Biomimetic screening of class-B G protein-coupled receptors

Article abstract of DOI:10.1021/ja200160s

The 41-amino acid peptide corticotropin releasing factor (CRF) is a major modulator of the mammalian stress response. Upon stressful stimuli, it binds to the corticotropin releasing factor receptor 1 (CRF₁R), a typical member of the class-B G-protein-coupled receptors (GPCRs) and a prime target in the treatment of mood disorders. To chemically probe the molecular interaction of CRF with the transmembrane domain of its cognate receptor, we developed a high-throughput conjugation approach that mimics the natural activation mechanism of class-B GPCRs. An acetylene-tagged peptide library was synthesized and conjugated to an azide-modified high-affinity carrier peptide derived from the CRF C-terminus using copper-catalyzed dipolar cycloaddition. The resulting conjugates reconstituted potent agonists and were tested in situ for activation of the CRF₁ receptor in a cell-based assay. By use of this approach we (i) defined the minimal sequence motif that is required for full receptor activation, (ii) identified the critical functional groups and structure-activity relationships, (iii) developed an optimized, highly modified peptide probe with high potency (EC₅₀ = 4 nM) that is specific for the activation domain of the receptor, and (iv) probed the behavioral role of CRF receptors in living mice. The membrane recruitment by a high-affinity carrier enhanced the potency of the tethered peptides by >4 orders of magnitude and thus allowed the testing of very weak initial fragments that otherwise would have been inactive on their own. As no chromatography purification of the test peptides was necessary, a substantial increase in screening throughput was achieved. Importantly, the peptide conjugates can be used to probe the endogenous receptor in its native environment in vivo.

Full text of DOI:10.1021/ja200160s

ARTICLE
pubs.acs.org/JACS
Biomimetic Screening of Class-B G Protein-Coupled Receptors
Christian Devigny,† Francisco Perez-Balderas,*,†,§ Bastiaan Hoogeland,† Serena Cuboni,† Rudolf Wachtel,†
Christoph P. Mauch,† Katharine J. Webb,†,‡ Jan M. Deussing,†,‡ and Felix Hausch*,†
†Max-Planck-Institute of Psychiatry, Kraepelinstrasse 2-10, 80804 Munich, Germany
‡Institute of Developmental Genetics, Helmholtz Zentrum M€unchen, Ingolstaeter Landstrasse1, 85764 Neuherberg, Germany
S Supporting Information
b
ABSTRACT: The 41-amino acid peptide corticotropin releasing factor (CRF) is a major modulator
of the mammalian stress response. Upon stressful stimuli, it binds to the corticotropin releasing factor
receptor 1 (CRF1R), a typical member of the class-B G-protein-coupled receptors (GPCRs) and a
prime target in the treatment of mood disorders. To chemically probe the molecular interaction of CRF
with the transmembrane domain of its cognate receptor, we developed a high-throughput conjugation
approach that mimics the natural activation mechanism of class-B GPCRs. An acetylene-tagged peptide
library was synthesized and conjugated to an azide-modified high-affinity carrier peptide derived from
the CRF C-terminus using copper-catalyzed dipolar cycloaddition. The resulting conjugates recon-
stituted potent agonists and were tested in situ for activation of the CRF1 receptor in a cell-based assay.
By use of this approach we (i) defined the minimal sequence motif that is required for full receptor
activation, (ii) identified the critical functional groups and structureꢀactivity relationships, (iii) developed an optimized, highly
modified peptide probe with high potency (EC50 = 4 nM) that is specific for the activation domain of the receptor, and (iv) probed the
behavioral role of CRF receptors in living mice. The membrane recruitment by a high-affinity carrier enhanced the potency of the tethered
peptides by >4 orders of magnitude and thus allowed the testing of very weak initial fragments that otherwise would have been inactive
on their own. As no chromatography purification of the test peptides was necessary, a substantial increase in screening throughput was
achieved. Importantly, the peptide conjugates can be used to probe the endogenous receptor in its native environment in vivo.
’ INTRODUCTION
domains. The latter is a key issue not only to better understand the
biochemistry of these receptors but also to advance drug discovery for
this class of receptors. Hereinwedescribetheuseofpeptideꢀpeptide
conjugates for a simple and rapid screening of the activation sequence
motif for class-B GPCRs, using CRF1R as a model system.
Class-B G-protein-coupled receptors (GPCRs) of the secretin
family (or subfamily B) comprise several important drug targets
that are implicated in severe human diseases including diabetes,
osteoporosis, neurodegeneration, inflammation, and psychiatric
disorders.1,2 Class-B GPCRs are without exception activated by
large endogenous peptide hormones. Among them, corticotro-
pin-releasing factor (CRF) is a 41-mer peptide that activates
corticotropin-releasing factor receptor 1 (CRF1R), a class-B
GPCR that has been validated as a prime target in psychiatric
disorders such as anxiety, depression, and addiction.3ꢀ7 The
general mechanism of ligand interaction with class-B GPCRs is
based on a low resolution model termed the two-domain model
(Figure 1a). All class-B GPCRs possess a large extracellular
domain (ECD; ∼100ꢀ160 residues) with which they bind with
high affinity to the carboxyl-terminal regions of their peptide
ligands. This interaction serves as an anchorage for the peptide
ligands.2,8 A second interaction must occur between the juxta-
membrane domain of the receptor and the amino-terminal portion
of the peptide ligand that induces a structural rearrangement
leading to intracellular G-protein activation.9 In this model the
intermediate part of the peptide ligands is thought to serve mainly
as a connector between these two receptor binding sites.10
’ RESULTS AND DISCUSSION
We propose that the coupling of a CRF1R extracellular
domain specific C-terminal peptide with an N-terminal peptidic
fragment could produce high potency agonists (Figure 1b). As
C-terminal peptide we synthesized a cyclic 13 amino-acid pep-
tide that possesses high affinity for the CRF1R extracellular
domain. This defined peptide is conjugated to variable probe
fragments by copper-catalyzed chemical ligation. The coupling of
the complementary azide and alkyne groups by “click-chemistry”
should yield a peptideꢀpeptide conjugate that effectively binds
and activates the CRF receptor.
