Requirements for Specific Binding of Low Affinity Inhibitor Fragments to the SH2 Domain of pp60Src Are Identical to Those for High Affinity Binding of Full Length Inhibitors

Article abstract of DOI:10.1021/jm020970s

Results from a novel approach which uses protein crystallography for the screening of a low affinity inhibitor fragment library are analyzed by comparing the X-ray structures with bound fragments to the structures with the corresponding full length inhibi

Full text of DOI:10.1021/jm020970s

5184
J . Med. Chem. 2003, 46, 5184-5195
Requ ir em en ts for Sp ecific Bin d in g of Low Affin ity In h ibitor F r a gm en ts to th e
SH2 Dom a in of p p 60Sr c Ar e Id en tica l to Th ose for High Affin ity Bin d in g of F u ll
Len gth In h ibitor s
Gudrun Lange,*†,§ Dominique Lesuisse,| Pierre Deprez,‡ Bernard Schoot,‡ Petra Loenze,† Didier Be´nard,‡
J ean-Pierre Marquette,‡ Pierre Broto,‡ Edoardo Sarubbi,‡ and Eliane Mandine‡
Aventis Pharma, 102 route de Noisy, 93235 Romainville, France, and Aventis Pharma, Structural Biology,
65926 Frankfurt, Germany
Received J une 18, 2002
Results from a novel approach which uses protein crystallography for the screening of a low
affinity inhibitor fragment library are analyzed by comparing the X-ray structures with bound
fragments to the structures with the corresponding full length inhibitors. The screen for new
phospho-tyrosine mimics binding to the SH2 domain of pp60src was initiated because of the
limited cell penetration of phosphates. Fragments in our library typically had between 6 and
30 atoms and included compounds which had either millimolar activity in a Biacore assay or
were suggested by the ab initio design program LUDI but had no measurable affinity. All
identified fragments were located in the phospho-tyrosine pocket. The most promising fragments
were successfully used to replace the phospho-tyrosine and resulted in novel nonpeptidic high
affinity inhibitors. The significant diversity of successful fragments is reflected in the high
flexibility of the phospho-tyrosine pocket. Comparison of the X-ray structures shows that the
presence of the H-bond acceptors and not their relative position within the pharmacophore
are essential for fragment binding and/or high affinity binding of full length inhibitors. The
X-ray data show that the fragments are recognized by forming a complex H-bond network
within the phospho-tyrosine pocket of SH2. No fragment structure was found in which this
H-bond network was incomplete, and any uncompensated H-bond within the H-bond network
leads to a significant decrease in the affinity of full length inhibitors. No correlation between
affinity and fragment binding was found for these polar fragments and hence affinity-based
screening would have overlooked some interesting starting points for inhibitor design. In
contrast, we were unable to identify electron density for hydrophobic fragments, confirming
that hydrophobic interactions are important for inhibitor affinity but of minor importance for
ligand recognition. Our results suggest that a screening approach using protein crystallography
is particularly useful to identify universal fragments for the conserved hydrophilic recognition
sites found in target families such as SH2 domains, phosphatases, kinases, proteases, and
esterases.
In tr od u ction
which scan a putative binding site using probes repre-
senting H-bond donors/acceptors or hydrophobic ele-
ments in order to identify H-bond partner positions in
the binding pocket and evaluate the size of hydrophobic
pockets. Somewhat between, there are programs such
as LUDI5 which selects fragments from a virtual
database and places them into the binding pocket.
Connecting these fragments ideally results in a full
length inhibitor. However, these approaches have not
always been successful, possibly due to the limited
quality of potentials and scoring functions used. In
addition, the flexibility of the protein is not taken into
account since it is unknown to which extent the protein
will adapt to the ligand.
Traditionally, structure-based drug design plays a
major role in the optimization of hits found by high
throughput screening of company-owned libraries or
derived from other sources. More recently it has emerged
that particularly for targets involving protein-protein
interactions, there is a high probability of not finding a
promising hit by high throughput screening,1 and
alternative approaches for lead generation are required.
With no hits available, computer programs have been
used to generate leads for structure-based drug design.
These approaches range from programs such as DOCK2
and FLExX3 which dock full length inhibitors into
putative binding sites and programs such as GRID4
So far, only a few experimental protein structure-
based approaches for lead generation have been re-
ported. They include NMR measurement of the target
protein mixed with inhibitor fragments.6,7 On the basis
of the difference in the NMR spectra of free and
complexed protein, inhibitor fragments are identified
which bind to pockets in the target protein (SAR by
NMR). Recently, it has been demonstrated that screen-
* To whom correspondence should be addressed: Industriepark
Ho¨chst, Building G837, Tel.: +49-69-30516207, Fax: +49-69-305-
17768, Email: gudrun.lange@Bayercropscience.com.
† Aventis Pharma, Frankfurt, Germany.
‡ Aventis Pharma, Romainville, France.
§ Present address: Bayer CropScience, Chemistry, 65926 Frankfurt,
Germany.
| Present address: Aventis Research Centre, 13 Quai J ules Guesde,
94403 Vitry, France.
10.1021/jm020970s CCC: $25.00 © 2003 American Chemical Society
Published on Web 10/28/2003

