Structure-based design of substituted biphenyl ethylene ethers as ligands binding in the hydrophobic pocket of gp41 and blocking the helical bundle formation

Article abstract of DOI:10.1016/j.bmcl.2009.08.018

A series of substituted biphenyl ethylene ether compounds has been designed to target the gp41 N-trimer in order to inhibit formation of the six-helical bundle that represents the end state of gp41-mediated viral fusion. A size exclusion HPLC based helica

Full text of DOI:10.1016/j.bmcl.2009.08.018

Bioorganic & Medicinal Chemistry Letters 19 (2009) 5693–5697
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Structure-based design of substituted biphenyl ethylene ethers as ligands
binding in the hydrophobic pocket of gp41 and blocking the helical
bundle formation
*
Bin Liu , Rhoda W. Joseph, Bruce D. Dorsey, Robert A. Schiksnis, Katrina Northrop, Marina Bukhtiyarova,
Eric B. Springman
Locus Pharmaceuticals, Inc., Four Valley Square, 512 Township Line Road, Blue Bell, PA 19422, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
A series of substituted biphenyl ethylene ether compounds has been designed to target the gp41 N-trimer
in order to inhibit formation of the six-helical bundle that represents the end state of gp41-mediated viral
fusion. A size exclusion HPLC based helical bundle formation (HBF) assay was developed to evaluate
in vitro inhibitory affinity of the inhibitors. The most potent compound 1 had an IC50 of 31 lM. The bind-
ing of compound 1 to the proposed hydrophobic pocket of gp41 was further validated by site-directed
Received 15 June 2009
Revised 3 August 2009
Accepted 4 August 2009
Available online 8 August 2009
peptide mutagenesis experiments.
Keywords:
HIV
Ó 2009 Elsevier Ltd. All rights reserved.
gp41
Inhibitor
The process by which HIV attaches to and enters cells provides
several potential targets for drug discovery.1 HIV viral fusion is ini-
tiated by interaction between the homotrimeric viral glycoprotein
gp120 and the CD4+ receptor of the host cell. This interaction is fol-
lowed by a conformational change in gp120 and then an additional
interaction with a cell surface chemokine coreceptor (CXCR4 or
CCR5). These events result in the exposure of a transmembrane
glycoprotein, namely gp41, which inserts its fusion peptide into
the cell membrane and generates a pre-hairpin intermediate. It is
then followed by the collapse of the HR-2 portion of gp41 onto
the HR-1 region to form a coiled-coil helical bundle of hairpins,
which brings the virus and host cell membrane into fusion proxim-
ity.2 Based on this model, inhibition of the gp41 six-helix bundle
formation process, the final step of HIV-1 membrane fusion, will
stop the viral entry. Enfuvirtide (Fuzeon), an FDA approved peptide
therapeutic that has proven effective in reducing the viral load of
patients, is an example of this newly validated treatment modality.
Several other peptides have been reported to inhibit viral fusion in
a similar way.3
Trp631, and Ile635, plays a crucial role in the stability of the hex-
amer (Fig. 1b). It has been proposed that small molecules which
bind with high affinity into this pocket could inhibit six-helix bun-
dle formation and, therefore, exhibit an antiviral effect.
Two lines of structural evidence support this fusion inhibition
mechanism. HIV-1 neutralization by a gp41 fusion intermediate-
directed antibody demonstrated that both binding and neutraliza-
tion events depend on residues in the antibody loop region that
protrude into the conserved gp41 hydrophobic pocket, as well as
a large pocket in the region surrounding core gp41 residues.5 Acy-
clic peptides, such as C34 and Enfuvirtide, and cyclic peptides, such
as 2 K-PIE1,6 which specifically bind to the conserved pocket have
been shown to block HIV-1 entry. Extensive efforts have been
made to identify small molecule compounds as gp41 inhibitors.7,8
However, there remain significant barriers to further optimization
and development of these initial lead compounds in order to over-
come low potency and toxicity issues.
Using Locus’ computational fragment-based discovery technol-
ogy,9 we applied a de novo ligand design process starting from the
gp41 N36/C34 six-helix bundle crystal structure.4 Briefly, C34-pep-
tides were computationally removed from the crystal structure
and binding of privileged fragments to the N36-peptide trimer core
protein was calculated and rank-ordered. A high affinity binding
site that centers around Lys574 (residue in green, Fig. 2) was iden-
tified from the computational fragment data. The fragment cluster
wraps over the top of the lysine and down into a deep hydrophobic
groove. It then continues through a narrow gap toward Arg579
(residue in red, Fig. 2).
The X-ray crystal structure of the post-fusion gp41 six-helix
bundle has been obtained using two peptides, N36 (a fragment of
HR-1) and C34 (a fragment of HR-2).4 In this structure, the inner
core parallel N-trimer is surrounded by three antiparallel C-pep-
tides (Fig. 1a). Evidence suggests that the cavity formed by the in-
ner core trimer, which is occupied by C-peptide residues Trp628,
* Corresponding author.
E-mail address: bliu@locuspharma.com (B. Liu).
0960-894X/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2009.08.018

