Small-molecule inhibitors of AF6 PDZ-mediated protein-protein interactions

Article abstract of DOI:10.1002/cmdc.201300553

PDZ (PSD-95, Dlg, ZO-1) domains are ubiquitous interaction modules that are involved in many cellular signal transduction pathways. Interference with PDZ-mediated protein-protein interactions has important implications in disease-related signaling process

Full text of DOI:10.1002/cmdc.201300553

CHEMMEDCHEM
FULL PAPERS
DOI: 10.1002/cmdc.201300553
Small-Molecule Inhibitors of AF6 PDZ-Mediated Protein–
Protein Interactions
Carolyn Vargas,[a, b] Gerald Radziwill,*[c] Gerd Krause,[a] Anne Diehl,[a] Sandro Keller,[a, b]
Nestor Kamdem,[a] Constantin Czekelius,[d] Annika Kreuchwig,[a] Peter Schmieder,[a]
Declan Doyle,[f] Karin Moelling,[e] Volker Hagen,[a] Markus Schade,*[g] and
Hartmut Oschkinat*[a]
PDZ (PSD-95, Dlg, ZO-1) domains are ubiquitous interaction
modules that are involved in many cellular signal transduction
pathways. Interference with PDZ-mediated protein–protein in-
teractions has important implications in disease-related signal-
ing processes. For this reason, PDZ domains have gained at-
tention as potential targets for inhibitor design and, in the
long run, drug development. Herein we report the develop-
ment of small molecules to probe the function of the PDZ
domain from human AF6 (ALL1-fused gene from chromo-
some 6), which is an essential component of cell–cell junctions.
These compounds bind to AF6 PDZ with substantially higher
affinity than the peptide (Ile-Gln-Ser-Val-Glu-Val) derived from
its natural ligand, EphB2. In intact cells, the compounds inhibit
the AF6–Bcr interaction and interfere with epidermal growth
factor (EGF)-dependent signaling.
Introduction
Most biological processes are governed by the transient forma-
tion of protein complexes that are established through weak
yet specific protein–protein interactions (PPIs).[1,2] Malfunctions
within these networks can bring about pathological conditions
and thus have important implications for the treatment of
human diseases. Such PPIs represent a particularly challenging
class of targets for the development of bioactive small mole-
cules.[3,4] One approach to their modulation is to target the re-
spective interaction-mediating protein domains.[5] Although
the task of finding and developing such modulators is formida-
ble,[4,6] there have been several examples of successful inhibi-
tion of protein–protein interactions by small molecules.[4,7] We
have undertaken a small-molecule approach toward under-
standing PPIs involving PDZ (postsynaptic density/discs large/
zona occludens-1) domains.[8] They consist of six b strands (bA–
bF) flanked by two a helices (aA and aB)[9] and expose a char-
acteristic, well-defined binding pocket that is an attractive
target for inhibitor design. For DVL and Pick1 PDZ domains,
small-molecule inhibitors have been identified, the binding af-
finities of which are similar to those of the endogenous pep-
tide ligands.[10,11] Herein we describe the development of inter-
action modulators for the disease-related cell-junction protein
AF6 (ALL-1 fused gene from chromosome 6, also known as
l-afadin). AF6 is also a scaffolding protein that connects mem-
brane-associated proteins to the actin cytoskeleton.[8,12] It
binds the protein kinase break-point-cluster region (Bcr) via its
PDZ domain and the GTPase Ras via its Ras binding domain,
which leads to down-regulation of Ras signaling at sites of
[a] Dr. C. Vargas, Dr. G. Krause, Dr. A. Diehl, Prof. Dr. S. Keller, Dr. N. Kamdem,
A. Kreuchwig, Dr. P. Schmieder, Dr. V. Hagen, Prof. Dr. H. Oschkinat
Leibniz Institute of Molecular Pharmacology (FMP)
Robert-Rçssle-Str. 10, 13125 Berlin (Germany)
Heinrich-Heine-Universitꢁt Dꢀsseldorf
Universitꢁtsstr. 1, 40225 Dꢀsseldorf (Germany)
[e] Prof. Dr. K. Moelling
Institute of Medical Virology
University of Zurich, Gloriastr. 30, 8006 Zurich (Switzerland)
E-mail: oschkinat@fmp-berlin.de
[b] Dr. C. Vargas, Prof. Dr. S. Keller
[f] Dr. D. Doyle
Current address: Molecular Biophysics, University of Kaiserslautern
Erwin-Schrçdinger-Str. 13, 67663 Kaiserslautern (Germany)
Structural Genomics Consortium
Oxford University, Botnar Research Center, Oxford, OX3 7LD (UK)
and
[c] Prof. Dr. G. Radziwill
Institute of Medical Virology, University of Zurich (Switzerland)
and
Current address: Trinity College Dublin
School of Biochemistry and Immunology, College Green, Dublin 2 (Ireland)
Current address: Faculty of Biology
[g] Dr. M. Schade
BIOSS Centre for Biological Signalling Studies, University of Freiburg
Schaenzlestr. 1, 79104 Freiburg (Germany)
E-mail: gerald.radziwill@biologie.uni-freiburg.de
Evotec AG, Berlin (Germany)
and
Current address: Grꢀnenthal GmbH, 52099 Aachen (Germany)
E-mail: Markus.Schade@grunenthal.com
[d] Prof. Dr. C. Czekelius
Institut fꢀr Chemie und Biochemie, Freie Universitꢁt Berlin (Germany)
and
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cmdc.201300553.
