Calorimetric and structural studies of 1,2,3-trisubstituted cyclopropanes as conformationally constrained peptide inhibitors of Src SH2 domain binding

Article abstract of DOI:10.1021/ja011746f

Isothermal titration calorimetry and X-ray crystallography have been used to determine the structural and thermodynamic consequences associated with constraining the pTyr residue of the pYEEI ligand for the Src Homology 2 domain of the Src kinase (Src SH2

Full text of DOI:10.1021/ja011746f

Published on Web 12/15/2001
Calorimetric and Structural Studies of 1,2,3-Trisubstituted
Cyclopropanes as Conformationally Constrained Peptide
Inhibitors of Src SH2 Domain Binding
James P. Davidson,‡ Olga Lubman,§ Thierry Rose,§ Gabriel Waksman,*,§ and
Stephen F. Martin*,‡
Contribution from the Department of Chemistry and Biochemistry and The Institute of Cellular
and Molecular Biology, The UniVersity of Texas, Austin, Texas 78712, and Department of
Biochemistry and Molecular Biophysics, Washington UniVersity School of Medicine,
St. Louis, Missouri 63110
Received July 17, 2001
Abstract: Isothermal titration calorimetry and X-ray crystallography have been used to determine the
structural and thermodynamic consequences associated with constraining the pTyr residue of the pYEEI
ligand for the Src Homology 2 domain of the Src kinase (Src SH2 domain). The conformationally constrained
peptide mimics that were used are cyclopropane-derived isosteres whereby a cyclopropane ring substitutes
to the N-CR-Câ atoms of the phosphotyrosine. Comparison of the thermodynamic data for the binding of
the conformationally constrained peptide mimics relative to their equivalent flexible analogues as well as
a native tetrapeptide revealed an entropic advantage of 5-9 cal mol-1 K-1 for the binding of the
conformationally constrained ligands. However, an unexpected drop in enthalpy for the binding of the
conformationally constrained ligands relative to their flexible analogues was also observed. To evaluate
whether these differences reflected conformational variations in peptide binding modes, we have determined
the crystal structure of a complex of the Src SH2 domain bound to one of the conformationally constrained
peptide mimics. Comparison of this new structure with that of the Src SH2 domain bound to a natural
11-mer peptide (Waksman et al. Cell 1993, 72, 779-790) revealed only very small differences. Hence,
cyclopropane-derived peptides are excellent mimics of the bound state of their flexible analogues. However,
a rigorous analysis of the structures and of the surface areas at the binding interface, and subsequent
computational derivation of the energetic binding parameters, failed to predict the observed differences
between the binding thermodynamics of the rigidified and flexible ligands, suggesting that the drop in enthalpy
observed with the conformationally constrained peptide mimic arises from sources other than changes in
buried surface areas, though the exact origin of the differences remains unclear.
Introduction
One of the assumed benefits of preorganizing a ligand into the
biologically active conformation is that this will give rise to an
Oligopeptides have been commonly used as lead compounds
to design ligands or drugs with high affinity and specificity for
a particular binding site. A number of strategies have been
developed to achieve this goal, and one that has proven
particularly meritorious involves incorporating conformational
constraints into the oligopeptide. This approach allows the
topography of the binding site and the biologically active
conformation, or the bound structure, of the ligand to be
explored,1-7 and ligands exhibiting increased affinity or speci-
ficity for the receptor or enzyme active site can be identified.
entropic advantage in the binding event. This favorable entropy
of binding would eventuate in tighter binding, provided there
were no enthalpic penalties that arose from a loss of attractive
interactions or the introduction of unfavorable steric interactions
in the complex.
A survey of the vast majority of peptide mimics reveals that
most are dedicated to controlling backbone organization; few
are capable of orienting the amino acid side chains, which
contribute critical recognition elements for binding and
specificity.8-12 Thus, there is a general need for peptide
replacements that predictably constrain both the peptide back-
* To whom correspondence should be addressed. G.W.: phone, (314)
362-4562; fax, (314) 362-7183; e-mail, waksman@biochem.wustl.edu.
S.F.M.: phone, (512) 471-3915; fax, (512) 471-4180; e-mail, sfmartin@
mail.utexas.edu.
(6) Olson, G. L.; Bolin, D. R.; Bonner, M. P.; Bos, M.; Cook, C. M.; Fry, D.
C.; Graves, B. J.; Hatada, M.; Hill, D. E.; et al. J. Med. Chem. 1993, 36,
3039-3049.
‡ The University of Texas.
§ Washington University School of Medicine.
(7) Patani, G. A.; LaVoie, E. J. Chem. ReV. 1996, 96, 3147-3176.
(8) Boumendjel, A.; Roberts, J. C.; Hu, E.; Pallai, P. V.; Rebek, J., Jr. J. Org.
Chem. 1996, 61, 4434-4438.
(1) Babine, R. E.; Bender, S. L. Chem. ReV. 1997, 97, 1359-1472.
(2) Ball, J. B.; Alewood, P. F. J. Mol. Recognit. 1990, 3, 55-64.
(3) Giannis, A.; Kolter, T. Angew. Chem., Int. Ed. Engl. 1993, 32, 1244-
1267.
(9) Chen, L.; Trilles, R. V.; Tilley, J. W. Tetrahedron Lett. 1995, 36, 8715-
8718.
(4) Gibson, S. E.; Guillo, N.; Tozer, M. J. Tetrahedron 1999, 55, 585-615.
(5) Liskamp, R. M. J. Recl. TraV. Chim. Pays-Bas 1994, 113, 1-19.
(10) Hagihara, M.; Anthony, N. J.; Stout, T. J.; Clardy, J.; Schreiber, S. L. J.
Am. Chem. Soc. 1992, 114, 6568-6570.
9
10.1021/ja011746f CCC: $22.00 © 2002 American Chemical Society
J. AM. CHEM. SOC. VOL. 124, NO. 2, 2002 205

A R T I C L E S
Davidson et al.
contrast to original expectations, the potency of the conforma-
tionally constrained ligands were at best equal, not superior, to
their flexible peptide counterparts. Hence, the primary goal of
preparing tighter binding pseudopeptides through conformational
constraint was not achieved, and the fundamental question at
this juncture was “Why?” Did the introduction of the cyclo-
propane ring into the inhibitors not result in the expected
entropic advantage, or did unfavorable enthalpic factors such
as loss of binding contacts or steric interactions override the
entropic gains, thereby resulting in a similar free energy of
binding? Because there have been no studies that directly assess
the thermodynamic consequences (∆G°, ∆H°, and ∆S°) of pre-
organizing a ligand into its biologically active conformation,
we designed a set of calorimetric and structural experiments to
evaluate the consequences of introducing this localized confor-
mational constraint into pseudopeptides.
After considering a number of possibilities, we concluded
that complexes of Src Homology 2 (SH2) domains with
phosphotyrosyl-derived ligands would constitute an excellent
biological system in which to probe the effects of structure and
energetics of binding. SH2 domains are small protein domains
containing ∼100 amino acids that play critical roles in a variety
of signal transduction pathways.21-23 These domains specifically
bind to tyrosine-phosphorylated sites, thereby facilitating re-
cruitment of SH2 domain-containing proteins to these sites.24
The design of potent and selective SH2 domain-binding
inhibitors that selectively target these signaling processes has
been of keen interest in the pharmaceutical industry.25,26 One
notable success in this area has been the design of compounds
that are selective for the SH2 domain of the Src kinase (Src
SH2 domain) and that selectively inhibit bone resorption in
vivo.27-29
Figure 1. Conceptual design of 1,2,3-trisubstituted cyclopropane replace-
ments of peptides (A) and formulas of the peptides and pseudopeptides
used in this study (B). In (A), the rotors constrained by the substitution are
indicated by φ and ø1. In (B), compounds 3-7 are shown.
bone and the side chains in orientations that correspond to the
biologically active conformation of the peptide. Toward this
end, we designed a novel class of cyclopropane-derived dipep-
tide isosteres related to 2 (Figure 1A).13 The cyclopropane ring
in 2 replaces two atoms in the peptide backbone of the native
dipeptide 1 as well as the â-carbon of the amino acid (Yaa)
being replaced (Figure 1A). These unique replacements were
designed to orient both the peptide backbone and the amino
acid side chain by varying the stereochemistry on the cyclo-
propane ring.
