Valentina Albanese1, Chiara Ruzza2,3, Erika Marzola1, Tatiana Bernardi1, Martina Fabbri1, Anna Fantinati1, Claudio Trapella1,3, Rainer K Reinscheid4,5, Federica Ferrari2, Chiara Sturaro2, Girolamo Calò6, Giorgio Amendola7, Sandro Cosconati7, Salvatore Pacifico1, Remo Guerrini1,3, Delia Preti1. 1. Department of Chemical, Pharmaceutical and Agricultural Sciences, University of Ferrara, Via Luigi Borsari 46, 44121 Ferrara, Italy. 2. Department of Neuroscience and Rehabilitation, Section of Pharmacology, University of Ferrara, Via Fossato di Mortara 17/19, 44121 Ferrara, Italy. 3. LTTA Laboratory for Advanced Therapies, Technopole of Ferrara, 44121 Ferrara, Italy. 4. Institute of Pharmacology and Toxicology, Jena University Hospital, Friedrich Schiller University, 07747 Jena, Germany. 5. Institute of Physiology I, University Hospital Münster, University of Münster, 48149 Münster, Germany. 6. Department of Pharmaceutical and Pharmacological Sciences, University of Padova, Largo Meneghetti, 2, 35131 Padova, Italy. 7. "DiSTABiF", Università della Campania "Luigi Vanvitelli", Via Vivaldi 43, 81100 Caserta, Italy.
Abstract
Neuropeptide S modulates important neurobiological functions including locomotion, anxiety, and drug abuse through interaction with its G protein-coupled receptor known as neuropeptide S receptor (NPSR). NPSR antagonists are potentially useful for the treatment of substance abuse disorders against which there is an urgent need for new effective therapeutic approaches. Potent NPSR antagonists in vitro have been discovered which, however, require further optimization of their in vivo pharmacological profile. This work describes a new series of NPSR antagonists of the oxazolo[3,4-a]pyrazine class. The guanidine derivative 16 exhibited nanomolar activity in vitro and 5-fold improved potency in vivo compared to SHA-68, a reference pharmacological tool in this field. Compound 16 can be considered a new tool for research studies on the translational potential of the NPSergic system. An in-depth molecular modeling investigation was also performed to gain new insights into the observed structure-activity relationships and provide an updated model of ligand/NPSR interactions.
Neuropeptide S modulates important neurobiological functions including locomotion, anxiety, and drug abuse through interaction with its G protein-coupled receptor known as neuropeptide S receptor (NPSR). NPSR antagonists are potentially useful for the treatment of substance abuse disorders against which there is an urgent need for new effective therapeutic approaches. Potent NPSR antagonists in vitro have been discovered which, however, require further optimization of their in vivo pharmacological profile. This work describes a new series of NPSR antagonists of the oxazolo[3,4-a]pyrazine class. The guanidine derivative 16 exhibited nanomolar activity in vitro and 5-fold improved potency in vivo compared to SHA-68, a reference pharmacological tool in this field. Compound 16 can be considered a new tool for research studies on the translational potential of the NPSergic system. An in-depth molecular modeling investigation was also performed to gain new insights into the observed structure-activity relationships and provide an updated model of ligand/NPSR interactions.
Neuropeptide S (NPS),
identified in 2002 by a reverse pharmacology
approach,[1] is the endogenous ligand of
a previous orphan G protein-coupled receptor (GPCR), now named neuropeptide
S receptor (NPSR). NPS is a 20 amino acid neuropeptide (primary sequence
in humans: SFRNGVGTGMKKTSFQRAKS) highly conserved among different
species, and it owes its name to the serine residue at the 1-position
of the peptide sequence. NPSR shows a moderate homology with the other
members of the GPCR family. The in vitro pharmacology
of the human and mouseNPSR showed that NPS increases both intracellular
calcium levels and cAMP accumulation with EC50 values in
the low nanomolar range. This indicates that NPSR can signal via both
Gq and Gs pathways to increase cellular excitability.[2,3] In the rodent brain, NPS is expressed only in few neurons in the
peri-locus coeruleus region. On the contrary, NPSR is widely expressed
in several brain regions (i.e., hypothalamus, endopiriform nucleus,
amygdala, subiculum, cortex, and nuclei of the thalamic midline).[4,5]In vivo, NPS has been shown to control several
biological functions in rodents including stress, anxiety, social
behavior, locomotor activity, wakefulness, food intake and gastrointestinal
functions, memory processes, pain, and drug abuse.[6,7] As
far as the therapeutic potential of selective NPSR ligands is concerned,
NPSR agonists may be useful as innovative anxiolytics devoid of sedative
effects, analgesics, and nootropics. On the other hand, NPSR antagonists
may be useful to treat substance abuse disorders against which there
is an urgent need for the exploration of novel potential drug targets
and for developing innovative therapeutic approaches.[8]NPSR antagonists with potent in vitro activity
have been developed in the last few years and a few compounds are
currently in use as pharmacological tools.[7] Among these, oxazolo[3,4-a]pyrazine derivatives
have been first reported in 2005 by Takeda Pharmaceuticals,[9] and SHA-68 (1, Figure ) is the most representative
member of this class.[10] Compound 1 was shown to display nanomolar antagonist potency values
(pA2/pKB)
ranging from 7.28 to 8.16 toward the hNPSR-Asn107 variant
and from 7.55 to 8.03 toward the hNPSR-Ile107 variant.
Also, compound 1 exhibited high affinity for the hNPSR
in radioligand-binding experiments (pKi = 7.32) and high selectivity over several unrelated GPCRs.[7]
Figure 1
SAR extension performed in this work around the oxazolo[3,4-a]pyrazine nucleus of known NPSR antagonists 1 and 2.
SAR extension performed in this work around the oxazolo[3,4-a]pyrazine nucleus of known NPSR antagonists 1 and 2.The in vivo pharmacological profile of 1 has been explored in various
animal models in which considerably
variable effectiveness was observed according to different assays,
which has been interpreted as due to suboptimal pharmacokinetic properties
of the molecule.[10−15] As a first attempt to overcome these limits, Hassler et al. developed
the piperidine derivative RTI-118 (2, Figure ) that exhibited
lower potency (hNPSR-Asn107 pA2 = 6.31; hNPSR-Ile107 Ca2+ pA2 =
6.96) in vitro(16) but a
slightly improved in vivo effectiveness in reducing
cocaine self-administration and seeking behavior in rats; these results
were ascribed to the higher water solubility of the molecule.[17] Nonetheless, there is a generally recognized
need for further optimizing the pharmacological profile and, above
all, the drug-likeness properties of oxazolo[3,4-a]pyrazine ligands to obtain even more potent NPSR antagonist tools
to be employed in vivo in preclinical studies. These
ligands could be extremely useful for understanding the real therapeutic
potential of the NPSergic system. This prompted us to extend the structure–activity
relationship studies in this field investigating new and unexplored
modifications of the bicyclic piperazine nucleus of compounds 1 and 2. Thus, in this work, we describe the
synthesis and the in vitro and in vivo biological evaluation of oxazolo[3,4-a]pyrazine
derivatives resulting from a series of substitutions at the 1-, 5-,
and 7-positions, as summarized in Figure . Moreover, molecular modeling studies were
performed to gain new insights into the structure–activity
relationships observed for the newly discovered NPSR ligands and provide
an updated atomistic model of ligand/NPSR interactions.
Results and Discussion
Chemistry
As depicted in Scheme , N-Fmoc-oxazolo-piperazines
with a general structure 26 were employed as synthetic
precursors to obtain the final compounds 3–12 in
analogy with the approach previously applied for the synthesis of 1 by Okamura et al.[10] Specifically,
intermediates 26 were obtained starting from unsubstituted
piperazine that was first monoalkylated with benzyl-bromide and next
Boc-protected on the second piperazinenitrogen to give compound 23. Subsequently, the desired N-benzyl-protected
oxazolo[3,4-a]pyrazines 25a–j were obtained through an ortho-lithiation reaction, in the presence
of sec-butyllithium (sec-BuLi) as
the base and various symmetric aromatic/aliphatic ketones (24a–j) as electrophiles. The benzyl function was next replaced with an
Fmoc-group by treatment with FmocCl and finally, in order to achieve
compounds 3–11, 26a–i were
reacted with 4-F-benzyl-isocyanate, while compound 12 was obtained from 26j by treatment with benzyl isothiocyanate.