Current address: Instiut fꢀr Organische Chemie und Makromolekulare
Chemie
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 1458 – 1462 1458

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
cell–cell contact to maintain cells in a non-proliferative state.[13]
The PDZ domain (AF6 PDZ) mediates clustering of Eph recep-
tor tyrosine kinases[14] and binds c-Src, a non-receptor tyrosine
kinase.[15] Furthermore, AF6 PDZ interacts with Jagged-1[16] and
connexin 36.[17] Although a number of these proteins that in-
teract with the AF6 PDZ have been identified, the physiological
roles of these interactions have yet to be established.[18] Low-
molecular-weight regulators may help address this issue. We
previously reported the three-dimensional NMR structure of
compound 1 (5-(4-trifluoromethylbenzyl)-2-thioxo-4-thiazolidi-
none) in complex with AF6 PDZ.[19] The binding affinity of 1 for
AF6 PDZ (KD =100 mm) is similar to that of the peptide (Ile-Gln-
Ser-Val-Glu-Val, KD =137 mm)[20] derived from its natu-
whereas no binding was detected for both Dlg2 PDZ1 and
Shank3 PDZ (Dd= <0.02 ppm). Evidently, 1 binds to various
PDZ domains. Nonetheless, the differential binding exhibited
by 1 is an indication that selectivity may be achieved by ration-
al derivatization. To this end, we synthesized derivatives of
1 with the aid of molecular modeling based on the AF6 PDZ–
1 complex structure (PDB code 2EXG) and NMR chemical shift
data. Selectivity was introduced through substituents that rec-
ognize a unique “Ala80 pocket” in the binding groove of AF6
PDZ, absent in other PDZ domains (Supporting Information,
Figure S1). This was achieved by extending the lead scaffold at
the 6-position by substituents of various bulk (Table 1). As ex-
ral ligand, EphB2.[14] In our previous study, compound
Table 1. Lead scaffold 1 and its derivatives 2–10 with dissociation constants and se-
lectivity indexes (SI) for AF6 and a1-syntrophin PDZ domains.
1 was found to bind a1-syntrophin PDZ in a similar
manner as to AF6 PDZ, with approximately threefold
lower affinity.[19] Therefore, in this work, we aimed at
developing compounds that bind more avidly to AF6
PDZ but not to a1-syntrophin PDZ. For selectivity
testing, we initially performed a screen of 1 against
a limited set of PDZ domains (Dlg2 PDZ-2, Dlg3 PDZ-
2, MPDZ-3, MPDZ-5, MPDZ-11, MPDZ-12, and
Shank3). Then, guided by molecular modeling, we
designed and synthesized derivatives of our lead
scaffold 1 and estimated binding affinity by NMR to
ascertain ligand selectivity and established structure–
activity relationships on the basis of the solved com-
plex structure. We investigated the selectivity of the
tightest-binding compounds further by testing them
against a second set of PDZ domains, namely DVL-1,
PSD95 PDZ-1, -2 and -3, and Shank3 PDZ. We then
tested the effects of the compounds in cellular
assays.
