Synthesis and optimization of peptidomimetics as HIV entry inhibitors against the receptor protein CD4 using STD NMR and ligand docking

Article abstract of DOI:10.1039/b605380g

We recently described the design and synthesis of a novel CD4 binding peptidomimetic as a potential HIV entry inhibitor with a K_D value of ～35 M and a high proteolytic stability [A. T. Neffe and B. Meyer, Angew. Chem., Int. Ed., 2004, 43, 2937-2940]. Based on saturation transfer difference (STD) NMR analyses and docking studies of peptidomimetics we now report the rational design, synthesis, and binding properties of 11 compounds with improved binding affinity. Surface plasmon resonance (SPR) resulted in a K_D = 10 M for the best peptidomimetic XI, whose binding affinity is confirmed by STD NMR (K_D = 9 M). The STD NMR determined binding epitope of the ligand indicates a very similar binding mode as that of the lead structure. The binding studies provide structure activity relationships and demonstrate the utility of this approach. The Royal Society of Chemistry 2006.

Full text of DOI:10.1039/b605380g

View Article Online / Journal Homepage / Table of Contents for this issue
PAPER
www.rsc.org/obc | Organic & Biomolecular Chemistry
Synthesis and optimization of peptidomimetics as HIV entry inhibitors
against the receptor protein CD4 using STD NMR and ligand docking†
Axel T. Neffe, Matthias Bilang and Bernd Meyer*
Received 18th April 2006, Accepted 20th June 2006
First published as an Advance Article on the web 26th July 2006
DOI: 10.1039/b605380g
We recently described the design and synthesis of a novel CD4 binding peptidomimetic as a potential
HIV entry inhibitor with a KD value of ∼35 lM and a high proteolytic stability [A. T. Neffe and
B. Meyer, Angew. Chem., Int. Ed., 2004, 43, 2937–2940]. Based on saturation transfer difference (STD)
NMR analyses and docking studies of peptidomimetics we now report the rational design, synthesis,
and binding properties of 11 compounds with improved binding affinity. Surface plasmon resonance
(SPR) resulted in a KD = 10 lM for the best peptidomimetic XI, whose binding affinity is confirmed by
STD NMR (KD = 9 lM). The STD NMR determined binding epitope of the ligand indicates a very
similar binding mode as that of the lead structure. The binding studies provide structure activity
relationships and demonstrate the utility of this approach.
resistant to a CD4 binding drug, the virus would have to switch
Introduction
to a totally different entry mechanism to infect macrophages and
The first step in the infection of a human cell with HIV is the
T-cells. However, this is not very probable as it would require the
interaction between the viral envelope glycoprotein gp120 and
mutation of many amino acids simultaneously forcing the virus to
human CD4.1 After this event, gp120 interacts with a coreceptor,2
establish a new and highly efficient entry mechanism.
normally CCR5 or CXCR43 to infect the human cell.4 The three
main classes of clinical anti HIV drugs, i.e. nucleoside reverse tran-
scriptase inhibitors (NRTIs), non nucleoside reverse transcriptase
inhibitors (nNRTIs), and protease inhibitors (PIs) inhibit viral
replication after infection of the cell.5 Despite advances in HIV
treatment,6 none of these drugs nor their combination can finally
cure HIV infection.7 Due to the high mutation rate of HIV and the
resulting resistance against drug treatment,8 new classes of drugs
have to be developed.9
A promising concept is the inhibition of one of the interac-
tions resulting in membrane fusion.10 These—so called—entry
inhibitors could assist in HIV treatment as is demonstrated by the
success of fuzeon (enfuvirtide, T20).11 Likewise, other compounds
preventing receptor or coreceptor binding of the virus could
potentially be utilized as HIV medication.12
Targeting the interaction of gp120 with CD4, others developed
gp120 or antibody related peptides,13 or gp120 binding molecules.14
The approach of developing CD4 ligands might potentially
interfere with the human immune system, as CD4 plays an
important role in the binding of MHC class II proteins to T cell
receptors.15 The region on the CD4 protein that binds to gp120
is overlapping the one that binds to the MHC class II proteins.16
However, the contact area of the gp120–CD4 interaction is much
larger (and therefore stronger) than that of the CD4–MHC class
II protein interaction.17
We reported a peptidomimetic compound I (cf. Fig. 1),18 derived
from the known CD4 binding peptide NMWQKVGTPL that
shows an antiviral activity in an HIV proliferation assay.19 This
was achieved by eliminating all amino acids of the lead compound
that do not contribute to binding, i.e. asparagine, methionine,
and glutamine. The two hydrophobic amino acids tryptophane
and leucine were replaced by generic hydrophobic non amino
acid residues. Additionally, non peptidic linkages have been
incorporated. I shows a 170 fold stronger binding (KD = 35 lM)
to CD4 than the original peptide, a 4–5 times higher proteolytic
stability, and a lower molecular weight of ≈800 compared to
the lead peptide. Therefore, I has much better pharmacological
properties compared to the lead decapeptide NMWQKVGTPL.