To evaluate the efficacy of such replacements in biological
systems, cyclopropane-derived isosteres related to 2 were
introduced into inhibitors of renin, HIV-1 protease, matrix
metalloproteinases, and Ras farnesyltransferase, as well as
enkephalin analogues and fibrinogen receptor antagonists.14-20
These studies generally established the viability of introducing
trisubstituted cyclopropanes into biologically active analogues
of peptides and provided evidence that these replacements could
be used to probe the topography of the binding site. High affinity
ligands were identified in a number of cases; however, in
The phosphotyrosine-containing tetrapeptide Ac-pTyr-Glu-
Glu-Ile-OH (pYEEI) (3) is a well-known antagonist of the SH2
domains of the Src family of kinases.30 Examination of known
three-dimensional structures of pTyr ligands in complex with
Src family SH2 domains31-33 revealed that the cyclopropane-
derived analogues 4 and 5, which are conformationally con-
strained derivatives of 3, were well suited to probe thermody-
(21) Cantley, L. C.; Auger, K. R.; Carpenter, C.; Duckworth, B.; Graziani, A.;
Kapeller, R.; Soltoff, S. Cell 1991, 64, 281-302.
(22) Pawson, T.; Schlessinger, J. Curr. Biol. 1993, 3, 434-442.
(23) Sadowski, I.; Stone, J. C.; Pawson, T. Mol. Cell. Biol. 1986, 6, 4396-
4408.
(24) Pawson, T.; Scott, J. D. Science 1997, 278, 2075-2080.
(25) Brugge, J. S. Science 1993, 260, 918-919.
(26) Sawyer, T. K. Biopolymers 1998, 47, 243-261.
(11) Janetka, J. W.; Raman, P.; Satyshur, K.; Flentke, G.; Rich, D. H. J. Am.
Chem. Soc. 1997, 119, 441-442.
(27) Charifson, P. S. S., L. M.; Rocque, W.; Hummel, C. W.; Jordan, S. R.;
Mohr, C.; Pacofsky, G. J.; Peel, M. J.; Rodriguez, M.; Sternback, D. D.;
Consler, T. G. Biochemistry 1997, 36, 6283-6293.
(12) Smith, A. B., III; Guzman, M. C.; Sprengeler, P. A.; Keenan, T. P.;
Holcomb, R. C.; Wood, J. L.; Carroll, P. J.; Hirschmann, R. J. Am. Chem.
Soc. 1994, 116, 9947-9962.
(28) Shakespeare, W.; Yang, M.; Bohacek, R.; Cerasoli, F.; Stebbins, K.;
Sundaramoorthi, R.; Azimioara, M.; Vu, C.; Pradeepan, S.; Metcalf, C.;
Haraldson, C.; Merry, T.; Dalgarno, D.; Narula, S.; Hatada, M.; Lu, X. D.;
van Schravendijk, M. R.; Adams, S.; Violette, S.; Smith, J.; Guan, W.;
Bartlett, C.; Herson, J.; Iuliucci, J.; Weigele, M.; Sawyer, T. Proc. Natl.
Acad. Sci. U.S.A. 2000, 97, 9373-9378.
(13) Martin, S. F.; Austin, R. E.; Oalmann, C. J. Tetrahedron Lett. 1990, 31,
4731-4734.
(14) Hillier, M. C.; Davidson, J. P.; Martin, S. F. J. Org. Chem. 2001, 66, 1657-
1671.
(15) Martin, S. F.; Austin, R. E.; Oalmann, C. J.; Baker, W. R.; Condon, S. L.;
deLara, E.; Rosenberg, S. H.; Spina, K. P.; Stein, H. H.; Cohen, J.; Kleinert,
H. D. J. Med. Chem. 1992, 35, 1710-1721.
(29) Violette, S. M.; Shakespeare, W. C.; Bartlett, C.; Guan, W.; Smith, J. A.;
Rickles, R. J.; Bohacek, R. S.; Holt, D. A.; Baron, R.; Sawyer, T. K. Chem.
Biol. 2000, 7, 225-235.
(16) Martin, S. F.; Dorsey, G. O.; Gane, T.; Hillier, M. C.; Kessler, H.; Baur,
M.; Mathae, B.; Erickson, J. W.; Bhat, T. N.; Munshi, S.; Gulnik, S. V.;
Topol, I. A. J. Med. Chem. 1998, 41, 1581-1597.
(30) Zhou, S. Y.; Shoelson, S. E.; Chaudhuri, M.; Gish, G.; Pawson, T.; Haser,
W. G.; King, F.; Roberts, T.; Ratnofsky, S.; Lechleider, R. J.; Neel, B. G.;
Birge, R. B.; Fajardo, J. E.; Chou, M. M.; Hanafusa, H.; Schaffhausen, B.;
Cantley, L. C. Cell 1993, 72, 767-778.
(17) Martin, S. F.; Dwyer, M. P.; Hartmann, B.; Knight, K. S. J. Org. Chem.
2000, 65, 1305-1318.
(31) Waksman, G.; Kominos, D.; Robertson, S. C.; Pant, N.; Baltimore, D.;
Birge, R. B.; Cowbrun, D.; Hanafusa, H.; Mayer, B. J.; Overduin, M.; Resh,
M. D.; Rios, C. B.; Silverman, L.; Kuryian, J. Nature 1992, 358, 646-
653.
(18) Martin, S. F.; Oalmann, C. J.; Liras, S. Tetrahedron 1993, 49, 3521-3532.
(19) Baker, W. R.; Jae, H. S.; Martin, S. F.; Condon, S. L.; Stein, H. H.; Cohen,
J.; Kleinert, H. D. Bioorg. Med. Chem. Lett. 1992, 2, 1405-1410.
(20) Bovy, P. R.; Tjoeng, F. S.; Rico, J. G.; Rogers, T. E.; Lindmark, R. J.;
Zablocki, J. A.; Garland, R. B.; McMackins, D. E.; Dayringer, H.; et al.
Bioorg. Med. Chem. 1994, 2, 881-895.
(32) Waksman, G.; Shoelson, S. E.; Pant, N.; Cowburn, D.; Kuriyan, J. Cell
1993, 72, 779-790.
(33) Eck, M. J.; Shoelson, S. E.; Harrison, S. C. Nature 1993, 362, 87-91.
9
206 J. AM. CHEM. SOC. VOL. 124, NO. 2, 2002

Studies of 1,2,3-Trisubstituted Cyclopropanes
A R T I C L E S
PAGE. The protein was then dialyzed twice (24 h, 4 °C, 2 L of buffer)
against the standard calorimetry buffer (20 mM HEPES, pH 7.3, 1 mM
â-mercaptoethanol, 1 mM EDTA, 100 mM NaCl). The published
extinction coefficient of 14 700 M-1 cm-1 was used to determine the
final protein concentrations for the calorimetry experiments.36 Typical
yields for this purification procedure were 40-50 mg/L of culture.
Syntheses of Phosphotyrosines 3-7. Compounds 3-7 were
synthesized as reported previously by Davidson and Martin,34 and
complete experimental details for the syntheses of 3-7 are included
in the Supporting Information. The experimentally determined extinction
coefficient of 775 M-1 cm-1 at 268 nm was used to determine the
final ligand concentrations after dissolution in the standard calorimetry
buffer, pH adjustment, and filtration.
namic and structural consequences of the cyclopropane-derived
conformational constraint. Because 4 and 5 lack the tyrosine
amide nitrogen atom of 3, the pseudopeptides 6 and 7 were
identified as the necessary flexible controls. Compounds 4-7
were prepared by standard synthetic methods and initially
screened in competitive binding experiments to evaluate the
relative affinities of these compounds for the SH2 domain of
the Src family kinase Lck (Lck SH2 domain).34 These results
indicated that both of the constrained ligands 4 and 5 bound to
the SH2 domain with higher affinity than their flexible
counterparts 6 and 7, but the energetic differences were small
(<0.6 kcal mol-1); the native tetrapeptide 3 and the synthetic
pseudopeptide 4 were equipotent.