KD [mm][a]
SI
R
Compd
AF6
a1-Syntrophin
H
1
100Æ12
270Æ21
3
2
2
256Æ90
481Æ128
3
274Æ47
>1000
>4
4
5
6
91Æ16
36Æ7
32Æ3
307Æ60
504Æ95
3
14
>1000
>31
7
8
9
30Æ6
27Æ5
25Æ5
773Æ137
400Æ152
26
15
Results and Discussion
>1000
>40
As a prerequisite for deriving extended derivatives by
molecular modeling, we screened lead scaffold 1 for
potential binding to a variety of selected PDZ do-
mains. Because the ligands are positioned largely in
the space occupied by the last two residues of pep-
tide ligands, PDZ domain classification is not relevant
for specificity testing (class I: S/T-X-F; class II: F-X-
F). In the absence of 15N-labelled protein, two-di-
mensional total correlated spectroscopy (TOCSY)
NMR experiments were used for this purpose, and
the AF6 PDZ chemical shift perturbations (CSP, Dd)
were taken as reference (Dd=0.06 ppm). Dlg2 PDZ2,
Dlg3 PDZ2, MPDZ-3, MPDZ-5, MPDZ-11, and MPDZ-
12 showed intermediate binding affinities (Dd=0.03,
0.02, 0.04, 0.06, 0.03, and 0.03 ppm, respectively),
10
10Æ2
71Æ16
7
10a
10b
no binding*
–
–
4.9Æ2*
[a] KD values were measured by 1H–15N HSQC NMR titration (Figure S2) using the rap-
idly interconverting racemate 1 and racemic mixtures of diastereomers 2–10; values
are reported as means ÆSD of at least six residues influenced by ligand binding.
:
*
Denotes the value obtained after enantiomeric separation of compound 10 into the
pairs 10a (5S,6S and 5R,6S) and 10b (5R,6R and 5S,6R). Mixtures 10a and 10b were
assigned on the basis of molecular modeling using the solution structure of 1 com-
plexed with AF6 PDZ (PDB code 2EXG) as template.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 1458 – 1462 1459

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
pected, the modification of 1 at
position 6 resulted in a number
of racemic mixtures of diastereo-
mers, 5–10. Separation of the
four stereoisomers of each dia-
stereomeric mixture proved diffi-
cult, because of rapid racemiza-
tion. However, this was per-
formed for one stereocenter of
10 at the state of precursors re-
sulting in the mixture (5R,6R)
and (5S,6R) or (5S,6S) and
(5R,6S), as discussed below. In
agreement with our initial hy-
pothesis, bulky substituents at
position 6 of compound 1 had
the most pronounced influence
on the binding affinity to AF6
PDZ. Short extensions such as
a methyl (2), N,N-dimethylcar-
boxymethyl (3), or carboxymeth-
yl (4) groups showed no im-
provement, whereas substitu-
ents such as ethoxycarbonyl-
methyl (5), isopropyloxycarbo-
Figure 1. Structural models of PDZ–ligand complexes: a) AF6 PDZ–(5R,6R)-10; b) a1-syntrophin PDZ–(5R,6S)-10;
and c) AF6 PDZ–(5R,6R)-9. d) Superposition of the surfaces of the AF6 PDZ–(5R,6R)-9 complex and the a1-syntro-
phin PDZ domain (violet surface, presence of bulkier Val68 instead of Ala80, black arrow). Surface representations
of AF6 PDZ are shown as hydrophobic color potential (yellow=hydrophobic, green= hydrophilic); ligands are de-
picted in orange. The indicated Val68 is of a1-syntrophin PDZ.
nylmethyl
(6),
morpholino-
carbonylmethyl (7), 2-furanylme-
thoxycarbonylmethyl (8), and
benzyloxycarbonylmethyl (9) improved binding. The binding
affinities of 4–9 clustered in the KD range of 25–36 mm, sug-
gesting that these molecules occupy a hydrophobic interaction
area of similar size. Additionally, the bulkier and longer mor-
pholinophenylcarbonylmethyl moiety of 10 shows the highest
binding affinity (KD =10 mm).
plates for modeling the PDZ–ligand complexes of all four ste-
reoisomers of 9 (which selectively binds AF6 PDZ over a1-syn-
trophin PDZ) and 10 (which has the highest affinity for AF6
PDZ). This yielded the (5R,6R) isomers of 9 and 10 as the best
binding partners (Figure 1a,c,d). Docking of isomers (5R,6R)
and (5S,6R) resulted in a larger number of productive interac-
tions than complexes involving isomers (5S,6S) and (5R,6S). For
the latter, no energetically relevant interactions could be found
in which the five-membered ring interacts appropriately with
the GLGF loop, the conserved sequence motif critical for PDZ
interactions. Therefore, we assigned 10b to consist of the
(5R,6R) and (5S,6R) isomers, whereas 10a consisted of (5S,6S)
and (5R,6S).
