 146-18-471-4284; fax: 146-18-471-
4474.
E-mail address: anders.hallberg@farmkemi.uu.se (A. Hallberg).
0928-0987/01/$ – see front matter
PII: S0928-0987(01)00109-9
2001 Elsevier Science B.V. All rights reserved.

204
M. Alterman et al. / European Journal of Pharmaceutical Sciences 13 (2001) 203–212
Uppsala, Sweden. Column chromatography was performed
on silica gel 60 (0.04–0.063 mm, Merck). Circular chro-
matography was performed on Chromatotron 7924T (TC
Research) with 1-mm thick silica gel 60 (0.04–0.063 mm,
Merck) plates and gradient elution. Thin-layer chromatog-
raphy was performed on precoated silica gel F-254 plates
(0.25 mm; E. Merck) and visualised with UV light and
H2SO4 in ethanol, or ninhydrin. The palladium tetrakis
triphenylphosphine was freshly made according to pro-
cedure described by Heck (1985). Standard workup:
organic layers were dried with MgSO4 and concentrated in
vacuum.
Scheme 1.
with Ki.5 nM or identification of inhibitors with Ki,100
nM. Unfortunately, the method is less practical for screen-
ing, even though an improvement in the methodology has
been reported recently (Markgren et al., 2001). Since the
SPR signal is proportional to the mass of the analyte
interacting with the immobilised molecule (Stenberg et al.,
1991) a greater degree of accuracy in the measurement
would be expected if high molecular weight molecules
were binding to the immobilised entity.
We herein report two different, experimentally simple
assay systems that can be used for the identification and
ranking of inhibitors in a wider affinity range. Both assays
take advantage of the previous discovery, as demonstrated
by X-ray crystallography, that the P1/P19 benzyloxy
groups of a C2-symmetric inhibitor could be extended into
the water phase, surrounding the protease, without serious-
ly affecting the binding affinity to the protease (Alterman
et al., 1999). In the first assay (assay 1) biotinamidocaproic
acids were attached via a linker to the P1/P19 benzyloxy
side chains of the parent inhibitor 1 to provide 7 which in a
mixture with a test compound was passed over the
immobilised HIV protease sensor surface. An anti-biotin
antibody was subsequently added to amplify the SPR
signal. In the second assay (assay 2) the inhibitor 1 was
immobilised to the sensor surface using the same linker
strategy as in assay 1, and the inhibition of the HIV-1
protease binding by a test compound was monitored. A
series of structurally diverse HIV protease inhibitors was
evaluated with the surface plasmon resonance assays and
the data were compared to those obtained previously with
a standard fluorometric assay (Scheme 1).
2.1.1. N-h2-[2-(2-tert-Butoxycarbonylaminoethoxy)-
ethoxy]ethyljacrylamide (4)
To a cooled (08C) solution of the mono-protected amine
(3) (50.0 mg, 0.201 mmol) and diisopropylethylamine (70
ml, 0.402 mmol) in dichloromethane (3 ml) under a
nitrogen atmosphere was slowly added acryloyl chloride
(32.7 ml, 0.402 mmol). The reaction mixture was allowed
to reach room temperature and was stirred for an additional
hour. The reaction mixture was concentrated and the crude
product was purified by circular chromatography (CH2Cl2 /
MeOH (97:3)) to give pure 4 in 92% yield: IR (KBr) n
1
3314, 3078, 2931, 1691, 1663 cm21; H NMR (CD3OD) d
6.24 (ddd, J515.5, 8.5, 3.5 Hz, 2H), 5.60 (dd, J58.5, 3.5
Hz, 1H), 3.59 (m, 6H), 3.51 (t, J55.6 Hz, 2H), 3.45 (m,
2H), 3.2 (t, J55.6 Hz, 2H), 1.43 (s, 9H); 13C NMR
(CD3OD) d 168.4, 158.5, 132.2, 126.8, 80.1, 71.4, 71.2,
70.6, 41.4, 40.7, 40.5, 28.9. Analysis, found: C, 55.0; H,
8.7; N, 9.2%; C14H26N2O51CHR.4.18; H2O (M5
306.871) requires: C, 54.8; H, 8.7; N, 9.1%.
2.1.2. N1,N6-bis[(1S)-2-methyl-1-(methyl-
carbamoyl)propyl]-(2R,3R,4R,5R)-2,5-bish4-[(E)-
2-(N-h2-[2-(2-t-butoxycarbonylaminoethoxy)-
ethoxy]ethylj)carbamoyletenyl]benzyloxyj-3,4-
dihydroxyhexanediamide (5)
A mixture of 2 (30 mg, 0.039 mmol), the olefin 4 (94
mg, 0.31 mmol), diisopropylethylamine (27 ml, 0.16
mmol), Pd(OAc)2 (0.87 mg, 3.9 mmol), P(o-Tol)3 (2.8 mg,
9.3 mmol), water (0.15 ml) and DMF (0.85 ml) was added
to a Pyrex tube and degassed under a nitrogen flow for 5
min. The tube was sealed with a screw cap and stirred for
8 h at 808C. After cooling, chloroform (50 ml) was added
to the reaction mixture. The organic layer was separated
and subsequently washed with saturated aqueous NaHCO3
(3320 ml), dried, filtered, and concentrated. The crude
product was purified by circular chromatography
(CH2Cl2→CH2Cl2 /MeOH (40:1)) to give pure 5 (35 mg,
0.029 mmol) in 74% yield: IR (KBr) n 3311, 2937, 1622
2. Materials and methods
2.1. Chemistry
1H and 13C NMR spectra were recorded on a Jeol
JNM-EX 270 spectrometer at 270.2 and 67.8 MHz,
respectively. Chemical shifts are given as d values (ppm)
downfield from tetramethylsilane. Infrared spectra were
recorded on a Perkin-Elmer 1600 series FTIR instrument.