Here we describe the optimization of the pharmacological
properties of this class of compounds. Starting from the lead
peptide we docked about 85 peptidomimetics that were created
by changing one residue at a time to test for the optimal residues
at each position. Of these, eleven compounds were chosen for
synthesis and nine of them were fully characterized by 2D
NMR techniques. The binding affinities of all substances to
CD4 were determined by surface plasmon resonance (SPR), the
best compound in this assay was also analyzed with saturation
transfer difference (STD) NMR for its KD and binding epitope.
The combined results gave information about structure activity
relationships.
On the other hand, blocking the CD4 protein on the human cell
is much less likely to cause viral resistance. In order to become
Results and discussion
Institute for Organic Chemistry, University of Hamburg, Martin Luther
King Platz 6, 20146, Hamburg. E-mail: bernd.meyer@chemie.uni-
hamburg.de, axel@neffe.org; Fax: +49 40 428382878
† Electronic supplementary information (ESI) available: SPR graphs,
characterization data and docking results of the compounds with variation
of the aromatic residues. See DOI: 10.1039/b605380g
Docking
Extensive docking studies were performed using the program
Flexidock of the Sybyl (version 6.9) software package (Tripos,
Inc.). Due to the simplifications of the docking procedure, e.g.
This journal is
The Royal Society of Chemistry 2006
Org. Biomol. Chem., 2006, 4, 3259–3267 | 3259
©

View Article Online
Fig. 1 a) The template on which the design of the newly synthesized peptidomimetics is based. The peptidomimetics consist of a core pentapeptide
carrying non-peptidic, hydrophobic residues on the N- and C-termini. b) Structure of the peptidomimetic I with a KD = 35 lM.
no consideration of solvent effects, the calculated binding energies
can only be regarded as semi quantitative indicators of the binding
affinity. The docked model of I (Fig. 2) served as starting position
for docking studies of other ligands. First the corresponding
functional group(s) were replaced and, subsequently, the resulting
structures were minimized for a short period each. The docking
results were interpreted in terms of the existing experimental data,
like X-ray analyses, STD NMR, and binding studies for related
compounds.
IV, Chart 1). Compounds with an isoquinoline, benzodiazole,
benzothiazole, indole, or biphenyl group on the N-terminus
did not show favorable interactions with CD4 in the docking
experiments (details see supporting information†).
We also varied the linkers of the peptidomimetics that connect
the core peptide and the aromatic ring (cf. Fig. 3). The best
interaction energies were observed for the oxyacetic acid linker
1 and for similar longer linkers 6a and 7, cf. Table 1. Because of
In the docking process, we studied different aromatic residues,
different linkages between the aromatics and the core peptide,
as well as different amino acids forming the core of the pep-
tidomimetics (cf. Fig. 1a) to improve the affinity of I. The N-
terminal linker and the aromatic residues were selected during
the design phase of the template without consideration of the
cyclohexylcarbamoyl subunit at the C-terminus. After selection
the complete peptidomimetics were docked and their energy
evaluated.
A variety of aromatic ring systems, selected by similarity
search in the available chemicals directory (ACD), were docked
to potentially improve the effect of the b-naphthyl residue.
Compounds with a quinolyl and 4-methylcoumarin group on the
N-terminus showed calculated binding affinities close to that of
the original aromatic b-naphthyl system and were therefore chosen
for synthesis, as was the a-naphthyl derivative to elucidate the
ideal substitution pattern at the aromatic ring (cf. II, III and
Fig. 3 Different types of linkers analyzed in the docking.
Fig. 2 Stereo view of the docked ligand I shown as ball and stick on the surface of the CD4 protein (green). The excellent complementarity of the surface
to the ligand is clearly visible.
3260 | Org. Biomol. Chem., 2006, 4, 3259–3267
This journal is
The Royal Society of Chemistry 2006
©

View Article Online
Chart 1 Structures of the peptidomimetics II–XII.
Table 1 Calculated binding energies of the ligand–protein complexes of
peptidomimetics with different linkages and energy differences relative to
the oxyacetic acid linker
these findings and the ease of synthesis, an amide bond between
the core and the aromatic residue was chosen for synthesis. The
three atom linker of 1 seemed to be a good compromise between
the right distance and flexibility, even though the longer linkers
gave a better score. To validate this hypothesis, two compounds
with the linkers 5 and 6b were synthesized (cf. V and VI). 6b was
chosen as it was commercially available contrary to 6a.
