Development of novel cyclic peptides as pro-apoptotic agents

Article abstract of DOI:10.1016/j.ejmech.2016.04.001

Our recent finding that paclitaxel behaves as a peptidomimetic of the endogenous protein Nur77 inspired the design of two peptides (PEP1 and PEP2) reproducing the effects of paclitaxel on Bcl-2 and tubulin, proving the peptidomimetic nature of paclitaxel.

Full text of DOI:10.1016/j.ejmech.2016.04.001

Accepted Manuscript
Development of novel cyclic peptides as pro-apoptotic agents
Margherita Brindisi, Samuele Maramai, Simone Brogi, Emanuela Fanigliulo, Stefania
Butini, Egeria Guarino, Alice Casagni, Stefania Lamponi, Claudia Bonechi, Seema
M. Nathwani, Federica Finetti, Francesco Ragonese, Paola Arcidiacono, Pietro
Campiglia, Salvatore Valenti, Ettore Novellino, Roberta Spaccapelo, Lucia Morbidelli,
Daniela M. Zisterer, Clive D. Williams, Alessandro Donati, Cosima Baldari, Giuseppe
Campiani, Cristina Ulivieri, Sandra Gemma
PII:
S0223-5234(16)30273-2
10.1016/j.ejmech.2016.04.001
EJMECH 8515
DOI:
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 19 February 2016
Revised Date: 31 March 2016
Accepted Date: 1 April 2016
Please cite this article as: M. Brindisi, S. Maramai, S. Brogi, E. Fanigliulo, S. Butini, E. Guarino, A.
Casagni, S. Lamponi, C. Bonechi, S.M. Nathwani, F. Finetti, F. Ragonese, P. Arcidiacono, P. Campiglia,
S. Valenti, E. Novellino, R. Spaccapelo, L. Morbidelli, D.M. Zisterer, C.D. Williams, A. Donati, C. Baldari,
G. Campiani, C. Ulivieri, S. Gemma, Development of novel cyclic peptides as pro-apoptotic agents,
European Journal of Medicinal Chemistry (2016), doi: 10.1016/j.ejmech.2016.04.001.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT
Graphical Abstract
Development of novel cyclic peptides as pro-apoptotic agents
*
Margherita Brindisi, Samuele Maramai, Simone Brogi, Emanuela Fanigliulo, Stefania Butini, Egeria
Guarino, Alice Casagni, Stefania Lamponi, Claudia Bonechi, Seema M. Nathwani, Federica Finetti,
Francesco Ragonese, Paola Arcidiacono, Pietro Campiglia, Salvatore Valenti, Ettore Novellino,
Roberta Spaccapelo, Morbidelli Lucia, Daniela M. Zisterer, Clive D. Williams, Alessandro Donati,
*
Cosima Baldari, Giuseppe Campiani, Cristina Ulivieri, Sandra Gemma.

ACCEPTED MANUSCRIPT
Development of novel cyclic peptides as pro-apoptotic
agents
1
,2
1,2
1,2
3
Margherita Brindisi,
Samuele Maramai,
Simone Brogi,
Emanuela Fanigliulo, Stefania
1
,2,*
1,2
1,2
1,2
1,2
Butini,
Egeria Guarino, Alice Casagni, Stefania Lamponi, Claudia Bonechi, Seema M.
4
2,3
5
5
6
Nathwani, Federica Finetti, Francesco Ragonese, Paola Arcidiacono, Pietro Campiglia, Salvatore
1
,2
1,7
5
2,3
4
Valenti, Ettore Novellino, Roberta Spaccapelo, Lucia Morbidelli, Daniela M. Zisterer, Clive D.
4
1,2
3
1,2,**
3
Williams, Alessandro Donati, Cosima Baldari, Giuseppe Campiani,
Cristina Ulivieri, Sandra
1
,2
Gemma.
1
European Research Centre for Drug Discovery and Development (NatSynDrugs), Siena University,
via Aldo Moro 2, I-53100 Siena, Italy
2
Istituto Toscano Tumori, Siena University, via Aldo Moro 2, I-53100 Siena, Italy
3
Department of Life Sciences, via Aldo Moro 2, I-53100 Siena, Italy
4
School of Biochemistry and Immunology, Trinity Biomedical Sciences Institute, Trinity College
Dublin, 152-160 Pearse Street, Dublin 2, Ireland
5
Department of Experimental Medicine, Università degli Studi di Perugia, P.le Gambuli, I-06132
Perugia, Italy
6
Dipartimento DIFARMA, Università di Salerno, I-84084 Fisciano, Salerno, Italy
7
Dipartimento di Farmacia, University of Naples Federico II, via D. Montesano 49, I-80131 Naples,
Italy
∗
Corresponding author. Tel.: +39 0577234161; e-mail: butini3@unisi.it
1