Elemental analyses were performed by Mikro Kemi,
1
cm21; H NMR (CD3OD) d 7.52 (d, J515.8 Hz, 0.9H),
7.47 (m, 9H), 6.73 (d, J512.6 Hz, 0.1H), 6.64 (d, J515.8
Hz, 0.9H), 6.03 (d, J512.6 Hz, 0.1H), 4.59 (m, 4H), 4.21

M. Alterman et al. / European Journal of Pharmaceutical Sciences 13 (2001) 203–212
205
(d, J56.6 Hz, 2H), 4.10 (m, 4H), 3.62 (m, 10H), 3.53 (m,
10H), 3.22 (m, 4H), 2.72 (s, 6H), 2.13 (m, 2H), 1.42 (s,
18H), 0.91 (m, 12H); 13C NMR (CD3OD) d 174.0, 173.6,
168.7, 158.3, 141.4, 140.7, 136.1, 129.8, 129.1, 122.2,
81.6, 72.3, 71.4, 71.2, 70.8, 60.1, 41.4, 40.7, 40.3, 31.9,
28.9, 26.4, 19.9, 18.6. Analysis, found: C, 58.4; H, 7.8; N,
9.0%; C60H94N8O181H2O (M51233.447) requires: C,
58.4; H, 7.8; N, 9.1%.
H, 7.6; N, 11.5%; C82H128N14O20S214H2O (M5
1766.169) requires: C, 55.8; H, 7.8; N, 11.1%.
2.1.5. N1,N6-Bis[(1S)-2-methyl-1-(methylcarbamoyl)-
propyl]-(2R,3R,4R,5R)-2-h4-[(E)-2-(N-h2-[2-(2-t-
Butoxycarbonylaminoethoxy)ethoxy]ethylj)carbamoyl-
etenyl]benzyloxyj)-5-(4-bromobenzyloxy)-3,4-
dihydroxyhexanediamide (8)
A mixture of 2 (200 mg, 0.26 mmol), the olefin 4 (102
mg, 0.34 mmol), diisopropylethylamine (180 ml, 1.0
mmol), Pd(OAc)2 (5.8 mg, 0.026 mmol), P(o-Tol)3 (18.9
mg, 0.062 mmol), water (1.8 ml) and DMF (10.2 ml) was
added to a Pyrex tube and degassed under a nitrogen flow
for 5 min. The tube was sealed with a screw cap and
stirred for 8 h at 808C. After cooling, chloroform (50 ml)
was added to the reaction mixture. The organic layer was
separated and subsequently washed with saturated aqueous
NaHCO3 (3320 ml), dried, filtered, and concentrated. The
crude product was purified by circular chromatography
(CH2Cl2→CH2Cl2 /MeOH (40:1)) to give pure 8 (39.5
mg, 0.040 mmol) in 15% yield: IR (KBr) n 3302, 2967,
2.1.3. N1,N6-Bis[(1S)-2-methyl-1-(methylcarbamoyl)-
propyl]-(2R,3R,4R,5R)-2,5-bish4-[(E)-2-(N-h2-[2-(2-
aminoethoxy)ethoxy]ethylj)carbamoyletenyl]benzyloxyj-
3,4-dihydroxyhexanediamide (6)
To a stirred solution of 5 (35.6 mg, 0.029 mmol) in
dichloromethane (3 ml) was added TFA (3 ml). The
reaction mixture was stirred for 10 min before evaporating
the solvent and coevaporating with acetonitrile to give the
pure TFA salt of 6 (36 mg, 0.029 mmol) in 98% yield: IR
(KBr) n 3152, 2961, 1660 cm21 1H NMR (CD3OD) d
;
7.45 (m, 8.9H), 6.74 (d, J512.5 Hz, 0.1H), 6.68 (d,
J515.8 Hz, 1.9H), 6.04 (d, J512.5 Hz, 0.1H), 4.6 (m,
4H), 4.22 (d, J56.8 Hz, 2H), 4.10 (m, 4H), 3.61 (m, 20H),
3.12 (m, 2H), 2.73 (s, 6H), 2.11 (m, 2H), 0.92 (m, 12H);
13C NMR (CD3OD) d 174.1, 173.7, 168.9, 141.5, 140.8,
136.0, 129.8, 129.1, 122.2, 81.5, 73.2, 72.2, 71.5, 70.8,
68.0, 60.2, 40.9, 40.6, 32.0, 26.4, 20.0, 18.7. High Res.
