Structure-activity based study of the Smac-binding pocket within the BIR3 domain of XIAP

Article abstract of DOI:10.1016/j.bmc.2007.02.010

A small series of peptide mimics was designed and synthesized to contain a heterocyclic ring in place of the potentially labile N-terminal peptide bond of the tetrapeptide containing the Smac-XIAP-binding motif. Two Smac mimics were shown to bind to the B

Full text of DOI:10.1016/j.bmc.2007.02.010

Bioorganic & Medicinal Chemistry 15 (2007) 2935–2943
Structure–activity based study of the Smac-binding pocket within
the BIR3 domain of XIAP
Aislyn D. Wist,a,*, Lichuan Gu,b Stefan J. Riedl,b,à Yigong Shib
and George L. McLendona,*,§
aDepartment of Chemistry, Princeton University, Princeton, NJ 08544, USA
bDepartment of Molecular Biology, Princeton University, Princeton, NJ 08544, USA
Received 19 December 2006; revised 31 January 2007; accepted 8 February 2007
Available online 11 February 2007
Abstract—A small series of peptide mimics was designed and synthesized to contain a heterocyclic ring in place of the potentially
labile N-terminal peptide bond of the tetrapeptide containing the Smac-XIAP-binding motif. Two Smac mimics were shown to bind
to the BIR3 domain of XIAP with moderate affinity and one displayed increased activity in cells relative to the Smac peptides. The
structures of BIR3-XIAP in complex with a Smac peptide and a peptide mimic were solved and analyzed to elucidate the structure–
activity relationship surrounding the Smac-binding domain within BIR3-XIAP.
ꢀ 2007 Elsevier Ltd. All rights reserved.
1. Introduction
from the mitochondria, Smac/DIABLO (the second
mitochondrial activator of apoptosis6/direct IAP-bind-
ing protein with low pI)7 binds to XIAP using the same
binding motif as caspase-9, thus causing its displace-
ment.8,9 Once released from XIAP, caspase-9 can acti-
vate downstream caspases (ꢀ3 and ꢀ7) which go on
to destroy cellular targets and promote cell death. How-
ever, due to the high level of XIAP present in many
types of cancer cells,5,10 the amount of Smac released
from the mitochondria may not be sufficient to over-
come the inhibitory effect of XIAP on caspase-9; thus
tipping the balance toward life rather than programmed
cell death. The anti-apoptotic behavior of XIAP and the
critical role it plays in the apoptotic program makes the
Smac-XIAP interaction an important drug target.
Apoptosis plays a major role in maintaining the balance
between cell proliferation and cell death. Imbalances in
the apoptotic pathway are linked to several different
types of diseases.1–3 For instance, it is generally recog-
nized that in many types of cancer the apoptotic pathway
is inactivated in some way. One mechanism by which this
occurs is the upregulation of anti-apoptotic proteins,
such as the IAP (inhibitor of apoptosis) family of pro-
teins, which prevent apoptosis from occurring by seques-
tering caspases (cysteine-aspartic acid protease).4,5
An IAP family member, XIAP (X-linked IAP), func-
tions by directly binding and inhibiting the active sub-
unit of caspase-9, an initiation caspase in the
mitochondrial cell death pathway.2 Upon its release
Molecules able to mimic Smac could counter an excess
of XIAP, restoring the ability of the apoptotic machin-
ery to fully execute programmed cell death. Typically
it is a greater challenge to target a protein–protein inter-
action compared to an enzyme active site with a small
molecule, because large and diverse surface areas are
generally involved and drug molecules are usually low
molecular weight.11,12 The design of small molecules to
target the Smac-XIAP interaction is facilitated by the
structural data13,14 which revealed that the interaction
consists primarily of the N-terminal tetrapeptide of
Smac binding to a surface groove on the BIR3 (baculo-
virus IAP repeat) domain of XIAP. However, since the
target is a protein–protein interaction, the structure
Keywords: XIAP; Smac; Peptide mimetics; Apoptosis; Rational design;
Oxazole.
*
Corresponding authors. Tel.: +1 919 684 4510; fax: +1 919 684
8503; e-mail addresses: aislyn.wist@mssm.edu; george.mclendon@
duke.edu
Present address: Department of Pharmacology and Biological
Chemistry, Mount Sinai School of Medicine, New York, NY
10029, USA.
à
Present address: Burnham Institute for Medical Research, La Jolla,
CA 10901, USA.
§
Present address: Department of Chemistry, 104 Allen Building, Duke
University, Durham, NC 27708, USA.
0968-0896/$ - see front matter ꢀ 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmc.2007.02.010

2936
A. D. Wist et al. / Bioorg. Med. Chem. 15 (2007) 2935–2943
activity relationship is particularly convoluted by struc-
tural dynamics contributed by both the small molecule
and the flexible protein-binding site. In order to eluci-
date the structure activity relationship, it is necessary
to capture the molecular movements during binding.
crystals and the structure of the bound complex was
˚
solved at 2.8 A resolution. The structure of AVPF
bound to BIR3-XIAP was also solved. These
structures are compared to the Smac:BIR3-XIAP
structure (PDB ID: IG3F).14 Images rendered using
PyMol20 from the X-ray crystallographic data of
AVPF or AoxSPW bound to BIR3-XIAP are shown
in Figure 1a–c.