Scheme 1
Synthesis of Final Compounds 3–12
Reagents
and conditions: (i)
BnBr, EtOH, 75 °C, 5 h; (ii) Boc2O, DMAP, tetraethylammonium,
CH2Cl2, rt, 0.5 h; (iii) sec-BuLi, TMEDA, THF, −78 °C, 6 h; (iv) FmocCl, MeCN, 90
°C, 5 h; (v) DBU, THF, rt, 2 h.
Synthesis of Final Compounds 3–12
Reagents
and conditions: (i)
BnBr, EtOH, 75 °C, 5 h; (ii) Boc2O, DMAP, tetraethylammonium,
CH2Cl2, rt, 0.5 h; (iii) sec-BuLi, TMEDA, THF, −78 °C, 6 h; (iv) FmocCl, MeCN, 90
°C, 5 h; (v) DBU, THF, rt, 2 h.The amide
derivatives 13–14 and the guanidine
analogues 15–16 were synthesized according to Scheme starting from 26j that was first deprotected by treatment with 1,8-diazabicyclo[5.4.0]undec-7-ene
(DBU). The N-alkylation of 28 with 2-chloro-N-(4-fluorophenyl)acetamide or 2-chloro-N-(4-fluorobenzyl)acetamide produced the final compounds 13 and 14, respectively. In order to obtain the guanidine
derivatives 15 and 16, we explored two different
synthetic strategies. The first approach, developed in the liquid
phase, involved the reaction of 28 with cyanogen bromide,
giving the key intermediate 29. Then, the addition of
benzylamine or 4-fluorobenzylamine in the presence of p-toluenesulfonic acid (p-TsOH) provided the desired
final products.
Scheme 2
Synthesis of Final Compounds 13–16
Reagents and conditions: (i)
DBU, THF, rt, 2 h; (ii) CH2Cl2–DMF (2:1),
50 °C, 4 h; (iii) HgCl2, CH3CN, 90 °C,
24 h; (iv) BrCN,CH2Cl2, NaHCO3, H2O, 30 min at 0 °C, 24 h rt; (v) benzyl amine or 4-fluoro
benzylamine, p-TsOH, DMSO, 60 °C, 18 h; (vi)
K2CO3, CH3CN, 90 °C, 4 h.
Synthesis of Final Compounds 13–16
Reagents and conditions: (i)
DBU, THF, rt, 2 h; (ii) CH2Cl2–DMF (2:1),
50 °C, 4 h; (iii) HgCl2, CH3CN, 90 °C,
24 h; (iv) BrCN,CH2Cl2, NaHCO3, H2O, 30 min at 0 °C, 24 h rt; (v) benzyl amine or 4-fluoro
benzylamine, p-TsOH, DMSO, 60 °C, 18 h; (vi)
K2CO3, CH3CN, 90 °C, 4 h.The second pathway resulted from the optimization
of a known solid-phase
approach.[18] In this case, a bromomethyl
polymeric resin was functionalized with 1-(4-fluorobenzyl)thiourea
or 1-(benzyl)thiourea affording 30a–b. Subsequently,
the loaded resin was reacted with 28 in the presence
of HgCl2 to produce the desired guanidine derivatives in
good yields.Finally, compounds 17–21 were
prepared as depicted
in Scheme starting
from five different commercially available l-amino acid methyl
esters (32a–e). Specifically, the piperazine-diones 34a–e were obtained in two simple steps involving a
first amino acid acylation with chloroacetyl chloride, followed by
cyclization with benzylamine.[19] The subsequent
reduction with LiAlH4 gave 35a–e that
were then protected with di-tert-butyl dicarbonate
(Boc2O).[20] The resulting orthogonally
protected piperazines36a–e were employed in the
ortho-lithiation reaction to give the benzyl derivatives 37a–e, followed by treatment with FmocCl and the final addition reaction
with 4-fluorobenzyl isocyanate as described above.
Scheme 3
Synthesis of Final
Compounds 17–21
Reagents and conditions:
(i)
NaHCO3, chloroacetyl chloride, toluene, 0 °C to rt,
overnight; (ii) Et3N, benzylamine, dioxane, reflux, 20
h; (iii) LiAlH4, THF, reflux, 3 h; (iv) Boc2O, THF, 0 °C to rt, 1 h; (v) benzophenone, sec-BuLi, TMEDA, THF, −78 °C, 3 h; (vi) FmocCl, MeCN, 90
°C, 5 h and then rt, 18 h; (vii) 4-fluorobenzyl isocyanate, DBU,
THF, rt, 2 h.
Synthesis of Final
Compounds 17–21
Reagents and conditions:
(i)
NaHCO3, chloroacetyl chloride, toluene, 0 °C to rt,
overnight; (ii) Et3N, benzylamine, dioxane, reflux, 20
h; (iii) LiAlH4, THF, reflux, 3 h; (iv) Boc2O, THF, 0 °C to rt, 1 h; (v) benzophenone, sec-BuLi, TMEDA, THF, −78 °C, 3 h; (vi) FmocCl, MeCN, 90
°C, 5 h and then rt, 18 h; (vii) 4-fluorobenzyl isocyanate, DBU,
THF, rt, 2 h.The ortho-lithiation step is
of key importance for the stereochemical
course of the synthetic approach leading to the final compounds 17–21. This reaction takes advantage of the defined
stereochemistry at C-2 of intermediates 36 that is imposed
by the choice of the starting amino acid as demonstrated in the literature
for analogous piperazine systems obtained through the same strategy.[21] The spatial orientation of the substituents
around the asymmetric C-8a, generated in the bicyclic
derivatives 37 during the ortho-lithiation reaction,
was driven by the absolute configuration previously introduced at
C-2. A single diastereoisomer was isolated in all cases in which an
antirelative stereochemistry between the substituent at the C-5 and
the oxazole ring fused at C-8a was expected according to previous
studies[21] and as confirmed by NOE spectroscopy
performed on the reference compound 17 (Figures S1 and S2). In this experiment, the irradiation of
the methyl protons at the 5-position produced an important enhancement
of the signal of the proton at the C-8a position
which is in accordance with a syn relationship. Thus, the absolute
(5S,8aR)-configuration was assigned
to 17 and to the final compounds 18–21, the latter obtained from α-amino acid with even more hindered
side chains. The maintenance of a significant in vitro activity of these derivatives would indirectly confirm that C-8a would assume the absolute configuration of the eutomer
of 1 that has been previously identified following the
separation of its enantiomers.[22] According
to this study, the interaction of 1 with the humanNPSR
would be markedly enantioselective with the R-isomer
showing a pKB value of 8.28 (hNPSR-N107) in calcium mobilization experiments, while the S-enantiomer would display a considerably reduced potency
(pKB < 6).
In Vitro Structure–Activity Relationships
In the calcium
mobilization assay, NPS increased intracellular
calcium levels in a concentration-dependent manner with pEC50 and Emax values of 8.95 and 287 ±
26% over the basal values, respectively. Inhibition response curves
to 1 (0.1 nM to 10 μM), used as an internal reference,
were performed against the stimulatory effect of 10 nM NPS, approximately
corresponding to NPS EC80. As shown in Figure , compound 1 concentration-dependently
inhibited 10 nM NPS stimulatory effects with a pKB value of 8.12. These results agree with previously reported
data.[11] The pharmacological activity of
compounds 3–21 was evaluated under the same experimental
conditions, and the corresponding results are reported in Table .
Figure 2
Calcium mobilization
assay performed on HEK293mNPSR cells.
Concentration–response curve to NPS [panel (A)] and inhibition–response
curves to 1 against the stimulatory effect of 10 nM NPS
[panel (B)]. Data are mean ± SEM of at least five separate experiments
made in duplicate.
Table 1
In Vitro Pharmacological
Activity of Compounds 1, 3–21 as
NPSR Antagonistsa
Calcium
mobilization assay performed
in HEK293mNPSR cells. Data are the mean of five separate
experiments made in duplicate.
Calcium mobilization
assay performed on HEK293mNPSR cells.