Mass. (C50H78N8O141H1) Calc. 1015.5716; Found
1015.5747
1
1649 cm21; H NMR (CD3OD) d 7.41 (m, 8.95H), 6.75
(d, J512.5 Hz, 0.05H), 6.64 (d, J515.8 Hz, 0.95H), 6.27
(d, J512.5 Hz, 0.05H), 4.56 (m, 4H), 4.21 (m, 2H), 3.87
(m, 4H), 3.55 (m, 10H), 3.21 (m, 2H), 2.76 (s, 6H), 2.13
(m, 2H), 1.46 (s, 9H), 0.92 (m, 12H); 13C NMR (CD3OD)
d 174.0, 173.6, 168.7, 158.5, 141.4, 140.7, 138.2, 136.0,
132.6, 131.1, 129.7, 129.1, 122.8, 122.2, 81.5, 73.2, 72.8,
72.1, 71.4, 71.2, 70.7, 60.2, 41.33, 40.7, 31.9, 28.9, 26.5,
26.4, 20.0, 18.7. Analysis, found: C, 53.2; H, 6.9; N, 8.2%;
C46H69BrN6O1312H2O (M51030.007) requires: C, 53.6;
H, 7.1; N, 8.2%.
2.1.4. N1,N6-Bis[(1S)-2-methyl-1-(methylcarbamoyl)-
propyl]-(2R,3R,4R,5R)-2,5-bis[4-((E)-2-hN-[2-(2-h2-[6-
(biotiamido)carbamoylpentyl]ethoxyjethoxy)ethyl]jcarba-
moyletenyl)benzyloxy]-3,4-dihydroxyhexanediamide (7)
To a stirred mixture of 6 (36 mg, 0.029 mmol) and
diisopropylethylamine (20 ml, 0.12 mmol) in DMF (1 ml)
was added succinimidyl-6-(biotinamido)hexanoate (33 mg,
0.072 mmol). The reaction mixture was stirred over night
and then concentrated. The crude product was purified by
circular chromatography (CH2Cl2→CH2Cl2 /MeOH (9:1))
to give pure 7 (42 mg, 0.025 mmol) in 86% yield: IR
2.1.6. N1,N6-Bis[(1S)-2-methyl-1-(methylcarbamoyl)-
propyl]-(2R,3R,4R,5R)-2-h4-[(E)-2-(N-h2-[2-(2-
aminoethoxy)ethoxy]ethylj)carbamoyletenyl]benzyloxyj)-
5-(4-bromobenzyloxy)-3,4-dihydroxyhexanediamide (9)
To a stirred solution of 8 (78.3 mg, 0.079 mmol) in
dichloromethane (9 ml) was added TFA (9 ml). The
reaction mixture was stirred for 10 min before evaporating
the solvent and coevaporating with acetonitrile to give the
pure TFA salt of 9 (78.0 mg, 0.077 mmol) in 98% yield:
(KBr) n 3252, 2933, 1684 cm21 1H NMR (CD3OD) d
;
7.52 (m, 9.9), 6.73 (d, J512.7 Hz, 0.1H), 6.67 (d, J515.9
Hz, 1.9H), 6.05 (d, J512.7 Hz, 0.1H), 4.61 (m, 4H), 4.47
(m, 2H), 4.29 (m, 2H), 4.21 (d, J56.6 Hz, 2H), 4.12 (m,
4H), 3.61 (m, 20H), 3.37 (m, 4H), 3.35 (m, 4H), 2.91 (m,
2H), 2.75 (s, 6H), 2.72 (m, 2H), 2.19 (m, 8H), 2.12 (m,
2H), 1.51 (m, 26H), 0.93 (m, 12H); 13C NMR (CD3OD) d
176.3, 176.0, 168.7, 166.1, 141.4, 140.7, 136.0, 129.8,
129.1, 122.2, 81.6, 73.9, 71.5, 70.7, 63.5, 61.8, 60.2, 57.1,
41.2, 40.7, 40.4, 40.3, 37.1, 37.0, 31.9, 30.2, 29.9, 29.6,
27.6, 27.1, 26.7, 26.5, 20.0, 18.8. Analysis, found: C, 55.8;
1
IR (KBr) n 3306, 2923, 1676 cm21; H NMR (CD3OD) d
7.45 (m, 8.95H), 6.74 (d, J512.5 Hz, 0.05H), 6.65 (d,
J515.8 Hz, 0.95H), 6.03 (d, J512.5 Hz, 0.05H), 4.57 (m,
4H), 4.22 (m, 2H), 4.09 (m, 4H), 3.55 (m, 10H), 3.09 (m,
2H), 2.73 (s, 6H), 2.10 (m, 2H), 0.92 (m, 12H); 13C NMR
(CD3OD) d 174.0, 173.6, 168.9, 141.5, 140.8, 138.2,
134.8, 132.7, 131.1, 129.7, 129.1, 124.8, 122.9, 122.1,
81.5, 73.2, 72.9, 72.1, 71.4, 70.8, 68.0, 60.2, 40.8, 40.6,
31.9, 26.4, 19.9, 18.7. Analysis, found: C, 49.4; H, 6.3; N,

206
M. Alterman et al. / European Journal of Pharmaceutical Sciences 13 (2001) 203–212
7.9%; C43H62BrF3N6O1312H2O (M51043.915) requires:
C, 49.5; H, 6.4; N, 8.1%.
