Inhibition of HIV-1 envelope-mediated fusion by synthetic batzelladine analogues

Article abstract of DOI:10.1021/np049958o

Marine natural products that feature polycyclic guanidine motifs, such as crambescidins and batzelladines, are known to have antiviral activities toward some viruses including HSV and HIV. In this study we evaluated a synthetic library containing 28 batze

Full text of DOI:10.1021/np049958o

J . Nat. Prod. 2004, 67, 1319-1324
1319
In h ibition of HIV-1 En velop e-Med ia ted F u sion by Syn th etic Ba tzella d in e
An a logu es†
Carole A. Bewley,*,‡ Satyajit Ray,‡ Frederick Cohen,§ Shawn K. Collins,§ and Larry E. Overman*,§
Laboratory of Bioorganic Chemistry, National Institute of Diabetes and Digestive and Kidney Diseases,
National Institutes of Health, DHHS, Bethesda, Maryland 20892, and Department of Chemistry, 516 Rowland Hall,
University of California, Irvine, California 92697-2025
Received J anuary 22, 2004
Marine natural products that feature polycyclic guanidine motifs, such as crambescidins and batzelladines,
are known to have antiviral activities toward some viruses including HSV and HIV. In this study we
evaluated a synthetic library containing 28 batzelladine analogues, the structures of which encompass
and surpass variations seen in natural batzelladines, for their ability to inhibit HIV-1 envelope-mediated
cell-cell fusion. Clear structure-activity relationships were revealed and indicated that the best inhibitors
of fusion were most similar in structure to natural batzelladine F, with IC50 values ranging from 0.8 to
3.0 µM. Proceeding from the earlier finding that some batzelladines block gp120-CD4 binding, modeling
studies of inhibitors binding to the CD4 binding site on gp120 were carried out. The lowest energy models
suggest a preferred orientation for inhibitor binding that is consistent with the observed structure-
activity relationships.
A number of marine invertebrates, especially sponges
belonging to the genera Batzella and Crambe, are known
to produce complex secondary metabolites that contain one
or more polycyclic guanidine units. Instructive examples
include members of the crambescidin family (such as 1 and
2) that feature a pentacyclic guanidine core coupled to alkyl
spermidine moieties,1-5 and the batzelladine family6,7
(examples include 5-10), members of which generically
comprise mixed bicyclic and/or tricyclic guanidine units,
with the bisguanidines tethered by an alkyl ester unit
(Figure 1). Fervent interest in this group of compounds
arises from the combination of notable biological activities
associated with these compounds, as well as the synthetic
challenges posed by these complex structures.
The biological activities associated with the crambesci-
dins are varied, and reports have included cytotoxicity
toward the cancer cell lines P388,1 L1210,2,3 and HCT-16,4
antifungal activity toward Candida albicans,1 and antiviral
activities toward Herpes simplex virus type 1 (HSV-1)2,3
and human immunodeficiency viruses (HIV).6,8 Batzella-
dines A-E, on the other hand, were originally discovered
in an enzyme-linked immunosorbent assay (ELISA) for
their ability to block interactions between the surface
envelope glycoprotein of HIV, namely gp120, and the
extracellular domains of its primary receptor CD4.6 Later,
the structures of batzelladines F-I were determined fol-
lowing the observation that crude extracts of the Batzella
sp. were able to induce dissociation of the complex between
the protein tyrosine kinase p56lck and CD4.7 It is worth
emphasizing that although both of these assays in which
the batzelladines proved active involve CD4, the regions
on CD4 with which gp120 and p56lck interact are distinct,
with gp120 binding to the extracellular regions of CD4 and
p56lck binding to the intracellular cytoplasmic tail of CD4.