Binding to the SH2 Domain of pp60Src
J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 24 5185
ing of a fragment library using protein crystallography
can be used for lead generation.8,9 Inhibitor fragments
are soaked into the binding pocket of the crystallized
protein andsin case of specific binding-the correspond-
ing X-ray structures are determined. Starting from the
high-resolution structure of a protein-fragment com-
plex, a full length inhibitor can be obtained by attaching
an inhibitor scaffold designed by structure-based drug
design. Alternatively, libraries can be screened to
replace fragments in already existing full length inhibi-
tors.9 These replacements can be selected in order to
modify a variety of features such as binding affinity,
cell penetration, solubility, or selectivity. However,
using protein crystallography only a limited number of
compounds can be investigated and thus either com-
pound cocktails were used8 or affinity-based filtering
techniques9 were applied.
Understanding the key principles according to which
fragments bind to proteins would allow us to limit the
experimental screening to fragment candidates which
have a high likelihood of binding to the protein. A
comparison between X-ray structures with bound frag-
ments to those with full length inhibitors should allow
us to extract some general rules according to which low
affinity fragments bind to proteins. In addition, the
superposition of a multitude of X-ray structures with
different inhibitor fragments would reveal at which
positions within the binding site specific functional
groups such as H-bond donors/acceptors or hydrophobic
groups are required. If a binding site is rigid, the
positions of functional groups should form well-defined
clusters in such a superposition of experimental struc-
tures. However, for flexible ligand binding sites, there
is additional information encoded in the size and shape
of the clusters for a given functional group. The topology
of the individual clusters and their relative position will
reflect the flexibility of the protein and show to what
extent the protein can stretch itself in order to accom-
modate a ligand into the binding site.
compound solution and prior to data collection with 10%
of glycerol as cryoprotectant. Most data were collected
at -170 °C on either a Mar345 imaging system or a
MarCCD mounted on a rotating anode X-ray generator
(GX21, Enraf Nonius). Some data were collected at
beam-lines at the ESRF or Lure. The data were pro-
cessed using XDS16 and refined using X-PLOR17 based
on a model provided by Ariad. Model building and
structure comparison was carried out using Quanta.18
Prior to their comparison, all structures were superim-
posed using LSQKAB.19 The LUDI5 predictions were
based on 1shd20 and calculated using standard param-
eters. The compounds including the most important
details of the data collection and structure refinement
and their IC50 value are listed in Charts 1 and 2. The
IC50 values of compounds 1-10 were measured as
described earlier using either Surface Plasmon Reso-
nance (SPR) or Scintillation Proximity Assay.9 The
X-ray structures (PDB codes 1o41, 1o42, 1o43, 1o44,
1o45, 1o46, 1o47, 1o48, 1o49, 1o4A, 1o4B, 1o4C, 1o4D,
1o4E, 1o4F, 1o4G, 1o4H, 1o4I, 1o4J , 1o4K, 1o4L, 1o4M,
1o4N, 1o4O, 1o4P, 1o4Q,1o4R,1o4S) have been deposited
at the Protein Data Bank.
Syn th eses. The synthesis of compound 1 is described
in reference.13 The synthesis of compounds 2, 3, and 6
is described in reference.9 The synthesis of compounds
5, 7, and 8 is described in reference.21,22 Compounds 4
and 9 have not been published, and their synthesis will
be reported here.
Starting malonate 11 was fluorinated using N-fluo-
robenzenesulfonimide in the presence of potassium
hexamethyldisilyl azide at low temperature. The result-
ing fluoro-malonate 12 was condensed with methyl-2-
acetamidoacrylate using Heck conditions yielding the
ester 13a , which was saponified using lithium hydroxide
as a base to generate the acid 13b. Condensation of the
resulting dehydrotyrosine analogue 13b with the aze-
pinone 14 afforded 15, which was subsequently hydro-
genated on palladium and saponified using formic acid
to yield 4 (Scheme 1).
Herein, we analyze our results of the screening of
inhibitor fragments9 using a combination of a Biacore-
based assay, computer predictions, and protein crystal-
lography for the design of inhibitors of the SH2 domain
of pp60src. On the basis of this analysis, more general
requirements for the binding of low affinity fragments
and thus for a successful application of this novel
methods will be proposed. pp60src kinase has been
implicated in bone resorption and represents a potential
target for therapeutic intervention of osteoporosis.10
Crystals structures of peptides show that there are two
major binding sites: a positively charged phospho-
tyrosine binding site and the hydrophobic pY+3 pocket.11
So far, few inhibitors have been described which com-
bine a nonpeptidic character with an affinity comparable
to the natural substrate pYEEI. Combining the ap-
proach described above with traditional structure-based
lead optimization12-15 has resulted within little more
than a year in nonpeptidic inhibitors with IC50 values
in the low nanomolar range.
The same azepinone 14 was condensed with 3-[3,4-
bis(diethoxyphosphoryl)phenyl-2-tert-butoxycarbony-
laminopropionic acid 17 synthesized according to W.
Shakespeare23 to afford the diphosphono derivative 18.
The latter was deprotected using trifluoroacetic acid
yielding the amine 19a which was subsequently acety-
lated in pyridine to give the acetamide 19b. The
resulting diphosphono ester 19b was deprotected using
trimethylsilyl iodide to provide the final diphosphonic
acid 9 (Scheme 2).
Resu lts
A fragment library was generated which included 150
compounds typically consisting of 6-30 atoms. The
compounds were either commercially available or were
part of small libraries prepared by parallel synthesis.
To reduce the number of compounds to a feasible
amount for the crystallographic experiments, a Biacore
assay was used as a filter. Phenyl phosphate, the
smallest fragment of the substrate pYEEI with a
detectable binding affinity for Src SH2, was used as a
reference. Fragments that displayed comparable or
better binding affinity than phenyl phosphate were kept
for soaking. In addition, fragments suggested by the
Ma ter ia ls a n d Meth od s
P r otein Cr ysta llogr a p h y. Human pp60src SH2 was
purified and crystallized as described earlier.9 The
crystals were soaked overnight with the corresponding

5186 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 24
Ch a r t 1. Chemical Structure and IC50 Values of Fragments
Lange et al.
program LUDI were included. These fragments showed
no biochemical activity within the range measured. The
citrate fragment was found bound to the protein after
a citrate buffer had been used during purification and
displayed no measurable affinity. The successfully
soaked fragments including their IC50 values are listed
in Chart 1.
had a fairly good chance to bind in a specific binding
mode. Except for phenyl phosphate which was clearly
bound at two sites, all other fragments were bound only
in the phospho-tyrosine pocket and are, therefore,
phospho-tyrosine mimics. The fragments identified dif-
fered significantly in their chemical nature and include
oxalate, malonate, phenylmalonate, sulfates, and vari-
ous compounds from an aldehyde library. We did not
observe well-defined electron density for the more
hydrophobic fragments predicted by LUDI which were
tried in order to identify inhibitor fragments pointing
into the hydrophobic pY+3 pocket. This may be due to
either the limited solubility of the compounds and/or an
undefined positioning within the pocket. No correlation
between affinity and fragment binding was found for
the polar ligands. Two fragments out of six suggested
by LUDI were found to bind in the corresponding
soaking experiments.9 However, the predicted binding
modes for oxalate and malonate was not confirmed in
the experiments. Figure 1 shows the electron density
of a selection of nine fragments binding into the phos-
pho-tyrosine pocket. The maps were calculated using
only coordinates of the protein model and omitting the
coordinates of the fragments (omit maps).
Soaking of src SH2 crystals was tried for about 200
compounds including fragments and full length inhibi-
tors. Since soaking is a fast but not always reliable
method, phenyl phosphate was used as an internal
standard. A total of 200 X-ray data sets including both
fragments and full length inhibitors were collected and
analyzed resulting in a total of 45 structures with bound
inhibitor fragments or full length inhibitors. They
included two structures with fragments predicted by
LUDI and 14 structures with fragments found using the
Biacore assay. All X-ray structures were better than 2.1
Å and refined to R-factors below 20%. The electron
density was for all structures well defined, though some
compounds displayed limited local disorder.
Our general experience was that soluble fragments
with a substantial number of putative H-bond partners

