Ligand preorganization may be accompanied by entropic penalties in protein-ligand interactions

Article abstract of DOI:10.1002/anie.200600844

(Figure Presented) Fact or fiction? It is generally assumed that preorganization of a flexible ligand in the 3D shape it adopts when bound to a macromolecular receptor will provide a derivative with an increased binding affinity that benefits entropically

Full text of DOI:10.1002/anie.200600844

Communications
Molecular Recognition
DOI: 10.1002/anie.200600844
Ligand Preorganization May Be Accompanied by
Entropic Penalties in Protein–Ligand
Interactions**
Aaron P. Benfield, Martin G. Teresk, Hilary R. Plake,
John E. DeLorbe, Laura E. Millspaugh, and
Stephen F. Martin*
A prevailing hypothesis in the field of molecular recognition
in chemistry and biology is that the preorganization of flexible
hosts and their guests in a manner corresponding to the three-
[*] A. P. Benfield, M. G. Teresk, Dr. H. R. Plake, J. E. DeLorbe,
L. E. Millspaugh, Prof. S. F. Martin
Department of Chemistry and Biochemistry
and The Institute of Cellular and Molecular Biology
University of Texas at Austin
1 University Station–A5300, Austin, TX 78712-1167 (USA)
Fax: (+1)512-471-4180
E-mail: sfmartin@mail.utexas.edu
Homepage: http://research.cm.utexas.edu/smartin/
[**] Acknowledgment is made to the National Institutes of Health, The
Robert A. Welch Foundation (F-652), the Texas Advanced Research
Program, Pfizer, Inc., and MerckResearch Laboratories for their
generous support of this research. We are also grateful to Professor
Gabriel Waksman (University College London, Birkbeck College),
Andrew Prongay (Schering-Plough Research Institute), and Dr.
John H. Clements (The University of Texas) for helpful discussions.
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
6830
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2006, 45, 6830 –6835

Angewandte
Chemie
dimensional structures they adopt upon complex formation
will lead to higher association constants. In simple host–guest
complexes, changes in free energy DG that result from
introducing such constraints can be highly favorable, an
observation that led Cram to characterize preorganization as
“a central determinant of binding power”.[1,2] Ligand preor-
ganization has also emerged as a strategy for structure-based
drug design, and restricting the flexibility of peptides and
other small molecules has indeed led to the discovery of
selective and bioavailable drug leads that may exhibit
increased potencies.[3–5]
Toward explicitly elucidating the energetic and structural
consequences of ligand preorganization in protein–ligand
interactions, we have prepared pseudopeptides in which
cyclopropane rings serve as rigid replacements for the Ca
and Cb carbon atoms and the NH groups of an amino acid
residue, thereby constraining the backbone and orienting the
side chain.[17] In one study, constrained and flexible deriva-
tives of the tetrapeptide Ac-pTyr-Glu-Glu-Ile-OH, the con-
sensus sequence in Src SH2 domain binding ligands,[18,19] were
prepared that contained substituted cyclopropane and suc-
cinyl replacements for the phosphotyrosine residue.[20,21] The
thermodynamic parameters for complex formation of these
ligands with the Src SH2 domain were determined by
isothermal titration calorimetry (ITC). Although the con-
strained ligands bound with more favorable entropies of
binding than their flexible counterparts, this expected
entropic advantage was always offset by a corresponding
enthalpic penalty that resulted in comparable binding affin-
ities of the various ligand pairs. There was thus no net
energetic advantage associated with ligand preorganization
Although the energetic consequences of ligand preorga-
nization on protein binding affinities are necessarily both
enthalpic and entropic, preorganization seems universally
regarded as having
a favorable entropic component.
Enhanced affinities arising from preorganization are thus
typically, albeit simplistically, regarded as being primarily
entropic in origin, provided that the constrained and flexible
molecules make the same pairwise interactions with the
protein and solvent. Fully restricting each independent rotor
in a flexible ligand should accordingly be accompanied by an
entropic advantage of about 0.7–1.6 kcalmolÀ1.[6,7] However,
increases in potency accompanying ligand preorganization
are commonly much less than those predicted based upon
reducing the entropic binding penalties that are associated
with conformational flexibility.