In the present study, novel Smac mimics were designed
and synthesized to contain a heterocycle in place of the
N-terminal bond. While there are several studies which
produced partially non-peptidic Smac mimics,15–18 this
is the first study to present Smac mimics where the N-ter-
minal peptide bond is modified to a non-peptidic struc-
ture and binding affinity for BIR3-XIAP is retained.
From a drug design perspective, it is important to pro-
duce a completely non-peptidic molecule to maximize
bioavailability and stability properties. This has been a
challenge with this particular target due to the small
hydrophobic pocket and intricate hydrogen bonding
network involved with the binding of the N-terminal
segment of the Smac peptide/mimic. In order to under-
stand the structural implications of this modification,
the crystal structures of Smac, a peptide ligand or a pep-
tide mimic bound to the BIR3 domain of XIAP (BIR3-
XIAP), are presented and analyzed for conformational
changes that occur within the binding pocket and in
the respective ligands. The structural basis for the differ-
ences in binding affinity for the Smac mimic compared
to the peptides is described. The biological activity of
the peptide mimic and peptides is compared in cell via-
bility assays.
2.2. Comparison of the hydrogen-bonding networks
In comparing the crystal structures of AVPI, AVPF,
and AoxSPW bound to BIR3-XIAP, it is apparent that
the hydrogen-bonding network has changed in the
AoxSPW structure relative to the peptide/protein struc-
tures. There are five hydrogen bonds (depicted by red
dotted lines in Figure 1) to BIR3 in the AoxSPW struc-
ture and seven hydrogen bonds to BIR3 each in the
AVPI and AVPF structures. The terminal amine of
AoxSPW maintains two hydrogen bonds with the car-
boxylic acid of BIR3 E314, but loses one with Q319 that
is present in the AVPI and AVPF structures. Two
hydrogen bonds involving the N-terminal peptide bond
are present, but shifted around slightly in the AoxSPW
relative to the AVPI and AVPF structures. The position
of the oxazole ring causes a loss of the hydrogen bond
with the indole N–H of W323. However, the carbonyl
oxygen between the proline and the oxazole ring of
AoxSPW forms a new hydrogen bond with the hydrox-
ide proton of T308, which is not present in the AVPI
and AVPF structures. A backbone hydrogen bond
formed between the amide proton in the second peptide
bond (P–I or P–F) and BIR3 G306 is not present in the
AoxSPW structure due to the orientation of AoxSPW in
the pocket. Overall, there is a loss of three hydrogen
bonds and a gain of one in the AoxSPW structure with
respect to the hydrogen bonds in the AVPI and AVPF
structures, which may partially account for the relative
loss of binding affinity with AoxSPW for BIR3-XIAP.
2. Results and discussion
2.1. Design and synthesis of Smac mimics
The Smac mimic chosen for this structural study was
designed based on a tetrapeptide (AVPW) that has been
shown to bind to BIR3-XIAP with high affinity.19 This
previously published study mapped out the residues that
are critical for binding to BIR3-XIAP by systematically
substituting the amino acids in the N-terminal region of
Smac. Specifically, modification of the position 1 Ala
dramatically reduced binding affinity; whereas, the posi-
tion 4 Ile was more amenable to modification and
switching to aromatic residues such as Phe and Trp
increased binding affinity for BIR3-XIAP. In order to
reduce the peptidic character of the Smac mimic, one
peptide bond was replaced with an oxazole ring to pro-
duce a series of Smac mimics (Table 1) with the anticipa-
tion that such modification would render the molecule
more ‘drug-like’ by increasing its uptake and stability
in a cellular context.
In addition to the hydrogen bonds, there are hydropho-
bic interactions between AoxSPW and BIR3-XIAP.
Similar to AVPI and AVPF, the A1 b-methyl group is
buried in the small hydrophobic pocket in BIR3 near
W310. The methylene chains of BIR3 K297 and K299
interact with the indole ring of W4 in AoxSPW, in a sim-
ilar manner as F4 of AVPF. The distance between the
heteroatoms in the oxazole ring in the nearby BIR3
residues, Q319 (sidechain) and T308 (backbone), indi-
cates there are also dipole–dipole interactions, which
may partially compensate for missing hydrogen bonds.
Taken together, it appears that several intramolecular
hydrogen bonds of BIR3 were interchanged in order
to accommodate AoxSPW into the binding pocket,
while maximizing hydrophobic and dipole–dipole
interactions.
The heterocycle insertion caused
a reduction in
binding affinity, but not as drastically as other modi-
fications to the first peptide bond, such as N-methyla-
tion.19 AoxSPW bound to BIR3-XIAP with an affinity
of 30 lM (Table 1), while AVPF and AVPI bind with
0.5, 0.04 lM, respectively. In order to understand the
structural differences that attributed to the difference
in binding affinity, AoxSPW (Table 1, entry 2) was
co-crystallized with BIR3-XIAP to yield X-ray quality
2.3. Comparison of ligand conformations
The positions of bound ligand were compared by over-
laying each of the bound structures with respect to the
residue positions in XIAP-BIR3 using XtalView. This
analysis allowed for further explanation of the relative

A. D. Wist et al. / Bioorg. Med. Chem. 15 (2007) 2935–2943
2937
Table 1. The position 1, 2 peptide bond in N-terminal Smac tetrapeptide was replaced by an oxazole
Entry
Smac mimic
AA
KD std deva (lM)
11
Relative to AVPIb
50 14
70
1
2
3
4
AoxSPF
AoxSPW
AoxSPY
AoxSPI
Phe
Trp
Tyr
Ile
3
30 12
120 60
300 60
8
170 80
400
The binding affinity for BIR3-XIAP and the binding affinity relative to AVPI are shown.
a Standard deviations for three or more separate experiments.
b Relative to the value for AVPI during each individual assay was calculated by dividing the KD of the Smac mimic by the KD of AVPI.