The lysine of the peptidomimetic I shows a tight interaction
with Asp63 of CD4 in docking experiments (Fig. 4). Side chains
without a terminal basic functional group at this position lead to a
complete loss of the binding affinity. Therefore, histidine, arginine
and citrulline were tested as potential replacements for the lysine
residue. The side chain of histidine is too short to interact with
Linker
Energy/kcal mol−1
DE/kcal mol−1
1
−274
−264
−261
−258
—
0
10
13
16
—
2
3
4
5a
6a
7
−292
−18
−290
−16
a For this compound, no binding energy could be determined, because of
spatial separation of the ligand protein complex during docking.
This journal is
The Royal Society of Chemistry 2006
Org. Biomol. Chem., 2006, 4, 3259–3267 | 3261
©

View Article Online
Synthesis and characterization
Chart 1 shows all synthesized peptidomimetics. All building
blocks were purchased in the highest possible purity. (8ꢀ-
Quinolyloxy)acetic acid, (1ꢀ-naphthyloxy)acetic acid, and (6ꢀ-(4ꢀ-
methyl)coumarinyloxy)acetic acid were synthesized by Williamson
ether synthesis of the respective aromatic alcohol and bromo acetic
acid.21 The peptidomimetics were synthesized in parallel, following
the published procedure as shown in Scheme 1 for I.18 The solid
phase synthesis gave yields for the unpurified primary alcohol
of 20–35% (equivalent to 72–80% per coupling–deprotection
cycle). The relatively low yields are due to problematic coupling
reactions connecting to the sterically demanding prolinol and
valine residues, respectively. The carbamates at the C-terminus
were formed in 62–80% yield. Deprotection and RP-HPLC
purification resulted in overall yields of 1–7% of isolated product,
respectively. XII (Chart 1) was the result of an incomplete coupling
of the isocyanate22 and was also tested for binding affinity. For the
peptidomimetic I, two stable conformers could be detected by
2D NMR spectra.18 The same findings were obtained for II–X.
However, this equilibrium was not observed for XI and XII which
both have an alaninol subunit substituting for the prolinol. This
reinforces the assumption that the conformers are the cis and trans
rotamers, respectively, of the threonine prolinol amide bond.
Fig. 4 Stereo view of the interaction of the lysine, valine and glycine
of I (ball and stick) with CD4 (shown as a surface with the patches of
the surface carrying the color of the associated atoms; red: oxygen, blue:
nitrogen, cyan: hydrogen, white: carbon).
Asp63. Arginine and citrulline were estimated to have roughly equal
binding energies, despite having different pKB values. We decided
to synthesize the citrulline containing compound, because of the
expected higher proteolytic stability (cf. VII, Chart 1).
Valine fits into a hydrophobic cavity formed of Trp62, Ser42,
Phe43, Arg59, Ser60, and Ser23 (Fig. 4). To some degree this cavity is
flexible and there may be place for bigger hydrophobic residues.20
Therefore, a peptidomimetic in which cyclohexylglycine (Chg)
substitutes for valine was synthesized (cf. VIII, Chart 1).
The glycine does not show a direct interaction with CD4 in
the STD NMR experiments of I. Therefore, it was an obvious
candidate for substitution (Fig. 4). An alanine substitution leads to
a reduced calculated binding affinity. Alternatively, peptidomimet-
ics containing amino acids with a basic (histidine) and an acidic
(glutamate) side chain in this position were docked. While the
incorporation of histidine resulted in a lower calculated binding
affinity, glutamate increases the calculated binding affinity because
of a possible interaction with Ser60 of the CD4 (cf. X, Chart 1).
Since the amidic proton of glycine is not participating in hydrogen
bonding to CD4, the substitution with sarcosine should not have
a significant influence on the binding energy but could increase
proteolytic stability (cf. IX, Chart 1).
Interactions of threonine and prolinol with CD4 are predicted
by Flexidock (cf. Fig. 5). All attempts to substitute the threonine
residue led to a worse binding energy. The in silico alanine scan of
the lead peptide showed that substitution of proline with alanine
resulted in a better binding energy. Therefore, we replaced prolinol
with alaninol (cf. XI, Chart 1).
Scheme 1 Synthesis of the peptidomimetics as exemplified for I.
SPR analysis
The KD-values of all substances have been determined by SPR. The
compounds were passed over a Biacore CM5 chip with immobi-
lized CD4. About 53 fmol (44 fmol for V and VI, respectively) of
Fig. 5 Stereo view of threonine and proline of I binding to CD4 in the
docking. The intensity of the proline Hb protons observed in the epitope
mapping is not explained by this model.