Isothermal Titration Calorimetry. Calorimetry experiments were
performed with an MCS titration calorimeter obtained from Microcal
Inc. (Northhampton, MA) as described in detail by Wiseman.45 Protein
and ligand solutions were degassed with stirring under reduced pressure
for 15 min prior to experiments. For a typical titration experiment, 50-
75 µM Src SH2 domain was placed in the 1.35 mL reaction cell, and
the 0.6-0.8 mM phosphotyrosine mimic was loaded into the 250 µL
injection syringe. Phosphotyrosine mimics were injected in 8-12 µL
increments. At least five injections were typically performed after
saturation was observed. All values in the results section are the result
of at least three independent titration experiments. The data for each
titration were collected by the ORIGIN software provided with the
calorimeter, and the titration curves were fit using the same software
to give ∆H° and Ka. For each ligand, at each experimental temperature,
a blank was run where ligand was injected into buffer alone to establish
the nonzero heat of dilution for the ligand. This heat was subtracted
from the raw titration data prior to fitting. The integrated data from all
experiments fit the single-site binding model with the stoichiometry
of binding being between 0.95 and 1.15 for all titrations. The c value,
defined as the product of the association constant Ka and the
macromolecule concentration M, was within the range of 1-1000, as
per the manufacturer’s recommendations for the determination of
accurate binding constants, for all experiments.45
Crystallization and Data Collection. Crystals of the Src SH2
domain complexed with compound 4 were grown by the hanging drop
method with 40 mg mL-1 solutions of 1:2 molar ratio of protein to
compound complex equilibrated against a reservoir containing 50%
MPD and 0.1 M HEPES, pH 8.0. The crystals diffracted to a resolution
of 1.9 Å in the laboratory setting (Rigaku Raxis IV image plate mounted
on a Rigaku RU200 rotating anode generator). A complete data set to
a resolution of 1.9 Å (Table 1) was collected using a single crystal
with an oscillation range of 1.0° and an exposure time of 30 min/frame.
Crystals were in the orthorhombic space group P212121, with cell
dimensions a ) 58.3 Å, b ) 56.5 Å, and c ) 69.3 Å, and two molecules
per asymmetric unit. All X-ray diffraction data were reduced using
the program DENZO and scaled using SCALEPACK.46
The observation that the constrained ligands 4 and 5 bound
to the Lck SH2 domain better than 6 and 7, respectively,
supported the original premise that introducing a conformational
constraint enhanced binding. Elucidating the energetic origin
of these differences, however, required calorimetric studies to
determine the complete thermodynamic profiles (∆∆G°, ∆∆H°,
∆∆S°, and ∆∆Cp) for forming the different protein-ligand
complexes. Unfortunately, some of the regions of the Lck SH2
domain that are intimately involved in phosphotyrosine binding
have been shown to undergo significant conformational changes
upon binding of phosphotyrosine-containing ligands similar to
3-7.35 Because such structural changes would complicate the
interpretation of the calorimetric data, we turned our atten-
tion to the Src SH2 domain, which was known not to under-
go these conformational changes upon binding.31,32,36 The
Src SH2 domain has been the subject of numerous structural
and calorimetric studies that have led to a detailed knowledge
of the binding energetics (∆G°, ∆H°, ∆S°, and ∆Cp) asso-
ciated with these protein-ligand interactions, thus providing a
broad knowledge base from which to launch our investiga-
tions.27,31,32,36-44
In this account, we report the thermodynamic profiles for the
binding of the conformationally constrained phosphotyrosyl
peptides 4 and 5, their flexible analogues 6 and 7, and the
tetrapeptide 3 to the Src SH2 domain. These profiles are
analyzed in the context of the crystal structures of the known
complex of the Src SH2 domain, an 11-mer natural tyrosyl
phosphopeptide containing the pYEEI motif,32 and a new
structure of the same SH2 domain bound to the conformationally
constrained pseudopeptide 4.
Experimental Section
Protein Expression and Purification. The Src SH2 domain was
expressed and purified as described previously by Waksman et al.31,36
The purity of the resulting material was assessed to be >98% by SDS-
Structure Determination and Refinement. The structure of the
Src SH2 domain-compound 4 complex was solved by the molecular
replacement method, using the program AMoRe.47 The search model
consisted of the coordinates of the Src SH2 domain from the refined
structure of the wild-type Src SH2 domain bound to the 11-mer peptide
EPQpYEEIPIYL (RCSB ID code 1sps)32 with the peptide and water
molecules omitted. The molecular replacement search yielded a clear
solution peak with high correlation coefficient (45.8) and low R-factor
(40.6%). The structure was refined using simulated annealing in
torsional angle space using the program CNS.48,49 Model inspection
and manual rebuilding were carried out in simulated annealing omit
maps using the program O.50,51 After bulk solvent correction, the
refinement converged to a final R-factor of 23.6% and free R-factor of
(34) Davidson, J. P.; Martin, S. F. Tetrahedron Lett. 2000, 41, 9459-9464.
(35) Tong, L.; Warren, T. C.; Lukas, S.; Schembri-King, J.; Betageri, R.;
Proudfood, J. R.; Jakes, S. J. Biol. Chem. 1998, 273, 20238-20242.
(36) Bradshaw, J. M.; Grucza, R. A.; Ladbury, J. E.; Waksman, G. Biochemistry
1998, 37, 9083-9090.
(37) Bradshaw, J. M.; Mitaxov, V.; Waksman, G. J. Mol. Biol. 2000, 299, 521-
535.
(38) Bradshaw, J. M.; Waksman, G. Biochemistry 1998, 37, 15400-15407.
(39) Bradshaw, J. M.; Waksman, G. Biochemistry 1999, 38, 5147-5154.
(40) Bradshaw, J. M.; Mitaxov, V.; Waksman, G. J. Mol. Biol. 1999, 293, 971-
985.
(41) Grucza, R. A.; Bradshaw, J. M.; Futterer, K.; Waksman, G. Med. Res. ReV.
1999, 19, 273-293.
(42) Grucza, R. A.; Bradshaw, J. M.; Mitaxov, V.; Waksman, G. Biochemistry
2000, 39, 10072-10081.
(43) Henriques, D. A.; Ladbury, J. E.; Jackson, R. M. Protein Sci. 2000, 9,
1975-1985.
(45) Brandts, J. F.; Wiseman, T.; Williston, S. Anal. Biochem. 1989, 179.
(46) Otwinowski, Z.; Minor, W. Macromol. Crystallogr., Part A 1997, 276,
307-326.
(44) Xu, R. X. W., M. J.; Davis, D. G.; Rink, M. J.; Willard, D. H.; Gampe, R.
T. Biochemistry 1995, 34, 2107-2121.
(47) Navaza, J. Acta Crystallogr., Sect. A 1994, 50, 157-163.
9
J. AM. CHEM. SOC. VOL. 124, NO. 2, 2002 207

A R T I C L E S
Davidson et al.
Table 1. Data Collection and Refinement Statistics
Data Collection
total/unique
reflections
radiation
resolution
completenessa (%)
R
sym (%)b
Cu KR
30-1.9 Å
101 439/18 121
Refinement
97.0 (95.3)
0.031 (0.11)
f
rms deviations
angles (deg)
1.8
number of
total number
of atomsd
resolution (Å)
reflectionsc
R-factor/Rfree (%)e
bonds (A)
B-values
30.0-1.9
16 124 (86.5/82.1%)
1873
23.6/26.7
0.013
1.2(main chain)
1.4(side chain)
b
a Completeness for I/σ(I) > 1, high-resolution shell (1.97-1.9 Å) in parentheses. Rsym )∑|I - I |/∑I, where I is the observed intensity, and I is the
average intensity from multiple observations of symmetry related reflections; high resolution (1.97-1.9 Å) shell in parentheses. c Numbers reflect “working”
e
set of reflections at F/σ(F) > 2.0, overall/last shell (1.97-1.9) completeness in parentheses. d Includes 189 water molecules. Rfree was calculated on the
basis of 10% of the total number of reflections randomly omitted from the refinement. f Deviations from ideal bond lengths and angles and in B factors of
bonded atoms.
26.7%. The refined structure contains 207 amino acids in the two protein
molecules, two molecules of compound 4, and 189 water molecules.
Details on data collection and refinement statistics are presented in
Table 1. The PDB entry code for this structure is 1IS0.