ACCEPTED MANUSCRIPT
*
*
Corresponding author. Tel.: +39-0577-234172; e-mail: campiani@unisi.it
Keywords: Pro-apoptotic agents; Anticancer agents; Cyclic peptides; Peptidomimetics; Nur77;
Tubulin; Bcl-2; NMR studies; Molecular modelling; Confocal microscopy; FACS analysis.
Abbreviations: ADT, AutoDock Tools; COSY, correlation spectroscopy; DBU, 1,8-
Diazabicyclo[5.4.0]undec-7-ene; DCM, dichloromethane; DHPro, 3,4-dehydro-(L)-proline; DIPEA,
N,N-Diisopropylethylamine; DMEM, Dulbecco's Modified Eagle's medium; DMF, N,N-
dimethylformamide;
DMSO,
dimethyl
sulfoxide;
EDCI,
1-Ethyl-3-(3-
dimethylaminopropyl)carbodiimide; EEDQ, N-Ethoxycarbonyl-2-ethoxy-1,2-dihydroquinoline; FCS,
Fetal Calf Serum; GA, Genetic algorithm; GB/SA, Generalized-Born/Surface-Area; HATU, (1-
[
Bis(dimethylamino)methylene]-1H-1,2,3-triazolo[4,5-b]pyridinium 3-oxid hexafluorophosphate);
HOBt, N-Hydroxybenzotriazole; MCMM, Monte Carlo Multiple Minimum; MTAs, microtubule
targeting agents; NMR, Nuclear Magnetic Resonance; NOESY, nuclear Overhauser enhancement
spectroscopy; OPLS-AA, Optimized Potentials for Liquid Simulations-all atom; PBS, Phosphate-
buffered saline; PRCG, Polak-Ribiere conjugate gradient; PTX, paclitaxel; RBL-2H3, rat basophilic
leukemia cell line; RIPA, radioimmunoprecipitation assay; TEA, trimethylamine; TFA, trifluoroacetic
acid; THF, tetrahydrofuran; TMRM, tetramethylrhodamine methyl ester perchlorate; TOCSY, total
correlation spectroscopy.
Abstract
Our recent finding that paclitaxel behaves as a peptidomimetic of the endogenous protein Nur77
inspired the design of two peptides (PEP1 and PEP2) reproducing the effects of paclitaxel on Bcl-2 and
tubulin, proving the peptidomimetic nature of paclitaxel. Starting from these peptide-hits, we herein
describe the synthesis and the biological investigation of linear and cyclic peptides structurally related
2

ACCEPTED MANUSCRIPT
to PEP2. While linear peptides (2a,b, 3a,b, 4, 6a-f) were found inactive in cell-based assays, biological
analysis revealed a pro-apoptotic effect for most of the cyclic peptides (5a-g). Cellular permeability of
5
a (and also of 2a,b) on HL60 cells was assessed through confocal microscopy analysis. Further
cellular studies on a panel of leukemic cell lines (HL60, Jurkat, MEC, EBVB) and solid tumor cell
lines (breast cancer MCF-7 cells, human melanoma A375 and 501Mel cells, and murine melanoma
B16F1 cells) confirmed the pro-apoptotic effect of the cyclic peptides. Cell cycle analysis revealed that
treatment with 5a, 5c, 5d or 5f resulted in an increase in the number of cells in the sub-G /G peak.
0
1
Direct interaction with tubulin (turbidimetric assay) and with microtubules (immunostaining
experiments) was assessed in vitro for the most promising compounds.
1
. Introduction
Microtubules play a key role in mitosis and as such represent an ideal target for anticancer drugs.[1]
Compounds able to affect microtubule dynamics are referred to as microtubule targeting agents
MTAs). MTAs, which may bind to different domains both on α- and β-tubulin, are classified as: i)
(
microtubule-depolymerizing agents (e.g. colchicine, nocodazole, vinca alkaloids,[2] and PBOX-
compounds[3-6]) and ii) microtubule-polymerizing agents (e.g. taxanes and epothilones).[7] While the
mechanism of action of these drugs is well described with respect to their anti-mitotic function, the
emerging effects of these compounds on cellular process other than mitosis (apoptosis, chemotaxis,
intracellular transport and intracellular signalling) are still under intense investigation.[8] Among
MTAs, taxanes are the most important class of antitumor agents and paclitaxel (PTX, 1, Fig. 1) inhibits
microtubule dynamics by interacting with β-tubulin leading to tubulin stabilization, G
2
/M cell cycle
arrest, and apoptotic cell death.[9] However it is also well known that PTX-induced apoptosis may
occur independently from G
2
/M arrest.[10] Horwitz and coworkers have indeed demonstrated that 1 at
3

ACCEPTED MANUSCRIPT
concentrations lower than that able to induce a G /M arrest induces the formation of an aneuploid G
2
1
cell population.[11] Further, Jordan demonstrated that at low concentrations 1 can block the mitotic
process by stabilizing microtubules and not by altering the mass of polymerized microtubules.[12]
Taxanes, besides targeting β-tubulin, also bind to Bcl-2 and behave as Bcl-2 molecular switchers,
thus inducing apoptosis.[13] Interestingly, the endogenous orphan nuclear receptor Nur77 (NGFI-B,
TR3, NR4A1) also interacts with β-tubulin and Bcl-2 and shares with 1 the same binding site on both
proteins.[13] Starting from this evidence, we demonstrated that taxanes mimic β-tubulin and Bcl-2
binding domains of Nur77.[13] To prove the concept two undecapeptides were built, reproducing
Nur77 sequences 405-406 and 580-587 linked by a glycine (PEP1) or a proline (PEP2, 2a, Fig. 1).
Biological analysis revealed that PEP2 (2a) was slightly more active than PEP1 in mimicking the
biological activities of 1 such as tubulin polymerization and opening of the permeability transition pore
channel in mitochondria (measured by transmembrane potential evaluated in isolated
mitochondria).[13]
Starting from the undecapeptide molecular templates PEP1 and PEP2 (2a), we have recently
embarked in the development of peptidomimetics mimicking the Nur77 binding domain in order to
develop novel pro-apoptotic agents (Table 1). We initially acquired the conformational information on
linear undecapeptides in solution by Nuclear Magnetic Resonance (NMR) using 2b (Fig. 1) as the
water soluble counterpart of 2a. Next, a small series of shorter linear peptides (3a,b, 4, and 6a,e, Fig. 1
and Table 1) and cyclic peptides (5a-g, Fig. 2 and Table 1) were obtained by combining both solid
phase and in solution synthesis. Here we have addressed the ability of the newly developed compounds
and the reference 2a to promote apoptosis in different cell lines.
Starting the from the molecular modeling analysis of 1 and 2a,b with respect of the Nur77 binding
domain we investigated the minimum required sequence for activity with the synthesis of shorter linear
4