As part of an effort to discover natural products that can
block protein-protein interactions comprising large sur-
faces, such as those involved in HIV-1 entry into its host
cell, we recently described new analogues of the crambes-
cidin family, crambescidin 826 (2) and dehydrocrambine
A (3).5 Demonstrations that these alkaloids inhibit HIV-1
envelope-mediated cell fusion (or HIV entry) in a time-
dependent manner and exert their effect through interac-
tions with HIV-1 envelope (Env) rather than cellular
receptors suggested that, like the structurally related
batzelladines, crambescidins may be able to block CD4-
gp120 interactions. To further evaluate the specific struc-
tural requirements for guanidine alkaloids as HIV-1 fusion
inhibitors, we describe here the results of testing a series
of 28 synthetic batzelladine analogues (11-38) in an HIV-1
cell fusion assay. Clear structure-activity relationships
were revealed suggesting a model for polycyclic guanidine
alkaloids binding to the HIV-1 surface envelope glycopro-
tein gp120. These results together with modeling studies
are presented.
Resu lts a n d Discu ssion
Str u ctu r e-Activity Rela tion sh ip s of Ba tzella d in e
An a logu es a s F u sion In h ibitor s. Common to all of the
batzelladines is the presence of a tricyclic guanidine unit.
Internally, these ring systems vary in the number and
positioning of double bonds within the pyrimidine rings,
as well as in the stereochemistry of the angular hydrogens
flanking the pyrrolidine nitrogen (Figure 1). In addition
to variations within the tricyclic unit itself, the natural
products differ in location and configuration of varied alkyl
substituents. Following the discovery of batzelladines and
crambescidins, several synthetic approaches to build-
ing these polycyclic guanidine ring systems have been
described.9-13 As a result of work from the Overman group,
a library of batzelladine analogues that encompasses and
surpasses variations seen in the natural products14,15 has
been synthesized and is illustrated in Figure 2. The
assemblage of structures comprising this library allowed
us to evaluate the contribution toward inhibitory activity
by several different structural features. These include the
type (bicyclic or tricyclic) and number of guanidine ring
systems (one or two) present, the absolute configuration
of these units, and the position and number of alkyl
† Dedicated to the late Dr. D. J ohn Faulkner (Scripps) and the late Dr.
Paul J . Scheuer (Hawaii) for their pioneering work on bioactive marine
natural products.
* To whom correspondence should be addressed. (C.A.B.) Tel: 301-594-
5187. Fax: 301-402-0008. E-mail: caroleb@mail.nih.gov. (L.E.O.) Tel: 949-
824-6793. Fax: 949-824-3866. E-mail: leoverma@uci.edu.
‡ National Institute of Diabetes and Digestive and Kidney Diseases.
§ University of California.
10.1021/np049958o CCC: $27.50
© 2004 American Chemical Society and American Society of Pharmacognosy
Published on Web 07/08/2004

1320 J ournal of Natural Products, 2004, Vol. 67, No. 8
Bewley et al.
F igu r e 1. Examples of crambescidins and batzelladines.
substituents decorating the guanidine core. For those
compounds containing two guanidine units, the rigidity of
the linker tethering the two ring systems also could be
evaluated.
comprise a core similar to that of batzelladine D (8) and
unlike 13-16 are substituted by carboxyl groups in addi-
tion to the C-1′ alkyl chain. Analogues 22-24 contain one
bicyclic and one tricyclic guanidine unit, with the bicyclic
ring system connected by an alkyl ester group off C-1′ of
the tricyclic motif, and compounds 25-30, which closely
resemble natural batzelladine F,15 are alike in that they
contain two tricyclic guanidine motifs where both ring
systems bear an alkyl group at C-1′. Seen in the right-most
column of Figure 2, the bistricyclic compounds 31-38 are
distinguished by the presence of a C7 or C9 alkyl chain at
C-1 rather than the C-1 methyl group present in all other
tricyclic guanidine analogues. Further variations among
compounds 31-38 not seen in the natural products include
the presence of biscarboxy esters in 31-33 and 36-38 and
a 1,4-benzenedimethanol linker connecting the two tricyclic
units in 37 and 38.
The fusion data in Figure 3 indicate that these com-
pounds generally fall within three activity levels, those
inhibiting greater than 90%, those inhibiting greater than
approximately 40%, and those inhibiting less than ap-
proximately 20%, at 5 µM concentrations. In terms of
structure-activity relationships, all compounds with C-1
alkyl substituents larger than methyl (20, 31-38) were
inactive, indicating that only a small substituent at this
position is tolerated. Notably, in the fusion assays employ-
ing 100 µM inhibitor, compounds 37 and 38 were also
inactive, indicating that the added rigidity is detrimental
to binding to gp120. Structures of compounds that were
modest inhibitors of fusion were varied and included mono-
Each of these compounds was tested in an HIV-1 Env-
mediated cell-cell fusion assay for their ability to inhibit
fusion.16 Compounds were tested against both X4 and R5
HIV-1 strains and included LAV and SF162, respectively.