In contrast to AF6 PDZ, KD values for a1-syntrophin PDZ
ranged from 70 mm to >1 mm (Table 1). The preferential bind-
ing of the compounds to AF6 PDZ relative to a1-syntrophin
PDZ is given in the right column of Table 1 by a selectivity
index (ratio of KD (a1-syntrophin PDZ) over KD (AF6 PDZ)). Com-
pounds 6 and 9 bound 30- and 40-fold more tightly to AF6
PDZ than to a1-syntrophin PDZ, respectively. Although 10 dis-
criminates only by a factor of 7, it has the highest affinity for
AF6 PDZ, showing a 10-fold improvement over compound 1.
Considering that 10 is a racemic mixture of diastereomers, we
attempted to separate the stereoisomers. Enantiomeric separa-
tion of 10 was performed by chromatography on the level of
a protected precursor and resulted in 10a and 10b, each a dia-
stereomeric mixture of enantiomerically pure material with re-
spect to the benzylic position (Figure S3). No binding was de-
tected for 10a (Table 1, Figure S4a), whereas 10b induced
strong CSPs with KD =4.9 mm (Table 1, Figure S4b). Because
1 and its derivatives 5–10 induced the same strong CSPs to
residues located along the binding pocket (Met23, Leu25,
Ile27, Met83, Val90, and Leu92), they are all likely to bind in
a very similar way.[19] This allowed us to use the AF6 PDZ–
1 complex structure (PDB code 2EXG) and the ligand-bound
NMR structure of a1-syntrophin PDZ (PDB code 2PDZ) as tem-
In the models of the complex AF6 PDZ–10 (Figure 1a), the R
group of the 5R,6R isomer extends across the peptide binding
groove toward the region where aB starts and where the mor-
pholino ring makes contact with the Gln76 side chain. The
morpholino ring occupies a surface indentation primarily
formed by Ala80, which we designate as the Ala80 pocket. The
experimental CSPs of Gln76, Arg78, Ala80, and Leu82 con-
firmed binding of 10b to the Ala80 pocket. Similar arguments
explain the binding of 9 to AF6 PDZ (Figure 1c).
In contrast, modeling complexes involving ligands 10 and
a1-syntrophin PDZ (PDB code 2PDZ) yielded a satisfying result
only for the (5R,6S) isomer of 10. The a1-syntrophin PDZ–10
models (Figure 1b) show the R group of 10 oriented toward
bB, where the oxygen atom of the morpholino ring could form
hydrophilic contacts with the NH groups of the Lys32 side
chain. In this region, confirmatory CSPs were observed for
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 1458 – 1462 1460

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
Leu29, Lys32, and Phe34. This alternative Lys32 binding geom-
etry is very likely caused by the occlusion of the Ala80 pocket
in a1-syntrophin PDZ with the side chain of residue Val68. Fig-
ure 1d shows a superposition of the AF6 PDZ and a1-syntro-
phin surfaces including the (5R,6R) isomer of 9, suggesting
that the low affinity between a1-syntrophin PDZ and 9 may be
due to both the absence of the Ala80 pocket and the lack of
other attractive protein contacts. Intriguingly, the À2 position
of the natural peptide ligand partially targets the Ala80 pocket
and is considered a critical determinant of PDZ subclass selec-
tivity.[21]
To obtain further selectivity profiles for 9 and 10, we
screened them against 15N-labeled PDZ domains of PSD95
1
(PDZ-1, -2, and -3), DVL1 and Shank3, applying H–15N hetero-
nuclear single quantum correlation (HSQC) spectroscopy and
comparing the observed CSPs to the changes observed on
AF6 PDZ (Dd=0.28 and 0.31 ppm for 9 and 10, respectively).