The change in enthalpy ∆H° was evaluated at 60 °C using the following
formula from Baker and Murphy:52
∆H°333K ) 29.19∆ASApolar - 7.27∆ASAnonpolar
Structure-Based Thermodynamic Calculation. The protein-
peptide binding thermodynamic parameters were calculated from
changes in solvent-accessible surface areas (∆ASA) during intermo-
lecular association according to the method of Baker and Murphy as
applied to the Src SH2 domain-peptide interaction by Henriques et
al.43,52 The empirical binding energies were evaluated for compound 4
from the crystal structure of the two Src SH2 domain-compound 4
complexes present in the asymmetric unit of the crystals described
herein. The structure of the Src SH2 domain complexes bound to
compounds 3 and 6 were modeled from the structure of the Src SH2
domain cocrystallized with the 11-mer peptide EPQYEEIPIYL.32 The
asymmetric unit of these crystals contained three such complexes, and
all three were used as starting point templates to build compounds 3
and 6 (program INSIGHTII, MSI, San Diego, CA). These models were
optimized by a four-step protocol using the program Discover and the
force field CFF91 (MSI, San Diego, CA) with no cut off and a dielectric
constant fixed to 1 × r. Potential energy minimization was applied
successively (1) to peptide side chains only, using 100 cycles of steepest
descent minimization, (2) to peptide and protein side chains, using 1000
cycles of conjugate gradient minimization, (3) to all the peptide atoms
and protein side chains, as in step 2, and (4) then to all atoms except
the Src SH2 backbone atoms located 10 Å or more away from the
peptide, as in step 2. Root mean square (rms) deviation calculations
between final and starting models and stereochemistry validation using
the program PROCHECK were systematically performed.53
and extrapolated at 25 °C using the following:
∆Cp,calc ) (∆H°298K - ∆H°333K)/(333 - 298)
∆H°298K ) ∆Cp,calc/(333 - 298) + ∆H°333K
The changes in entropy were calculated as the sum of the conforma-
tional entropy ∆S°conf, the association entropy ∆S°asso, and the solvation
entropy ∆S°solv defined as follows:
∆S°conf
)
n(∆ASASC,i/ASAAXA-SC,i)S°BuEx
+
i
∑
n(∆ASABB,i/ASAAXA-BB,i)S°BB
+
i
∑
n (∆ASABB,i/ASAAXA-BB,i)S°Ex-U
i
∑
where ∆ASASC,i or ∆ASABB,i are the changes in ASA of the side chain
or the backbone of residue i on binding. ASAAXA-SC,i or ASAAXA-BB,i
are the ASA of the side chain or the backbone of residue i in an
extended Ala-X-Ala tripeptide.43,52 Values for the backbone entropy
S°BB and S°Ex-U were from D’Aquino et al.,54 and the values for side-
chain entropy (S°BuEx) were from Lee et al.55
∆S°asso ) -7.89 cal K-1 mol-1
∆S°solv ) ∆Cp,calc ln(298/385)
∆ASAs were calculated using a 1.4 Å radius probe and the following
definitions:
Finally the binding free energy was computed from ∆G° ) ∆H° -
T∆S°.
∆ASAtotal ) ASASrc SH2:peptide - ASASrc SH2 - ASApeptide
∆ASA ) ∆ASApolar + ∆ASAnonpolar
Results
Thermodynamic Profiles for Complexes of Src SH2
Domain with Compounds 3-7. Compounds 4-7 are peptide
mimics that contain rigid structural replacements for the
phosphotyrosine (pTyr). For example, compounds 4 and 5
contain a cyclopropane ring as a replacement for the N-CR-Câ
atoms of the pTyr in the parent peptide compound 3. In
compounds 6 and 7, the amide nitrogen of pTyr in compound
ASASrc
is calculated from the structure of the complex where the
SH2
ligand has been removed. The changes in heat capacity ∆Cp,calc were
calculated from the area variations upon binding of the polar and
nonpolar water accessible surfaces according to Baker and Murphy:52
∆Cp,calc ) -0.26∆ASApolar + 0.43∆ASAnonpolar
(51) Jones, T. A.; Zou, J. Y.; Cowan, S. W.; Kjeldgaard, M. Acta Crystallogr.,
Sect. A 1991, 47, 110-119.
(48) Brunger, A. T.; Adams, P. D.; Clore, G. M.; DeLano, W. L.; Gros, P.;
Grosse-Kunstleve, R. W.; Jiang, J. S.; Kuszewski, J.; Nilges, M.; Pannu,
N. S.; Read, R. J.; Rice, L. M.; Simonson, T.; Warren, G. L. Acta
Crystallogr., Sect. D 1998, 54, 905-921.
(49) Rice, L. M.; Brunger, A. T. Proteins 1994, 19, 277-290.
(50) Hodel, A.; Kim, S. H.; Brunger, A. T. Acta Crystallogr., Sect. A 1992, 48,
851-858.
(52) Baker, B. M.; Murphy, K. P. Methods Enzymol. 1998, 295, 294-315.
(53) Laskowski, R. A.; Moss, D. S.; Thornton, J. M. J. Mol. Biol. 1993, 231,
1049-1067.
(54) D’Aquino, J. A.; Gomez, J.; Hilser, V. J.; Lee, K. H.; Amzel, L. M.; Freire,
E. Proteins 1996, 25, 143-156.
(55) Lee, K. H.; Xie, D.; Freire, E.; Amzel, L. M. Proteins 1994, 20, 68-84.
9
208 J. AM. CHEM. SOC. VOL. 124, NO. 2, 2002

Studies of 1,2,3-Trisubstituted Cyclopropanes
A R T I C L E S
Table 2. Thermodynamic Parameters Determined at 25 °C and Change in Heat Capacity Valuesa
compound
K
a
∆G°obs (kcal mol-1
)
∆H°obs (kcal mol-1
)
∆S°obs (cal mol-1 K-1
)
∆Cp,obs (cal mol-1 K-1
)
3
4
6
5
7
4.1 ((0.1) × 106
1.0 ((0.1) × 107
1.7 ((0.6) × 107
6.3 ((0.6) × 106
1.4 ((0.1) × 107
-9.01 ( 0.01
-9.55 ( 0.07
-9.8 ( 0.2
-9.26 ( 0.06
-9.72 ( 0.06
-6.06 ( 0.05
-5.91 ( 0.04
-7.33 ( 0.03
-5.01 ( 0.05
-6.92 ( 0.09
9.9 ( 0.2
17 (1
8.3 (0.5
14.3 ( 0.4
9.4 ( 0.2
-225 ((9)
-213 ((17)
a Errors represent the standard deviations between multiple measurements.
3 is substituted with a methylene, CH2, group. The additional
carbon-carbon bond in the pTyr replacements of 4 and 5
introduce conformational constraints that are absent in the more
flexible succinate derivatives 6 and 7. The other difference is
that the N-terminal acetyl (CH3CO-) group in 3 is replaced
with a methyl amide (CH3NHCO-) or dimethyl amide [(CH3)2-
NCO-] group (Figure 1B). Owing to their structural similarities,
any observed differences in binding energetics for the con-
strained compounds 4 and 5 relative to the corresponding
flexible counterparts 6 and 7 would be expected to result solely
from the conformational constraints imposed by the cyclopro-
pane ring.
The energetics of binding of compounds 3-7 were deter-
mined by isothermal titration calorimetry. The experimentally
determined and calculated thermodynamic data for these
compounds are summarized in Table 2. All compounds in this
study bound to the Src SH2 domain at 25 °C with a favorable
enthalpy (∆H° < 0) and favorable entropy (∆S° > 0) as shown
in Table 2. This trend was expected on the basis of previous
studies of SH2 domain-phosphopeptide interactions.36,56,57 Each
of the pseudopeptides 4-7 bound to the Src SH2 domain with
slightly higher affinity than the parent tetrapeptide 3 (Table 2).
Neither cyclopropane-containing compound 4 nor 5 exhibited
a higher affinity than the flexible analogue 6 or 7, respectively,
as had been observed previously.34 The tetrapeptide 3 binds with
only a 0.5 kcal mol-1 less favorable free energy change than
an 8-mer peptide having the sequence Ac-PQpYEEIPI-NH2
under similar conditions.36 Hence, residues beyond the pYEEI
motif do not appear to make significant contributions to the
binding free energy.32
The thermodynamic profiles for the binding of both cyclo-
propane-containing ligands 4 and 5 showed a significant entropic
advantage (∆∆S° ) 5-9 cal mol-1 K-1) over the tetrapeptide
3 and their flexible analogues 6 and 7. The favorable ∆∆S°
corresponds approximately to that predicted for restricting two
“rotors” (ø1 and Φ in Figure 1A).58,59 This result supports the
hypothesis that the preorganization of a ligand in its active
conformation does give rise to a favorable entropic contribution
to binding. Both cyclopropane-containing compounds 4 and 5
bound to the SH2 domain with significantly less favorable
enthalpies of binding (∆∆H° ) 1.4-1.9 kcal mol-1 relative to
their flexible analogues 6 and 7). Thus, the introduction of the
conformational constraint resulted in significant differences in
both the enthalpy and the entropy changes of binding without
effecting ∆G° significantly.