ACCEPTED MANUSCRIPT
peptides 3a,b (featuring the PTX-superimposable region) and 4 (Fig. 1). The lack of efficacy in cell
based assays of these latter analogues prompted the development of cyclic peptides (5a-g), with
compound 5f characterized by a relevant reduction of the peptidic nature (Fig. 2 and Table 1).
Cyclic peptides (5a, 5c, 5d or 5f), but not the linear peptides or the reference compound 2a, are able
to induce apoptosis in different leukemic cell lines (Jurkat, MEC, and EBV-B) and in solid tumor cell
lines (MCF-7, A375, 501Mel, and B16F1). Interaction with tubulin was assessed in vitro by
turbidimetric assay and also by immunostaining of microtubules for the most promising compounds.
Since confocal microscopy experiments confirmed cell permeability for both the cyclic peptide 5a and
the linear peptide 2a, we performed mass analysis of cellular lysates for examining the apoptotic effect
of 5a (pro-apoptotic) and 2a (neither pro-apoptotic, nor cytotoxic).
Fig. 1. Reference compounds, paclitaxel, (1) and PEP2 (2a), and outline of the approach to develop
novel PEP2-derivatives 2b, 3a-b, and 4.
5

ACCEPTED MANUSCRIPT
Fig. 2. General structure of the title cyclic peptides 5a-e and peptidomimetics 5f,g. (DHPro = 3,4-
dehydro-(L)-proline).
6

ACCEPTED MANUSCRIPT
Table 1. Chemical structure of the developed peptides 2a,b, 3a,b, 4, 5a-g, 6a-f and type of synthesis.
Type of
synthesis
-
Cmpd
Structure
2
a
Ac-L-V-P-P-P-P-I-V-P-Y-F-CONH
Ac-L-V-P-P-P-P-I-V-P-Y-F-COOH
2
2
b
Solid phase
Solid phase &
in solution
chemistry
Solid phase
Solid phase &
in solution
chemistry
3
a
Ac-P-P-I-V-P-Y-F-CON(Me)
Ac-P-P-I-V-P-Y-F-COOH
2
3
b
4
Ac-V-P-P-P-P-I-CONH
2
Solid phase &
in solution
chemistry
5
a
Solid phase & in
solution
5
b
chemistry
Solid phase & in
solution
5
c
chemistry
Solid phase &
in solution
chemistry
5
d
7

ACCEPTED MANUSCRIPT
Solid phase &
in solution
chemistry
5
e
In solution
chemistry
5
f
In solution
chemistry
5
6
g
a
NH
NH
NH
NH
2
2
2
2
(CH
(CH
(CH
2
2
2
)
)
)
3
3
2
CO-P-P-I-V-P-Y-F-COOH
CO-P-P-I-V-G-Y-F-COOH
CO-P-P-I-V-G-Y-F-COOH
Solid phase
Solid phase
Solid phase
Solid phase
Solid phase &
in solution
chemistry
6
b
6
c
6
d
-G-P-P-I-V-G-Y-F-COOH
6
e
NH
NH
2
2
(CH
2
2
)
3
2
CO-P-P-I-V-(DH)P-Y-F-COOH
O(CH NHCO-I-V-(DH)P-Y(Bn)-F-COOH
In solution
chemistry
6
f
(CH
)
)
2 2
2
. Results and discussion
2
.1. Chemistry
The linear and the cyclic peptides object of this study were synthesized by combining solid phase and
in solution synthesis. Table 1 summarizes all the linear and cyclic peptides herein described and the
type of synthesis employed for their development. Compounds 2a,b, 3a,b, 4, and 6a-e were
synthesized by solid phase synthesis under microwave irradiation (Schemes 1-3) by applying a
classical Fmoc chemistry while 6f was synthesized as described in Scheme 4, entirely by in solution
8

ACCEPTED MANUSCRIPT
chemistry. The cyclic peptides 5a-g were obtained from their linear precursors by classical in solution
chemistry using the appropriate condensing agent (Schemes 3 and 4).
Scheme 1. Synthesis of the linear undecapeptides 2a and 2b. Reagents and conditions: (a) TFA/(i-
Pr) SiH/H O 95:2.5:2.5, 25 °C, 3 h; (b) EEDQ, NH HCO , MeCN, 25 °C, 12 h.
3
2
4
3
9

ACCEPTED MANUSCRIPT
Scheme 2. Synthesis of the “disconnected” derivatives 3a,b and 4. Reagents and conditions: (a)
TFA/(
i
-Pr)
3
SiH/H
2
O 95:2.5:2.5, 25 °C, 3 h; (b) EDCI, HOBt, DIPEA, NH(Me)
2
·HCl, dry DCM, 0 °C
4 3
to 25 °C, 12 h; (c) EEDQ, NH HCO , MeCN, 25 °C, 12 h.
tBu-O
O
O
H
O
O
O
O
O
H
N
H
N
N
N
N
O
Fmoc
N
N
AA
N
NH2
O
( )n
H
H
O
O
(a)
HO
O
(b)
O
5a-e
O
O
O
H
H
N
N
N
N
NH3+ CF CO
-
HO
N
N
AA
N
3
2
O
( )n
H
H
O
O
6
6
6
6
6
a, AA = Pro, n = 3
b, AA = Gly, n = 3
c, AA = Pro, n = 2
d, AA = Pro, n = 1
e, AA = DHPro, n = 3
HO
I
(f)
(
c)
(d)
(e)
CO Me
N
CO Me
N
CO Me
N
N
CO Me
N
Fmoc
11
g)
CO Me
2
2
2
2
2
Cl H
2
Boc
Boc
Boc
9
7
8
10
(
(g)
(h)
(i)
N
CO H
N
CO H
N
CO H
2
2
2
Cl H
2
Boc
Fmoc
12
13
14
Scheme 3. Synthesis of the cyclic peptides 5a-e, their linear precursors 6a-e, and the dehydroproline
4. Reagents and conditions: (a) TFA/( -Pr) SiH/H O 95:2.5:2.5, 25 °C, 3 h; (b) HATU, HOBt,
1
i
3
2
DIPEA, dry DCM, 25 °C, 12 h; (c) MeI, PPh , DIAD, dry THF, 0 °C to 25 °C, 12 h; (d) DBU, dry
3
toluene, 85 °C, 3 h then 25 °C, 12 h; (e) AcCl, MeOH, 40 °C, 15 min; (f) Fmoc-Cl, TEA, dry DCM, 0
°
C to 25 °C, 12 h; (g) LiOH, THF, MeOH, H O, 25 °C, 3 h; (h) TFA, dry DCM, 25 °C, 2 h; (i) Fmoc-
2
Cl, Na CO , 1,4-dioxane, 25 °C, 12 h.
2
3
1
0

