Selective inhibition of Src SH2 by a novel thiol-targeting tricarbonyl-modified inhibitor and mechanistic analysis by 1H/13C NMR spectroscopy

Article abstract of DOI:10.1016/S0960-894X(01)00269-4

Detailed analysis of Src SH2 binding by peptides containing a novel tricarbonyl-modified pTyr moiety is described. We envisaged that Src SH2 selectivity might be obtained by exploiting the thiol group of Cys188 present in the pTyr binding pocket of the pr

Full text of DOI:10.1016/S0960-894X(01)00269-4

Bioorganic & Medicinal Chemistry Letters 11 (2001) 1665–1669
Selective Inhibition of Src SH2 by a Novel Thiol-Targeting
Tricarbonyl-Modified Inhibitor and Mechanistic Analysis by
1H/13C NMRSpectroscopy
Raji Sundaramoorthi,* Chris Siedem, Chi B. Vu, David C. Dalgarno, Ellen C. Laird,
Martyn C. Botfield, Amanda B. Combs, Susan E. Adams, Ruth W. Yuan,
Manfred Weigele and Surinder S. Narula
ARIAD Pharmaceuticals, Inc., 26 Landsdowne Street, Cambridge, MA 02139, USA
Received 9 February 2001; accepted 12 April 2001
Dedicated to Professor Michael P. Cava, Robert Ramsay Professor of Chemistry, UAL, Tuscaloosa, AL, on his 75th birthday
Abstract—Detailed analysis of Src SH2 binding by peptides containing a novel tricarbonyl-modified pTyr moiety is described. We
envisaged that Src SH2 selectivity might be obtained by exploiting the thiol group of Cys188 present in the pTyr binding pocket of
the protein at the bC3 position. Peptidyl as well as non-peptidyl compounds 1–4 possessing a 4-a,b-diketoester-modified pTyr
mimic exhibited micromolar affinity to Src SH2. Furthermore, these tricarbonyl compounds were selective for Src SH2 to the extent
they showed no significant affinity for either Cys188Ser or Cys188Ala Src SH2 mutants. Upon closer examination of the binding of
these tricarbonyls to Src SH2 using NMR of 13C-labeled compounds (6a, 6b, and 6c), we found that after the initial binding event
the molecule disproportionated in a ‘retro-Claisen’ fashion to provide benzoic acid 16 and, following hydrolysis of the methyl ester
17, the hemiketal adduct of glyoxalic acid 18. # 2001 Elsevier Science Ltd. All rights reserved.
The Src family of tyrosine kinases play an important
role in regulating intracellular signalling. The SH2
domain binds to pTyr-containing proteins and peptides
in a sequence dependent manner,1 which provides the
basis for differentiation of the phosphorylation/depho-
sphorylation events surrounding a multitude of intra-
cellular signalling pathways.2 This intricate role of SH2
domains coupled with the postulated involvement of the
nonreceptor tyrosine kinase Src in various diseases pro-
vides an incentive to develop Src SH2 binding small
molecules as therapeutic agents to inhibit the signalling
activity associated with diseases such as cancer and
osteoporosis.3 The SH2 domain of Src differs from
other closely related SH2 domains in that it contains a
unique cysteine (Cys188) located in the pTyr binding
pocket. A Cys at the bC3 position is unique to Src, even
within the Src family, as all the other family members
possess Ser in this position. Careful examination of the
Src SH2 crystal structure reveals that this Cys associates
with the imidazole of an adjacent His, which should
serve to enhance the nucleophilicity of the Cys. Model-
ing predicted a 3.9 A gap into which a small electrophile
appropriately positioned might engage the thiol to
enhance binding affinity and selectivity. Synthesis of
reversible, high affinity ligands for Cys-proteases in
which the active site contains a reactive thiol has been
reported earlier.4 Although it is unlikely that the thiol of
Cys188 is fully ionized at neutral pH, the concentration
of positively-charged residues in the pocket is likely to
decrease the pKa. In fact, in an elegant study Singer and
Forman-Kay5 determined the pKs of His57 and pTyr in
a PLCg-1 C-terminal SH2 domain-pY1021 complex to
be significantly reduced due to the proximity of these
groups to the highly charged pTyr binding pocket.