Figure 2. Temperature dependence of the binding enthalpy. ∆Hobs versus
temperature was evaluated for compounds 4 (black square) and 6 (blue
triangle). Error bars on the enthalpy values represent 95% confidence
intervals. A linear least-squares best fit of the compounds 4 and 6 enthalpy
data gives the heat capacity change for each compound, which are listed in
the text and Table 2. The uncertainty in the heat capacity values represents
a 95% confidence interval on the best linear fit.
solvation-desolvation processes during complex formation, the
changes in heat capacities (∆Cp,obs) for compounds 4 and 6 were
determined.60,61 Titrations were carried out at intervals of 5 °C
at temperatures from 15 to 35 °C. When ∆H°obs is plotted versus
temperature (Figure 2), the plot is linear, and the slope of the
line is ∆Cp,obs for the binding event. The ∆Cp,obs values for the
binding of each compound were found to be identical within
experimental error: 4, -225 ((9); 6, -213 ((17) cal mol-1
K-1 (Table 2). Hence, differences in the entropic contributions
do not appear to arise from differences in solvation-desolvation
or hydrophobic effects, but rather they are likely due to the
preorganization and restriction of rotors by the cyclopropane.
Because these values of ∆Cp,obs are very similar to that observed
for the binding of an 8-mer pYEEI-containing peptide to this
Src SH2 domain,36 additional sequences at the N- or C-termini
of the pYEEI motif do not appear to contribute significantly to
∆Cp,obs
.
Crystal Structure of the Src SH2 Domain-Compound 4
Complex. The difference of 1.4 kcal mol-1 in ∆H°obs between
compounds 4 and 6 suggested that these two compounds might
bind to the Src SH2 domain differently. We therefore attempted
to crystallize complexes of the Src SH2 domain with both
compounds. Although crystals of a complex of the Src SH2
domain bound to compound 4 were readily obtained, we were
unable to crystallize the Src SH2 domain bound to compound
6. Data for the complex of 4 with the Src SH2 domain were
collected to 1.9 Å resolution. Molecular replacement using a
search model of the Src SH2 domain obtained previously was
used to solve the structure that was subsequently refined to a
To probe whether incorporating the cyclopropane ring would
contribute significantly to the hydrophobic interactions or the
(56) Ladbury, J. E.; Lemmon, M. A.; Zhow, M.; Green, J.; Botfield, M. C.;
Schlessinger, J. Proc. Natl. Acad. Sci. U.S.A. 1995, 92, 3199-3203.
(57) Lemmon, M. A.; Ladbury, J. E. Biochemistry 1994, 33, 5070-5076.
(58) Jencks, W. P.; Page, W. I. Proc. Natl. Acad. Sci. U.S.A. 1971, 68, 1678-
1683.
(60) Robertson, A. D.; Murphy, K. P. Chem. ReV. 1997, 97, 1251-1267.
(61) Stites, W. E. Chem. ReV. 1997, 97, 1233-1250.
(59) Williams, D. H.; Searle, M. S. J. Am. Chem. Soc. 1992, 114, 10690-10697.
9
J. AM. CHEM. SOC. VOL. 124, NO. 2, 2002 209

A R T I C L E S
Davidson et al.
final R-factor of 23.6% and a final free R-factor of 26.7%
(resolution range: 30-1.9 Å; |F|/σ(F) > 2). The crystals
contained two molecules in the asymmetric unit, yielding two
independent views of the complex of the Src SH2 domain and
compound 4. The two complexes align with an rms deviation
of 1.5 Å for all protein atoms and 1.0 Å for backbone atoms.
The ligands align with an rms deviation of 1.2 Å for all atoms.
The significant deviation between the ligand structures in the
two complexes is due to differences in side-chain conformations
at the +1 and +2 position. When these side chains are excluded
from the calculation, the rms deviation between ligand atoms
is only 0.3 Å.
An overlay of the two structures from the asymmetric unit
of the Src SH2 domain-4 complex crystals is depicted in Figure
3A. A close-up overlay of the two bound structures of compound
4 is shown in Figure 3B. As shown in these figures, no
significant differences are observed in the binding of the +3
Ile and pTyr residues between these two structures. The
differences between the two complexes in the asymmetric unit
of the crystal lie in the positioning of the first (+1 Glu) and
second (+2 Glu) glutamate residues C-terminal to the pTyr in
the ligand and in some of the residues in the protein interacting
with the side chains of these two amino acids.
Differences between the two complexes at the +2 position
are significant; in one molecule, the side chain of the +2 Glu
interacts directly with Arg âD′1 (Figures 3A and 7), whereas
in the second molecule, the +2 Glu side chain interacts with
residues in a symmetry related molecule, and Arg âD′1 interacts
instead with Thr BC3 in the pTyr-binding loop (the BC loop).
In the crystal structure of the Src SH2 domain bound to an 11-
mer peptide containing the pYEEI motif,32 the +2 Glu was
involved in stabilizing a water network extending to Arg âD′1
located 5 Å away. This water network was originally interpreted
as important for determining the preferential recognition of Glu
at the +2 position of the ligand by the Src SH2 domain.30,32 In
the two complexes of the Src SH2 domain bound to compound
4, no such water network is observed.
Figure 3. Superposition of the structures of the two Src SH2 domain-
compound 4 complexes present in the asymmetric unit of the crystal. (A)
Superposition of the two complexes. The protein backbone is shown in
ribbon representation. The ligand and the residues of the protein involved
in contacts with the ligand are shown in thick and thin ball-and-stick
representation, respectively. The yellow and magenta protein ribbon models
correspond to models 4.1 and 4.2 of Table 3, respectively. Color-coding
for atoms in the ligand and the protein are the following: red for oxygen,
blue for nitrogen, either light (model 4.1) or dark (model 4.2) gray for
carbon, and pink for phosphorus. Protein residues are labeled according to
previous convention.32 Labeling of the pseudopeptide is PTR for the
cyclopropane substituted phosphotyrosine, +1 for +1 Glu, +2 for +2 Glu,
and +3 for +3 Ile. (B) Superposition of the two compound 4 structures
from the two complexes in the asymmetric unit of the crystals. The structures
corresponding to models 4.1 and 4.2 of Table 3 are shown in magenta and
green, respectively. (C) Interaction of the Lys âD3 and Asp âC8 and Asp
CD2. The pseudopeptide is represented as in (A). Lys âD3 is in blue, and
the two aspartates are in red. The +1 position in the peptide is labeled +1
Glu, while the rigidified phosphotyrosine is labeled PTR.
The differences at the +1 Glu position are more localized
and involve only the orientation of the Glu side chain of the
ligand. Namely, the +1 Glu side chain of 4 adopts at least two
different conformations as shown in Figure 3A and B, indicating
that this side chain may interact weakly with the protein. From
the interactions observed in the original crystal structure of the
11-mer peptide bound to the Src SH2 domain, it had been
hypothesized that the Glu +1 side chain interacts specifically
with the side chain of Lys âD3, which is the basic residue closest
to the carboxylic group of +1 Glu. Interestingly, the conforma-
tion of Lys âD3 in this new structure is nearly identical in both
complexes within the asymmetric unit, and the lysine ammonium
group is directed toward a region of the Src SH2 domain that
contains a patch of aspartate residues (Asp CD2 and Asp âC8)
rather than toward +1 Glu (Figure 3C).
Comparison of Complexes of SH2 Domains with Rigidi-
fied and Flexible Ligands. Although the complex of the Src
SH2 domain bound to the flexible peptide mimic 6 could not
be crystallized, some qualitative insights regarding whether this
flexible analogue binds in a mode different from the confor-
mationally constrained peptide mimic may be drawn by
comparing the structure of the complexes of the Src SH2 domain
bound to compound 4 and to the 11-mer peptide that was
reported previously.32 This should be a valid comparison because
only the four residues comprising the pTyr and the EEI sequence
immediately C-terminal to pTyr made contact with the SH2
domain in the latter complex.32 Hence, residues outside the
pYEEI motif are not expected to significantly affect the
conformation of the pYEEI sequence.
The crystal structure of the Src SH2 domain bound to the
11-mer peptide contained three complexes in the asymmetric
unit, so three independent views of this complex were obtained.
