Design, synthesis, and computational studies of inhibitors of Bcl-X L

Article abstract of DOI:10.1021/ja0650347

One of the primary objectives in the design of protein inhibitors is to shape the three-dimensional structures of small molecules to be complementary to the binding site of a target protein. In the course of our efforts to discover potent inhibitors of Bc

Full text of DOI:10.1021/ja0650347

Published on Web 11/30/2006
Design, Synthesis, and Computational Studies of Inhibitors of
Bcl-XL
Cheol-Min Park,*,† Tetsuro Oie,‡ Andrew M. Petros,‡ Haichao Zhang,†
Paul M. Nimmer,† Rodger F. Henry,‡ and Steven W. Elmore†
Contribution from Global Pharmaceutical R&D, Abbott Laboratories, 100 Abbott Park Road,
Abbott Park, Illinois 60064
Received July 14, 2006; E-mail: cheol-min.park@abbott.com
Abstract: One of the primary objectives in the design of protein inhibitors is to shape the three-dimensional
structures of small molecules to be complementary to the binding site of a target protein. In the course of
our efforts to discover potent inhibitors of Bcl-2 family proteins, we found a unique folded conformation
adopted by tethered aromatic groups in the ligand that significantly enhanced binding affinity to Bcl-XL.
This finding led us to design compounds that were biased by nonbonding interactions present in a urea
tether to adopt this bioactive, folded motif. To characterize the key interactions that induce the desired
conformational bias, a series of substituted N,N′-diarylureas were prepared and analyzed using X-ray
crystallography and quantum mechanical calculations. Stabilizing π-stacking interactions and destabilizing
steric interactions were predicted to work in concert in two of the substitution patterns to promote the bioactive
conformation as a global energy minimum and result in a high target binding affinity. Conversely,
intramolecular hydrogen bonding present in the third substitution motif promotes a less active, extended
conformer as the energetically favored geometry. These findings were corroborated when the inhibition
constant of binding to Bcl-XL was determined for fully elaborated analogues bearing these structural motifs.
Finally, we obtained the NMR solution structure of the disubstituted N,N′-diarylurea bound to Bcl-XL
demonstrating the folded conformation of the urea motif engaged in extensive π-interactions with the protein.
Introduction
fication often results in the loss of enthalpic energy that can
offset the gain in the entropic term.14-16 Nonbonding interactions
In recent years, the manipulation of conformations of flexible
molecules has attracted great interest.1-3 The ability to achieve
specific conformations for proteins and small organic molecules
is crucial for their chemical and biological functions. A number
of bioactive natural products with conformational flexibility rely
upon specific conformations to present functional groups in
specific orientations for their activity, and for optimal binding,
the functional groups should be complementary to those in the
protein binding site.4 The concept of preorganization has been
widely utilized in many areas, including molecular recognition
and drug design.5-7 Constraining flexible ligands by covalent
bonds in such a way that they represent bound conformations
may benefit from an entropic factor.8-13 However, the difficulty
to precisely mimic the bound conformation by covalent modi-
have also been utilized in biasing the conformations of flexible
molecules.17,18 These conformationally biased flexible molecules
may be more likely to find low energy conformations in
complexation processes without significant enthalpic penalty
over rigid systems through dynamic complementarity.19
The Bcl-2 family of proteins is the key regulator of
programmed cell death.20 Bcl-2 family members contain from
one to four Bcl-2 homology (BH) domains and can be broadly
divided into two classes; anti-apoptotic and pro-apoptotic
(10) Barboni, L.; Lambertucci, C.; Appendino, G.; Vander Velde, D. G.; Himes,
R. H.; Bombardelli, E.; Wang, M.; Snyder, J. P. J. Med. Chem. 2001, 44,
1576-1587.
(11) Bartlett, S.; Beddard, G. S.; Jackson, R. M.; Kayser, V.; Kilner, C.; Leach,
A.; Nelson, A.; Oledzki, P. R.; Parker, P.; Reid, G. D.; Warriner, S. L. J.
Am. Chem. Soc. 2005, 127, 11699-11708.
