Bridged peptide macrocycles as ligands for PDZ domain proteins

Article abstract of DOI:10.1021/ol0475966

(Chemical Equation Presented) Conformationally constrained side chain-bridged cyclic peptides were prepared using bis-carboxylic acid ring spacers. These macrocyles were designed to inhibit protein-protein interactions mediated by the third PDZ domain (PDZ3) of a mammalian neuronal protein, PSD-95. Isothermal titration calorimetry (ITC) experiments measured dissociation constants in the low micromolar range. For each compound, the change in entropy (TΔS) of binding either is comparable in magnitude to the enthalpy change (ΔH) or is the predominant driving force for association.

Full text of DOI:10.1021/ol0475966

ORGANIC
LETTERS
2005
Vol. 7, No. 7
1203-1206
Bridged Peptide Macrocycles as
Ligands for PDZ Domain Proteins
Gomika Udugamasooriya, Dorina Saro, and Mark R. Spaller*
Department of Chemistry, Wayne State UniVersity, Detroit, Michigan 48202
mspaller@chem.wayne.edu
Received November 22, 2004 (Revised Manuscript Received February 18, 2005)
ABSTRACT
Conformationally constrained side chain-bridged cyclic peptides were prepared using bis-carboxylic acid ring spacers. These macrocyles
were designed to inhibit protein protein interactions mediated by the third PDZ domain (PDZ3) of a mammalian neuronal protein, PSD-95.
Isothermal titration calorimetry (ITC) experiments measured dissociation constants in the low micromolar range. For each compound, the
−
change in entropy (T
association.
∆S) of binding either is comparable in magnitude to the enthalpy change (∆H) or is the predominant driving force for
Cyclic peptides are an enticing molecular class for the
development of protein-binding ligands for a variety of
reasons: the prospect of high affinity and selectivity; cellular
stability from proteolytic resistance; and relative ease of
assembly compared to nonpeptide organic macrocycles.1
These were all motivating factors when we initiated a
program to design cellular probes for the PDZ domain protein
family, which led to the development of a nontraditional form
of macrocyclic peptide for the third PDZ domain (PDZ3) of
the neuronal postsynaptic density-95 protein (PSD-95).2
Several compounds from this first generation exhibited good
in vitro affinity, and a selected member demonstrated efficacy
in vivo.3 We now report further progress in this direction,
presenting the design, synthesis, and thermodynamic binding
analysis for a new series of PDZ domain-targeting macro-
cycles employing a modified ring assembly strategy.
in neuronal cells.5 The development of stable, selective, high-
affinity chemical probes for PDZ domains will aid cellular
studies investigating endogenous activity of proteins contain-
ing these binding modules.
The three PDZ domains of PSD-95 are of particular
neurobiological interest, given the functional roles exhibited
by the parent protein at the synaptic membrane. From a
pharmacological perspective, targeting PSD-95 may prove
to be useful for therapeutic intervention in conditions such
as stroke-associated ischemic brain damage6 and psycho-
stimulant addiction.7
Coupled to our interest in developing tools for PDZ cell
biology was the desire to address a fundamental biophysical
question: what are the thermodynamic consequences of
attempting to improve the binding affinity of a linear ligand
though cyclization? With their structural stability and
penchant for recognizing relatively small epitopes, many
PDZ domains may serve as “nonartificial model systems”
for investigating protein-protein thermodynamic binding
properties and serve as useful platforms for the development
and testing of new constrained ligand designs.
As a large class of autonomously folding and binding
substructure found within larger proteins, PDZ domains are
active participants in mediating protein-protein interactions
in numerous mammalian proteins,4 most notably those found
In this study, we expand upon a strategy developed for
generating cyclic peptides in which R-, â-, and γ-amino acids
(1) (a) Kessler, H. Angew. Chem., Int. Ed. Engl. 1982, 21, 512-523.
(b) Rizo, J.; Gierasch, L. M. Annu. ReV. Biochem. 1992, 61, 387-418.
(2) Li, T.; Saro, D.; Spaller, M. R. Bioorg. Med. Chem. Lett. 2004, 14,
1385-1388.
(3) Piserchio, A.; Salinas, G. D.; Li, T.; Marshall, J.; Spaller, M. R.;
Mierke, D. F. Chem. Biol. 2004, 11, 469-473.
(5) Kim, E.; Sheng, M. Nat. ReV. Neurosci. 2004, 5, 771-781.
(6) Aarts, M.; Liu, Y.; Liu, L.; Besshoh, S.; Arundine, M.; Gurd, J. W.;
Wang, Y. T.; Salter, M. W.; Tymianski, M. Science 2002, 298, 846-850.
(7) Roche, K. W. Trends Neurosci. 2004, 27, 699-700.
(4) Sheng, M.; Sala, C. Annu. ReV. Neurosci. 2001, 24, 1-29. (b) Harris,
B. Z.; Lim, W. A. J. Cell. Sci. 2001, 114, 3219-3231.
10.1021/ol0475966 CCC: $30.25
© 2005 American Chemical Society
Published on Web 03/08/2005