3262 | Org. Biomol. Chem., 2006, 4, 3259–3267
This journal is
The Royal Society of Chemistry 2006
©

View Article Online
Table 2 The calculated binding energies from the docking and the
respective experimental KD values including the standard deviation of
the newly synthesized peptidomimetics II–XII in comparison to I. The
structures of II–XII are shown in Chart 1. Also shown is the highest
concentration that was used for the regression analysis. Depending on the
hydrophobic nature of the peptidomimetics I–XII, aggregation and thus
unspecific binding starts at different concentrations for each molecule thus
requiring different cut-offs for the calculation of the binding constant
Calculated binding
energy/kcal mol−1
Highest fitted
concentration/lM
Compound
KD/lM
I
−376
−370
−379
−382
−374
−376
−344
−334
−387
−400
−372
−359
39
55 19
16
26
105 33
53 102
31
5
32
30
30
100
100
25
50
50
100
30
30
III
III
1V
V
4
4
VI
VII
VIII
IX
X
XI
XII
6
146 94
26
30
10
5
7
4
103 50
500
slow binding kinetic, possibly indicating an induced fit and/or
conformational changes of the ligand.
STD NMR
The most promising compound XI has been analyzed in detail
by 1D 1H-STD NMR spectroscopy (Fig. 7 and 8).23 The KD
values (Table 3) determined by this homogeneous NMR based
method are in good agreement with the value obtained by the
Fig. 6 Fit of the SPR data points to a one site binding model of the
compounds IV (KD = 26 lM) and XI (KD = 10 lM). The data show the
results from two independent concentration dependent SPR experiments.
Response units (RU) given by the SPR experiment are directly correlated
to the amount of ligand bound (1 RU is approximately equivalent to 1 pg
of substance).
CD4 were active after immobilization. A concentration dependent
binding assay with 4, 6, 8, 10, 12, 16, 20, 25, 30, 50, 100, 500 and
1000 lM of the corresponding ligand in buffer was performed. The
data points of the lower concentrations of all ligands can be fitted
to a one-site binding model, while for higher concentrations a
linear correlation of the binding affinity with the concentration
was found. This behaviour indicates a non specific (i.e. non
saturable) binding, resulting from secondary binding sites on the
protein of very low affinity or aggregation of the ligands at higher
concentrations. KD values were therefore determined only with the
data points following a saturable binding behaviour. Regression
of these data points using a one-site binding model to eqn (3)
yielded the binding constants. All data are presented in Table 2.
The results of the binding assays for IV and XI are depicted in
Fig. 6, the others are shown in the supporting information†. In
general, the fit is good, with exception of VI, whose data points
have a great variance (see supporting information†).
The best compound in this assay, XI, has a KD value of 10 lM,
which is about 4-fold better than that of I (KD = 39 lM from
SPR, KD = 31 lM from STD NMR) and 600 fold better than
the lead NMWQKVGTPL (KD = 6 mM). The kon and koff values
could be determined by the analysis of individual SPR curves. The
kon values for the peptidomimetics range from 6 × 103 s−1 M−1 to
3 × 104 s−1 M−1, while koff is about 0.1–0.2 s−1. This is a relative
Fig. 7 a) 1H NMR spectrum of CD4. The typical broad lines for proteins
are observed with a few sharp signals resulting from amino acids on the
surface of CD4. b) 1H NMR spectrum of a mixture of 4.4 lM CD4 and
75 lM XI containing 20 lM DSS as internal standard. c) Corresponding
1H STD NMR spectrum of the same sample with T1q filter for the reduction
of most protein signals. Most of the remaining signals could be identified
as signals of XI. A titration experiment with different concentrations of
the ligand was used to determine KD. d) 1H NMR reference spectrum of
XI.
This journal is
The Royal Society of Chemistry 2006
Org. Biomol. Chem., 2006, 4, 3259–3267 | 3263
©

View Article Online
Table 3 KD values of individual protons of XI, calculated through a
regression analysis of the data points of the STD NMR titration by fitting
a one site binding model
Proton
KD/lM
Naph-H4/5/8
Naph-H6/7
Naph-H1/3
Val-Hb
Cyclohexyl-H4
Cyclohexyl-H3
Thr-Hc
12
30
16
47
35
23
9
Val-Hc
25
It shows comparable regions of the molecule to be in close contact
to CD4 as in the lead peptide NMWQKVGTPL and in I.18
Fig. 8 STD NMR titration curves for a choice of individual protons of
XI. All data are presented in Table 3.
heterogeneous SPR system. The different values for individual
protons are due to varying distances to the protein surface and
because of diverse relaxation times of the protons. KD values from
those protons yielding the lowest KD values and thus having the
tightest binding to the receptor are presented. This results in a KD
value of 9 lM by STD NMR in comparison to 10 lM by SPR.
The most intensive peaks in the 1D STD NMR 1H NMR
spectrum were used to derive the binding epitope of the ligand,
which is shown in Fig. 9 and is further visualized in Fig. 10 and 11.