Binding to the SH2 Domain of pp60Src
J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 24 5187
Ch a r t 2. Chemical Structure and IC50 Values of
Inhibitors
36-42. The difference of 20° between the chi-values of
S36OG found in the oxalate and the malonate structure
results in the considerable differences in loop conforma-
tion described above. The observed differences for cor-
responding cR position in the malonate and oxalate
structure are 0.45 Å for T37, 1.69 Å for T38, and 0.39 Å
for N39 which is well above the coordinate error. For
other ligands, the observed conformation of this loop
does not dependent on the size or the affinity of the
particular ligand. However, it seems that a loop con-
formation is assumed which allows the system to
complete the H-bond network, and thus the loop con-
formation depends on the relative positions of H-bond
partners within the ligand.
The requirements for specific binding which need to
be met by putative millimolar fragments can be ex-
tracted from the superposition of all fragment struc-
tures. The superposition of the fragment H-bond accep-
tors (Figure 2c) based on the structure with bound
citrate shows that these acceptors form well-defined
clusters when interacting with rigid H-bond donors such
as R34. However, the cluster formed by the H-bond
acceptors to S36OG is quite diffuse, indicating that the
exact location of the H-bond partner of S36OG within
the phospho-tyrosine pockets is less crucial. The dis-
tances of up to 3 Å between individual cluster members
reflect the extent of the hinge movement which can be
exercised by the side chain of S36. A closer inspection
of the fragment binding reveals that in all cases complex
H-bond networks are formed between the fragments and
SH2 as seen in Figure 2d for the citrate structure. All
networks are identical in the rigid part of the pocket
and include H-bonds/salt-bridges between the fragments
and the side chains of R14, R34, and the backbone
amide of E37. In addition, an H-bond between the
fragments and S36OG is present in all structures. The
details of the individual networks involving the more
flexible protein H-bond partners such as T38OG, T39OG,
K62NH2, and the backbone amide of T38 differ some-
what and include in some cases a well defined water
molecule. However, there are no uncompensated H-
bonds neither within the protein nor within the frag-
ment. Hydrophobic interactions seem to be of minor
importance at this site since fragments such as citrate
and malonate, which have no hydrophobic groups, were
unambiguously bound to the protein. However, if dif-
ferent binding modes are accessible to the fragment,
additional hydrophobic interactions influence which
mode is assumed. For example, the phenyl phosphate
fragment binds in a different mode to the phospho-
tyrosine in the peptide structures. In both modes the
requirements with respect to the H-bond network are
fully met but additional hydrophobic interactions
inaccessible in the peptide complexessfavor an alter-
nate binding mode of the phenyl phosphate fragment.
Both binding modes have been found subsequently in
complex structures with full length inhibitors.15
The superposition of the fragment structures shows
that the protein in all structures superimposes excellent
with rms deviations of typically 0.2 Å for most main
chain atoms. However, closer inspection reveals that
there are significant differences in the phospho-tyrosine
pocket (Figure 2a). The conformational changes induced
upon binding of the individual fragments result in
differences of up to 3 Å for the position of identical
protein atoms in different structures. For instance,
comparison of the oxalate and malonate structures
shows that both fragments interact in an identical
manner. In both structures, one carboxylate moiety
interacts with the side chains of R14 and R34 and the
main chain nitrogen of E37 while the second carboxylate
moiety forms H-bonds to the side chains of S36, T38,
and the main chain nitrogen of T38 (Figure 2b). The
positions of protein H-bond donors (R14NH2, R34NH2,
and E37N) interacting with the first carboxylate moiety
are identical within the experimental error. However,
the positions of the protein H-bond donors interacting
with the second carboxylate moiety differ significantly
in both structures i.e., 0.6 Å for S36OG, 1.1 Å for T38N,
and 2.6 Å for T36OG.
A closer analysis of the superimposed structures
shows that the loop consisting of amino acids 36-42
displays a significant flexibility. The structure of this
loop is stabilized by H-bonds between the backbone
amides of S36 and A42 and backbone carbonyls of A42
and T39, respectively. In addition, the side chain of S36
interacts via its two free electron pairs with the amide
backbones of T38 and T39 while the proton of S36OG
acts in all fragment structures as an H-bond donor to
the fragment. This means, that a rotation of the serine
side chain induces a conformational change of the loop
In parallel to the identification of fragments pointing
into the phospho-tyrosine pocket, a nonpeptidic inhibitor
containing a phospho-tyrosine unit was designed and
optimized.14 The best phospho-tyrosine inhibitor had an
activity of 9 nM (inhibitor 1). Subsequently, replace-
ments for the phospho-tyrosine were selected based on
the X-ray structures of the fragments. In particular,

5188 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 24
Lange et al.
Sch em e 1
Sch em e 2
different phenyl malonate fragments incorporating the
structural information obtained from the malonate and
citrate complexes were designed and subsequently
tested in the Biacore assay. Replacement of phospho-
tyrosine by 2-carboxy phenyl malonate resulted in an
inhibitor with an IC50 value of 3 nM (inhibitor 2). The
omit-map displaying the electron density for inhibitor
2 is shown in Figure 3a. The structures of the complexes
with inhibitor 1 and 2 including their common inhibitor
atoms superimpose extremely well (Figure 3b). As
expected, there are some changes within the phospho-
tyrosine pocket due to the different induced fit of the
phospho-tyrosine and 2-carboxy phenyl malonate frag-
ment. Comparison of the X-ray structures containing
the fragment with that containing inhibitor 2 (Table 1)
shows that the malonate, citrate, and phenyl malonate
fragments had the same binding mode as the corre-
sponding group in inhibitor 2, indicating that the
fragments were bound in a mode relevant for drug
design (Figure 4a-c). The protein structures including
the H-bond acceptors in inhibitor 2 and in the fragments
superimpose within the experimental error. In particu-
lar, the distances between corresponding H-bond accep-
tors in the structures of inhibitor 2 and citrate were
between 0.2 and 0.4 Å.
Further analysis of complex structures with full
length inhibitors showed that the requirements for
fragment binding are identical to those for high affinity
full length inhibitors. The phospho-tyrosine fragment
in inhibitor 1 was replaced by other putative mimics,
and subsequently the crystal structures of some inter-
esting inhibitors, 3-10, were determined. The affinity
of inhibitors 1-10 differs by several orders in magni-
tude, i.e., between micromolar and picomolar. The
superposition of the fragment and full length inhibitor
structures shows that the common atoms interact
identically within the SH2 domain (Figure 5a). Thus,
the differences in affinity can be correlated with the
different interactions of the inhibitors with the phospho-
tyrosine pocket.
The positions of all H-bond acceptors in full length
inhibitors superimposed based on the protein structure
of inhibitor 2 shows that most of the H-bond acceptors
of the fragments and full length inhibitors are located