protease, Q7K, concentration approximately 10 mM, was
diluted 6 times in acetate buffer (10 mM, pH 4.5). This
was then allowed to react with the activated esters for 25
min. After this the active N-hydroxysuccinimide esters
were deactivated using ethanolamine (1 M, pH 8.5) for 5
min. Finally an additional activation of the matrix was
2.1.7. N1,N6-Bis[(1S)-2-methyl-1-(methylcarbamoyl)-
propyl]-(2R,3R,4R,5R)-2[4-((E)-2-hN-[2-(2-h2-[6-
(biotiamido)carbamoylpentyl]ethoxyjethoxy)ethyl]j-
carbamoyletenyl)benzyloxy]-5-(4-bromobenzyloxy)-
3,4-dihydroxyhexanediamide (10)
made
using
N-ethyl-N9-(dimethylaminopropyl)-car-
bodiimide and N-hydroxysuccinimide for 5 min. The
resulting crosslinking is a stabilisation of the surface. This
procedure resulted in 5500 RU of immobilised protease on
the surface. This amount corresponds to approximately
0.25 pmol of HIV protease/mm2 of the sensor surface.
To a stirred mixture of 9 (36 mg, 0.036 mmol) and
diisopropylethylamine (16 ml, 0.09 mmol) in DMF (2 ml)
was added succinimidyl-6-(biotinamido)hexanoate (20.3
mg, 0.045 mmol). The reaction mixture was stirred over
night and then concentrated. The crude product was
purified by circular chromatography (CH2Cl2→CH2Cl2 /
MeOH (9:1)) to give pure 10 (42.1 mg, 0.034 mmol) in
2.2.2. Direct binding assay
HIV-1 protease was immobilised to the active channel.
Each sample cycle consisted of a 60-s buffer injection,
followed by a 60-s sample injection. For the screening
experiments the sample concentration was 10 mM. The
signal was adjusted for non-specific binding of the samples
to the dextran matrix by subtracting the signal in the
reference channel from the signal in the active channel.
The report points were; b1510 s before injection start and
d155 s after injection stop. The last injection consisted of
a 30-s regeneration solution. A washing step was per-
formed at the end of each cycle. It contained aqueous
DMSO (50%) that was used to wash the needle and the
sample line of the integrated fluid cartridge. The direct
binding assay needs a bulk compensation method to be
able to account for differences in DMSO concentration in
the sample injections. The calibration procedure was
performed with nine consecutive injections of buffer
samples with different DMSO concentrations. The DMSO
concentration varied equidistantly between 4 and 6%. A
report point was placed on each sample pulse and the
difference in response in the active and the reference
channel was compared. The difference was plotted against
the response in the reference channel and an exponential
function was fitted to the points. The function was then
used to compensate all measured responses for differences
depending on variation in DMSO concentration of the
sample.
1
96% yield: IR (KBr) n 3291, 2932, 1636 cm21; H NMR
(CD3OD) d 7.44 (m, 8.95H), 6.74 (d, J512.6 Hz, 0.05H),
6.66 (d, J512.5 Hz, 0.95H), 6.04 (d, J512.6 Hz, 0.05H),
4.57 (m, 4H), 4.48 (m, 1H), 4.27 (m, 1H), 4.21 (m, 2H),
4.09 (m, 4H), 3.59 (m, 10H), 3.34 (m, 2H), 3.17 (m, 2H),
2.90 (m, 1H), 2.73 (s, 6H), 2.71 (m, 1H), 2.27 (m, 6H),
1.45 (m, 13H), 0.92 (m, 12H); 13C NMR (CD3OD) d
176.0, 174.0, 168.8, 166.2, 141.4, 140.8, 138.3, 136.0,
132.7, 131.2, 129.8, 129.1, 122.8, 122.2, 81.5, 74.1, 73.8,
73.2, 72.8, 72.2, 72.1, 71.5, 71.4, 70.7, 63.5, 61.7, 60.2,
57.1, 44.0, 41.2, 41.2, 40.7, 40.4, 40.3, 37.0, 37.0, 31.93,
30.2, 29.9, 29.6, 27.7, 27.1, 26.8, 26.6, 26.4, 19.9, 18.7.
Analysis, found: C, 53.2; H, 7.3; N, 9.9%;
C57H86BrN9O14S1H2O (M51287.361) requires: C, 53.2;
H, 7.2; N, 9.8%.
2.2. SPR assays: instrumental setup
All experiments were performed on a BIACORE 3000
system (Biacore AB, Uppsala, Sweden) at 258C. A stan-
dard carboxymethylated Sensor Chip (CM5) was used.
Flow channel 1 was an unmodified CM5 surface and flow
channel 2 contained the active surface. The running buffer
contained phosphate (50 nM, pH 7.4), sodium chloride
(150 nM), polyoxyethylenesorbitan (0.01%). DMSO was
added to a concentration of 5% in both the running buffer
and the sample solution, to increase the solubility of the
tested compounds. The flow rate was 40 ml/min for all
assays and the binding responses was recorded continuous-
ly (1 Hz). The samples were automatically injected from
96-well microplates (Greiner, Frickenhausen, Germany).
2.2.3. Assay 1
HIV-1 protease was immobilised to the active channel.
Each sample cycle consisted of a 90-s injection of the
tested inhibitor (10 mM) and compound 7 (30 nM). The
first report point was b1510 s before the antibody in-
jection. This was followed by a 90-s injection of a
Monoclonal antibody against biotin (5 mg/ml). The sec-
ond report point was d155 s after the end of antibody
injection. The last injections consist of a 30-s regeneration
solution. A washing step was performed at the end of each
cycle. It contained aqueous DMSO (50%) that was used to
wash the needle and the sample line of the integrated fluid
cartridge.