Identification of the specific origin of differences in the
protein binding affinities of preorganized ligands relative to
their flexible analogues is problematic for several reasons.
Firstly, there are few cases where association constants are
determined for a pair of constrained and flexible ligands
having the same number and type of heavy atoms, the same
functional groups, and the same number of hydrogen-bond
donors and acceptors. Appropriate controls are thus generally
absent, although there are reports where the ligands either
meet the above criteria[8–10] or differ only slightly in elemental
composition and functionality.[7,11–15] Secondly, structural
information for complexes of the con-
because of balancing enthalpy–entropy compensation,[22,23]
a
ubiquitous phenomenon in host–guest and protein–ligand
systems. Comparison of the X-ray crystal structures of the Src
SH2 domain complexed with the cyclopropane-derived ligand
and an 11-mer peptide containing the pTyr-Glu-Glu-Ile
sequence revealed that both bound similarly, and all intera-
tomic distances between the domain and the ligand in each of
the two structures were in close agreement. Hence, the origin
of the observed enthalpic disadvantage attending ligand
preorganization was enigmatic, and investigations of other
protein–ligand complexes would be required to gain further
insights.
Phosphotyrosine peptides with the consensus sequence of
Ac-pTyr-Xaa-Asn-NH2, wherein Xaa is typically a hydro-
phobic residue, bind to the SH2 domain of the growth
receptor binding protein 2 (Grb2).[18,19] The pseudopeptides 2
and 3 (see Table 1) were thus prepared as constrained and
Table 1: Thermodynamic parameters for complex formation between the Grb2 SH2 domain and
pseudopeptides 2–7.[a]
strained and flexible ligands with the bio-
logical target is typically lacking, so whether
both ligands interact similarly with the
biomacromolecule is unknown. Ligands
also frequently bind to proteins in confor-
mations that are higher in energy than their
global minima in solution,[16] and the anal-
ysis of how differential conformational
strain energies affect affinities of similar
compounds is a complex matter. Finally, the
specific contributions to DS and DH of
binding are rarely determined, so the extent
to which entropic and enthalpic factors play
a role in the relative potencies of con-
strained and flexible ligand pairs cannot be
assessed.[15] There is thus little compelling
scientific evidence supporting the widely
asserted hypothesis that ligand preorgani-
zation will lead to more favorable binding
entropies in protein–ligand interactions.
Cmpd.
Ka [mÀ1
]
DG [kcalmolÀ1
]
DH [kcalmolÀ1
]
DS [calmolÀ1 K]
2
3
4
5
6
7
(1.0Æ0.1)”106
(4.4Æ0.4)”105
(1.6Æ0.1)”106
(3.7Æ0.1)”105
(2.3Æ0.1)”105
(1.7Æ0.1)”105
À8.2Æ0.1
À7.7Æ0.1
À8.5Æ0.1
À7.6Æ0.1
À7.3Æ0.1
À7.1Æ0.1
À7.0Æ0.1
À5.3Æ0.1
À6.7Æ0.1
À4.9Æ0.1
À5.6Æ0.1
À4.6Æ0.1
4.2Æ0.1
8.2Æ0.2
6.0Æ0.1
9.1Æ0.2
5.9Æ0.1
8.6Æ0.2
[a] ITC experiments were conducted at 258C in duplicate with the same batch of ligand and Grb2 SH2
domain in 2-[4-(2-hydroxyethy1)piperazin-1-yl]ethanesulfonic acid (HEPES, 50 mm) with NaCl (150 mm)
at pH 7.5. Uncertainties in Ka, DG, DH, and DS values represent deviations from the average. Similar
experiments conducted with 4 and 5 in Tris buffer gave comparable results.
Angew. Chem. Int. Ed. 2006, 45, 6830 –6835
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
6831

Communications
flexible analogues of Ac-pTyr-Val-Asn-NH2 (1).[24] As in our
previous studies of Src SH2 binding ligands, the cyclopropane
ring in 2 serves as a rigid mimic of the pTyr residue in 1,
whereas a benzyl succinyl moiety is a flexible pTyr replace-
ment in the control 3. The thermodynamic parameters (DG,
DH, DS) for binding of 2 and 3 to the Grb2 SH2 domain were
determined by ITC (Table 1), and the constrained ligand 2
was observed to bind approximately twofold better than its
flexible counterpart 3. However, in an unprecedented finding
that is inconsistent with the conventional wisdom regarding
the putative effects of preorganization, the entropy of binding
for 2 was significantly less favorable than for 3.