ACCEPTED MANUSCRIPT
The synthesis of the linear peptides outlined in Schemes 1, 2, and 3 was performed following a solid
phase approach, starting from a Wang resin functionalized with
Fmoc methodology in a CEM Liberty Automated Microwave Peptide Synthesizer. The peptides were
released from the solid support by treatment with a mixture of TFA/( -Pr) SiH/H O (95:2.5:2.5). The
free carboxylic acid moiety of peptide 2b was then converted into the corresponding primary amide
2a) by using EEDQ in presence of NH HCO (Scheme 1) The same protocol was followed for the
N-Fmoc-(L)-Phe, and using a standard
i
3
2
(
4
3
.
synthesis of peptide 4 while peptide 3a was converted into the corresponding dimethylamide derivative
b, by classical coupling reaction with the hydrochloride salt of dimethylamine (Scheme 2). Scheme 3
describes the synthesis of 6a-e and their cyclic counterpart 5a-e. The peptides 6a-e, obtained as
3
-5
described for 2b, were dissolved in dry dichloromethane (DCM) at a concentration of 10 M and were
cyclized by using HATU and HOBt as coupling agents. The -Fmoc protected -dehydroproline 14
N
S
needed for the synthesis of compound 6e was prepared following the procedure reported in Scheme 3.
Starting from the commercially available 4-hydroxy proline 7, Mitsunobu reaction allowed its
transformation into the corresponding iodinated derivative 8 which underwent an elimination reaction
in the presence of DBU, to give the
needed for the synthesis of 5f) the protection as 9-fluorenylmethyl carbamate was not successful and
only afforded traces of the ester 11. Thus, compound 14 was obtained by and alternative route where
N-Boc protected S-dehydroproline 9. After Boc removal (10,
N
the hydrolysis of ester 9, with an aqueous solution of LiOH (12), followed by Boc removal (13) and N-
Fmoc protection led to the desired intermediate in good yield.
1
1

ACCEPTED MANUSCRIPT
2
Scheme 4. Synthesis of the cyclic peptides 5f and 5g. Reagents and conditions: (a) H , Pd/C, MeOH, 2
h; (b) Di-tert-butyl dicarbonate, TEA, dry DCM, rt, 12 h; (c) (
L)-Ile-OBn, COCl 20% in toluene,
2
NaHCO ss, DCM, 0 °C to rt, 15 min then TEA, dry DCM, rt, 12 h; (d)
N
-Boc-(
L)-Valine for 20,
3
EDCI, HOBt, DIEA, dry DCM, 0 °C to rt, 12 h; (e) LiOH, THF, MeOH, H O, rt, 3 h; (f) AcCl, MeOH,
2
4
0 °C, 15 min; (g) TFA, DCM, rt, 2 h; (h) HATU, HOBt, DIEA, dry DCM, 0 °C to rt, 12 h; (i) BCl3,
dry DCM, -78 °C to rt, 12 h.
The synthesis of compounds 5f and 5g was performed as outlined in Scheme 4. The diazidoethylether
1
5, synthesized according to a reported procedure,[14] was subjected to a catalytic hydrogenation,
followed by protection of the amine functionality with di-tert-butyl dicarbonate, and the obtained
mono-protected intermediate 16 underwent a second reduction to afford the free amine 17. This latter
1
2

ACCEPTED MANUSCRIPT
was used as the nucleophile in the reaction with the isocyanate synthesized from H N-
L-Ile-OBn with
2
phosgene, for obtaining the intermediate urea 18 which was debenzylated by catalytic hydrogenation.
The unnatural aminoacid -dehydroproline 10, obtained as reported in Scheme 3,[15] was employed in
a classical in-solution coupling reaction with -Boc-( )-Val to afford dipeptide 20, which was
L
N
L
converted into the corresponding free acid (21) by hydrolysis with aqueous LiOH. In parallel the
hydrochloride salt of the dipeptide 22 was obtained after coupling of the appropriately orthogonally
protected amino acids (Boc-L-Tyr(OBn)-OH and NH2-L-Phe-OMe) and deprotection of the amine
functionality, by a methanolic solution of HCl. The dipeptides 21 and 22 were classically coupled, in
the presence of EDCI and HOBt as condensing agents, and deprotection of the obtained compound
afforded amine 23 to be used in the last coupling reaction with the previously synthesized intermediate
1
9. The obtained linear compound 24, after deprotection at both the C and N terminal ends (with
aqueous LiOH, and with TFA respectively) yelded 6f that was submitted to the cyclization reaction
with HATU and HOBt as coupling agents. The obtained cyclic peptide 5f was then treated with BCl to
3
afford the debenzylated compound 5g.
2
.2. Conformational flexibility of linear peptide 2a versus the cyclic peptide 5a: NMR and
computational studies
In order to define the key structural features that could be relevant for the activity of our lead 2a, we
performed a detailed NMR structural analysis aiming at clarifying its structure in solution and we
compared the data with the mimicked portion of Nur77. To perform NMR studies, we synthesized 2b
as a water soluble counterpart of 2a. Preliminarily a primary 3D peptide structure of 2a from sequence
information was generated by submitting its amino acid sequence to PEP-FOLD [16, 17] and PEPstr
[
18] servers. The sequence analysis performed by means of PEP-FOLD resulted in six clusters of
1
3