Initially, compounds were tested at single concentrations
of 100 and 5 µM toward both viral strains. Data generated
from the assays carried out in the presence of 100 µM
concentrations yielded little to no structural information
because most of the compounds tested inhibited fusion
greater than 80% (data not shown). However, in the
presence of 5 µM inhibitor, clear structure-activity rela-
tionships were evident, as the percent fusion observed
ranged from 0 to 100% at these concentrations (Figure 3).
After these initial assays, dose-response curves were
measured for selected compounds to better deduce struc-
ture-activity relationships (Table 1). Representative in-
hibition curves for compounds 13, 22, 25, and 27 are shown
in Figure 4.
Analogues viewed as monomers in terms of the number
of guanidine motifs appear in the left panel of Figure 2.
Compounds 11 and 12 contain a bicyclic ring system and
terminal benzyloxy group, and compounds 13-21 feature
a single tricyclic guanidine motif. Of these tricyclics, the
â-keto esters 13, 14, and 16 bear a C-1′ alkyl side chain,
and compound 15 contains a C-2′ ester substituent in
addition to a C-1′ alkyl side chain. Compounds 17-19

Inhibition of HIV-1 Envelope-Mediated Fusion
J ournal of Natural Products, 2004, Vol. 67, No. 8 1321

1322 J ournal of Natural Products, 2004, Vol. 67, No. 8
Bewley et al.
F igu r e 3. Inhibition of HIV-1 envelope-mediated fusion for analogues
11-38 tested at 5 µM concentration.
F igu r e 4. Representative inhibition curves for HIV-1 Env-mediated
fusion. Curves for compounds 13, 22, 25, and 27 are represented by
solid black, blue, red, and green circles, respectively.
F igu r e 5. Complexes and binding orientations for gp120. (a) CD4-
gp120 complex with gp120 shown in a charged surface representation
with negative and positive regions colored red and blue, respectively;
CD4 shown in a CR worm representation, with Phe 43 and Arg 59
shown as rods with nitrogen and oxygen atoms colored blue and red,
respectively. (b) Four possible orientations in which bisguanidine
analogues can bind to gp120. Note that manual docking indicates this
type of inhibitor can best be accommodated in the CD4 binding site in
a nearly horizontal position, as viewed in this figure. Positions closer
to vertical are precluded by the amino acid residues appearing at the
top of the binding site, in the views shown here. (c) Model of the
preferred orientation (box 3 of 5b) of compound 27 binding to gp120.
The guanidine group of rings A and B are within hydrogen-bonding
distances of the carboxylate of Asp 368.
Ta ble 1. IC50 Values for Selected Analoguesa
compound
IC50 valuesb
13
14
15
19
22
24
25
27
29
30
17.0 ( 3.3
5.3 ( 0.7
4.5 ( 0.9
3.5 ( 0.6
7.7 ( 1.5
6.2 ( 1.2
1.8 ( 0.4
0.8 ( 0.2
3.0 ( 0.6
2.5 ( 0.6
Mod elin g of Ba t zella d in e An a logu es Bin d in g t o
gp 120. The initial steps leading to HIV-1 virus-cell or
cell-cell fusion include attachment of the viral surface
envelope glycoprotein gp120 to the cellular receptor CD4,
which further permits binding of gp120 to one of two classes
of co-receptors. This assembly ultimately facilitates fusion
between the virus and cell membranes.17 The first series
of batzelladines, A-E, were discovered in an ELISA format
that detects binding of the HIV-1 surface envelope (Env)
glycoprotein gp120 binding to the primary receptor CD4.6
Batzelladines A and B inhibited gp120-CD4 binding at low
micromolar concentrations, as well as inhibiting HIV
infectivity, as would be expected for a fusion inhibitor. The
crystal structure of a ternary complex comprising CD4, a
deglycosylated core construct of gp120, and a neutralizing
antibody known as 17b revealed unique features of the
CD4-gp120 interface (Figure 5a).18 In terms of gp120, the
CD4 binding site (CD4bs) comprises a substantial binding
pocket that is formed by both hydrophobic and polar amino
a
Dose-response curves were determined on analogues for
b
which sufficient quantities of compound were available. Micro-
molar.