In an extension of these studies, the amino acid at the
pY+ 1 position of 2 and 3 was changed from Val to Ile and
Leu, thus giving the corresponding ligand pairs 4–7 to assess
the influence that variations of this hydrophobic residue
might have on the energetics of binding. The thermodynamic
parameters for binding of 4–7 to the Grb2 SH2 domain were
determined (Table 1), and the constrained pseudopeptides 4
and 6 were observed to bind with slightly higher affinities than
their corresponding flexible controls 5 and 7. However, the
increased affinities for the preorganized ligands 4 and 6 were
again the consequence of more favorable binding enthalpies,
and the binding entropies for 4 and 6 were unfavorable
relative to those of their flexible counterparts 5 and 7. The
DCp values for 4 and 5 were determined to be À88 and
À94 calmolÀ1 K, respectively, so differences in binding free
energies do not appear to arise from solvation/desolvation or
hydrophobic effects.[25] Based upon these unprecedented
findings, it is now apparent that ligand preorganization may
be accompanied by unfavorable binding entropies, yet more
favorable binding free energies.
To explore whether structural factors might be respon-
sible for the differential binding enthalpies of the constrained
and flexible ligands, complexes of the monomeric Grb2 SH2
domain with 2 and 3 were co-crystallized for X-ray analysis.
Data for the complexes of 2 and 3 with the Grb2 SH2 domain
were collected to 1.9 and 1.7 Š resolution, respectively, and
the structures were solved by molecular replacement using a
known structure.[26,27] Similar to other Grb2 SH2 binding
ligands,[26,28,29] 2 and 3 bind to the domain in a b-turn-like
conformation. There are two Grb2 SH2–2 complexes in the
asymmetric unit, and the backbone atoms of the domain in
these align with a root-mean-square deviation (rmsd) of
0.3 Š; atoms of 2 in the two complexes align with an rmsd of
0.2 Š. An overlay of the two Grb2 SH2–2 complexes with that
of the Grb2 SH2-3 complex is shown in Figure 1a. The
backbone atoms of the Grb2 SH2 domain in the complexes of
2 align with those in the complex of 3 with rms deviations of
0.5 and 0.6 Š. The variations occur primarily in the flexible
loops, but the most notable difference, which is revealed by
the close-up superimposition in Figure 1b, is the position of
the BC loop that forms part of the phosphate binding pocket.
In particular, the Ca and nitrogen backbone atoms of Ser90
and Glu89 in the BC loop in each of the complexes of 2 are
displaced toward the ligand so they are about 2.0 Š closer to
the phosphate group of 2 than they are in the corresponding
complex with 3. The atoms of the Val and Asn residues in the
complexes of 2 and 3 superimpose with an rmsd of 0.1 Š, and
Figure 1. Superimposition of the structures of the complexes of 2 and
3 with the monomeric Grb2 SH2 domain. The protein is shown as a
ribbon representation, whereas the ligands are shown as stickrepre-
sentations. The complex between 2 and the Grb2 SH2 domain crystal-
lized with two molecules in the asymmetric unit, and the structure was
refined to a Rcryst of 19.4% and a Rfree of 22.5%. The complex between
the Grb2 SH2 domain and 3 crystallized with one molecule in the
asymmetric unit, and the structure was refined to a Rcryst of 20.7% and
a Rfree of 23.4%. a) Alignment of the two complexes in the asymmetric
unit of 2 (magenta and violet) with the complex of 3 (orange).
b) Close-up overlay of the structures of the two molecules of 2 from
the two complexes in the asymmetric unit (magenta and violet) with
the structure of 3 (orange) showing their relationship with residues
Ser88-Asp94 in the BC loop. The constrained and flexible phosphotyr-
osine replacements are labeled cpTyr and fpTyr, respectively.
the direct interactions between the domain and the Val-Asn
dipeptide subunits of 2 and 3 are virtually identical. However,
there are significant variations in the positions of the atoms in
the constrained and flexible phosphotyrosine replacements of
2 and 3, as these align with an rmsd of 0.6 Š. Dissimilarities in
the positioning of the N-terminal methyl amide moieties of 2
and 3 appear inconsequential, whereas variations in the
orientations of the phosphate groups lead to significant
differences in the interactions of the phosphate oxygen atoms
of 2 and 3 with the SH2 domain.