Binding to the SH2 Domain of pp60Src
J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 24 5189
F igu r e 1. Omit electron density maps of fragments bound to the SH2 domain of src contoured at 3σ.
within the clusters defined by the fragments structures
(Figure 5b). In particular, the H-bond acceptors of high
affinity inhibitors are located well within these clusters
and form a complete H-bond network with the phospho-
tyrosine pocket. In contrast, in complexes with poor
inhibitors such as 6 or 7 the central H-bond to the side
chain of S36 is not made. The importance of this
particular H-bond is highlighted by the presence of a
second conformation of the phenyl acetic acid moiety in
the complex structure with compound 6. In one confor-
mation, the carboxylic group interacts as expected with
R14 and R34. In the second, much lower occupied
conformation (10-20%), the H-bond to S36OG is made.9
In other complex structures with low affinity inhibitors,
the electron density of the phospho-tyrosine replacement
is disordered even though the electron density of the
common inhibitor part is well defined.15 In these cases,
the H-bond acceptors are not located within the clusters
defined by the fragments and are thus unable to
complete the H-bond network. Examples include inhibi-
tor 9 and 10 with an IC50 of 2000 and 2200 nM.
Discu ssion
The lack of promising hits after high throughput
screening of huge compound libraries has intensified the
interest in fragment-based approaches using computer
algorithms, SAR by NMR or crystallography and coop-
erative combinatorial chemistry for fragment combina-
tion. For instance, it has been demonstrated, that a
combination of two weakly binding fragments from a
combinatorial library can result in a nanomolar inhibi-
tor.24 Recently, it has been shown that a screening
procedure using protein crystallography can identify
new lead compounds and that their X-ray structures
form an excellent basis for the design of high affinity
inhibitors.8,9 In addition, we have shown, that fragment
libraries can be screened in order to identify replace-
ments such as phospho-tyrosine mimics for already
existing inhibitors.9 Compared to NMR, protein crystal-
lography has less high-throughput potential since col-
lecting a complete data set takes in our case about 6 h
at a home source and 10 min at a synchrotron source,
and we have been able to finalize two structures per
day. This lack has been compensated for by either using
a cocktail of compounds8 or by filtering the fragment
library using a biacore assay.9 Also fragments found by
other techniques such as mass spectrometry,25 NMR
experiments,5,26 and in particular computer prediction27
can be used as input for crystallographic screening.
In summary, fragments are bound specific and rep-
resent high affinity phospho-tyrosine mimics if their
H-bond acceptors are located within the limits high-
lighted by the fragment H-bond acceptors and thus able
to complete the H-bond network present in the phospho-
tyrosine pocket.

5190 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 24
Lange et al.
F igu r e 2. (a) Superposition of X-ray structures with fragments bound to the SH2 domain of src. (b) The binding mode of oxalate
(white) and malonate (pink) in the complex with the SH2 domain of src. (c) The superposition of the fragment H-bond acceptors
shows that they form well defined clusters when interacting with rigid protein parts (red) and less well defined clusters when
interacting with flexible protein parts (pink). For easier orientation only the full structure of the citrate complex is displayed. (d)
Citrate is bound in a complex H-bond network to the SH2 domain involving a total of seven H-bonds and salt-bridges indicated
in red. The H-bonds stabilizing the conformation of loop 36-42 are shown in white
In the particular case of SH2, however, purely com-
puter-based approaches have limitations in proposing
new phospho-tyrosine mimics because the phospho-
tyrosine binding pocket displays a high flexibility. Even
in the best case, computational methods could have
picked up only compounds with binding modes charac-
terized by a pharmacophore closely related to the
phosphate when using X-ray structures with phospho-
tyrosine peptides as templates. Surprisingly, two out of
the six fragments proposed by the program LUDI were
predicted correctly. The binding of oxalate and malonate
was successfully predicted using a structure containing
a phospho-tyrosine peptide. However, only the salt-
bridge to the rigid arginines 14 and 34 was correctly
proposed while the interactions with the flexible loop
were not correctly predicted. In addition, the superposi-
tion of the X-ray structures suggests that the positions
of the H-bond acceptors are not as well defined as a
program such as Grid would have predicted and that
the distances between functional groups in the phar-
macophor model are not as restrained as usually is
assumed in computer-based approaches.
Our experiments show that neither ligand affinity nor
the exact positioning of functional groups within the
pharmacophore are an essential requirement for frag-
ment binding and thus good selection criteria for polar
fragments to the SH2 domain. However, it is important
that the fragments complete the H-bond network within
the phospho-tyrosine pocket. The importance of the
H-bond network is highlighted by the significantly lower

Binding to the SH2 Domain of pp60Src
J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 24 5191
F igu r e 3. (a) Omit electron density map of inhibitor 2 contoured at 3σ. (b) Superposition of the X-ray structures of inhibitors
1(white) and 2 (pink) bound to the SH2 domain.
to inhibitor 230 and bisphosphonates23 maintained their
Ta ble 1. X-ray Data for Complexes of the SH2 Domain
activity compared to their phospho-tyrosine analogues
if the correct scaffold is attached. Both compound classes
are able to complete the H-bond network. The bispho-
sphonates were even designed to mimic the interaction
of the citrate bound in the phospho-tyrosine pocket.23
compound
resolution [nm]
Rmerge [%]
Rfactor [%]
With Fragments
fragment 1
fragment 2
fragment 3
fragment 4
fragment 5
fragment 6
fragment 7
fragment 8
fragment 9
fragment 10
fragment 11
fragment 12
fragment 13
fragment 14
fragment 15
fragment 16
fragment 17
fragment 18
0.180
0.155
0.160
0.160
0.170
0.190
0.170
0.150
0.170
0.185
0.200
0.200
0.155
0.225
0.175
0.165
0.170
0.155
8.6
4.8
9.5
3.5
6.8
9.8
4.4
3.2
6.5
8.1
5.6
9.9
5.0 19.9
6.7 15.6
4.0 20.7
5.0 20.7
6.1 18.7
4.2 19.3
17.7
19.1
18.2
20.9
20.2
16.9
17.9
19.0
18.7
17.6
18.5
16.4
Even though complex H-bond networks between
inhibitors and proteins have been extensively described
in the literature, the number of made H-bonds was
emphasized rather than the number of uncompensated
H-bonds. However, the importance of a complete H-
bonding network has been highlighted for other biologi-
cal systems. It has been shown that there is a correla-
tion between uncompensated protein H-bonds and the
buried surface area of hydrophobic atoms in high
resolution X-ray structures. This correlation was suc-
cessfully used to predict the stability and structure of
proteins.31 Recently, a term was introduced in a scoring
function for protein-protein interactions which penal-
izes buried charged groups that do not form stabilizing
hydrogen bonds.32 The importance of a complete H-bond
network in nucleotide recognition has been investigated.
It was proposed that the correct codon:anticodon du-
plexes are those whose formation and interaction with
the ribosomal decoding center are not accompanied by
uncompensated losses of H-bonds.33
With Inhibitors
0.170
0.170
0.180
0.200
0.180
0.155
0.150
0.150
inhibitor 1
inhibitor 2
inhibitor 3
inhibitor 4
inhibitor 5
inhibitor 6
inhibitor 7
inhibitor 8
inhibitor 9
inhibitor 10
6.8
4.5
7.7
7.7
7.7
5.4
5.5
5.1
5.0
5.5
19.2
19.0
20.0
19.2
19.2
19.4
19.6
20.2
19.6
19.6
0.185
0.150
The experiments described above have limitations in
identifying hydrophobic fragments. Hydrophobic inter-
actions are nondirectional and fragments not exactly
matching the shape of the pocket can bind in a multi-
tude of different conformations unless they have a
functional group which form directional interactions
such as H-bonds with the protein. For this and experi-
mental reasons such as limited solubility, it seems that
it might be difficult to identify hydrophobic fragments
and their binding modes using protein crystallography.
affinity of full length inhibitors which form incomplete
H-bond networks within the phospho-tyrosine pocket.
This is also true for the phospho-tyrosine mimics
described in the literature such as carboxymethylphe-
nylalanine28 which lack a H-bond acceptor to S36OG.
Benzylmalonates29 do not seem to be able to position
their H-bond acceptors within the required limits.
Recently, it was shown, that phenylmalonates related