ACCEPTED MANUSCRIPT
structures all characterized by similar energies, thus indicating the convergence towards affordable
conformations (Fig. S1 of the Supplementary Information (SI)).
As expected for an undecapeptide lacking secondary structure elements, we got divergent results by
applying PEPstr analysis. PEPstr predicts a peptide secondary structure by applying the PSIPRED
algorithm, [19, 20] and it generates a starting conformation with
φ
and
ψ
dihedral angle values
energy
corresponding to the previously predicted secondary structure. Using AMBER 14[21]
,
minimization and molecular dynamics simulations of the initial conformation were performed. This
analysis revealed a more “extended” conformation of 2a compared to the corresponding Nur77 region.
Our data revealed that 2a conformation also embeds a type IV β-turn in the poly-prolyl region (Fig.
S2).
This conformational analysis was essentially confirmed by NMR studies. As reported in the
Experimental Section 2b spectra were acquired in D O and H O/D O as solvent. All the proton
2
2
2
resonances were assigned using a combination of correlation spectroscopy (COSY), total correlation
spectroscopy (TOCSY) and nuclear Overhauser enhancement spectroscopy (NOESY) experiments.
Residue-specific assignment was achieved from TOCSY spectra, by recognizing the unique spin
system for each residue. For most of the residues, starting from the amidic proton, it was possible to
reconstruct the connectivity pathway, by comparing the TOCSY and the COSY spectra (Fig. S3).
Complete assignment is given in Table S1.
The main feature of NMR conformational analysis is the accurate estimation of structural restraints.
In fact, NOE intensities and coupling constants (
respectively.
J) account for distance and torsional restraints
1
4

ACCEPTED MANUSCRIPT
Overall our NMR analysis revealed for 2b a random coil structure, indicating the need of analogues
with reduced conformational flexibility to address the binding mode. Consequently, the cyclic
conformationally constrained peptides typified by 5a were synthesized.
A total of 82 and 99 peak volumes were integrated from the NOESY spectra of 2b and 5a
respectively. In the spectrum of 2b a region with opposite-phased peaks (compared to all the other
peaks) was observed (Fig. S4). The related resonances belong to the Phe11 residue; this accounted for a
distinct flexibility of this residue with respect to the overall tumbling rate of the peptide.
3
The
JNHα for both 2b and 5a were directly measured from the splitting observed on the amide proton
resonances. All the NH groups of 2b showed similar coupling constants in the range of 6÷7 Hz (Table
3
2
), indicating a random coil structure. In fact, small linear peptides usually present averaged JNHα
values of 7.4 Hz, indicating that the peptide lacks a single stable conformation. On the contrary, with
3
5
a the observed JNHα values (Table 2) strongly suggested that the molecule adopts a preferred
conformation.
3
Table 2. Measured
JNHα for 2b and 5a amide hydrogen atoms. All the values are characteristic of
random coil dihedral angles, suggesting the absence of a single favored conformation.
2
b
3
5a
Residue
Spacer 1-a
Spacer 1-b
Ile3
3
J
NHα
J
NHα
Residue
Leu1
Val2
Ile7
Val8
6.58
7.67
7.45
7.67
7.23
7.89
1.53
9.43
5.70
Val3
Tyr6
Phe7
overlapped
6.58
Tyr10
Phe11
9.57
To further confirm the low structural flexibility of 5a compared to 2a,b, we performed temperature
3
dependence experiments. This analysis demonstrated that the
JNHα coupling constants were not
affected by temperature variation, thus indicating that 5a adopts a particularly stable conformation.
1
5

ACCEPTED MANUSCRIPT
Notably, at 323 K and 328 K, the spacer amide peak of 5a did not overlap to aromatic resonances, an
3
event that allowed us to identify the two well-defined J
diasterotopic γ methylenes hydrogen atoms of the spacer).
constants (between the NH and the two
NHγ
H-bond is one of the major interactions contributing to the stabilization of secondary structures in
proteins, and delineation of H-bonds is relevant for assessing peptide conformations. Accordingly, H-
bond network involving amide protons has been defined by temperature dependence experiments.
From NH chemical shift analysis, we could identify the exposed or shielded (H-bonded) NH in our
peptide sequences. In fact, amide shielded protons show a lower temperature dependence of their
chemical shifts (usually defined as δ(ppm)/T(K) ratio or Temperature coefficient, Tc) compared to
exposed protons.[22] A |Tc| < 3 ppb/K is indicative of intra-molecular H-bond formation of the
shielded protons, whereas a |Tc| > 5 ppb/K likely represents exposed amide protons.[22, 23]
Fig. 3. Temperature dependence of the chemical shifts of the amide protons of 2b (left) and 5a (right).
The quantitative plot of the analysis of these data (Fig. 3) further underlines the different behavior of
2
b and 5a amide H atoms.
The Tc were calculated for each residue and defined as the slope of the best-fit line (Table S2).
Concerning 5a, amide NH protons of the spacer, Val5 and Phe8 residues, are likely intra-molecular H-
1
6