were used as bridging elements to tether amino- and carboxy-
functionalized residue side chains (Figure 1, route A).2 This
Lys-Gln-Thr-Ser-Val) of the CRIPT protein.8 The bisacids
were chosen on the basis of their propensities to impart
varying degrees of length and rigidity to the resulting
macrocyclic rings. Although there are nonproteinogenic
amino acids that possess amine-bearing side chains of
different lengths that can be used to further attenuate the
ring size,2 for this initial investigation these were kept fixed
with lysine.
Two synthetic routes were devised to achieve the required
protecting group orthogonality, depending on whether the
bisacid tether was purely carbogenic (1-4, Scheme 1) or
Scheme 1. Synthetic Route for Ligands with
Nonfunctionalized Bisacid Bridging Elements (1-4).
Figure 1. Design strategies for side chain-bridged ligands.
modular approach allows for the rapid and convenient
manipulation of functionality as well as bridge length and
ring size. The present design employs an organic bis-
carboxylic acid as a bridging constraint, in which both
carboxylates form amide bonds with the two amine-present-
ing side chains within the peptide backbone (Figure 1, route
B). As in keeping with a design criterion from our prior
generation of ligands, the critical residues at the “0” and
“-2” positions remain exposed and available for binding to
the PDZ domain.
Seven macrocyclic peptides (1-7) were prepared using
bisacid bridges to connect the side chains of lysine residues
occupying the “-1” and “-3” positions in the hexapeptide
Tyr-Lys-Lys-Thr-Lys-Val (Figure 2). This peptide is based
upon a known PDZ3-binding sequence, the C-terminus (Tyr-
whether an amine functionality was present (5-7, Scheme
2). Bisacids 5 and 7 bore Fmoc amino protection prior to
coupling to the Lys side chain; the former was prepared by
adapting a known procedure for protecting a structurally
related aniline.9 A linear control, 8, was prepared using
Scheme 2, except that the second coupling step (amidation
of the “-1” Lys) was not performed.
PDZ3-ligand interactions were analyzed using isothermal
titration calorimetry (ITC), a solution technique that provides
the full panel of thermodynamic binding parameters; these
include the changes in Gibbs free energy (∆G), enthalpy
(8) Niethammer, M.; Valtschanoff, J. G.; Kapoor, T. M.; Allison, D.
W.; Weinberg, T. M.; Craig, A. M.; Sheng, M. Neuron 1998, 20, 693-
707.
(9) Neustadt, B. R.; Smith, E. M.; Nechuta, T.; Zhang, Y. Tetrahedron
Lett. 1998, 39, 5317-5320.
Figure 2. Peptide scaffold and bisacid bridging components for
ligands 1-8.
1204
Org. Lett., Vol. 7, No. 7, 2005