“Click” PeptideꢀPeptide Conjugates Ligands Are Full
Activators of CRF1R. Peptide 1 was designed as a high-affinity
peptidic carrier based on an optimized C-terminal CRF analog
and was synthesized by standard Fmoc SPPS (see Supporting
Information).11 This compound bears a fluorescent tag for
quantification purposes and a triethylene glycol linker containing
an azide group. Peptide 1retains its high affinity (Kd = 18.8 (2.6 nM)
While the molecular binding mode between the ligand
C-termini and the extracellular domains of class-B GPCRs is very
well understood,8 relatively little is known about the interaction
between the peptide ligand N-termini and the juxtamembrane
Received: January 7, 2011
Published: May 02, 2011
r
2011 American Chemical Society
8927
dx.doi.org/10.1021/ja200160s J. Am. Chem. Soc. 2011, 133, 8927–8933
|

Journal of the American Chemical Society
ARTICLE
to the extracellular domain of CRF1R (Figure S1 of Supporting
Information). N-Terminal fragments, such as peptide 2, were
synthesized using a backbone amide linker (BAL) strategy in order
to introduce an alkyne moiety at the peptide C-terminus.12,13
The isolated peptide fragment 2 did not show any activity at the
CRF1R by itself up to a concentration of 10 μM in a cellular
cAMP stimulation assay. The conjugation of peptide 2 with
peptide 1 by copper catalyzed alkyneꢀazide cycloaddition,
however, reconstituted a full agonist with a potency of 2.82 (
0.23 nM (Figure 2). Thus the presented strategy constitutes a fast
and efficient alternative to the synthesis of full length hormone
derivatives. Moreover, the conjugation to the high-affinity carrier
dramatically enhanced the detection limits for very weak ago-
nists, such as fragment 2, by at least 4 orders of magnitude.
Having proven that “clicked” conjugates can act as high
potency agonists, we went on to determine the minimal peptide
sequence that is necessary to fully activate the receptor. Bidir-
ectionally truncated derivatives of peptide 2 were synthesized
and coupled to peptide 1. C-Terminal deletion of Arg15 led to a
substantial decrease in potency. Surprisingly, N-terminal trunca-
tion of the conjugates initially increased potency. Truncation
beyond Leu5, however, completely abrogated the stimulating
potency of the conjugates (Figure S2 and Table S1 of Support-
ing Information). These results established the 12-amino acid
Ucn4ꢀ15 sequence motif (peptide 7) as a good balance between
size and potency. Conjugate 7 was thus used as a minimized
activation segment and a starting point to further investigate the
characteristics of CRF1R activation.
Medium Throughput Screening of the Urocortin1 Activa-
tion Sequence. To explore the contribution of individual
residues to CRF1R stimulation in more detail we engaged on a
medium throughput screening of peptideꢀpeptide conjugates.
On the basis of the minimized reference template 7 identified
above, we synthesized a library of 96 peptidic fragments contain-
ing a series of single amino acid substitutions. Conveniently, the
synthesis of this library was achieved using standard SPPS in a 96
deep-well plate allowing the fast and parallel synthesis of the 96
C-terminal propargylated activation fragments. Each position of
the Ucn14ꢀ15 peptide sequence was systematically changed to
alanine, aminoisobutyric acid (Aib), the corresponding D-amino
acid, as well as five structurally related natural or unnatural amino
acids (Table 1).
The crude Ucn14ꢀ15 analogs were coupled via their C-term-
inal alkyne group to the purified carrier peptide 1 via the
previously described method. Gratifyingly, the peptideꢀpeptide
conjugates were routinely found to be much less soluble than
their precursors allowing simple precipitation for their purifica-
tion (Figure S3 of Supporting Information). HPLC analysis of
the crude conjugates showed complete conversion of the carrier
peptide 1 for all 96 conjugation reactions and acceptable purities
(g65%) for most crude conjugates (89 out of 96).14 At this stage
we decided to test these crude peptideꢀpeptide conjugates
directly in a cellular assay without any further purification. This
was possible as any residual reaction components (e.g., Cuþ, t-
butanol, or any remaining unconjugated peptide fragments) were
minimal and were found not to disturb the cellular assay.
