Investigation of the PDZ domain ligand binding site using chemically modified peptides

Article abstract of DOI:10.1016/S0960-894X(02)00345-1

Several chemically modified analogues to a tightly binding ligand for the second PDZ domain of MAGI-3 were synthesized and evaluated for their ability to compete with native peptide ligands. N-methyl scanning of the ligand backbone amides revealed the energetically important hydrogen bonds between the ligand backbone and the PDZ domain. Analogues to the ligand's conserved threonine/serine(-2) residue, involved in a side chain to side chain hydrogen bond with a conserved histidine in the PDZ domain, revealed that the interaction is highly sensitive to the steric structure around the hydroxyl group of this residue. Analogues of the ligand carboxy terminus revealed that the full hydrogen bond network of the GLGF loop is important in ligand binding.

Full text of DOI:10.1016/S0960-894X(02)00345-1

Bioorganic & Medicinal Chemistry Letters 12 (2002) 2471–2474
Investigation of the PDZ Domain Ligand Binding Site Using
Chemically Modified Peptides
Kathleen A. P. Novak, Naoaki Fujii and R. Kiplin Guy*
Departments of Pharmaceutical Chemistry and Cellular and Molecular Pharmacology, University of California,
513 Parnassus Avenue, San Francisco, CA 94143-0446, USA
Received 26 February 2002;accepted 22 April 2002
Abstract—Several chemically modified analogues to a tightly binding ligand for the second PDZ domain of MAGI-3 were synthe-
sized and evaluated for their ability to compete with native peptide ligands. N-methyl scanning of the ligand backbone amides
revealed the energetically important hydrogen bonds between the ligand backbone and the PDZ domain. Analogues to the ligand’s
conserved threonine/serine(À2) residue, involved in a side chain to side chain hydrogen bond with a conserved histidine in the PDZ
domain, revealed that the interaction is highly sensitive to the steric structure around the hydroxyl group of this residue. Analogues
of the ligand carboxy terminus revealed that the full hydrogen bond network of the GLGF loop is important in ligand binding.
# 2002 Elsevier Science Ltd. All rights reserved.
The identification of adaptor proteins has led to the idea
that organization of protein complexes at the cellular
membrane is a crucial aspect of signal transduction.1
Adaptors normally utilize clusters of modular protein
interaction domains to control protein complex assem-
bly. One important class of interaction domain in
adaptors is the PDZ domain, a 100 amino acid module
that normally binds a short carboxy terminal motif on
its cognate ligands.2 PDZ domains segregate into three
major classes: Class I, which binds the S/T-X-V motif;
Class II, which binds F/L-X-V/L;and Class III, which
binds D/E-X-V. Structural3 and biochemical4 studies of
PDZ domains have suggested some requirements for
interactions between PDZ domains and their ligands.
The most commonly observed contacts for class I PDZ
domains (Fig. 1) are: (1) a hydrogen bond network
between the ligand carboxylate anion and both amide
protons from the backbone of the PDZ domain’s
GLGF motif and the side chain of a conserved basic
residue on the PDZ domain, (2) specific hydrogen bonds
between the backbone amides of the PDZ domain and
the ligand, (3) hydrophobic interactions between the
ligand’s terminal V(0) side chain and a PDZ domain
pocket lined with hydrophobic side chains, and (4) a
hydrogen bond between the ligand S/T(À2) side chain
and a H in a helix 2 of the PDZ domain. Other points of
interaction that might allow for discrimination between
related ligands by the PDZ domain include (1) specific
interactions between the (À1) residue side chain of the
ligand and those of b strand C of the PDZ domain, (2)
the ligand’s (À3) residue interacting with b strand B,
and (3) ligand residues past (À3) interacting with a
variable loop region between b strands B and C. As part
of our program aimed at specifically blocking function
of individual PDZ domains, we desired to carefully
evaluate the role of these elements.
Figure 1. Recognized features of the ligand–PDZ domain interaction
and proposed interactions under investigation as adapted from Doyle
et al.3a The peptide sequence shown is that of the native PTEN car-
boxy terminus.