Structure activity relationships
The analysis of the KD values and the docking results provide
insight into structure activity relationships (for ligand structures
see Chart 1). Aromatic systems containing hetero atoms (III and
IV) bind better than I. Also, by comparing the KD values of I
and II it becomes clear that b-naphthyl residues bind better than
a-naphthyl residues. The oxyacetic acid linker proved to be the
best choice accommodating the requirements for distance and
Fig. 9 Ligand epitope mapping of XI. Highlighted protons shown in red are in close contact to CD4.
Fig. 10 XI (ball and stick, with the atom colored surface of the binding epitope) docked to CD4 (green).
Fig. 11 The surface of the binding epitope of XI determined by 1H STD NMR. In fact, only these parts of the ligand are needed for binding.
3264 | Org. Biomol. Chem., 2006, 4, 3259–3267
This journal is
The Royal Society of Chemistry 2006
©

View Article Online
flexibility as is obvious from the binding affinity of I, V, and VI.
This was already suspected during the design stage. Compound
VII containing citrulline instead of lysine binds slightly better than
I. The worst binding affinity was measured for VIII, indicating that
the hydrophobic cavity that normally encloses the valine residue
is not large enough for the cyclohexyl residue. Glycine may be
substituted by either sarcosine or glutamate to improve the binding
affinity slightly, while the replacement of prolinol with alaninol in
XI clearly results in an improved binding affinity. This is partly
due to the fact that XI has only one stable conformer in solution.
The 10 fold decrease in binding affinity of XII in comparison to XI
shows the importance of the cyclohexylcarbamoyl group. Further
modifications of this carbamoyl residue may result in even better
binding compounds.
Synthesis of the peptidomimetics
The synthesis of the peptidomimetics has been described before.18
The synthesis starts with an amino alcohol linked to a 2ꢀ-
chlorotrityl resin via the hydroxy function. The following amino
acids are coupled using standard peptide coupling protocols
(activation by TBTU–DIPEA, Fmoc as N-terminal protecting
group and Boc and tBu, respectively, as side chain protecting
groups). The coupling of the terminal carboxylic acid was
accomplished in the same way via a normal amide bond. Mild
acidic conditions (1% TFA in DCM) cleave the molecule from
the resin and result in a side chain protected free alcohol. The
alcohol function reacts under copper(I) catalysis with cyclohexyl
isocyanate forming a carbamate linkage to give the protected
peptidomimetics. Subsequent treatment with 95% TFA gives the
unprotected peptidomimetic, which all have been characterized
by MALDI TOF MS. Overall yield of the products was 1–7%
after HPLC purification. All NMR data of II, III and V–XI are
shown in the supporting information†, IV and XII have not been
characterized by NMR due to low yield.
Experimental
Synthesis
Synthesis of the (aryloxy)acetic acids. 1 eq. aromatic alcohol
was refluxed in ethanol (2 mL mmol−1). A solution of KOH
(2.17 eq.) in ethanol (0.8 mL mmol−1) was added and subse-
quently a solution of bromo acetic acid (1.03 eq.) in ethanol
(0.5 mL mmol−1) was added dropwise. The solution was cooled
down to room temperature and stirred for another 12 hours ((8-
quinolyloxy)acetic acid forms a colorless precipitate during this
time and can be used without further workup). The ethanol was
evaporated, the residue dissolved in 20 mL ethyl acetate and
extracted three times with 10 mL of 5% aqueous NaHCO3. The
combined aqueous phases were acidified with conc. HCl to pH 2–
3 and extracted three times with 10 mL ethyl acetate. The organic
layer was dried over MgSO4 and the solvents were evaporated.
Drying in high vacuum gave the desired product which needed no
further purification.
Experimental details of the docking
Docking studies were carried out on Silicon Graphics Octane or
Octane2 computers using the program Flexidock of the Sybyl
(version 6.9) software package (Tripos, Inc.). The structure of CD4
was obtained from a Protein Data Bank file (code 1gc1.pdb)2b and
was treated as described before.18 The force field calculations take
into account van der Waals, electrostatic, torsional, and constraint
energy terms. All bonds of the ligand were freely rotatable, as well
as the amino acids of the CD4 binding epitope to allow an induced
fit docking.
Experimental details of SPR experiments
The SPR data were determined on a BIACORE 3000 instru-
ment using CM5 chips. 106 fmol and 220 fmol, respectively,
sCD4 were covalently attached to the activated dextrane matrix.
