β-Turn mimetic-based stabilizers of protein-protein interactions for the study of the non-canonical roles of leucyl-tRNA synthetase

Article abstract of DOI:10.1039/c5sc03493k

For the systematic perturbation of protein-protein interactions, we designed and synthesized tetra-substituted hexahydro-4H-pyrazino[2,1-c][1,2,4]triazine-4,7(6H)-diones as β-turn mimetics. We then devised a new synthetic route to obtain β-turn mimetic scaffolds via tandem N-acyliminium cyclization and constructed a 162-member library of tetra-substituted pyrazinotriazinediones with an average purity of 90% using a solid-phase parallel synthesis platform. Each library member was subjected to ELISA-based modulator screening for the LRS-RagD interaction, which plays a pivotal role in the nutrient-dependent mTORC1 signalling pathway. Western blot analysis of phosphorylated S6K1 as well as FRET-based imaging confirmed that 5c{3,9} stabilizes the direct interaction between LRS and RagD and activates mTORC1 in live cells under leucine-deprived conditions. Thus, 5c{3,9} can be used as a new research tool for studying the non-canonical role of LRS.

Full text of DOI:10.1039/c5sc03493k

Chemical
Science
View Article Online
View Journal | View Issue
EDGE ARTICLE
b-Turn mimetic-based stabilizers of protein–
protein interactions for the study of the non-
Cite this: Chem. Sci., 2016, 7, 2753
canonical roles of leucyl-tRNA synthetase†
a
b
a
ab
*
Chanwoo Kim,‡ Jinjoo Jung,‡ Truong T Tung and Seung Bum Park
For the systematic perturbation of protein–protein interactions, we designed and synthesized tetra-
substituted hexahydro-4H-pyrazino[2,1-c][1,2,4]triazine-4,7(6H)-diones as b-turn mimetics. We then
devised a new synthetic route to obtain b-turn mimetic scaffolds via tandem N-acyliminium cyclization
and constructed a 162-member library of tetra-substituted pyrazinotriazinediones with an average purity
of 90% using a solid-phase parallel synthesis platform. Each library member was subjected to ELISA-
based modulator screening for the LRS–RagD interaction, which plays a pivotal role in the nutrient-
dependent mTORC1 signalling pathway. Western blot analysis of phosphorylated S6K1 as well as FRET-
based imaging confirmed that 5c{3,9} stabilizes the direct interaction between LRS and RagD and
activates mTORC1 in live cells under leucine-deprived conditions. Thus, 5c{3,9} can be used as a new
research tool for studying the non-canonical role of LRS.
Received 16th September 2015
Accepted 14th December 2015
DOI: 10.1039/c5sc03493k
www.rsc.org/chemicalscience
strategies have been used to construct diverse molecular
frameworks, such as polyheterocycles and macrocycles, with
Introduction
different characteristics compared to conventional inhibitors at
the active site of druggable targets as substrate analogs.9–14
A recent structural analysis of PPI interfaces revealed that
not all residues at the PPI interface were critical, but rather that
Protein–protein interactions (PPIs) play critical roles in
a myriad of biological processes and many complex diseases,
including cancer1 and metabolic and neurodegenerative
diseases that are oen caused by aberrant PPIs.2–6 Given their
signicance in biological systems, PPIs have been investigated
to identify novel therapeutic targets for various diseases.
Despite progress over the past 20 years, at most 40 PPIs have
been targeted among the ꢀ650 000 pair-wise interactions in the
human interactome, and a limited number of modulators have
reached clinical trials.4–6 It is probably due to the fact that PPI
interfaces do not typically bind to endogenous ligands that may
serve as lead structures for drug discovery. PPI interfaces also
present inherent physical challenges: the small-molecule
binding region is likely noncontiguous to the interface between
interacting proteins, and the PPI interface itself is relatively
large and/or at.3,4,7 In fact, few studies have reported successful
and/or sufficiently potent PPI modulators because PPI modu-
lators do not fully overlap with the criteria of drug-likeness
based on FDA-approved orally available drugs.5,6,8 For the
systematic perturbation of diverse PPIs, various synthetic
small “hot spots” conferred most of the binding energy.4–6,15–17
A
hot spot can be dened as a specic structural region that
assembles at the PPI interface with conformational adap-
tivity.4,18 Hot spot areas are generally hydrophobic and can be
comparable to the size of a small molecule, suggesting that PPIs
can be modulated via the specic binding of small molecules at
the hot spots of the protein partners.4,7 The structural elements
for specic binding at hot spots include three major secondary
structures such as a-helix, b-turn and b-strand, and small-
molecule-based mimetics may mediate PPIs by their binding at
the PPI interface.17,19 Therefore, these sites could be used for the
design and synthesis of drug-like small molecules that mimic
these recognition motifs for the systematic identication of
novel modulators of protein–protein interactions.
