Caged mono- and divalent ligands for light-assisted disruption of PDZ domain-mediated interactions

Article abstract of DOI:10.1021/ja309870q

We report a general method for light-assisted control of interactions of PDZ domain binding motifs with their cognate domains by the incorporation of a photolabile caging group onto the essential C-terminal carboxylate binding determinant of the motif. The strategy was implemented and validated for both simple monovalent and biomimetic divalent ligands, which have recently been established as powerful tools for acute perturbation of native PDZ domain-dependent interactions in live cells.

Full text of DOI:10.1021/ja309870q

Communication
pubs.acs.org/JACS
Caged Mono- and Divalent Ligands for Light-Assisted Disruption of
PDZ Domain-Mediated Interactions
Matthieu Sainlos,†,‡,§ Wendy S. Iskenderian-Epps,† Nelson B. Olivier,† Daniel Choquet,‡,§
and Barbara Imperiali*,†
†Departments of Chemistry and Biology, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, United States
‡University of Bordeaux, IINS, CNRS, UMR 5297, F-33000 Bordeaux, France
§CNRS, IINS, UMR 5297, F-33000 Bordeaux, France
S
* Supporting Information
(AMPAR12a and NMDAR12b) to macromolecular complexes in
ABSTRACT: We report a general method for light-
assisted control of interactions of PDZ domain binding
motifs with their cognate domains by the incorporation of
a photolabile caging group onto the essential C-terminal
carboxylate binding determinant of the motif. The strategy
was implemented and validated for both simple mono-
valent and biomimetic divalent ligands, which have
recently been established as powerful tools for acute
perturbation of native PDZ domain-dependent interac-
tions in live cells.
the postsynaptic density in live cells. While these ligands have
provided valuable insights into PDMIs, the task of deciphering
the various modes of regulation and their functional implications
would greatly benefit from new tools that allow spatiotemporal
control over these binding events.
Herein we report the design, synthesis, and characterization of
caged PDZ domain ligands (PDLs) for the photochemical
control and study of PDMIs. Indeed, this strategy should provide
excellent spatiotemporal resolution of the acute disruption of
endogenous PDZ domain binding events by controlling when
and where a ligand can bind to its cognate protein partner. Our
goals were both to develop a general caging method for PDLs
and to implement the approach with recently reported
biomimetic ligands.12 To establish the general strategy and
assess the caging efficiency, we exploited solvatochromic
fluorophores for direct monitoring of the interaction of
monovalent and bivalent peptide-based ligands and their cognate
PDZ domains.13,14
As a first step in the design process, we obtained structural
evidence that the fluorogenic probes described previously14
constitute canonical PDLs, thereby providing a paradigm for this
class of protein interactions. Crystallographic analysis of a PDZ
domain bound to a prototypical fluorescent ligand established
that the binding mode is comparable to that of other previously
reported ligands,8,15 thus excluding any perturbing role of the
fluorophore and validating the use of the probes as representative
PDLs (Figure 1 and Figure S1 in the Supporting Information).