The surface was activated by N-hydroxysuccinimide and N-
(3-dimethylaminopropyl)-N-ethylcarbodiimide hydrochloride for
the coupling of the protein. After coupling of the protein
the matrix was capped with ethanolamine. The activity of the
immobilized protein was checked by gp120 binding; extrapolation
of the data showed an activity of 50% (53 fmol) and 20%
(44 fmol), respectively, of the immobilized CD4.24 sCD4 was
obtained from Progenics Pharmaceuticals, Tarrytown, New York,
USA. Catalogue-No.: PRO1008-1, Lot 48 (recombinant human
soluble CD4 (CHO), purity >95%, amino acids 1–370 of the
natural CD4 (45 kDa)), gp120 was purchased from the National
Institute for Biological Standards and Control, catalogue-No.
EVA607, http://www.nibsc.ac.uk/catalog/ aids-reagent (recom-
binant HIV-1 IIIB GP120, 120 kDa. purity >95%, expression
system: baculovirus). The KD value of the gp120–CD4 interaction
is in the low nM range (depending on the method).2b,25 The activity
was calculated following eqn (1) and (2).
(1ꢀ-Naphthoxy)acetic acid. 1.3 g (6.4 mmol) (1ꢀ◦-naphthoxy)-
acetic acid (46.5%) as colorless solid, mp 194–196 C (Lit: 193–
◦
195 C); dH (400 MHz, MeOH-d4) 8.39–8.34 (1H, m, H8), 7.87–
7.82 (1H, m, H5), 7.55–7.47 (3H, m, H4/6/7), 7.43–7.37 (1H, m,
H3), 6.89–6.85 (1H, m, H2), 4.89 (2H, s, CH2); dC (400 MHz,
MeOH-d4–CDCl3 2 : 1) 175.2 (CO2H), 154.9 (C1), 134.9 (C4a),
127.7 (C5), 126.9 (C6), 125.9 (C3), 125.7 (C7), 124.8 (C8a), 122.4
(C8), 121.8 (C4), 105.4 (C2), 65.9 (CH2); m/z (%) 202 (95), 143
(100), 127 (15), 115 (58), 77 (12), 40 (18).
6-Carboxymethoxy-4-methylcoumarin. 0.5 g (2.1 mmol) 6-
carboxymethoxy-4-methylcoumarin (18.7%) as colorless solid, mp
104–106 ◦C; dH (400 MHz, MeOH-d4) 7.38–7.26 (3H, m, H5/7/8),
6.39 (1H, d, 4JH3–CH3 1, H3), 4.80 (2H, s, CH2), 2.51 (3H, d, 4JH3–CH3
1, CH3); dC (400 MHz, MeOH-d4) 121.3 (C5), 119.3 (C7), 116.2
(C8), 110.8 (C3), 19.1 (CH3); m/z (%) 234 (100), 175 (50), 147 (69),
91 (18).
(8-Quinolyloxy)acetic acid. 0.7 g (3.4 mmol) (8-quinolyloxy)-
acetic acid (24.6%) as yellow solid, mp 168 ◦C; dH (400 MHz,
MeOH-d4) 9.27 (1H, m, H2), 9.24 (1H, m, H4), 8.21 (1H, m, H3),
8.02–7.92 (2H, m, H5 + H6), 7.73 (1H, m, H7); m/z (%) 203 (3),
158 (100), 145 (8), 129 (55), 102 (15), 89 (10).
RU[gp120] maxmeasured
activity (CD4) =
(1)
RU[gp120] maxcalculated
active CD4 [fmol] = activity (CD4) × immobilized CD4
[fmol]
(2)
This journal is
The Royal Society of Chemistry 2006
Org. Biomol. Chem., 2006, 4, 3259–3267 | 3265
©

View Article Online
We used a dilution series with concentrations of the pep-
tidomimetics of 4, 6, 8, 10, 12, 16, 20, 25, 30, 50, 100, 500 and
1000 lM in HBS-EP (10 mM HEPES, 150 mM NaCl, 3 mM
EDTA, 0.005% surfactant P20, pH 7.4) buffer. Contact time was
120 s with a flow rate of 15 lL min−1 and a dissociation time of 60 s.
Every concentration was measured twice starting from the lowest
concentration. The surface was regenerated with two injections of
100 mM H3PO4 for 32 s with the same flow rate. The dependence
of the response units (RU; 1 RU ∼ 1 pg) on the concentration can
be used to obtain KD values using eqn (3):
frequency of the presaturation pulses alternated each scan.
All STD experiments were recorded with WATERGATE water
suppression. For determination of the relative STD% and the
STD amplification factor, a reference experiment was recorded
with the same conditions, processed and phased with the same
parameters. The integrals were compared in the dual display mode
in XWINNMR (version 3.1, Bruker). The temperature of the
samples was 285 K.
Conclusions
[c] RUmax
KD + [c]
RU ([c]) =
(3)
We could improve the binding affinity of the CD4 binding pep-
tidomimetics and gain insight in structure activity relationships.