9
210 J. AM. CHEM. SOC. VOL. 124, NO. 2, 2002

Studies of 1,2,3-Trisubstituted Cyclopropanes
A R T I C L E S
Figure 5. Arginine residues of the phosphotyrosine binding pocket;
representation and labeling of residues in the peptide and in the protein are
as in Figure 4A.
pTyr residue in SH2 domain binding ligands. The only
significant differences found in the bound structures of 4 and
the 11-mer peptide were in the orientations of the +1 and +2
Glu side chains. However, this is not surprising given that
similar variations are observed in the orientations of these side
chains of 4 and the 11-mer peptide in different complexes within
asymmetric units.
The structures of the SH2 domain in each of these complexes
are largely identical, but there is one notable difference in the
pTyr binding pockets of the two. The constrained pTyr residue,
labeled PTR, and the two Arg residues in the pTyr binding
pocket, labeled Arg RA2 and Arg âB5, are depicted in Figure
5. The Arg residues in one of the complexes of 4 with the Src
SH2 domain are shown in gray, while the Arg residues in one
of the Src SH2 domain-11-mer peptide complexes reported
previously are shown in green.32 As is evident from Figure 5,
the orientation of the guanidinium group in the side chain of
Arg RA2 in the SH2 complex with 4 is flipped by approximately
180° relative to the complex with the 11-mer peptide; Arg âB5
remains unchanged in the two complexes.
This deviation in the conformation of the Arg RA2 side chain,
which is consistent through the various structures within the
asymmetric unit, may be significant because it results in a
change in the hydrogen-bonding network within the complex.
Namely, Nꢀ of Arg RA2 in the 11-mer peptide-SH2 domain
complex forms a hydrogen bond with the phosphate group of
pTyr, whereas in the SH2 domain complex with 4, the Nꢀ of
Arg RA2 is directed away from the phosphate moiety making
instead a hydrogen bond with a water molecule. Since the net
number of hydrogen bonds is maintained, the impact of the
change in Arg RA2 side-chain conformation upon binding
affinity cannot be easily predicted.62 Another possible conse-
quence of the different orientation of the Arg RA2 side chain
in the SH2 domain-4 complex is the disruption of a π-cation
interaction with the tyrosine ring, which had been observed in
the Src SH2 domain-11-mer peptide complex structure,32
because the distance between the guanidine nitrogen closest to
the centroid of the tyrosine ring has increased to 3.8 from 3.3
Å in the original complex.
Figure 4. Superposition of the structures of one of the Src SH2 domain-
compound 4 complex and one of the Src SH2 domain-11-mer flexible
peptide described previously in Waksman et al.32 (A) Superposition of the
two complexes. Representation of the backbone structure and of the ligand
is as in Figure 3A. The yellow and green ribbon models correspond to
models 4.1 and 3.3 of Table 3. Color-coding for atoms in the ligand and
the protein are as in Figure 3A except for carbons in model 3.3 which are
in green. All residues in the protein are labeled according to Waksman et
al.32 Residues in the ligand are labeled PTR for the conformationally
constrained phosphotyrosine, pTyr for the flexible phosphotyrosine of model
3.3, +1 for the +1 Glu, +2 for the +2 Glu, and +3 for the +3 Ile. (B)
Overlay of the three bound structures of the 11-mer peptide (truncated for
clarity to show only the pYEEI core) and the two conformations of 4 (purple
and magenta).
A superimposition of these three ligand structures, together with
those of the two Src SH2 domain-compound 4 complexes, is
depicted in Figure 4B, while a superimposition of one of each
of the two types of complexes is shown in Figure 4A. Inspection
of these overlays reveals that there are no significant differences
in the structures of the pTyr-+3 Ile segment in the backbones
of 4 and the 11-mer peptide, and the orientations of these peptide
backbones relative to the protein are essentially identical. The
side chains of pTyr and Ile in the two ligands are also projected
into their respective pockets of the SH2 domain in a nearly
identical fashion.
Several other conformational aspects of the bound structure
of 4 are noteworthy. For example, the amide carbonyl group
that links the pTyr to the +1 Glu bisects the cyclopropane ring,
whereas the carbonyl group of the N-terminal methyl amide is
deflected from this bisecting orientation by approximately 30°
(Figure 6). Structural and conformational studies of cyclopropyl
For the purposes of the present study, it is significant that
the spatial orientations of the carbonyl groups and the phosphate
on the aromatic ring of the pTyr residue in the 11-mer peptide-
SH2 domain complex are very similar to those of the corre-
sponding substituents on the cyclopropane ring of 4 bound to
the SH2 domain. The cyclopropane replacement in 4 thus serves
as a reliable structural mimic of the bound conformation of the
(62) Fersht, A. R. Trends Biochem. Sci. 1987, 12, 301-304.
9
J. AM. CHEM. SOC. VOL. 124, NO. 2, 2002 211

A R T I C L E S
Davidson et al.
Figure 6. Conformation of the carbonyl group of cyclopropane-rigidified
phosphotyrosine (B); representation and labeling of residues in the peptide
and in the protein are as in Figure 4A. The top panel represents the bisecting
orientation of the pTyr-+1 Glu amide carbonyl, while the lower panel
represents the significant deviation from the “preferred” bisecting orientation
of the methylamide-pTyr carbonyl that is observed in the crystal structure.
carbonyl compounds suggest that the preferred orientation of
the carbonyl group is the one in which the carbon-oxygen
double bond bisects the cyclopropane ring.63-66 Indeed, we have
observed this preference in the crystal structures of several
cyclopropane-derived peptide mimics,67 including the trisub-
stituted cyclopropane subunit in 4 (see Supporting Information).
We have previously observed similar deviations from this
“ideal” bisecting conformation in earlier structural studies of
free and bound cyclopropane-containing inhibitors of HIV
protease,16 so the energetic consequences of such small torsional
changes are difficult to predict. The methyl group of the
N-terminal amide is also rotated out of the plane of the carbonyl
group by approximately 22°. This deviation from planarity,
which results in the loss of some amide resonance, may result
from a steric interaction with a guanidinium nitrogen of Arg
RA2 only 3.3 Å away (Figure 7).
Surface Area Calculation and Derivation of Thermo-
dynamic Parameters from Structure. Despite the overall
similarities in the putative binding modes of compounds 4 and
6, the introduction of a cyclopropane ring in 4 could result in
differences in surface contact areas that might help explain the
observed differences in binding enthalpy. To address this
question, the structure of the pYEEI core of the 11-mer peptide
was used to generate two models.32 In the first one, the
N-terminal amino group of the pYEEI peptide was acetylated
to constitute compound 3. In the other model, a CH3-
NHCOCH2- moiety was substituted for the amino group on
the Cr atom of pTyr to give compound 6, and this structural
subunit of 6 was then manually oriented into a conformation
that corresponded to that of the trisubstituted cyclopropane in
the SH2 domain-4 complex. These models were then mini-
mized using conjugate gradient protocols. Polar and nonpolar
contact surface areas were calculated for 4, 6, and 3 using the
Figure 7. Schematic diagram of the interactions between the rigidified
phosphotyrosine and the pTyr binding pocket (A) and between the +1 and
+2 positions and the protein (B). Schematic drawings of the ligand and
protein residues are indicated in thick red and thin black lines, respectively.
Dashed lines connect atoms of the ligand and the protein that are within
contact distances. Numbers on the dashed lines indicate corresponding
distances. In (B), PTR indicates the conformationally constrained phos-
photyrosine, and “PTR O” indicates the carbonyl oxygen of PTR. Arg âD′1
is in light blue, because its guanidinium group is observed in contact with
the +2 Glu in only one molecule in the asymmetric unit crystal. In (A) and
(B), * labels the same water molecule.
method described above (see Experimental Section), the ther-
modynamic parameters were computed,52,54 and the results are
summarized in Table 3.
This computational method has been applied successfully in
a number of studies involving the prediction of binding
energetics of protein-peptide or protein-protein interactions.68-70
However, in the present instance, these calculations do not
predict accurately the observed energetic parameters for the
binding of phosphotyrosyl peptides to SH2 domains. For
example, the calculated free energy for the binding of compound
3, a tetrapeptide, varies between -7.3 and -4.3 kcal mol-1, a
value that compares poorly against the experimentally deter-
mined value of -9.0 kcal mol-1. There is also a striking lack
of convergence among the predicted thermodynamic values
derived from equivalent structures. For example, the free energy
of binding of compound 4 calculated from one structure in the
asymmetric unit yields a free energy of -11.3 kcal mol-1
,
whereas the same calculation using the other structure for the
(63) Allen, F. H. Acta Crystallogr., Sect. B 1979, 36, 81-96.