Because of the conformational adaptivity of hot spots under
physiological conditions, small-molecule-based mimetics of
secondary structures may be useful in studies of the biological
processes of PPIs with unknown structures or as potential
candidates for rst-in-class therapeutics. Although a few b-turn
mimetics have been reported as PPI modulators, these motifs
are frequently observed in the loop regions of protein domains
and are thought to be important in PPIs.19,20 Therefore, b-turn
mimetics should be constructed for use as PPI modulators and
in studies of unexplored biological processes. It would be
particularly valuable to construct polar polyheterocyclic core
aDepartment of Chemistry, Seoul National University, Seoul 151-747, Korea. E-mail:
sbpark@snu.ac.kr; Fax: +82 2 884 4025
bDepartment of Biophysics and Chemical Biology, Seoul National University, Seoul
151-747, Korea
† Electronic supplementary information (ESI) available: Detailed procedures for
synthesis, bioassay, and biophysical experiments, and full characterization data
of all new compounds. See DOI: 10.1039/c5sc03493k
‡ These authors contributed equally to this study.
This journal is © The Royal Society of Chemistry 2016
Chem. Sci., 2016, 7, 2753–2761 | 2753

View Article Online
Chemical Science
Edge Article
skeletons that can accommodate a series of hydrophobic exibility to mimic various secondary structures such as b-turns
substituents, which is critical for specic binding at hot spots, and g-turns.
without sacricing the solubility and membrane permeability of
the resulting b-turn-mimetic-based small molecules with mimetic
multiple hydrophobic substituents.17,19,21,22
dione 1. A small-molecule library of b-turn mimetics was con-
Herein, we report the design and synthesis of a b-turn
hexahydro-4H-pyrazino[2,1-c][1,2,4]triazine-4,7(6H)-
Among the PPIs involved in various biological processes, we structed using a parallel synthetic strategy involving both solid-
are interested in specic PPIs related to the mechanistic target and solution-phase reactions. We also maximized the molecular
of rapamycin complex 1 (mTORC1), which is a serine/threonine diversity of our peptidomimetic library to occupy the potential
kinase that controls mRNA transcription and translation by chemical space for the specic modulation of protein–protein
integrating various environmental changes and signals such as interactions. In order to determine whether the b-turn mimetics
growth factors, nutrients and energy status. Thus, mTORC1 would perturb protein–protein interactions, we conducted an
modulates cell growth and proliferation and the dysregulation ELISA-based biological evaluation of our b-turn mimetic library
of mTORC1 is closely related to many diseases such as cancer against the specic interaction between LRS and RagD and
and diabetes, making mTORC1 an attractive therapeutic identied a novel PPI stabilizer targeting the LRS–RagD
target.23 A recent study demonstrated that Ras-related GTPases interaction.
(Rag) mediate amino acid-based activation of mTORC124 and
that leucyl-tRNA synthetase (LRS) is thought to be a binding
partner of Rag proteins, particularly RagD, in a leucine-depen-
Results and discussion
dent manner. Along with its canonical role in the conjugation of
leucine to its cognate tRNA for leucyl-tRNA synthesis, LRS can
also act as a leucine sensor, bind to RagD-GTP, and form a LRS–
RagD complex, which translocates mTORC1 from the cytosol to
the lysosome surface for subsequent activation of the mTORC1
signalling pathway (Fig. 1).25 The nutrient sensing mechanism
of mTORC1, particularly for Leu, an essential biomarker for
nutrient status in cellular systems,26 may be regulated by PPIs
between LRS and RagD and directly mediate mTORC1 activa-
tion. Therefore, we hypothesized that small-molecule PPI
modulators between LRS and RagD can be used as powerful
research tools for studying the nutrient-dependent activation of
mTORC1 and the subsequent biological outcomes. The protein
structures of both human LRS and RagD are unknown and the
binding mode of the LRS–RagD complex has not been deter-
mined. Thus, we attempted to identify PPI modulators of the
LRS–RagD interaction through the systematic construction of
polar molecular frameworks with limited conformational
Design of a tetra-substituted pyrazinotriazinedione scaffold as
a b-turn mimetic
The b-turn is one of the three major secondary structural motifs
found in proteins and peptides, particularly in the reverse
direction of polypeptide strands. The peptidomimetics of b-
turns have been investigated to identify small-molecule modu-
lators that disrupt b-turn-mediated recognition events.19,27–31 As
illustrated in Fig. 2A, the b-turn consists of four amino acid
residues, designated as i, i + 1, i + 2, and i + 3. One of the
important features of the b-turn structure is that the distance
˚
between the Cai and Cai+3 atoms is less than 7 A. Depending on
the dihedral angles j and 4 of the i + 1 and i + 2 residues,
respectively, nine b-turn types dened by Hutchinson and
Thornton are widely used: types I, I0, II, II0, VIa1, VIa2, VIb, VIII,
and IV.19,32 To successfully construct b-turn structures, it is
Fig. 2 (A) Schematic diagram of a b-turn structure. (B) Overlay of
a potential mode of b-turn mimicry with a tetra-substituted hexahy-
dro-4H-pyrazino[2,1-c][1,2,4]triazine-4,7(6H)-dione. (C) Distance
between substituents in the energy-minimized conformer of its tetra-
Fig. 1 Roles of leucyl-tRNA synthetase. (i) Canonical role (left circle) of substituted representative analog
7 containing 4-methylphenyl,
leucyl-tRNA synthetase; LRS conjugates leucine and cognate tRNA. (ii) benzyl, butyl, and 4-chlorophenyl at the R1–R4 positions, respectively.
