Identification of Novel Fragments Binding to the PDZ1-2 Domain of PSD-95

Article abstract of DOI:10.1002/cmdc.202000865

Inhibition of PSD-95 has emerged as a promising strategy for the treatment of ischemic stroke, as shown with peptide-based compounds that target the PDZ domains of PSD-95. In contrast, developing potent and drug-like small molecules against the PSD-95 PDZ domains has so far been unsuccessful. Here, we explore the druggability of the PSD-95 PDZ1-2 domain and use fragment screening to investigate if this protein is prone to binding small molecules. We screened 2500 fragments by fluorescence polarization (FP) and validated the hits by surface plasmon resonance (SPR), including an inhibition counter-test, and found four promising fragments. Three ligand efficient fragments were shown by ¹H,¹⁵N HSQC NMR to bind in the small hydrophobic P⁰ pockets of PDZ1-2, and one of them underwent structure-activity relationship (SAR) studies. Overall, we demonstrate that fragment screening can successfully be applied to PDZ1-2 of PSD-95 and disclose novel fragments that can serve as starting points for optimization towards small-molecule PDZ domain inhibitors.

Full text of DOI:10.1002/cmdc.202000865

Accepted Article
Title: Identification of novel fragments binding to the PDZ1-2 domain of
PSD-95
Authors: Jie Zang, Fei Ye, Sara Solbak, Lars Jakobsen Høj, Mingjie
Zhang, and Anders Bach
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: ChemMedChem 10.1002/cmdc.202000865
Link to VoR: https://doi.org/10.1002/cmdc.202000865
01/2020

10.1002/cmdc.202000865
ChemMedChem
COMMUNICATION
Identification of novel fragments binding to the PDZ1-2 domain of
PSD-95
Jie Zang,[a] Fei Ye,[b] Sara M. Ø. Solbak,[a] Lars J. Høj,[a] Mingjie Zhang,[b] and Anders Bach*[a]
[a]
Dr. J. Zang, Dr. S. M. Ø. Solbak, Dr. L. J. Høj, Assoc. Prof. A. Bach
Department of Drug Design and Pharmacology
Faculty of Health and Medical Sciences
University of Copenhagen
Universitetsparken 2, 2100 Copenhagen (Denmark)
E-mail: anders.bach@sund.ku.dk
[b]
Dr. F. Ye, Prof. M. Zhang
Division of Life Science
Hong Kong University of Science and Technology
Clear Water Bay, Kowloon, Hong Kong (China)
Supporting information for this article is given via a link at the end of the document.
Abstract: Inhibition of PSD-95 has emerged as a promising strategy
for treatment of ischemic stroke as shown with peptide-based
compounds that target the PDZ domains of PSD-95. In contrast,
developing potent and drug-like small-molecules against the PSD-95
PDZ domains has so far been unsuccessful. Here, we explore the
druggability of the PSD-95 PDZ1-2 domain and use fragment
screening to investigate if this protein is prone to bind small-
molecules. We screened 2,500 fragments by fluorescence
polarization (FP) and validated the hits by surface plasmon
resonance (SPR) including an inhibition counter-test and found four
promising fragments. Three ligand efficient fragments were shown
Small-molecule inhibitors of PDZ domains could be
advantageous over peptide-based compounds as such
molecules are generally more permeable to membranes and
biological barriers and less prone to cause immunologic
reactions, biological degradation, and kidney excretion. However,
development of small-molecule PDZ domain inhibitors has
proven to be very difficult.[8] Except for Biogen’s tripeptide-like
nanomolar PICK1 PDZ domain inhibitors,[9] potent (<1 µM)
small-molecules are lacking for the ~260 human PDZ domains,
including those of PSD-95.[10] A better understanding of PDZ
domains and their ligand-binding capabilities could open up for
new chemical probes and drug candidates towards a plethora of
biological relevant proteins.
1
by H−15N HSQC NMR to bind in the small hydrophobic P0 pockets
of PDZ1-2, and one of them underwent structure-activity relationship
(SAR) studies. Overall, we demonstrate that fragment screening
can successfully be applied to PDZ1-2 of PSD-95 and disclose novel
fragments that can serve as starting points for optimization towards
small-molecule PDZ domain inhibitors.