Concentration–response curve to NPS [panel (A)] and inhibition–response
curves to 1 against the stimulatory effect of 10 nM NPS
[panel (B)]. Data are mean ± SEM of at least five separate experiments
made in duplicate.Calcium
mobilization assay performed
in HEK293mNPSR cells. Data are the mean of five separate
experiments made in duplicate.None of the novel compounds stimulated calcium mobilization up
to 10 μM. On the other hand, the substitution or the replacement
of the 1,1-diphenyl moiety of 1, such as in compounds 3–11, significantly affected the antagonist potency
although to a different extent. In particular, the para-substitution
of both the aromatic rings at the 1-position was slightly tolerated
only in the case of the 4-fluoro derivative 3 which was
only 3-fold less potent than 1, while a progressive reduction
of potency was observed by increasing the steric hindrance of the
para-substituents. For example, the bulky dimethylamino groups of
compound 6 resulted in a complete loss of activity (KB > 10 μM). These data suggest that
the
phenyl rings at the 1-position could occupy the NPSR binding pocket
in a region with highly stringent steric requirements.Our results
indicated that also a proper spatial orientation of
the geminal phenyl groups relative to the oxazolidinone ring seems
to be of particular importance to promote activity. Indeed, the ortho-substitution
of the 1,1-diaryl moiety such as in compounds 7 and 8 should induce a conformational distortion with respect to
the nonsubstituted 1, determining a marked or total loss
of potency. This observation was further confirmed by the inactivity
of compound 9 in which the 1,1-phenyl rings were forced
into a coplanar arrangement due to their inclusion in the spiro-fluorene
fusion.The aromaticity of 1,1-substituents seems to be important
as well
since the 1,1-dicyclohexyl derivative 10 was more than
50-fold less potent than 1. Even more unfavorable was
the replacement of the diaryl template with linear propyl chains (compound 11, KB > 10 μM).In compounds 12–16, we explored the effect
of a few modifications at the 7-position of the oxazolo[3,4-a]pyrazine core that has not been explored before.[23] In particular, we introduced side chains containing
thiourea (12), N-substituted acetamide (13, 14) and guanidine (15, 16) functions. In this subset of molecules, compounds 13–16 have been specifically designed to modulate the hydrophilic/lipophilic
balance of 1, which might be important for its in vivo effectiveness as suggested in different studies.[10,11] In particular, it has been demonstrated that 1, at
the high dose of 50 mg/kg, can only partially counteract NPS effects,
with different levels of efficacy, depending on the assay used.[10−15] These findings have been hypothetically attributed to suboptimal
physicochemical properties of the compound, in particular, its high
lipophilicity.[7] Thus, in a first attempt
to overcome these limits, the acetamide derivatives 13 and 14 have been synthesized as possible bioisosteres
of 1 in which a methylene spacer was interposed between
the piperazinenitrogen and the carbonyl function of the 7-side chain.
The modification was conceived to increase the basicity of the piperazinenitrogen thus opening the possibility to obtain hydrochloride salts
with improved water solubility. Nonetheless, the compounds were shown
to display very low (14, pKB = 6.33) or null activity (13, KB > 10 μM) in the calcium mobilization assay. However,
the partial recovery of activity of compound 14, in which
the 4-F phenyl moiety is not directly linked to the amidenitrogen,
suggested that also the conformational freedom of this pharmacophoric
portion may be important for the interaction with NPSR.In compounds 15 and 16, we replaced the
urea moiety of 1 with a guanidine function as an alternative
strategy to obtain NPSR antagonists with improved hydrophilicity.
The fluorinated derivative 16 was conceived as a close
analogue of 1 bearing an NH-group in place of the ureaoxygen atom. This modification does not interfere with the ability
of the ligand to establish polar interactions with the receptor. Of
note, compounds incorporating a guanidine moiety have aroused an increasing
interest for their potential in the development of novel drugs due
to the ability of the guanidinium group to form strong noncovalent
interactions and to provide obvious advantages in terms of hydrophilicity.[24−26] Compound 16 displayed low nanomolar potency in antagonizing
the stimulatory activity of NPS with a pKB value of 7.38. The nonfluorinated guanidine derivative 15 was slightly less potent (pKB = 7.21)
indicating some importance of the fluorine atom at the para position
of the terminal benzyl moiety.Finally, we introduced different
substitutions at the 5-position
of the oxazolo[3,4-a]pyrazine core whose effect on
NPSR modulation has not been explored before. To this aim, we developed
a highly accessible diastereoselective synthesis that provided compounds 17–21 in which the 5-position was functionalized with
the side chains of a series of l-amino acids employed as
the starting material. The introduction of a −CH3 (compound 17) or an isopropyl chain (compound 18) determined about a 10-fold reduction of potency if compared
to 1. Even more detrimental was the introduction of a
bulkier branched alkyl moiety (19) or a benzyl group
(20). In contrast, the l-phenylglycine derivative 21 showed a recovery in activity becoming the most active
compound of the newly reported series.These data indicated
that the 5-position tolerates substitutions
with hydrophobic chemical groups of different size generating derivatives
with 2 (21) to 30 (compound 20)—fold
reduction of potency. Intriguingly, in the latter compounds, a subtle
chemical modification such as the introduction of a methylene spacer
between C5 and the phenyl ring promoted a consistent reduction of
bioactivity. Given these data, we cannot exclude that the C5 phenyl
ring of 21 might be recognized by a previously unexplored
region of the NPSR binding pocket.To confirm and better define
the antagonist properties of compounds 16 and 21, the concentration–response
curve of NPS has been reassessed in the absence and presence of 100
nM of 1, 16, and 21 (Figure ). 1, 16, and 21 shifted the concentration–response
curve of NPS to the right without changing its maximal effects. The
following pA2 values have been derived
from these experiments: 7.82 (7.40–8.24) for 1; 7.10 (6.65–7.55) for compound 16; and 7.59
(7.08–8.10) for compound 21. Thus, the rank order
of potency of these NPSR antagonists is 1 > 21 > 16. These results are superimposable
to those obtained
in inhibition experiments.
Figure 3
Calcium mobilization assay performed on HEK293mNPSR cells.
Concentration–response curve to NPS in the absence and in the
presence of 100 nM of 1, 16, and 21. Data are mean ± SEM of at least five separate experiments
made in duplicate.
Calcium mobilization assay performed on HEK293mNPSR cells.
Concentration–response curve to NPS in the absence and in the
presence of 100 nM of 1, 16, and 21. Data are mean ± SEM of at least five separate experiments
made in duplicate.
Molecular Modeling Studies
To gain major insights into
the reasons for the structure–activity relationships (SARs)
observed for the newly discovered NPSR ligands, molecular modeling
studies were attempted. So far, the three-dimensional (3D) structure
of the target receptor has not been determined so that homology-modeling
techniques had to be employed to first construct a viable model of
NPSR in its inactive state. The available SAR data were obtained from
humanHEK293 cells expressing the mouseNPSR (mNPSR). On the other
hand, the sequence alignment between the human (hNPSR) and mNPSR revealed
that the two proteins share 89.22% sequence identity with all the
differences residing in the N-terminal region distant from the putative
ligand-binding site. Thus, considering the high structural homology
of NPSR across the two species, we decided to model the pharmacologically
relevant hNPSR in its I107 variant in the present work. In 2010, Dal
Ben et al.[27] published the first model
of the two NPSR variants, namely, NPSR-N107 and -I107. In this seminal
work, the authors modeled the NPSR receptors starting from the X-ray
crystal structure of bovinerhodopsin.[28] The choice of using this latter structure as a template was dictated
by preliminary modeling studies indicating that the NPSR extracellular
loop 2 (ECL2) had a propensity to adopt a β-sheet conformation
which was partially present in the bovinerhodopsin structure rather
than in the structures of the human β1, β2 adrenergic
and adenosineA2A receptors that were available at that
time. Since then, more than 340 structures of GPCR have been deposited
in the protein data bank (PDB) thereby allowing for a re-evaluation
of the optimal template to employ in the in silico construction of
the NPSR receptor variants. Thus, the hNPSR sequence (Uniprot entry
code Q6W5P4) was used to interrogate the PDB and select the solved
X-ray structures sharing the highest homology with the target structures.
In this analysis, we decided to retain all the structures that shared
with NPSR a sequence identity higher than 20%, a sequence coverage
higher than 70%, and that were crystallized in their inactive states
(i.e., bound to an antagonist ligand). These criteria allowed selecting
7 humanGPCR structures in which, interestingly, 6 of them turned
out to be receptors for endogenous peptides and 5 of these feature
a twisted β-hairpin in the ECL2 region (Table S1). Indeed, the β-hairpin motif is usually found
in the ECL2 of peptide-activated GPCRs such as Neuropeptide Y Y1 receptor,[29] orexin receptor type 1 (OX1R),[30] chemokine receptor type 4 (CXCR4),[31] delta,[32] and kappa[33] opioid receptors, protease-activated receptor 1 (PAR1),[34] neurotensin receptor 1 (NTSR1),[35] endothelin ETB receptor,[36] and
angiotensin receptors AT1[37] and AT2.[38] Unfortunately, for one of the selected 7 templates
(Table S1), the humanCC chemokine receptor
type 9 (CCR9), ECL2 was unresolved; thus, this template structure
was not considered further. Subsequently, the primary sequences of
the remaining six GPCRs were all pairwise aligned to the one of the
hNPSR-I107 variant, and the phylogenetic tree was calculated (see Figures S3–S9). Then, these templates
were all used to construct six models of hNPSR, one for each template,
using the Prime software within the Schrodinger’s Maestro suite.