Figure 1. Image rendered from crystal structure of the (a) AVPF (green) bound to BIR3-XIAP (mauve) and (b) AoxSPW (gold) bound to BIR3-
XIAP (blue). Hydrogen bonds are shown as red dotted lines. Note that a strand of a b-sheet (between G306 and T308) is shown in detail instead of a
cartoon arrow for clarity. (c) Surface depiction of AoxSPW (gold) bound to BIR3-XIAP (blue). Red and blue represent oxygen and nitrogen,
respectively. Images were rendered using PyMol.20
binding affinities. Overlaid images of the bound ligands
(AVPI overlaid with AoxSPW and AVPI overlaid with
AVPF) derived from the respective superimposed
BIR3-XIAP structures are shown in Figure 2. The
ligands are in the bound conformations and shown with
several surrounding BIR3-XIAP residues. The oxazole,
an aromatic, planar ring, is more sterically constrained
than an amide bond and is therefore expected to induce
some significant shifts in atom positioning relative to
AVPI. Indeed, from the perspective in the plane of the
oxazole ring, the atoms are shifted significantly
(Fig. 2a). The N-terminal amines are 0.80 A apart, caus-
ing the AoxSPW amine to form two hydrogen bonds
˚
with BIR3-XIAP instead of three. The alanine b-methyl

2938
A. D. Wist et al. / Bioorg. Med. Chem. 15 (2007) 2935–2943
Figure 2. Images of the superimposed conformations of (left panel) AoxSPW (gold) and AVPI (pink), and (right panel) AVPI (pink) and AVPF
(turquoise). (a) Perspective from the plane perpendicular to the plane of the proline ring. (b) Perspective from the plane of the proline ring.
Surrounding BIR3-XIAP residues are shown from each structure: AoxSPW (blue), AVPI (yellow), and AVPF (mauve). Several distances between
parallel atoms are labeled as dashed black lines. Hydrogen bonds formed by AVPI are shown as dashed red lines and by AVPF or AoxSPW as
dashed brown lines. Elements are shown as follows: C is gold/pink, O is red, N is blue. Images were rendered as sticks using PyMol and the
superimposed coordinates were produced using XtalView.
˚
groups are 1.03 A apart (measurement not shown), most
likely causing a loss in hydrophobic interaction with
BIR3-XIAP for AoxSPW relative to AVPI. The combi-
nation of orientation of the alanine b-methyl and the
amine to retain some hydrophobic and hydrogen bond-
ing interactions and the steric constraint of the oxazole
causes the Trp-containing portion of AoxSPW to buckle
outward relative to AVPI. This is highlighted by the
BIR3 indole ring amine, whereas the oxazole oxygen
does not. Such small movements may be able to disrupt
hydrogen bonding since the Smac mimic has access to a
more limited conformational space compared to AVPI,
due to the increased steric constraint of the oxazole rel-
ative to an amide bond. On the other hand, BIR3-XIAP
residues are shifted to allow for some new contacts in
the AoxSPW structure. For example, the carbonyl
groups of AVPI and AoxSPW (between Val and Pro
and the oxazole and Pro) are in close proximity
˚
3.15 A distance between the b-methyls of Trp and Ile
from AoxSPW and AVPI, respectively (Fig. 2a). In con-
trast, the two peptide ligands, AVPI and AVPF, map
relatively well onto each other in the binding pocket
(Fig. 2b). The largest displacement is between the
˚
(0.78 A), yet the AoxSPW carbonyl participates in an
extra hydrogen bond with BIR3 T308; therefore move-
ment of BIR3 T308 group was most likely responsible
for this interaction.
˚
b-methyls, which are 1.27 A apart. The movement is
most likely caused by the insertion of the AVPF phenyl
ring between the two nearby BIR3 lysines. The hydro-
phobic interaction is less effective with the branched iso-
butyl group of AVPI, which is more sterically hindered
and less able to properly insert between the two Lys
methylene chains.
2.4. Comparison of BIR3 residue positioning
The protein is not static and residue and/or backbone
movements in the protein also contribute to the overall
binding. From a visual comparison of the positions and
distances in the crystal structures, it is apparent that the
protein is displaced in the AoxSPW structure relative to
the AVPI structure. For example, G304 through L307
participate in b-sheet formation in the AoxSPW struc-
ture while this b-sheet is from G306 through T308 in
the AVPI structure; this causes the nearby loop to be
longer in the AVPI structure (not shown in images).
Thus, several intramolecular hydrogen bonds of BIR3
were switched around in order to accommodate
AoxSPW into the binding pocket.
There are some atoms that map relatively well between
AVPI and AoxSPW, when viewed from the perspective
perpendicular to the plane of the oxazole ring (Fig. 2a).
