Synthesis, characterization and anti-cancer activity of a peptide nucleolipid bioconjugate

Article abstract of DOI:10.1016/j.bmcl.2016.06.020

The synthesis, characterization and anti-cancer activity of a novel peptide nucleolipid bioconjugate is reported in this study. The prerequisite 5′-carboxy derived nucleolipid was synthesized following a five-step solution-phase approach and then coupled

Full text of DOI:10.1016/j.bmcl.2016.06.020

Bioorganic & Medicinal Chemistry Letters xxx (2016) xxx–xxx
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Synthesis, characterization and anti-cancer activity of a peptide
nucleolipid bioconjugate
Niki Rana a, Suiying Huang b,c, Pradeepkumar Patel a, Uri Samuni b,c, David Sabatino a,
⇑
a Department of Chemistry and Biochemistry, Seton Hall University, 400 South Orange Avenue, South Orange, NJ 07079, United States
b Department of Chemistry, Queens College, City University of New York, 65-30 Kissena Blvd, Flushing, New York 11367, United States
c The Ph.D. Programs in Chemistry and Biochemistry, The Graduate Center of the City University of New York, New York, NY 10016, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
The synthesis, characterization and anti-cancer activity of a novel peptide nucleolipid bioconjugate is
reported in this study. The prerequisite 50-carboxy derived nucleolipid was synthesized following a five-
step solution-phase approach and then coupled to the cytotoxic D-(KLAKLAK)2 sequence by solid-phase
bioconjugation. The biophysical and structural properties of the peptide-nucleolipid bioconjugate were
evaluated and compared to the peptide controls. These characterization studies revealed that the amphi-
philic peptides favored helical-type secondary structures and well-defined nanoparticle formulations that
were found to be contributive towards their biological activity. The peptide-nucleolipid bioconjugate
displayed greater lethality in comparison to the native D-(KLAKLAK)2AK sequence when treated within
the human A549 non-small cell lung carcinoma cell line. Thus, the amphiphilic peptide-nucleolipid forms
a new class of anti-cancer peptides that may be developed into promising leads in the fight against cancer.
Ó 2016 Elsevier Ltd. All rights reserved.
Received 30 March 2016
Revised 31 May 2016
Accepted 8 June 2016
Available online xxxx
Keywords:
Nucleolipid
Amphiphilic peptides
Peptide bioconjugate
Anti-cancer activity
The so-called ‘‘killer” peptide sequence D-(KLAKLAK)2 has been
originally designed and developed as an antimicrobial agent.1 This
peptide sequence was found to maintain selective cytotoxicity
towards gram positive and negative bacteria, but not in
mammalian cell types. The chemical basis for its selectivity has
been attributed to its poly(cationic) amphiphilic nature, which
facilitates cell translocation across the negatively charged bacterial
membrane, but with limited permeability across the zwitterionic
lipid bilayer of mammalian cell types.2 Once internalized within
cells, the positively charged D-(KLAKLAK)2 sequence has been
found to accumulate on the surface of the mitochondria causing
dissipation of the negatively charged mitochondrial membrane
potential.3 This depolarization resulted in a loss of mitochondrial
integrity which caused membrane rupture and the release of mito-
chondrial contents into the cytoplasm. The release of the cell death
effectors from the mitochondria, including cytochrome c, the sec-
ond mitochondrial-derived activator of caspase (Smac/DIABLO)
and the apoptosis-inducing factor (AIF) ultimately resulted in pro-
grammed cell death.4 In an effort to exploit the cell death effects of
D-(KLAKLAK)2, modifications to the primary sequence5 and conju-
gation with cell penetrating/targeting peptides and proteins6–14
have resulted in improved cell translocation and cytotoxic activity
within malignant mammalian cell types, such as in human cancers.
In spite of these successful examples, effective methods for the safe
and long-lasting administration of peptide based therapeutics are
still in widespread demand. Limitations such as poor peptide cell
permeability, minimum exposure and resident time at the target
site for activity have halted the translation of bio-active peptides
such as the D-(KLAKLAK)2 sequence from pre-clinical to clinical
use. In an effort to mitigate these limitations and improve the
‘drug-like’ properties of the pro-apoptotic D-(KLAKLAK)2 sequence,
a chemically robust and structurally pre-organized amphiphilic
nucleolipid is proposed to enhance cell permeability and
mitochondria localization in tumors, resulting in potent and
long-lasting anti-cancer effects.
The nucleolipids represent an interesting class of bioconjugates
owning the ability to form stable, higher-ordered structures for
potential applications in medicinal chemistry.15,16 For example,
they have been shown to form complexes with short-interfering
RNA (siRNA), facilitating siRNA transfection leading to the down-
regulation of mRNA and protein expression in cancer cells.17–19
Moreover, we have recently reported the synthesis, GRP78
oncogene binding and selective anti-cancer activity of a novel
aminoacyl nucleolipid within the SR human leukemia cancer cell
line.20 Thus, the nucleolipids have effectively served as robust
delivery vehicles for biologicals and as potent cytotoxic agents
for enhancing the cell death response in cancer. In this study, a
thymidine-derived nucleolipid is rationally designed to contain a
reactive carboxy group for coupling with the pro-apoptotic
⇑
Corresponding author. Tel.: +1 973 313 6539.
E-mail address: david.sabatino@shu.edu (D. Sabatino).
http://dx.doi.org/10.1016/j.bmcl.2016.06.020
0960-894X/Ó 2016 Elsevier Ltd. All rights reserved.
Please cite this article in press as: Rana, N.; et al. Bioorg. Med. Chem. Lett. (2016), http://dx.doi.org/10.1016/j.bmcl.2016.06.020

