Discovery of thioazepinone ligands for Src SH2: From non-specific to specific binding

Article abstract of DOI:10.1016/S0960-894X(01)00386-9

The structure-based design and synthesis of new thioazepinones as ligands for Src SH2 protein is presented. From benzothioazepinones, ligands with somewhat unspecific binding properties, simpler thioazepinones were designed, the best ones demonstrated nanomolar affinity for Src SH2. A few of these new ligands were crystallized with the protein and demonstrated a specific binding mode with the protein.

Full text of DOI:10.1016/S0960-894X(01)00386-9

Bioorganic & Medicinal Chemistry Letters 11 (2001) 2127–2131
Discovery of Thioazepinone Ligands for Src SH2: From
Non-specific to Specific Binding
Dominique Lesuisse,a,* Pierre Deprez,a Eva Albert,a Tran Thien Duc,a Benoit Sortais,a
Dominique Gofflo,b Veronique Jean-Baptiste,b Jean-Pierre Marquette,c Bernard Schoot,c
Edoardo Sarubbi,c Gudrun Lange,c Pierre Brotoa and Eliane Mandineb
aDepartment of Medicinal Chemistry, Aventis, 102 route de Noisy, 93235 Romainville Cedex, France
bBone Disease Group, Aventis, 102 route de Noisy, 93235 Romainville Cedex, France
cCore Research, Aventis, 102 route de Noisy, 93235 Romainville Cedex, France
Received 20 March 2001; accepted 21 May 2001
Abstract—The structure-based design and synthesis of new thioazepinones as ligands for Src SH2 protein is presented. From ben-
zothioazepinones, ligands with somewhat unspecific binding properties, simpler thioazepinones were designed, the best ones
demonstrated nanomolar affinity for Src SH2. A few of these new ligands were crystallized with the protein and demonstrated a
specific binding mode with the protein. # 2001 Elsevier Science Ltd. All rights reserved.
Introduction
fold–hydrophobic). The study of the X-ray structure of
Tyrosine kinases are involved in many pathologies such
as cancer, inflammation or osteoporosis and are conse-
quently important targets for the pharmaceutical
industry.1 In particular, Src has been shown to be
essential in the process of bone resorption.2 Src is the
member of a family of 10 homologous proteins featur-
ing several functional domains, including SH2 and SH3
domains involved in protein–protein interactions, and a
tyrosine kinase catalytic domain.3 Rescue experiments4
suggested that the SH2 domain is critical to the bone
resorbing activity of Src and therefore we embarked in a
program of finding inhibitors of the SH2 domain of Src.
The SH2 domain of the Src family specifically recog-
nizes a sequence of tetrapeptide featuring a phospho-
tyrosine and a lipophilic amino acid at the +1 and +3
positions. In particular, pYEEI, a sequence found on
the PDGF receptor upon activation5 has been shown to
specifically recognize Src SH2 domain. The approach
that we have used to discover nonpeptidic inhibitors of
this tetrapeptide sequence has been typically rational
drug design driven from a peptidomimetic modular
approach where the pYEEI peptide was viewed as a
three-component ligand (phosphotyrosine–central scaf-
pYEEI with Src SH26 reveals that the middle glutamate
amino acids do not make a strong interaction with the
peptide backbone, and therefore, mainly serve the pur-
pose of presenting the phosphotyrosine and the hydro-
phobic group to their respective pockets. However, the
aliphatic chain of the +1 glutamate makes an impor-
tant hydrophobic interaction with a tyrosine of the
protein and we decided to search our library in order to
identify potential scaffolds capable of restoring this
interaction in a more rigid frame. On the other hand,
the +2 glutamate is solvent oriented and molecular
modeling suggested that a glycine would be sufficient to
present the hydrophobic moiety in the desired position.
Chemistry and Biology
For the +1 glutamate replacement, we explored by
molecular modeling several scaffolds we had at hand in
our library. One of them, the benzothioazepinone scaf-
fold found in diltiazem,7 was chosen first. Initially, a
compound where the +1 glutamate had been replaced
by a benzothioazepinone was synthesized. Starting ben-
zothioazepinone 18 was condensed with protected l-
tyrosine using EDC and BOBt in dichloromethane and
the resulting compound 2 was alkylated on the cyclic
nitrogen with various substituted derivatives of
*Corresponding author. Tel.: +33-1-4991-4731; fax: +33-1-4991-
5087; e-mail: dominique.lesuisse@aventis.com
0960-894X/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(01)00386-9