(12) Dinsmore, C. J.; Bogusky, M. J.; Culberson, J. C.; Bergman, J. M.;
Homnick, C. F.; Zartman, C. B.; Mosser, S. D.; Schaber, M. D.; Robinson,
R. G.; Koblan, K. S.; Huber, H. E.; Graham, S. L.; Hartman, G. D.; Huff,
J. R.; Williams, T. M. J. Am. Chem. Soc. 2001, 123, 2107-2108.
(13) Hanessian, S.; Moitessier, N. Curr. Top. Med. Chem. 2004, 4, 1269-1287.
(14) Babine, R. E.; Bender, S. L. Chem. ReV. 1997, 97, 1359.
(15) Wacowich-Sgarbi, S. A.; Bundle, D. R. J. Org. Chem. 1999, 64, 9080-
9089.
† Cancer Research.
‡ Advanced Technologies.
(1) Galeazzi, R.; Mobbili, G.; Orena, M. Curr. Org. Chem. 2004, 8, 1799-
1829.
(2) Bartlett, P. A.; Yusuff, N.; Pyun, H.-J.; Rico, A. C.; Meyer, J. H.; Smith,
W. W.; Burger, M. T. Spec. Publ.-R. Soc. Chem. 2001, 264, 3-15.
(3) Hruby, V. J. Acc. Chem. Res. 2001, 34, 389-397.
(4) Houk, K. N.; Leach, A. G.; Kim, S. P.; Zhang, X. Angew. Chem., Int. Ed.
2003, 42, 4872-4897.
(16) Davidson, J. P.; Lubman, O.; Rose, T.; Waksman, G.; Martin, S. F. J. Am.
Chem. Soc. 2002, 124, 205-215.
(5) Cram, D. J. Angew. Chem., Int. Ed. Engl. 1986, 25, 1039-1134.
(6) Ciani, B.; Jourdan, M.; Searle, M. S. J. Am. Chem. Soc. 2003, 125, 9038-
9047.
(17) Knust, H.; Hoffmann, R. W. Chem. Rec. 2002, 2, 405-418.
(18) Hodge, C. N.; Lam, P. Y. S.; Eyermann, C. J.; Jadhav, P. K.; Ru, Y.;
Fernandez, C. H.; De Lucca, G. V.; Chang, C.-H.; Kaltenbach, R. F., III;
Holler, E. R.; Woerner, F.; Daneker, W. F.; Emmett, G.; Calabrese, J. C.;
Aldrich, P. E. J. Am. Chem. Soc. 1998, 120, 4570-4581.
(19) Morgan, B. P.; Holland, D. R.; Matthews, B. W.; Bartlett, P. A. J. Am.
Chem. Soc. 1994, 116, 3251-3260.
(7) Ilioudis, C. A.; Tocher, D. A.; Steed, J. W. J. Am. Chem. Soc. 2004, 126,
12395-12402.
(8) Reichelt, A.; Gaul, C.; Frey, R. R.; Kennedy, A.; Martin, S. F. J. Org.
Chem. 2002, 67, 4062-4075.
(9) Meyer, J. H.; Bartlett, P. A. J. Am. Chem. Soc. 1998, 120, 4600-4609.
(20) Danial, N. N.; Korsmeyer, S. J. Cell 2004, 116, 205-219.
9
16206
J. AM. CHEM. SOC. 2006, 128, 16206-16212
10.1021/ja0650347 CCC: $33.50 © 2006 American Chemical Society

Computational Studies of Inhibitors of Bcl-XL
A R T I C L E S
Figure 1. Chemical structures of inhibitors of Bcl-XL 1 and ABT-737.