To design efficient caged ligands, we focused on conserved
residues within canonical PDZ domain binding motifs to ensure
both disruption of the critical interactions and broad applicability
of this general strategy to different domain classes. Structural
analyses8,15 have consistently revealed that canonical binding of
PDLs involves their C-terminal sequence, typically via the
terminal amino acid hydrophobic side chain and carboxylate as
well as the residue at the −2 position (Figures 1 and S2). The
nature of the residue at the −2 position changes among various
PDZ domains and is used to distinguish specific domains into
classes.16 Therefore, among these conserved interactions, the C-
terminal carboxylate is the ideal target for caging because of its
presence across all PDZ domain classes, essentiality for binding,
nderstanding the complex dynamics of protein networks at
Uthe molecular level requires tools that allow for temporal
and spatial control of specific protein−protein interactions
(PPIs). The use of light as an effective switch to turn biological
events on or off in live cells is a powerful approach because of its
noninvasive nature and potential for high resolution. For
instance, photolabile caging groups provide an efficient method
for masking critical functionality in small bioactive molecules and
protein-binding partners, preventing a functional interaction
until the group is liberated through photoactivation.1−3 This
approach has been successfully applied to masking of neuro-
transmitters, nucleic acids, phosphopeptides, and proteins2−5
and used to modulate specific PPIs by exploiting caged variants of
molecules that compete with native interactions.5,6 In this
context, we were interested in developing a chemical caging
approach for interactions mediated by PDZ (PSD-95/DLG/ZO-
1) domains, which constitute an important and abundant PPI
system in mammals.7 Small globular PDZ domains are found in
multidomain repeats or associated with other protein interaction
domains and occur principally at cell junctions, where they are
key players in the assembly and localization of macromolecular
complexes involved in signal transduction pathways.8−10 These
types of interactions play a pivotal role in governing the dynamic
localization of ion channels and receptors, particularly when they
occur at synaptic junctions.9 Therefore, they have been the focus
of numerous studies involving the application of peptide-based
ligands to acutely disrupt native interactions and define their
functional roles in transient macromolecular complexes.11,12 We
recently introduced a series of synthetic biomimetic ligands that
efficiently disrupt the synaptic PDZ domain-mediated inter-
actions (PDMIs) involved in anchoring glutamate receptors
Received: October 5, 2012
Published: March 8, 2013
© 2013 American Chemical Society
4580
dx.doi.org/10.1021/ja309870q | J. Am. Chem. Soc. 2013, 135, 4580−4583

Journal of the American Chemical Society
Communication
Figure 1. Caged PDL design and synthesis. (a) Structure of fluorescent probe 1 [stick representation in blue color theme, with the non-natural amino
acid Dab(4-DMAP) in orange] bound to the third PDZ domain of SAP102 in green (PDB entry 3JXT). The hydrogen bonds between the ligand and the
domain are highlighted in yellow. (b) Conserved PDZ domain−ligand interactions. In addition to less specific interactions of the PDL backbone, amino
acids at position 0 (circled in black) and −2 (circled in blue) strongly interact with the domain surface residues in a conserved manner. (c) General
caging strategy. Incorporation of a caging group such as NPE on the C-terminal carboxylate of a PDL should strongly impair its capacity to bind partner
PDZ domains, which can then be recovered, when desired, by photorelease. Application of this strategy to PDZ domain fluorescent reporters allows
direct monitoring of the interaction. (d) General synthesis of caged PDLs.
and intrinsic chemical reactivity. The introduction of a
photolabile group at this position on a PDL should abolish key
electrostatic interactions involved in binding to the domains and
introduce steric hindrance, significantly impairing the inter-
actions with PDZ domains (Figure 1).
Table 1. Fluorescent Peptide-Based PDLs
The method used to obtain ligands caged at the C-terminal
carboxylate is based on the introduction of a photolabile group
after solid-phase peptide synthesis (SPPS). This strategy relies
on activation of the caging group as a diazoalkane rather than the
carboxylate as an activated ester to minimize potential
epimerization of the terminal amino acid. The diazoalkane
approach has been reported for caging of phosphate deriva-
tives17a and amino acids17b and could be modified for use with
peptide-based ligands. We chose the 1-(2-nitrophenyl)ethyl
(NPE) caging group for its compatibility with biological
applications2 and facile chemical activation for reaction with
the target carboxylates.17 Peptides were first synthesized by
standard Fmoc-based SPPS on a highly acid-sensitive resin (e.g.,
4-carboxytrityl linker) and derivatized with the 4-dimethylami-
nophthalimide (4-DMAP) fluorophore18 as reported previ-
ously.19 The resulting peptides were cleaved from the resin with
0.5% trifluoroacetic acid (TFA), thus preserving the side-chain
protecting groups to allow selective coupling of the free C-
terminal carboxylate with a diazo-activated NPE group (Figures
1d and S3). The caging strategy was applied to the recently
reported probes derived from the C-terminal sequences of
Stargazin, GluN2A, and GluA1 that target representative class-I
PDZ domains from synaptic PSD-95-like proteins (PSD-95,
PSD-93, SAP97, and SAP102) and Shank3 (Table 1).14
a
φ = Dab(4-DMAP), where Dab = 2,4-diaminobutyric acid; λ =
norleucine; Ac = acetyl group. The critical residues at positions 0 and
b
−2 (blue) and the NPE caging group (red) are highlighted. n.a.: not
applicable. PLP-3 and PLP-12 stand for the third and the first two
PDZ domains of PSD-95-like proteins, respectively.