The constructed models were all used to perform docking calculations
of all the newly identified analogues employing the Glide program.
The results of these simulations were then analyzed in light of the
available SAR data. In this step, we first verified whether Glide
was able to find a viable binding pose for each active compound reported
in Table (namely, 1, 3–5, 7, 10, 12, and 14–21) in each of the
NPSR models constructed employing the aforementioned 6 template structures.
This first analysis was instrumental for the selection of the best
model structure that was able to host all the newly discovered NPSR
ligands. In particular, the NPSR model constructed starting from the
human neuropeptide Y Y1 receptor (hNPY1R, PDB code 5ZBH)[29] was the only one able to fulfill the above-mentioned selection
criteria. Subsequently, we decided to analyze the docking results
achieved for the most potent antagonists 16 and 21 as well as the control compound 1. Interestingly,
for all three ligands, Glide was able to suggest two possible binding
poses (i.e. featuring comparable docking scores) in which the ligand
pendant benzyl substituent is alternatively pointing downward [inside
the transmembrane (TM) bundle] or upward (toward the NPSR extracellular
region). In this work, the two alternative docked positions will be
referred to as binding mode 1 (BM1) and 2 (BM2), respectively (Figure ).
Figure 4
Energy-minimized docked
poses of compound 1 in BM1
and BM2 (panels A and B, respectively) in the model of NPSR constructed
starting from the human neuropeptide Y Y1 receptor (hNPY1R, PDB code 5ZBH).[29]1 and the protein are represented as violet
sticks and multicolored ribbons, respectively.
Energy-minimized docked
poses of compound 1 in BM1
and BM2 (panels A and B, respectively) in the model of NPSR constructed
starting from the human neuropeptide Y Y1 receptor (hNPY1R, PDB code 5ZBH).[29]1 and the protein are represented as violet
sticks and multicolored ribbons, respectively.In BM1, the 1,1-diphenyl moiety of 1 is pointing toward
TM7 and TM2 and establishes several π–π interactions
with aromatic residues present in the outer region of NPSR (W108,
W198, F273, and F289) (Figures A and 5A). The limited extension of
the cleft lodging this moiety should explain why incrementing its
steric hindrance has a detrimental effect on the antagonist potency
of its analogues (compounds 3–8). Moreover, the
nature of the established ligand–protein interactions (charge
transfer contacts) as well as the relative position of the two phenyl
rings (not coplanar) explains why compounds 9–11 are less active or devoid of an evident antagonist potency. The 1 bicyclic piperazine core orients its pendant benzylamide
chain inside the TM bundle between TM3 and TM6 where its NH forms
a charged-reinforced H-bond with D274 and the terminal fluorophenyl
ring is lodged in a well-defined lipophilic gorge establishing π–π
interactions, reinforced by the electron-withdrawing effect of the
fluorine atom, with W267, Y270, F271, and van der Waal contacts with
L132 (Figure A). The
tight interactions established by the benzylamide chain in this receptor
region would explain why modifications in this position result in
a reduction (compound 14) or abrogation (compound 13) of the antagonist potency thereby indicating a limited
tolerance for 7-position substituents.[23]
Figure 5
Energy-minimized
docked poses of compound 1 [panel
(A) for BM1 and (D) for BM2], 16 [panel (B) for BM1 and
(E) for BM2], and 21 [panel (C) for BM1 and (F) for BM2]
in the model of NPSR constructed starting from the human neuropeptide
Y Y1 receptor (hNPY1R, PDB code 5ZBH).[29]1, 16, and 21 are represented as
violet, orange, and red sticks, respectively. The protein is represented
as cyan ribbons and sticks. H-bonds are represented as dashed yellow
lines.
Energy-minimized
docked poses of compound 1 [panel
(A) for BM1 and (D) for BM2], 16 [panel (B) for BM1 and
(E) for BM2], and 21 [panel (C) for BM1 and (F) for BM2]
in the model of NPSR constructed starting from the human neuropeptide
Y Y1 receptor (hNPY1R, PDB code 5ZBH).[29]1, 16, and 21 are represented as
violet, orange, and red sticks, respectively. The protein is represented
as cyan ribbons and sticks. H-bonds are represented as dashed yellow
lines.Almost the same binding orientation
was also found for 16 with all the aforementioned interactions
being maintained (Figure B). Here, the protonated
guanidine group is now engaged in an ionic interaction with the D274
side chain. 21 is also predicted to adopt BM1 where the
presence of the additional phenyl ring on the core structure allows
to form supplementary charge-transfer and lipophilic interactions
with F185, W198, A199, and W207 (Figure C). These contacts should explain why the
substitution of the phenyl ring with alkyl or benzyl chains (compounds 17–20) still allows observing an antagonist activity
at NPSR, although resulting in less potent antagonists if compared
to 21.In BM2, the 1,1-diphenyl moiety of 1 is predicted
to point toward the inner part of NPSR making contact with TM2, TM6,
and TM7 (Figure B).
In this position, this moiety established π–π and
cation−π contacts with W108 and R109, respectively (Figure D). The carbonyl
oxygen of the bicyclic piperazine scaffold accepts an H-bond from
the A199 backbone NH while the pendant fluorobenzylamide chain laced
beneath ECL3 (Figure B) in a cleft lined by D284, T285, R288, F289, and S292. Almost the
same interaction pattern is also predicted for 16 in
BM2 (Figure E), and
the presence of the protonated guanidine group allows the ligands
to establish charge-reinforced H-bonds with N280 and P283 backbone
COs as well as a cation−π interaction with W198. 21 recapitulates the same interactions established by 1 in BM2 (Figure F) and takes advantage of its additional phenyl ring in position
7 of the core scaffold to establish a π–π interaction
with F185. Also in this case, BM2 would match the SAR data acquired
in this manuscript as well as the ones already present in the literature.[23]The binary complexes calculated using
the docking program for 1, 16, and 21 in both BM1 and BM2
were then subjected to 100 ns molecular dynamics (MD) simulations
with Desmond[39] to refine the predicted
binding geometries. Most importantly, given the dichotomy of binding
modes predicted for these ligands, results of MD simulations could
suggest a preferential binding orientation for each ligand. Analysis
of the 6 MD simulations was attained through the Desmond SID tool
which allowed us to analyze the ligand–receptor interactions
during the MD trajectory. Attention was given to the ligand root-mean-square
fluctuation (RMSF) (Figures S10–S15) and the stability of the ligand–residue interactions (Figures S16–S27).The ligand RMSF
was useful for characterizing changes in the ligand
atom positions during the MD. Analysis of this parameter demonstrated
that in BM1 and BM2, the three ligands display a different degree
of flexibility. In particular, regardless of the adopted binding mode,
the pendant fluorobenzylamide chain is the most flexible part of the
molecule with BM1 being more stable than BM2 (see Figures S10–S15). In the latter binding position, the
fluorobenzylamide moiety rapidly loses its interactions with the residues
belonging to ECL3 to point toward the external part of the receptor
without taking stable contacts with any NPSR residue. This is further
outlined by plotting the most frequent (>30%) ligand–protein
contacts (Figures S16–S27) showing
that for the three ligands in BM2, the benzyl chain is always solvent
exposed. On the contrary, in BM1, the same portion remains anchored
to the receptor although experiencing a partial relocation during
the MD. All in all, MD results would suggest that only in BM1, the
fluorobenzylamide chain would play a role in the ligand–receptor
recognition as underscored by the experimental SAR data[23] available for this ligand moiety, thereby suggesting
that 1, 16, and 21 should adopt
this binding orientation. Plotting of the ligand–receptor interactions
also shows that while binding of 1 is mainly governed
by hydrophobic and charge-transfer contacts with the receptor, 16 is stably anchored to NPSR also through the polar interaction
with D274 while 21 finds additional contacts with W207
and F185.