Scheme 2. Synthetic Route for Ligands with
Table 1. Thermodynamic Values for the Calorimetric Titration
of Bisacid-Bridged Cyclic Peptides into PDZ3 from PSD-95a
Amine-Functionalized Bisacid Bridging Elements (5-7)a
∆G
∆H
T∆S
compd
Kd (µM)
(kcal/mol)
(kcal/mol)
(kcal/mol)
1
15.4
-6.59
-1.85
4.74
((4.1)
7.17
((0.55)
4.87
((0.56)
3.78
((0.02)
3.76
((0.31)
6.34
((0.17)
15.0
((0.3)
14.9
((0.15)
-7.02
((0.05)
-7.25
((0.07)
-7.41
((0.02)
-7.40
((0.05)
-7.09
((0.01)
-6.58
((0.01)
-6.60
((0.02)
-3.44
((0.30)
-3.88
((0.21)
-1.87
((0.16)
-1.85
((0.05)
-3.49
((0.01)
-2.66
((0.02)
-3.32
((0.18)
3.58
((0.25)
3.37
((0.14)
5.53
((0.14)
5.55
((0.10)
3.60
((0.03)
3.92
((0.03)
3.28
2
3
4
5
6
7
8
((3.6)
((0.15)
((0.00)
((0.15)
a Values are the mean of at least two independent experiments (error
shown below reflects the range). Binding stoichiometry values (n) ranged
from 0.93 to 1.11.
experienced by the 1,3-phenyl-linked 4 over less-constrained
3 is compensated by a nearly matched change in ∆∆H
(+2.01 kcal/mol).
a For 7, the bisacid bridge (Glu) is protected as a phenylisopropyl
ester (Pip) at its side chain carboxylate.
(∆H), and entropy (T∆S), as well as the stoichiometry (n).10
While the affinity (Kd or ∆G) is of primary interest when
assessing and ranking the efficacy of ligands, changes in
enthalpy and entropy, reflecting specific molecular interac-
tions, ultimately determine the observed binding constants.
Since thermodynamic values can often be rationalized with
different molecular explanations, in the absence of explicit
structural characterization the interpretations are tentative.
Data from the separate ITC experiments, in which the
solubilized ligands were individually titrated into a solution
of PDZ3, are shown in Table 1 and Figure 3. The structural
modification of 1f2, as well as that of 3f4, is expected to
increase both the rigidity and the hydrophobic surface area
of the ligand. Either preorganization into the bound confor-
mation or desolvation of nonpolar regions of a molecule
would lead to a gain in the observed change in entropy, if
not the actual affinity. But should T∆S increase as desired,
the confounding and often unpredictable effect of entropy-
enthalpy compensation frequently dampens the anticipated
improvement in ∆G by somehow arranging molecule inter-
relationships, including those with solvent, to reduce the net
favorable contribution of enthalpy.11 So while the slight
2-fold improvement in Kd of the 1,2-phenyl-linked 2 over
the flexible ethyl chain of 1 is due to a gain in -∆H at the
expense of T∆S, the favorable T∆∆S (+2.16 kcal/mol)
Figure 3. Bar graph of thermodynamic parameters for ligands 1-8.
Note that T∆S is plotted as the negative value.
It is possible to analyze compounds 1-4 from a different
perspective, as ring expansion (1f3) and ring isomerization
(2f4) perturbations. The addition of a methylene equivalent
to 1 adds a partially rotatable bond and leads to an expected
unfavorable decrease in binding entropy. Yet the affinity
improves 3-fold, due to a surge in the ∆H contribution. This
may result from a ring that is able to adopt a conformation
(10) Ward, W. H.; Holdgate, G. A. Prog. Med. Chem. 2001, 38, 309-
376.
(11) Sharp, K. Protein Sci. 2001, 10, 661-667.
Org. Lett., Vol. 7, No. 7, 2005
1205