Figure 2. Generation of uncaged ligand by photolysis. Relative
fluorescence increases (RFIs) for the caged probes in the presence of
cognate PDZ domain(s) with increasing durations of 365 nm
irradiation. RFIs were obtained by comparison with levels of
fluorescence obtained with the same amount of the corresponding
noncaged probes. The last entry in each graph (300 s*) shows the RFI
obtained in the absence of PDZ domain(s) after irradiation for 300 s.
The photolysis of the C-terminal caged peptides was first
studied via HPLC by submitting a solution of caged ligand 3 to
various durations of exposure to 365 nm UV light (Figure S4).
Under these conditions, the amount of uncaged free acid 2
increased incrementally from 0 to >80% over 300 s while the
amount caged of peptide decreased to 5%.
In parallel, the capacity of the caged and uncaged ligands to
bind cognate PDZ domains was evaluated by monitoring the
relative fluorescence increases in comparison with changes
observed with noncaged probes (Figures 2 and S5). In the
absence of protein, the caged and noncaged ligands showed
similar baseline fluorescence levels, indicating that the probe
fluorescence was not affected by the presence of the NPE group
in proximity to the reporter fluorophore (Figure S5). In the
presence of the cognate PDZ domain, the caging group clearly
resulted in significant loss of fluorescence, as the relative
fluorescence increases were generally <12% of those of the
noncaged ligand at t = 0 s (Figure 2), except for GST-PSD-95-12
(see Figure S5). Since we previously showed a correlation
between fluorescence increase and affinity within a given series of
ligands for the same PDZ domain, these observations reflect a
lower binding affinity of the caged ligand.14 Furthermore,
4581
dx.doi.org/10.1021/ja309870q | J. Am. Chem. Soc. 2013, 135, 4580−4583

Journal of the American Chemical Society
Communication
exposure of the caged ligands to various durations of UV
irradiation (t = 120 and 300 s) to release the caging group by
photolysis and reveal the free C-terminal carboxylate resulted in
the stepwise recovery of the fluorescence signal in the presence of
PDZ domains. At t = 300 s, all of the ligands yielded fluorescence
increases corresponding to more than ∼70% of the maximum
possible increase obtained with uncaged ligands, consistent with
the uncaging levels observed by HPLC (Figure S4). UV
irradiation at 365 nm did not result in significant photobleaching
or other changes in fluorescence, thereby validating the use of
fluorescence-based measurements to follow the modulation of
PDMIs.
perturbation of the interaction, with a decrease of <7% in the
fluorescence. In parallel, the same concentration of the noncaged
version 8 reduced the fluorescence by >50%. However, after UV
irradiation and uncaging of 9, most of the competing properties
of the peptide sequence were recovered, as judged by the
significant loss of fluorescence comparable to that obtained with
the noncaged ligand (40%). Application of the same treatment to
the probe−PDZ domain pair did reveal a small amount of
fluorophore photobleaching, albeit at levels that could not
account for the effect observed after photoactivation of the caged
competitor. This confirms the light-induced disruption of the
probe−PDZ domain interactions.
We recently showed that a new class of biomimetic ligands
comprising two covalently linked PDZ domain binding motifs
acutely and specifically disrupt the synaptic stabilization of
endogenous glutamate receptors in situations where monovalent
ligands proved ineffective.12 We were therefore motivated to
implement the carboxyl caging strategy to develop efficient
competitors of PDMIs in live cells. To take advantage of the
validated bivalent biomimetic design while also maintaining
optimal uncaging efficiency (i.e., requiring a single uncaging
event for full recovery of the binding properties, as opposed to
double uncaging if the divalent ligands were symmetrically
caged), the original synthetic scheme of the ligand12a required a
new strategy to allow for the introduction of a single caging group
per divalent ligand (Figure 4a). After on-resin ligation, the
Next, the validity of the caging strategy was established by
determining the relative binding affinities of the caged and
uncaged peptides by fluorescence titration with their respective
cognate domains (Table 2 and Figure S6). Binding saturation
Table 2. Dissociation Constants of Caged and Noncaged
a
Probes for Their Respective Cognate PDZ Domains
noncaged
D (μM)
caged
KD (μM)
b
b
PDZ domain(s)
ligand
K
ligand
GST-PSD-95-3
GST-SAP102-3
GST-PSD-95-12
GST-Shank3
2
2
4
6
1.66 0.32
5.45 0.59
1.97 0.26
3
3
5
7
37.07 4.95
148.4 82.8
82.9 25.8
59.9 29.6
1.44 0.30
a
b
See Figure S6 for the titration curves. KD standard error.