To ensure the suitability of this procedure we purified 14 con-
jugates, including sequences with enhanced activity as well as
peptide conjugates with <65% purity, by HPLC and compared the
activities of the purified conjugates in the cellular assay with their
crude product counterparts. In all the cases, only minimal differ-
ences in activity were found between the crude and the purified
peptide conjugates, thus validating the in situ screening approach
(Figure S4 and Table S5 of Supporting Information).14
Figure 1. Mechanism of class-B GPCR signaling (a) and principle of
biomimetic screening (b). (a) Mechanism of class-B GPCR signaling:
The C-terminal part of the peptide ligand (dark green) binds with high
affinity to the extracellular domain of the GPCR (dark blue). This
recruits the N-terminal activation domain of the peptide ligand (light
green) to the juxtamembrane domain of the GPCR leading to intracel-
lular signal transduction.9 (b) Biomimetic screening: A library of alkyne-
tagged peptides (light green) is conjugated to a constant peptide
fragment (dark green) that has high-affinity for the extracellular domain
of the class-B GPCR. These conjugates are probed for modulation of the
GPCR transmembrane domain.
The biomimetic screening of the peptide conjugate library
resulted in 70 conjugates where the stimulatory activity for CRF1R
was reduced or completely abolished (Table 1 and Table S4 of
Figure 2. Stimulation of CRF1R by a peptideꢀpeptide conjugate. (a) Synthesis of the activation segment 2 (light green) using BAL resin 22 and “click”
conjugation with membrane recruition segment 1 that is derived from CRF1R ligands. X = cyclohexyl alanine, EAEK = lactam bridge, NBD = 4-(7-
nitro)benzofurazanyl. (b) Stimulation of cAMP production in HEK293 cells stably overexpressing CRF1R by peptide conjugate 3 and CRF.
8928
dx.doi.org/10.1021/ja200160s |J. Am. Chem. Soc. 2011, 133, 8927–8933

Journal of the American Chemical Society
ARTICLE
Table 1. Summary of Single-Point Substitutions of the Ucn14ꢀ15 Sequence
a These conjugates were additionally purified by high-performance liquid chromatography (HPLC) and retested for activity. Aib = R-aminoisobutyric
acid; Cha = 3-cyclohexyl-L-alanine; Chg = L-R-cyclohexylglycine; Cit = L-citrulline; Dap = L-2,3-diaminopropionic acid; Dab = L-2,4-diaminobutyric acid;
h-Ser = homoserine; Nle = norleucine; Orn = ornithine; PhGly = L-R-phenylglycine; pI-Phe = para-iodo phenylalanine; tBuAla = β-tert-butyl-L-alanine;
tBuGly = L-R-tert-butylglycine. Crude peptideꢀpeptide conjugates carrying the indicated amino acid change were assayed at 2 nM for cAMP stimulation
of HEK293 cells stably overexpressing CRF1R. Inactive sequences are indicated in red, minimal activity is indicated in orange, equal activity compared to
the original Ucn14ꢀ15 conjugate 7a is indicated in yellow, and activity-enhancing substitutions are highlighted in green.
Table 2. Binding to or Stimulation of CRF1R (Cellular cAMP Production) by Multisubstituted Peptide Fragments and Conjugates
(Mean ( Std Dev)
R = carrier 1a
EC50 (nM) c
R = NH2b
EC50 (nM) c
IC50 (nM) d
IC50 (nM) d
þ15
þI125sauvagine
þ15
þI125sauvagine
ꢀ15
(5 μM)
(20pM)
ꢀ15
Inact.g
487 ( 144I
181 ( 15
48 ( 2
(5 μM)
(20pM)
7
Ac-Ucn14ꢀ15-R
Ac-[Thr4Cha5]Ucn14ꢀ15-R
Ac-[Thr4Cha5D-Phe11]Ucn14ꢀ15-R
Ac-[Thr4Cha5,13D-Phe11] Ucn14ꢀ15-R
Ac-[Thr4Cha5,13Chg7D-Phe11] Ucn14ꢀ15-R
Ac-[Thr4Cha5,13Chg7tBuAla9D-Phe11] Ucn14ꢀ15-R
1.12 ( 0.05
0.053 ( 0.001
0.040 ( 0.001
0.074 ( 0.001
0.037 ( 0.001
0.049 ( 0.001
Inact.f
7.1 ( 0.5
5.1 ( 0.5
8.4 ( 0.5
5.5 ( 0.3
5.9 ( 0.5
5.4 ( 2.6
3.6 ( 1.8
1.8 ( 0.9
2.2 ( 1.3
1.6 ( 1.0
1.6 ( 0.6
Inact.h
Inact.h
Inact.e
Inact.e
10
11
12
13
14
396 ( 31
113 ( 7
50 ( 6
10188 ( 6611
3738 ( 1604
2977 ( 1602
400 ( 190
25 ( 1
4.0 ( 0.1
10.0 ( 1.0
a Propargylated peptides conjugated to peptide 1. b C-terminally amidated. c cAMP stimulation of a stable CRF1R overexpressing cell line in the absence
or presence of 5 μM of the extracellular domain-specific antagonist 15. d Radioactive binding assay using membrane preparations from CRF1R-
overexpressing HEK293 cells. e Inactive up to 80 μM. f Inactive up to 50 nM . g Inactive up to 10 μM. h Inactive up to 5 μM. I Partial agonist.