Even though the calculated binding energies by Flexidock rep-
resent the experimental binding affinity only semi quantitatively,
theoretical and analytical data show a positive correlation. Overall
we could show that this protocol for a rational drug design
yielded 7 out of 11 compounds with an improved binding constant
compared to the lead.
The RUs for each ligand and concentration were determined
by subtracting the RUs of pure buffer solutions (blanks) from the
individually observed RUs. The peptidomimetics did not show
an asymptotic behavior at high concentrations in the Biacore
assay. This may be due to unspecific binding of the ligands to
the protein surface or because of a second binding site with a
lower binding affinity. Therefore, only data points representing a
typical association curve were used for data analysis (cf. Table 3).
Acknowledgements
Experimental details of STD NMR experiments
This work was supported by grants from the Deutsche
Forschungsgemeinschaft (DFG) through Graduiertenkolleg 464
and SFB 470/B2. We acknowledge that the AIDS reagent program
of the WHO provided us with generous gifts of the CD4 and the
gp120 proteins.
Preparation of the samples. Commercially available sCD4 (see
above) (200 lg, 4.4 nmol) was dissolved in buffer (200 lL D2O,
containing 8 mM Na2HPO4, 2 mM NaH2PO4, 140 mM NaCl,
3 mM KCl and 6 mM NaN3, adjusted to pH = 7–7.5 (not
corrected) with 0.1 M DCl). Through repeated dilution and
membrane centrifugation, the sample was rebuffered in deuterated
PBS, resulting in 250–300 lL solution (exactly measured by
weighing). The protein solution was split into two halves and
5 lL 1 mM DSS solution in D2O was added to each portion. The
samples were filled to 250 lL each, one with buffer and the other
one with buffer containing XI to produce the highest concentration
of 80 lM. The samples were transferred into Shigemi NMR tubes
and measured on a Bruker Avance 700 MHz spectrometer with a
cryo probe. The other concentrations of the ligand were achieved
by diluting the ligand solution with the buffered protein solution
from the other sample and vice versa. The exact concentrations of
the ligand were determined by comparison with the internal DSS
standard.
References
1 A. G. Dalgleish, P. C. Beverly, P. R. Clapham, D. H. Crawford, M. F.
Greaves and R. A. Weiss, Nature, 1984, 312, 763–767; D. Klatzmann,
E. Champagne, S. Chemaret, J. Gruest, D. Guetard, T. Hercend, J. C.
Gluckman and L. Montagnier, Nature, 1984, 312, 767–768.
2 (a) Q. J. Sattentau, J. P. Moore, F. Vignaux, F. Traincard and P. Poignard,
J. Virol., 1993, 67, 7383–7393; (b) P. D. Kwong, R. Wyatt, J. Robinson,
R. W. Sweet, J. Sodroski and W. A. Hendrickson, Nature, 1998, 393,
648–659.
3 A. Bjorndal, H. Deng, M. Jansson, J. R. Fiore, C. Colognesi, A.
Karlsson, J. Albert, G. Scarlatti, D. R. Littman and E. M. Fenyo,
J. Virol., 1997, 71, 7478–7487.
4 S. Bar and M. Alizon, J. Virol., 2004, 78, 811–820.
5 K. E. Squires, Antiviral Ther., 2001, 6(Suppl. 3), 1–14; S. Ren and E. J.
Lien, Prog. Drug Res., 1998, 51, 1–31; M. Bowers, Bull. Exp. Treat.
AIDS, 1996, Jun., 19–22.
The on resonance frequency was determined by measuring the
ligand sample only. At −1.5 ppm, all discoverable artefacts were
ꢁ1 rel. STD%. The number of scans of the reference spectra and
the STD spectra was adjusted to the concentration and was up to
11k for the STD spectrum at the lowest ligand concentration of
16 lM. The off resonance frequency was at 28.6 ppm.
6 B. Autran, G. Carcelain, T. S. Li, C. Blanc, D. Mathez, R. Tubiana, C.
Katlama, P. Debre and J. Leibowitch, Science, 1997, 277, 112–116.
7 R. J. Pomerantz, Curr. Opin. Investig. Drugs (Thomson Sci.), 2002, 3,
1133–1137.
8 C. Tamalet, J. Fantini, C. Tourres and N. Yahi, AIDS (London), 2003,
17, 2383–2388.
9 J. H. Condra, M. D. Miller, D. J. Hazuda and E. A. Emini, Annu. Rev.
Med., 2002, 53, 541–555.
Data analysis. The 1D 1H NMR spectra were acquired at
a spectral width of 10 ppm and with 32k data points. Before
Fourier transformation, these were filled up to 64k with zeros.