˚
However, even seemingly small movements (<1 A) cor-
respond to changes in interactions that potentially con-
tribute to overall binding affinity. For example, the
oxazole ring oxygen and the AVPI carbonyl oxygen
˚
are relatively close in space (0.91 A apart); however,
the AVPI oxygen donates a hydrogen bond to the

A. D. Wist et al. / Bioorg. Med. Chem. 15 (2007) 2935–2943
2939
Figure 3. (a) Stereo view of the superimposed conformations of the binding site residues of BIR3-XIAP in the AoxSPW (cyan), AVPI (yellow) or
AVPF (blue–gray) bound states. The space that the ligand occupies is represented as dots, rendered using AoxSPW. (b) Plot of the root-mean-square
difference value for residue positions in the structures of AVPI, AVPF, and AoxSPW bound to BIR3-XIAP. The values for the AVPI structure
superimposed on the AoxSPW structure are blue, AVPI superimposed on AVPF are pink and AVPF to AoxSPW are yellow. The residues closest to
the binding site are highlighted in red. The RMSD values were calculated for residues 280–340 using XtalView. The cartoon image of AoxSPW:BIR3
structure with AoxSPW in dots is displayed for residue number reference (rendered in Pymol.).
There are other more subtle spatial shifts that are nota-
ble between the three structures, depicted in the overlay
images in Figure 3a, which contains several views of the
superimposed BIR3-XIAP structures from the AVPI-,
AVPF-, and AoxSPW-bound states. To examine the
movement of each residue more closely, root-mean-
square difference (RMSD) value for the movement of
each residue was calculated for BIR3 residues in the
AVPI structure with respect to the AoxSPW structure,
the AVPF structure with respect to the AoxSPW struc-
ture, and the AVPI structure with respect to the AVPF
structure. The three sets of values were plotted in on the
same axes, shown in Figure 3b. Proteins are not static in
solution, some movement is ‘natural;’ however, larger
displacements (beyond what is natural RMSD >ꢁ0.4)
were detected and were most likely due to the ligand
binding. Examples of residues that are displaced in the
AoxSPW structure include Pro 312, Ser 313, Glu 314,
and Asp 315. Not all of these residues have a direct
interaction with the ligand, such as a hydrogen bond
or van der Waals. They are most likely displaced in a
compensatory affect to accommodate the ligand in the
binding pocket. Closer examination of the RMSD val-
ues reveals that there may be a distal effect of the
AoxSPW ligand binding. There is high deviation near
BIR3-XIAP 337–340 which is most likely due to an
artifact caused by the relative disorder in this part of
the protein, which is an a-helix that juts out and has
relatively little interaction with the rest of the protein.
Outside of this small cluster of residues and the binding
region, the AVPI- and AVPF-bound structures overlaid
with the AoxSPW-bound structure show about three
times as many significant movements (RMSD >ꢁ0.4)
compared to the AVPI-bound overlaid with the
AVPF-bound structure.
The loss of binding affinity caused by replacing the
N-terminal peptide bond with an oxazole ring can be
explained by the comparisons made between the pep-
tide-bound and mimic-bound structures. Residues
surrounding and distal from the binding site in the pro-
tein are relatively more displaced in the AoxSPW-bound
structure, which is most likely energetically unfavorable.
Some individual interactions between the Smac peptides
and the protein, such as hydrogen bonds and hydropho-
bic contacts, are lost or reduced with AoxSPW. The
U-shaped conformation of bound AoxSPW most likely
costs entropic energy to adopt. From a rational design

2940
A. D. Wist et al. / Bioorg. Med. Chem. 15 (2007) 2935–2943
perspective, synthetic modifications to render the
U-shaped configuration more energetically favorable
would reduce the entropic penalty in adopting such a
configuration and could thereby potentially increase
binding affinity. For example, the pyrolidine ring could
be bridged to the oxazole ring or the indole ring, which
would allow the ligand to access the pocket while intro-
ducing less conformational strain and possibly gaining
some hydrophobic interaction. Also, modifications in
the C-terminal portion of the molecule, which have been
shown by other studies15–18 to increase binding affinity,
may be adaptable to the oxazole scaffold. Overall, the
observations that several BIR3 residues within and sur-
rounding, and distal to the binding site were adjusted to
accommodate AoxSPW relative to the tetrapeptides
demonstrate the flexibility of the Smac-binding groove
in BIR3, and have clear implications in structure func-
tion design for the BIR3-XIAP target and in general.
binding affinity of AVPI, no reduction in cell viability
was observed when cells were dosed with up to
500 lM. Relative to cisplatin, higher concentrations of
the Smac mimic were required to reduce cell viability
(Fig. 4). This result indicates that introduction of the
oxazole moiety enabled cellular entry and most likely
prevented immediate degradation once inside the cell.
3. Conclusions
A Smac mimic, AoxSPW, was characterized for binding
to BIR3-XIAP biochemically and structurally, and for
activity in cancer cells. AoxSPW displayed only moder-
ate affinity for BIR3-XIAP, which was explained by the
structural data; however, AoxSPW shows potentially
more activity in cells compared to its peptide counter-
parts and it is the first Smac mimic where the N-terminal
peptide bond was replaced. By our results, AoxSPW
makes a reasonable scaffold in rational design efforts
toward a completely non-peptidic Smac mimic. The
scope of this study was to expand the structure activity
relationship between the Smac peptides and XIAP to
include modification of the N-terminal peptide bond
and further synthetic rational design was outside of
our aims.