2.2.1. Immobilisation of HIV-1 protease
The HIV protease mutant Q7K (Markgren et al., 2001)
was immobilised by direct coupling to primary amines of
the protease. Carboxyl groups in the surface matrix were
activated by treatment with a fresh mixture of N-ethyl-N9-
(dimethylaminopropyl)-carbodiimide (0.2 M) and N-hy-
droxysuccinimide (0.05 M) in water for 10 min. HIV-1

M. Alterman et al. / European Journal of Pharmaceutical Sciences 13 (2001) 203–212
207
2.2.4. Regeneration of HIV-1 protease immobilised
surface
mono protected diamine 3 (Zuckermann et al., 1994) with
acryloyl chloride as shown in Scheme 3. The dibromo
compound 2 was synthesised by a three-step procedure
from L-mannonic-g-lactone, by dibenzylation and ring
To ensure complete removal of bound high affinity
inhibitors from the HIV protease surface, a regeneration
mixture was developed. This mixture contains, ethyl-
eneglycol, sodium dodecyl sulphate (25 mM) in ethanol,
and MgCl2 (1 M), 1:1:1. The regeneration solution was
utilised in a 30-s pulse. A small decline of the surface
capacity was observed during repeated measurements, over
100 cycles the surface capacity diminished with approxi-
mately 30%.
opening of
a
dilactone intermediate with valine
methylamide, as previously described (Alterman et al.,
1999). The Boc-derivative 5 was obtained by a Heck
reaction (Heck, 1982; Beletskaya and Cheprakov, 2000) of
the dibromo derivative 2 with the unsaturated amide 4 in
large excess. The Heck reaction, performed with palladium
acetate/tri-o-tolylphosphine as the catalytic system and
with diisopropylethylamine as base, proceeded smoothly
and afforded the Boc amide 5 in good yield (Scheme 2).
This trans isomer, contaminated by a minor amount (5–
2.2.5. Immobilisation of compounds 6 and 9
Compounds 6 and 9 were immobilised by direct cou-
pling to the primary amines. Carboxyl groups in the
surface matrix were activated by treatment with a fresh
1
10%) of the cis isomer as determined by H NMR, was
found to exhibit a sufficiently high affinity, in the nanomo-
lar range (Ki51.3 nM), to encourage further studies.
Treatment with TFA afforded the amine 6, and subsequent
mixture
of
N-ethyl-N9-(dimethylaminopropyl)-car-
bodiimide and N-hydroxysuccinimide for 10 min. Then a 1
mM solution of 6 or 9 in sodium borate buffer (pH 8.5)
was allowed to react for 12 min. Finally the remaining
active N-hydroxysuccinimide esters were deactivated using
1 M ethanolamine for 5 min. This resulted in approximate-
ly 1500 RU of 6 or 9 immobilised surface. This corre-
sponds to approximately 1.2 pmol of 6 or 9/mm2 on the
sensor surface.
reaction
with
succinimidyl-6-(biotinamido)hexanoate
(NHS-LC-Biotin) in DMF with diisopropylethylamine as
base resulted in a 96% yield of the bisbiotinylated C2-
symmetric dihydroxyethylene transition-state mimic 7
(Scheme 2), able, in principle, to bind antibodies with both
of the biotin units connected to the P1/P19 arms.
By employing only 1.3 equivalents of the olefin 4 in the
palladium-catalyzed reaction the mono Boc-amide 8 could
be isolated, but in a low yield. This was attributed mainly
to the fact that a mixture of 8, 2, and 5 was formed, and to
the difficulties encountered in the subsequent purification
of 8 by chromatography (Scheme 2). Compound 8 was
contaminated by approximately 5% of the cis compound.
Deprotection by TFA provided the monoamine 9, in good
yield. Reaction with NHS-LC-Biotin in DMF with di-
isopropylethylamine as base ultimately resulted in a 96%
yield of the mono-biotinylated compound 10
2.2.6. Assay 2
Compounds 6 or 9 were immobilised to the active
channel. Each sample cycle consisted of a 90-s injection of
the tested inhibitor (10 mM) and HIV-1 protease (10 nM).
The report points were; b1510 s before injection start and
d155 s after injection stop. This was followed by a 30-s
injection of regeneration solution. A washing step was
performed at the end of each cycle. It contained aqueous
DMSO (50%) that was used to wash the needle and the
sample line of the integrated fluid cartridge.
The amino groups of 6 and 9 provided the ‘handles’ for
linkage to the carboxyl anchor group of the surface matrix
(Assay 2). The amide bond was formed after activation of
the carboxyl group with N-ethyl-N9-(dimethylaminop-
ropyl)-carbodiimide and N-hydroxysuccinimide.
2.2.7. Regeneration of the 6 or 9 immobilised surface
The surface was regenerated using an aqueous NaOH
solution (250 mM). This solution was in contact with the
surface for 30 s. No decline in the binding capacity was
observed for .100 cycles. This made it possible to use the
same chip for 3–4 weeks.
3.2. Assay
In the previously developed optical biosensor assay
(direct binding assay), HIV protease was immobilised to
the surface and the binding of inhibitors was monitored
(Markgren et al., 1998), as depicted in Fig. 1a.