Previous
assessments
using
computational
and
experimental methods deemed PDZ domains non-druggable or
non-amenable to small-molecule binding.[11] Noticeably, 11,759
fragments were screened against PSD-95 PDZ1 by 1H-15N
HSQC NMR without finding a single hit.[11b] This result was
attributed to the structure of the PDZ domain pocket, which is
elongated and shallow in shape.[11] These features correlate with
the natural role of PDZ domains, which is to interact with the C-
terminal peptide parts of their protein interacting partner often
with micromolar affinities,[12] but makes small-molecule drug
discovery challenging.
Here, we examined the druggability of the PDZ1-2 domain of
PSD-95 by computational methods and experimental fragment
screening using fluorescence polarization (FP) as screening
assay, and surface plasmon resonance (SPR) and 1H-15N HSQC
NMR as sensitive methods for validation and characterization of
the fragment-protein interactions.
First, an FTMap analysis of PDZ1-2 was conducted. This
computational method identifies likely hot spots, i.e. regions of
the protein that contribute the most to ligand binding affinity, by
sampling a series of small organic probe molecules across the
protein and calculating their binding energy. Regions that bind
several kinds of probe molecules (consensus clusters, CC)
indicate hot spot areas.[13] FTMap identified two main clusters in
PDZ1 and PDZ2 of PDZ1-2, respectively, which correlate with
the hydrophobic P0 pockets that are normally occupied by the
extreme C-terminal hydrophobic amino acid (often Val or Ile) of
the peptide ligand (Figure 1).[14] However, the two main clusters
(CC1 and CC2) are rather small and very far apart, and the
The neuronal scaffolding protein postsynaptic density protein-95
(PSD-95) forms protein-protein interactions (PPIs) with the N-
methyl-D-aspartate (NMDA) receptor and neuronal nitric oxide
synthase (nNOS) through its PSD-95/Discs-large/ZO-1 (PDZ)
domains. Thereby, glutamate-mediated NMDA receptor
activation and the resulting Ca2+ influx are linked to the
production of nitric oxide (NO) by nNOS.[1] During pathological
over-activation of NMDA receptors excessive and harmful levels
of Ca2+ and NO contribute to excitoxicity and cellular damage.[2]
The 20-mer peptide Tat-NR2B9c binds the PDZ1 and PDZ2
domains of PSD-95 and is thereby able to inhibit the
nNOS/PSD-95/NMDA receptor complex and uncouple NMDA
receptor activation from downstream toxic events such as NO
production.[2] Tat-NR2B9c has shown neuroprotective effects in
several animal stroke models and even in recent clinical stroke
trials.[3] Also, high-affinity dimeric peptide-based inhibitors that
bind the PDZ1-2 domain of PSD-95 are neuroprotective in
mouse and rat stroke models,[4] and other peptide-based PSD-
95 inhibitors have shown effects in animal models of pain[5] and
neurodegenerative diseases.[6] Since PSD-95 inhibitors do not
affect the ion transportation through the NMDA receptor they are
believed to be safer drug candidates compared to NMDA
receptor antagonists.[2, 7]
1
This article is protected by copyright. All rights reserved.

10.1002/cmdc.202000865
ChemMedChem
COMMUNICATION
Figure 1. FTMap results showing two consensus clusters (CC1–2) in PSD-95 PDZ1-2. Consensus cluster strength (S) is defined as the number of probe clusters
within the consensus cluster. CC1 is found in the hydrophobic P0 pocket of PDZ2 (21 probe clusters; maximum dimension is 7.0 Å; magenta sticks) and CC2 is
likewise seen in P0 of PDZ1 (17 probe clusters; maximum dimension is 6.1 Å; yellow sticks). The distance between CC1 and CC2 is 38 Å. Four smaller consensus
clusters (containing 7–15 probe clusters) are seen on the other side (salmon, cyan, blue, and orange sticks) and are thus disconnected from CC1–2. Tripeptide
TAV (1) docked into PDZ2 is shown in green sticks to illustrate typical peptide-PDZ interactions (yellow dashes).
additional even smaller clusters are found on the other side of
the protein where ligands normally do not bind (Figure 1).