Compound 9 did not bind PSD95 PDZ-1, -2, and -3, and
Shank3 PDZ (Dd<0.02 ppm), whereas very weak binding to
both DVL1 and a1-syntrophin PDZ (Dd=0.04 ppm) was ob-
served. Compound 10 did not bind Shank3 and PSD95 PDZ-
1 and -3 (Dd<0.02 ppm), but showed weak binding to DVL1
(Dd=0.07 ppm), PSD95 PDZ-2 (Dd=0.12 ppm), and a1-syntro-
phin PDZ (Dd=0.10 ppm). Taken together, we have shown ex-
amples of the preferential binding of 9 to AF6 PDZ within
a panel of PDZ domains investigated, and binding of 10b to
the AF6 PDZ domain with considerable affinity (KD =4.9 mm),
suggesting that thiazolidinone derivatives may be developed
as specific binders for PDZ domains.
Figure 2. Effect of compounds 1, 5–7, and 10 on the interaction of AF6 PDZ
with Bcr. a) Pull-down assay: Incubated GST–AF6 PDZ and GST bound to glu-
tathione sepharose with HEK293 lysate overexpressing Bcr pretreated with
1, 5–7, and 10 or DMSO (1 mm, 1 h) as indicated. Bound Bcr was detected
by immunoblotting with anti-Bcr antibody (top panel). Levels of GST pro-
teins and Bcr were controlled by immunoblotting (middle and bottom
panels). b) Co-immunoprecipitation: HEK293 cells transiently transfected
with full-length AF6–FLAG and Bcr were treated before cell lysis with 1, 5,
and 10 (1 mm, 1 h). Interactions were monitored by immunoprecipitation
with anti-FLAG antibody and immunoblotting with anti-Bcr antibody (left
panels); right panels show control of protein expression. All compounds
were soluble under the assay conditions.
Compounds 1 (the lead scaffold), 5–7 (KD ~30 mm), and 10
(highest-affinity AF6 PDZ binder in the series) were tested in
biological assays. Compounds 5, 6, and 7 were included in in-
dividual assays for comparison, to obtain data points with
weaker binding ligands. In pull-down assays, we observed that
1, 5–7, and 10 interfere with the binding of Bcr (one of the
AF6 PDZ interaction partners) to the isolated AF6 PDZ (Fig-
ure 2a). Follow-up co-immunoprecipitation assays conducted
on 1, 5, and 10 demonstrated that 1 and 5 indeed interfere
with the interaction between full-length proteins AF6 and Bcr
in HEK293 cells (Figure 2b), thus supporting the results of pull-
down assays. Surprisingly, 10 induced degradation of AF6 and
Bcr in cells (Figure 2b, right panel; Figure S5a), a case that re-
quires further elaborative studies.
ble that the negative influence of 1 and 6 on EGF-induced sig-
naling involves receptor-type protein tyrosine phosphatases
such as RPTPb/z and RPTPg.[22] Both carry a C-terminal se-
quence, Glu-Ser-Leu-Val, that binds to AF6 PDZ,[20] which in
effect may restrict their phosphatase activity. We hypothesize
that compounds competing for binding of RPTPs to AF6 PDZ
could release RPTPs, thereby inhibiting receptor tyrosine phos-
phorylation and ERK activation, as depicted in a model
(Figure 3).
Furthermore, we used epidermal growth factor (EGF)-stimu-
lated cells to analyze the effect of compounds on signaling,
i.e., EGFR phosphorylation and ERK (extracellular signal-regulat-
ed kinase) phosphorylation. Compound 1 was most effective
and decreased EGF-induced ERK1/2 phosphorylation at
100 mm, whereas 1 and 6 inhibited ERK1/2 phosphorylation at
300 mm and 1 mm, respectively (Figure S5a). Furthermore, 6
and 10 were shown to abrogate EGF-induced receptor tyrosine
phosphorylation (Figure S5b). Therefore, we postulate that
under these conditions the compounds act on the level of
EGFR phosphorylation, interfering with downstream phosphor-
ylation of ERK. Generally, phosphorylation of both EGFR and
ERK correlates with increased cell proliferation. Accordingly,
both 1 and 6 decreased cell proliferation (Figure S6). It is possi-
Compound 10 was also shown to inhibit ERK1/2 and EGF re-
ceptor phosphorylation (Figures S5a and S5b). Additionally, 10
induces degradation of both AF6 and Bcr (Figure 2b and Fig-
ure S5a). Compound-induced degradation was observed in ex-
periments with intact cells, and it was not detectable in pull-
down assays (Figure 2a). Additionally, in contrast to 1 and 6,
10 slightly elevated the HEK293 cell count, presumably
through disruption of cell–cell contacts and overgrowth of
cells, as previously observed after down-regulation of AF6.[23]
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 1458 – 1462 1461

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
Drewes, M. Raida, T. Bouwmeester,
P. Bork, B. Seraphin, B. Kuster, G.
Neubauer, G. Superti-Furga, Nature
2002, 415, 141–147.