5694
B. Liu et al. / Bioorg. Med. Chem. Lett. 19 (2009) 5693–5697
Figure 1. (a) The structure of the six-helix bundle. (b) The conserved hydrophobic binding pocket on the inner core trimer. The amino acid residues from C-peptides that
occupy the pocket are also shown.
N
HN
N
N
O
COOCH3
Figure 3. Structure of compound 1.
2-(4-bromonaphthalen-1-yl)-ethanol and the hydroxyl group was
protected with TBDMS to give intermediate 2. Pd-mediated car-
boxylation followed by deprotection afforded compound 3. A sub-
sequent Mitsunobu reaction generated intermediate 4. Suzuki
coupling between 4 and 3-cyanophenylboronic acid afforded com-
pound 5, which was then treated with azidotributylstannane to
give the targeted molecule 1.
In addition to naphthalene compounds, indole analogs of com-
pound 1 proposed by the simulation were also synthesized
(Scheme 2). To this end, 4-bromoindole-1-carboxylic acid tert-bu-
tyl ester 6 was converted to tert-butyl 4-(2-hydroxyethyl)-1H-in-
Figure 2. Predicted fragment cluster map (blue) on the N36-peptide trimer core of
HIV-1 gp41 (purple) with key residues Lys574 (green) and Arg579 (red).
dole-1-carboxylate 7 via a Stille coupling and hydroboration
sequence. As for the synthesis of 1, Mitsunobu reaction followed
by Suzuki coupling gave compound 8, which was subsequently
converted to tetrazole 9. Removal of the Boc protecting group
and subsequent treatment with toluene sulfonyl chloride yielded
compound 10.
In the binding area near Lys574, a cluster of fragments with
high affinity to the positively charged amino group has been iden-
tified. In the hydrophobic binding pocket to the right of Lys574,
fragments such as substituted benzene, biphenyl and biphenyl
ether showed good computed affinity. Continuing past the hydro-
phobic pocket, the simulation predicts a narrow groove defined by
Gln577, Trp571 and Arg579, where aromatic moieties such as
biphenyl ether, biphenyl, substituted benzene, naphthalene and in-
dole, are predicted to interact. Based on this fragment map, several
series of target molecules were designed and synthesized. Among
them, a series of small molecules represented by compound 1
(Fig. 3) demonstrated a reasonable inhibitory effect against gp41
helical bundle formation. In addition, the interaction of this series
of compounds with gp41 was further validated by a peptide muta-
genesis study. Herein, we report the synthesis, possible binding
mode, and biochemical and biophysical study of this series of com-
pounds represented by compound 1.
In order to evaluate the in vitro binding affinity of small mole-
cules to the gp41 inner core trimer and their ability to disrupt hex-
amer formation, we developed a gp41 binding assay. This assay is
based on size exclusion chromatography (SEC) separation of the
six-member bundle formed by mixing of the HR-1 derived N36-
peptide with the HR-2 derived C34-peptide. This six-helix bundle
formation mimics the function of HIV gp41 in the biological setting
which is required for the HIV virus to enter into the host cell. The
basic experimental design and typical results are illustrated in
Figure 4. In the first chromatogram (green), the N36-peptide is
mixed with C34-peptide, and the hexamer bundle is identified by
SEC. In the second chromatogram (magenta), the N36-peptide is
first mixed with a helix bundle inhibitor, such as compound 1.
After a 30 min pre-incubation period, C34-peptide is added and
the reaction mixture is applied on the HPLC size exclusion column.
The synthesis of biphenyl ether compound 1 is shown in
Scheme 1. Briefly, 1,4-dibromonaphthalene was first converted to