Thus, a rational approach to the design of Src specific
ligands would involve engineering a pTyr mimic with an
appropriately positioned electrophile that could cova-
lently engage the reactive Cys. The synthesis of many
such Cys-targeted ligands is reported in the literature.5ꢀ9
Critical to the success of this approach for discovery of
viable drug candidates is that the electrophilic moiety of
the ligand be highly selective and not reactive to other
*Corresponding author. Fax: +1-617-494-8144; e-mail: raji.
sundaramoorthi@ariad.com
0960-894X/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(01)00269-4

1666
R. Sundaramoorthi et al. / Bioorg. Med. Chem. Lett. 11 (2001) 1665–1669
nucleophiles, particularly other thiols, or alternatively,
that the interaction with the thiol be reversible. The
present study shows a tricarbonyl-modified pTyr mimic
provides such interactions with Cys188 of Src SH2 (Fig.
1a and b).
synthesized from a common intermediate 7, which was
prepared from N-Boc-l-Tyr-OBn. Palladium mediated
carbonylation of the triflate 715 followed by hydrolysis
of the TMSE group with TBAF gave compound 8,
which was converted to the protected tricarbonyl com-
pound 9 using the Wasserman conditions.14 Hydro-
genolysis of the benzyl ester 9, followed by removal of
the Boc group with TFA and reprotection of this amine
with acylimidazole in DME gave 10. Coupling of the
carboxylic acid 10 with the respective amines containing
protected tripeptide R0NH2 gave 11a and 11b. Oxidative
deprotection of the protected tricarbonyl moiety by
ozonolysis in methanol gave 12a and 12b. Deprotection
of the benzyl ether of 12a gave 1 and deprotection of the
t-butyl ester of 12b gave 2. Compound 6 was synthe-
sized from benzoic acid.14 In a similar manner 3 and 4
were synthesized using Scheme 2, starting from 4-iodo
benzoic acid TMSE ester 13 for 3 and starting from 3-
iodobenzoic acid TMSE ester for 4 (not shown in
Scheme 2).
Hatada et al.10 determined the crystal structure of Src
SH2 at 1.5 A resolution and found that the protein
contained a well complexed citrate ion in the pTyr
binding pocket, when crystallized from citrate buffer.
The citrate was held by a number of hydrogen bonds to
the protein including both side-chain and main-chain
atoms. A vicinal tricarbonyl moiety could function as a
phosphate mimic when superimposed with the citrate
structure in the pTyr binding pocket of SH2 domain.
Independently, another group also claimed that the tet-
rahedral center of phenyl phosphate resembles the
hydrated form of the tricarbonyl compound.11 Mole-
cular modeling reported earlier12 also suggested that the
polar tricarbonyl moiety may occupy the phosphate
binding pocket of SH2 domains in a manner similar to
that of phosphonates and malonates.13 None of these
papers has delineated the fragile nature of the tricarbo-
nyls or the molecular basis of their mechanistic selec-
tivity. Here, we report our synthesis of the tripeptides
containing the tricarbonyl moiety and their selective
interaction with the Cys188 of Src SH2. Furthermore
we describe the mechanistic details of decompositon of
the tricarbonyl moiety that involve a glyoxalic acid
intermediate by analysis of the 13C-labeled tricarbonyl
compounds 6a–c by NMR.
Compounds 1–6 (Schemes 1 and 2 and Fig. 2) as well as
the tetrapeptide Ac-pYEEI-NH2 were analyzed in the
BIAcore binding assay16 against wild-type Src as well as
Cys188Ala Src SH2 mutant. Table 1 shows the IC50
values for these compounds against both these proteins.
All compounds exhibited a good affinity for wild-type
Src SH2 and were selective for Src SH2 relative to Fyn
SH217 as well as Cys188Ser17 and Cys188Ala Src SH2
mutants. They were reversible, suggesting that the elec-
tophilic carbonyl engaged the Cys of Src SH2. Ac-
pYEEI showed similar affinity to both wild-type Src as
well as Cys188Ala Src SH2 to further confirm the
selectivity of the tricarbonyls for Src SH2. Similarly, the
simple p-carboxy substituted tripeptide 5 did not show
any selectivity between wild-type Src SH2 and Cys188Ala
Src SH2. Binding affinities of these compounds were
To test the above hypothesis, p-tricarbonyl-substituted-
l-phenyl alanine containing tetrapeptides 1 and 2 were
synthesized according to Scheme 1 using the earlier
reported method of vicinal tricarbonyl synthesis by
Wasserman and others.11ꢀ14 Compounds 1 and 2 were
Figure 1. (a) A portion of 1H–15N HSQC spectrum of 15N labeled Src SH2 complexed with 6a and recorded at different time intervals is shown. The
spectrum recorded within 1 h of complex formation is shown in blue and those recorded at 24 and 72 h intervals are shown in red and green,
respectively. The horizontal and vertical axes represent 1H and 15N chemical shifts, respectively. The cross peaks shown in this portion represent the
backbone HN/N of the marked residues of the 15N labeled protein. (b) A portion of 1H–13C HSQC spectrum of double (15N, 13C) labeled Src SH2
complexed with 6c and recorded at 30 h shown in red is compared with uncomplexed protein shown in green. The horizontal and vertical axes
1
represent H and 13C chemical shifts, respectively. The marked peaks are arising from the CysCb–Hb resonances. Each cysteine is expected to give
two cross peaks in this spectrum.