Non-canonical role (right circle) of LRS–leucine-dependent mTORC1 (D) Alignment of an energy-minimized structure of representative
activation; LRS binds to RagD and activates mTORC1 in a leucine- compound 7 with the peptide backbone structure of phospholipase A2
dependent manner.
b-turn motif (PDB: 4BP2).
2754 | Chem. Sci., 2016, 7, 2753–2761
This journal is © The Royal Society of Chemistry 2016

View Article Online
Edge Article
Chemical Science
essential to mimic the back bone alignment and functionality of
the side chains.19 Dihedral angles and the mean distances
between the a-carbon centers of b-turn mimetics are considered
to be geometric indicators for evaluating the mimetic capability
in 3-dimensional space because of their crucial roles in b-turn
structures.5,6,33 Using this approach, Whitby et al. reported
a strategy for designing trans-pyrrolidine-3,4-dicarboxamide as
a b-turn mimetic structure as well as investigating its biological
activities.29 Since protein–protein interactions primarily occur
at the hydrophobic “hot spot” region of the PPI interfaces, we
also considered the general characteristics of substituents
appended to the b-turn skeleton to mimic the hydrophobic side
chains of the residues at hot spots.
For b-turn mimicry, we designed and synthesized a fused
bicyclic scaffold, tetra-substituted hexahydro-4H-pyrazino[2,1-c]
[1,2,4]triazine-4,7(6H)-dione (1), as a b-turn mimetic to produce
the desired conformation that could maintain its b-turn struc-
tural features with limited conformational exibility under
physiological conditions. The rigid bicyclic scaffold 1 allowed
several slightly different conformers to cover the geometric b-
turn structures and the resulting conformational restriction
may have induced a higher binding affinity toward biopolymers
because of its prepaid entropic penalty.27 As shown in Fig. 2B,
each a-carbon of the b-turn motif matched well with each atom
of our scaffold 1. When we calculated the distances between
each atom of the energy-minimized conformer of its represen-
tative analog 7, the distance and geometry of the substituents
were well-matched with the designated a-carbon of a b-turn
motif, conrming that our b-turn mimetic scaffold 1 ts well
into the geometric analysis data of b-turns observed in 10 245
reported structures in the protein data bank (Fig. 2C).29 As nal
conrmation, the energy-minimized 3-dimensional structure of
7 was overlapped with the X-ray crystal structure of phospholi-
pase A2 containing the b-turn motif in its biologically active
conformation.34,35 As shown in Fig. 2D, 7 overlaps with the b-
turn backbone structure of phospholipase A2 (in pink) showing
excellent conformational similarity. The root-mean-square
deviation (rmsd) value of compound 7 compared with each a-
Fig. 3 Retrosynthetic analysis of tetra-substituted hexahydro-4H-
pyrazino[2,1-c][1,2,4]triazine-4,7(6H)-dione (1).
containing the R3 and R4 groups. To ensure the efficiency of the
high-throughput parallel synthesis, we synthesized the carbox-
ylic acid partner (I) in the solution phase, and the subsequent
amide coupling with the amine partner (II) allowed for the
formation of the linear precursor on a solid support. The nal
acid-catalyzed cleavage step from the solid support was
designed for the in situ formation of N-acyliminium interme-
diates from aldehydes and an amide nitrogen, followed by their
tandem stereoselective cyclization to obtain the b-turn mimetic
scaffold 1 with a new stereogenic center. This pyrazino-
triazinedione 1 contained four modiable sites to maximize the
molecular diversity, which could mimic the functionalities of
side-chain moieties in natural amino acids for the specic
recognition events in various PPIs.