2.5. Cell viability data
In searching for a drug, a primary goal is not only to
optimize binding affinity but also to optimize properties
that allow for membrane permeability and resistance to
degradation in cells or biological fluids. Despite a lower
binding affinity, the peptide mimic may display advanta-
ges relative to the peptides when assayed in cells, from a
presumably better ability to penetrate and/or remain
stable in a cellular context. In fact, previous studies
which examined the effect of dosing cancer cells with
N-terminal Smac peptides showed that, in order to have
any effect, the peptides needed to be fused with a cell
permeating sequence, such as polyarginine or penetra-
tin.21–23 In order to assay the ability of the oxazole-con-
taining Smac mimics to enter cells without fusion to a
cell-penetrating sequence, cell viability assays were per-
formed. AoxSPW caused a reduction in viability of
MBA-MD-231 cells above 30 lM. Despite the higher
4. Experimental
4.1. Chemistry
All protected amino acids were purchased from Calbio-
chem-Novabiochem Corp. (San Diego, CA) or
Advanced Chemtech (Louisville, KY). HATU and
Hunig’s base (N,N-diisopropylethylamine, Biotech
¨
grade, 99.5%) were purchased from Aldrich (Milwau-
kee, WI).
The synthesis of oxazole-containing portion of the pep-
tide mimics has been described previously24 and is
described in Scheme 1. The starting dipeptide (Ala-Ser)
was formed by a standard method using the commercially
available Ala-succinimide ester (Calbiochem-Novabio-
chem) with Boc protection on the amine and methyl ester
protection on the carboxyl group of Ser (NH2-Ser-OMe,
Calbiochem-Novabiochem).
4.1.1. tert-Butyl(S)-1-(4-(methoxycarbonyl)oxazol-2-ine)ethylc-
arbamate (2). The oxazoline was synthesized by adding
1.05 equiv of Burgess’ reagent ((methyl (carboxysulfa-
moyl)triethylammonium hydroxide, Aldrich or Acros)
in one portion to a stirring solution of the Ala-Ser
methyl ester in dry THF under an inert atmosphere.
The resulting solution was then heated at 70 ꢁC for
ꢁ12 h. The THF was evaporated, the resulting oil was
dissolved in EtOAc, and the mixture was washed three
times each with 5% (w/v) aqueous citric acid, saturated
aqueous NaHCO3, and brine. The aqueous layers were
re-extracted with ethyl acetate and the organic layer was
dried with MgSO4. The product was purified by silica
gel flash chromatography in 70% ethyl acetate in
hexanes. The oxazoline, 2, was formed in 38% yield:
Figure 4. MBA-MD-231 (breast cancer) cells were dosed with cisplatin
(black), AoxSPW (gray), AVPI (white), or vehicle at varying concen-
trations for 72 h. At the end of the dosing period the cell viability was
measured using an MTT assay kit and reported relative to control.
Standard errors are for three separate experiments run in triplicate
(with the exception of the cisplatin which was run once in triplicate).

A. D. Wist et al. / Bioorg. Med. Chem. 15 (2007) 2935–2943
2941
4.1.4. Boc and methyl-ester deprotection methods. Prior
to amine deprotection, the methyl ester was stirred at
0 ꢁC in a minimal volume of 3:1 CH3OH:H2O. Three
equivalents of LiOH was added in one portion. After
several hours, the reaction was complete, as indicated
by the TLC spot moving to the baseline in 1:10
CH3OH:CHCl3. The mixture was brought up in EtOAc
and washed three times with 5% (w/v) aqueous citric
acid. The aqueous layer was re-extracted three to five
times with EtOAc and the combined organic layers were
washed once with brine and dried with MgSO4.
To deprotect the amine, a solution of the tert-butyl car-
bamate in CH2Cl2 was cooled to 0 ꢁC. Approximately
three equivalents of TFA was added in one portion.
Upon completion (generally 1–2 h), the reaction mixture
was dried by rotoevaporation. The mixture was re-dis-
solved in CH2Cl2 and re-rotoevaporated. This process
was repeated three times or until the scent of TFA was
no longer detected.
4.1.5. Smac mimic purification and characterization.
Smac mimics and peptides were purified by reversed
phase HPLC (C18 column, Vydac) or by silica gel chro-
matography before deprotection and verified on an ana-
lytical reversed phase column for purity. The HPLC
method used was as follows: Solvent A 99% H2O, 1%
CH3N, 0.1% TFA, Solvent B 90% CH3N, 10% H2O,
0.1% TFA, 10 mL/min (semi-prep) or 1 mL/min (analyt-
ical), C18 Vydac columns, 230 nm detection, gradient of
1% B/min (0.5%/min analytical), t0 = 10% B. Smac mim-
ics were dried by lyophilization after purification, col-
lected as TFA salts, and characterized by mass
Scheme 1. Synthesis of oxazole containing Smac mimics.24,25
Rf = 0.51 (70% EtOAc/hexanes); 1H NMR 400 MHz
(CDCl3) d 1.29–1.46 (s and d overlapping, 12H), 3.69–
3.82 (s, 3H), 4.38–4.60 (m, 2H), 4.67–4.79 (m, 1H),
5.15–5.38 (m, 1H).
4.1.2. tert-Butyl(S)-1-(4-(methoxycarbonyl)oxazol-2-
yl)ethylcarbamate (3). The 4,5 oxazoline bond was
oxidized with BrCCl3 and DBU (diazobicyclounde-
cane) as base.25 A solution of 2 in CH2Cl2 was cooled
to 0 ꢁC and stirred under an inert atmosphere. 1.1
equiv of DBU was added in one portion, then 1.3
equiv BrCCl3 was added dropwise over ꢁ10 min.
The mixture was washed twice with aqueous saturated
NH4Cl, the aqueous layer was re-extracted with
EtOAc, and the organic layer was dried with MgSO4.