6832
www.angewandte.org
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2006, 45, 6830 –6835

Angewandte
Chemie
The detailed interactions between the Grb2 SH2 domain,
ordered water molecules, and the phosphotyrosine replace-
ments in 2 and 3 are presented in Figure 2a and b. An analysis
of these interactions reveals that the aforementioned dis-
the phosphate group, but there are no comparable direct
interactions in the complex with 3. The side chain of Glu89 in
the complex of 2 is oriented toward the ligand, so the carboxy
group interacts via a water molecule with a nonbridging
oxygen atom of the phosphate group. There is no interaction
between the carboxy group of Glu89 and the phosphate group
in the complex of 3.
The spatial dispositions of the atoms in the constrained
and flexible phosphotyrosine replacements in 2 and 3 are
different, so we queried whether these replacements might be
uniquely responsible for the observed differences in the
pairwise interactions with the Grb2 SH2 domain and the
conformational perturbation of the BC loop. These com-
plexes were thus compared with the structures of 8, in which a
22-membered macrocycle constrains the pTyr-Val-Asn
sequence in its biologically active conformation,[9] and the
linear nonapeptide Ala-Pro-Ser-pTyr-Val-Asn-Val-Gln-Asn
(9) bound to the Grb2 SH2 domain (Figure 3a,b).[26] The
backbone atoms of the domain in the two complexes of 2 and
the complex of 8 align with an rmsd of 0.3–0.4 Š; the
corresponding atoms in 2 and 8 align with an rmsd fit of about
0.6 Š (Figure 3b). The polar interactions and their respective
contact distances between the domain and the phosphate
groups of 2 and 8 in these complexes are comparable. Hence,
two different tactics for introducing a conformational con-
straint into a pYVN-derived ligand result in similar displace-
ment of the BC loop toward the ligand. When the complexes
of the Grb2 SH2 domain with 3 and the nonapeptide 9 are
superimposed, the backbone atoms of the domain in the two
complexes align with an rmsd of 0.5 Š; the corresponding
atoms in 3 and 9 align with an rmsd of 0.3 Š (Figure 3c).
There are slight variations in the pairwise polar interactions
and contact distances between the ligands and the SH2
domain in these two complexes, but the differences are similar
to those observed when comparing the structures of other
phosphotyrosine ligands bound to the Grb2 SH2
domain.[9,26,28,29] These comparisons suggest that the cyclo-
propane ring in 2 is a good mimic of the phosphotyrosine
residue in the macrocycle 8, and that the succinyl moiety in 3
serves as a reasonable replacement of the phosphotyrosine in
the linear peptide 9. Hence, the specific phosphotyrosine
replacements in 2 and 3 do not themselves appear to induce
any unusual structural changes in the Grb2 SH2 domain upon
binding.
The greater number and closer contact distances associ-
ated with the pairwise interactions in the Grb2 SH2–2
complex relative to those in the Grb2 SH2–3 complex are
consistent with the more favorable enthalpy of binding
observed for 2. However, identification of the specific origins
of this enthalpic advantage is problematic because of the
complexity of these interactions, some of which involve
bridging water molecules. Further complicating the analysis is
the flexible nature of proteins and the energetic consequences
thereof. Namely, the formation of a protein–ligand complex
can result in an overall decrease or increase in protein
flexibility that is accompanied by corresponding changes in
nonbonded interactions and order throughout the com-
plex.[23,30] The effect these changes have upon the relative
magnitudes of the compensating enthalpies and entropies of
Figure 2. Hydrogen-bonding and polar interactions in the phosphotyr-
osine binding site of complexes of 2 and 3 with the Grb2 SH2 domain
showing those water molecules, W, involved in binding networks. Polar
contact distances between heavy atoms of ꢀ3.1 Š are indicated, but
some hydrogen atoms involved in these interactions are omitted for
clarity. a) One complex in an asymmetric unit of 2; the corresponding
pairwise distances in the other complex of 2 in the asymmetric unit
are Æ0.1 Š. b) Complex of 3.