R. Sundaramoorthi et al. / Bioorg. Med. Chem. Lett. 11 (2001) 1665–1669
1667
Scheme 1. (i) Pd(OAc)2, dppp, CO, TMSE, Et3N, 80 ꢁC, DMSO, 2 h; (ii) Bu4NF, THF, rt, 5 h; (iii) PPh3 ¼CHCOOMe, EDC, THF, DMAP, rt,
28 h; (iv) MeOH, Pd/C, H2, rt, 30 min; (v) TFA/CH2Cl2, rt, 30 min; (vi) N-acylimidazole, DME, Et3N, rt, 18 h; (vii) EDC/HOBt,CH2Cl2/DMF,
R00NH2, rt, 4.5 h; (viii) O3, MeOH, ꢀ78 ꢁC to rt, 30 min; (ix) For 12a, MeOH, Pd/C, H2, rt, 16 h: For 12b. TFA/CH2CH2Cl2 (0.5% water), rt, 30
min.
ꢁ
.
Scheme 2. (i) PdCl2, CH¼CHCOOBn, Bu4NCl 6H2O, DMF, 80 C, 2.5 h; (ii) Bu4NF, THF, rt, 5 h; (iii) PPh3¼CHCOOMe, EDC, THF, DMAP, rt,
28 h; (iv) MeOH, Pd/C, H2, rt, 30 min; (v) EDC, HOBt, CH2Cl2/DMF, R00NH2, rt, 4.5 h; (vi) O3, MeOH, ꢀ78 ꢁC to rt, 30 min; (vii) For 3, TFA/
CH2Cl2 (0.5% water), rt, 30 min; for 4, MeOH, Pd/C, H2, rt, 16 h.
Figure 2. Compounds synthesized besides the compounds in Schemes 1 and 2.
observed to be time-dependent. The tricarbonyl group
attached to the molecular recognition regions, namely 1
and 2, were found to bind better than the simple tri-
carbonyl compound 6. With the des-aminoacetyl 3 and
4, the binding to Src was one order of magnitude less,
and altering the tricarbonyl substitution from the 4- (3)
to 3-position (4) of the phenyl ring did not change the
binding significantly. This selectivity of the tricarbonyl
compounds to Src SH2 prompted us to investigate the
binding further by NMR. Three of the 13C labeled aro-
matic tricarbonyl compounds (viz. 6a–c) were synthe-
sized18 in order to study the reactions and the nature of

1668
R. Sundaramoorthi et al. / Bioorg. Med. Chem. Lett. 11 (2001) 1665–1669
Table 1. Comparative binding of tricarbonyl compounds to wild-
type Src SH2 and mutant Cys188Ala Src SH2
with 15N Src-SH2 is shown in blue in Figure 1a. A few
cross peaks arising from the backbone HN/N of Src are
1
shown marked with their residue number. 2-D H–15N
Compounds
WT Src
IC50 (mM)
Src Cys188Ala
IC50 (mM)
Selectivity
HSQC spectrum acquired in 24 h time frame is shown in
Figure 1a. The cross peaks that have moved from their
positions (absence of red cross peak from the equivalent
positions of blue cross peak) represent residues which
are affected by the complex formation. All these resi-
dues are a part of the phosphotyrosine binding pocket
of Src SH2. Noticeable among these residues are the
cross peaks of Glu181 (bC1), Cys188 (bC3), Leu189
(bC4), Tyr205 (bD5), Lys206 (bD6), Ile207 (bD7), and
Arg208 (bD1). The peak of Ser192 (bC7), which is not a
part of any pocket is not affected at all. Moreover, cross
peaks of Cys248 (bG3), and Cys241 (BG5 loop) are not
1
2
3
4
8.6
2.3
20
>1000
>1000
>1000
Yes
Yes
Yes
Yes
No
18
>1000
5
250
100
1.2
15% @500 mM
>1000
1.33
6
pYEEI
Yes
No
intermediate species formed. Compounds 6a–c were
subjected to NMR studies with single-labeled (15N) and
double-labeled (13C,15N) Src SH2.19
1
affected at all (Fig. 1b). The 2-D H–15N HSQC spec-
trum acquired at 72 h is shown in Figure 1a in green.