(64) Allen, F. H. Acta Crystallogr., Sect. B 1981, 37, 890-900.
(65) Allen, F. H.; Kennard, O.; Taylor, R. Acc. Chem. Res. 1983, 16, 146-
153.
(66) de Meijere, A. Angew. Chem., Int. Ed. Engl. 1979, 18, 809-886.
(67) Lynch, V. M.; Austin, R. E.; Martin, S. F.; George, T. Acta Crystallogr.,
Sect. C 1991, 47, 1345-1347.
complex yields the very different value of -6.2 kcal mol-1
.
(68) Baker, B. M.; Murphy, K. P. J. Mol. Biol. 1997, 268, 557-569.
(69) Gomez, J.; Freire, E. J. Mol. Biol. 1995, 252, 337-350.
(70) Murphy, K. P.; Xie, D.; Garcia, K. C.; Amzel, L. M.; Freire, E. Proteins
1993, 15, 113-120.
9
212 J. AM. CHEM. SOC. VOL. 124, NO. 2, 2002

Studies of 1,2,3-Trisubstituted Cyclopropanes
A R T I C L E S
Table 3. Calculated Binding Energies of the Src SH2 Domain Interaction with Compounds 4, 3, and 6a
∆ASA total
∆ASA polar
∆ASA nonpolar
∆C°p,calc
∆H°calc
∆S°solv
∆S°conf
∆S°calc
∆G°calc
model
(Å)
(Å)
(Å)
cal/mol/K
kcal/mol
cal/mol/K
cal/mol/K
cal/mol/K
kcal/mol
4.1
4.2
3.1
3.2
3.3
6.1
6.2
6.3
-989
-948
-521
-477
-470
-504
-494
-500
-516
-488
-468
-471
-493
-462
-507
-469
-494
-543
-105.1
-84.8
-76.0
-99.1
-83.1
-95.4
-95.8
-70.8
-15.5
-13.4
-12.8
-14.8
-13.6
-14.5
-14.2
-12.7
27.0
21.7
19.4
25.3
21.2
24.4
24.5
18.0
-32.9
-37.9
-39.9
-42.6
-41.1
-29.0
-44.1
-36.8
-13.8
-24.0
-28.3
-25.1
-27.7
-28.3
-27.4
-27.7
-11.3
-6.2
-4.3
-7.3
-5.3
-10.8
-6.6
-4.8
-963
-967
-1001
-970
-1010
-1032
a Surface area calculations and derivation of binding energetics were carried out as described in the Experimental Section. Model 4.1 and 4.2 refer to the
two experimental Src SH2 domain structures bound to compounds 4 present in the asymmetric unit of the crystal. Models 3.1, 3.2, and 3.3 refer to the
structures of the Src SH2 domain bound to compound 3 (Figure 1B). Here, the starting structures for modeling compound 3 were those of the three Src SH2
domain-11-mer peptide complexes determined previously.32 Models 6.1, 6.2, and 6.3 refer to the structures of the Src SH2 domain bound to compound 6
(Figure 1B). The starting structures of these models were as for models 3.1, 3.2, and 3.3.
The range of calculated free energy values is narrower for
compound 3, but it is still unusually large. Similarly, the
calculated enthalpy and entropy changes do not correlate well
with the experimental values (Tables 2 and 3). It is perhaps
significant that even though the calculated values for ∆S for
the conformationally constrained pseudopeptide 4 are very
different (-13.9 and -24.0 cal mol-1 K-1), they are less
negative than for the more flexible compounds 6 and 3.
result supports the hypothesis that the preorganization of a ligand
in its active conformation gives rise to an entropic advantage
in binding to its biological target. Entropies of internal rotations
in the gas phase are known experimentally, and they appear to
be similar in solution.58,71 For example, the loss of two rotational
rotors has been estimated to be 7-8 cal mol-1 K-1 at 298 K.58,71
This value corresponds to that predicted for the restriction of
the two rotors about the ø1 and Φ angles by introducing the
1,2,3-trisubstituted cyclopropane ring (Figure 1B). Hence, in
the present instances, the experimental results agree with
predictions.
Discussion
Introducing conformational constraints into a flexible ligand
to enforce or stabilize its bound structure has been widely used
as a strategy for the design of new ligands that will have higher
affinity for a selected macromolecular target. The rationale for
this method is straightforward: reducing the conformational
entropic penalty that must be paid upon binding would be
predicted to result in a net gain in the free energy of binding
provided all favorable binding interactions remain and no
unfavorable steric contacts are introduced. While there are a
number of examples that support this hypothesis,1-7 there have
been no reports of detailed thermodynamic studies using a
structurally well-characterized biological system.
We have used cyclopropanes as dipeptide replacements in
designing a number of conformationally constrained enzyme
inhibitors having high potencies.15-17,19,20,34 In the present study,
1,2,3-trisubstituted cyclopropanes were prepared to replace the
CR, Câ, and amide nitrogen of a phosphotyrosine residue in
the pYEEI tetrapeptide 3, which is a ligand for the SH2 domain
of the Src kinase. These substitutions led to the pseudopeptides
4 and 5 in which the phosphotyrosine side chain and backbone
were constrained in an orientation that modeling had suggested
would closely mimic that of a pTyr residue bound to the Src
SH2 domain. The effects of thus rigidifying the pTyr subunit
of pYEEI upon binding to the Src SH2 domain were examined
using calorimetry and X-ray crystallography. The thermody-
namic parameters, ∆H°, ∆S°, ∆G°, were obtained for complex
formation between the Src SH2 domain and compounds 3-7,
and the ∆Cp of binding for the conformationally constrained
pseudopeptide 4 and its flexible analogue 6 was determined.
These binding parameters were then interpreted in light of crystal
structures of the complexes of the Src SH2 domain with 4 and
with a natural 11-mer peptide.
This study reveals a number of important and interesting
features regarding the use of substituted cyclopropanes in the
design of conformationally constrained phosphopeptide inhibi-
tors having high affinity for the Src SH2 domain. Introducing
a cyclopropane as a conformational constraint into compounds
4 and 5 did have the expected effect upon the entropy of binding,
but the constrained (4, 5) and flexible analogues (6, 7) had
comparable binding affinities for the Src SH2 domain because
the favorable gain in entropy was compensated by an unfavor-
able enthalpic change. We have shown through cocrystallization
with the Src SH2 domain that the binding modes of compound
4 and an 11-mer peptide containing the pYEEI sequence do
not differ. Superimposition of these two complexes does not
reveal dramatic changes in the way the pseudopeptide and the
peptide are bound. The bound structures of the ligands are
similar, and both ligands make comparable interactions with
the protein. Including the differing conformations observed in
Arg RA2 and the slight deviations observed around the
phosphotyrosine, the differences in contacts between the two
ligands and the SH2 domain are qualitatively no greater than
the differences observed between similar complex structures in
the asymmetric units of the crystals. A computational study of
the binding interface between the pYEEI core of the 11-mer
peptide (i.e., 3), 4, and 6 and the Src SH2 domain did not reveal
any significant differences in either the mode of ligand binding
or the buried accessible surface area. Hence, although the slight
deviations observed may influence the enthalpy term unfavor-
ably, it is not possible to assign a value to the individual
contributions, and the underlying causes for the observed
difference in enthalpy contribution between ligands remain
unclear.
Constraining the pTyr of 3 with a substituted cyclopropane
replacement had the expected beneficial effect upon the observed
∆S° as the binding of 4 and 5 to the Src SH2 domain was
entropically favored over 6 and 7 by 5-9 cal mol-1 K-1. This
A large body of data has been accumulated on recognition
and binding of tyrosyl phosphopeptides by the Src SH2
(71) Williams, D. H.; Gerhard, U.; Searle, M. S. Bioorg. Med. Chem. Lett. 1993,
3, 803-808.