To generate carboxylic acid partners (A-1 & A-2), we used two
different hydrazines (methyl and tolyl) for diversication at the
R1 position. Since a large quantity of acid partners is required
for library construction, the synthesis of the acid partners was
carried out in the solution phase (Scheme 1A). The synthetic
route that we used for methylhydrazine differed from that for
tolylhydrazine because methylhydrazine requires Boc protec-
tion (iii) of its internal nitrogen to differentiate the nucleophi-
licity of its two nitrogens. In the case of tolylhydrazine, we
treated it with three different isocyanates (benzyl, 4-uo-
rophenyl, and allyl) via nucleophilic addition at 0 ꢁC to modify
the external nitrogen (i). The remaining internal nitrogen was
further decorated with t-butyl bromoacetate via SN2 reaction in
the presence of KHCO3 at an elevated temperature. Due to the
limited nucleophilicity of the internal nitrogen in tolylhy-
drazine, its modication step with t-butyl bromoacetate
suffered with the formation of a diazene moiety as a major by-
product, which lowered the overall yields regardless of the types
of isocyanates used. For methylhydrazine, Boc-protected
methylhydrazine was modied with three different isocyanates
(benzyl, 4-uorophenyl and 2-biphenyl) (iii), followed by Boc
deprotection using 4 N HCl in 1,4-dioxane (iv). Aer the SN2-
type modication of the internal nitrogen with t-butyl bro-
moacetate, six different carboxylic acid partners (A-1 & A-2) were
obtained through acidic hydrolysis on a gram scale in 3–4 steps
with moderate overall yields (30–51%).
˚
carbon of phospholipase A2 was 0.643 A for ve atom positions
(see the ESI†). We also conrmed the moderate rmsd values
˚
(0.544–0.916 A) of compound 7 toward ve different types of b-
turn structures (I, II, II0, III and VIII). Based on these results, we
were condent that our pyrazinotriazinedione scaffold 1
mimicked the natural b-turn structures with limited confor-
mational exibility. In addition, our b-turn mimetic 1 showed
3
a high Fsp value (0.67) with two stereogenic centers, which may
allow selective binding toward certain target proteins to perturb
PPIs.
High-throughput library synthesis of tetra-substituted
pyrazinotriazinediones as b-turn mimetics
The retrosynthetic analysis of tetra-substituted hexahydro-4H-
pyrazino[2,1-c][1,2,4]triazine-4,7(6H)-dione 1 is shown in Fig. 3.
To construct the b-turn mimetic library, we designed two amide
coupling partners; the carboxylic acid partner (I) containing the
R1 and R2 groups and the amine partner (II) on a solid support
This journal is © The Royal Society of Chemistry 2016
Chem. Sci., 2016, 7, 2753–2761 | 2755

View Article Online
Chemical Science
Edge Article
Scheme 1 General synthesis scheme for acid partners (A) and tetra-substituted hexahydro-4H-pyrazino[2,1-c][1,2,4]triazine-4,7(6H)-dione (B).
Reagents and conditions: (i) R2-isocyanate, TEA, THF, 0 ꢁC to r.t., 5 h, then t-butyl bromoacetate, KHCO3, DMF, 80 ꢁC, 12 h; (ii) 4 N HCl, 1,4-
dioxane, r.t., overnight; (iii) Boc2O, NaHCO3, THF/H2O (1 : 1 v/v), r.t., overnight, then R2-isocyanate, TEA, THF, 0 ꢁC to r.t., 5 h; (iv) 4 N HCl, 1,4-
dioxane, r.t., then t-butyl bromoacetate, K2CO3, toluene/DMF (8 : 1 v/v), reflux, 12 h; (v) R3-NH2, DMSO, 60 ꢁC, 12 h; (vi) Fmoc-amino acid, HCTU
[O-(6-chlorobenzotriazol-1-yl)-N,N,N0,N0-tetra-methyluronium hexafluorophosphate], DIPEA, DMF, r.t., 4 h; (vii) 20% piperidine, DMF, r.t., 10
min; (viii) acid coupling partner (A-1), HOBt [hydroxybenzotriazole], DIC [N,N0-diisopropylcarbodiimide], DMF, r.t., 3 h; (ix) acid coupling partner
(A-2), HOBt, DIC, DMF, r.t., 3 h; (x) formic acid, r.t., 18 h.
As shown in Scheme 1B, library synthesis was initiated by nal acidolysis step (x) involving neat formic acid allowed the
preparing solid-supported amine partners 4 with various R3 and formation of bicyclic b-turn mimetic scaffold 1 in excellent
R4 substituents. First, bromoacetal resin was loaded with three purities via the N-acyliminium intermediate shown in Fig. 3.