*Corresponding author. Tel.: +1-415-502-7051;fax: +1-415-514-
0689;e-mail: rguy@cgl.ucsf.edu
0960-894X/02/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(02)00345-1

2472
K. A. P. Novak et al. / Bioorg. Med. Chem. Lett. 12 (2002) 2471–2474
To date, there have been no reports of detailed investi-
gation of the hydrogen bond network between domain/
ligand backbones or the size and shape of the PDZ
domain’s hydrophobic pockets. To address these issues,
we prepared a library of peptides and peptidomimetics
designed from the homology model of PDZ2 of MAGI-
3 bound to a peptide ligand representing the carboxy
terminus of tumor suppressor PTEN.5 We chose the
following peptide modifications (Fig. 2): (1) serial trun-
cation of the peptide to identify the minimum length for
the efficient binding (Panel A), (2) serial substitution of
peptide-bond amide hydrogens with methyl groups (N-
methyl scanning) to identify important hydrogen bond
donors (Panel B), (3) replacement of the carboxy termi-
nus with other functionalities (Panel C), and (4) modifi-
cation of the hydrophobic regions of the X(À1) and S/
T(À2) side chains (Panel C). We compared the binding
affinity of these modified ligands toward PDZ2 of
MAGI-3 with that of a native ligand using a fluorescence
polarization competition assay.4d,6
weaker binding than 3c. N-Methylation may change
amide bond geometry from E-form to Z-form11 thus
inducing the overall reduction in binding ability. Dis-
parity in the affinity of these peptides may reveal the
importance of the amide hydrogens or may reflect a
greater propensity for b branched residues to allow a
conformational change. The order of affinity of the
peptides: 4e/4b >4d >4c > >4a may indicate that
amide hydrogens on V(0) and T(À2) are important and
the I(À3) amide hydrogen is also contributing to effi-
cient binding. Meanwhile, amide hydrogens on W(À1)
and Q(À5) have significantly less contribution. These
results are consistent with the hydrogen bond networks
inferred from the co-crystal structure of KQTSV
peptide binding to the PSD-95 PDZ3 domain.3a
We produced OregonGreenTM-tagged fluorescent pep-
tides 1a (the native sequence of PTEN,7 a tumor sup-
pressor that binds to MAGI-2/3 PDZ2 domain8) and 1b
(a high affinity peptide for the MAGI-3 PDZ2 domain5)
by standard Fmoc peptide synthesis protocols. Con-
sistent with prior ELISA results, peptide 1b showed
almost a 10-fold tighter affinity for MAGI-2/3 PDZ2
than 1a.5 Anticipating relatively weak binding from our
library of modified peptides, we chose the natural
sequence 1a as our probe for the competition binding
assay. We synthesized several modified versions of 1b in
order to test hypotheses about the role in ligand binding
of the various moieties.
Few reports have addressed the significance of residues
further amino terminal than S/T(À2) in the ligand con-
sensus motif.4c To investigate the contribution of other
residues in the extended motif we truncated the AcHT-
QITWV-OH peptide (3d) sequentially (Fig. 2A) by one
amino acid. All of the truncated peptides 3a–3d showed
weaker affinity than longer peptide 2b. Presumably, the
contribution by residues more amino terminal than the
H(À6) imparts tighter affinity as seen with 2b. This is
also evident in the weak binding shown by the negative
control peptide 2a, which has an incomplete pharmaco-
phore. However, peptides 3b, 3c and 3d of five, six, and
seven amino acids, respectively, showed similar affinity,
while 3a, a peptide of only four amino acids, showed
even weaker binding. While these data are consistent
with reported observations that five residues confer
PDZ domain recognition, they do indicate a role for
the more extended amino acids as has been previously
proposed.4a,c
Figure 2. Designed peptides from the carboxy terminus sequence of
PTEN. Unless otherwise specified, peptides were synthesized from
commercially available amino acids using standard Fmoc protocols.9
All were purified to homogeneity by RP-HPLC. Structures were con-
a
.