3. Results and discussion
3.1. Chemistry
Assay 1 also utilises the surface with immobilised HIV
protease. However, in this new assay system the biotin-
labelled HIV protease inhibitor 7 is mixed with a test
compound and thereafter passed over the surface. The
amount of compound 7 bound to the HIV protease is
inversely proportional to the affinity and concentration of
the potential HIV protease inhibitor. The signal from the
binding of compound 7 is amplified using an anti-biotin
antibody that is passed over the surface in a second
Prior to embarking on a more extensive program we
decided to prepare and evaluate the a,b-unsaturated Boc-
protected derivative 5, characterised by an extended tether
arm aimed as a linker to biotin, or to a solid support on a
sensor chip. The synthesis of 5 is outlined in Scheme 2.
The a,b-unsaturated amide 4, a reactant for the preparation
of 5, was afforded in high yield by the reaction of the

208
M. Alterman et al. / European Journal of Pharmaceutical Sciences 13 (2001) 203–212
Scheme 2.
injection (Fig. 1b). The response from the antibody is in
turn proportional to the amount of bound biotin-labelled
compound (7).
compared to what is observed in the direct binding assay,
are demonstrated in assay 1 and 2 after binding high
molecular weight ligands (inhibitor/antibody and protease,
respectively) to the sensor surface.
In the second assay (assay 2) the mono amino-linker
substituted HIV protease inhibitor 9 is immobilised to the
surface and the binding of the HIV protease is measured.
The HIV protease is here mixed and pre-incubated with the
potential HIV protease inhibitor prior to the injection. The
amount of HIV protease that is bound to the surface is
inversely proportional to the affinity and the concentration
of the tested compound (Fig. 1c).
A schematic sensorgram, representing the signal as a
function of time for a complete analysis cycle of assay 1, is
shown in Fig. 3a. The sensorgram consists of data from
three sequential injections. The sample pulse contained the
tested compound and the biotin-labelled compound 7. The
second pulse contained the anti-biotin antibody used to
enhance the signal. Regeneration of the sensor surface was
necessary to remove bound inhibitors and biotin-labelled
compound (7)–antibody complex from the surface. A
subsequent system wash rinsed the autosampler and the
A comparison of the responses achieved, with nelfinavir
as test compound, in the three SPR assay systems is
presented in Fig. 2. Strong signal enhancements, as
Scheme 3.

M. Alterman et al. / European Journal of Pharmaceutical Sciences 13 (2001) 203–212
209
Fig. 1. Schematic picture of the different assay formats. Assay 1, surface competition assay; Assay 2, inhibition in solution assay.
injection unit. Report points were placed at appropriate
times in the sensorgram, to facilitate fast and easy evalua-
tion of experimental data.
Sensorgram data from a complete analysis cycle for
assay 2 is shown in Fig. 3b. The sensorgram consists of
data from two sequential injections. Firstly the sample
pulse that contained the tested compound and HIV protease
was injected. The second injection contained the regenera-
tion solution. Also in this assay report points were used for
easy evaluation of experimental data.
reported direct binding assay, and with the two new assay
formats. The response as a function of the inhibitor
concentration was measured (Fig. 4). Firstly the direct
binding assay was evaluated. The response as a function of
the concentration was measured for each inhibitor. The
immobilised HIV protease slowly decomposed and there-
fore indinavir at 500 nM was injected as a reference every
12th sample, to account for an observed decline in the
binding capacity of the surface. The decline of the surface
capacity after 100 cycles was estimated at approximately
30%. An exponential function was fitted to the reference
responses. The responses from all samples were normal-
ised according to this function and plotted against the
concentration (Fig. 4a). The direct binding assay did not
allow for ranking of the affinity of the four inhibitors, as
the differences in observed IC50 values were smaller than
the deviations caused by instrument noise.
Four inhibitors in clinical use, indinavir, nelfinavir,
ritonavir, and saquinavir were analysed with the previously
In assay 1, compound 7 and the anti-biotin antibody
were tested every 12th cycle, to account for the decline in
the binding capacity of the enzyme surface. The response
for all samples was compared to the uninhibited response
and the percent inhibition was calculated. (Fig. 4b) Assay
1 permitted a clear distinction between the four protease
inhibitors: the differences in the measured IC50 values
were significantly greater than the errors caused by back-
ground noise. This made it possible to rank the different
commercial inhibitors with respect to their affinities for the
Fig. 2. Dose–response curves for nelfinavir in the three assays formats
(corrected). RU, resonance unit.

210
M. Alterman et al. / European Journal of Pharmaceutical Sciences 13 (2001) 203–212
Fig. 3. Schematic sensorgram illustrating the experimental design for identification of HIV protease inhibitors. Thick arrows represent injections. Thin
arrows represent reporter points used to provide a reduced data set. (a) Assay 1: Sample, injection of test compound and compound 7; Antibody, anti-biotin
antibody injection; Regeneration, wash pulse with ethylene glycol, sodium dodecyl sulphate, MgCl2; b1, report point of baseline; A, start of antibody
injection; D, end of antibody injection; d1, report point of dissociation phase. (b) Assay 2: Sample, injection of test compound and HIV protease;
Regeneration, wash pulse with 250 mM NaOH; b1, report point of baseline; A, start of sample injection; D, end of sample injection; d1, report point of
dissociation phase.
surface bound HIV protease. The order of the inhibitors,
sorted from high to low affinity: saquinavir, ritonavir,
indinavir, and nelfinavir.