different amines containing methylfuranyl, n-butyl and benzyl
As shown in Table 1, eight representative compounds (7–14)
moieties at the R3 position via SN2 reaction (v). Next, solid- were obtained without further purication with good purities
supported secondary amines 2 were coupled to various Fmoc- and were fully characterized using 1H, 13C, and gCOSY NMR
protected natural or unnatural amino acids [Phe, Val, Met, Ile, spectroscopy along with LC/MS and chiral HPLC analysis (see
Tyr, Leu, Arg, Cys(bzl) and Phe(4-Cl)] under the HCTU-based the ESI†). Because PPI interfaces generally contain some
amide coupling conditions (vi). Aer the deprotection of Fmoc hydrophobic regions, various aromatic and hydrophobic
amines 3 on a solid support using 20% piperidine (vii), the substituents were adopted at the R1–R4 position of this b-turn
resulting amine partners 4 were coupled with carboxylic acid mimetic scaffold 1 to target these hot spots. Each substituent at
partners (A-1 or A-2) under HOBt and DIC-mediated amide four different diversity points was selected to mimic the side-
coupling conditions (viii or ix), respectively. Aer completion of chains in the amino acids of the b-turn structures from natural
the amide coupling monitored using ninhydrin testing, the to unnatural amino acids as well as to maximize the molecular
2756 | Chem. Sci., 2016, 7, 2753–2761
This journal is © The Royal Society of Chemistry 2016

View Article Online
Edge Article
Chemical Science
Table 1 Purity and mass conformation of representative compounds
MS[M + H]+
R1
R2
R3
R4
Puritya (%)
91
Calcd
Found
574.22
7
8
574.25
96
99
93
98
492.29
530.27
524.26
558.22
492.04
530.03
524.15
558.20
9
10
11
12
13
90
753.31
753.67
90
95
512.26
514.24
512.32
514.06
14
a
Purities in percentage were obtained using PDA (photodiode array)-based LC/MS analysis of crude nal compounds aer a short lter column. Pbf
¼ 2,2,4,6,7-pentamethyldihydrobenzofuran.
diversity and drug-likeness of the resulting b-turn mimetic library members were measured using LC/MS equipped with
library via principal component analysis (see the ESI†). It is a PDA detector aer short silica-gel ltration (Table 2). The
worth mentioning that our pyrazinotriazinedione skeleton itself average purity was found to be 90%.
is polar (clog P ¼ ꢂ1.26) enough to accommodate a series of
hydrophobic substituents and maintain the drug-likeness of the
resulting b-turn mimetics (clog P range from 3.45 to 7.47).
Using this synthetic route, a new chiral center was generated
in bicyclic b-turn mimetic scaffold 1 in a stereoselective
manner. In fact, a new stereogenic center was generated via two
different trajectories of the nitrogen nucleophile (pathways
a and b) during the tandem N-acyliminium cyclization step
under neat formic acid (see Fig. 4A). However, we only observed
a single diastereomer in the reaction mixture, which was
conrmed using crude LC/MS and crude 1H NMR spectroscopy.
Aer purication of representative compound 9, we performed
a 2D nuclear Overhauser effect (NOE) NMR experiment to
conrm the absolute conguration of the b-turn mimetic pyr-
azino-triazinedione scaffold. As shown in Fig. 4B (red color), we
clearly observed NOE interactions between Ha at the new ster-
eogenic center and Hb at the R4 position, and concluded that
the nitrogen nucleophile attacked the top face of the N-acyli-
minium intermediate (pathway a) to avoid the steric repulsion
of bulky R4 residues during tandem cyclization.36 Using a solid-
Fig. 4 (A) Two possible trajectories of the nitrogen nucleophile
phase parallel synthetic strategy, we efficiently constructed
(pathways a & b) via tandem N-acyliminium cyclization. (B) Repre-
a 162-member b-turn mimetic library and the purities of all
sentative compound 9 and its 2D-NMR NOE data.
This journal is © The Royal Society of Chemistry 2016
Chem. Sci., 2016, 7, 2753–2761 | 2757

View Article Online
Chemical Science
Edge Article
RagD interaction in a dose-dependent manner. It is difficult to
identify small-molecule modulators of specic PPIs, but it is
even more difficult to identify PPI stabilizers from a small-
molecule library.37 Thus, our results conrm the quality and
novelty of our b-turn mimetic library. Through our structure–
activity relationship analysis, we clearly demonstrated that our
pyrazinotriazinedione scaffold 1 with hydrophobic and aromatic
substituents stabilized the LRS–RagD interaction more effec-
tively (Fig. 5A), which is in accordance with our library design
rationale, in which hydrophobic substituents on a single b-turn
mimetic scaffold enhanced the binding affinity toward the
hydrophobic region of hot spots at the PPI interfaces.
Screening the b-turn mimetic library to identify LRS–RagD
interaction modulators
To identify small-molecule modulators of the LRS–RagD inter-
action, we conducted ELISA-based screening with puried LRS
coated on 96-well plates and each well was treated with each b-
turn mimetic library member and GST (Glutathione S-trans-
ferase)-tagged RagD simultaneously for 3 h. Due to its in vitro
instability, GST-RagD was roughly puried using a glutathione
column and used immediately without freezing. The specic
interaction between LRS and RagD can capture RagD-GST at the
surface of each well, which was detected using anti-GST anti-
body. The level of small-molecule-based modulation toward the
LRS–RagD interaction was quantied based on the quantity of
bound RagD-GST, which was measured by determining the
enzyme activity conjugated to the secondary antibody.