proteins. The anti-apoptotic proteins (e.g., Bcl-2, Bcl-XL) are
characterized by four BH domains, BH1-4. The pro-apoptotic
proteins can be further subdivided into those that incorporate
three BH domains (e.g., Bax, Bak) and the BH3-only proteins
(e.g., Bad, Bid, Bim, Bik). The interplay between these three
groups of proteins serves as the gateway to the intrinsic
apoptosis pathway.21-24 Under conditions of cellular stress, pro-
apoptotic BH3-only proteins (e.g., Bid, Bim) are mobilized to
activate the direct mediators of apoptosis, Bax and Bak. Upon
activation, cytochrome C release from the mitochondria leads
to formation of the apoptosome which initiates the caspase-9
pathway. The anti-apoptotic proteins, Bcl-2 and Bcl-XL, prevent
this activation by directly binding to and sequestering these pro-
apoptotic proteins. Due to the overexpression of Bcl-2 and Bcl-
XL in numerous types of tumors, compounds selectively
inhibiting Bcl-2 and Bcl-XL would be of great therapeutic value
for cancer. A number of approaches have been reported in the
literature, including peptide conjugates, natural products, virtual
screening, high-throughput screening, and structure-based de-
sign.25
We have recently reported the discovery of the potent Bcl-
XL ligand 1 by an NMR-based fragment screening technique
(SAR by NMR) and structure-guided parallel synthesis.26 Further
elaboration of 1 led to the discovery of ABT-737, which
displayed subnanomolar binding affinities to Bcl-2, Bcl-XL, and
Bcl-w, which regressed established tumors in murine xenograft
models.27,28 Figure 1 shows the chemical structures of inhibitors
of Bcl-XL 1 and ABT-737; Figure 2 shows a schematic diagram
of critical binding interactions between compound 1 and Bcl-
XL.26 Intramolecular π-stacking interactions between the nitro-
phenyl and thiophenyl groups of 1 induce a folded conformation
that fits nicely into the binding pocket formed by BH1 and BH3
domains of the protein. Furthermore, these two aromatic rings
engage in intermolecular interactions with the aromatic groups
of the surrounding protein residues: (a) the nitrophenyl group
with Tyr194 in “herring bone” structure (intersection angle of
about 45° between the two planes of rings) and (b) the
thiophenyl group with Phe97 in a parallel-displaced structure
and with Tyr101 in a displaced T-shape (intersection angle of
about 90° between the two planes of rings). This extensive intra-
Figure 2. Schematic diagram of an NMR-based structure of compound 1
bound to Bcl-XL showing the key residues lining the binding site.
Figure 3. Four representative conformations of N,N′-diarylurea.
and intermolecular π-π interaction network is essential for tight
binding to this large surface area, hydrophobic binding site and
serves as one of the key anchoring points for the ligand.
Given the significant contribution to binding affinity from
this π-π interaction network created by the folded conforma-
tion, we were prompted to explore alternative tethers that might
bias the molecule to adopt this conformation by utilizing
nonbonding interactions. We chose to explore N,N′-diarylureas
due to their unique preference to adopt folded conformations.29
The four possible N,N′- diarylurea conformations are shown in
Figure 3 and are defined on the basis of the orientation of the
aryl groups relative to the urea carbonyl group. To examine
whether a urea tether would preorganize the compound to adopt
an appropriate folded conformation, compounds 2, 3, and 4
containing systematic alterations in nitrogen substitution were
prepared. Their binding affinities were compared with that of
the parent bent-back compound 18, and the conformational
preferences and flexibilities were further analyzed using X-ray
crystallography, quantum mechanical calculations, and NMR
spectroscopy.
Results and Discussion
Design and Synthesis. The three closely related compounds
2, 3, and 4 shown in Figure 4 were chosen to probe the effect
of nonbonding interactions arising from simple nitrogen sub-
stitution on overall N,N′-diarylurea conformation. Figure 5
shows atom and torsional angle numbering. We reasoned that
(21) Cory, S.; Adams, J. M. Nat. ReV. Cancer 2002, 2, 647-656.
(22) Hanahan, D.; Weinberg, R. A. Cell 2000, 100, 57-70.
(23) Borner, C. Mol. Immunol. 2003, 39, 615-647.
(24) van Delft, M. F.; Huang, D. C. S. Cell Res. 2006, 16, 203-213.
(25) Elmore, S. W.; Oost, T. K.; Park, C.-M. Annu. Rep. Med. Chem. 2005, 40,
245.
(26) Petros, A. M.; et al. J. Med. Chem. 2006, 49, 656-663.
(27) Oltersdorf, T.; et al. Nature 2005, 435, 677-681.
(28) Wendt, M. D.; et al. J. Med. Chem. 2006, 49, 1165-1181.
(29) Kurth, T. L.; Lewis, F. D.; Hattan, C. M.; Reiter, R. C.; Stevenson, C. D.
J. Am. Chem. Soc. 2003, 125, 1460-1461 and references therein.
9
J. AM. CHEM. SOC. VOL. 128, NO. 50, 2006 16207

A R T I C L E S
Park et al.