Through multiplication of the FID with an exponential function
(line broadening: 5 Hz), an increase of the signal noise ratio was
achieved, and subsequently the phase was corrected. A T1q filter
of 15 ms with an attenuation of 12 dB eliminated nearly all protein
signals. The phase cycle of the STD experiments was selected, such
that the subtraction needed for the difference spectra occurred
on alternating scans. In this way, artifacts due to temperature or
magnetic inhomogeneities are minimized. On- and off-resonance
10 M. P. D’Souza, J. S. Cairns and S. F. Plaeger, JAMA, J. Am. Med.
Assoc., 2000, 284, 215–222; J. A. Turpin, Expert Rev. Anti-infect. Ther.,
2003, 1, 97–128.
11 I. G. Williams, Int. J. Clin. Pract., 2003, 57, 890–897.
12 (a) J. M. Jacobson, I. Lowy, C. V. Fletcher, T. J. O’Neill, D. N. Tran,
T. J. Ketas, A. Trkola, M. E. Klotman, P. J. Maddon, W. C. Olson and
R. J. Israel, J. Infect. Dis., 2000, 182, 326–329; (b) C. W. Hendrix, C.
Flexner, R. T. MacFarland, C. Giandomenico, E. J. Fuchs, E. Redpath,
G. Bridger and G. W. Henson, Antimicrob. Agents Chemother., 2000,
44, 1667–1673.
13 (a) C. Boussard, V. E. Doyle, N. Mahmood, T. Klimkait, M. Pritchard
and I. H. Gilbert, Eur. J. Med. Chem., 2002, 37, 883–890; (b) C. Monnet,
D. Laune, J. Laroche-Traineau, M. Biard-Piechaczyk, L. Briant, C.
Bes, M. Pugniere, J.-C. Mani, B. Pau, M. Cerutti, G. Devauchelle,
3266 | Org. Biomol. Chem., 2006, 4, 3259–3267
This journal is
The Royal Society of Chemistry 2006
©

View Article Online
C. Devaux, C. Granier and T. Chardes, J. Biol. Chem., 1999, 274,
3789–3796.
19 J. Wu¨lfken, PhD Thesis, Hamburg, 2001.
20 H. M. Mo¨ller, PhD Thesis, Hamburg, 2003.
14 (a) S. Ramurthy, M. S. Lee, H. Nakanishi, R. Shen and M. Kahn,
Bioorg. Med. Chem., 1994, 2, 1007–1013; (b) G. P. Allaway, K. L. Davis-
Bruno, G. A. Beaudry, E. B. Garcia, E. L. Wong, A. M. Ryder, K. W.
Hasel, M. C. Gauduin, R. A. Koup and J. S. McDougal, AIDS Res.
Hum. Retroviruses, 1995, 11, 533–539.
15 J. D. Lifson and E. G. Engleman, Immunol. Rev., 1989, 109, 93–117.
16 J. H. Wang, R. Meijers, Y. Xiong, J. H. Liu, T. Sakihama, R. Zhang,
A. Joachimiak and E. L. Reinherz, Proc. Natl. Acad. Sci. USA, 2001,
98, 10799–10804.
17 D. G. Myszka, R. W. Sweet, P. Hensley, M. Brigham-Burke, P. D.
Kwong, W. A. Hendrickson, R. Wyatt, J. Sodroski and M. L. Doyle,
Proc. Natl. Acad. Sci. USA, 2000, 97, 9026–9031.
18 A. T. Neffe and B. Meyer, Angew. Chem., Int. Ed., 2004, 43, 2937–
2940.
21 W. Rasshofer, W. M. Mu¨ller and F. Vo¨gtle, Chem. Ber., 1979, 112,
2095–2119.
22 M. E. Duggan and J. S. Imagire, Synthesis, 1989, 131–132.
23 (a) M. Mayer and B. Meyer, Angew. Chem., Int. Ed., 1999, 38, 1784–
1788; (b) M. Mayer and B. Meyer, J. Am. Chem. Soc., 2001, 123,
6108–6117; (c) R. Meinecke and B. Meyer, J. Med. Chem., 2001, 44,
3059–3065; (d) B. Meyer and T. Peters, Angew. Chem., Int. Ed., 2003,
42, 864–890.
24 H. Wu, D. G. Myszka, S. W. Tendian, C. G. Brouilette, R. W. Sweet,
I. M. Chaiken and W. A. Hendrickson, Proc. Natl. Acad. Sci. USA,
1996, 93, 15030–15035.
25 S. M. Schnittman, H. C. Lane, J. Roth, A. Burrows, T. M. Folks, J. H.
Kehrl, S. Koenig, P. Berman and A. S. Fauci, J. Immunol., 1988, 141,
4181–4186.
This journal is
The Royal Society of Chemistry 2006
Org. Biomol. Chem., 2006, 4, 3259–3267 | 3267
©