Overall, the absence of secondary hot spots close to CC1/2, the
large distance between the hot spots, and their small
dimensions makes PDZ1-2 a challenging target for small-
molecule drug discovery. However, a previous FTMap analysis
of PDZ1 suggested a secondary cluster around the conserved
His130 (His225 in PDZ2),[13a] but no experimental evidence for
such a potential small-molecule prone hot-spot has been
provided.
0.6% (Figure 2). Still, the inhibition effects were in most cases
small or sometimes showing a sharp dose-dependency.
We next screened our library of 2,500 commercial available
'rule of 3' compliant fragments[15] by our well-established FP
competition assay.[4a, 14] This assay measures the ability of the
fragments to inhibit the interaction between PDZ domain and a
fluorescent-labelled peptide probe derived from the NMDA
receptor GluN2B subunit. The assay was optimized for
sensitivity towards weak binders by exploring various
combinations of PDZ domains (PDZ2 or PDZ1-2 of PSD-95) and
Cy5-labelled peptide probes (either a monomeric 11-mer derived
from the C-terminus of the NMDA receptor GluN2B subunit,[14] or
a high-affinity dimeric probe[4a, 16]) and checked for DMSO
tolerability. Tripeptides TAV (1) and the corresponding TAV
peptide where the amino group was replaced with hydrogen
(2)[14] were used as weak-affinity inhibitors during optimization.
The combination of PDZ1-2 and the monomeric probe gave the
lowest IC50 values for 1 and 2 (220 and 2600 µM, respectively),
and DMSO from 1–8% did not affect the sensitivity of the assay
(Figure S1). The 2,500 fragments were then screened at 4 mM
(4% DMSO), and 63 fragments showed >10% inhibition and
were defined as primary hits (2.5% hit rate). These hits were
next validated in dose-response experiments (0.5–8 mM, 0.5–
8% DMSO) in the presence and absence of PDZ1-2 in order to
identify false-positives related to fluorescence inner-filter
effects.[17] Sixteen fragments showed dose-dependent inhibition
without a concomitant decrease in FP when PDZ1-2 was
omitted from the assay, corresponding to a validated hit rate of
Figure 2. Screening cascade and Venn diagram of our fragment screening
campaign against the PSD-95 PDZ1-2 domain. Venn diagram shows number
of hits after validation. Numbers in parenthesis indicate how many of the
FP/SPR validated hits that also showed dose-dependent and fragment-like
binding in the SPR dose-response test.
To further characterize the 63 primary FP hits and exclude
other types of false positives, such as aggregation-based
2
This article is protected by copyright. All rights reserved.

10.1002/cmdc.202000865
ChemMedChem
COMMUNICATION
promiscuous inhibitors,[18] we conducted
a
series of SPR
Next, binding of fragment hits 5–8 to PDZ1-2 was explored
by NMR chemical shift perturbation (CSP) analysis based on
1H−15N HSQC NMR titration experiments. NMR titration of
peptide controls 1 and 3 to PDZ1-2 showed obvious chemical
shifts indicating their interaction with PDZ1-2 (Figure S5A-B, E).
The slow exchange between the ligand-bound and ligand-free
PDZ1-2 in the titration is consistent with the high potency of
compound 3 (Figure S5B). Multiple peaks of PDZ1-2 underwent
fast exchange upon the binding of fragment 5, 8 and 9 (Figure
3A-C), indicating their interaction with PDZ1-2. Fragments 6 and
7 only gave negligible chemical shifts (Figure S5C-D). However,
6 was found unstable under the conditions of the NMR
experiments and 7 was hampered by low solubility preventing
testing at high excess of ligand. Mapping of binding-induced
chemical shift changes to the structure of PDZ1-2 suggested
that fragment 8 binds to PDZ1-2 by showing fast exchange of
chemical shifts specifically at the canonical GLGF loops and
hydrophobic P0 pockets of the βB–αB grooves of PDZ1-2 with
PDZ2 showing the strongest signals (Figure 3B, D). This
correlated with the suggested binding pose of 8 by molecular
docking (Figure 3E). Docking of 5–7 and 9 suggested that these
were also interacting with PDZ1-2 at these motifs (Figure S6).