ACCEPTED MANUSCRIPT
bonded in the majority of conformers. On the contrary, Tc values from 2b amides were all
characteristic of solvent exposed hydrogen (|Tc| > 5ppb/K), with the only exception of Phe11 NH (|Tc|
=
4 ppb/K). In the latter case, this intermediate Tc value did not unambiguously indicate the presence of
a H-bond. For random coil peptides in aqueous medium (as 2b) the Tc parameter is often
unreliable.[22, 24] However, since Phe11 amide peak of 2b was also the only NH peak which did not
undergo line broadening at high temperatures, the hypothesis of a H-bond involving this residue could
not be completely ruled out; alternatively we can hypothesize a solvent protection of Phe11 amide for a
minor population of conformers of 2b. Moreover, the narrow resonances recorded on the NMR spectra,
at variable concentrations, suggest that the linear peptide did not undergo aggregation.
2
.3. Molecular Modelling Studies.
In order to understand the potential ability of our lead cyclic peptide (5a) to mimic the region of
Nur77 which can bind Bcl-2 and β-tubulin[13] (see Fig. 4A), we initially performed an extensive
conformational search analysis by means of MacroModel,[25] to evaluate the possibility of lower
energy conformers of 5a to form an intramolecular H-bond as highlighted by NMR studies and which
reproduces the same H-bond as evidenced in Nur77 X-ray structure (PDB ID 1YJE) (Fig. 4A,B).
Fig. 4. (A) Crystal structure of Nur77 (PDB ID 1YJE) highlighting the β-tubulin and Bcl-2 interacting
portion represented as cyan sticks; (B) Low energy conformer (-542.454 kcal/mol) of 5a represented by
magenta sticks. Intramolecular H-bonds are represented by dotted lines. The pictures were generated by
means of PyMOL.[26]
1
7

ACCEPTED MANUSCRIPT
Our studies highlighted a high degree of 3D arrangement similarity between the lower energy
conformer of 5a (Fig. 4B) and the Nur77 binding region. In fact, the region of Nur77 where the H-bond
is detected (PDB ID 1YJE) (between V587 and P584 mouse numbering) is superimposable to the
corresponding area of the low energy conformer of 5a where the intramolecular H-bond between Pro7
and Val4 was proved by NMR analysis. This feature allows 5a to reproduce the relevant motif of
Nur77 for binding its targets.[13]
Since we already demonstrated that 1 mimics the Nur77 binding region,[13] we hypothesized that the
cyclic peptide 5a might share (although partially) with 1 the same binding site on β-tubulin. To confirm
this hypothesis we have performed a blind docking calculation considering both α- and β-tubulin
subunits by using Autodock.[27-32] Autodock software is able to accomplish the blind docking of
compounds and to select the correct complexes based on energy without prior knowledge of the
binding site. The output of this approach (Fig. 5) clearly shows that the preferred binding site for the
lower energy poses of 5a overlaps to that of 1. On the β-tubulin subunit, no poses were found in the
colchicine binding site, while a very limited number of solutions was found matching with the
vinblastine binding site, the analysis of which revealed much less favourable docking scores when
compared to those found for 5a into the PTX-binding site. Accordingly, the vinblastine binding site
appeared not suitable for 5a/β-tubulin interaction. Notably, a very limited number of docked solutions
were found on α-tubulin. These poses did not cover the binding site of known drugs (pironetin, and
griseofulvin) and, being devoid of relevant docking scores, allowed us to exclude the α-tubulin subunit
as a putative target for 5a. Overall, these findings indicate the PTX-binding site on β-tubulin as the
most reliable binding site for 5a.
1
8

ACCEPTED MANUSCRIPT
Fig. 5. Blind docking output for 5a (coloured sticks) against α/β-tubulin (orange and white surfaces,
respectively). In the picture the known tubulin binding sites have been reported. The picture was
generated by Autodock.[32]
Additional computational studies were performed for investigating the influence of the cycle size on
the binding mode of a set of cyclic peptides when gradually shortening the butyric chain of 5a (as in
compounds 5c, and 5d, Fig.1 and Table 1). The potential binding modes of compounds 5a, 5c, and 5d
on β-tubulin were then analysed by means of GOLD software that employs genetic algorithm[33-35]
on the binding site of 1. These studies were coupled to the estimation of free binding energy (Prime
MM/GBSA[36] implemented in Maestro suite[37]) as previously reported by us[38-42] also for
evaluating the binding modes of peptide-like compounds.[43, 44] The output of this calculation is
reported in Fig. 6A-D for 5a, 5c, 5d, and 5f. Analysis of the data revealed that 5a interacts within its
1
9

ACCEPTED MANUSCRIPT
putative binding site through a series of polar and hydrophobic contacts. In particular, we observed a
series of H-bonds with R278 and T276, while a cation-π stacking could be established between the Tyr
residue of 5a and R369 while the closer Pro was easily accommodated into a hydrophobic sub-pocket
formed by residues F272, P360 and L371. The Phe residue of 5a was involved in a π-π stacking with
H229. These major contacts might support a good fitting of 5a in the PTX-binding site. Notably, as a
further validation of this data, the same pose was found as the top-ranked pose during the blind docking
calculation. The superposition of the pose of 5a from the blind docking (Autodock) and the docking
calculation (GOLD) is provided in Fig. S5. The used protocol was able to reproduce the crystallized
pose of 1 (PDB ID 1JFF) further supporting our computational approach.
For compound 5c, the obtained docking output (Fig. 6B) was slightly different from that found for
5
a. Particularly, 5c was found to be more deeply accommodated into the binding pocket. This may be
due to a strong H-bond network found between R369 and the Tyr residue of 5c which in turn allowed
an increase of contacts especially in terms of hydrophobic interactions. In fact, in addition to the H-
bonds and the hydrophobic contacts above mentioned for 5a, 5c was also able to establish a π-π
stacking with F272 by its Tyr residue. Moreover the Pro closer to the Tyr residue is involved in a series
of hydrophobic contacts with P360 and V23. The two other Pro residues were found to form
hydrophobic contacts with P274, L286, and L371.
2
0