and bisguanidines 12-16, 19, and 21-24. By far the most
active compounds among this library are compounds 25-
30, all of which contain two tricyclic guanidine motifs
connected by an alkyl ester linkage comprising eight heavy
atoms. Although compounds 25-30 can be viewed as
having the same carbon skeletons, the absolute and relative
configuration of the left-hand guanidine ring system differs.
When considered together, these results suggest that for
this class of compounds the structures of the best fusion
inhibitors are pseudo-dimers incorporating two tricyclic
guanidine motifs connected by a flexible linker with each
of the guanidine units bearing a methyl group at C-1 and
an alkyl group at C-1′. These structures are most similar
to the natural products batzelladines F and G.7,15

Inhibition of HIV-1 Envelope-Mediated Fusion
J ournal of Natural Products, 2004, Vol. 67, No. 8 1323
Ta ble 2. Energiesa of Compound 27 upon Binding to gp120
acid residues including a key aspartic acid, Asp 368. On
the CD4 side, although numerous amino acids are involved
in binding to gp120, two amino acids, namely, Phe 43 and
Arg 59, dominate the intermolecular contacts. Phe 43 is
involved in van der Waals contacts with hydrophobic
regions of the CD4bs, while the guanidine group of Arg 59
makes double hydrogen bonds to Asp 368. In an effort to
further our understanding of the structural features lend-
ing to the bioactivity of batzelladines and analogues, we
carried out modeling studies of inhibitors binding to gp120.
Mindful of the structural features of CD4-gp120 binding
and the structure-activity relationships gleaned from the
HIV-1 fusion assays, modeling was accomplished by dock-
ing and minimization of the complexes between gp120 and
several of the batzelladine analogues as described next.
orientation
VDW
torsion
dipole-dipole
total
3
4
11.3
14.3
14.6
28.6
2.3
3.0
28.2
45.9
a
kcal mol-1
.
left-hand guanidine unit of that of 27, was therefore
minimized, docked, and subjected to torsion angle dynamics
using the same protocol as that used for the model of 27.
Results similar to those obtained for compound 27 were
observed for 28, where the structures largely converged to
orientation 3. Although the opposite stereochemistry leads
to a pucker of the tricyclic ring system opposite that of 28,
the pyrrole is directed to the outside of the CD4 binding
site, and modeling suggests that the large binding pocket
(relative to the tricyclic guanidine motif) can accommodate
either shape.
We began the modeling studies with compound 27, which
is representative of the bistricyclic guanidines found to be
the best inhibitors. The structure of 27 was first minimized
and then positioned randomly about 20 Å outside of the
CD4bs (Figure 5a). Manual docking attempts demonstrated
to us that the shape and concavity of the CD4bs would
require that the inhibitor occupy a nearly horizontal
position in the view seen in Figure 5a,c. That is, the
overlying residues of the CD4bs preclude docking the
bisguanidines in a more-nearly vertical orientation. Thus,
as seen in Figure 5b, four possible modes of binding could
occur owing to the presence of two guanidine ring systems
and two possible binding orientations. In generating a set
of distance restraints that would facilitate electrostatic
interactions, we made the assumption that either of the
guanidine units residing within the two tricyclic motifs in
27 would mimic the guanidine group of the side chain of
Arg69 and participate in electrostatic interactions with
Asp368 of gp120. Thus, four sets of distance restraints that
account for the four possible orientations of the ligand were
prepared. To allow the program to sample all four binding
orientations (Figure 5b), these restraints were initially
introduced as ambiguous distance restraints, where re-
straints are satisfied as long as one of the four sets of
distances is satisfied. For compound 27, the structures
converged surprisingly well (∼80%) to orientation 3 (lower
left panel), where the guanidine group of the terminal ring
system is within hydrogen-bonding distance of the car-
boxylate of Asp368. Of the remaining 20% of the structures,
the majority converged to orientation 4. Owing to the
proximity to gp120 of the alkyl groups positioned on ring
B or D in orientations 1 and 2, respectively, it was
impossible to position the guanidine units near Asp368 of
gp120 without other portions of the ligand making bad
contacts with gp120, even among the multiple conforma-
tions observed for the alkyl chains. Additional calculations
were then carried out for 27 bound to gp120 in orientation
3 wherein conservative and ambiguous restraints were
introduced so as to allow favorable van der Waals interac-
tions between any hydrophobic region of compound 27 and
gp120. The family of structures for which no violations were
observed was subsequently averaged, and the restrained
regularized mean structure of 27 docked to gp120 is shown
in Figure 5c. From the model, it appears likely that this
conformation can facilitate hydrophobic interactions be-
tween the ligand and the V1/V2 stem loop of gp120, which
comprises predominantly hydrophobic amino acids.