validation experiments. SPR is a sensitive technique that can
detect direct binding to the protein and unusual sensorgram
shapes or response levels above those of comparative control
compounds are indicative of false positives.[17, 19] In SPR, the
fragments were injected in a OneStep concentration gradient at
0.5 mM and 1 mM (4% DMSO) over immobilized PSD-95 PDZ1-
2 in a 384-well plate format. An 8-mer GluN2B-derived peptide
(LSSIESDV-OH, 3[14]) and 2 were used as a positive controls
(Figure S2). To discriminate true binding events from non-
specific binding and bulk effects, a follow-up counter-test was
performed where the running and ligand buffers contained 10
μM of a high-affinity dimeric PDZ1-2 inhibitor (4).[20] The affinity
of 4 has previously been determined to be 10 nM by ITC (Kd)
and FP (Ki);[4a, 20] here a Kd of 21–73 nM was determined by SPR
(Figure S2). Thereby, the PDZ1-2 domain binding pocket is
blocked by 4 during the counter-test and fragment binding to the
site is hindered leading to reduced response.[19, 21] Out of the 63
primary FP hits, 28 were considered SPR validated as they gave
a MW normalized response level higher than 40% of the
response level of the weak control compound 2 in the primary
SPR experiment (2 and the hits showed response levels around
10 RU and 4–20 RU, respectively). Also, these fragments had
an ‘S‘ value ≥0.3, where S is a calculated metric based on the
responses from the primary test and the counter-test[21] (S = 0.3
corresponds to a reduction in response level of 50%). Among
the 28 SPR validated fragments, six were also FP validated hits
(Figure 2). The 38 FP/SPR validated hits were further
characterized by a SPR dose-response experiment by injecting
the fragments as two-fold serial dilutions (0.125–2 mM, 4%
DMSO). This allowed for a more careful evaluation of fragment-
like binding behaviour (e.g. fast kinetics and weak affinities) and
provided another measurement of the binding affinity to PDZ1-2.
Five out of the six fragments that were validated by both FP and
SPR showed dose-dependent and fragment-like binding, as did
15 out of the remaining 22 SPR validated hits (Figure 2). Four
fragments (5–8) showed affinities below 10 mM in the dose-
response SPR experiment, and these Kd values generally
correlated well with those suggested by OneStep SPR injections
(Table 1 and Figure S3–4). Interestingly, fragments 5, 6, and 8
were also among the pool of hits that were validated in both the
FP and SPR validation step (Figure 2). Fragment 7 was FP
validated, but, however, not defined as an SPR validated hit
(Figure 2) due to a low S value.
Table 1. SPR validation data of fragment hits[a]
SPR – OneStep
(Kd/mM)
SPR – MCI
(Kd/mM) (LE)
F#
Structure
5
6.2
2.3
7.1 (0.29)
7.7 (0.22)
6
7
2.8
7.3 (0.21)
8
9
3.2
4.2 (0.20)
6.3 (0.27)
The native peptide ligands of PSD-95 PDZ1-2 inserts the
side chain of the C-terminal amino acid (most often Val) into the
small hydrophobic P0 pocket while the C-terminal carboxylic acid
simultaneously makes hydrogen bonds with the backbone of the
conserved Gly-Leu-Gly-Phe (GLGF) motif (Figure 1).[22] Based
on the structures of 5–8, we imagined a similar binding mode of
these fragments. Thus, to compare the binding efficiencies of 5–
8 we designed the smallest possible truncation product of the
GluN2B peptide ligand, i.e. N-acetyl-L-Val (9), and tested it by
SPR. From this it was clear that the best fragment hits bind with
similar (5–7) or slightly improved (8) affinities compared to 9
(Table 1 and Figure S3–4). Notable, the ligand efficiency (LE)[23]
of 5 was better than for 9 (0.29 vs 0.27), indicating that 5 fits well
into the PDZ domain pocket. For fragment 6–8, LE was lower
than 9 (Table 1), but still at a reasonable level considering that
PDZ1-2 is a PPI target.
-
[a] SPR Kd values of fragment hits 5–8 and control fragment 9 as found by
OneStep (1 mM) or dose-response multi-cycle injections (MCI) (2-fold serial
dilutions) during the validation step (both n=1). LE values shown in
parenthesis are based on dose-response Kd values. Sensorgrams are shown
in Figure S3–4.
A series of 20 analogues of fragment 8 (10–29) were
purchased or synthesized (Scheme S1) and tested by SPR to
gain structural insight of the binding mode of 8 within the PDZ1-
2 pockets. From this it became clear that the cyclopentyl group
was important for binding as removing it (10), increasing or
decreasing the alkyl ring (11–12), or opening (13–15) or
aromatizing (16–17) it resulted in weaker compounds (Table 2).