was not reached with the caged variants of the probes at
concentrations well above those needed to achieve complete
binding with free C-terminal carboxylate-containing ligands. The
dissociation constants of the caged peptides were estimated to be
at least 20-fold weaker than those of the parent peptides. We also
observed that the fluorescence was much lower for all of the
caged ligands, again confirming the weak affinities and the
efficacy of the C-terminal carboxylate caging strategy.
Having assessed the efficiency of C-terminal carboxylate
caging in blocking PDZ domain−ligand interactions as well as
the light-assisted recovery of the interactions upon photorelease
of the caging group, we next probed the caged ligands for their
capacity to compete against prototypical PDMIs upon exposure
to light. For this purpose, we compared the effects of
nonfluorescent caged and uncaged ligands at disrupting the
interaction of a fluorescent probe with its cognate PDZ domain
(2 vs GST-PSD-95-3; Figure 3). Consistent with the observed
loss of affinity after introduction of the caging group in the
absence of 365 nm irradiation, the addition of a 20-fold excess of
the caged nonfluorescent competing ligand 9 led to a minimal
Figure 4. (a) Synthesis of monocaged divalent PDLs (see Figure S7 for
details). (b) Fluorescence-based competition assay. The fluorescence
resulting from the interaction of GST-PSD-95-12 and 2 was monitored
as the concentration of nonfluorescent ligand (monovalent 8,
monocaged divalent 10, or divalent 11) was increased.
original synthesis provided a symmetrical ligand dimer, which
could not be easily used to introduce a caging group on only one
of the C-terminal carboxylates. To generate the desired singly
caged divalent ligand, we used copper-catalyzed azide−alkyne
click chemistry to ligate the two forms of the ligand (caged and
noncaged) in solution rather than on resin. The resulting
monocaged divalent ligand derived from the Stargazin sequence
was evaluated by a competition assay with the tandem PDZ
domains of PSD-95 in comparison with both noncaged divalent
and monovalent ligands (Figure 4b). The affinity constants
confirmed that the insertion of a single caging group on a divalent
ligand resulted in behavior comparable to that of the
corresponding monovalent ligand (KI = 0.49 0.01, 14.2
3.1, and 17.9 0.7 μM for noncaged divalent 11, monovalent 8,
and monocaged divalent 10, respectively). In light of our
previous studies,12 it is evident that the caging and photorelease
of the masking group provides a simple method to control the
activity of these potent ligands for acute disruption of PDZ
domain-mediated anchoring of the glutamate receptors at the
synapse. These results also indicate that the general C-terminal
Figure 3. Light-assisted disruption of PDMIs. Competing properties of
caged and noncaged ligands were estimated by observing the disruption
of the binding of a fluorescent probe (5 μM) to its target PDZ domain (5
μM). Competing ligands were used in a 20-fold molar excess (t test: **,
P > 0.1; ***, P < 0.02; n ≥ 3).
4582
dx.doi.org/10.1021/ja309870q | J. Am. Chem. Soc. 2013, 135, 4580−4583

Journal of the American Chemical Society
Communication
caging approach as a method to control PDZ domain−ligand
binding can be extended to more complex systems.
ACKNOWLEDGMENTS
■
This research was supported by NSF CHE-0414243 (to B.I.), the
Cell Migration Consortium (GM064346), the Marie Curie
Postdoctoral Fellowship (PICK-CPP to M.S.), and the ANR
(ChemTraffic to M.S. and D.C.). We thank the staff at NSLS
beamline X6A for access via the General User program.