Supporting Information). For 15 of these peptide conjugates, the
CRF1R stimulation was not significantly different compared to
the wild-type sequence of Ucn14ꢀ15. However, an enhanced
stimulation was observed for 11 peptide conjugates analogs.
Analysis of the activities for the single amino acid substitution
analogs allowed several conclusions about structureꢀactivity
relationship of CRF1R agonist. First, the hydrogen bonding
pattern of Arg15 seems to contribute to CRF1R stimulation,
consistent with the truncation study. Generally, positions 5, 7, 9,
13, and 14 tolerate only conservative aliphatic substitutions
which can partially enhance potency. His12 seems to contribute
comparatively less to the interaction with the CRF1R receptor.
An L- to D-Phe substitution at position 11 substantially improved
the activity of the peptide conjugate. Interestingly, this substitu-
tion is also present in the high-affinity CRF1R antagonist
astressin.15 The polar residues Ser6, Asp8, and to a lesser extent
Thr10 are essential for activation, which is consistent with prior
studies on full length CRF.16,17 Ser4 could be advantageously
substituted for several amino acids.
Discovery of Agonists Specific for the Transmembrane
Domain of CRF Receptors. The next step was to investigate
whether the enhancing effects of the single substitutions identi-
fied above were additive. In the medium throughput screening,
the substitutions Ser4Thr, Leu5Cha, and Phe11D-Phe provided
the greatest individual enhancement in potency. Combination of
the first two mutations (10a) enhance potency 24-fold compared
to the original urocortin4ꢀ15 sequence (Table 2). The triple
mutation in conjugate 11a further slightly increased potency to
provide a 40 pM agonist. Importantly, the combination of these
point mutations showed for the first time detectable CRF1R
agonism in an untethered setup (peptides 10b and 11b, Table 2
and Figure S9 of Supporting Information). The successive
incorporation of three additional mutations identified as bene-
ficial in the single substitution screen (Ile7Chg, Leu9tBuAla, and
8929
dx.doi.org/10.1021/ja200160s |J. Am. Chem. Soc. 2011, 133, 8927–8933

Journal of the American Chemical Society
ARTICLE
peptide 15,11 which is an antagonist specific for the extracellular
domain of CRF1R (Table 2 and Figure 3). The cAMP stimula-
tion potency of the conjugates was strongly reduced by a factor of
100 in the presence of 5.0 μM peptide 15, a 200-fold excess
compare to its Ki determined in the radioactive binding experi-
ment (Figure S1 of Supporting Information). This is consistent
with peptide 15 competitively inhibiting the binding of the
carrier part of the conjugates to the extracellular domain, the
latter being the dominating recognition event. For conjugate 7a,
the weakest of the conjugate series, no activity could be detected
in the presence of inhibitor 15. Likewise, CRF was strongly
inhibited by peptide 15 (Figure 3). In contrast to the conjugates
and CRF, the activities of the untethered peptides 10bꢀ14b
were only mildly affected by a factor of 2.2 by the presence of 5.0
μM peptide 15 (Table 2 and Figure 3).
We next tested the conjugates and untethered peptides in
HEK293 cells transiently overexpressing the full length CRF1R
or a truncated mutant lacking the extracellular domain (Table S2
of Supporting Information and Figure 3). While the results
obtained using the full length construct nicely replicated those
obtained using the stable CRF1R overexpressing cell line for the
conjugates and the untethered peptide series, substantial differ-
ences were found for the truncated mutant. In the latter system,
none of the conjugates showed any remaining activity; neither
did CRF (Table S2 of Supporting Information and Figure 3).
The untethered peptides 10bꢀ14b, however, all activated the
extracellular domain-deficient CRF1R mutant, albeit with about
4-fold reduced potency. This clearly and directly shows that the
transmembrane domain is the primary interaction site for these
N-terminally derived CRF analogs, with only minor contribu-
tions from the extracellular domain, possibly indirectly by
stabilizing the correct conformation of the juxtamembrane
domain. It is noteworthy that the untethered peptides were able
to stimulate the truncated CRF1R mutant with good potency
(e.g., 15.6 ( 0.5 nM for peptide 14b), while their conjugated
counterparts (e.g., 14a) were inactive in the same setting. This
could indicate that the carrier peptide 1 could actually negatively
affect their conjugation partners when not bound to the extra-
cellular domain.