ACCEPTED MANUSCRIPT
Fig. 6. (A) 5a (magenta sticks) (B) 5c (light pink sticks) (C) 5d (gold sticks) and (D) 5f (green sticks)
into the β-tubulin PTX-binding site (cyan cartoon). The residues located in the binding site are
represented by lines. H-bonds are represented by dotted lines. Nonpolar hydrogens were omitted for the
sake of clarity. The pictures were generated by PyMOL.
A progressive restriction of the ring dimension led to compound 5d (Fig. 6C) for which our
computational studies revealed a different orientation of the docked poses belonging to the most
populated cluster. In fact the aromatic moieties of 5d were accommodated at the boundary of the
binding site, strongly interacting by H-bonds with R278, G279, and Q281 by its Tyr portion. The Phe
portion established a cation-π stacking with R278. Moreover H-bond with H229 and hydrophobic
interactions of Ile moiety (5d) stabilized the docked pose. In fact, Ile moiety is involved in a series of
2
1

ACCEPTED MANUSCRIPT
hydrophobic contacts with P360, F272, and V23, while the Pro residues closer to Phe portion
established hydrophobic contacts with L217, L219 and L275. Compounds 5f and 5g were conceived to
further reduce the peptide character of our cyclic compounds. Two further proline residues and the
spacer of 5a were replaced by an ethereal chain linked by an ureidic bond to the Ile5, while maintaining
the DHPro into the structure. Notably, as shown in Fig. 6D, 5f spans the entire binding site on tubulin,
interacting with the backbone of G370 by H-bond, while its benzylated Tyr residue enables a double
cation-π stacking with R278. Furthermore, the central region of 5f produced a series of hydrophobic
interactions principally by its Phe moiety which is deeply located in a hydrophobic sub-pocket formed
by P274, A285 and L286. Moreover, the Phe lateral chain of 5f may form another cation-π stacking
with R284. In compound 5g the lack of the benzyl group did not allow to establish a binding mode as
that found for 5f. In fact, the interactions around R278 were completely lost by 5g and these contacts
were not replaced by other interactions. These features did not allow 5g to occupy the entire binding
cleft thus resulting in a relevant solvent exposed portion of the compound and in a large number of
clustered docking solutions. This event produced an inconsistent binding mode which is nonetheless in
agreement with the poor pro-apoptotic potential of the compound (see Table S4). In general all the
computational studies are in agreement with the cellular data reported in Table S4 Table 3 and Figures
7
and 8.
2
.4. Cellular studies
2
.4.1. Cyclic peptides 5a-d,f promote apoptosis in cancer cell lines
Preliminary flow cytometric sub-G /G analysis EBV-B (Epstein-Barr Virus (EBV)-immortalized B
0
1
cells), Jurkat (human T lymphocyte), and HL60 (human promyelocytic leukemia) cell lines treated
under identical conditions with either 5a and 2a (25 µM) evidenced that the cyclic peptide but not the
linear peptide was able to induce apoptosis after 24 h of incubation (Table S3). Following this initial
2
2

ACCEPTED MANUSCRIPT
screening, and in order to understand if the novel compounds might promote apoptosis,
intranucleosomal DNA fragmentation, a major hallmark of apoptotic cell death, was determined by
flow cytometry [45] on HL60 cells treated with both cyclic (5a-g) or linear (2a,b and 6a-f) peptides
(
Fig. 7 and Table S4). The results show that cyclic peptides 5a,c,d,f, but not 5b, 5e and 5g or linear
peptides, induced a significant enhancement in the percentage of apoptotic cells (with hypodiploid or
‘sub-G /G ’’ DNA content) compared with vehicle (Fig. 7A). In fact, compound 5e, which
‘
0
1
incorporates the unnatural amino acid 3,4-dehydro-(L)-proline (DHPro) into its structure in place of
Pro3 of 5a, almost completely lost its pro-apoptotic potential (Table S4). Moreover, in agreement with
our computational studies, 5f improved the pro-apoptotic effect of our lead 5a while no appreciable
activity was observed with 5g (Table S4).
Interestingly, differently from 1, treatment with compounds 5a, 5c, 5d and 5f resulted in a higher and
dose dependent percentage of cells in sub-G
0 1
/G phase which was paralleled by a significant reduction
in the percentage of cells in G /G and G /M phases confirming that these novel compounds induce cell
0
1
2
death by apoptosis (Fig. 7A). The efficacy of these compounds, when tested at 50 µM for 24 h, ranged
from 33.1% induction of apoptotic cells for compound 5c to the 77% for compound 5f (Table S4 and
Fig. 7A) when compared to untreated cells.
Interestingly our prototypical compound 5a, which behaved as one of the best performing cyclic
peptides, was found to induce apoptosis in various tumor cell lines (Table 3) thus indicating a broad-
spectrum activity.
Consistent with the pro-apoptotic effect of 5a, 5c, 5d, and 5f on HL60 cells, externalization of
phosphatidylserine, which normally resides in the inner plasma membrane of healthy cells, was
significantly enhanced as compared to cells treated with vehicle, as assessed by flow cytometric
analysis of Annexin-V (Fig. 7A).
2
3

ACCEPTED MANUSCRIPT
Mitochondria are at the crossroads of the two main apoptotic signaling pathway, namely the intrinsic
and extrinsic pathways, and their integrity is moreover tightly controlled by the Bcl-2 family of
proteins.[46-48] To understand if 5a, 5c, 5d and 5f induce apoptosis by targeting mitochondria we
measured the dissipation of mitochondrial inner transmembrane potential (∆ Ψ m) by flow cytometry
in tetramethylrhodamine methyl ester perchlorate (TMRM) loaded cells. A significant increase in the
percentage of cells with loss of mitochondrial inner transmembrane potential was found following
treatment with cyclic peptides 5a, 5c, 5d and 5f compared with vehicle (Fig. 7B) indicating that these
novel compounds induced apoptosis by targeting the mitochondria.
Collectively, the results provide evidence that cyclic peptides 5a, 5c, 5d and 5f are able to induce cell
death by apoptosis through a mitochondrial dependent pathway with 5a and 5f being the most active of
the series.
To further asses the efficacy profile of this newly identified class of pro-apoptotic agents our
prototypical compound 5a, and some analogues, were also tested on small set of solid tumor cell lines
(
Fig. 8). Flow cytometry analysis of the MCF-7 breast cancer cell lines treated with 5a (50 µM) for 24
h, showed a modest pro-apoptotic effect (11.8% apoptosis compared to 8.5% in vehicle-treated cells).
This was slightly lower than the level of apoptosis observed when the same cell line was treated with
compounds 5c and 5d (18.9% and 17.3% apoptosis, respectively). Compound 2a was inactive when
tested on the MCF-7 cell line.
The effect of 5a and 2a was also evaluated in murine and human melanoma cell lines. Flow
cytometric analysis of both B16F1 (less metastatic) and B16F10 (more metastatic) mouse melanoma
cells incubated with 5a (50 µM) for 24 h revealed a small percentage of apoptotic cells (as sub-G
) when compared to the respective controls (Table 3). Furthermore, flow cytometric analysis of two
human melanoma cell lines (the less metastatic A375 and more metastatic 501Mel cell line) treated
0 1
/G
%
2
4