For assay 2 we first designed the mono substituted
amino compound 9 for attachment to the dextran surface of
the biosensor. The next step was to evaluate whether the
bis-functionalised compound 6, that was considerably
more accessible synthetically than 9, could be employed
and if cross-linking to the dextran matrix could be avoided.
The second linker arm did not exert any pronounced effect
on the result.
Inhibitor evaluation by monitoring of concentration
series is time consuming and uses a significant amount of
protein. The time taken to perform an analysis could be
shortened considerably if single runs at fixed concen-
trations could provide accurate data. To address this issue
the results obtained with the optical biosensor were
compared to the Ki values obtained with a standardised
fluorometric assay. The analysis set consisted of 17 test
compounds that were taken from five different structural
series spanning an affinity range from 0.1 nM to 10 mM
(Fig. 5). In all the three assay formats the test compounds
were injected 3 times in three consecutive series with a
reference injection every 12th cycle.
In assay 2, HIV protease was injected as every 12th
cycle to allow any decline in the protease activity or for
the surface immobilised inhibitor to be accounted for. The
surface was found to be very durable and could therefore
be utilised more than 100 cycles and for a long period of
time with no significant degradation. The responses for all
samples were compared to the uninhibited response and
the percent inhibition was calculated. (Fig. 4c) The con-
centration series could be used to calculate the observed
IC50 values. Minor errors due to instrument noise and
small differences in protease concentrations were signifi-
cant in relation to IC50 values. Thus, the assay 2 did not
allow any ranking of the high affinity binders regarding
their affinity for HIV protease.
In assay 1 the bis-biotinylated compound 7 can bind
either one or two anti-biotin antibodies. To evaluate the
importance of the second biotin-subunit, for enhancement
of the SPR-signal, the mono functionalised compound 10
was used in assay 1. The result clearly showed a better
response with compound 7 than with 10, although the
signal was not double the magnitude, as would have been
expected if both of the arms of 7 were engaged in antibody
binding simultaneously. In fact only a 30% enhancement
of the response was encountered with 7.
In the direct binding assay an exponential function, to
account for the surface decline, was fitted to the d1
responses of the indinavir injections, the first report point
in the dissociation phase. This function was used to
normalise the responses for all samples. The normalised
responses were then plotted against the Ki values (Fig. 6a).
Fig. 4. The dose–response curves for indinavir (d), nelfinavir (j), ritonavir (m), and saquinavir (3).

M. Alterman et al. / European Journal of Pharmaceutical Sciences 13 (2001) 203–212
211
Fig. 5. Indinavir, nelfinavir, ritonavir, saquinavir, and 17 test compounds.
The responses from direct binding assay showed correla-
tion with the Ki values for compounds with Ki values
between 1 and 100 nM. Compounds with higher Ki values
than 100 nM were impossible to rank due to background
noise. These compounds could hence only be characterised
as low affinity binders of HIV protease.
Fig. 6. The normalised responses or percent inhibition for the 17 compounds plotted against their Ki values.

212
M. Alterman et al. / European Journal of Pharmaceutical Sciences 13 (2001) 203–212
In assay 1 an exponential function was fitted to the d1
References
responses from the reference injections. This function was
then used to normalise all responses. The percent inhibi-
tion was calculated from the normalised responses and
plotted against the Ki values from the standardised fluoro-
metric assay (Fig. 6b). The percent inhibition from assay 1
showed good correlation (R250.93) to the Ki values for
compounds with Ki values over the tested range.
Additionally in assay 2 the d1 responses from the
reference injection were used. The d1 response for all
samples was then normalised against this function and the
percent inhibition was calculated. The percent inhibition
was then plotted against the Ki values (Fig. 6c). The
percent inhibition from assay 2 showed correlation to the
Ki values for inhibitors with Ki between 1 and 100 nM.
Compounds with Ki values outside this range could be
characterised as high and low affinity binders, respectively,
but it was not possible to determine the internal ranking
within these groups.
´
¨
Alterman, M., Andersson, H.O., Garg, N., Ahlsen, G., Lovgren, S.,
¨
Classon, B., Danielson, U.H., Kvarnstrom, I., Vrang, L., Unge, T.,
Samuelsson, B., Hallberg, A., 1999. Design and fast synthesis of
C-terminal duplicated potent C2-symmetric P1/P19-modified HIV-1
protease inhibitors. J. Med. Chem. 42, 3835–3844.
Beletskaya, I.P., Cheprakov, A.V., 2000. The Heck Reaction as
a
sharpening stone of palladium catalysis. Chem. Rev. 100, 3009–3066.
Floyd, C.D., LeBlanc, C., Whittaker, M., 1999. In: King, F.D., Oxford,
A.W. (Eds.). Combinatorial Chemistry as a Tool for Drug Discovery.
Progress in Medicinal Chemistry, Vol. 26. Elsevier, Amsterdam, pp.
91–168.
Heck, R.F., 1982. In: Dauben, W.G. (Ed.). Palladium-Catalyzed Vinyla-
tion of Organic Halides. Organic Reactions,Vol. 27. Wiley, New York,
pp. 345–390.