Scheme 1 a
Figure 4. Compounds containing N,N′-diarylurea motifs.
a Reagents and conditions: (a) (DMB)2NH, TEA, dichloromethane, 0
°C to room temperature (rt), 2 h; (b) for R1 ) Me/MeNH2, MeOH/THF
(3.5:1), 80 °C, overnight, sealed tube; for R1 ) H/NH3, MeOH, 125 °C, 16
h, 120 PSI; (c) phosgene, PS-DIEA, dichloromethane/toluene, 50 °C, 24 h,
sealed tube; (d) (i) for R2 ) Me/PhNH(Me), dichloromethane, rt, overnight;
for R2 ) H/PhNH2, dichloromethane, rt, overnight; (ii) dichloromethane/
TFA/triethylsilane (1:0.9:0.1), rt, 2 h; (e) (i) 13, DMAP, DIPEA, 1,2-
dichloroethane, 80 °C, 36 h; (ii) dichloromethane/TFA/triethylsilane (1:
0.9:0.1), rt, 2 h. DMB ) 2,4-dimethoxybenzyl.
Figure 5. Atom and torsional angle numbering.
the desired anti-anti folded conformation would be most
favored by the disubstitution found in 2 because of (a) the
presence of reinforcing π-stacking interaction between the
aromatic groups in 2 and (b) the presence of unfavorable steric
interactions between methyl aryl or methyl alkyl groups in the
anti-syn, syn-anti, and syn-syn conformers (Figure 3).
Compounds 3 and 4, both of which contain a single nitrogen
methyl group, would have less potential for such steric interac-
tions. We sought to explore how modifications of the urea
substitution pattern influenced the energetic preference for
conformation and the binding affinity to Bcl-XL.
Scheme 2 a
The synthesis of compound 2 commenced with the prepara-
tion of sulfonamide 6 by coupling of the sulfonyl chloride 5
with bis-2,4-dimethoxybenzylamine (Scheme 1). Nucleophilic
aromatic substitution of chloride 6 with methylamine gave
aniline 7, which was treated with phosgene to afford carbamoyl
chloride 9. Reaction of 9 with N-methylaniline, followed by
deprotection with trifluoroacetic acid, yielded compound 10.
Carboxylic acid 16 was prepared by reductive amination of
piperazine 14 with â-phenylcinnamaldehyde followed by sa-
ponification (Scheme 2). Carbodiimide-mediated coupling of
sulfonamide 10 with carboxylic acid 16 yielded compound 2.
Compound 3 was prepared in a manner similar to compound 2
by reacting carbamoyl chloride 9 with aniline and the resulting
urea 11 processed as above. Compound 4 was prepared by the
reaction of aryl chloride 6 with ammonia and the resulting
aniline condensed with carbamoyl chloride 13.30 Removal of
the dimethoxybenzyl groups under acidic conditions and
coupling of the resulting sulfonamide 12 with carboxylic acid
16 yielded compound 4. Compound 18 was prepared by
carbodiimide-mediated coupling of carboxylic acid 16 and
sulfonamide 17.26
a Reagents and conditions: (a) (Ph)2C)CHCHO, Na(OAc)3BH, 1,2-
dichloroethane, rt, overnight; (b) LiOH, THF/MeOH/H2O (2:1:1), rt; (c)
EDCI, DMAP, dichloromethane, rt, overnight, 10, 11, 12 for 2, 3, and 4,
respectively.
polarization assay as described previously.26 Compounds 2 and
3 exhibited binding affinities of 5.0 and 7.4 nM, respectively,
which is comparable to that of the parent compound 18 (1.1
nM) (Table 1). However, compound 4, where R1 ) H, shows
a significant loss of affinity (550 nM). These results prompted
a more in-depth conformational analysis.
Binding Study. The binding constants for compounds 2, 3,
4, and 18 to Bcl-XL were determined using a fluorescence
(30) Freer, R.; McKillop, A. Synth. Commun. 1996, 26, 331-350.