2128
D. Lesuisse et al. / Bioorg. Med. Chem. Lett. 11 (2001) 2127–2131
Scheme 1. (i) BOC-Tyr-O(PO3Bn2), EDC, HOBt, CH2Cl2 Hunig’s base, DMF (49%); (ii) (1) R1Br, NaH, DMF; (2) NaI, acetone.
bromoacetamide using sodium hydride in DMF with
good yields. The resulting benzylphosphate esters were
mono deprotected using NaI in acetone to afford the
new derivatives 3a–k (Scheme 1).
with the protein. One main problem was solubility of
the inhibitors and we began to question the specificity of
the binding. This was also suggested by the shape of the
binding curves. For instance, while pYEEI showed a
straight binding curve with a Hill coefficient10 around 1,
4e displayed a steeper curve with a Hill coefficient of
1.75 suggestive of cooperative binding (Fig. 1). In addi-
tion, when we performed binding with increasing
DMSO concentration which did not affect the binding
affinity of pYEEI, the affinity of the nonpeptides was
sometimes dramatically decreased (data not shown)
which also alluded to the higher lipophilic component
of the binding affinity (in agreement with the high log P
values calculated for these compounds).
Table 1 presents the hydrophobic moieties that were
explored and the binding affinities of the new derivatives
on Src SH2 (determined with an assay relying on fluo-
rescence polarization9). The best one, compound 3k,
presented binding affinity comparable to the one of
pYEEI in this assay.
Alternatively, several analogues where the tyrosine
group had been replaced by another moiety were syn-
thesized as shown in Scheme 2. The resulting new deri-
vatives are presented in Table 2, the best analogue being
the one where the tyrosine replacement was a 4-mono-
benzylphosphate-phenyl acetate (4e).
It has been shown from the early work of Gilmer et al.11
that changing the +1 l-glu of pYEEIE peptide for a d-
Glu residue resulted in a 250-fold loss in binding affi-
nity. In the case of 3c, the four separated isomers gave
about the same binding affinity when comparing the
pairs varying at +1 position (Table 3, cf. 3c with 3e and
3m with 3n). This further suggested that the binding
displayed some unspecific component (Table 3).
However, after several rounds of optimization around
compounds 3k or 4e (data not shown), binding affinity
could not be further improved. In addition, none of
these inhibitors gave rise to successful co-crystallization
Table 1. Variation of the hydrophobic moieties of compound 3
R1
Compd
Isomers
IC50a(mM)
Table 2. Variation of the phosphotyrosine moieties of compound 3
pYEEI
—
3a
3b
3c
3d
3e
3f
3g
3h
3i
—
trans, 2 isos
cis, 2 isos
6.5
128
118
189
95
14.2
873
23
a
R2
Compd
IC50 (mM)
CH2CONHCH2-1-naphthyl
CH2CONHCH2-1-naphthyl
CH2CONH-1-naphthyl
CH2CONHPh-2-(OMe)
CH2CONHPh-3-(OPh)
H
pYEEI
l-Tyr(NBOC)OPO(OBn)(OH)
—
3a
6.5
trans, iso 1
trans, iso 1
trans, iso 1
trans, iso 1
trans, iso 1
trans, iso 1
trans, iso 1
trans, iso 1
trans, iso 1
128
4a
4b
4c
4d
4e
4f
339
27
23
32
5
CH(Ph)CONH-1-Naphthyl
CH2Ph
46
29
18
CH2CONHPh-4-(OPh)
CH2Ph-4-Ph(20-COOtBu)
(CH2) Cyclohexyl
3j
3k
6.4
3
aIC50’s of inhibition determined using the FP assay9 in a medium
containing 2% DMSO.
20
Scheme 2. (i) R2COOH, EDC, BOBt, CH2Cl2, Hunig’s base, DMF;
(ii) (1) BrCH2CONHCH2-1-naphthyl, NaH/DMF; (2) NaI/acetone.
aIC50’s of inhibition determined using the FP assay9 in a medium
containing 2% DMSO.