improving specific protein-ligand interactions, whereas the
presumably more constrained 1 cannot access such a state
without some energetic cost. The strategy of methylene unit
augmentation (as in 1f3) was a key feature of our earlier
PDZ-targeting macrocycles2 but has also been applied in
diverse ways to ligand design for the SH2 domain,12 in
peptide mimics of the arginine-rich motif of the HIV-1 Tat
protein,13 in agonists for receptors,14 and for enzyme inhibi-
tors.15
mol. This might be the price of restricting the additional
rotors introduced with the Phe residue.
Along similar lines, the amino analogue of 3 was also
prepared (7), although in this case the affinity decreased.
This ligand demonstrates that our design strategy is not
restricted to symmetrical bisacids, although the use of
asymmetrical bridge components does require the protection
of one of the carboxylates to ensure that assembly yields a
single isomer.
The comparison between ligands 2 and 4 attempts to
diminish the contribution of desolvation differences that
might be due to nonequivalent hydrophobic surface areas,
since the pair are isomers. While 4 possesses a slightly
elongated internal ring size, it is tied with 5 for largest
entropy change of the series, with a T∆∆S of almost 2 kcal/
mol over 2. In the absence of further structural or confor-
mational characterization, however, it cannot be concluded
that 4 is necessarily more rigid or constrained than 2, since
there may be additional global restraints present in the
former.
The next step in development was to functionalize the
bridge to permit the rapid formation of derivatives. This
specific design strategy was implemented with the idea of
preparing ligands that might exhibit selectivity between
various PDZ domains by allowing for unique interactions
with regions distal from the canonical binding site. NMR
structural studies of an earlier developed amino acid-bridged
macrocyle with a PDZ domain showed that contacts can
occur between the protein surface and the bridging unit.3
Placement of an amine on the bridge (5) maintains the
same affinity as the unmodified parent, allowing for potential
“expansion” of the ligand binding surface through various
chemical transformations. As an example, a simple amidation
reaction was performed in which Fmoc-phenylalanine was
coupled to the amine in a manner analogous to peptide
synthesis (6; Scheme 2). With an enthalpic enhancement of
∆∆H ≈ 1.6 kcal/mol, this suggests that additional binding
interactions may be formed with 6 that are not possible with
5. Even if this speculation is correct, the energetic enhance-
ment is effectively abrogated by a more than compensating
decrease in entropy, with a T∆∆S of approximately -2 kcal/
A control ligand (8) was also synthesized, which represents
a linear analogue of 4. Compound 8 can be thought of as a
bond-cleavage product of 4 if the latter were hydrolyzed at
the “-1” lysine side chain. The macrocycle 4 does, in fact,
exhibit a favorable change in entropy over the linear 8 of
T∆∆S ≈ 2.3 kcal/mol, which is the source for a 3-fold
improvement in affinity. Although 8 is linear, in fact two
changes have been wrought: ring rupture and formation of
ionizable amine and carboxylic acid functional groups.
Ideally, an appropriate control would entail the cleavage of
an aliphatic C-C bond, so as to minimize complicating
effects due to potential changes in bonding interactions.
In summary, the presented modular ligand design enables
the synthetic addition or subtraction of single-carbon units,
thus allowing for incremental expansion or contraction of
ring size, which can impact affinity and, perhaps eventually,
selectivity. Combining the results from this report with those
of our previous study,2 we have demonstrated that such
modulation can significantly influence the binding strength
of PDZ domain-ligand interactions. Further, the PDZ
macrocycles prepared provide useful information about the
nature of constrained protein-binding molecules and can
potentially serve as compounds for the design of cellular
probes.
Finally, this macrocyclic scaffold may also prove to be of
use for ligand development outside the realm of PDZ
domains, since the design is in keeping with the notion of
creating a broader, wider binding surface area that may be
necessary to inhibit association with certain proteins. Many
protein-protein interaction interfaces have larger surface
areas that may not succumb to disruption in the presence of
a competing linear inhibitor; this report provides one possible
solution by more densely functionalizing a compound without
simply resorting to elongation.
(12) (a) Lee, K.; Zhang, M.; Liu, H.; Yang, D.; Burke, T. R., Jr. J. Med.
Chem. 2003, 46, 2621-2630. (b) Long, Y. Q.; Lung, F. D.; Roller, P. P.
Bioorg. Med. Chem. 2003, 11, 3929-3936.
(13) Friedler, A.; Friedler, D.; Luedtke, N. W.; Tor, Y.; Loyter, A.; Gilon,
C., J. Biol. Chem. 2000, 275, 23783-23789.
(14) (a) Lindman, S.; Lindeberg, G.; Gogoll, A.; Nyberg, F.; Karlen,
A.; Hallberg, A. Bioorg. Med. Chem. 2001, 9, 763-772. (b) Condon, S.
M.; Darnbrough, S.; Burns, C. J.; Bobko, M. A.; Morize, I.; Uhl, J.; Jariwala,
N. U.; Burke, K.; Labaudiniere, R. F. Bioorg. Med. Chem. 2002, 10, 731-
736.
(15) (a) Andrews, M. J.; McInnes, C.; Kontopidis, G.; Innes, L.; Cowan,
A.; Plater, A.; Fischer, P. M. Org. Biomol. Chem. 2004, 2, 2735-2741. (b)
Hu, X.; Nguyen, K. T.; Jiang, V. C.; Lofland, D.; Moser, H. E.; Pei, D. J.
Med. Chem. 2004, 47, 4941-4949.
Acknowledgment. Funding for this research was pro-
vided by the National Institutes of Health (GM63021).
Supporting Information Available: Experimental pro-
cedures for preparation and characterization of peptide
ligands. This material is available free of charge via the
Internet at http://pubs.acs.org.
OL0475966
1206
Org. Lett., Vol. 7, No. 7, 2005