9
16208 J. AM. CHEM. SOC. VOL. 128, NO. 50, 2006

Computational Studies of Inhibitors of Bcl-XL
A R T I C L E S
Table 1. Binding Affinity of the Compounds
intermolecular hydrogen bonds, compound 12 also contains an
intramolecular hydrogen bond between the aryl nitro group and
the adjacent urea N-H that might provide a barrier to adopting
a folded conformation. This is consistent with the dramatically
lower binding affinity of 4. The inconsistency between the
conformation of 11 found in its crystal structure and the high
binding affinity of 3 may suggest a small energy difference
between the various conformations. To better examine this, we
next determined global energy minimum conformations for 10-
12 employing quantum mechanical calculations.
Quantum Mechanical Calculations. To assess the relative
stability of the conformations of sulfonamides 10, 11, and 12,
quantum mechanical calculations were conducted.33 Figure 8
illustrates the low energy conformations of the anti-anti (A1),
anti-syn (B1), syn-syn (C1), and syn-anti (D1) forms for
compounds 10, 11, and 12. In addition, fragment 19 of the parent
compound 18 containing the NHCH2CH2S linker was subjected
to calculations, and the structures were compared with that found
in complex with the protein (Figure 7).
For fragment 19 representing the parent compound 18, the
relative energies of the folded and extended conformers are of
particular interest. The input geometries for the optimization
of these two conformers were taken directly from the experi-
mental NMR structure, but one torsional angle of N-C-C-S
was adjusted to 180° in the input geometry of the extended
conformer. The folded conformation 19A1 is more stable than
the extended conformation 19B1 by 1.69 kcal/mol and exhibits
conformation very similar to that found in the bound state
(Figure 7).
The lowest energy conformation for 10 identified in these
calculations (10A1, Figure 8A) matches very closely that found
in the crystal structure except for the orientation of the
sulfonamide group. The conformation (10A2, Figure 9) corre-
sponding to the crystal structure (10Xr) is only 0.12 kcal/mol
higher in energy than 10A1. The conformations found in 10B1-
10D1 are energetically disfavored (5.65-7.16 kcal/mol) likely
due to the unfavorable methyl-methyl and methyl-phenyl
steric interactions. It is not apparent from these studies the extent
to which other intramolecular forces, such as π-stacking
interactions, contribute to the folded conformation.
Calculations on compound 11 that lacks the N3 methyl group
reveal that the folded conformation is still favored by 2.69-
4.57 kcal/mol even in the absence of significant steric bias
(Figure 8B, 11A-11D). This may explain the nearly identical
potency of 3 in comparison to that of 2. The next higher energy
conformation 11B1 (2.69 kcal/mol) differs only in the orientation
of the terminal aryl group and underscores the propensity to
form π-stacking interactions34,35 even at the expense of partial
loss of conjugation between N3 and the phenyl ring in 11A1.
It is noteworthy that the calculated energy of the conformation
corresponding to the crystal structure (11C2, Supporting Infor-
mation) is 4.58 kcal/mol higher than that of the global energy
minimum conformation (11A1). This highlights the extent to
which the intermolecular interactions inherent to the crystal
lattice can bias conformational preference in the solid state.
The lowest energy conformation of 12 (12D1) found by
calculations is in good agreement with that found in the crystal
compd
Ki (nM)a
18
2
3
1.1 ( 0.4
5.0 ( 0.8
7.4 ( 2.3
550 ( 170
4
a The Ki values represent the mean ( SEM (n g 3).
X-ray Crystallography. To examine the conformational
preference of these compounds, we first focused our effort on
determining the X-ray crystal structures of the fully elaborated
compounds 2-4. However, difficulties in obtaining suitable
crystals of these molecules led us to examine the structures of
the sulfonamide fragments 10, 11, and 12 as suitable surrogates.
High-resolution single-crystal X-ray diffraction data were suc-
cessfully obtained for 10 and 11, but the data for 12 were of
lower quality due to small crystal size. Consequently, the
structure of 12 was independently determined using structure
determination from powder diffraction (SDPD) techniques.31
The unit cell from the single-crystal data was fit to the
experimental powder data using Pawly refinement to extract
intensities. The starting model was based on the single-crystal
structure, but all of the rotatable bonds and the orientation of
molecules in the unit cell were perturbed. In the SDPD
determination, all nonrigid bonds were allowed to rotate freely.
The resulting structure was very similar in conformation to the
single-crystal structure. The average root-mean-square deviation
between equivalent atoms in the two models was 0.272 Å, with
the greatest difference occurring in one of the two sulfonamide
oxygen atoms.