5192 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 24
Lange et al.
F igu r e 5. (a) Superposition of all X-ray structures of 18
fragments and 10 full length inhibitors based on inhibitor 1.
(b) Superposition of the H-bond acceptors of the 18 fragments
and 10 full length inhibitors show that most acceptors are
located within the clusters defined by the fragments shown
in Figure 2a. The H-bond acceptor positions of the low affinity
H-bond inhibitors 9 and 10 are indicated in purple. For easier
orientation, only the protein structure of the complex with
inhibitor 12 is displayed.
conserved recognition sites. Recognition fragments iden-
tified for one particular family member should have,
therefore, a significant probability to fit into the recog-
nition sites of other family members. Currently, there
are more than 200 sequences of human SH2 domains
in the SMART database.34 They represent putative
targets implicated in breast and colon cancer, T-cell
activation, and osteoporosis. Some of the phospho-
tyrosine mimics described here will only bind if the
flexibility of loop 36-42 induced by the side chain of
S36 is present. Sequence alignment shows that these
residues belong to the most highly conserved residues
F igu r e 4. (a) Superposition of the X-ray structures of inhibitor
2 (pink) and (a) fragment 3 (malonate), (b) fragment
(phenylmalonate) and (c) fragment 2 (citrate) bound to the SH2
6
domain of src.
Our results suggest that a screen using protein
crystallography might be best used in identifying inhibi-
tor fragments for hydrophilic recognition sites. These
include the phosphate binding sites in phosphatases or
SH2 domains, the catalytic triad of esterases and
proteases and the ATP binding site in kinases. Family
members sharing the same fold usually have highly

Binding to the SH2 Domain of pp60Src
J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 24 5193
within the SH2 domains,34 indicating that a substantial
number of other SH2 domains should display similar
flexibility and bind the fragments identified for the src-
SH2 domain. Other structures of SH2 domains depos-
ited in the Brookhaven database20 confirm that the
structure of the recognition site is highly conserved.
Examples include: the SH2 domains of tyrosine phos-
phatases such as syp (1aya) and shp-2 (2shp), kinases
such as tyrosine kinase Fyn and cell kinase Hck (1adp5),
and other proteins such as the xlp protein sap (1diz)
which has been proposed to regulate the recruitment
of SH2 signaling proteins to specific docking sites.
The huge number of new targets coming from genom-
ics has made the fast turnover of projects into a
competitive issue within the pharmaceutical industry.
The recycling of general recognition fragments for a
particular protein family identified by a protein crystal-
lography-based screening should result in a significant
speed-up of the lead generation and optimization pro-
cess.
2-acetamidoacrylate was added to the mixture followed by 0.61
mL (4.35 mmol) of triethylamine, 45.15 mg (0.20 mmol) of Pd-
(OAc)2, and 0.105 mg (0.348 mmol) of tri-o-tolylphosphine. The
suspension was refluxed for 17 h and then brought to room
temperature. The suspension was poured into a saturated
aqueous solution of NaHCO3 and extracted with EtOAc. The
organic extracts were washed with brine, dried, and evapo-
rated to dryness to afford 2.65 g of crude 13a as yellow oil.
Flash chromatography (265 g silica, CH2Cl2/MeOH, 97.5/2.5)
gave 0.317 g (18%) 13a as an orange oil (mixture Z/ E 50/50).
Rf ) 0.19 (SiO2F254Merck60 CH2Cl2/MeOH, 97.5/2.5); NMR
(CDCl3), δ 1.52 (s, 18H), 2.16 (s, 1H), 3.84 (s, 3H), 3.87 (s, 3H),
7.07 (1s, 1H), 7.37 (1s, 1H), 7.40 (d, 1H), 7.54 (dl, 1H), 7.95
(sl, 1H). Compound 13a (0.3 g, 0.6 mmol) was dissolved in 12.2
mL THF and treated with a solution of 32.5 mg (0.774 mmol)
of LiOH.H2O in 4 mL of H2O. The resulting yellow solution
was stirred for 18 h at room temperature and then poured into
25 mL of H2O and 10 mL of 1 M HCl and extracted with
EtOAc. The organic extracts were washed with brine, dried,
and concentrated under vacuum to afford 0.265 g of 13b as a
cream-colored foam (90%). Rf ) 0.14 (SiO2F254Merck60 CH2-
Cl2/MeOH, 90/10);MS m/z 494 (M - H)-, 518 (M + Na)+.
F lu or op r op a n ed ioic Acid , [4-[2-(Acetyla m in o)-3-[[(3S)-
1-[[(1,1-biph en yl)-4-yl]m eth yl-h exah ydr o-2-oxo-1H-azepin -
3-yl]a m in o]-3-oxo-1-p r op en yl]-2-(m eth oxyca r bon yl)p h e-
n yl]-, Bis(1,1-d im eth yleth yl) Ester (15). An amount of 0.265
g (0.535 mmol) 13b was dissolved in 7.6 mL of CH2Cl2 and
2.55 mL of DMF in a 50 mL flask under argon and cooled in
an ice bath. An amount of 0.102 g (0.535 mmol) of 1-(3-
dimethylaminopropyl)-3-ethylcarbodiimide hydrochloride (EDCI)
was added followed by 0.072 mg (0.535 mmol) of 1-hydroxy-
benzotriazole (HOBT). The solution was stirred at room
temperature for 15 min and then added dropwise to a cooled
solution of 0.232 mg (0.535 mmol) of amine (3S)-amino-1-
[[(1,1′-biphenyl)-4-yl]methyl]hexahydro-2H- azepin-2-one 1413
and 0.185 mL (1.07 mmol) of N,N-diisopropylethylamine in 5
mL of CH2Cl2 (0 °C). The resulting solution was stirred for 5
Exp er im en ta l Section
Gen er a l. Analytical data were recorded for the compounds
described using the following general procedures. Proton NMR
spectra were recorded on a Bruker AC 300 MHz spectrometer,
and chemicals shifts were recorded in ppm (δ) from an internal
tetramethylsilane standard in dimethyl sulfoxide-d6 unless
otherwise specified. Coupling constants (J ) were recorded in
hertz. Mass spectra (MS) were recorded on an MS platform
(micromass) and MS Autospec TOF (micromass) in electron
impact, electrospray, and chemical ionization modes. Flash
chromatography was performed on silica gel 60H (Merck) using
the solvent systems indicated below. For mixed-solvent sys-
tems, the volume ratios are given. Analytical HPLC was
performed on a Merck HPLC system equipped with a Merck
Hitachi L-4000A UV detector and reversed-phase C18 column,
kromasil 5 µm. All reactions were performed under a nitrogen
atmosphere using magnetic stirring. Reactions were monitored
by thin-layer chromatography (TLC) on precoated plates of
silica gel 60F254 (layer thickness, 0.25 mm; E. Merck, Darm-
stadt). Anhydrous magnesium sulfate (MgSO4) was used
routinely to dry the combined organic layers from extractions.
Solvent was routinely removed in vacuo using rotary evapora-
tor followed by evacuation with a vacuum pump. Methyl
2-acetamidoacrylate was purchased from Aldrich. Commonly
used abbrevations are EtOAc (ethyl acetate), MeOH (metha-
nol), DMF (N,N-dimethylformamide), THF (tetrahydrofuran),
and CH2Cl2 (dichloromethane).
F lu or op r op a n ed ioic Acid , 4-Iod o-2-(m eth oxyca r bon -
yl)p h en yl-, Bis(1,1-d im eth yleth yl) Ester (12). An amount
of 2.0 g (4.20 mmol) of 2-(4-iodo-2-methoxycarbonylphenyl)-
propanedioic acid, 1,3-bis(1,1-dimethylethyl) ester 119 was
dissolved in 88 mL of THF. The solution was cooled to -78
°C, and then 3.35 g (16.78 mmol) of potassium hexamethyl-
disilazide 95% was added. The solution was stirred for 1 h
under argon to -70 °C. An amount of 5.23 g (16.77 mmol) of
N-fluorobenzenesulfonimide was added. The mixture was
stirred for 15 min at -70 °C and then warmed to RT and
stirred for 2 h. The mixture was poured into a saturated
aqueous solution of NaCl and extracted with EtOAc. The
organic extracts were washed with brine, dried, and evapo-
rated to dryness to afford 5.54 g of crude 12 as a cream powder.
Flash chromatography (320 g silica, CH2Cl2/EtOAc, 99:1) gave
h
at room temperature and then poured into H2O and
extracted with EtOAc. The organic extracts were washed with
an aqueous solution of NaHCO3 and then brine, dried, and
concentrated to afford 0.391 g of crude product. Flash chro-
matography (50 g of silica, CH2Cl2/MeOH, 90/10) gave 0.275
g of the expected compound as a white foam (67%). Rf ) 0.13
(SiO2F254Merck60 CH2Cl2/MeOH, 97.5/2.5); MS m/z 770 (M -
H)-,794 (M + Na)+.
Flu or opr opan edioic Acid, [4-[-2-(Acetylam in o)-3-[[(3S)-
1-[[(1,1-biph en yl)-4-yl]m eth yl-h exah ydr o-2-oxo-1H-azepin -
3-yl]a m in o]-3-oxo-1-p r op en yl]-2-(m eth oxyca r bon yl)p h e-
n yl]- (4). A hydrogenating flask was charged with 0.264 g
(0.342 mmol) of 15 and dissolved in 30 mL of methanol, and
85 mg of 10% palladium on carbon (E10ND charge 7389) was
added. The mixture was placed under 1760 mbar of hydrogen
for 48 h, and then the suspension was filtered over Celite and
concentrated to afford 0.241 g of the pure hydrogenated
intermediate as a white foam (91%, 60/40 mixture of diaste-
reoisomers). Rf ) 0.44 (SiO2F254Merck60 CH2Cl2/MeOH, 96.5/
3.5) HPLC (CH3CN/H2O, 75:25, 0.01%,TFA, 1 mL/min, 63 bar,
230 nm, 9.45 min, 92.5%; NMR (DMSO-d6) δ 1.17-1.67 (m,
2H), 1.3-1.9 (m, 4H), 1.46 (s, 18H), 1.75 (1s, 3H), 1.76 (1s,
3H), 2.81 and 3.06 (2d, 1H), 2.81 and 3.06 (2dd, 1H), 3.28 and
3.58 (2m, 2H), 3.75 (s, 3H), 4.60-4.67 (m, 1H), 4.61 (s, 1H),
4.52 and 4.69 (d, 1H), 4.47 and 4.72 (d, 1H), 7.19, 7.36, 7.46,
7.65, 7.54, 7.75 (12H), 8.20 (dd, 1H), 8.24 (dd, 1H). 0.204 g
(0.264 mmol) of this material was dissolved in 6 mL of 98%
HCOOH (157 mmol). The solution was stirred for 6 h at room
temperature and then poured into H2O and extracted with
AcOEt. The organic extracts were washed with an aqueous
solution of NaHCO3 and then brine, dried, and concentrated
to afford 0.187 g of crude 4 as a white foam. Crude 4 was
triturated with cold diethyl ether and filtered to afford 0.156
g (96%) of 4 as a white powder (60/40 mixture of diastereoi-
somers). Rf ) 0.13 (SiO2F254Merck60 CH2Cl2/MeOH/AcOH, 80/
20/5). NMR (MeOD), δ 1.23-1.86 ((m, 6H), 1.94-1.96 (s, 3H),
2.96-3.40 (d, 2H), 3.37 and 3.59 (dd, 2H), 3.81 (s, 3H), 4.57
1.74 g (83.7%) of 12 as a colorless oil. Rf ) 0.20 (SiO2F254
-
Merck60 CH2Cl2/MeOH, 99/1); NMR (CDCl3) δ 1.48 (s, 18H),
3.88 (s, 3H), 7.28 (d, 1H), 7.84 (d, 1H), 8.30 (d, 1H).
F lu or op r op a n ed ioic Acid , 4-[2-(Acetyla m in o)-2-ca r -
b oxyet h en yl]-2-(m et h oxyca r b on yl)p h en yl-, Bis(1,1-d i-
m eth yleth yl) Ester (13b). An amount of 1.72 g (3.4 mmol)
of 12 was introduced into a 500 mL flask and dissolved in 65
mL of CH3CN. An amount of 1 g (6.95 mmol) of methyl