firmed by HRMS. Conditions for generating 4a: (a) HCl H-V-OMe (1
equiv), 2-(O2N)PhSO2Cl (1.1 equiv), DIPEA (3 equiv), DMF, rt, 3
days: then added 4-(O2N)PhSO3Me (2.6 equiv), MTBD (6 equiv), rt, 4
h, 97% overall;(b) MeOH, NaOH aq (10%), 50 ꢀC, 4 h;(c) PMB-Cl
(1.2 equiv), K2CO3 (2.2 equiv), DMF, rt, 12 h;(d) HOCH 2CH2SH (5.0
equiv), DBU (2.5 equiv), DMF, rt, 10 min, 3 steps 85% overall yield;
(e) Fmoc-W(Boc)-OH (1 equiv), HATU (1.3 equiv), DIPEA (3 equiv),
DMF, rt, 1 h;(f) 96% TFA, 2% TIS, 2% H 2O, rt;(g) 2-chlorotrityl
chloride resin, DIPEA , CH2Cl2, rt, 3 days;(h) repeat the following
protocols: Fmoc amino acid (2.5 equiv), HBTU (2.4 equiv), DIPEA (5
equiv), DMF, rt, 0.5 h then 20% piperidine in DMF, rt, 10 min, three
times;(i) Ac 2O (4 equiv), DIPEA (5 equiv), DMF, rt, 0.5 h;(j) 96%
TFA, 2% TIS, 2% H2O, rt; bPreparation of 5a: (a) H2NOTHP (10
equiv), AcT(tBu)Q(Tr)IT(tBu)W(Boc)V-OH (1 equiv), HATU (9
equiv), DIPEA (20 equiv), DMF, 50 ꢀC, 12 h;(b) 96% TFA, 2% TIS,
2% H2O, rt. cPreparation of 5b: (a) l-valinol (3 equiv), AcT(tBu)Q-
(Tr)IT(tBu)W-(Boc)-OH (1 equiv), HBTU (2.4 equiv), DIPEA (8
equiv), DMF, 50 ꢀC, 1 h;(b) 96% TFA, 2% TIS, 2% H 2O, rt. dPre-
paration of 6: (a) H-W-OH (1 equiv), PhCH2Br (3.2 equiv), K2CO3 (4
equiv), DMF, 50 ꢀC, 12 h;(b) MeI (large excess), NaH (large excess),
DMF, rt, 0.5 h, two steps quantitative;(c) HCO 2NH4 (10 equiv), 10%
Pd/C, MeOH, THF, H2O, reflux, 12 h;(d) Fmoc-OSu (2.2 equiv),
THF, H2O, pH 9, rt, 12 h, two steps 57% overall;(e) H-V-OWang
resin, HBTU (2.4 equiv), DIPEA (5 equiv), CH2Cl2, rt, 1 h then 20%
piperidine in DMF, rt, 10 min, three times;(f) repeat the following
protocols: Fmoc amino acid (2.5 equiv), HBTU (2.4 equiv), DIPEA (5
equiv), DMF, rt, 0.5 h then 20% piperidine in DMF, rt, 10 min, 3
Based on these results, we carried out N-methyl scan-
ning of 3c at the five amide nitrogens in the peptide
backbone (Fig. 2B) by a reported methodology.10 The
steric effects of the valine isopropyl precluded 4a from
being synthesized through this method. A liquid-phase
version of the solid-phase method was applied to syn-
thesize N-methyl valine derivatives as described in
Figure 2. All of the methylated peptides, 4a–4e, showed
times;(g) Ac O (4 equiv), DIPEA (5 equiv), DMF, rt, 0.5 h;(h) 96%
2
TFA, 2% TIS, 2% H2O, rt.

K. A. P. Novak et al. / Bioorg. Med. Chem. Lett. 12 (2002) 2471–2474
2473
We next modified individual residues to investigate the
flexibility and character of the PDZ binding pocket. The
carboxy terminus was substituted with other functional
groups (Fig. 2C). A neutral alcohol 5b almost com-
pletely abolished the peptide’s binding while a weakly
acidic hydroxamide 5a retains weak binding ability.
These results suggest that the electrostatic interaction of
the ligand carboxylate and a conserved cationic lysine in
the MAGI-3 PDZ2 domain which has been suggested to
mediate ligand binding are more important than pre-
serving size. Terminating the peptide as an amide, a
modification that increases bulk and decreases electro-
static potential, has previously been reported to sig-
nificantly decrease affinity of the ligand.4d,5 There are
several reports about the lack of firm preferences for a
particular side chain at the X(À1) position.4a,4b To
investigate this residue, the W(À1) of was minimally
modified by methylation on its indole-1 position (Fig.