Hit compound selection of LRS–RagD stabilizer and its
biophysical evaluation
Using ELISA-based screening, we tested all 162 compounds in
the b-turn mimetic library to determine whether they could
affect the interaction between LRS and RagD. We identied
a series of b-turn mimetic compounds that stabilized the LRS–
Aer selecting a series of hit compounds from our ELISA-based
screening, we performed western blot analysis of
Table 2 Purity table of all final compoundsa
Fig.
5 (A) Structure–activity relationship study of PPI stabilizers
between LRS–RagD among our b-turn mimetic compounds.
Compounds with hydrophobic and aromatic substituents (black)
showed better stabilizing activity than others (grey). (B) HEK293T cells
were starved of leucine for 1 h and treated with 20 mM compounds for
3 h under leucine-deprived conditions. Phospho-S6K1 (T389), S6K1,
and GAPDH were determined using western blot. HEK293T cells
treated with 0.6 mM leucine were used as a control. (C) Quantification
of western blotting against a DMSO control.
a
Purities (%) were obtained using PDA-based LC-MS analysis of crude
nal compounds aer a short lter column (average purities: 90%).
2758 | Chem. Sci., 2016, 7, 2753–2761
This journal is © The Royal Society of Chemistry 2016

View Article Online
Edge Article
Chemical Science
phosphorylated S6K1, one of the mTORC1 substrates, to of the FRET signal between CFP and YFP (Fig. 6C). To precisely
determine whether these compounds could affect the activity of quantify the FRET efficiency, we selected some single cells to
mTORC1 by stabilizing the LRS–RagD interaction. As shown in track over time and conducted live cell imaging with CFP/CFP,
Fig. 5B, the presence of leucine can initiate the LRS and RagD CFP/YFP, and YFP/YFP lter sets (Fig. S3†). Finally, the captured
interaction, which is essential for the activation of mTORC1 and images were analyzed to exclude false-positive signals such as
phosphorylation of the subsequent S6K1. Under leucine- CFP and YFP crosstalk. As shown in Fig. 6D and E, we observed
deprived conditions, phosphorylated S6K1 (T389) decreased that the FRET efficiency between CFP and YFP was increased
because of the inactivation of mTORC1. However, the phos- over time aer treatment with 5c{3,9}, conrming the 5c{3,9}-
phorylation level of S6K1 was successfully restored upon treat- mediated stabilization of the LRS and RagD interaction under
ment with our b-turn mimetic compounds, as the LRS–RagD live cell conditions. For the less potent compound 5a{2,9}, we
PPI conferred stability even under leucine-deprived conditions. observed marginal changes in the FRET efficiency that were
Among the initial hit compounds, we identied the most similar to those in the DMSO control. This FRET-based
potent compound based on the relative intensity of phosphor- biophysical experiment supported that our b-turn mimetic 5c
ylated S6K1 in the Leu-deprived media compared to that in {3,9} stabilized the direct LRS–RagD interaction in the cellular
normal media (Fig. 5C) and selected 5c{3,9} for further system and modulated the activity of mTORC1.
biophysical analysis (Fig. 6A and B). Using ELISA and western
blot analysis, we conrmed that 5c{3,9} stabilized the in vitro
interaction between LRS and RagD and regulated the activity of
mTORC1, leading to the restoration of phosphorylated S6K1
levels under leucine-deprived conditions. But, this observation
does not necessarily conrm the stabilization of the direct LRS–
RagD interaction. To conrm this, we established a novel
biophysical assay using uorescence resonance energy transfer
(FRET) to examine the interaction between CFP (Cyan Fluores-
cent Protein)-fused LRS (LRS-CFP) and YFP (Yellow Fluorescent
Protein)-fused RagD (RagD-YFP) in live cells: the extent of the
LRS and RagD direct interaction was measured as the intensity
Small-molecule stabilizer of LRS–RagD interaction as a tool
compound
Our b-turn mimetic bicyclic compound 5c{3,9} selectively acti-
vated mTORC1, a key protein complex of cell growth and
proliferation, with a novel mode of action—direct stabilization
of the LRS–RagD interaction. 5c{3,9} maintained the cells under
normal conditions even in the leucine-deprived media. In fact,
nutrient sensing, particularly amino acid sensing, is extremely
important for cell survival, but there are no effective methods
for dissecting the amino acid-mediated activation of mTORC1.