To further characterize the interaction of peptide 14b or
conjugate 7a with the transmembrane domain, we studied
their sensitivity toward DMP696,19 a nonpeptidic antagonist
specific for the juxtamembrane domain (Figure 3, Figure 4, and
Table S3 of Supporting Information). Increasing concentrations
of DMP696 incrementally depressed the maximal stimulation
(Emax) induced by 7a while only weakly affecting the potency
(EC50) of the latter (Figure 4a), consistent with a noncompeti-
tive mode of inhibition. This is in line with the two domain model
were the receptor occupancy is largely determined by the
extracellular domain and its high affinity to the carrier peptide
(unaffected by DMP696), while efficacy is primarily determined
by the interaction of the transmembrane domain and the con-
jugated N-terminal peptide. Importantly, the two-domain model
predicts that concentrations of conjugate 7a that fully saturate
the extracellular domain constitute an upper limit to the effective
(“intramolecular”) concentration of the conjugated peptide 7
that can be “perceived” by the transmembrane domain. Further
increasing concentrations of free conjugated 7a unligated to the
extracellular domain can be neglected in this context (Table 2
and Table S3 of Supporting Information, dECD). This maximal
effective concentration of 7 is eventually unable to overcome the
increasing concentrations of DMP696 (IC50 = 67 ( 5 nM for
Figure 3. Stimulation of CRF1R and extracellular domain-truncated
CRF1R by an optimized peptide agonist. HEK293 cells transiently
overexpressing full length CRF1R24ꢀ415 (brown bars), a truncated
version CRF1R112ꢀ415 (yellow bars), or an empty vector (gray bars)
were stimulated with CRF (filled) or peptide 14b (hatched) at the
indicated concentration in the absence or presence of the antagonists
DMP696 (2 μM) or peptide 15 (Ac-EAEKNRKLXDII-NH2,11 X = Cha,
underlined = lactam bridge, 10 μM). After 30 min, cAMP accumulation
was determined and normalized to a negative (DMSO, 0%) and positive
control (forskolin, 100%).
Leu13Cha) further enhanced the potency of untethered peptides
to provide the hexasubstituted peptide 14b as a low nanomolar
agonist. From a structural point of view, it seems that this
increase in potency was largely achieved by fine-tuning the
hydrophobic interactions through incorporation of the unnatural
amino acids side chains (Cha, Chg, and tBuAla). Intriguingly,
these additional substitutions did not further increase the po-
tency in the context of the high-affinity carrier. This is in contrast
to the cooperative behavior classically expected for conjugated
two-site binding fragments.18 The molecular underpinnings for
the apparent lower limit of approximately 50 pM for peptide-
carrier 1 conjugates are currently unknown.
To confirm the specificity of peptide 14b for CRF1R in our test
system, we first performed experiments using HEK293 cells
transiently transfected with CRF1R or mock controls. No
significant cAMP stimulation could be detected in the latter
setup (Figure 3). We further characterized the specificity of
peptide 14b by testing its ability to stimulate others class-B
GPCRs. Peptide 14b was active at the CRF2R (34.1 ( 1.4 nM),
the closest homologue of CRF1R (>70% amino acid sequence
identity). CRF2R is also potently stimulated by the endogenous
ligand Urocortin1 from which peptide 14b was derived from,
explaining the low degree of selectivity between the two CRF
receptor subtypes. However, peptide 14b was inactive in an assay
using HEK293 cells transiently transfected with PTH1R and
GLP1R, showing that there is no cross-reactivity for other closely
related receptors (data not shown).
Pharmacological Characterization of Conjugates and
Transmembrane-Specific Peptides. The N-terminal part of
CRF corresponding to peptide 14b is believed to primarily interact
with the juxtamembrane domain of CRF1R. To further characterize
the binding mode of peptides 7bꢀ14b or of the corresponding
conjugates we first performed cellular competition assays using
8930
dx.doi.org/10.1021/ja200160s |J. Am. Chem. Soc. 2011, 133, 8927–8933

Journal of the American Chemical Society
ARTICLE
Figure 4. Mechanistic analysis of DMP696 inhibition vs peptide agonists 7a and 14b. Full doseꢀresponse curves for (a) conjugate 7a and (b) peptide
14b stimulated cAMP production in absence or presence of increasing concentrations of DMP696 using HEK293 cells stably overexpressing CRF1R.
Figure 5. Acoustic startle response (ASR) in mice following intracerebroventricular injection of a peptideꢀpeptide conjugate. Conjugate 7a (0.16 μg,
filled blue triangles), CRF (0.16 μg, filled black circles), and their corresponding vehicle controls (open triangles and circles, respectively) were injected
directly into the lateral ventricle of mice, and the ASR was measured using increasing stimulus intensity (noise bursts). (a) CRF and conjugate 7a injection
led to significantly higher ASR than vehicle controls (105 and 115 dB, p < 0.05, respectively), but ASR increase by CRF did not differ significantly from
that caused by conjugate 7a (p > 0.05; statistical analyses were performed by 2-way repeated measures ANOVA, followed by Newman-Keulsposthoc test).
(b) In comparison to wild-type mice conjugate 7a was unable to increase the ASR in CRF1R knockout mice. All data are shown as mean ( SEM.
depression of Emax of peptide 7a) thus providing a mechanistic
rationale for the observed noncompetitive mode of inhibition.
The sensitivity toward DMP696 was also investigated for the
untethered multisubstituted agonist 14b (Figures 3 and 4b)
revealing an interaction reminiscent of a mixed competitive
mode of inhibition. DMP696 strongly affected the EC50 of
peptide 14b even at low concentrations, while at higher con-
centrations it also reduced the Emax of the latter.
Finally, to probe the interaction the conjugates and untethered
peptides to CRF1R more directly competitive binding assays
were performed with the natural peptide ligand I125sauvagine
(Table 2). All conjugates fully competed with the labeled tracer
in the low nanomolar range, with only a modest increase in
affinity with increasing degree of substitution in the conjugated
N-terminal segment (compare IC50 = 5.4 ( 2.6 nM for 7a to
1.6 ( 0.6 nM for 14a). This is consistent with the constant carrier
peptide 1 being the major driving force for binding and competition.