ACCEPTED MANUSCRIPT
with 2a and 5a, showed that 2a was inactive in both cell lines, when tested at 100 µM, while 5a
displayed significant pro-apoptotic effects in the less metastatic A237 cell line eliciting 48% apoptosis
when tested at 50 µM with an incubation time of 72 h (Table 3).
2
5

ACCEPTED MANUSCRIPT
Fig. 7. Cyclic peptidomimetics induce apoptosis in HL60 cells. (A) Cells were treated with 50 µM
2
a,b, 5a-d,f for 24 h and DNA content was measured by flow cytometry. Control samples were treated
with vehicle (0.1% DMSO). Alternatively cells were treated with 1, 10, 25, 50 and 100 µM 5a-d,f. The
2
6

ACCEPTED MANUSCRIPT
bar graphs show the percentage of cells in the sub-G /G phase. Data are mean ± SD (n > 3), ****p<
0
1
0
.0001 5a vs vehicle, ***p= 0.0002 5c vs vehicle, ***p= 0.0001 5d vs vehicle, ****p< 0.0001 5f vs
vehicle. Representative histograms are shown. (B) Quantitation by flow cytometry of mitochondrial
integrity in TMRM-loaded cells treated with 50 µM 5a-d,f or with vehicle (0.1% DMSO) for 24 h. The
bar graphs show the percentage of cells with depolarized mitochondria (TMRM low) cells. Data are
mean ± SD (n > 3). **p = 0.0073 5a vs vehicle, *p = 0.016 5c vs vehicle, ***p = 0.0004 5d vs vehicle,
*
**p = 0.0008 5f vs vehicle. Quantitation by flow cytometry of Annexin V/PI staining of cells treated
with 50 µM 5a-d,f or with vehicle (0.1% DMSO) for 24 h. The bar graphs show the percentage of
+
Annexin V cells. Data are mean ± SD (n > 3). ****p < 0.0001 5a vs vehicle, **p = 0.011 5c vs
vehicle, ***p = 0.0002 5d vs vehicle, ****p < 0.0001 5f vs vehicle. Representative dot plots are
shown.
Fig. 8. Effect of cyclic peptides 5a,c,d on apoptosis in breast carcinoma cells. MCF-7 cells were treated
with vehicle (0.1% DMSO) or 50 ꢀM 5a, 5c, or 5d for 48 h. The linear peptide 2a was found inactive
when tested under the same experimental conditions. Their DNA content was then analysed as
described in the Experimental Section. Cells with less than 2N DNA content were deemed apoptotic.
Values represent the mean ± SEM for three independent experiments.
Table 3. Efficacy of compounds 2a and 5a (at 50 µM concentration) on different melanoma cell lines
as % of apoptosis (sub-G /G ).
0
1
Apoptosis %
2a
Cmpd→
Cell line↓
B16F1
B16F10
A375
Vehicle
5a
b
1.9±1.6
3.7±3
7±1
NAb
13.9±5.4
13.6±1.8
NAb
NA
c
48±0.5c
b
5
01Mel
9±1.8
NA
17±1.7
a
b
c
vehicle (0.1% DMSO); NA: not active; tested at 50 µM with 72 h of incubation time. All the
values are the mean of at least three independent experiments.
2
.4.2. Both the cyclic peptide 5a and the linear peptide 2a enter into the cells and do not undergo
metabolic inactivation.
2
7

ACCEPTED MANUSCRIPT
The differing efficacy of the cyclic versus the linear peptides was further investigated, by means of
confocal microscopy and mass analysis of cell lysates, in order to understand whether the observed
difference in cellular efficacy might be dependent upon cell permeability and/or on different metabolic
stability of the (internalized) compounds.
Confocal microscopy analysis capitalized on the fact that these novel compounds behaved as
fluorescent probes (with two emission maxima around 340 and 500 nm, not shown) thus allowing us to
follow their cell permeability in confocal microscopy studies on HL60 cells without any derivatization.
These experiments clearly showed that the cyclic peptide 5a and the linear analogues 2a and 2b were
all cell permeable (Fig. 9). We hypothesized that the lack of cellular effects of the linear peptides could
be due to metabolic susceptibility. In fact, the use of peptides in the cellular environment is plagued by
unwanted and rapid proteolysis exerted by exo- and endo-peptidases.[49] These enzymes are an
integral part of the protein recycling machinery of cells and ensure that most peptides are rapidly
hydrolyzed to their amino acid constituents. To address this issue we incubated 5a and 2b in HL60
cells at different concentrations and incubation times (6 h, 24 h and 48 h). Mass spectra analysis of the
cellular lysates evidenced that both compounds were present inside the cells (Fig. S6), thus confirming
cell permeability of these analogues, and also showed that both compounds were found within the cells
up to 48 h of incubation. Although not quantitative, this methodology allow us to exclude that
inefficacy of the linear undecapeptide 2a could be ascribable to its more rapid metabolism as compared
to 5a.
2
8