In Vivo Characterization in the Mouse Locomotor
Activity Test
Among the synthesized molecules, 16 and 21 have been selected to be tested in vivo in the mouse locomotor activity assay. As shown in Figure and in line with previous
findings,[11,40−42] NPS injected by the
i.c.v. route at 0.1 nmol concentration was able to stimulate mouse
locomotor activity by increasing the distance traveled (panel A) and
the number of rearings (panel C) and reducing the immobility time
(panel B) with statistically significant effects. 1 (50
mg/kg) did not significantly modify mice locomotor activity and only
partially counteracted NPS-induced stimulant effects, confirming previous
studies.[10,11] All mice treated with 50 mg/kg of 16 displayed an important reduction of locomotor activity
(data not shown); thus, in the present study, the 10 mg/kg dose was
used for this compound. The doses of 10 mg/kg for 16 and
50 mg/kg for 21 did not elicit by themselves statistically
significant effects on mouse locomotor activity (Figure ).
Figure 6
Effects of NPS, 1, 16, and 21 on mouse locomotor
activity. The cumulative effects exerted on distance
traveled are shown in panel (A), while the total time immobile and
number of rearings over the 30 min observation period are shown in
panels (B,C), respectively. Data are mean ± SEM of four separate
experiments (vehicle + saline, 12 mice; 1 + saline, 8
mice; 16 + saline, 8 mice; 21 + saline,
7 mice; vehicle + NPS, 11 mice, 1 + NPS, 8 mice; 16 + NPS, 7 mice; and 21 + NPS, 7 mice). The
two-way ANOVA NPS × antagonist revealed for the total distance
traveled, an effect of NPS F(1,59) = 6.63 and of
the interaction NPS × antagonist F(3,59) = 4.48;
for the immobility time, an effect of NPS F(1,59)
= 8.09 and of the interaction NPS × antagonist F(3,59) = 4.26; for the number of rearings, an effect of NPS F(1,59) = 7.35, of antagonist F(3,59) =
10.65, and of the interaction NPS × antagonist F(3,59) = 4.32. *p < 0.05 vs saline, #p < 0.05 vs vehicle according to Bonferroni’s
test for multiple comparisons.
Effects of NPS, 1, 16, and 21 on mouse locomotor
activity. The cumulative effects exerted on distance
traveled are shown in panel (A), while the total time immobile and
number of rearings over the 30 min observation period are shown in
panels (B,C), respectively. Data are mean ± SEM of four separate
experiments (vehicle + saline, 12 mice; 1 + saline, 8
mice; 16 + saline, 8 mice; 21 + saline,
7 mice; vehicle + NPS, 11 mice, 1 + NPS, 8 mice; 16 + NPS, 7 mice; and 21 + NPS, 7 mice). The
two-way ANOVA NPS × antagonist revealed for the total distance
traveled, an effect of NPS F(1,59) = 6.63 and of
the interaction NPS × antagonist F(3,59) = 4.48;
for the immobility time, an effect of NPS F(1,59)
= 8.09 and of the interaction NPS × antagonist F(3,59) = 4.26; for the number of rearings, an effect of NPS F(1,59) = 7.35, of antagonist F(3,59) =
10.65, and of the interaction NPS × antagonist F(3,59) = 4.32. *p < 0.05 vs saline, #p < 0.05 vs vehicle according to Bonferroni’s
test for multiple comparisons.This result is in line with previous studies performed with different
NPSR antagonists[10,11,41] and with the lack of locomotor phenotype of NPSR knockout mice,[43−45] collectively suggesting that the endogenous NPS does not control
mouse locomotor activity in the open field. When administered 30 min
before NPS, both 16 and 21 were able to
completely block the stimulant effects of the peptide. Compound 21 was in vivo slightly more effective than 1 when injected intraperitoneally (ip) at the same dose (50
mg/kg). Compound 16, despite its lower potency in vitro, was at least 5-fold more potent than 1in vivo. It must be remarked that compound 2 (see the Introduction section) was
shown to exert a significant blockade of NPS-induced locomotor activity
at an i.p. dose of 50 mg/kg, thus being equipotent to the parental
compound 1.[16] Of note, the
ability of both 16 and 21 to completely
block the stimulant NPS effects confirms previous studies demonstrating
that the stimulant NPS effects are selectively due to the activation
of the NPSR receptor (see Tables 1 and 2 of the recent review by Ruzza
et al.).[7] It is worth noting that at the
higher doses tested, compound 1 was able to only partially
block NPS effects. This may somewhat limit the usefulness of compound 1 as a research tool to explore the biological actions modulated
by the endogenous NPS. Both compounds 16 and 21 seem to overcome this limitation; in fact, at doses statistically
inactive, they were able to fully block NPS stimulant effects. This
feature, that is, complete NPSR occupation and blockage, makes these
compounds essential pharmacological tools for investigating those
conditions in which the endogenous NPS/NPSR signaling is activated,
for example, cocaine/alcohol seeking and relapse.[46−48] On the other
hand, in proposing compound 16 as an innovative research
tool, we should also underline that selectivity concerns must be taken
into account. In fact, at the high dose of 50 mg/kg, 16 strongly reduces mouse locomotor activity. This is most probably
an off-target effect since NPSR selective peptide antagonists did
not modify mouse locomotion at doses able to completely block NPS
stimulatory effects,[41] and NPSR(−/−)
mice did not display a locomotor phenotype in this assay.[43] Thus, the high potency and antagonist effectiveness
of compound 16 is associated with somewhat limited selectivity
of action. Further studies, for example, CEREP receptogram, are needed
to eventually identify the mechanisms involved in the putative off-target
effects of compound 16 at high doses. Based on these
considerations, we certainly recommend the use of compound 16 but with special caution in selecting the range of doses to be used
for NPSR physiopathological investigations. The in vivo action of 21 reflects the in vitro potency of the compound that was very similar to that of the reference
tool 1. On the other hand, the improved in vivo antagonist effectiveness of 16, deriving from the bioisosteric
replacement of the urea function of 1 with a guanidine
moiety, may be attributed to its relatively higher hydrophilic properties
that could compensate for the slight loss of in vitro potency. Lipophilicity is indeed considered one of the most important
physicochemical properties to be addressed for drug design purposes.
Typically, highly hydrophilic compounds suffer from poor membrane
permeability and faster renal clearance. On the other hand, water
solubility and metabolism are more likely to be compromised at high
lipophilicity values.[49] Noteworthily, the
optimum region of lipophilicity for candidate drug molecules has been
generally suggested to lie within a narrow range of log D7.4 that has been approximately determined between 1 and
3.[49] Thus, the improved in vivo potency of 16 (Clog P = 3.43 ±
0.89, ACDLabs; Clog D7.4 = 1.87, ChemAxon
predictor) with respect to the known NPSR antagonist 1 (Clog P = 4.32 ± 0.86, ACDLabs; Clog D7.4 = 4.05, ChemAxon predictor) can be reasonably
justified in light of the optimization of lipophilicity parameters
that are closer to the recommended values. This could also account
for the lower in vivo potency of compound 21 (Clog P = 6.18 ± 0.87, ACDLabs; Clog D7.4 = 5.83, ChemAxon predictor) despite its
higher in vitro activity. However, we would like
to underline the fact that these hypotheses are only supported by
theoretical calculations; firm conclusions on this issue can be drawn
only after performing experiments investigating the pharmacokinetic
profile of compound 16 eventually in comparison with
compounds 1 and 21.
Conclusions
The therapeutic potential of selective NPSR ligands in psychiatric
disorders is supported by a series of preclinical studies.[7] In particular, NPSR antagonists were shown to
reduce cocaine/alcohol seeking and relapse in animal models, and this
makes them potentially useful for treating drug addiction. Patients
suffering from such a condition typically show resistance to the very
few treatments currently available. Thus, the identification of innovative
drugs able to ameliorate these conditions, which are essentially untreated,
is still urgent. Additionally, it can be speculated that NPSR antagonists
may be useful also for the treatment of other types of drug abuse,
although no preclinical data are currently available. NPSR antagonists
may open new perspectives for addressing this unmet clinical need.
Moreover, NPSR antagonists still represent important research tools
to investigate the neurobiology of the NPS/NPSR system and to study
those biological functions for which the NPSergic tone is important.
However, even though in vitro potent NPSR antagonists
have been reported, NPSR antagonists with a good in vivo pharmacological profile are still missing. This is probably due
to pharmacokinetic issues and represents an undeniable limit for preclinical
studies on the NPSergic system aimed at translating results from basic
pharmacology into clinical utility. In the present study, the in vitro pharmacological activity of a new series of oxazolo[3,4-a]pyrazines as NPSR antagonists was investigated and a molecular-modeling
study helped to rationalize the resulting SARs. The most promising
compounds (16 and 21) in terms of in vitro potency and/or drug-likeness properties have been
also evaluated in vivo for their capability to counteract
NPS-induced stimulant effects on mouse locomotor activity. Our findings
demonstrated that these compounds behave in vitro as pure NPSR antagonists with nanomolar potency in inhibiting the
NPS stimulatory effects in the calcium mobilization assay (pKB values of 7.38 and 7.82 for 16 and 21, respectively). Importantly, the guanidine derivative 16 exhibited a significantly (5-fold) improved potency and
increased antagonist effectiveness in vivo when compared
to the reference compound 1, although this is associated
with somewhat reduced selectivity of action. Collectively, our efforts
can be considered an important advancement in this research field
culminating in the identification of a new pharmacological tool that
combines in vitro and in vivo potency
in blocking NPSR. This could be useful to investigate possible pharmacological
treatments in all the pathological conditions in which the endogenous
NPSergic system is activated.