In the suggested docking pose (Figure 3E), the cyclopentyl
3
This article is protected by copyright. All rights reserved.

10.1002/cmdc.202000865
ChemMedChem
COMMUNICATION
Figure 3. A-C) NMR-based titrations of PDZ1-2 with increasing concentrations of fragment hits 5 and 8 and control fragment 9. 1H,15N HSQC spectra of 15N-
labelled PDZ1-2 (150 µM) without or with the respective compounds are shown. D) Mapping of binding-induced chemical shift changes to PDZ1-2 by 8 illustrates
that 8 binds to the GLGF loop and P0 pocket of the conserved βB–αB PDZ binding groove in both PDZ domains. PDZ1 and PDZ2 of PDZ1-2 are presented
2
separately for clarity. The combined 1H and 15N chemical shift changes are defined as Δppm = [(ΔδHN
)
+ (ΔδN × αN)2]1/2, where ΔδHN and ΔδN represent the
chemical shift difference. The scaling factor αN to normalize the 1H and 15N chemical shifts is 0.17. E) Fragment hit 8 (green sticks) docked into PDZ2 of PSD-95.
The highest ranking pose is shown of (R)-8. Polar contacts are shown in yellow dashes, and a suggested π-cation interaction between 8 and K233 is shown in
orange dashes.
group of 8 binds into the P0 hydrophobic pocket, which is
normally occupied by a Val side chain from the native peptide
ligand (Figure 1). Apparently, for the small fragment 8, the
cyclopentyl group provides stronger binding than this branched
alkyl group. Esterification or amidation of the carboxylic acid of
8, as in 18–19, led to basically inactive compounds. This
correlates with the docking of 8 where the carboxylic acid group
forms essential hydrogen bonds to the GLGF motif similar to
peptide ligands.[22] Interestingly, the hydroxyl group of 8 also
engages in hydrogen bonding to the PDZ GLGF motif according
to docking (Figure 3E) - similar to the first amide NH group of the
peptide ligand (Figure 1). Accordingly, removing the hydroxyl
group led to a much weaker compound (20), and so did changes
of the central C–OH group into amides or sulfonamides as in
21–24 (Table 2). Finally, ortho or para-substitutions of the
phenyl ring of 8 with electron withdrawing or donating groups
resulted in 25–28 showing clear binding to PDZ1-2 only slightly
weaker than 8. Also, 29 was equipotent to 8 demonstrating that
thiophene could replace the phenyl ring (Table 2). In docking,
the phenyl ring points towards the αB' helix and forms a π-cation
interaction with Lys233 (Figure 3E). This part of the ligand and
corresponding protein area could be focus of future
optimizations.
facilitated the identification of the most promising hits, which
were also confirmed and further characterized by dose-response
SPR. Four fragments (5–8) bound PDZ1-2 with Kd values of 2–8
mM. Although these affinities are low, they are in the same
range as the native peptide-derived fragment N-acetyl-L-Val (9),
and the ligand efficiencies are reasonable to good and
comparable to 9. This indicates a good fit into the binding pocket,
and suggests they could provide starting points for further ligand
design. Also, fragments 5–8 are novel in structure in relation to
PDZ domain drug discovery, and can be docked to show binding
into the P0 pockets of the PSD-95 PDZ1-2 domain. The location
of the fragments into P0 was confirmed for 5, 8, and 9 by 1H−15N
HSQC NMR. Several derivatives of 8 were made, which gave
insight into the structural requirements for affinity, confirmed the
suggested docking pose, and provided equipotent analogues
thereby guiding future optimizations.
This work is one of the rare examples of successful fragment
screening studies related to a PDZ domain target[24] and to the
best of our knowledge the only one related to PSD-95. We show
that indeed it is possible to find ligand efficient fragments against
PDZ1-2 of PSD-95 by screening a commercial fragment library
using a well-optimized biochemical assay followed by thorough
counter-tests and biophysical assessments. The nature of the
target – especially the small size of the main hydrophobic pocket
and the lack of closely related secondary pockets – makes
In conclusion, we here report a screening campaign testing
2,500 fragments against the PDZ1-2 domain of PSD-95 using
FP and SPR as screening and main validation methods. This
included counter-tests where PDZ1-2 was omitted from the FP
assay to identify fluorescence artefacts, and an SPR step where
the target was blocked by a high-affinity ligand. This strategy
optimization
very
challenging.