Finally, since the most important application of the tools
presented here is to provide enhanced temporal control over the
intracellular interactions mediated by proteins containing PDZ
domains, we sought to estimate the stability of the ester bond
connecting the ligand to the caging group in complex
environments. Stability studies carried out using brain lysate
preparations to approximate the conditions to which the caged
probes would be exposed in cellular experiments provided clear
evidence that the caged ligands, which are esters of peptides
containing C-terminal valines, present a significantly higher
resistance toward cellular esterases than do simple unhindered
acetate esters (Figure S8). Specifically, after incubation for 20
min at 37 °C, the background hydrolysis of the caged ligand
approached only ∼10%. In the context of our previous successful
studies with noncaged ligands in cultured neurons,12 this time
window is sufficient to allow for efficient cellular internalization
followed by uncaging experiments with minimal background
hydrolysis. Moreover, the nature of the side chain of the C-
terminal residue in most PDZ domain binding motifs (in this
case valine) contributes significantly to the lower esterase
susceptibility, thereby allowing for the generalization of our
caging strategy. Finally, if additional resistance to cellular
hydrolases is needed, the stability of the caged ligand could be
further improved by a straightforward modification of the NPE
caging group, as shown previously.20
In summary, we have described a general and versatile method
to control the binding of a PDZ domain ligand to its cognate
PDZ domains using photoactivation. The introduction of a
caging group on the critical C-terminal carboxylate of a ligand-
binding motif strongly impairs the interactions with targeted
PDZ domains until a photorelease event is triggered. In view of
the prevalence of C-terminal binding motifs for PDLs over less
common internal sequences,21 the approach should be generally
applicable to most PDZ domain-mediated interactions. In a
complementary approach, photoswitchable cyclic peptides for
noncanonical internal binding motifs were recently reported.22
Importantly, the caging method can be applied to more
sophisticated bivalent ligands, and we have shown that a single
caging group is sufficient to confer upon these ligands tailored
binding properties that can be revealed by a single uncaging
event. Such caged ligands will constitute unique tools for
investigating systems where sharp spatial and temporal control of
the release of the active ligand is critical, such as studies of the
local effect of glutamate receptor trafficking on neighboring
synapses and of the receptor life cycle with respect to diffusion or
endo- or exocytosis. The new tools presented here provide high
spatiotemporal resolution and can be used to study transient and
localized biological events involving PDMIs.
REFERENCES
■
(1) Adams, S. R.; Tsien, R. Y. Annu. Rev. Physiol. 1993, 55, 755. Brieke,
C.; Rohrbach, F.; Gottschalk, A.; Mayer, G.; Heckel, A. Angew. Chem.,
Int. Ed. 2012, 51, 8446.
(2) Ellis-Davies, G. C. Nat. Methods 2007, 4, 619.
(3) Mayer, G.; Heckel, A. Angew. Chem., Int. Ed. 2006, 45, 4900.
(4) Goguen, B. N.; Hoffman, B. D.; Sellers, J. R.; Schwartz, M. A.;
Imperiali, B. Angew. Chem., Int. Ed. 2011, 50, 5667. Lee, H. M.; Larson,
D. R.; Lawrence, D. S. ACS Chem. Biol. 2009, 4, 409. Riggsbee, C. W.;
Deiters, A. Trends Biotechnol. 2010, 28, 468. Banghart, M. R.; Sabatini, B.
L. Neuron 2012, 73, 249.
(5) Nguyen, A.; Rothman, D. M.; Stehn, J.; Imperiali, B.; Yaffe, M. B.
Nat. Biotechnol. 2004, 22, 993.
(6) Humphrey, D.; Rajfur, Z.; Vazquez, M. E.; Scheswohl, D.; Schaller,
M. D.; Jacobson, K.; Imperiali, B. J. Biol. Chem. 2005, 280, 22091. Kuner,
T.; Li, Y.; Gee, K. R.; Bonewald, L. F.; Augustine, G. J. Proc. Natl. Acad.
Sci. U.S.A. 2008, 105, 347.
(7) Bhattacharyya, R. P.; Remenyi, A.; Yeh, B. J.; Lim, W. A. Annu. Rev.