Leucine is an essential amino acid in cells and plays a vital role
Fig. 6 (A) Structure of the hit compound, 5c{3,9}. (B) ELISA-based evaluation of LRS and RagD stabilization by 5c{3,9} in a dose-dependent
manner. (C) Schematic representation of FRET between LRS-CFP and RagD-YFP. When LRS directly interacts with RagD, LRS-fused CFP is
physically close to RagD-fused YFP, generating a FRET signal. Fluorescence of 475 nm from CFP with excitation at 433 nm transfers to YFP and
emits fluorescence at 527 nm. (D) Relative FRET efficiency ratio between LRS-CFP and RagD-YFP in HEK293T cells co-transfected with LRS-CFP
and RagD-YFP. Images were taken at 10 min intervals over 3.5 h in a single cell. Cells were randomly selected from 3 independent experiments (P
¼ 0.002; Mann–Whitney Test). 5c{3,9} (clear square) or 5a{2,9} (reverse black triangle) was treated at a 40 mM concentration after an initial 30 min
of live cell imaging. Images were captured with CFP/CFP, CFP/YFP, and YFP/YFP filters (excitation/emission). Captured FRET images were
analyzed using Softworks (imaging software) to exclude false-positive signal such as CFP crosstalk and YFP crosstalk. (E) Representative
calculated FRET within cells. The color scale indicates the range of FRET intensity, from low (blue) to high (red). Scale bar, 20 mm.
This journal is © The Royal Society of Chemistry 2016
Chem. Sci., 2016, 7, 2753–2761 | 2759

View Article Online
Chemical Science
Edge Article
in anabolic pathways such as skin wound healing and restoring
muscle loss.38 Therefore, 5c{3,9} could be an excellent research
tool for activating the mTORC1 signalling pathway and used to
reveal the non-canonical role of LRS in leucine-decient
conditions. Furthermore, 5c{3,9} can be used as a novel lead
compound for the selective activation of mTORC1 via LRS–
RagD stabilization without targeting mTORC1 itself.
3 M. J. Perez de Vega, M. Martin-Martinez and R. Gonzalez-
Muniz, Curr. Top. Med. Chem., 2007, 7, 33–62.
4 M. R. Arkin, Y. Tang and J. A. Wells, Chem. Biol., 2014, 21,
1102–1124.
5 M. R. Arkin and J. A. Wells, Nat. Rev. Drug Discovery, 2004, 3,
301–317.
6 T. Clackson and J. A. Wells, Science, 1995, 267, 383–386.
7 V. Azzarito, K. Long, N. S. Murphy and A. J. Wilson, Nat.
Chem., 2013, 5, 161–173.
8 P. Buchwald, IUBMB Life, 2010, 62, 724–731.
9 S. K. Ko, H. J. Jang, E. Kim and S. B. Park, Chem. Commun.,
2006, 2962–2964.
Conclusions
For the systematic perturbation of PPIs, we designed and
synthesized tetra-substituted hexahydro-4H-pyrazino[2,1-c]
[1,2,4]triazine-4,7(6H)-diones to mimic the b-turn structure, 10 S. Oh and S. B. Park, Chem. Commun., 2011, 47, 12754–
a key secondary structural motif. Distance calculations of 12761.
energy-minimized conformers and the alignment of model 11 H. Kim, T. T. Tung and S. B. Park, Org. Lett., 2013, 15, 5814–
compound 7 with side chains of a biologically active b-turn 5817.
motif conrmed that our tetra-substituted b-turn mimetic 12 J. Kim, H. Kim and S. B. Park, J. Am. Chem. Soc., 2014, 136,
bicycle 1 was conformationally similar to the natural b-turn
14629–14638.
structure. We then developed a robust synthetic pathway to 13 H. S. G. Beckmann, F. Nie, C. E. Hagerman, H. Johansson,
obtain the b-turn mimetic scaffolds via tandem N-acyliminium
cyclization and constructed a 162-member library of tetra-
Y. S. Tan, D. Wilcke and D. R. Spring, Nat. Chem., 2013, 5,
861–867.
substituted pyrazinotriazinediones as b-turn mimetics with an 14 J. Kim, W. S. Lee, J. Koo, J. Lee and S. B. Park, ACS Comb. Sci.,
average purity of 90% using a solid-phase parallel synthesis 2014, 16, 24–32.
platform. In this synthetic route, a new chiral center was 15 W. Guo, J. A. Wisniewski and H. Ji, Bioorg. Med. Chem. Lett.,
generated in a diastereoselective manner, which was conrmed 2014, 24, 2546–2554.
using 2D NOE. Aer library construction, we examined whether 16 H. Yin and A. D. Hamilton, Angew. Chem., Int. Ed., 2005, 44,
these b-turn mimetic compounds could modulate PPIs and 4130–4163.
identied a series of compounds that could effectively stabilize 17 A. J. Wilson, Chem. Soc. Rev., 2009, 38, 3289–3300.