There are no significant differences in the spectra
acquired at 24 h (red) and 72 h (green). The 2-D 1H–15N
HSQC spectra were identical to those discussed for 6a.
1
1
The H, 13C, and 2-D H–15N HSQC NMR20 spectra
were acquired initially every 1 h to check the time
dependence in the binding event. The initial event was
the addition of the thiol of Cys in the protein to the
electrophilic carbon of 6, which was seen by the dis-
appearance of the peak at 91.2 ppm (13C NMR is
assigned to the only labeled carbon of 6a) (Fig. 3, Table
2). Further analysis indicated that this adduct under-
went a retro-Claisen rearrangement to form 16 and 17
in 24 h. Thus, the peak at 191.7 ppm, which was of a
keto group attached to the phenyl ring, disappears ear-
lier to give the carboxylic acid carbonyl at 173.0 ppm.
The appearance of peaks at 83.7 ppm by 13C NMR and
1
The 2-D H–13C HSQC spectrum was acquired on 6c
after 30 h of mixing with double-labeled (13C,15N) Src
SH2 and a part of the spectrum is shown in Figure 1bin
red. The spectrum compared to the uncomplexed Src
data is shown in green. The peaks corresponding to
CysCb-Hb for other Cys residues of Src SH2 (at Cys241
and Cys248) are still present, indicating clearly that only
Cys188 is targeted by the tricarbonyl compound 6c (6a
and 6b as well). Similar decompositions of the tricarbo-
nyl compounds were reported in the literature.21 This
disproportionation also occurred in aqueous solution in
1
5.40 ppm by H NMR indicates the presence of a sec-
Table 2. 1H and 13C NMR data for 6a–c with double-labeled (13
and 15N) Src SH2 with the rearranged products 16, 17, and 18
C
ondary alcohol of 17 and also confirms the retro-Claisen
rearrangement. At 70 h, the ester group was hydrolyzed
to the carboxylic acid 18 and methanol. Initially we
made the double-labeled (13C) tricarbonyl compound
6b, which provided almost all the information needed,
but to confirm the hydrolysis of the ester at pH 7.4 as a
function of time, we also synthesized the other two
Functional group
6a
6b
6c
16
17
18
-COOMe
-CHOH
-PhCO
-C(OH)2
–COOMe
3.77
3.77
3.77
3.82
5.40 5.10
169.0
191.7 191.7 173.0
91.2 91.2
168.0 168.0 168.0
1
compounds 6a and 6c. A portion of the 2-D H–15N
HSQC spectrum acquired within 1 h of mixing of 6a
Figure 3. Sequence of events when 13C labeled 6 was complexed with Src SH2 with the 1H/13C chemical shifts of each species. Blue: 1H NMR
chemical shifts in ppm; red: 13C NMR chemical shift in ppm.

R. Sundaramoorthi et al. / Bioorg. Med. Chem. Lett. 11 (2001) 1665–1669
1669
the presence of glutathione alone. Thus, it became clear
that tricarbonyls would not be physiologically viable
molecules. Nevertheless, in this paper we have demon-
strated that the tricarbonyl compounds showed the fea-
sibility of selectively inhibiting the Src SH2 with a
reversible thiol-targeting group
9. Shakespeare, W. C.; Bohacek, R. S.; Narula, S. S.; Azi-
moiara, M. D.; Yuan, R. W.; Dalgarno, D. C.; Madden, L.;
Botfield, M. C.; Holt, D. A. Bioorg. Med. Chem. Lett. 1999, 9,
3109.
10. Bohacek, R. S.; Dalgarno, D. C.; Hatada, M.; Jacobson,
V. A.; Lynch, B. A.; Macek, J. A.; Taylor, M.; Metcalf, C. A.,
III; Narula, S. S.; Sawyer, T. K.; Shakespeare, W. C.; Violette,
S. M.; Weigele, M. J. Med. Chem. 2001, 44, 660.
11. Fretz, H. Tetrahedron Lett. 1996, 37, 8475.
Acknowledgements
12. Rahuel, J.; Gay, B.; Erdmann, D.; Straus, A.; Garcia-
Echeverria, C.; Furet, P.; Caravatti, G.; Fretz, H.; Schoepfer,
J.; Grutte, M. G. Nat. Struct. Biol. 1996, 3, 586.