Experimental Section
Materials
and Methods
The chemicals, including the
2-(4-bromomethyl-phenoxy)ethyl polystyrene HL resin for solid-phase
synthesis, were purchased from Fluorochem, Novabiochem Iris Biotech
GmbH, or Sigma-Aldrich. Reaction progress and product mixtures were
monitored by thin-layer chromatography (TLC) on silica gel (precoated
F254 Macherey-Nagel plates) and visualized with a UV lamp (254 nm
light source). Compounds were purified through silica gel flash chromatography
(silica gel 60, 40–63 μm) using appropriate eluent mixtures
or on a reverse-phase Waters Prep 600 HPLC system equipped with a
Jupiter column C18 (250 × 30 mm, 300 Å, 15 μm spherical
particle size). Reverse-phase purification of crude compounds was
carried out using a gradient of CH3CN/H2O [with
0.1% trifluoroacetyl (TFA)] programed time by time, with a flow rate
of 20 mL/min and a UV detector with a wavelength of 220 nm. Analytical
HPLC analyses were performed on a Beckman 116 liquid chromatograph
equipped with a Beckman 166 diode array detector. Analytical purity
of the final compounds were assessed using a XBridge C18 column (4.6
× 150 mm, 5 μm particle size) at a flow rate of 0.7 mL/min
with a linear gradient from 100% of solvent A (H2O + 0.1%
TFA) to 100% of solvent B (CH3CN + 0.1% TFA) over 25 min.
Analytical determinations were reported as column retention time (TR) in minutes, and the purity of final compounds
was >95% as determined by HPLC analysis carried out at a wavelength
of 220 nm. Mass spectra were recorded with a Waters ESI Micromass
ZQ dissolving the samples in a solution of H2O/CH3CN/TFA (40:60:0.1). Melting points for purified products 3–21 were determined using glass capillaries on a Stuart Scientific electrothermal
apparatus SMP3 and are uncorrected. NMR analyses were performed in
CDCl3 or DMSO-d6 at ambient
temperature using a Varian 200 or 400 MHz spectrometer. Chemical shifts
(δ) are reported in parts per million (ppm) using the peak of
tetramethylsilane as an internal standard in deuterated solvents,
and coupling constants (J) are reported in Hertz.
Splitting patterns are designed as s, singlet; d, doublet; t, triplet;
q, quartet; m, multiplet; and b, broad. Optical rotations were measured
on a Jasco P-2000 polarimeter dissolving the samples in methanol,
with a path length of 1 dm, using sodium D line, 589 nm.
General Procedure
for the Synthesis of 25a–j
Tetramethylethylenediamine
(TMEDA, 2.7 mmol) was added
under an argon atmosphere at room temperature to a stirring solution
of 23(10) (1 mmol) in freshly
distilled tetrahydrofuran (THF) (5 mL). After cooling at −78
°C, sec-BuLi (2.7 mmol) was added and the reaction
was allowed to reach −30 °C over 2 h. A solution of appropriate
benzophenone 24a–j (2 mmol) in THF (5 mL) was
added and the reaction solution was left stirring at −30 °C
for 30 min, then it was slowly warmed to rt and stirred for 16 h.
The reaction was quenched with a saturated solution of NH4Cl (15 mL), and the solvents were concentrated under vacuum to half
volume and the aqueous phase was extracted with EtOAc (3 × 15
mL). The organic layers were combined, dried over Na2SO4, and the solvent was removed under vacuum. The resulting
crude product was purified by flash column chromatography on silica
gel using a mixture of EtOAc/PEt 1:2 as the eluent.
9-Fluorenylmethoxycarbonyl chloride (FmocCl, 1.1 mmol)
was added to a solution of 25a–j (1 mmol) in CH3CN (5 mL). The reaction solution was heated at 90 °C
for 5 h and then stirred at room temperature for 18 h. The solvent
was evaporated giving a residue that was dissolved in EtOAc (30 mL),
and the resulting organic phase was washed with water (3 × 15
mL), dried over anhydrous Na2SO4, and concentrated
to dryness. All crude residues were finally purified via flash column
chromatography on a silica gel using the appropriate mixture of petroleum
ether and EtOAc as an eluent (see below).
General Procedure for the Synthesis of Final Compounds 3–12
The appropriate iso(thio)cyanate 27a–b (2 mmol) and DBU (1.2 mmol) were sequentially
added at room temperature to a stirring solution of 26a–j (1 mmol) in anhydrous THF (5 mL). The reaction solution was continued
stirring for 2 h after which it was quenched with a saturated solution
of NH4Cl (10 mL). The mixture was concentrated under vacuum
to half volume, and the aqueous phase was extracted with EtOAc (3
× 15 mL). The organic layers were combined, dried over Na2SO4, and concentrated under vacuum. The crude products
were purified by preparative RP-HPLC.
DBU (1.2 mmol)
was added to a stirring solution of 26j (1 mmol) in anhydrous
THF (10 mL). The reaction solution was allowed stirring at room temperature
for 18 h and monitored by TLC. The solvent was removed under vacuum,
and the residue was dissolved in EtOAc (20 mL) and washed with water
(20 mL). The organic phase was separated, dried over Na2SO4, and the solvent was evaporated to give a residue
that was purified via flash column chromatography on silica gel using
a 1:3 mixture of petroleum ether and EtOAc as the eluent.Off-white
solid (256 mg, 87% yield). 1H NMR (200 MHz, DMSO-d6): δ (ppm) 7.52 (d, J = 7.2 Hz, 2H); 7.41–7.24 (m, 8H); 4.32 (d, J = 10.1 Hz, 1H); 4.10–3.71 (m, 2H); 3.16 (t, J = 5.6 Hz, 1H); 2.97–2.76 (m, 1H); 2.28 (td, J = 11.6, 3.4 Hz, 1H); 1.83–1.75 (m, 1H), 1.21 (br s, 1H).
MS (ESI): m/z calcd for C18H19N2O2 [M + H]+, 295.36;
found, 295.38.
General Procedure for the Synthesis of Final
Compounds 13 and 14
2-Chloro-N-(4-fluorobenzyl)acetamide or 2-chloro-N-(4-fluorophenyl)acetamide
(1 mmol) was added to a mixture of compound 28 (1 mmol)
and K2CO3 (1.5 mmol) in CH3CN (15
mL). The reaction mixture was heated at 90 °C for 4 h after which
the solvent was removed under vacuum, and the residue was partitioned
between water (15 mL) and CH2Cl2 (15 mL). The
aqueous phase was further extracted with CH2Cl2 (2 × 15 mL), and the combined organic layers were washed with
brine (10 mL) and dried over Na2SO4. After evaporation,
the residue was purified by flash column chromatography on silica
gel using a 1:1 mixture of EtOAc/PEt as the eluent.
Method A for the Synthesis
of Guanidine Derivatives 15 and 16
A solution of NaHCO3 (1.7
mmol) in H2O (1 mL) was added at 0 °C to a stirring
solution of 28 (1 mmol) in CH2Cl2 (5 mL). At the same temperature, a solution of cyanogen bromide
(1.2 mmol) in CH2Cl2 (5 mL) was added. The heterogeneous
mixture was vigorously stirred at 0 °C for 30 min, then warmed
to room temperature, and stirred for further 24 h. After this time,
the layers were separated and the organic phase was washed with a
saturated solution of NaHCO3 (2 × 10 mL), dried with
anhydrous Na2SO4, and concentrated under vacuum
to give a residue from which compound 29 was purified
via flash column chromatography on silica gel using a 1:1 mixture
of petroleum ether and EtOAc as an eluent.White solid (160
mg, 50% yield). 1H NMR (200 MHz, DMSO-d6): δ (ppm) 7.52–7.25 (m, 10H), 4.60 (dd, J = 11.2, 3.6 Hz, 1H), 3.98–3.88 (m, 1H), 3.37–3.22
(m, 2H), 3.10–3.02 (m, 2H), 1.27–1.18 (m, 1H); MS (ESI): m/z calcd for C19H18N3O2 [M + H]+, 320.37; found, 320.40.Benzylamine or 4-fluorobenzylamine (3 mmol) was added to a stirring
solution of 29 (1 mmol) in DMSO (3 mL) in the presence
of a catalytic amount of p-toluenesulfonic acid.