D. Lesuisse et al. / Bioorg. Med. Chem. Lett. 11 (2001) 2127–2131
2129
Figure 1. Displacement curves of pYEEI from Src SH2 by pYEEI (a) and 4e (b) using fluorescence polarization assay.
In this series, monobenzylphosphates appeared with
better binding affinities than the corresponding free
phosphates (i.e., compare 3k with its deprotected ana-
logue 3o in Table 4) and molecular modeling suggested
that this could arise from an additional beneficial
hydrophobic interaction around the phosphotyrosine
binding pocket (data not shown). However, the mono-
benzylphosphate ester of the pYEEI peptide was syn-
thesized and its very poor binding affinity for Src SH2
(ꢀ100 mM) further suggested that the binding mode of
the nonpeptidic inhibitors was substantially different
from the peptides.
esis of scaffolds 512 and 713 devoid of one and the other
aromatic rings, respectively. Benzothioazepinones 5 and
6 were prepared from orthofluoronitrobenzene and l-
or d-cysteine, respectively, as described12 while the
thioazepinone 7 was prepared via a one-step condensa-
tion between l-cysteine and aminochloroethane as
reported.13 From 5, 6 and 7, the synthesis of 8, 9, 10 and
11 was straightforward (Scheme 3).
In the meantime, because our FP assay was at its
detection limits and required high concentrations of
compounds giving rise to solubility problems, we devel-
oped a more sensitive binding assay using SPA
technology.14 In this assay, the pYEEI reference peptide
Molecular modeling suggested that neither of the two
aromatic rings on the benzothioazepinone moiety were
involved in the binding, so in order to lower the lipo-
philicity of the compounds, we embarked in the synth-
Table 3. Binding of the four isomers of 3c with Src SH2
Scheme 3. (i) C6H11(CH2)3I, NaH, DMF (100%); (ii) HClg, AcOH
then NaHCO3 (50%); (iii) l- or d-BOC-Tyr-O(PO3Bn2), EDC, HOBt,
DMF, CH2Cl2, Hunig’s base, DMF; (iv) NaI/acetone (72%) (8) or H2,
Pd–C, MeOH (74%) (9 or 10). (v) l-BOC-Tyr-O(PO3Bn2), EDC,
HOBt, DMF CH2Cl2, Hunig’s base; (vi) C6H11(CH2)3I, NaH, DMF
(16%); (vii) H2, Pd–C, MeOH (94%).
a
Isomer
Compd
IC50 (mM)
trans iso 1
trans iso 2
cis iso 1
3c
3l
3m
3n
189
133
26
cis iso 2
23
Table 5. Optimization of the hydrophobic moiety on compound
aIC50’s of inhibition determined using the FP assay9 in a medium
containing 2% DMSO.
Table 4. Removal of the aromatic rings from compound 3k
a
b
Compd
IC50 (mM)
IC50 (mM)
IC50c (mM)
a
R
Compd
IC50 (mM)
pYEEI
6.5
6.4
20
26
10
5
28
164
87
11
nd
nd
0.15
>100
nd
3k
3o
8
9
10
11
CH2COPhNMe2
CH2PhOCF3
CH2PhCF3
CH2PhCOPh
CH2PhPh
CH2PhClOCH2O
CH2PhCOOMe
CH2PhPh-2-CN
12
13
14
15
16
17
18
19
9.3
0.4
1.6
1.6
0.09
1.2
0.13
0.3
13
1
18
422
115
1.4
aFP assay using 2% DMSO in the buffer solution.9
bFP assay using 20% DMSO in the buffer solution.9
cSPA assay.14
aSPA assay.