Before moving on to other analogues, we sought to
further examine the alternate binding mode (orientation
4, Figure 5b) where the central (comprising rings C/D),
rather than the terminal (rings A/B), guanidine motif is
centered in the CD4bs. We therefore carried out additional
calculations again including ambiguous distance restraints
that would allow for favorable protein-ligand interactions.
Although a model for 27 binding to gp120 in this alternate
mode could be obtained (model presented in Supporting
Information), the torsion angle and van der Waals energies
of the ligand bound in this mode were approximately double
those obtained for 27 binding in orientation 3 (Table 2).
Following on the results of docking and modeling for
compounds 27 and 28, we also attempted to dock com-
pounds 33 and 37, which showed little to no activity. After
numerous attempts, however, we were unable to model
either of these compounds into the CD4bs on gp120, even
if we positioned the analogues (in each of the four possible
orientations) near the CD4bs by hand. As discussed above,
compounds 31-38 differ from 25-30 in that the methyl
group at C-1 has been replaced by an additional C7 or C9
alkyl chain. With respect to accessibility to the guanidine
nitrogens, the net result is that each guanidine motif
bears two or more bulky substituents at C-1/C-1′ and/or
C-2/C-2′, obstructing these compounds from entering deeply
enough into the binding pocket to participate in hydrogen-
bonding interactions with Asp368. Compounds 37 and 38,
both of which possess a methyl phenyl bis ester linkage,
were the worst inhibitors of this library and failed to inhibit
fusion even at 1 mM concentrations (data not shown). This
implies that the imposed rigidity of the methyl phenyl
linker, the added mass between the guanidine motifs, or
both abrogate fusion-blocking activity. We were also unsuc-
cessful with docking either of these compounds to the
CD4bs.
Con clu sion s
The crystal structures of gp120 solved to date are all
ternary complexes between the gp120 core, CD4, and anti-
gp120 antibody.18,19 The construct used for crystallization
of gp120 comprises what are believed to be the three
domains corresponding to the central core of gp120 and the
stem of the V1/V2 loop. In terms of binding affinity,
interactions between this core gp120 molecule and both
CD4 and anti-gp120 antibodies are preserved, evidence
that this construct truly represents the main structural
features of gp120. In this study we have carried out HIV-1
fusion assays and modeled batzelladine analogues binding
to the CD4bs of gp120 by docking and minimization. Clear
structure-activity relationships were revealed from the
These results led us to next examine the effect of
stereochemistry of the terminal guanidine ring system on
the model since several different arrangements are present
in compounds 25-30, all of which are good inhibitors.