Biochem. 2006, 75, 681. Luck, K.; Charbonnier, S.; Trave, G. FEBS Lett.
2012, 586, 2648.
(8) Hung, A. Y.; Sheng, M. J. Biol. Chem. 2002, 277, 5699.
(9) Kim, E. J.; Sheng, M. Nat. Rev. Neurosci. 2004, 5, 771.
(10) Nourry, C.; Grant, S. G. N.; Borg, J.-P. Sci. STKE 2003, No. 179,
re7. Feng, W.; Zhang, M. J. Nat. Rev. Neurosci. 2009, 10, 87.
(11) 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.
Gerges, N. Z.; Backos, D. S.; Rupasinghe, C. N.; Spaller, M. R.; Esteban,
J. A. EMBO J. 2006, 25, 1623. Bertaso, F.; Zhang, C.; Scheschonka, A.;
de Bock, F.; Fontanaud, P.; Marin, P.; Huganir, R. L.; Betz, H.; Bockaert,
J.; Fagni, L.; Lerner-Natoli, M. Nat. Neurosci. 2008, 11, 940.
(12) (a) Sainlos, M.; Tigaret, C.; Poujol, C.; Olivier, N. B.; Bard, L.;
Breillat, C.; Thiolon, K.; Choquet, D.; Imperiali, B. Nat. Chem. Biol.
2011, 7, 81. (b) Bard, L.; Sainlos, M.; Bouchet, D.; Cousins, S.;
Mikasova, L.; Breillat, C.; Stephenson, F. A.; Imperiali, B.; Choquet, D.;
Groc, L. Proc. Natl. Acad. Sci. U.S.A. 2010, 107, 19561.
(13) Loving, G. S.; Sainlos, M.; Imperiali, B. Trends Biotechnol. 2010,
28, 73.
(14) Sainlos, M.; Iskenderian, W. S.; Imperiali, B. J. Am. Chem. Soc.
2009, 131, 6680.
(15) Cabral, J. H. M.; Petosa, C.; Sutcliffe, M. J.; Raza, S.; Byron, O.;
Poy, F.; Marfatia, S. M.; Chishti, A. H.; Liddington, R. C. Nature 1996,
382, 649. Doyle, D. A.; Lee, A.; Lewis, J.; Kim, E.; Sheng, M.;
MacKinnon, R. Cell 1996, 85, 1067.
(16) 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.
(17) (a) Walker, J. W.; Reid, G. P.; McCray, J. A.; Trentham, D. R. J.
Am. Chem. Soc. 1988, 110, 7170. (b) Wilcox, M.; Viola, R. W.; Johnson,
K. W.; Billington, A. P.; Carpenter, B. K.; McCray, J. A.; Guzikowski, A.
P.; Hess, G. P. J. Org. Chem. 1990, 55, 1585.
ASSOCIATED CONTENT
■
S
* Supporting Information
(18) Vazquez, M. E.; Rothman, D. M.; Imperiali, B. Org. Biomol. Chem.
2004, 2, 1965.
(19) Sainlos, M.; Imperiali, B. Nat. Protoc. 2007, 2, 3219. Sainlos, M.;
Imperiali, B. Nat. Protoc. 2007, 2, 3201.
Supporting figures and experimental methods. This material is
available free of charge via the Internet at http://pubs.acs.org.
(20) Kusaka, N.; Maisch, J.; Nick, P.; Hayashi, K.-i.; Nozaki, H.
ChemBioChem 2009, 10, 2195.
(21) Ivarsson, Y. FEBS Lett. 2012, 586, 2638.
(22) Hoppmann, C.; Schmieder, P.; Domaing, P.; Vogelreiter, G.;
Eichhorst, J.; Wiesner, B.; Morano, I.; Ruck-Braun, K.; Beyermann, M.
Angew. Chem., Int. Ed. 2011, 50, 7699.
AUTHOR INFORMATION
■
Corresponding Author
imper@mit.edu
Notes
The authors declare no competing financial interest.
4583
dx.doi.org/10.1021/ja309870q | J. Am. Chem. Soc. 2013, 135, 4580−4583