13. Relative to the phosphate group, the phosphonate group
occupies approximately 22%, difluoromethylene phosphonate
32%, dioxopropionate hydrate 52%, the 3-malonate ether
55%, a-methylenemalonate 73% more volume.
The authors thank Tomi K. Sawyer for his useful sug-
gestions in writing this manuscript. Our special thanks to
Ms. Chris Stearns for her help in editing this document.
References and Notes
14. Wasserman, H. H.; Vu, C. B. Tetrahedron Lett. 1990, 31,
5205.
1. Songyang, Z.; Shoelson, S. E.; Chaudhuri, M.; Gish, G.;
Pawson, T.; Neel, B. G.; Birge, R. B.; Fajardo, J. E.; Chou,
M. M.; Hanufusa, H.; Schaffhausen, B.; Cantley, L. C. Cell
1993, 72, 767. Thomas, S. M.; Brugge, J. S. Annu. Rev. Cell.
Dev. Biol. 1997, 13, 513.
15. Compounds 1–6 and all the intermediates were character-
ized by 300 MHz 1H NMR and MS. All compounds have been
shown to be homogeneous by TLC and HPLC.
16. Lynch, B. A.; Loiacono, K. A.; Tiong, C. L.; Adams, S. E.;
MacNeil, I. A. Anal. Biochem. 1997, 247, 77.
17. Data not shown for Fyn and Cys188Ser.
18. These three labeled compounds were synthesized from
appropriately labeled starting material as described in ref 15.
19. Buchanan, J. L.; Bohacek, R. S.; Luke, G. P.; Hatada, M.;
Lu, X.; Dalgarno, D. C.; Narula, S. S.; Yuan, R.; Holt, D.
Bioorg. Med. Chem. Lett. 1999, 9, 3985.
2. Pawson, T. Nature 1995, 373, 573. Cohen, G. B.; Ren, R.;
Baltimore, D. Cell 1995, 80, 237.
3. Sawyer, T. K. Biopolymers (Peptide Science) 1998, 47, 243.
Dalgarno, D. C.; Metcalf, C. A., III.; Shakespeare, W. C.;
Sawyer, T. K. Curr. Opin. Drug Dis. Develop. 2000, 3, 549.
4. Shaw, E. J. Protein Chem. 1984, 3, 109. Shaw, E. Adv.
Enzymol. Relat. Area Mol. Biol. 1990, 63, 271. Otto, H. H.;
Schirmeister, T. Chem. Rev. 1997, 97, 133.
5. Singer, A. U.; Forman-Kay, J. D. Protein Sci. 1997, 6,
1910.
6. Charifson, P. S.; Shewchuk, L. M.; Rocque, W.; Hummel,
C. W.; Jordon, S. R.; Moher, C.; Pacofsky, G. J.; Peel, M. R.;
Rodriguez, M.; Sternbach, D. D.; Counsler, T. G. Biochem-
istry 1997, 36, 6283.
7. Allligwood, K. J.; Charifson, P. S.; Crosby, T. G. C.;
Feldman, P. L.; Gampe, R. T.; Gilmer, T. M.; Jordan, S. R.;
Milstead, M. W.; Mohr, C.; Peel, M. R.; Rocque, W.; Rodri-
guez, M.; Rusnak, D. W.; Shewchuk, L. M.; Sternbach, D. D.
Bioorg. Med. Chem. Lett. 1998, 7, 1189.
20. NMR sample preparation: A typical NMR sample had
protein concentration of 0.5 mM and the concentration of
ligand varying from 0.5–1.5 mM. A sample volume of 0.5 mL
has 250 mM NaPi buffer of pH 7.4 in 10% D2O in water as
solvent. NMR Experiments: All the NMR data were acquired
on an AMX600 Bruker instrument equipped with triply tuned
probe. Proton NMR spectrum were recorded with a typical
spectrum width of 15 ppm and relaxation delay of 1.5 s. 13C
NMR spectrum were recorded with a typical spectrum width
1
of 236 ppm and relaxation delay of 3.0 s. 2-D H–15N HSQC
spectrum were recorded using the Watergate technique to get
rid of the huge water signal. The peaks were assigned with
reference to TMS in ppm.
8. Holt, D. A.; Veber, D. F.; Yamashita, D. S.; PCT Int.
Appl. WO9840093, 1998; Chem. Abstr. 1998, 129, 239893.
21. Rode-Gowal, H.; Dao, H. L.; Dahn, H. Helv. Chim. Acta
1974, 57, 240.