placement of the BC loop and the different orientations of the
phosphate group in the phosphotyrosine replacements in 2
and 3 result in a number of closer contacts between the ligand
and the Grb2 SH2 domain in the complex with 2. For
example, the backbone amide N-H of Glu89 is directly
hydrogen-bonded to a nonbridging oxygen atom of the
phosphate group in the Grb2 SH2–2 complex, whereas a
water molecule mediates this interaction in the complex with
3. The amide N-H of Ser90 in the Grb2 SH2–2 complex forms
a direct hydrogen bond with a nonbridging oxygen atom of
Angew. Chem. Int. Ed. 2006, 45, 6830 –6835
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
6833

Communications
domain with 2 and 3 suggests there is more thermal motion in
the BC loop of the Grb2 SH2–2 complex. This observation is
opposite to that which might be expected from the tighter
interactions that are evident between this loop and the
phosphate moiety of 2 relative to those in the Grb2 SH2–3
complex. It is also inconsistent with the relative binding
entropies of 2 and 3.
Although preorganized ligands may bind to proteins with
higher affinities than those of their flexible counterparts, we
are not aware of any convincing evidence that the process
must be entropically favored, as is widely assumed. Indeed,
we have shown that the entropies of binding of preorganized
ligands may be disfavored relative to their less potent, flexible
controls. Increased affinities arising from ligand preorganiza-
tion may thus result solely from more favorable enthalpies of
binding. This study suggests that the enhanced enthalpy of
binding can arise from an unexpected increase in the number
and proximity of polar contacts between the protein and
ligand; however, changes in individual nonbonded interac-
tions throughout the complex will also contribute. These
results clearly indicate that the prevailing view of the
energetic consequences associated with ligand preorganiza-
tion, which focuses largely if not exclusively on the ligand
itself, in protein–ligand interactions must be modified. Future
models must include explicit consideration of the enthalpic
and entropic contributions arising from changes in non-
bonded interactions and order that occur in the protein upon
complex formation, and how these changes vary as a function
of ligand structure and flexibility. Interactions with water
molecules must also be taken into account.
Received: March 4, 2006
Revised: August 3, 2006
Published online: September 26, 2006
Keywords: ligand effects · molecular recognition · peptides ·
.
proteins · thermodynamics
[1] D. J. Cram, Angew. Chem. 1986, 98, 1041 – 1060; Angew. Chem.
Int. Ed. Engl. 1986, 25, 1039 – 1057.
[2] K. N. Houk, A. G. Leach, S. P. Kim, X. Zhang, Angew. Chem.
2003, 115, 5020 – 5046; Angew. Chem. Int. Ed. 2003, 42, 4872 –
4897.
Figure 3. Complexes of different ligands with the Grb2 SH2 domain,
[3] H. Nakanishi, M. Kahn in The Practice of Medicinal Chemistry,
2nd ed. (Ed.: C. G. Wermuth), Academic Press, London, 2003,
pp. 477 – 500.
[4] D. F. Veber, S. R. Johnson, H. Y. Cheng, B. R. Smith, K. W.
Ward, K. D. Kopple, J. Med. Chem. 2002, 45, 2615 – 2623.
[5] A. Mann in The Practice of Medicinal Chemistry, 2nd ed. (Ed.:
C. G. Wermuth), Academic Press, London, 2003, pp. 233 – 250.
[6] U. Gerhard, M. S. Searle, D. H. Williams, Bioorg. Med. Chem.
Lett. 1993, 3, 803 – 808.
[7] A. R. Khan, J. C. Parrish, M. E. Fraser, W. W. Smith, P. A.
Bartlett, M. N. G. James, Biochemistry 1998, 37, 16839 – 16845.
[8] B. P. Morgan, D. R. Holland, B. W. Matthews, P. A. Bartlett, J.
Am. Chem. Soc. 1994, 116, 3251 – 3260.
with the protein shown in ribbon representation and the ligands
shown in stickrepresentation. a) Chemical structure of 8. b) Alignment
of the two complexes in the asymmetric unit of 2 (magenta and violet)
with the published structure (2.1 Š resolution) of the complex of
macrocycle 8 (cyan) in reference [9] (1BM2), which shows their
relationship with residues Ser88-Asp94 in the BC loop of the respective
Grb2 SH2 domains. c) Overlay of structures of 3 (orange) with the
published structure (1.55 Š resolution) of the linear nonapeptide 9
(green) in reference [26] (1JYR), which has been truncated for clarity
and shows their relationship with residues Ser88-Asp94 in the BC
loop.