The untethered peptides competed with I125sauvagine only
weakly, likely due to the lack of the high-affinity extracellular
interaction and, only partially, reflecting a competition only with
the fraction of I125sauvagine that is bound to the high-affinity, γS-
GTP-sensitive CRF1R binding sites. The binding affinities of
peptides 11bꢀ14b, however, were substantially improved with
increasing degree of substitution, in line with the trend observed
in the cAMP stimulation experiment. In all cases the affinities
measured in the radioactive binding assays were substantially
weaker than the potencies observed in the cAMP stimulation
assays. This has been repeatedly observed also for the natural
ligands CRF, urocortin, or sauvagine and possibly reflects a sub-
stantial receptor reserve in our CRF1R overexpressing cells
where only a small fraction of stimulated receptor is sufficient
to fully activate the cellular secondary messenger machinery.
Biomimetic Targeting of Endogenous CRF1R in Living Mice.
A hallmark of the biomimetic screening approach presented here
8931
dx.doi.org/10.1021/ja200160s |J. Am. Chem. Soc. 2011, 133, 8927–8933

Journal of the American Chemical Society
ARTICLE
is the possibility to target the unmodified native receptors. The
peptideꢀcarrier conjugates should therefore also be applicable to
the endogenous receptors in their native environment. To test
the effects of conjugates in a physiologically relevant system, the
conjugate 7a (Ucn14ꢀ15) was administered directly to the brain
of wild-type mice via intracerebroventricular injection. The
CRF1R-mediated modulation of the stress response was then
evaluated using the acoustic startle response (ASR) paradigm
(Figure 5a).20 Gratifyingly, the sensitizing effects of the peptide
conjugate on the startle response were qualitatively similar to
exogenously applied CRF controls. Moreover, CRF1R knockout
mice did not respond to the injection of peptide conjugate 7a
(Figure 5b), thus proving the specificity of the synthetic peptide
conjugates for the endogenous CRF1 receptor in vivo. Likewise,
the potentiation in startle response was also observed after
injection of the optimized, juxtamembrane-specific agonist
14b, again in a CRF1R-dependent manner (Figure S5 of
Supporting Information).
approach is inherently limited to genetically encoded amino acids.
It is noteworthy that—at least in this example for CRF1R—five
unnatural amino acids substitutions were necessary to achieve
low nanomolar, juxtamembrane-specific ligands. As such the
“carrier-conjugate” approach presented in this study might be
better suited to produce hits for drug development, as this
method can be readily adapted to unnatural or even nonpeptidic
test substances.
’ EXPERIMENTAL METHODS
Standard Procedure for the Conjugation of Peptides. The
peptidic carrier 1 (1 mM solution in acetonitrile/water 50/50, v/v)
(15 nmol) and the desired propargylated N-terminal peptide (1 mM
solution in acetonitrile/water 50/50, v/v) (4 equiv, 60 nmol) were
transferred to a reaction tube (1.5 mL). The mixture was evaporated
under reduced pressure. The residue was dissolved in tert-butanol/water
(43/1, v/v, 722 μL); sonication was used to help peptide dissolution
where necessary. To this solution were added sequentially TBTA (1 mM
solution in tert-butanol/water 43/1, v/v, 1.3 equiv., 20 nmol, 20 μL),
sodium ascorbate (71 mM solution in water, 28 equiv, 412 nmol, 5.8
μL), and copper sulfate (140 mM solution in water, 21 equiv, 308 nmol,
2.2 μL). The mixture was shaken for 40 h at 37 °C. The tube was cooled
down to room temperature and then centrifugated (13000 rpm, 5 min).
The supernatant was removed by decantation and the precipitate was
washed with tert-butanol/water (43/1, v/v, 100 μL). The precipitate
was dried under vacuum. The crude conjugates were dissolved in DMSO
and analyzed by reverse-phase HPLC.
cAMP Stimulation Assay. For the stimulation assay, cells grown
to 50ꢀ60% confluency in DMEM (þ FCS 10%, penicillin/
streptomycin) were detached from the culture dish with Cell dissocia-
tion solution (Sigma, Germany), resuspended in stimulation buffer (SB,
5 mM HEPES, 0,1% BSA, 0.1 mM IBMX in HBSS), and seeded in 5 μL
into a Corning 384 well plate (#3572) at a density of 3000 cells/well.
After equilibration for 10 min at room temperature, 5 μL of SB
containing peptides and/or DMSO, forskolin, or DMP696 (generous
gift by Bristol-Myers-Squibb) were added. After incubation for 30 min at
room temperature, cells were lysed, and the cAMP content was
measured using the cAMP dynamic HTRF detection kit (CisBio,
France) according to the manufactures instructions and using a Tecan
Genios Pro equipped with 320/35 nm excitation and 620/10 nm and
665/8 nm emission filters. The ratiometric fluorescence signals obtained
were normalized to the positive (forskolin, 100%) and negative controls
(DMSO, 0%). Additional information on cell culture, transfection
conditions, and constructs used are available in the Supporting
Information.