The three crystal structures thus obtained are shown in Figure
6 and reveal that the relative orientations of the two phenyl
groups with respect to the carbonyl group are all unique (anti-
anti, syn-syn, and syn-anti, for compounds 10, 11, and 12,
respectively). Compound 10 displays the desired folded con-
formation in the crystal structure. The comparable Bcl-XL
binding affinity of compound 2 compared to that of compound
18 suggests the crystal structure may represent its protein-bound
conformation. However, both sulfonamides 11 and 12 exhibit
extended conformations in the crystal structure. This extended
conformation is not consistent with the high binding affinity of
3.
X-ray crystal structures may not necessarily represent the low
energy conformations of flexible molecules due to the potentially
significant contributions from intermolecular interactions involv-
ing adjacent and solvent molecules that are inevitable in the
condensed states.32 In particular, the difference in conformations
found in these compounds may be attributed in part to the
difference in intermolecular hydrogen-bonding patterns. For
example, the crystal structure of compound 10 reveals that its
sulfonamide group is engaged in only one intermolecular
hydrogen bond interaction (hydrogen bonds are represented as
dotted lines in crystal packing lattice, Figure 6B). Compound
11, on the other hand, has a much more extensive intermolecular
hydrogen-bonding network involving both the sulfonamide and
urea subunits that likely influences the preference for an
extended conformation in the solid state. In addition to these
(33) Details are provided in the Supporting Information.
(34) Rashkin, M. J.; Waters, M. L. J. Am. Chem. Soc. 2002, 124, 1860-1861.
(35) Tsuzuki, S.; Honda, K.; Uchimaru, T.; Mikami, M.; Tanabe, K. J. Am.
Chem. Soc. 2002, 124, 104-112.
(31) MS Modeling, Powder SolVe Module, version 3.2; Accelrys Software: San
Diego, CA, 2004.
(32) Dunitz, J. D.; Gavezzotti, A. Acc. Chem. Res. 1999, 32, 677-684.
9
J. AM. CHEM. SOC. VOL. 128, NO. 50, 2006 16209

A R T I C L E S
Park et al.
Figure 6. (A) Chem3D representations of the crystal structures. (B) Views of crystal packing generated by Mercury for 10, 11, and 12. Hydrogen atoms
are omitted for clarity.
hydrogen bond to the stabilization of the structures.38 The
geometries devoid of this hydrogen bond (12A2-D2) were
generated by rotating the nitrophenyl ring of 12A1-D1 by 180°
followed by energy minimization, and the resulting conforma-
tions were compared to those containing the hydrogen bond
(Figure 8D). The energy differences for 12A1-12A2 and
12B1-12B2, 0.58 and 1.65 kcal/mol, respectively, are signifi-
cantly smaller than those for 12C1-12C2 and 12D1-12D2,
4.49 and 3.91 kcal/mol, respectively. These apparently dimin-
ished contributions of the hydrogen bond in 12A1 and 12B1
can be attributed to the increased repulsive interaction between
the oxygen atoms of carbonyl and nitro groups (respective O‚
‚‚O distances of 4.275 and 4.334 Å in 12A1 and 12B1 versus
4.923 and 4.941 Å in 12C1 and 12D1). Additional contribution
to the stability of 12C1 and 12D1 can be identified in the
attractive electrostatic interaction between the aromatic hydrogen
atom at C6 and the carbonyl oxygen atom (H‚‚‚O distances of
2.089 and 2.080 Å, respectively).39,40
Thus, the local conformation around the nitro group appears
to be one of the key factors in governing the overall conforma-
tion (folded versus extended). The methyl groups on N2 in
compounds 10 and 11 steer the nitrophenyl ring out of amide
bond conjugation, leading to a folded conformation being
favored over the extended. The presence of the additional N3
methyl group in 10 enhances the preference for the folded
conformation compared to 11 (5.65 versus 2.72 kcal/mol
difference from the lowest energy extended conformation) due
to unfavorable steric interactions in the alternative conformers.
On the other hand, compound 12, which lacks the N2 methyl,
is able to engage in an intramolecular hydrogen bond that allows
a high degree of π-conjugation that highly favors an extended
conformation.