[9] P. Ettmayer et al., J. Med. Chem. 1999, 42, 971 – 980; see
Supporting Information.
[10] Y. S. Tsantrizos et al., Angew. Chem. 2003, 115, 1394 – 1398;
Angew. Chem. Int. Ed. 2003, 42, 1356 – 1360; see Supporting
Information.
binding cannot be evaluated, because crystallographic data do
not provide a reliable measure of protein flexibility in
solution. Nevertheless, it is interesting in this context that a
comparison of B-factors in the complexes of the Grb2 SH2
6834 www.angewandte.org
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2006, 45, 6830 –6835

Angewandte
Chemie
[11] R. W. Marquis et al., J. Med. Chem. 2001, 44, 1380 – 1395; see
Supporting Information.
[12] S. K. Sia, P. A. Carr, A. G. Cochran, V. N. Malashkevich, P. S.
Kim, Proc. Natl. Acad. Sci. USA 2002, 99, 14664 – 14669.
[13] F. J. Dekker, N. J. de Mol, M. J. E. Fischer, J. Kemmink, R. M. J.
Liskamp, Org. Biomol. Chem. 2003, 1, 3297 – 3303.
[14] A. K. Ghosh, L. M. Swanson, H. Cho, S. Leshchenko, K. A.
Hussain, S. Kay, D. E. Walters, Y. Koh, H. Mitsuya, J. Med.
Chem. 2005, 48, 3576 – 3585.
[15] T. Li, D. Saro, M. R. Spaller, Bioorg. Med. Chem. Lett. 2004, 14,
1385 – 1388.
[16] E. Perola, P. S. Charifson, J. Med. Chem. 2004, 47, 2499 – 2510.
[17] A. Reichelt, S. F. Martin, Acc. Chem. Res. 2006, 39, 433 – 442.
[18] J. M. Bradshaw, G. Waksman, Adv. Protein Chem. 2002, 61, 161 –
210.
[19] K. Machida, B. J. Mayer, Biochim. Biophys. Acta 2005, 1747, 1 –
25.
[20] J. P. Davidson, O. Lubman, T. Rose, G. Waksman, S. F. Martin, J.
Am. Chem. Soc. 2002, 124, 205 – 215.
[21] J. P. Davidson, PhD thesis, The University of Texas, 2001.
[22] J. D. Dunitz, Chem. Biol. 1995, 2, 709 – 712.
[23] D. H. Williams, E. Stephens, D. P. OꢀBrien, M. Zhou, Angew.
Chem. 2004, 116, 6760 – 6782; Angew. Chem. Int. Ed. 2004, 43,
6596 – 6616.
[24] H. R. Plake, T. B. Sundberg, A. R. Woodward, S. F. Martin,
Tetrahedron Lett. 2003, 44, 1571 – 1574.
[25] J. M. Sturtevant, Proc. Natl. Acad. Sci. USA 1977, 74, 2236 –
2240.
[26] P. Nioche, W.-Q. Liu, I. Brouitin, F. Charbonnier, M.-T. Latreille,
M. Vidal, B. Roques, C. Garbay, A. Ducruix, J. Mol. Biol. 2002,
315, 1167 – 1177.
[27] X-ray crystallographic data are deposited in the Protein Data
Bank under the accession codes 2HUW and 2HUY.
[28] J. Rahuel et al., Nat. Struct. Biol. 1996, 3, 586 – 589; see
Supporting Information.
[29] J. Rahuel, C. Garcia-Echeverria, P. Furet, A. Strauss, G.
Caravatti, H. Fretz, J. Schoepfer, B. Gay, J. Mol. Biol. 1998,
279, 1013 – 1022.
[30] C.-Y. Yang, R. Wang, S. Wang, J. Med. Chem. 2005, 48, 5648 –
5650.
Angew. Chem. Int. Ed. 2006, 45, 6830 –6835
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
6835