B. Liu et al. / Bioorg. Med. Chem. Lett. 19 (2009) 5693–5697
5695
N
HN
N
CN
OTBDMS
OH
O
Br
O
N
O
Br
Br
a,b
c,d
g
e
f
Br
COOCH3
COOCH3
COOCH3
COOCH3
5
2
3
4
1
Scheme 1. Reagents and conditions: (a) BuLi, hexane/ether, À78 °C, then Oxirane, THF, 0 °C; (b) TBDMSCl, imidazole, THF; (c) Pd(OAc)2, dppp, CO, DMSO/MeOH, 80 °C; (d)
TBAF, THF, rt; (e) 3-bromophenol, DIAD, PPh3, THF, rt; (f) 3-cyanophenylboronic acid, Pd(PPh3)4, Na2CO3, DME, 80 °C; (g) nBu3SnN3, 80 °C.
N
HN
N
HN
N
N
CN
N
O
O
OH
N
O
Br
f, g
e
a, b
c, d
N
S
N
Boc
N
Boc
N
Boc
O
O
N
Boc
9
7
6
8
10
Scheme 2. Reagents and conditions: (a) tributylvinylstannane, Pd(PPh3)4, DMF, 85 °C; (b) 9-BBN, THF, rt, then H2O2, NaOH, À20 °C; (c) 3-bromophenol, DIAD, PPh3, THF, rt;
(d) 3-cyanophenylboronic acid, Pd(PPh3)4, Na2CO3, DME, 80 °C; (e) nBu3SnN3, 80 °C; (f) TBAF, THF, reflux; (g) TosCl, TEA, DCM.
Figure 4. An overlay of three chromatograms illustrating the hexamer bundle formation (HBF) assay by SEC–HPLC (see text for description).
The peak size of the hexamer bundle is diminished presumably due
to the binding of compound 1 to inhibit the formation of helical
bundle. In the control experiment without N36- or C34-peptides,
as shown in the last chromatogram (blue), no hexamer peak at
the corresponding retention time is observed. The hexamer bundle
formation (HBF) assay was cross-validated by comparing the
results with those of an ELISA method reported by Jiang and co-
workers.10 In the ELISA assay, an antibody that recognizes the in-
tact six-helical bundle was developed. A porphyrin compound
(mesotetra (4-carboxyphenyl) porphine, MTCPP) that is known to
prevent helical bundle formation was reported to have an IC50 of
Unlike published circular dichroism (CD) methods, this assay
method is compatible with dimethyl sulfoxide (DMSO) solvent that
is used to solubilize most small organic compounds for drug dis-
covery. The major limitation is the narrow range of useful concen-
tration of the N36-peptide (1–6
N36 does not efficiently form a six-helix bundle with C34, even
when C34 is in excess. Above 6 M, the N36-peptide begins to
lM). We found that below 1 lM,
l
aggregate and six-helix bundle formation is maximally saturated,
that is, addition of more N36 and C34 does not result in more hex-
amer bundle formation. The assay range could be extended by cre-
ation of N36-peptide derivatives with better physical properties,
such as increased trimer stability and solubility. A subsequent
dose–response experiment was used to determine IC50’s, the
in vitro binding of compounds 1, 9 and 10 were tested for their
ability to disrupt the hexamer bundle and the results are listed
ca. 1
the same compound disrupts gp41 bundle hexamer formation with
l
M in this ELISA sandwich assay.10 In our HPLC–SEC assay,
an IC50 of 0.9
assays.
lM and hence serves as a positive control in all of our