Solution Structure of the Protein-Ligand Complex. These
studies suggest that compounds such as 2 containing N,N′-
disubstituted diarylurea would prefer a folded conformation and
easily adopt the putative bioactive conformation. To confirm
this, we obtained a solution structure of compound 2 bound to
Bcl-XL.
Figure 7. Comparison of the NMR structure for the fragment 19 of the
parent compound 18 containing NHCH2CH2S linker bound to Bcl-XL with
low energy conformations found by calculations.
structure (12Xr) (Figure 8C). This conformation provides
maximal overlap of the π-systems of the nitro and urea nitrogens
and carbonyl group. This extended conjugation contributes to
the energetic differences between 12D1 and the other potential
conformers (4.10-10.81 kcal/mol). Compound 12, which lacks
the N2 methyl group, displays features drastically different from
those of compound 11 in that the proton on N2 is poised to
form an intramolecular hydrogen bond (1.812 Å between
hydrogen and oxygen atoms in the calculated structure) with
the oxygen atom of the nitro group. The most accessible folded
conformation of 12 (12A1) is significantly higher in energy (5.15
kcal/mol) than the global minimum (12D1), rendering it much
less available as a bioactive conformation. This presumably
explains why the binding affinity of 4 is reduced by more than
2 orders of magnitude relative to that of either 2 or 3, which
are easily able to adopt, and in fact prefer, the folded motif.
The two other extended conformations 12B1 and 12C1 are
9.16 and 4.10 kcal/mol higher in energy than the global
minimum, respectively. The significantly higher energy of 12B1
in comparison to 12C1 appears, at least in part, to be due to a
loss of conjugation of the π-system involving N2-C1 (torsional
angle t4 of 61.6° in comparison to -176.7° in 12C1). Although
the origin of the large energy difference (4.10 kcal/mol) between
12C1 and 12D1 is not immediately apparent, it might be
rationalized by considering a potentially stabilizing electrostatic
interaction between the electron-deficient proton on N2 and the
π-electrons of the phenyl group on N3 (the distance of 2.31 Å
between hydrogen atom and phenyl ring) in 12D1.36,37
All structure calculations were carried out with the program
CNX.41 Initially, compound 2 was randomly positioned near
The hydrogen bonds between the nitro group and the proton
on N2 found in all of the local energy minimum structures of
12 prompted us to investigate the contribution of this type of
(38) Allen, F. H.; Baalham, C. A.; Lommerse, J. P. M.; Raithby, P. R.; Sparr,
E. Acta Crystallogr., Sect. B 1997, 53, 1017-1024.
(36) Rashkin, M. J.; Hughes, R. M.; Calloway, N. T.; Waters, M. L. J. Am.
Chem. Soc. 2004, 126, 13320-13325.
(37) Lavieri, S.; Zoltewicz, J. A. J. Org. Chem. 2001, 66, 7227-7230.
(39) Pierce, A. C.; terHaar, E.; Binch, H. M.; Kay, D. P.; Patel, S. R.; Li, P. J.
Med. Chem. 2005, 48, 1278-1281.
(40) Taylor, R.; Kennard, O. J. Am. Chem. Soc. 1982, 104, 5063-5070.
9
16210 J. AM. CHEM. SOC. VOL. 128, NO. 50, 2006

Computational Studies of Inhibitors of Bcl-XL
A R T I C L E S
Figure 8. Low energy conformations of the anti-anti (A1), anti-syn (B1), syn-syn (C1), and syn-anti (D1) forms for compounds 10, 11, and 12 (A, B,
and C, respectively) and conformations for compound 12 lacking hydrogen bonding (D).44
Figure 9. Comparison of the crystal structure (10Xr) and the two lowest
energy structures (10A1, 10A2).
the binding groove of Bcl-XL and the observed intermolecular
NOEs were used to dock the ligand into the groove. This
docking was followed by energy minimization and a standard
simulated annealing protocol.42 During the simulated annealing,
the coordinates of the protein were held fixed, with the exception
of those residues that line the binding groove (residues 96-
112, 127-142, and 191-194). This protocol was based on the
structural studies of Bak peptide binding to Bcl-XL, for which
Figure 10. NMR-based solution structure of compound 2 bound to Bcl-
XL.
we observed structural rearrangements only for residues lining
the hydrophobic groove upon peptide binding.43 The conforma-
lecular NOEs. For compound 2, a total of 100 intermolecular
NOEs were used to dock the ligand to Bcl-XL.
tion of these residues was governed by the observed intermo-
(41) Brunger, A. T. X-PLOR, version 3.1; Yale University Press: New Haven,
CT, 1992.