5194 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 24
Lange et al.
and 4.72 (m, 2H), 4.66 (m, 1H), 7.28-7.65 (m, 1H), 7.73-7.78
(m, 11H), 7.93 (m, 1H), 8.10 (m, 1H), 8.30 (m, 1H). MS m/z
660 (M - H)-, 662 (M + H)+.
Ack n ow led gm en t. The authors would like to thank
Herman Schreuder, Steve Lindell, and Klaus Haaf for
helpful discussion and Alexander Liesum for excellent
technical assistance. This work was done during the
course of a collaboration with ARIAD Pharmaceuticals
and we are grateful to T. Sawyer and M. Hatada for
stimulating discussions.
Ca r b a m ic Acid , [(1S)-2-[[(3S)-1-([1,1′-Bip h en yl]-4-yl-
m eth yl)h exah ydr o-3-m eth yl-2-oxo-1H-azepin -3-yl]am in o]-
1-[[3,4-b is(d iet h oxyp h osp h in yl)p h en yl]m et h yl]-2-oxo-
eth yl]-, 1,1-Dim eth yleth yl Ester (17). An amount of 0.3 g
(0.56 mmol) of 3,4-bis(diethoxyphosphinyl)-N-[(1,1-dimethyl-
ethoxy)carbonyl]-L-phenylalanine 1623 was dissolved in 6 mL
of CH2Cl2 and 0.6 mL of DMF in a 50 mL flask under argon
and cooled in an ice bath. An amount of 0.106 g (0.558 mmol)
of 1-(3-dimethylaminopropyl)-3-ethylcarbodiimide hydrochlo-
ride (EDCI) was added followed by 0.075 mg (0.558 mmol) of
1-hydroxybenzotriazole (HOBT). The solution was stirred for
15 min at room temperature and then added dropwise to a
cooled solution of 0.140 mg (0.558 mmol) of 14.13 The resulting
solution was stirred for 2 h at room temperature and then
poured into H2O and extracted with CH2Cl2. The organic
extracts were washed with an aqueous solution of NaHCO3
and then brine, dried, and concentrated to afford 0.300 g of
crude 17. Flash chromatography (30 g of silica, CH2Cl2/MeOH,
95/5) gave 0.260 g of 17 (57%). Rf ) 0.37 (SiO2F254Merck60 CH3-
COCH3/ACOEt/H2O, 50/40/10); MS m/z 812 (M - H)-, NMR
(CDCl3), δ 1.36 (m,12H), 1.14-2.12 (m, 6H), 3.10 (m, 1H), 3.27
(m, 2H), 3.51 (dd, 1H), 4.20 (m, 8H), 4.51 (m, 1H), 4,55 and
4.75 (dd, 2H), 4.63 (dd, 1H), 4.94 (d, 1H), 7.30 and 7.56 (m,
4H), 7.35 (m, 1H), 7.44 (dd, 1H), 7.47 (dd, 1H), 7.57 (dd, 1H),
7.96 (dd, 1H), 8.08 (dd, 1H).
Su p p or tin g In for m a tion Ava ila ble: 1H NMR spectra of
compounds 4 and 9. This material is available free of charge
via the Internet at http://pubs.acs.org.
Refer en ces
(1) Spencer, R. W. High-throughput screening of historic collec-
tions: observations on file size, biological targets, and file
diversity. Biotechnol. Bioeng. 1998, 61, 61-67.
(2) Kuntz, I. D.; Blaney, J . M.; Oatley, S. J .; Langridge, R.; Ferrin,
T. E. A geometric approach to macrormolecule-ligand interac-
tions. J . Mol. Biol. 1982, 161, 269-88.
(3) Rarey, M.; Kramer, B.; Lengauer, T.; Klebe, G. A fast flexible
docking method using an incremental construction algorithm.
J . Mol. Biol. 1996, 261, 470-489.
(4) Goodford, P. J . A. Computational Procedure for determining
energetically favorable binding sites on biologically important
interaction sites. J . Med. Chem. 1985, 28, 849-857.
(5) Boehm, H. J ., The computer program LUDI: a new method for
the de novo design of enzyme inhibitors. J . Comput.-Aided Mol.
Des. 1992, 6, 61-78.
(6) Shuker, S. B.; Hajduk, P. J .; Meadows, R. P.; Fesik, S. W.
Discovering high-affinity ligands for proteins: SAR by NMR.
Science 1996, 274, 1531-4.
(7) Hajduk P. J .; Zhou, M.-M.; Fesik; S. W. NMR-based discovery
of phosphyotyrosine mimetics that bind to the Lck SH2 domain.
Bioorg. Med. Chem. Lett. 1999, 9, 2403-2406.
(8) Nienaber; V. L.; Richardson, P. L. Klighofer, V.; Bouska, J . J .;
Giranda, V. L.; Greer; J . Discovering novel ligands for macro-
molecules using X-ray crystallography screening. Nat. Biotech-
nol. 2000, 18, 1105-1107.
(9) Lesuisse, D.; Lange, G.; Deprez, P.; Schoot, B.; Benard, D.;
Delettre, G.; Marquette, J .-P.; Broto, P.; J ean-Baptiste, V.;
Bichet, P.; Sarubbi, E.; Mandine, E. J . SAR and X-ray- a new
approach combining fragment-based screening and rational drug
design: Application to the discovery of nanomolar inhibitors of
src SH2. J . Med. Chem. 2002, 45, 2379-2387.
(10) Soriano, P.; Montgomery, C.; Geske, R.; Bradley, A., Targeted
disruption of the c-src proto-oncogene leads to osteopetrosis in
mice. Cell 1991, 64, 693-702.
(11) Waksman, G.; Shoelson, S. E.; Pant, N.; Cowburn, D.; Kuriyan,
J . Binding of a high affinity phosphotyrosyl peptide to the src
sh2 domain: crystal structures of the complexed and peptide-
free form. Cell 1993, 72, 779-790.
(12) Deprez, P.; Mandine, E.; Gofflo; D.; Meunier, S.; Lesuisse, D.
Small ligands interacting with the phospho-tyrosine binding