The oxazole was purified by silica gel chromatography
in 30–50% EtOAc/hexanes. The oxazole, 3, was ob-
tained in 73% yield: Rf: 0.37 (1:1 EtOAc/hex); 1H
NMR (CDCl3) d 1.38–1.49 (s, 9H), 1.51–1.58 (d,
3H), 3.87–3.96 (s, 3H), 4.91–5.10 (m, 1H), 5.10–5.31
(m, 1H), 8.12–8.22 (s, 1H).
1
spectroscopy and H NMR (400 MHz).
4.1.5.1. NH2-Ala-oxazole-Pro-Phe-OH (Table 1,
entry 1; AoxSPF). White powder. H NMR (300 MHz,
1
D2O) d 1.60, 1.70 (dd, J = 5 Hz, 3H), 1.85–2.25 (m,
3H), 2.90–3.22 (m, 3H), 3.60 (m, 1H), 4.00 (m, 1H),
4.50–4.80 (m, 3H), 5.18 (ad, 1H), 7.18–7.34 (m, 5H),
8.42, 8.45 (ss, 1H). HPLC retention time 32 min
(ramp = 1% B/min). MS m/z 401 (M+H)+. Exact mass
(C20H24O5N4): 400.17442. Mass by HRMS: 400.17442.
4.1.5.2. NH2-Ala-oxazole-Pro-Trp-OH (Table 1, entry
2; AoxSPW). Off-white powder. H NMR (400 mMHz.
1
4.1.3. Coupling of oxazole piece to Pro-X dipeptides. As
in Scheme 1, the oxazole piece, 3, was coupled to the
Pro-X dipeptide or peptide derivative in a 50–100 mg
scale (with respect to the oxazole). The free-acid oxazole
was dissolved in a minimum amount of DMF (<5 mL)
and brought to 0 ꢁC. 1.4 equiv of HATU and 1 equiv
of Hunig’s base were added in one portion and the solu-
tion was allowed to stir for ꢁ30 min to allow formation
of the activated ester. To that solution, the pyrrolidine
derivative (free amine, methyl ester) and 2–3 equiv of
Hunig’s base dissolved in DMF were added. The solu-
tion was stirred allowing warming to RT until the reac-
tion was complete. The reaction was worked up by
diluting 10-fold with ether, washing five times with
saturated NH4Cl and re-extracting the aqueous layer
three to five times. The organic layer was dried with
MgSO4. Yields of four ranged from 40% to 80%.
D2O) d 1.48,1.52 (dd, J = 6 Hz, 3H), 1.52–2.15 (m,
3H), 2.90–3.41 (m, 3H), 4.05 (q, J = 6 Hz, 1H), 4.20–
4.80 (m, 4H), 6.90–7.58 (m, 5H), 8.05, 8.20 (ss, 1H).
HPLC retention time 43 min (ramp = 0.5% B/min).
Exact mass (C22H26O5N5)+: 440.1928. Mass by HRMS:
440.1933.
4.1.5.3. NH2-Ala-oxazole-Pro-Tyr-OH (Table 1, entry
3; AoxSPY). White powder. H NMR (400 MHz. D2O)
1
d 1.43, 1.56 (dd, J = 7 Hz, 3H), 1.62–1.82 (m, 2H),
1.97–2.15 (m, 1H), 2.62–3.08 (m, 3H), 3.45 (m, 1H),
3.78 (ad, 1H), 4.38–4.45 (m, 2H), 4.85 (ad, 1H), 6.68
(dd, J = 4.7 Hz, 2H), 6.96 (dd, J = 8.25 Hz, 2H), 8.05,
8.20 (ss, 1H). HPLC retention time 24 min
(ramp = 0.5% B/min). MS m/z 418 (M+H)+. Exact
mass (C20H24N4O6): 416.16958. Mass by HRMS:
416.16818.

2942
A. D. Wist et al. / Bioorg. Med. Chem. 15 (2007) 2935–2943
Table 2. Data collection and statistics from crystallographic analysis
4.1.5.4. NH2-Ala-oxazole-Pro-Ile-OH (Table 1, entry
4; AoxSPI). Off-white crystalline solid. 1H NMR d
(400 MHz, D2O) d 0.68–0.83 (m, 6H), 0.94–1.38 (m,
1H), 1.52, 1.58 (dd, J = 8 Hz, 3H), 1.60–2.25 (m, 4H),
3.75 (m, 1H),), 3.42–3.60 (m, 1H), 3.98, 4.14 (dd,
J = 6 Hz, 1H), 4.45 (m, 1H), 5.04 (m, 1H), 8.22, 8.24
(ss, 1H). HPLC retention time 40 min (ramp = 0.5%
B/min). MS m/z 367 (M+H)+. Exact mass (C17H26N4
O5): 366.19032. Mass by HRMS: 366.18863.