Figure 10 shows the solution structure of 2 bound to the
protein. Indeed, the urea moiety of compound 2 exhibits a folded
conformation and a very similar arrangement of the π-interaction
network found in compound 1 complexed with the protein
(Figure 2); the nitrophenyl group on N2 with Tyr194, the phenyl
group on N3 with Phe97 and Tyr101. These results unequivo-
(42) Kuszewski, J.; Nilges, M.; Brunger, A. T. J. Biomol. NMR 1992, 2, 33-
56.
(43) Sattler, M.; Liang, H.; Nettesheim, D.; Meadows, R. P.; Harlan, J. E.;
Eberstadt, M.; Yoon, H. S.; Shuker, S. B.; Chang, B. S.; Minn, A. J.;
Thompson, C. B.; Fesik, S. W. Science 1997, 275, 983-986.
(44) The Cartesian coordinates along with torsional angles and energies of all
minimized geometries are provided in the Supporting Information.
9
J. AM. CHEM. SOC. VOL. 128, NO. 50, 2006 16211

A R T I C L E S
Park et al.
cally establish the bioactive conformation of the urea motif of
compound 2 being a folded conformer.
tuted ureas 2, 3, and 4. While 2 and 3 (corresponding to
sulfonamide 10 and 11, respectively) bind to the protein with
high affinity (5.0 and 7.4 nM), 4 (corresponding to sulfonamide
12) shows a dramatic loss of potency (550 nM). The high
affinity of 2 and 3 can be explained by the heavy population of
the global energy minimum conformers that adopt folded
conformations of the urea subunits. The folded conformations
10A1 and 11A1 are separated from the next higher energy
conformations 10D1 and 11B1 by 5.65 and 2.69 kcal/mol,
respectively. The significantly lower binding affinity of 4 arises
from the large preference (5.15 kcal/mol) for a highly conju-
gated, extended urea conformation (12D1) that is due, at least
in part, to the presence of an intramolecular hydrogen bond.
The folded conformation of 2 that was predicted by both the
X-ray crystal structure studies and energy minimization calcula-
tions on the sulfonamide surrogate 10 to be favored was indeed
found to be its protein-bound conformation by obtaining the
NMR solution structure of 2 bound to Bcl-XL.
To gain an idea on the bound conformation of the urea in
compound 3, we briefly investigated the compound in the
presence of Bcl-XL by NMR. The similarity of the NOE pattern
in conjunction with the results from binding, crystallography,
and calculations led us to conclude that the folded conformation
of the urea group in compound 3 also represents its bound
conformation. The intermediate exchange rate of compound 4
for Bcl-XL prohibited us from obtaining high quality spectra.
However, the weak binding affinity of 4 is consistent with the
diminished population of the bioactive conformation due to the
high energy of the folded conformation of the urea motif.
Conclusions
The concept of preorganization has been widely utilized in
the area of molecular recognition, including small molecule-
protein interactions. The binding affinity of a ligand for a protein
could be improved with higher complementarity between a
ligand and a receptor binding site. The folded conformation
found in inhibitors bound to Bcl-2 family proteins led us to
design compounds with alternative motifs to favor the protein-
bound conformation. To circumvent some of the pitfalls
associated with constraining a conformation by covalent modi-
fications, we have instead exploited nonbonding interactions.
Evaluation of structurally related analogues using target protein
binding affinity, X-ray crystallography, and quantum mechanical
calculations allowed us to gain insight into factors influencing
the bioactive conformations of a series of differentially substi-
Acknowledgment. We thank Thomas B. Borchardt for the
powder X-ray study and the staff of Structural Chemistry and
Pressure Lab for NMR, MS, and high-pressure reactions.
Supporting Information Available: Experimental details,
X-ray crystallographic data for 10, 11, and 12 in CIF format,
Cartesian coordinates and energies of structures 10, 11, 12, and
19, and complete citation for refs 26-28. This material is
available free of charge via the Internet at http://pubs.acs.org.
JA0650347
9
16212 J. AM. CHEM. SOC. VOL. 128, NO. 50, 2006