pocket of the Src SH2 protein. Bioorg. Med. Chem. Lett. 2002,
12, 1295-1298.
P h osp h on ic Acid , [4-[(2R)-2-(Acetyla m in o)-3-[[(3S)-1-
([1,1′-b ip h en yl]-4-ylm et h yl)h exa h yd r o-3-m et h yl-2-oxo-
1H -a zep in -3-yl]a m in o]-3-oxop r op yl]-1,2-p h en ylen e]b is,
Tetr a eth yl Ester (18b). An amount of 0.24 g (0.294 mmol)
of17 was dissolved in 4 mL of CH2Cl2 and 1.5 mL of trifluo-
roacetic acid in a 50 mL flask under argon. The mixture was
stirred for 1 h at room temperature and then concentrated
under reduced pressure. The residue was dissolved in CH2Cl2
and washed with an aqueous solution of NaHCO3 and then
brine, dried, and concentrated to afford 0.200 g of the amine
18a (95%). Rf ) 0.33 (SiO2F254Merck60 CH2Cl2/MeOH, 90/10);
MS m/z 712 (M - H)-. An amount of 0.20 g (0.28 mmol) of
18a was dissolved in 4 mL of pyridine under argon and cooled
at 0 °C, and then 0.027 mL (0.336 mmol) of acetic anhydride
was added dropwise. The resulting solution was stirred for 2
h
at room temperature and then poured into H2O and
extracted with CH2Cl2. The organic extracts were washed with
an aqueous solution of NaHCO3 and then brine, dried, and
concentrated to afford 0.200 g of crude product. Flash chro-
matography (20 g of silica, CH2Cl2/MeOH, 95/5) gave 0.170 g
of 18b (81%). Rf ) 0.46 (SiO2F254Merck60 CH2Cl2/MeOH, 90/
10); MS m/z 754(M - H)-. NMR (CDCl3),δ 1.20 (m, 1Η), 1.34-
1.37 (t, 12H), 1.47 (m, 1H), 1.73 (m, 2H), 1.92 (m, 1H), 2.00 (s,
3H), 2.03 (m, 1H), 3.12 (dd, 1H), 3.27 (dd, 1H), 3.28 (m, 1H),
3.52 (dd, 1H), 4.19 (m, 8H), 4,61 and 4.70 (d, 2H), 4.61 (m,
1H), 4.81 (dd, 1H), 6.05 (d, 1H), 7.33 and 7.57 (m, 7H), 7.41
(m, 1H), 7.44 (m, 3H), 7.96 (dd, 1H), 8.07 (dd, 1H).
(13) Deprez, P.; Baholet, I.; Burlet, S.; Lange, G.; Schoot, B.;
Vermond, A.; Mandine, E.; Lesuisse, D. Discovery of highly
potent Src SH2 binders: Structure-Activity studies and X-ray
structures. Bioorg. Med. Chem. Lett. 2002, 12, 1291-1294.
(14) Lesuisse, D.; Deprez, P.; Albert, E.; Duc, T. T.; Sortais, B.; Gofflo,
D.; J ean-Baptiste, V.; Marquette, J .-P.; Schoot, B.; Sarubbi, E.;
Lange, G.; Broto, P.; Mandine, E. Discovery of Thioazepinone
Ligands for Src SH2: From Nonspecific to Specific Binding.
Bioorg. Med. Chem. Lett. 2001, 11(16), 2127-2131.
(15) Lange, G.; Lesuisse, D.; Deprez.; P.; Schoot, B.; Loenze, P.;
Be´nard, D.; Marquette, J .-P.; Broto, Sarubbi, E.; Mandine, E.
Principles governing the binding of a class of nonpeptidic
inhibitors to the SH2 domain of src studied by X-ray analysis.
J . Med. Chem 2002, 45, 2915-2922.
[4-[(2R)-2-(Acetyla m in o)-3-[[(3S)-1-([1,1′-bip h en yl]-4-yl-
m eth yl)h exah ydr o-3-m eth yl-2-oxo-1H-azepin -3-yl]am in o]-
3-oxop r op yl]-1,2-p h en ylen e]bisp h osp h on ic Acid (9). An
amount of 0.165 g (0.218 mmol) of18b was dissolved in 3.3
mL of CH2Cl2 in a 25 mL flask under argon and cooled in an
ice bath, and then 0.290 mL (2.18 mmol) of iodotrimethylsilane
was added. The resulting solution was stirred for 3 h, warmed
to room temperature, and finally concentrated under reduced
pressure. This material was dissolved in MeOH, concentrated
under reduced pressure, and filtered to afford 0.130 g of crude
9. Preparative HPLC (CH3CN/H2O, 75:25, 0.01%,TFA, 3 mL/
min, 95 bar, 230 nm gave 0.070 g of 9 (50%); MS m/z 642 (M
- H)-; NMR (DMSO-d6) δ 1.17 (m, 1H), 1.45 (1H), 1.68 (dl,
2H), 1.82 (m, 2H), 1.79 (s, 3H), 2.82 (dd, 1H), 3.19 (dd, 1H),
3.28 (dd, 1H),3.60 (dd, 1H), 4.52 (dd, 1H), 4.60 (m, 1H), 4.69
(dd, 1H), 7.37 and 7.63 (dd, 4H), 7.37 (m, 1H), 7.46 (t, 2H),
7.54 (d, 1H), 7.65 (d, 1H), 7.82 (m, 1H), 7.89 (m, 1H), 8.23 (m,
2H).
(16) Kabsch, W., Automatic processing of rotation diffraction data
from crystals of initially unknown symmetry and cell constants.
J . Appl. Crystallogr. 1993, 26, 795-800.
(17) Brunger, A. T.; Kurkowski, A.; Erickson, J . W. Slow-cooling
protocols for crystallographic refinement by simulated annealing.
Acta Crystallogr. 1990, A46, 46-57.
(18) Oldfield, T. J . A number of real-space torsion-angle refinement
techniques for proteins, nucleic acids, ligands and solvent. Acta
Crystallogr. D 2001, D57, 82-94.
(19) W. Kabsch. A discussion of the solution for the best rotation to
relate two sets of vectors. Acta Crystallogr. 1978, A34, 827-
828.
(20) Berman H. M. et al. The protein data bank. Nucleic Acids Res.
2000, 28, 235-242.
(21) Deprez, P.; Lesuisse, D.; Be´nard, D.; Caprolactam derivatives
and uses thereof. WO 01/68655 A2, 2001.