After 18 h of stirring at 60 °C, the reaction solution was diluted
with water (10 mL) and extracted with EtOAc (3 × 15 mL). The
organic layers were dried over Na2SO4 and concentrated in vacuum after which crude products were purified by flash
column chromatography on silica gel using a 4:1 mixture of CH2Cl2 and MeOH.
Method B for the Synthesis
of Guanidine Derivatives 15 and 16
Compounds 15 and 16 were alternatively
synthetized according to a manual solid-phase
synthesis approach described previously.[18] Briefly, compounds 31a or 31b (0.62 mmol)
was added to a suspension of 2-(4-bromomethyl-phenoxy)ethyl polystyrene
HL resin (substitution: 1.23 mmol/g, 0.62 mmol) in a 2:1 mixture of
CH2Cl2/DMF (3 mL). The mixture was heated at
50 °C until starting material consumption was observed (4 h).
After that, each of the two differently functionalized resins was
filtered, washed with DMF (2 × 5 mL) and CH2Cl2 (2 × 5 mL), and dried. Subsequently, the respective
functionalized resin (0.62 mmol) was suspended in CH3CN
(2 mL) before adding compound 28 (1.55 mmol) and HgCl2 (0.93 mmol). After heating at 90 °C for 24 h, a simple
filtration was performed and the filtrates were purified by flash
column chromatography on silica gel using a 4:1 mixture of CH2Cl2 and MeOH.
Enantiomerically pure monosubstituted
piperazines 35a–e were prepared starting from
commercial chiral
amino-esters following previously reported procedures, and the analytical
data for intermediates 33–35a–e are in
agreement with data from the literature.[19,50] A solution of di-tert-butyl dicarbonate (1.1 mmol)
in anhydrous THF was added at 0 °C to a solution of compounds 35a–e (1 mmol) in dry THF (10 mL). The reaction mixture
was warmed at room temperature and further stirred for 1 h. The solvent
was removed in vacuo, and the residue was dissolved in CH2Cl2 (15 mL) and washed with water (3 × 10 mL) and
brine (1 × 10 mL). The organic phase was dried over anhydrous
Na2SO4 and concentrated to dryness. Flash column
chromatography on silica gel using a 0.5:9.5 mixture of EtOAc/PEt
as the eluent provided the desired compounds with good yields.
TMEDA (2.7
mmol) was added under an argon atmosphere at
room temperature to a stirring solution of 36a-e (1 mmol)
in freshly distilled THF (5 mL). After cooling at −78 °C, sec-BuLi (2.7 mmol) was added, and the reaction solution
was allowed to reach −30 °C over 2 h. A solution of benzophenone
(2 mmol) in THF (5 mL) was added, and the reaction mixture was left
stirring at −30 °C for 30 min, then slowly warmed to room
temperature, and stirred for 18 h. The reaction was quenched with
a saturated solution of NH4Cl (15 mL), and the solvents
were concentrated under vacuum to half volume giving a residue, which
was extracted with EtOAc (3 × 15 mL). The organic layers were
combined, dried over Na2SO4, and the solvent
was removed under vacuum. The resulting crude product was purified
by flash column chromatography on silica gel using a 1:9 mixture of
EtOAc/PEt as the eluent.
9-Fluorenylmethoxycarbonyl chloride (FmocCl, 1.1 mmol)
was added to a solution of 37a–e (1 mmol) in CH3CN (5 mL). The reaction solution was heated at 90 °C
for 5 h and then stirred at room temperature for 18 h. The solvent
was evaporated giving a residue that was dissolved in EtOAc (15 mL),
and the resulting organic phase was washed with water (3 × 10
mL), dried over anhydrous Na2SO4, and concentrated
to dryness. All crude residues were finally purified via flash column
chromatography on silica gel using a 1:4 mixture of EtOAc/PEt as the
eluent.
General Procedure for the Synthesis of Final Compounds 17–21
4-Fluorobenzyl isocyanate (2 mmol) and
DBU (1.2 mmol) were sequentially added at room temperature to a stirring
solution of 38a–e (1 mmol) in anhydrous THF (5
mL). The reaction solution was stirred for 2 h and then it was quenched
with a saturated solution of NH4Cl (10 mL). The mixture
was concentrated under vacuum to half volume and the aqueous phase
was extracted with EtOAc (3 × 15 mL). The organic layers were
combined, dried over Na2SO4, and concentrated
in vacuum. The desired products were purified by preparative RP-HPLC.
HEK293mNPSR cells
were generated as previously described[3] and maintained in DMEM supplemented with 10% fetal bovine serum,
2 mM l-glutamine, and hygromycin B (100 mg/L) and cultured
at 37 °C in 5% CO2 humidified air. HEK293mNPSR cells were seeded at a density of 50,000 cells/well into poly-d-lysine coated 96-well black, clear-bottom plates. The following
day, the cells were incubated with medium supplemented with 2.5 mM
probenecid, 3 μM of the calcium sensitive fluorescent dye Fluo-4
AM, and 0.01% pluronic acid, for 30 min at 37 °C. After that
time, the loading solution was aspirated and 100 μL/well of
assay buffer (Hanks’ balanced salt solution; HBSS) supplemented
with 20 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES),
2.5 mM probenecid, and 500 μM Brilliant Black (Sigma-Aldrich)
was added. Concentrated solution (1 mM) of NPS was made in bidistilled
water and kept at −20 °C. Concentrated solutions (10 mM)
of NPSR antagonists were made in DMSO and kept at −20 °C.
Serial dilutions were carried out in HBSS/HEPES (20 mM) buffer (containing
0.02% bovine serum albumin fraction V). After placing both plates
(cell culture and master plate) into the fluorometric imaging plate
reader FlexStation II (Molecular Devices, Sunnyvale, CA), fluorescence
changes were measured. On-line additions were carried out in a volume
of 50 μL/well. To facilitate drug diffusion into the wells in
antagonist type experiments, the present studies were performed at
37 °C, and three cycles of mixing (25 μL from each well
moved up and down 3 times) were performed immediately after antagonist
injection to the wells. Inhibition response curves were determined
against the stimulatory effect of 10 nM NPS. Additionally, the concentration–response
curve to NPS has been tested in the absence and in the presence of
100 nM of 1, compound 16, and compound 21 (Figure ). NPSR antagonists were injected into the wells 24 min before adding
NPS.
Mouse Locomotor Activity Test
All animal care and experimental
procedures conformed with the European Communities Council Directives
(2010/63/EU) and national regulations (D.L. 26/2014). Studies involving
animals are reported in accordance with the ARRIVE guidelines.[51] This study was approved by the Italian Ministry
of Health (authorization number 120/2014-PR). The experiments were
performed with CD-1mice (2–4 month old, from the Laboratory
for Preclinical Research (LARP) of the University of Ferrara, Italy).
Mice were housed under standard conditions (22 °C, 55% humidity,
12 h light/dark cycle, light on at 7:00 am), with free access to food
and water. Appropriate environmental enrichment was present in each
cage. Mice were killed with CO2overdose. Each animal was
used only once. Experiments were performed during the light cycle
(between 09.00 and 13.00) according to Guerrini et al. (2009).[42] For in vivo studies, 1, 16, and 21 were solubilized in
water containing 1% DMSO and 10% Cremophor EL (Sigma-Aldrich). NPS
was solubilized in saline solution. Vehicle, 1, 16, and 21 were injected by the ip route 30 min
before saline or NPS injection. NPS or saline were given by the i.c.v.
route 15 min before the beginning of the test and the locomotor activity
was recorded for 30 min. The i.c.v. injections (2 μL/mouse)
were given under light (just enough to produce loss of the righting
reflex) isofluorane anesthesia into the left ventricle according to
the procedure described by Laursen and Belknap (1986)[52] and routinely adopted in our laboratory. For these experiments,
the ANY-maze video tracking system was used (Ugo Basile, application
version 4.52c Beta). Mice were positioned in a square plastic cage
(40 × 40 cm), one mouse per cage. Four mice were monitored in
parallel. Mouse horizontal activity was monitored by a camera while
vertical activity was measured by an infrared beam array. The parameters
measured were cumulative distance traveled (total distance in m that
the animal traveled during the test), immobility time (the animal
is considered immobile when 90% of it remains in the same place for
a minimum of 2.5 s), and the number of rearings (the number of beam
breaks due to the vertical movements; this input is triggered only
when the beam is interrupted for at least 200 ms). Previous studies
performed under the present experimental conditions demonstrated that
NPS stimulated locomotor activity in a dose-dependent way;[42] from these studies, the dose of 0.1 nmol was
selected as the lower dose inducing statistically significant effects.