However,
combining
peptidomimetic strategies with the insight and results obtained
from this study could perhaps facilitate the development of more
potent PSD-95 inhibitors with useful drug-like properties.
4
This article is protected by copyright. All rights reserved.

10.1002/cmdc.202000865
ChemMedChem
COMMUNICATION
Table 2. SPR activity of analogues of 8[a]
SPR
(Kd/mM)
SPR
(Kd/mM)
SPR
Structure
F#
Structure
F#
17
Structure
F#
24
(Kd/mM)
10
Very weak
Very weak
No binding
No binding
Very weak
Very weak
11
12
13
12
18
19
20
25
26
27
7.0
No binding
Very weak
9.6
8.4
14
15
Very weak
Very weak
21
22
No binding
No binding
28
29
8.6
6.1
16
Very weak
23
No binding
[a] Compounds were tested by dose-response multi-cycle injections (MCI) in 2-fold serial dilutions (125 µM – 2 mM) (n=1). A confirmatory OneStep injection (1
mM) experiment was also conducted (n=1, not shown). Kd values are determined from steady-state affinity analysis by fitting to a 1:1 model and by fixing Rmax to
the value of control peptide 3. Kd values were determined for compounds showing dose-dependent responses across all tested concentrations. Compounds
where only 1–2 of the highest concentrations led to responses >2 RU are designated 'Very weak'. And those only showing responses <2 RU are termed 'No
binding'.
[1]
[2]
[3]
R. Sattler, Z. Xiong, W. Y. Lu, M. Hafner, J. F. MacDonald, M.
Tymianski, Science 1999, 284, 1845-1848.
M. Aarts, Y. Liu, L. Liu, S. Besshoh, M. Arundine, J. W. Gurd, Y. T.
Wang, M. W. Salter, M. Tymianski, Science 2002, 298, 846-850.
a) D. J. Cook, L. Teves, M. Tymianski, Nature 2012, 483, 213-217; b)
M. D. Hill, R. H. Martin, D. Mikulis, J. H. Wong, F. L. Silver, K. G.
Terbrugge, G. Milot, W. M. Clark, R. L. Macdonald, M. E. Kelly, M.
Boulton, I. Fleetwood, C. McDougall, T. Gunnarsson, M. Chow, C. Lum,
R. Dodd, J. Poublanc, T. Krings, A. M. Demchuk, M. Goyal, R.
Anderson, J. Bishop, D. Garman, M. Tymianski, E. t. investigators,
Lancet Neurol. 2012, 11, 942-950; c) M. D. Hill, M. Goyal, B. K. Menon,
R. G. Nogueira, R. A. McTaggart, A. M. Demchuk, A. Y. Poppe, B. H.
Buck, T. S. Field, D. Dowlatshahi, B. A. van Adel, R. H. Swartz, R. A.
Shah, E. Sauvageau, C. Zerna, J. M. Ospel, M. Joshi, M. A. Almekhlafi,
K. J. Ryckborst, M. W. Lowerison, K. Heard, D. Garman, D. Haussen,
S. M. Cutting, S. B. Coutts, D. Roy, J. L. Rempel, A. C. Rohr, D. Iancu,
D. J. Sahlas, A. Y. X. Yu, T. G. Devlin, R. A. Hanel, V. Puetz, F. L.
Silver, B. C. V. Campbell, R. Chapot, J. Teitelbaum, J. L. Mandzia, T. J.
Kleinig, D. Turkel-Parrella, D. Heck, M. E. Kelly, A. Bharatha, O. Y.
Bang, A. Jadhav, R. Gupta, D. F. Frei, J. W. Tarpley, C. G. McDougall,
Acknowledgements
This research was supported by the Lundbeck Foundation (grant
R190-2014-3710 for A.B.); the A. P. Møller Foundation for the
Advancement of Medical Science (grant 14-28 for A.B.); the
Hørslev Foundation (grant 203866-MIA for A.B.); the Augustinus
Foundation (grant 14-1571 for A.B.), and a grant from RGC of
Hong Kong (AoE-M09-12 to M.Z). We also acknowledge access
to NMR measurements at the University of Copenhagen
(supported by grant #10-085264 from The Danish Research
Council for Independent Research | Nature and Universe and
grant R77-A6742 from the Lundbeck Foundation).