Binding to the SH2 Domain of pp60Src
J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 24 5195
(22) Kawahata, N.; Yang, M. Y.; Luke, G. P.; Shakespeare, W. C.;
Sundaramoorthi, R.; Wang, Y.; J ohnson, D.; Merry, T.; Violette,
S.; Guan, W.; Bartlett, C.; Smith, J .; Hatada, M.; Lu, X.;
Dalgarno, D. C.; Eyermann, C. J .; Bohacek, R. S.; Sawyer, T. K.
A novel phosphotyrosine mimetic 4′-carboxymethyloxy-3′-phospho-
nophenylalanine (Cpp): exploitation in the design of non peptide
inhibitors of pp60src SH2 Domain. Bioorg. Med. Chem. Lett. 2001,
11, 2319-23.
(23) Shakespeare, W. et al. Structure-based design of an osteoclast-
selective, nonpeptide src homology 2 inhibitor with in vivo
antiresorptive activity. Proc. Natl. Acad. Sci. 2000, 97, 9373-
9378.
(24) Maly, D. J .; Choong, I. C.; Ellman, J . A. Combinatorial target-
guided ligand assembly: identification of potent subtype-selec-
tive c-Src inhibitors. Proc. Natl. Acad. Sci. U.S.A. 2000, 97,
2419-2424.
(25) Kelly, M. A. et al. Characterization of SH2-Ligand Interactions
via Library Affinity Selection with Mass Spetrometric Detection.
Biochemistry 1996, 35, 11747-11755.
R.; Proudfoot, J .; J akes, S. Carboxymethyl-phenylalanine as a
replacement for phospho-tyrosine in SH2 domain binding J . Biol.
Chem. 1998, 273, 20238-20242.
(29) Charifson, P. S. et al. Peptide Ligands of pp60c-src SH2
domains: a thermodynamic and structural study. Biochemistry
1997, 36, 6283-6293.
(30) Gao, Y.; Wu, L.; Luo, J .; Guo, R.; J ang, D.; Zhang, Z.-Y.; Burke,
T. Examination of novel nonphosphorous-containing phospho-
tyrosyl mimetics against protein-tyrosine phosphatase 1B and
demonstration of differential affinities toward Grb2 SH2. Bioorg.
Med. Chem. Lett 2000, 10, 923-927.
(31) Savage, H. J .; Elliott, C. J .; Freeman, C. M. Finney, J . L. Lost
Hydrogen bonds and buried Surface Area: Rationalizing stabil-
ity in globular proteins. J . Chem. Soc., Faraday Trans. 1993,
89, 2609-17.
(32) Norel.; R:, Sheinerman, F.; Petrey, D.; Honig, B. Electrostatic
contributions to protein-protein interactions: Fast energetic
filters for docking and their physical basis. Protein Sci.. 2001,
10, 2147-2161.
(33) Lim, V. I. Curran, J . F. Analysis of codon; anticodon interactions
within the ribosome provides new insight into codon reading and
the genetic code structure. RNA 2001, 7, 942-957.
(34) Schultz, J .; Milpetz, F.; Bork, P.; Ponting, C. P. SMART, a simple
modular architecture research tool: identification of signaling
domains. Proc. Natl. Acad. Sci. U.S.A. 1998, 95, 5857-5864.
(26) Hajduk, P. J .; Meadows, R. P.; Fesik, S. W. Privileged molecules
for protein binding identified from NMR-based screening. J .
Med. Chem. 2000, 43, 3443-3447.
(27) Grueneberg, S.; Wendt, B.; Klebe, G. Subnanomolar Inhibitors
from Computer Screening: A model study using Human Car-
bonic Anhydrase II Angew. Chem., Int. Ed. 2001, 40, 389-393.
(28) Tong, L.; Warren, T. C.; Lukas; S.; Schembri-King, J .; Betageri,
J M020970S