Data Analysis and Terminology
Data are expressed as
means ± standard error of the mean (SEM) of n experiments. Nonlinear
regression analysis using GraphPad Prism software (v.4.0) allowed
logistic iterative fitting of the resultant responses and the calculation
of agonist potencies and maximal effects. Agonist potency was expressed
as pEC50, which is the negative logarithm to base 10 of
the agonist molar concentration that produces 50% of the maximal possible
effect of that agonist. In inhibition response experiments (i.e.,
increasing concentrations of antagonist vs a fixed concentration of
agonist), the antagonist potency was expressed as pKB, derived from the following equationwhere IC50 is the concentration
of antagonist that produces 50% inhibition of the agonist response,
[A] is the concentration of agonist, EC50 is the concentration
of agonist producing a 50% maximal response, and n is the Hill coefficient of the concentration–response curve
to the agonist. When the concentration–response curve to NPS
has been tested in the absence and in the presence of antagonists,
the antagonist potency was expressed as pA2, derived from the following equationwhere CR means the ratio
between agonist potency
in the presence and absence of antagonist and [A] the molar concentration
of the antagonist.In vivo data are expressed
as mean ± SEM of n animals. Data were analyzed using two-way
ANOVA, followed by Bonferroni’s post-hoc test. Differences
were considered statistically significant when p <
0.05.
Molecular Modeling
hNPSR Model Construction
The query sequence of the
hNPSR in its I107 variant (Q6W5P4, FASTA format) was downloaded from
the Universal Protein Resource.[53] BLAST
(Basic Local Alignment Search Tool) was used to search the homologous
sequences to be used as template structures. The sequences of templates
were obtained from the Uniprot web server. The human C5a anaphylatoxin
chemotactic receptor 1 (P21730, PDB 6C1R),[54] the human
κ opioid receptor (P41145, PDB 4DJH),[33] the human
M2 muscarinic receptor (P08172, PDB 5ZKC),[55] the human
neuropeptide Y Y1 Receptor (Q15761, PDB 5ZBH),[29] the human
orexin-1 receptor (O43613, PDB 6TOD),[56] and the
human type-2 angiotensin receptor (P50052, PDB 4ZUD)[38] were chosen as the templates. The sequence identity and
coverage between hNPSR and the six templates are reported in Table S1 along with the phylogenetic tree (Figure S3) and pairwise sequence alignments (Figures S4–S9). The choice of templates
was dictated by the resolution and sufficient similarity sequence
coverage with the target. The alignments reported in Figures S4–S9 were used to build six hNPSR models using
the Prime module within Schrodinger (Prime; Schrödinger, LLC,
version 2020-1). The X-ray structures of the templates were obtained
from the Protein Data Bank. Homology built models were obtained using
the knowledge-based method. Validation of the model was carried by
generating the Ramachandran plots for each model. This analysis allowed
supporting the viability of the constructed models that all had >90%
of the residues in the allowed regions of the plot.
Docking Calculations
Docking calculations were attained
employing the Glide tool implemented in Maestro 12.4.[57] The 3D structures of 1, 3–5, 7, 10, 12, and 14–21 were generated with the Maestro fragment Build tool and then energetically
minimized with Macromodel.[58] The 6 models
were all prepared through the Protein Preparation Wizard of the Maestro
12.4 graphical user interface, which assigns bond orders, adds hydrogen
atoms, and generates appropriate protonation states.The docking
grid box was centered on the residues lining the putative binding
orthosteric binding, with a grid box dimension equal to 31 Å
× 31 Å × 31 Å. Finally, docking runs were carried
out using the standard precision method. Pictures were rendered employing
UCSF Chimera software.[59]
MD Simulation
System Setup
The complexes obtained from
docking experiments of 1, 16, and 21 in both BM1 and BM2 employing the hNPSR model built using
the human neuropeptide Y Y1 receptor (hNPY1R, PDB code 5ZBH) structure as a
template were used to build an MD simulation system. The complex was
embedded in a membrane of phosphatidylcholinelipids[60] using Maestro’s system builder and was located in
the membrane using the default parameters. Next, the system was solvated
in an orthorhombic water box with a buffer distance of 10 Å.[61] The system was neutralized with 9 Cl– ions for the complexes with 1 and 21 while
10 Cl– ions were included in the 16/hNPSR complexes. The salt concentration was set to 0.15 M NaCl.
The OPLS3 force field was used for the constructed receptor/ligand/membrane
system.[62]
MD Simulation Protocols
The Desmond module within the
Schrödinger suite was used for MD simulations.[63−65] First, the system was relaxed using the following relaxation protocol
for membrane proteins:minimization with restraints on all
solute-heavy atoms;unrestrained
minimization;Brownian
Dynamics in NVT ensemble (constant
number of particles, constant volume, constant temperature of 300
K) using small timesteps and restraints on solute heavy atoms (50
kcal/mol) at 10 K for 50 ps;NPT ensemble simulation
(constant number of particles, constant pressure of 1 bar, and constant
temperature of 100 K) Gaussian barrier potential on water molecules,
membrane restrained in z, protein restrained (20
kcal/mol) for 20 ps;NPγT ensemble simulation
(constant number of particles, constant pressure
of 1 bar, constant temperature of 100 K, and lateral surface tension
of membranes) Gaussian Barrier potential on water molecules, membrane
restrained in z, protein restrained (10 kcal/mol)
for 100 ps;NPγT ensemble simulation with increasing temperature
from 100 to 300
K, Gaussian Barrier potential on water molecules, and gradual release
of restraints for 150 ps.NVT production run
(constant number of particles, constant volume, and constant temperature
of 300 K) without any restraint for 100 ps.After the relaxation, 100 ns production runs were conducted
under the NPγT ensemble for
each of the six systems using the default protocol. In detail, the
temperature was controlled using the Nosé–Hoover thermostat[66,67] with a coupling constant of 1.0 ps. The pressure was controlled
using the Martyna–Tuckerman–Klein barostat[68,69] with a coupling constant of 2.0 ps. The cutoff distance for short-range
nonbonded interactions was 9 Å, and the long-range van der Waals
interactions were based on a uniform density approximation. To minimize
the computation time, nonbonded forces were calculated using an r-RESPA
integrator[70] where the short-range forces
were updated every two steps and the long-range forces were updated
every six steps. The trajectories were saved at 100.0 ps. The Desmond
SID tool was used to analyze the receptor–ligand interactions
during the MD trajectory. Particular attention was given to ligand–residue
interactions and ligand RMSF.
Authors: Edward Harder; Wolfgang Damm; Jon Maple; Chuanjie Wu; Mark Reboul; Jin Yu Xiang; Lingle Wang; Dmitry Lupyan; Markus K Dahlgren; Jennifer L Knight; Joseph W Kaus; David S Cerutti; Goran Krilov; William L Jorgensen; Robert Abel; Richard A Friesner Journal: J Chem Theory Comput Date: 2015-12-01 Impact factor: 6.006
Authors: Hongyan Zhu; Melissa K Mingler; Melissa L McBride; Andrew J Murphy; David M Valenzuela; George D Yancopoulos; Michael T Williams; Charles V Vorhees; Marc E Rothenberg Journal: Psychoneuroendocrinology Date: 2010-02-19 Impact factor: 4.905
Authors: Stewart D Clark; Dee M Duangdao; Stefan Schulz; Li Zhang; Xiaobin Liu; Yan-Ling Xu; Rainer K Reinscheid Journal: J Comp Neurol Date: 2011-07-01 Impact factor: 3.215
Authors: Edward Lyman; Chris Higgs; Byungchan Kim; Dmitry Lupyan; John C Shelley; Ramy Farid; Gregory A Voth Journal: Structure Date: 2009-12-09 Impact factor: 5.006
Authors: Cheng Zhang; Yoga Srinivasan; Daniel H Arlow; Juan Jose Fung; Daniel Palmer; Yaowu Zheng; Hillary F Green; Anjali Pandey; Ron O Dror; David E Shaw; William I Weis; Shaun R Coughlin; Brian K Kobilka Journal: Nature Date: 2012-12-09 Impact factor: 49.962
Figure 1
Scheme 1
Scheme 2
Scheme 3
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6