’ CONCLUSIONS
In this work we developed a rapid approach to chemically
probe the activation domain of class-B GPCRs. We show that the
conjugation of a high affinity carrier results in a dramatic
enhancement in activity. This allows the initial testing of peptides
whose activity would otherwise have been too weak to be
detectable. The synthesis and coupling of short length peptides
to reconstitute a fully functional CRF1R modulator has several
advantages compared to the classical synthesis of whole, full
length peptide ligands. For example, it is quicker with higher
product purities. Importantly, the fragment conjugation ap-
proach obviated the need for HPLC purification of the initial
test peptides, thereby enabling a substantially higher screening
throughput. The crude peptide conjugates were compatible with
functional assays in living cells. This was likely facilitated in part
by the exquisite target receptor specificity of the conjugates
imparted by the carrier. This also enabled the straightforward
extension of this concept to very complex biological systems,
such as whole animal models.
By application of this methodology to the class-B GPCR
CRF1R we discovered a low nanomolar agonist that is specific for
the activation domain of this receptor. The activation mechanism
of peptide 14b is almost independent of the extracellular domain
and resembles the signaling mechanism of canonical class A
GPCRs. We assume that peptide 14b binds at the endogenous
hormone orthosteric site and derives its activity from optimized
hydrophobic interactions. Our results are inconsistent with a
necessary allosteric change of the extracellular domain which had
been postulated as the initiating step of signal transduction for
other class-B GPCRs.21,22 CRF1R therefore resembles more
closely the parathyroid hormone receptor 1—the only other
class-B GPCR for which juxtamembrane-specific ligands have
been described.23,24 Whether transmembrane-specific agonists
for other class-B GPCRs can be identified by the method des-
cribed in this manuscript remains to be determined.
ASR. ASR testing was carried out as described in Golub et al.20
Briefly, mice injected with either peptide or vehicle were subjected to
noise impulses of increasing intensities (75, 90, 105, and 115 dB(A)) in a
SR-LAB (San Diego Instruments) apparatus with background noise set
to 50 dB(A). The data depicted on the graphs represent the mean peak
startle amplitude in mV ( SEM in response to 30 pulses of each intensity
as well as 12 background (BG) noise measurements. The effect of the
conjugate 7a on the ASR in comparison to CRF was studied in Crhr1flox/
flox control mice obtained from breedings of conditional CRF1R knock-
out mice as previously described (n = 9ꢀ12).26 The data were subse-
quently subjected to 2-way repeated measures ANOVA followed by
Newman-Keuls posthoc test. Moreover, to assess the specificity of
peptides 7a and 14b, the ASR response of constitutive CRF1R knockout
mice (n = 2ꢀ3)27 was measured in comparison to wild-type littermates
(n = 3ꢀ4). The guide cannulas for the intracerebroventricular injections
were implanted in anaesthetised mice secured in a stereotaxic frame. The
skull surface was exposed, and a 23 gauge guide cannula, 8 mm in length,
was inserted using the following coordinates: 0.3 mm posterior from
Given the high mechanistic and structural homology within
the class-B GPCR family,2,8 we expect the biomimetic screening
approach to be applicable to many class-B GPCRs. The key
principle of this study—testing class-B ligands after membrane
recruitment—is related in some aspects to a recently published
“mammalian class-B ligand display”.25 However, while poten-
tially superior in the number of tested peptides, the latter
8932
dx.doi.org/10.1021/ja200160s |J. Am. Chem. Soc. 2011, 133, 8927–8933

Journal of the American Chemical Society
ARTICLE
bregma (AP), 1 mm lateral distance from the midline (L), and 1.2 mm
deep from the level of the skull surface (DV). The animals were left to
recover for at least ten days before testing.
(10) Beyermann, M.; Rothemund, S.; Heinrich, N.; Fechner, K.;
Furkert, J.; Dathe, M.; Winter, R.; Krause, E.; Bienert, M. J. Biol. Chem.
2000, 275, 5702.
(11) Yamada, Y.; Mizutani, K.; Mizusawa, Y.; Hantani, Y.; Tanaka,
M.; Tanaka, Y.; Tomimoto, M.; Sugawara, M.; Imai, N.; Yamada, H.;
Okajima, N.; Haruta, J. J. Med. Chem. 2004, 47, 1075.
(12) Jensen, J. A.; Songster, M. F.; Vꢀagner, J.; Albericio, F.; Barany,
G. J. Am. Chem. Soc. 1998, 120, 5441.
Data Analysis. SigmaPlot software version 11.0 was used for
sigmoidal curve fitting of ligand concentrationꢀresponse curves and
for calculating the EC50 values as an index of ligand potency. The
statistical parameters were generated from a composite nonlinear
regression (four parameter logistic curve, SigmaPlot) of pooled data
from two independent samples. In general, the mean and the standard
deviation were expressed for values obtained from duplicate samples.
The cAMP production response was normalized to a positive control
(Forskolin, 100%) and a negative control (DMSO, 0%). GraphPad
Prism software version 5.0 was used for curve fitting of the acoustic
startle response in mice and statistical analysis of variance.
(13) Springer, J.; Kimberly, R. d. C.; Sandrine, C. V.; Jan, A. J. G.;
Pedro, H. H. H.; Henk, H.; Jan, H. v. M. Eur. J. Org. Chem. 2008, No. 15,
2592.
(14) The Ucn4ꢀ15 sequence had a pronounced tendency to form an
isomer with higher polarity which was the major impurity for most Ucn
analogs. Purified samples of selected isomeric conjugates were found to
stimulate CRF1R with equal potency compared to their corresponding
correct isomers.