Keywords: Fragment screening • Drug discovery • PDZ
domains • Protein–protein interactions • PSD-95
5
This article is protected by copyright. All rights reserved.

10.1002/cmdc.202000865
ChemMedChem
COMMUNICATION
S. Holmin, J. H. Rha, A. S. Puri, M. C. Camden, G. Thomalla, H. Choe,
S. J. Phillips, J. L. Schindler, J. Thornton, S. Nagel, J. H. Heo, S. I.
Sohn, M. N. Psychogios, R. F. Budzik, S. Starkman, C. O. Martin, P. A.
Burns, S. Murphy, G. A. Lopez, J. English, M. Tymianski, ESCAPE-
NA1 Investigators, Lancet 2020, 395, 878-887.
[24] a) T. P. Kegelman, B. Wu, S. K. Das, S. Talukdar, J. M. Beckta, B. Hu,
L. Emdad, K. Valerie, D. Sarkar, F. B. Furnari, W. K. Cavenee, J. Wei,
A. Purves, S. K. De, M. Pellecchia, P. B. Fisher, Proc. Natl. Acad. Sci.
U. S. A. 2017, 114, 370-375; b) 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.
[4]
a) A. Bach, B. H. Clausen, M. Møller, B. Vestergaard, C. N. Chi, A.
Round, P. L. Sørensen, K. B. Nissen, J. S. Kastrup, M. Gajhede, P.
Jemth, A. S. Kristensen, P. Lundström, K. L. Lambertsen, K.
Strømgaard, Proc. Natl. Acad. Sci. U. S. A. 2012, 109, 3317-3322; b) A.
Bach, B. H. Clausen, L. K. Kristensen, M. G. Andersen, D. G. Ellman,
P. B. L. Hansen, H. Hasseldam, M. Heitz, D. Ozcelik, E. J. Tuck, M. V.
Kopanitsa, S. G. N. Grant, K. Lykke-Hartmann, F. F. Johansen, K. L.
Lambertsen, K. Strømgaard, Neuropharmacology 2019, 150, 100-111.
a) F. Tao, Q. Su, R. A. Johns, Mol Ther 2008, 16, 1776-1782; b) B. W.
LeBlanc, M. Iwata, A. P. Mallon, C. N. Rupasinghe, D. J. Goebel, J.
Marshall, M. R. Spaller, C. Y. Saab, Neuroscience 2010, 167, 490-500;
c) R. D'Mello, F. Marchand, S. Pezet, S. B. McMahon, A. H. Dickenson,
Mol. Ther. 2011, 19, 1780-1792.
[5]
[6]
[7]
[8]
L. M. Ittner, Y. D. Ke, F. Delerue, M. Bi, A. Gladbach, J. van Eersel, H.
Wolfing, B. C. Chieng, M. J. Christie, I. A. Napier, A. Eckert, M.
Staufenbiel, E. Hardeman, J. Gotz, Cell 2010, 142, 387-397.
F. X. Soriano, M. A. Martel, S. Papadia, A. Vaslin, P. Baxter, C.
Rickman, J. Forder, M. Tymianski, R. Duncan, M. Aarts, P. Clarke, D. J.
Wyllie, G. E. Hardingham, J. Neurosci. 2008, 28, 10696-10710.
a) C. N. Chi, A. Bach, K. Strømgaard, S. Gianni, P. Jemth, BioFactors
2012, 38, 338-348; b) N. R. Christensen, J. Calyseva, E. F. A.
Fernandes, S. Luchow, L. S. Clemmensen, L. M. Haugaard-Kedström,
K. Strømgaard, Adv. Ther. 2019, 2, 1800143.
[9]
a) E. Y. S. Lin, L. F. Silvian, D. J. Marcotte, C. C. Banos, F. Jow, T. R.
Chan, R. M. Arduini, F. Qian, D. P. Baker, C. Bergeron, C. A. Hession,
R. L. Huganir, C. F. Borenstein, I. Enyedy, J. Zou, E. Rohde, M.
Wittmann, G. Kumaravel, K. J. Rhodes, R. H. Scannevin, A. Dunah, K.