[2] P. Bork, L. J. Jensen, C. Mering, A. K.
Ramani, I. Lee, E. M. Marcotte, Curr.
Opin. Struct. Biol. 2004, 14, 292–
299.
[3] A. J. Wells, C. L. McClendon, Nature
2007, 450, 1001–1009.
[4] M. R. Arkin, M. Randal, W. L.
Delano, J. Hyde, T. N. Luong, J. D.
Oslob, D. R. Raphael, L. Taylor, J.
Wang, R. S. McDowell, J. A. Walls,
A. C. Braisted, Proc. Natl. Acad. Sci.
USA 2003, 100, 1603–1608.
Figure 3. Model depicting how AF6 PDZ and chemical interference may act on signaling. a) In non-stimulated
cells, binding of Bcr to AF6 PDZ down-regulates Ras signaling.[13] In addition, RPTP inhibits activation of RTK-de-
pendent ERK activation. b) Upon stimulation of cells, phosphorylated Bcr is released from the AF6 PDZ domain
that now sequesters RPTP, resulting in activation of ERK. c) Inhibitory compounds (grey) interfere with the RPTP–
AF6 PDZ interaction and allow RPTP-dependent inhibition of RTK/ERK signaling.
[5] K. K. Dev, Nat. Rev. Drug Discovery
2004, 3, 1047–1056.
[6] P. M. Fischer, D. P. Lane, Trends
Pharmacol. Sci. 2004, 25, 343–346.
[7] a) L. Pagliaro, J. Felding, K. Audouze, S. J. Nielsen, R. B. Terry, C. Krog-
Jensen, S. Butcher, Curr. Opin. Chem. Biol. 2004, 8, 442–449; b) A. G. Co-
chran, Curr. Opin. Chem. Biol. 2001, 5, 654–659; c) H. Yin, A. D. Hamil-
ton, Angew. Chem. Int. Ed. 2005, 44, 4130–4163; Angew. Chem. 2005,
117, 4200–4235.
Therefore, AF6 degradation may overcompensate for ERK path-
way inhibition, resulting in enhanced proliferation in the pres-
ence of 10 (Figure S6).
[8] a) R. Prasad, Y. Gu, H. Alder, T. Nakamura, O. Canaani, H. Saito, K. Hueb-
ner, R. P. Gale, P. C. Nowell, K. Kuriyama, Y. Miyazaki, C. M. Croce, E. Can-
aani, Cancer Res. 1993, 53, 5624–5628; b) A. B. Zhadanov, D. W. J. Prov-
ance, C. A. Speer, J. D. Coffin, D. Goss, J. A. Blixt, C. M. Reichert, J. A.
Mercer, Curr. Biol. 1999, 9, 880–888; c) T. Yamamoto, N. Harada, K.
Kano, S. Taya, E. Canaani, Y. Matsuura, A. Mizoguchi, C. Ide, K. Kaibuchi,
Y. Miyazaki, C. M. Croce, E. Canaani, J. Cell Biol. 1997, 144, 361–371.
[9] a) B. Z. Harris, W. A. Lim, J. Cell Sci. 2001, 114, 3219–3231; b) M. Sheng,
C. Sala, Annu. Rev. Neurosci. 2001, 24, 1–29.
[10] H. Lee, N. X. Wang, D. Shi, J. J. Zheng, Angew. Chem. Int. Ed. 2009, 48,
6448–6452; Angew. Chem. 2009, 121, 6570–6574.
[11] T. S. Thorsen, K. L. Madsen, N. Rebola, M. Rathje, V. Anggono, A. Bach,
I. S. Moreira, N. Stuhr-Hansen, T. Dyhring, D. Peters, T. Beuming, R. Huga-
nir, H. Weinstein, C. Mulle, K. Stromgaard, L. C. B. Ronn, U. Gether, Proc.
Natl. Acad. Sci. USA 2010, 107, 413–418.