(15) Gulyas, J.; Rivier, C.; Perrin, M.; Koerbe, S. C.; Sutton, S.;
Corrigan, A.; Lahrichi, S. L.; Craig, A. G.; Vale, W. W.; Rivier, J. E. Proc.
Natl. Acad. Sci. U.S.A. 1995, 92, 10575.
(16) Kornreich, W. D.; Galyean, R.; Hernandez, J. F.; Craig, A. G.;
Donaldson, C. J.; Yamamoto, G.; Rivier, C.; Vale, W. W.; Rivier, J. E.
J. Med. Chem. 1992, 35, 1870.
’ ASSOCIATED CONTENT
S
Supporting Information. Information about experimental
b
procedures, full references, analytical and spectral characteriza-
tion data, cAMP stimulation assays, and animal experiments.
This material is available free of charge via the Internet at http://
pubs.acs.org.
(17) Beyermann, M.; Fechner, K.; Furkert, J.; Krause, E.; Bienert, M.
J. Med. Chem. 1996, 39, 3324.
(18) Murray, C. W.; Verdonk, M. L. J. Comput. Aided Mol. Des. 2002,
16, 751.
’ AUTHOR INFORMATION
(19) He, L.; Gilligan, P. J.; Zaczek, R.; Fitzgerald, L. W.; McElroy, J.;
Shen, H. S.; Saye, J. A.; Kalin, N. H.; Shelton, S.; Christ, D.; Trainor, G.;
Hartig, P. J. Med. Chem. 2000, 43, 449.
Corresponding Author
hausch@mpipsykl.mpg.de; francisco.perez-balderas@rob.ox.ac.uk
(20) Golub, Y.; Mauch, C. P.; Dahlhoff, M.; Wotjak, C. T. Behav.
Brain Res. 2009, 205, 544.
(21) Dong, M.; Pinon, D. I.; Asmann, Y. W.; Miller, L. J. Mol.
Pharmacol. 2006, 70, 206.
(22) Dong., M.; Gao, F. D.; Pinon, D. I.; Miller, L. J. Mol. Endocrinol.
2008, 22, 1489.
Present Addresses
§CR/UK Gray Institute for Radiation Oncology and Biology,
Department of Oncology, University of Oxford, Old Road
Campus Research Building, Off Roosevelt Drive, Churchill
Hospital, Oxford OX3 7DQ, United Kingdom
(23) Shimizu, M.; Potts, J. T., Jr.; Gardella, T. J. J. Biol. Chem. 2000,
275, 21836.
’ ACKNOWLEDGMENT
(24) Carter, P. H.; et al. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 6846.
(25) Fortin, J.-P.; Zhu, Y.; Choi, C.; Beinborn, M.; Nitabach, M. N.;
Kopin, A. S. Proc. Natl. Acad. Sci. U.S.A. 2009, 106, 8049.
(26) M€uller, M. B.; et al. Nat. Neurosci. 2003, 6, 1100.
(27) Timpl, P.; Spanagel, R.; Sillaber, I.; Kresse, A.; Reul, J. H. M.;
Stalla, G. K.; Blanquet, V.; Steckler, T.; Holsboer, F.; Wurst, W. Nat.
Genet. 1998, 19, 162.
We thank Mrs. E. Weyher and Dr. S. Uebel (MPI of
Biochemistry, Martinsried, Germany), Dr. G. Maccarone and
Pr. Dr. C. Turck (MPI of Psychiatry, Munich, Germany) for MS
analysis, and Mrs. C. Dubler (Ludwig-Maximilians-Universit€at
M€unchen) for NMR measurements. We are indebted to Prof. F.
Holsboer for generous support. This work was partially sup-
ported by the Chemical Genomics Centre (Dortmund), by the
Bundesministerium f€ur Bildung und Forschung within the frame-
work of the NGFN-Plus (FKZ: 01GS08155), and by the
Initiative and Networking Fund of the Helmholtz Association
in the framework of the Helmholtz Alliance for Mental Health in
an Ageing Society (HA-215).
’ REFERENCES
(1) Foord, S. M.; Jupe, S.; Holbrook J. Biochem. Soc. Trans. 2002,
30, 473.
(2) Hoare, S. R. Drug Discov. Today 2005, 10, 417.
(3) Holsboer, F. Neuropsychopharmacology 2000, 23, 477.
(4) Bale, T. L.; Vale, W. W. Annu. Rev. Pharmacol. Toxicol. 2004,
44, 525.
(5) Grigoriadis, D. E. Expert Opin. Ther. Targets 2005, 9, 651.
(6) Holsboer, F.; Ising, M. Annu. Rev. Psychol. 2010, 61, 81.
(7) Koob, G. F.; Zorrilla, E. P. Curr. Opin. Invest. Drugs 2010, 11, 63.
(8) Parthier, C.; Reedtz-Runge, S.; Rudolph, R.; Stubbs, M. T.
Trends Biochem. Sci. 2009, 34, 303.
(9) The juxtamembrane domain is defined as the extracellular loops
and the extracellular facing parts of the transmembrane helices
of GPCRs.
8933
dx.doi.org/10.1021/ja200160s |J. Am. Chem. Soc. 2011, 133, 8927–8933