of AoxSPW bound to BIR3-XIAP
Data set
Space group
Unit cell dimension
P4122
71.2, 71.2, 105.1, 90.0, 90.0, 90.0
˚
Resolution (A)
Total observations
20–2.8
215,384
Unique observations
Data coverage (outer shell)
Rsym (outer shell)
7126
99.9%(100%)
0.076(0.447)
Refinement
Resolution (A)
4.2. Binding assay conditions
˚
20–2.8
R (Rfree
)
0.213 (0.235)
0.007
1.34
Before the compounds were assayed, the solid was
brought up in D2O and the concentration was measured
by 1H NMR with an internal dioxane reference of
known concentration. The assay is a competitive fluo-
rescent-binding assay described previously.19 Briefly,
the dye is comprised of a Smac peptide (AVPC) conju-
gated to a fluorescent dye, BADAN (Molecular Probes,
Invitrogen Corp., Carlsbad, CA) through a thiol-ester
linkage at Cys. When the dye is bound to BIR-XIAP,
the fluorescence emission (centered at 550 nm) is of
higher intensity and a second peak appears centered at
470 nm. As the dye is displaced from the protein by a
competing peptide or peptide mimic the fluorescence
emission is reduced and resembles that of the AVPC-ba-
dan dye in solution alone. The AUC for the emission
spectra of AVPC-badan alone, AVPC-badan, and for
the Smac mimic bound to BIR3-XIAP are used to calcu-
late a relative KD for the Smac mimic. The relative KD is
then converted to an absolute KD using the KD value for
AVPC-badan, which was determined in titration
assay.19
Rmsd bonds
Rmsd angles
Total number of reflections used 6878
Working set
Test set
Solvent
6502
376
75.00%
64
Water molecules
Zn-MAD data sets were collected at the Brookhaven National
Laboratory.
Rsym ¼ RhRijIh;i ꢀ Ihj=RhRiIh;i, where Ih is the mean intensity of the i
observations
of
symmetry
related
reflections
of
h.
P
P
R ¼ jF obs ꢀ F calcj=
F obs, where Fobs = FP, and Fcalc is the calcu-
lated protein structure factor from the atomic model (Rfree was cal-
culated with 5% of the reflections). Rmsd in bond lengths and angles
are the deviations from ideal values, and the rmsd deviation in B
factors is calculated between bonded atoms.
Table 3. Data collection and statistics from crystallographic analysis
of AVPF bound to BIR3-XIAP
Data set
Space group
Unit cell dimension
I213
170.4, 170.4, 170.4, 90.0, 90.0, 90.0
4.3. Crystallization and data collection
˚
Resolution (A)
50–3.0
276,259
15,951
Crystals were grown by the hanging-drop vapor-diffu-
sion method by mixing an equal volume of reservoir
solution. The reservoir contained 0.1 M Hepes, pH
7.5, and 4.3 M NaCl for AoxSPW with BIR3-XIAP
and 0.1 M MES, pH 6.5; 1.6 M magnesium sulfate hep-
tahydrate for AVPF with BIR3-XIAP. The BIR3-XIAP
constructs included residues 249–354 for AoxSPW and
249–357 for AVPF. The peptide or Smac mimic was
co-crystallized with the protein in 10:1 molar excess of
AVPF to BIR3-XIAP or 25:1 molar excess of AoxSPW
to BIR3-XIAP. Cryobuffer was the same as the reservoir
buffer but contained 10–15% glycerol. The crystallo-
graphic data and refinement statistics are shown in
Table 2 for AoxSPW:BIR3 and in Table 3 for AVPF:-
BIR3. The structures can be found in the Protein Data
Bank (PDBID 2OPY AoxSPW:BIR3 and 2OPZ
AVPF:BIR3).
Total observations
Unique observations
Refinement
˚
Resolution(A)
R (Rfree
500.0–3.0
0.231 (0.277)
0.011326
1.78703
)
Rmsd bonds
Rmsd angles
Total number of
reflections used
Working set
Test set
15951
14352
1599
Data sets were collected at the Brookhaven National Laboratory.
Rsym ¼ RhRijIh;i ꢀ Ihj=RhRiIh;i, where Ih is the mean intensity of the i
observations
of
symmetry
related
reflections
of
h.
P
P
R ¼ jF obs ꢀ F calcj=
F obs, where Fobs = FP, and Fcalc is the calcu-
lated protein structure factor from the atomic model (Rfree was cal-
culated with 5% of the reflections). Rmsd in bond lengths and angles
are the deviations from ideal values, and the rmsd deviation in B
factors is calculated between bonded atoms.
4.4. Cell culture and MTT assay
The compound to be tested was dissolved in DMSO at
100 times the highest concentration to be tested. A
1:50 dilution into cellular medium (phenol red-free
DMEM, no antibiotic, 10% fetal bovine serum) was
made to yield a solution that is twice the highest test
dose and 2% DMSO. The dose solution was pipetted
into a tissue culture treated 96-well plate, n = 3 for each
MBA-MD-231 cell line was purchased from the ATCC
(American Type Culture Collection). The cells are cul-
tured in DMEM containing 10% fetal bovine serum,
penicillin, and streptomycin at the recommended doses
and stored in humidified incubators regulated at 37 ꢁC
and 5% CO2. Cells are passed by trypsination once every
5–7 days.

A. D. Wist et al. / Bioorg. Med. Chem. 15 (2007) 2935–2943
2943
H.; Monks, A.; Andreeff, M.; Reed, J. C. Clin. Cancer Res.
2000, 6, 1796–1803.
11. Archakov, A. I.; Govorun, V. M.; Dubanov, A. V.;
Ivanov, Y. D.; Veselovsky, A. V.; Lewi, P.; Janssen, P.
Proteomics 2003, 3, 380–391.
12. Schade, M.; Oschkinat, H. Curr. Opin. Drug Discovery
Dev. 2005, 8, 365–373.
13. Liu, Z. H.; Sun, C. H.; Olejniczak, E. T.; Meadows, R. P.;
Betz, S. F.; Oost, T.; Herrmann, J.; Wu, J. C.; Fesik, S. W.