2C). Unexpectedly, 6 demonstrated reduced affinity
compared with 3c suggesting that there is a possibility
to design selective ligands for PDZ domains by modifi-
cation of the X(À1) region.4c Finally, we modified the
T(À2) side chain region by changing the methyl group
to other small hydrophobic moieties (Fig. 2C). As
expected, serine-substituted peptide 7a binds with simi-
lar extent to 3c. Otherwise, all other modifications
including an inversion of chirality of the threonine
hydroxyl group (7b), dimethyl substitution (7c), ethyl
substitution (7d), and elongation (7e) nearly abolished
binding. Therefore, interaction between the S/T(À2)
hydroxyl group and histidine side chain in the a helix 2
in class I PDZ domain is strictly regulated by the
approximate size of the binding pocket (Table 1).
Our results indicate that the ligand binding pocket of
the MAGI-3 PDZ2 domain is highly sensitive to alter-
ations in the carboxylate terminus and the individual
residue side chains. In particular, the S/T(À2) residue,
which has been shown to interact with a well conserved
histidine in class I binding PDZ domains, could not be
substituted with electrostatically similar but more bulky
alcohol side chains. Additionally, the carboxylate bind-
ing pocket cannot accommodate large changes in elec-
trostatic potential or added bulk.4d Therefore, it is
essential for affinity to a class I PDZ domain that these
interactions be fulfilled. Any successful competitor must
satisfy these requirements and incorporate features that
will discriminate between PDZ domains. The study of
the modified (À1) side chain shows that alterations in
this position can greatly affect binding affinity. Making
use of this region in concert with more extended inter-
actions with b strand C and the variable loop between
b strands B and C may be important to gain specificity.
The information obtained by this work would be useful
in the design of novel inhibitors of the PDZ domain/
ligand interaction, especially between MAGI-3 PDZ2
domain and PTEN.
Acknowledgements
We thank the Sidney Kimmel Foundation for Cancer
Research, the HHMI Research Program (Grant #
76296–549901) and UCSF for financial assistance.
Table 1. Affinity of modified peptides for MAGI-3 PDZ2
Direct binding
Kd (mM)Æstandard errorb
1.0Æ0.05
a
Ligand
1a
1b
References and Notes
0.02Æ0.0005
Competition binding ligand
IC50c (mM)Æstandard errorb
2a
2b
3a
3b
3c
3d
4a
4b
4c
4d
4e
5a
5b
6
14.4Æ0.6d
1. (a) Pawson, T.;Scott, J. D. Science 1997, 278, 2075. (b)
Schillace, R. V.;Scott, J. D. J. Clin. Invest. 1999, 103, 761.
2. (a) Harris, B. Z.;Lim, W. A. J. Cell Sci. 2001, 114, 3219.
(b) Sheng, M.;Sala, C. Annu. Rev. Neurosci. 2001, 24, 1.
3. (a) Doyle, D. A.;Lee, A.;Lewis, J.;Kim, E.;Sheng, M.;
MacKinnon, R. Cell 1996, 85, 1067. (b) Morais Cabral, J. H.;
Petosa, C.;Sutcliffe, M. J.;Raza, S.;Byron, O.;Poy, F.;
Marfatia, S. M.;Chishti, A. H.;Liddington, R. C. Nature
1996, 382, 649. (c) Daniels, D. L.;Cohen, A. R.;Anderson,
J. M.;Brunger, A. T. Nat. Struct. Biol. 1998, 5, 317. (d) Hil-
lier, B. J.;Christopherson, K. S.;Prehoda, K. E.;Bredt, D. S.;
Lim, W. A. Science 1999, 284, 812. (e) Kozlov, G.;Gehring,
K.;Ekiel, I. Biochemistry 2000, 39, 2572. (f) Tochio, H.;Hung,
0.74Æ0.2
14.8Æ1.5d
4.5Æ0.1e
5.2Æ0.2e
4.1Æ0.6e
Not detectedg
10.6Æ0.2f
29.0Æ1.8
20.4Æ2.1
9.8Æ0.7f
42.1Æ7.9
Not detected
34.0Æ0.6
7a
7b
7c
7d
7e
5.1Æ0.6e
F.;Li, M.;Bredt, D. S.;Zhang, M.