M. Guckian, Sci. Rep. 2018, 8; b) D. J. Marcotte, J. C. Hus, C. C.
Banos, C. Wildes, R. Arduini, C. Bergeron, C. A. Hession, D. P. Baker,
E. Lin, K. M. Guckian, A. W. Dunah, L. F. Silvian, Protein Sci. 2018, 27,
672-680.
[10] A. Bach, T. B. Pedersen, K. Strømgaard, MedChemComm 2016, 7,
531-536.
[11] a) D. C. Fry, L. T. Vassilev, J. Mol. Med. 2005, 83, 955-963; b) P. J.
Hajduk, J. R. Huth, S. W. Fesik, J. Med. Chem. 2005, 48, 2518-2525.
[12] M. A. Stiffler, J. R. Chen, V. P. Grantcharova, Y. Lei, D. Fuchs, J. E.
Allen, L. A. Zaslavskaia, G. MacBeath, Science 2007, 317, 364-369.
[13] a) D. Kozakov, D. R. Hall, R. L. Napoleon, C. Yueh, A. Whitty, S. Vajda,
J. Med. Chem. 2015, 58, 9063-9088; b) D. Kozakov, L. E. Grove, D. R.
Hall, T. Bohnuud, S. E. Mottarella, L. Luo, B. Xia, D. Beglov, S. Vajda,
Nat. Protoc. 2015, 10, 733-755.
[14] A. Bach, C. N. Chi, T. B. Olsen, S. W. Pedersen, M. U. Røder, G. F.
Pang, R. P. Clausen, P. Jemth, K. Strømgaard, J. Med. Chem. 2008,
51, 6450-6459.
[15] M. Congreve, R. Carr, C. Murray, H. Jhoti, Drug Discov. Today 2003,
8, 876-877.
[16] K. B. Nissen, L. M. Haugaard-Kedstrom, T. S. Wilbek, L. S. Nielsen, E.
Aberg, A. S. Kristensen, A. Bach, P. Jemth, K. Strømgaard, PloS One
2015, 10, e0117668.
[17] K. T. Tran, J. S. Pallesen, S. M. Ø. Solbak, D. Narayanan, A. Baig, J.
Zang, A. Aguayo-Orozco, R. M. C. Carmona, A. D. Garcia, A. Bach, J.
Med. Chem. 2019, 62, 8028-8052.
[18] S. L. McGovern, E. Caselli, N. Grigorieff, B. K. Shoichet, J. Med. Chem.
2002, 45, 1712-1722.
[19] S. M. Ø. Solbak, J. Zang, D. Narayanan, L. J. Høj, S. Bucciarelli, C.
Softley, S. Meier, A. E. Langkilde, C. H. Gotfredsen, M. Sattler, A.
Bach, J. Med. Chem. 2020, 63, 1156-1177.
[20] A. Bach, C. N. Chi, G. F. Pang, L. Olsen, A. S. Kristensen, P. Jemth, K.
Strømgaard, Angew. Chem. Int. Ed. 2009, 48, 9685-9689.
[21] A. M. Giannetti, H. N. Gilbert, D. P. Huddler, M. Reiter, C. Strande, K.
E. Pitts, B. J. Bravo in Fragment-Based Drug Discovery (Royal Society
of Chemistry), 2015, pp. 19-48.
[22] S. W. Pedersen, S. B. Pedersen, L. Anker, G. Hultqvist, A. S.
Kristensen, P. Jemth, K. Strømgaard, Nat. Commun. 2014, 5, 3215.
[23] A. L. Hopkins, C. R. Groom, A. Alex, Drug Discov. Today 2004, 9, 430-
431.
6
This article is protected by copyright. All rights reserved.

10.1002/cmdc.202000865
ChemMedChem
COMMUNICATION
Entry for the Table of Contents
PDZ domains are intriguing but challenging drug targets. Here, we investigate the druggability of PDZ1-2 of PSD-95 by a
computational method and fragment-based screening using biochemical and biophysical assays. This results in new well validated
fragments shown to bind the PDZ domains of PSD-95.
Institute and/or researcher Twitter usernames: @group_bach; @UCPH_health
7
This article is protected by copyright. All rights reserved.