Compound 28, which has the opposite configuration of the

1324 J ournal of Natural Products, 2004, Vol. 67, No. 8
Bewley et al.
dihedral angle violations (corresponding to those angles that
were fixed) greater than 5° were observed.
fusion assay data and indicated that inhibitors with low
to sub-micromolar IC50 values incorporate two tricyclic
guanidine ring systems tethered by a flexible alkyl linker
wherein C-1 of the terminal guanidine motif is substituted
by a methyl group only. Consistent with these results,
modeling studies of compounds representing the best and
worst fusion inhibitors demonstrated that a terminal
tricyclic guanidine motif, lacking substitution at C-1,
converged to one preferred orientation of binding to the
CD4bs on gp120, permitting electrostatic interactions
between the guanidine nitrogens and the carboxylate of
Asp368, as well as extensive hydrophobic interactions
between the alkyl chains and predominantly hydrophobic
regions of gp120. Compounds comprising essentially one-
half of the bisguanidines were less potent inhibitors of
fusion than the so-called dimers. These monomers can be
docked to the CD4bs in a similar manner as compounds
25-30, but lack the second alkyl unit that likely contrib-
utes favorable hydrophobic interactions. These studies
therefore provide another example whereby dimerization
of ligands enhances potency20,21 and provide a structural
framework for further development of useful HIV-1 fusion
inhibitors.
Ack n ow led gm en t. This work was supported in part by
the Intramural AIDS Targeted Antiviral Program of the Office
of the Director, National Institutes of Health (C.A.B.) and NIH
grant HL-25854 (L.E.O.).
Su p p or tin g In for m a tion Ava ila ble: The source of the batzella-
dine analogues studied and experimental procedures for preparing 17-
19, 21, and 27 together with characterization data for these com-
pounds; figure illustrating model of 27 in complex with gp120 in
orientation 4. This material is available free of charge via the Internet
at http://pubs.acs.org.
Refer en ces a n d Notes
(1) Kashman, Y.; Hirsh, S.; McConnell, O. J .; Ohtani, I.; Kusumi, T.;
Kakisawa, H. J . Am. Chem. Soc. 1989, 111, 8925-8926.
(2) J ares-Erijman, E. A.; Sakai, R.; Rinehart, K. L. J . Org. Chem. 1991,
56, 5712-5715.
(3) J ares-Erijman, E. A.; Ingrum, A. L.; Carney, J . R.; Rinehart, K. L.;
Sakai, R. J . J . Org. Chem. 1993, 58, 4805-4808.
(4) Berlinck, R. G. S.; Braekman, J . C.; Daloze, D.; Bruno, I.; Riccio, R.;
Ferri, S.; Spampinato, S.; Speroni, E. J . Nat. Prod. 1993, 56, 1007-
1015.
(5) Chang, L.-C.; Whittaker, N. F.; Bewley, C. A. J . Nat. Prod. 2003, 66,
1490-1494.
(6) Patil, A. D.; Kumar, N. V.; Kokke, W. C.; Bean, M. F.; Freyer, A. J .;
De Brosse, C.; Mai, S.; Truneh, A.; Faulkner, D. J .; Carte, B.; Breen,
A. L.; Hertzberg, R. P.; J ohnson, R. K.; Westley, J . W.; Potts, B. C.
M. J . Org. Chem. 1995, 60, 1182-1188.
Exp er im en ta l Section
(7) Patil, A. D.; Freyer, A. J .; Taylor, P. B.; Carte´, B.; Zuber, G.; J ohnson,
R. K.; Faulkner, D. J . J . Org. Chem. 1997, 62, 1814-1819.
(8) Palagiano, E.; Demarino, S.; Minale, L.; Riccio, R.; Zollo, F.; Iorizzi,
M.; Carre, J . B.; Debitus, C.; Lucarain, L.; Provost, J . Tetrahedron
1995, 51, 3675-3682.
HIV-1 En v-Med ia ted F u sion Assa ys. Assays were carried
out in duplicate on at least two separate days for both LAV
and SF162 viruses using protocols identical to those already
described.22 In the single-concentration assays, inhibitors were
added to each well to a final concentration of either 100 or 5
µM. For dose-response experiments, inhibitors were added
in increasing concentrations to each well and inhibition curves
were best fit to the equation % fusion ) 100/(1 + KA[I]) where
[I] is the concentration of inhibitor and KA is the equilibrium
association constant for inhibitor binding to gp120. Curves
were fit by nonlinear least-squares optimization using the
program Kaleidagraph 3.5 (Synergy Software, Reading, PA).