J. Mol. Biol. 2000, 295,
Not detected
Not detected
Not detected
Not detected
225. (g) Karthikeyan, S.;Leung, T.;Birrane, G.;Webster, G.;
Ladias, J. A. J. Mol. Biol. 2001, 308, 963. (h) Karthikeyan, S.;
Leung, T.;Ladias, J. A. J. Biol. Chem. 2001, 276, 19683.
4. (a) Schultz, J.;Hoffmuller, U.;Krause, G.;Ashurst, J.;
Macias, M. J.;Schmieder, P.;Schneider-Mergener, J.;Osch-
kinat, H. Nat. Struct. Biol. 1998, 5, 19. (b) Songyang, Z.;
Fanning, A. S.;Fu, C.;Xu, J.;Marfatia, S. M.;Chishti, A. H.;
Crompton, A.;Chan, A. C.;Anderson, J. M.;Cantley, L. C.
Science 1997, 275, 73. (c) Niethammer, M.;Valtschanoff, J. G.;
Kapoor, T. M.;Allison, D. W.;Weinberg, T. M.;Craig, A. M.;
Sheng, M. Neuron 1998, 20, 693. (d) Harris, B. Z.;Hillier, B. J.;
Lim, W. A. Biochemistry 2001, 40, 5921.
aDissociation constants determined by plotting data to y=bottom+
(topÀbottom) *x/(Kd+x).
bValues are means of three experiments.
cCompetitor concentration that inhibits probe binding by 50% as
determined by fitting data to one site competition equation y=bottom
+(topÀbottom)/(1+x/IC50).
e,fResults are not statistically different (pꢁ0.05) by Mann–Whitney
test.
gIndicates IC50>50 mM and unable to be determined by our assay.

2474
K. A. P. Novak et al. / Bioorg. Med. Chem. Lett. 12 (2002) 2471–2474
5. Fuh, G.;Pisabarro, M. T.;Li, Y.;Quan, C.;Lasky, L. A.;
Sidhu, S. S. J. Biol. Chem. 2000, 275, 21486.
6. (a) Parker, G. J.;Law, T. L.;Lenoch, F. J.;Bolger, R. E. J.
Biomol. Screen 2000, 5, 77. (b) Seethala, R. Methods 2000, 22,
61.
9. Wellings, D. A.;Atherton, E. In Methods in Enzymology;
Fields, G., Ed.;Academic: New York, 1997;Vol. 289, p 44.
10. (a) Miller, S. C.;Scanlan, T. S. J. Am. Chem. Soc. 1997,
119, 2301. (b) Miller, S. C.;Scanlan, T. S. J. Am. Chem. Soc.
1998, 120, 2690.
7. (a) Cantley, L. C.;Neel, B. G. Proc. Natl. Acad. Sci. U.S.A.
1999, 96, 4240. (b) Maehama, T.;Dixon, J. E. Trends Cell
Biol. 1999, 9, 125. (c) Vazquez, F.;Sellers, W. R. Biochim.
Biophys. 2000, 1470, M21.
11. (a) Portnova, S. L.;Bystrov, V. F.;Balashova, T. A.;
Ivanov, V. T.;Ovchinnikov, Y. A. Izv. Akad. Nauk SSSR,
Ser. Khim 1970, 825. (b) Fogarasi, G.;Pulay, P.;Torok, F.;
Boggs, J. E. J. Mol. Struct. 1979, 57, 259. (c) Vitoux, B.;
8. (a) Wu, X.;Hepner, K.;Castelino-Prabhu, S.;Do, D.;
Kaye, M. B.;Yuan, X.-J.;Wood, J.;Ross, C.;Sawyers, C. L.;
Whang, Y. E. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 4233. (b)
Wu, Y.;Dowbenko, D.;Spencer, S.;Laura, R.;Lee, J.;Gu,
Q.;Lasky, L. A. J. Biol. Chem. 2000, 275, 21477.
Boussard, G.;Manh Thong, C.;Marraud, M.;Aubry, A.
C.
R. Hebd. Seances Acad. Sci., Ser. C 1979, 289, 417. (d) Aubry,
A.;Vitoux, B.;Marraud, M. J. Chim. Phys. Phys.-Chim. Biol.
1984, 81, 179. (e) Takeuchi, Y.;Marshall, G. R. J. Am. Chem.
Soc. 1998, 120, 5363.