Conclusions
In summary, we have demonstrated inhibition of the Bcr–AF6
PDZ interaction, EGF-dependent signaling, and cell prolifera-
tion by our compounds. These findings show that small-mole-
cule binders can be developed to differentially modulate the
activity of individual PDZ domains. In spite of the rather
modest affinities of the improved compounds, they have pro-
vided us with a better assessment of AF6 PDZ-mediated PPIs,
thereby opening up new avenues for the development of tail-
ored chemical probes and novel chemotherapeutics.
[12] A. B. Zhadanov, D. W. J. Provance, C. A. Speer, J. D. Coffin, D. Goss, J. A.
Blixt, C. M. Reichert, J. A. Mercer, Curr. Biol. 1999, 9, 880–888.
[13] G. Radziwill, R. A. Erdmann, U. Margelisch, K. Moelling, Mol. Cell. Biol.
2003, 23, 4663–4672.
[14] B. Hock, B. Bohme, T. Karn, T. Yamamoto, K. Kaibuchi, U. Holtrich, S. Hol-
land, T. Pawson, H. Rubsamen-Waigmann, K. Strebhardt, Proc. Natl.
Acad. Sci. USA 1998, 95, 9779–9784.
Experimental Section
Chemical synthesis, NMR protein sample preparation, quantifica-
tion of ligand binding, molecular modeling, and cell-based assays
are described in the Supporting Information.
[15] G. Radziwill, A. Weiss, J. Heinrich, M. Baumgartner, P. Boisguerin, K.
Owada, EMBO J. 2007, 26, 2633–2644.
[16] M. Popovic, J. Bella, V. Zlatev, V. Hodnik, G. Anderluh, P. N. Barlow, A.
Pintar, S. Pongor, J. Mol. Recognit. 2011, 24, 245–253.
Acknowledgements
This research was supported by the Deutsche Forschungsgemein-
schaft (DFG) Research Group 806 and EXC-294 BIOSS (G.R.). We
thank M. Leidert and S. Radetzki for cloning and protein prepara-
tion, Dr. V. Martos for help in some HPLC runs, and B. Schlegel
and J. Fritz for technical assistance. We thank Prof. M. Stubbs
(Martin Luther University Halle-Wittenberg) and Dr. D. Geissler
(FMP Berlin) for constructive discussions.
[17] X. Li, B. D. Lynn, J. I. Nagy, Eur. J. Neurosci. 2012, 35, 166–181.
[18] a) G. Udugamasooriya, D. Saro, M. R. Spaller, Org. Lett. 2005, 7, 1203–
1206; b) M. Paduch, M. Biernat, P. Stefanowicz, Z. S. Derewenda, Z.
Szewczuk, J. Otlewski, ChemBioChem 2007, 8, 443–452.
[19] M. Joshi, C. Vargas, P. Boisguerin, A. Diehl, G. Krause, P. Schmieder, K.
Moelling, V. Hagen, M. Schade, H. Oschkinat, Angew. Chem. Int. Ed.
2006, 45, 3790–3795; Angew. Chem. 2006, 118, 3874–3879.
[20] U. Wiedemann, P. Boisguerin, R. Leben, D. Leitner, G. Krause, K. Moel-
ling, R. Volkmer-Engert, H. Oschkinat, J. Mol. Biol. 2004, 343, 703–718.
[21] Q. Chen, X. Niu, Y. Xu, J. Wu, Y. Shi, Protein Sci. 2007, 16, 1053–1062.
[22] M. Marzec, M. Kasprzycka, X. Liu, P. N. Raghunath, P. Wlodarski, M. A.
Wasik, Oncogene 2007, 26, 813–821.
Keywords: chemical shift perturbation · chemical synthesis ·
molecular modeling · NMR spectroscopy · signal transduction
[23] M. Lorger, K. Moelling, J. Cell Sci. 2006, 119, 3385–3398.
[1] A. C. Gavin, M. Bosche, R. Krause, P. Grandi, M. Marzioch, A. Bauer, J.
Schultz, J. M. Rick, A. M. Michon, C. M. Cruciat, M. Remor, C. Hofert, M.
Schelder, M. Brajenovic, H. Ruffner, A. Merino, K. Klein, M. Hudak, D.
Dickson, T. Rudi, V. Gnau, A. Bauch, S. Bastuck, B. Huhse, C. Leutwein,
M. A. Heurtier, R. R. Copley, A. Edelmann, E. Querfurth, V. Rybin, G.
Received: December 23, 2013
Revised: February 13, 2014
Published online on March 25, 2014
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 1458 – 1462 1462