Mod elin g of In h ibitor -gp 120 Com p lexes. Three-dimen-
sional coordinates for inhibitors used in this study were
generated using the program Chem3D. Parameter and topol-
ogy files for each set of coordinates were subsequently auto-
generated using the program XPLO2D23 and minimized using
the program Xplor-NIH.24 Inhibitors were randomly positioned
approximately 20 Å from the CD4bs on gp120 using the
program GRASP,25 and models were obtained by conjoined
rigid body/torsion angle dynamics.26,27 During the dynamics
protocol, gp120 is treated as a fixed rigid body (that is, the
coordinates for gp120 are held fixed), while the inhibitor is
free to translate and rotate relative to the protein, and the
alkyl chain and linker are given torsional degrees of freedom.
Force constants for bonds, angles, and impropers were set to
1000, 500, and 750 kcal mol-1, respectively. Note that no
dihedral angle force constants were employed since dihedral
angles are either held fixed or allowed free rotation if residing
within the alkyl chains or linker. Force constants for distance
restraints were increased from 2 to 30 kcal mol-1 during the
course of the run. Each restrained minimized model was
derived from the average of a set of 20 structures for which
no intermolecular distance violations greater than 0.3 Å or
(9) Snider, B. B.; Chen, J . S.; Patil, A. D.; Freyer, A. J . Tetrahedron Lett.
1996, 37, 6977-6980.
(10) Black, G. P.; Murphy, P. J .; Walshe, N. D. A. Tetrahedron 1998, 54,
9481-9488.
(11) Overman, L. E.; Rabinowitz, M. H.; Renhowe, P. A. J . Am. Chem.
Soc. 1995, 117, 2657-2658.
(12) Nagasawa, K.; Georgieva, A.; Koshino, H.; Nakata, T.; Kita, T.;
Hashimoto, Y. Org. Lett. 2002, 4, 177-180.
(13) Reviewed in: Aron, Z. D.; Overman, L. E. Chem. Commun. 2004, in
press.
(14) Cohen, F.; Collins, S. K.; Overman, L. E. Org. Lett. 2003, 5, 4485-
4488.
(15) Cohen, F.; Overman, L. E. J . Am. Chem. Soc. 2001, 123, 10782-
10783.
(16) Nussbaum, O.; Broder, C. C.; Berger, E. A. J . Virol. 1994, 68, 5411-
5422. Salzwedel, K.; Smith, E. D.; Dey, B.; Berger, E. A. J . Virol.
2000, 74, 326-333.
(17) Reviewed in: Berger, E. A.; Murphy, P. M.; Farber, J . M. Annu. Rev.
Immunol. 1999, 17, 657-700.
(18) Kwong, P. D.; Wyatt, R.; Robinson, J .; Sweet, R. W.; Sodroski, J .;
Hendrickson, W. A. Nature 1998, 393, 648-659.
(19) Kwong, P. D.; Wyatt, R.; Majeed, S.; Robinson, J .; Sweet, R. W.;
Sodroski, J .; Hendrickson, W. A. Struct. Fold. Des. 2000, 8, 1329-
1339.
(20) Carlier, P. R.; Du, D. M.; Han, Y. F.; Liu, J .; Perola, E.; Williams, I.
D.; Pang, Y. P. Angew. Chem., Int. Ed. 2000, 39, 1775-1777.
(21) Carlier, P. R.; Cho, E. S.; Han, Y.; Liu, J .; El Yazal, J .; Pang, Y. P. J .
Med. Chem. 1999, 42, 4225-4231.
(22) Louis, J . L.; Nesheiwat, I.; Chang, L. C.; Clore, G. M.; Bewley, C. A.
J . Biol. Chem. 2003, 278, 20278-20285.
(23) Kleywegt, G. T.; J ones, T. A. Methods Enzymol. 1997, 277, 208-230.
(24) Schwieters, C. D.; Kuszewski, J . J .; Tjandra, N.; Clore, G. M. J . Magn.
Reson. 2003, 160, 66-74.
(25) Nichols, A.; Sharp, K. A.; Honig, B. Proteins Struct. Func. Genet. 1991,
11, 281-296.
(26) Bewley, C. A.; Clore, G. M. J . Am. Chem. Soc. 2000, 122, 6009-6016.
(27) Schwieters, C. D.; Clore, G. M. J . Magn. Reson. 2001, 152, 288-302.
NP049958O