Heck, R.F., 1985. In: Palladium Reagents in Organic Syntheses. Academ-
ic Press, London, pp. 2–3.
Jonsson, U., Fagerstam, L., Ivarsson, B., Johnsson, B., Karlsson, R.,
¨
¨
¨ ˚
¨
¨
Lundh, K., Lofas, S., Persson, B., Roos, H., Ronnberg, I., Sjolander,
¨
˚
S., Stenberg, E., Stahlberg, R., Urbaniczky, C., Ohstlin, H.,
Malmqvist, M., 1991. Real-time biospecific interaction analysis using
surface plasmon resonance and a sensor chip technology. Biothech-
niques 11, 620–627.
Karlsson, R., 1994. Real-time competitive kinetic analysis of interactions
between low-molecular-weight ligands in solution and surface-im-
mobilized receptors. Anal. Biochem. 221, 142–151.
4. Conclusion
¨ ˚
Lofas, S., Johnsson, B., 1990. Novel hydrogel matrix on golden surfaces
In summary, high molecular weight ligands enhance the
response of the surface plasmon resonance signal, a
phenomenon that both of the assay formats take advantage
of to increase the sensitivity. X-ray structural analysis
guided the design of high affinity HIV protease inhibitors
exploited as tools in two new screening assays used for the
identification of low molecular weight compounds that
interact with the HIV protease. The most sensitive assay
was obtained by immobilising the enzyme and amplifying
the signal with an antibody, giving a detection range
between 0.1 nM and 10 mM. Immobilisation of the
inhibitor resulted in a narrower detection range (1–100
nM). Immobilisation of the inhibitor as compared to
immobilisation of the protease to the surface provided a
more durable and stable system. The new assay formats
permit highly automated screening over a wide affinity
range. The deviations in the test results are small when
compared to those encountered with the standardised
fluorometric assays. We foresee a utility of the method
presented herein also in other binder/protein systems
where there are no natural or stable substrates available
and in particular in systems where the access of the protein
is limited.
in surface plasmon resonance sensors for fast and efficient covalent
immobilization of ligands. J. Chem. Soc., Chem. Commun. 21, 1526–
1528.
Malmquist, M., 1993. Biospecific interaction analysis using biosensor
technology. Nature 361, 186–187.
Markgren, P.-O., Hamalainen, M., Danielson, U.H., 1998. Screening of
compounds interacting with HIV-1 proteinase using optical biosensor
technology. Anal. Biochem. 265, 340–350.
Markgren, P.-O., Lindgren, M.T., Gertow, K., Karlsson, R., Hamalainen,
M., Danielson U.H., 2001. Determination of interaction kinetic
constants for HIV-1 protease inhibitors using optical biosensor tech-
nology. Anal. Biochem. (in press).
Matayoshi, E.D., Wang, G.T., Krafft, G.A., Erickson, J., 1990. Novel
fluorogenic substrates for assaying retroviral proteases by resonance
energy transfer. Science 247, 954–958.
Nashed, N.T., Louis, J.M., sayer, J.M., Wondrak, E.M., Mora, P.T.,
Orsoszlan, S., Jerina, D.M., 1989. Continuous spectrophotometric
assay for retroviral proteases of HIV-1 and AMV. Biochem. Biophys.
Res. Commun. 163, 1079–1085.
Stenberg, E., Persson, B., Roos, H., Urbaniczky, C., 1991. Quantitative
determination of surface concentration of protein with surface plasmon
resonance using radiolabeled proteins. J. Colloid Interface Sci. 143,
513–526.
Stolz, L., 1998. In: Bristol, J.A. (Ed.). The Use of Surface Plasmon
Resonance Based Biosensor in Drug Discovery. Annual Reports in
Medicinal Chemistry, Vol. 33. Academic Press, San Diego, CA, pp.
293–299.
¨
¨ ¨
¨
¨ ¨
Tomasselli, A.G., Heinrikson, R.L., 2000. Targeting the HIV-protease in
AIDS therapy: a current clinical perspective. Biochim. Biophys. Acta
1477, 189–214.
Tomaszek, Jr. T.A., Magaard, V.W., Bryan, H.G., Moore, M.L., Meek,
T.D., 1990. Chromophoric peptide substrate for the spectrophotometric
assay of HIV-1 protease. Biochem. Biophys. Res. Commun. 168,
274–280.
Zuckermann, R.N., Martin, E.J., Spellmeyer, D.C., Stauber, G.B.,
Shoemaker, K.R., Kerr, J.M., Figliozzi, G.M., Goff, D.A., Siani, M.A.,
Simon, R.J., Banville, S.C., Brown, E.G., Wang, L., Richter, L.S.,
Moos, W.H., 1994. Discovery of nanomolar ligands for 7-transmem-
Acknowledgements
We gratefully acknowledge support from the Swedish
Foundation for Strategic Research (SSF) and the Swedish
National Board for Technical Development (NUTEK). We
¨
also gratefully acknowledge Anna Muhlman, Jonas
˚
´
Branalt, Gunilla Niklasson, and Rolf Noreen for synthesis
of some of the test compounds.
brane G-protein-coupled receptors from
a
diverse N-(sub-
stituted)glycine peptoid library. J. Med. Chem. 37, 2678–2685.