5696
B. Liu et al. / Bioorg. Med. Chem. Lett. 19 (2009) 5693–5697
Table 1
Inhibition of listed compounds in SEC–HBF assay
Compound
% Inhibition screening at 50
l
M
IC50, lM
1
9
10
95
100
100
31.0
na
32.9
in Table 1. Compounds 1, 9 and 10 all showed ca. 100% inhibition at
the screening concentration of 50 M. The IC50’s of compounds 1
and 10 were estimated to be 31 M and 32.9 M, respectively.
l
l
l
Computationally predicted interactions of compound 1 with the
inner trimer core of gp41 are shown in Figure 5. In this binding
mode, the tetrazole moiety interacts with Lys574, and the biphenyl
groups occupy the neighboring hydrophobic pocket. The naphthyl-
ene ester projects into the narrow pocket defined by Trp571 and
Arg579. Ideally,
a co-crystal structure of compound 1 with
gp41 N-trimer constructs would validate the predicted binding.
However, due to the nature of the target protein, its binding pock-
et, and compound properties, attempts to obtain an X-ray co-crys-
tal structure of compound 1 with several gp41 N-trimer constructs
were unsuccessful.11
Figure 6. Position of gp41 amino acid residues Trp571, Lys574 and Arg579 (green)
in the mutation study.
Despite the inhibition of the helical bundle formation, com-
In lieu of a co-crystal structure, we conducted site-directed
mutagenesis studies to assess gp41 interactions with compound
1 and gain some insight into the mode of binding. In this study,
each of the three amino acid residues that were proposed to
interact with compound 1 in N36 were mutated to Ala one at
a time (Fig. 6). The effect of the mutation on hexamer formation
and on the inhibition potency of compound 1 was then measured
in the HBF assay. Consistent with the predicted binding mode of
the biphenyl compounds, the results demonstrated that interac-
tion of biphenyl compound 1 with N-trimer is highly dependent
on gp41 residues Trp571 and Arg579. Although Trp571Ala and
Arg579Ala mutations did not affect the formation of the hexa-
meric bundle, the mutation of these residues to Ala reduces com-
pound 1 inhibitory activity by 20% and 60%, respectively. Another
important residue is Lys574, which is predicted to provide a key
interaction point with the tetrazole moiety of compound 1. How-
ever, we found that in Lys574Ala mutant, N36 was incapable of
forming a hexameric bundle with C34, so it was not possible
to measure the effect of this mutation on inhibitor binding.
Additionally, the binding of compound 1 to the gp41 hydropho-
bic pocket was further evaluated in a fluorescence assay developed
to quantify small molecule binding in the hydrophobic pocket.12
Compound 1 showed significant binding to the designed pocket
in this assay.13
pound 1 did not show cell fusion inhibition activity up to 70 lM
in cell–cell fusion inhibitory assay.14 The disconnection between
the inhibition of helical bundle formation observed for this series
of compounds and the in-activity in cell–cell fusion and cytopro-
tection assays is not clearly understood at this time, but it may
be due to the poor physical properties of the synthetic ligand.
In summary, we have applied de novo design to synthesize a
series of substituted biphenyl ethylene ether compounds targeting
the gp41 inner core trimer in order to disrupt the formation of the
helical bundle that facilitates viral fusion. The compounds showed
reasonable binding affinity to the gp41 inner core and prevented
the helical bundle formation. The interaction between the small
molecules and gp41 was also validated by targeted mutagenesis
of the gp41 peptide. The current study provides an important
example for structure-based drug design for protein–protein inter-
action targets.
Acknowledgments
This work was supported, in part, by grant funding from Inter-
national Partnership for Microbicides (IPM) and USAID, and by SBIR
Grant No. NIH2088. The authors acknowledge the synthetic sup-
port from Solvias AG for providing the intermediates 5 and 8. We
thank Professor Miriam Gochin for conducting the fluorescence
screening assay in which compound 1 was evaluated. We also
thank Dr. Enrique Michelotti and Dr. Martha Kelly for their helpful
suggestions on the Letter.
References and notes
1. For recent reviews, see: (a) Finnegan, C.; Blumenthal, R. Infect. Disord.: Drug
Targets 2006, 6, 355; (b) Sterjovski, J.; Churchill, M. J.; Wesselingh, S. L.; Gorry,
P. R. Curr. HIV Res. 2006, 4, 387.
2. Chan, D. C.; Kim, P. S. Cell 1998, 93, 681.
3. For recent reviews, see: (a) Liu, S.; Wu, S.; Jiang, S. Curr. Phar. Des. 2007, 13, 143;
(b) Debnath, A. K. Exp. Opin. Invest. Drugs 2006, 15, 465.
4. Chan, D. C.; Fass, D.; Berger, J. M.; Kim, P. S. Cell 1997, 89, 263.
5. Luftig, M. A.; Mattu, M.; Di Giovine, P.; Geleziunas, R.; Hrin, R.; Barbato, G.;
Bianchi, E.; Miller, M. D.; Pessi, A.; Carfi, A. Nat. Struct. Mol. Biol. 2006, 13, 740.
6. Welch, B. D.; VanDemark, A. P.; Heroux, A.; Hill, C. P.; Kay, M. S. Proc. Natl. Acad.
Sci. U.S.A. 2007, 104, 16828.
7. Frey, G.; Rits-Volloch, S.; Zhang, X. Q.; Schooley, R. T.; Chen, B.; Harrison, S. C.
Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 13938.
8. Liu, K.; Lu, H.; Hou, L.; Qi, Z.; Teixeira, C.; Barbault, F.; Fan, B.; Liu, S.; Jiang, S.;
Xie, L. J. Med. Chem. 2008, 51, 7843.
9. Moore, W. R., Jr. Curr. Opin. Drug Disc. Dev. 2005, 8, 355.
10. Jiang, S.; Lin, K.; Zhang, L.; Debnath, A. J. Virol. Method 1999, 80, 85.
11. Unpublished results.
Figure 5. Predicted binding of biphenyl compound 1 to gp41 N36 trimer surface.

B. Liu et al. / Bioorg. Med. Chem. Lett. 19 (2009) 5693–5697
12. Cai, L.; Gochin, M. Antimicrob. Agents Chemother. 2007, 51, 2388. to the response produced by 25
5697
lM control peptide C18Aib (Kd = 1
l
M) and
13. This experiment was conducted by the Miriam Gochin lab using that lab’s
previously reported FRET-based assay method, which measures the ability of a
67% response relative to 25
of compound 1 was not determined in this assay. For assay details and control
compounds, see Ref. 12.
l
M control compound ADS-J1 (Kd = 0.4 lM). The Kd
test compound to disrupt six-helix bundle formation. When tested in
a
screening experiment at 25 M, compound 1 produced 75% response relative
l
14. Cell–cell fusion test was conducted at ImQuest, Inc.