the LRS–RagD interaction using ELISA-based screening. The 18 A. Whitty and G. Kumaravel, Nat. Chem. Biol., 2006, 2, 112–
cellular interaction between LRS and RagD directly mediated
the leucine-dependent activation of mTORC1 and resulting 19 M. Pelay-Gimeno, A. Glas, O. Koch and T. N. Grossmann,
biological outcomes. The western blot analysis of phosphory- Angew. Chem., Int. Ed., 2015, 54, 8896–8927.
lated S6K1 and FRET imaging in live cell revealed that 5c{3,9} 20 J. Gavenonis, B. A. Sheneman, T. R. Siegert, M. R. Eshelman
stabilized the direct interaction between LRS and RagD and
and J. A. Kritzer, Nat. Chem. Biol., 2014, 10, 716–722.
activated the leucine-dependent mTORC1 signalling pathway in 21 S. M. Biros, L. Moisan, E. Mann, A. Carella, D. Zhai,
118.
living cells. With this mechanism of action, 5c{3,9} can serve as
an excellent research tool for studying the non-canonical role of
J. C. Reed and J. Rebek Jr, Bioorg. Med. Chem. Lett., 2007,
17, 4641–4645.
LRS under leucine-decient conditions and as a lead structure 22 S. Thompson and A. D. Hamilton, Org. Biomol. Chem., 2012,
for examining disease models that require the selective activa- 10, 5780–5782.
tion of anabolic processes for wound healing or the restoration 23 M. Laplante and D. M. Sabatini, Cell, 2012, 149, 274–293.
of muscle loss.
24 Y. Sancak, T. R. Peterson, Y. D. Shaul, R. A. Lindquist,
C. C. Thoreen, L. Bar-Peled and D. M. Sabatini, Science,
2008, 320, 1496–1501.
25 J. M. Han, S. J. Jeong, M. C. Park, G. Kim, N. H. Kwon,
H. K. Kim, S. H. Ha, S. H. Ryu and S. Kim, Cell, 2012, 149,
410–424.
Acknowledgements
We thank Prof. Sunghoon Kim for generously providing
pAmCyan1-N1-LRS and pZsYellow1-N1-RagD plasmids. This
work was supported by the Creative Research Initiative Grant 26 A. Efeyan, W. C. Comb and D. M. Sabatini, Nature, 2015, 517,
(2014R1A3A2030423) and the Bio & Medical Technology Devel- 302–310.
opment Program (2012M3A9C4048780), funded by the National 27 K. Burgess, Acc. Chem. Res., 2001, 34, 826–835.
Research Foundation of Korea (NRF).
28 L. R. Whitby and D. L. Boger, Acc. Chem. Res., 2012, 45, 1698–
1709.
29 L. R. Whitby, Y. Ando, V. Setola, P. K. Vogt, B. L. Roth and
D. L. Boger, J. Am. Chem. Soc., 2011, 133, 10184–10194.
Notes and references
1 A. A. Ivanov, F. R. Khuri and H. Fu, Trends Pharmacol. Sci., 30 A. Glas, D. Bier, G. Hahne, C. Rademacher, C. Ottmann and
2013, 34, 393–400.
2 L. Jin, W. Wang and G. Fang, Annu. Rev. Pharmacol. Toxicol.,
2014, 54, 435–456.
T. N. Grossmann, Angew. Chem., Int. Ed., 2014, 53, 2489–
2493.
31 A. Glas and T. N. Grossmann, Synlett, 2015, 26, 1–5.
2760 | Chem. Sci., 2016, 7, 2753–2761
This journal is © The Royal Society of Chemistry 2016

View Article Online
Edge Article
Chemical Science
32 E. G. Hutchinson and J. M. Thornton, Protein Sci., 1994, 3, 35 B. C. Finzel, P. C. Weber, D. H. Ohlendorf and F. R. Salemme,
2207–2216. Acta Crystallogr., Sect. B: Struct. Sci., 1991, 47, 814–816.
33 S. L. Garland and P. M. Dean, J. Comput.-Aided Mol. Des., 36 M. Eguchi, M. S. Lee, H. Nakanishi, M. Stasiak, S. Lovell and
1999, 13, 469–483. M. Kahn, J. Am. Chem. Soc., 1999, 121, 12204–12205.
34 K. Gunasekaran, L. Gomathi, C. Ramakrishnan, 37 F. Giordanetto, A. Schafer and C. Ottmann, Drug Discovery
J. Chandrasekhar and P. Balaram, J. Mol. Biol., 1998, 284,
1505–1516.
Today, 2014, 19, 1812–1821.
38 X.-J. Zhang, D. L. Chinkes and R. R. Wolfe, J. Nutr., 2004, 134,
3313–3318.
This journal is © The Royal Society of Chemistry 2016
Chem. Sci., 2016, 7, 2753–2761 | 2761