9
J. AM. CHEM. SOC. VOL. 124, NO. 2, 2002 213

A R T I C L E S
Davidson et al.
domain.36-39,41,43 All of these data underscore the difficulty of
interpreting thermodynamic parameters in light of structure. In
a recent study, about 30 Src SH2 domain mutants containing
alanine substitutions were examined in an attempt to correlate
the structural changes that accompany these mutations with
variations in the thermodynamic parameters ∆G°, ∆H°, and
T∆S°. The van der Waals contacts and buried surface areas upon
removal of a side chain were examined, and no correlation could
be established.37 Similar conclusions were drawn from a much
more exhaustive study by Henriques et al.43 who attempted to
predict thermodynamic parameters from a number of known
structures of complexes of the Src SH2 domain.43 Five different
area-based models were tested in which ligand conformational
flexibility and proximal ordered solvent molecules were treated
in various ways in the surface area calculations. None of these
models predicted the thermodynamic parameters for binding of
a pYEEI-containing peptide to the Src SH2 domain with any
degree of accuracy. In the present study, a similar computational
method not only failed to predict any of the thermodynamic
binding parameters, but there was also a striking lack of
convergence among the predicted thermodynamic values derived
from several similar complexes of an asymmetric unit of the
crystal. Such discrepancies may be caused by slight but
cumulative differences between complexes in the asymmetric
unit. These differences probably arise from packing forces that
can disturb side-chain orientations and local structures at and
around the crystal packing interfaces. Hence, the present work
further illustrates the deficiencies in existing computational
methods for predicting binding of phosphotyrosine-containing
peptides by SH2 domains.
peptide binding may be to neutralize the negative charge of this
cluster of carboxylic groups.
The binding of the +2 Glu was originally thought to be
influenced by a water network bridging the +2 Glu carboxyl
group to Arg âD′1,32 and other studies have also claimed a role
for this water network.72 However, when Arg âD′1 of the SH2
domain was mutated to either Phe or Ala, there was only a very
small loss in binding free energy of 0.3 kcal mol-1 for Phe âD′1
and 0.6 kcal mol-1 for Ala âD′1.37 This observation suggested
that the water-network-mediating interactions between +2 Glu
and Arg âD′1 reported by Waksman et al.32 may not play a
critical role in determining selectivity at the +2 position. This
conclusion is supported by the new crystal structure presented
herein where there are no such bridging water networks between
the âD′1 position in the protein and the +2 position in the
ligand.
Because the structures of the complex of 4 with the Src SH2
domain are of high resolution, the solvent substructure around
the complex and within the interface was defined with precision.
Two water molecules are found at the ligand-protein binding
interface, and by the nature of the interactions these make, both
the protein and the ligand indicate that they may play an
important role in binding (Figure 7). Both of these waters are
involved in extensive hydrogen-bonding interactions with the
ligand and the Src SH2 domain. The first water molecule, which
is labeled with an asterisk (*) in Figure 7, lies near the
conformationally constrained pTyr group (indicated as PTR)
and +1 Glu. This water makes H-bond interactions with the
N-terminal carbonyl oxygen of pTyr (or PTR), the carbonyl
group of the +1 Glu, and the amide nitrogen of Lys âD6. The
second water molecule in Figure 7B is positioned between the
+2 Glu and +3 Ile. It also is hydrogen bonded with the carbonyl
oxygen of +2 Glu, with the carbonyl oxygen of Gly BG2 in
the BG loop, and with the terminal hydroxyl of the side chain
of Tyr âD5. These hydrogen bonds, notably the one with the
hydroxyl group of Tyr âD5, were not observed previously.31,32
Tyr âD5 has been identified previously as one of the two hotspot
residues in the binding interface of the Src SH2 domain. Tyr
âD5 appears to play several roles at this interface. It forms a
portion of the hydrophobic +3 Ile-binding cavity, the so-called
+3 binding pocket where the +3 Ile of the pYEEI peptide
inserts.37 It also forms a van der Waals contact with the Câ of
the +1 position residue as well as a platform for the backbone
of the phosphoryl peptide ligand.32 The structures of the Src
SH2 domain-compound 4 complex suggest that Tyr âD5 may
play an additional role in anchoring the BG loop to the ligand
backbone through the involvement of a water molecule. In any
case, it is interesting to note that the water structure around the
binding site implicates mostly main chain atoms in the ligand
and therefore is not expected to play a role in determining the
specificity of the phosphopeptide-SH2 domain interaction.
The crystal structure of the Src SH2 domain bound to the
constrained pseudopeptide 4 further illuminates the roles of some
of the side chains in binding. Although the structures described
here are very similar to those already reported,32 there are some
small, yet likely significant, differences. Previous studies have
shown that, among the four ligand residues involved in binding,
the pTyr and the +3 Ile contribute most to the free energy of
binding.39,40 This finding is consistent with the two-pronged
model of SH2 domain binding and is reflected in the fact that
in all structures of the Src SH2 domain bound to a pYEEI
peptide, little variation in structure around the pTyr and the +3
Ile is observed.32 Most of the structural variations observed in
the ligands are localized around the +1 and +2 positions and
in the water structure surrounding the +2 position, suggesting
that these side chains contribute only modestly to binding.39
Deviations in the +1 Glu side-chain orientations observed
in the cocrystal structure described here confirm a previous study
that showed that mutating the +1 Glu position in the ligand to
Ala has little impact on the binding of a pYEEI-containing
peptide.39 It was originally hypothesized that the carboxyl group
of +1 Glu interacted favorably with the side chain of Lys âD3,
located nearby on the surface of the SH2 domain.32 However,
the crystal structure described here suggests another role: Lys
âD3 appears to balance other surface charges presented by a
cluster of aspartate carboxyl groups, Asp CD2 and Asp âC8
(Figure 3C). It has been shown that mutating Lys âD3 to Ala
results in a significant drop in the free energy of binding,
presumably due to the repulsion that the cluster of aspartate
carboxyl groups on the SH2 domain exerts on the +1 Glu
position of the peptide.37 Hence, the role of Lys âD3 in pYEEI
This study reveals a number of important and interesting
features regarding the design and evaluation of conformationally
constrained peptide mimics. Clearly there is an entropic
advantage that accompanies restricting the number of rotatable
bonds in closely related peptide-like ligands as evidenced by
the favorable change in ∆S° for 4 relative to 6. This observation
(72) Chung, E.; Henriques, D.; Renzoni, D.; Zvelebil, M.; Bradshaw, J. M.;
Waksman, G.; Robinson, C. V.; Ladbury, J. E. Struct. Fold. Des. 1998, 6,
1141-1151.
9
214 J. AM. CHEM. SOC. VOL. 124, NO. 2, 2002

Studies of 1,2,3-Trisubstituted Cyclopropanes
A R T I C L E S
validates one of the original premises that led to the invention
and use of 1,2,3-trisubstituted cyclopropanes as rigid replace-
ments of extended peptide secondary structure. However, this
entropic advantage was unexpectedly compensated by a loss in
binding enthalpy. Entropy-enthalpy compensation, which
involves corresponding changes in ∆H° and ∆S° so that the
changes in ∆G° are minimized, is beginning to appear as a
general phenomenon in interactions of proteins with peptides
and peptide-like ligands.73,74 The similarity of the crystal
structures of the Src SH2 domain with the conformationally
constrained pseudopeptide 4 and an 11-mer peptide coupled with
the magnitudes of the deviations in ∆H° and ∆S° between the
flexible and cyclopropane-containing ligands in this study
convincingly demonstrate that complete thermodynamic profiles
and structural data must be obtained to evaluate the conse-
quences of introducing a conformational constraint into a
molecule. Developing a paradigm for designing high-affinity
ligands so that favorable enthalpic interactions are maintained
while minimizing the unfavorable entropy of binding will be a
major challenge for the future.
Acknowledgment. This work was funded by NIH grants
GM60231 (to G.W.) and GM43473 (to S.F.M.) and by The
Robert A. Welch Foundation (to S.F.M.) and the Texas
Advanced Research Program (to S.F.M.). J.P.D. and O.L.
contributed equally to this work.
Supporting Information Available: Complete experimental
details and characterization (1H and 13C NMR spectra, IR
spectra, and mass spectra) of 3-7 and all new compounds
leading to their synthesis together with X-ray crystal structural
data for the cyclopropyl pTyr replacement in 4 (PDF). This
material is available free of charge via the Internet at
http://pubs.acs.org.
(73) Sharp, K. Protein Sci. 2001, 10, 661-667.
(74) Liu, L.; Guo, Q.-X. Chem. ReV. 2001, 101, 673-695.
JA011746F
9
J. AM. CHEM. SOC. VOL. 124, NO. 2, 2002 215