Nat. Struct. Biol. 2000, 408, 1004–1008.
concentration. Serial 1:2 dilutions were made with cellu-
lar medium containing 2% DMSO. The cells were tryp-
sinized, reconstituted in the cellular medium (DMSO
free), and counted using a hemacytometer. If necessary,
appropriate dilutions were made so that the concentra-
tion became 1 · 105 cells/mL. Fifty microliters of cell
suspension was added to each well (1:2 dilution was
made when cells are added to each well). The dosed cells
are incubated for 72 h. The MTT assay (Promega,
G4000) protocol is followed to obtain cell viability data.
The assay was repeated three times for each cell line
assayed.
14. Wu, G.; Chai, J. J.; Suber, T. L.; Wu, J. W.; Du, C. Y.;
Wang, X. D.; Shi, Y. G. Nat. Struct. Biol. 2000, 408, 1008–
1012.
15. Oost, T. K.; Sun, C. H.; Armstrong, R. C.; Al-
Assaad, A. S.; Betz, S. F.; Deckwerth, T. L.; Ding,
H.; Elmore, S. W.; Olejniczak, E. T.; Oleksijew, A.;
Oltersdorf, T.; Rosenberg, S. H.; Shoemaker, A. R.;
Tomaselli, K. J.; Zou, H.; Fesik, S. W. J. Med. Chem.
2004, 47, 4417–4426.
16. Li, L.; Thomas, R. M.; Suzuki, H.; De Brabander, J.
K.; Wang, X.; Harran, P. G. Science 2004, 305, 1471–
1474.
17. Sun, H.; Nikolovska-Coleska, Z.; Yang, C. Y.; Xu, L.; Liu,
M.; Tomita, Y.; Pan, H.; Yoshioka, Y.; Krajewski, K.;
Roller, P. P.; Wang, S. J. Am. Chem. Soc. 2004, 2004, 51.
18. Zobel, K.; Wang, L.; Varfolomeev, E.; Franklin, M. C.;
Elliott, L. O.; Wallweber, H. J. A.; Okawa, D. C.; Flygare,
J. A.; Vucic, D.; Fairbrother, W. J.; Deshayes, K. ACS
Chem. Biol. 2006, 1, 525–533.
Acknowledgments
The authors acknowledge Dr. E. Shiozaki for assistance
with early crystallographic work. Dr. M. Semmelhack
and Dr. M. Zhou for indispensable advice and critical
reading of the manuscript. Dr. I. Pelczer for assistance
with NMR spectroscopy, and Drs. D. Little and J.
Eng for mass spectroscopy.
References and notes
1. Green, D. R.; Reed, J. C. Science 1998, 281, 1309–1312.
2. Shi, Y. G. Nat. Struct. Biol. 2001, 8, 394–401.
3. Thompson, C. B. Science. 1995, 267, 1456–1462.
4. Deveraux, Q. L.; Reed, T. C. Gene Dev. 1999, 13, 239–252.
5. LaCasse, E. C.; Baird, S.; Korneluk, R. G.; MacKenzie,
A. E. Oncogene 1998, 17, 3247–3259.
6. Du, C. Y.; Fang, M.; Li, Y. C.; Li, L.; Wang, X. D. Cell
2000, 102, 33–42.
7. Verhagen, A. M.; Ekert, P. G.; Pakusch, M.; Silke, J.;
Connolly, L. M.; Reid, G. E.; Moritz, R. L.; Simpson, R.
J.; Vaux, D. L. Cell 2000, 102, 43–53.
8. Ekert, P. G.; Silke, J.; Hawkins, C. J.; Verhagen, A. M.;
Vaux, D. L. J. Cell Biol. 2001, 152, 483–490.
9. Srinivasula, S. M.; Hegde, R.; Saleh, A.; Datta, P.;
Shiozaki, E.; Chai, J. J.; Lee, R. A.; Robbins, P. D.;
Fernandes-Alnemri, T.; Shi, Y. G.; Alnemri, E. S. Nat.
Struct. Biol. 2001, 410, 112–116.
19. Kipp, R. A.; Case, M. A.; Wist, A. D.; Cresson, C. M.;
Carrell, M.; Griner, E.; Wiita, A.; Albiniak, P. A.; Chai, J.
J.; Shi, Y. G.; Semmelhack, M. F.; McLendon, G. L.
Biochemistry 2002, 41, 7344–7349.
20. DeLano, W. The PyMol Molecular Graphics System.
DeLano Scientific LLC: San Carlos, CA, USA, <http://
www.pymol.org.>.
21. Arnt, C. R.; Chiorean, M. V.; Heldebrant, M. V.; Gores,
G. J.; Kaufmann, S. H. J. Biol. Chem. 2002, 277, 44236–
44243.
22. Fulda, S.; Wick, W.; Weller, M.; Debatin, K. M. Nat.
Med. 2002, 8, 808–815.
23. Yang, L. L.; Mashima, T.; Sato, S.; Mochizuki, M.;
Sakamoto, H.; Yamori, T.; Oh-hara, T.; Tsuruo, T.
Cancer Res. 2003, 63, 831–837.
24. Wipf, P.; Miller, C. P. Tetrahedron Lett. 1992, 33, 907–
910.
10. Tamm, I.; Kornblau, S. M.; Segall, H.; Krajewski, S.;
Welsh, K.; Kitada, S.; Scudiero, D. A.; Tudor, G.; Qui, Y.
25. Williams, D. R.; Lowder, P. D.; Gu, Y.; Brooks, D. A.
Tetrahedron Lett. 1997, 38, 331–334.