2130
D. Lesuisse et al. / Bioorg. Med. Chem. Lett. 11 (2001) 2127–2131
compound 11 with approximately the same binding
affinity in the SPA binding assay as 9. These series of
lower molecular weight and lipophilicity were chosen
for further optimization.
Exploration of the hydrophobic moiety was revisited.
The compounds described in Table 5 have been synthe-
sized in an analogous fashion as described above, albeit
using solid-phase chemistry.15 Aside from the para-
amino substituted aromatic ring which gave rise to
lower binding affinity (12), the other substituents gave
comparable or better IC50’s with Src SH2. Best affinity
was obtained with the biphenylmethyl substitution, with
an IC50 of inhibition of 90 nM (15).
Figure 2. X-ray of 15 (turquoise) superimposed with pYEEI.11
The binding mode of some of these compounds with Src
SH2 was established by X-ray studies. In particular,
compounds 15 and 16 (Figs 2 and 3) gave rise to crystals
with a 1.8 A resolution. The comparison of both crys-
tals with pYEEI revealed that the inhibitors indeed bind
Src SH2 in the same way as the peptide with main
interactions between the phosphate and Arg14 and 34
and in the hydrophobic pocket with Tyr89, Ile73 and
Leu96. Both inhibitors are deeper in the pocket and
make a closer contact with these amino acids than the
+3 Ile of pYEEI. The +1 carbonyl interacts with the
protein backbone through a water molecule (78) and the
same type of interaction is found in both crystals with
the thioazepinone carbonyl. The thioazepinone ring
mimics the hydrophobic contact between Y61 and the
+1 chain. The better affinity of 16 versus 15 could be
accounted for by the additional hydrophobic interaction
between the first phenyl ring and Tyr61.
Figure 3. X-ray of 16 (blue) superimposed with pYEEI.11
Conclusion
gives an IC50 about 40-fold lower than in the FP assay,
partly because the protein concentration was lowered.
The compounds were validated in this assay (Table 4).
To our surprise, compound 3k that displayed a binding
affinity comparable to the one of pYEEI in the FP
assay, appeared almost inactive in the SPA assay.
Compound 8 devoid of the 2-position aromatic ring
displayed a slightly lower binding affinity in the FP
assay, however its affinity in the SPA binding assay
showed a 10-fold improvement suggesting that this
assay was better suited to differentiate specific binding
from unspecific hydrophobic interactions. In addition,
for the first time in these series, the free phosphate 9
demonstrated better binding affinity than the mono-
benzylphosphate ester (compare 9 and 8 vs 3o and 3k)
which suggested binding was taking place in the same
way as for pYEEI. Also, this new compound now did
not present any difference in binding affinity between
the 2 and 20% DMSO conditions in the FP assay which
also suggested that the lipophilic component of the
binding was less important. In addition, Hill coefficients
for the binding curves were around 1 suggesting specific
displacement (data not shown). Finally, the analogue 10
incorporating a d stereochemistry at the C3 position
was synthesized and found about 20-fold less active
than the l-analogue 9, confirming a specific binding
mode. Removing the fused aromatic ring afforded
A family of Src SH2 inhibitors was designed starting
from the benzothioazepinone scaffold found in the
cardiovascular drug diltiazem. From there, simpler and
more specific nanomolar inhibitors were discovered. X-
ray analysis established these inhibitors bound to Src
SH2 in the predicted mode.
Acknowledgements
The authors are grateful to M. Zoller, D. Holt, T. Saw-
yer, R. Bohacek, M. Weigele (Ariad Pharmaceuticals)
for helpful discussions. We thank our Department of
Analytics for physical analyses, and our Protein
Department for protein production.
References and Notes
1. Brugge, J. S. Science 1993, 260, 918.
2. Soriano, P.; Montgomery, C.; Geske, R.; Bradley, A. Cell
1991, 64, 693.
3. For a recent review on SH2 domains see: Sawyer, T. K.
Biopolymers 1998, 47, 243

D. Lesuisse et al. / Bioorg. Med. Chem. Lett. 11 (2001) 2127–2131
2131
4. (a) Schwartzberg, P. L.; Xing, L.; Hoffmann, O.; Lowell,
C. A.; Garrett, L.; Boyce, B. F.; Varmus, H. E. Genes Develop-
ment 1997, 11, 2835. (b) Kaplan, K. B.; Swedlow, J. R.; Morgan,
D. O.; Varmus, H. E. Genes Development 1995, 9, 1505.
5. Twamley, G. M.; Kypta, R. M.; Hall, B.; Courtneidge, S. A.
Adv. Second Messenger Phosphoprotein Res. 1993, 28, 187.
6. Waksman, G.; Shoelson, S. E.; Pant, N.; Cowburn, D.;
Kuriyan, J. Cell 1993, 72, 779.
7. Kiowski, W.; Linder, L.; Buhler, F. R. J. Cardiovasc.
Pharmacol. 1989, 13 (Suppl.), 1S25.
8. Levai, A.; Duddeck, H. Pharmazie 1983, 38, 827.
9. Lynch, B. A.; Loiacano, K. A.; Tiong, C. L.; Adams, S. E.;
McNeil, I. A. Anal. Biochem. 1996, 247, 77.
11. Gilmer, T.; Rodriguez, M.; Jordan, S.; Crosby, R.; Alli-
good, K.; Green, M.; Kimery, M.; Wagner, C.; Kinder, D.;
Charifson, P.; Hassel, A. M.; Willard, D.; Luther, M.; Rus-
nak, D.; Sternbach, D. D.; Mehrotra, M.; Peel, M.; Shampine,
L.; Davis, R.; Robbins, J.; Patel, I. R.; Kassel, D.; Burkhart,
W.; Moyer, M.; Bradshaw, T.; Berman, J. J. Biol. Chem. 1994,
269, 31711.
12. Morton, G. C.; Salvino, J. M.; Labaudiniere, R. F.; Her-
pin, T. F. Tetrahedron Lett. 2000, 41, 3029.
13. Ahmed, S. A.; Esaki, N.; Tanaka, H.; Soda, K. FEBS
Lett. 1984, 174, 76.
14. Mandine, E.; Gofflo, D.; Jean-Baptiste, V.; Gaillard-
Kelly, M.; Zoller, M.; Baron, R.; Lesuisse, D. J. Bone Miner.
Res. 1998, 23, S221.
15. Deprez, P.; Lesuisse, D. Tetrahedron, submitted for pub-
lication.
10. Limbird, L. E. Cell Surface Receptors: A Short Course on
Theory and Methods; Martinus Nijhoff: Boston, 1986; pp 51–
96