2
N. Rana et al. / Bioorg. Med. Chem. Lett. xxx (2016) xxx–xxx
D-(KLAKLAK)2 sequence, thereby affording the desired
peptide-nucleolipid bioconjugate (Fig. 1A). The thymidine
nucleoside provides
a chemically resilient and structurally
pre-organized scaffold,21 while the long chain alkylamine is antic-
ipated to enhance cell permeability and mitochondria localization
of the D-(KLAKLAK)2 sequence for potent anti-cancer activity.
The synthesis of nucleolipid 6 (Scheme 1) was initiated by the
regioselective attachment of the allyl group at the 30-hydroxyl of
commercially available 50-O-(4-dimethoxytrityl) DMT-thymidine,
1. The allylation reaction proceeded smoothly with sodium hydride
(NaH) as base, and allyl bromide as alkylating reagent resulting in
87% yield of the mono-allylated nucleoside, 2, and without any
observable allylation on thymine. The allyl group in 2 was then
oxidized to the aldehyde, 3, and isolated in 84% yields following
silica gel column chromatography. The reactive 30-aldehyde was
subsequently condensed with hexadecylamine and reduced using
sodium cyanoborohydride to yield the alkylamine –derived
thymidine, 4. This two-step one-pot reductive amination reaction
favored the formation of the desired intermediate, 4, in 43% yields.
In this reaction, the conversion of the imine to secondary amine
was closely monitored by 1H NMR which confirmed the
disappearance of the iminium proton (d: 6.85 ppm) upon complete
reduction of the Schiff base. Following quantitative removal of the
50-DMT group using trichloroacetic acid (TCA), a TEMPO-mediated
oxidation reaction converted the 50-hydroxy to the carboxylic acid,
Scheme 1. Synthesis of 50-carboxy-derived thymidine nucleolipid, 6. Conditions:
(a) allyl bromide, NaH, THF, sonication, rt, 3 h, 87%, (b) 4% OsO4 in t-BuOH, NMO,
Na2S2O5, NaIO4, acetone:phosphate buffer (3:1 v/v), 84%, (c) (i) 1-hexadecylamine,
THF, rt, 20 min, (ii) NaBH3CN, THF, rt, 4 h, 43% (d) 3% TCA:DCM, rt, 4 h, >99%, (e) (i)
TEMPO, BAIB, DCM, rt, 2 h, (ii) MeCN:H2O, rt, 24 h, 50%.
6, in 50% yield. Each reaction intermediate and the final product, 6,
were characterized by NMR, IR spectroscopy and by MS to confirm
purities and identities (see Supporting Information).
Figure 1. Biophysical and structure properties of the D-(KLAKLAK)2-AK analogues. (A) Peptide structures. The D-(KLAKLAK)2-AK peptide scaffold with R1(red) = H or FITC,
R2(blue) = H, Nucleolipid or Palmitamide. R1, R2 = H, 8, R1 = FITC, R2 = H, 9, R1 = H, R2 = Palmitamide, 11, R1 = H, R2 = Nucleolipid, 12, R1 = FITC, R2 = Palmitamide, 13, R1 = FITC,
R2 = Nucleolipid, 14. All amino acids are in their D-configuration. (B) Comparison of the retention times of D-(KLAKLAK)2-AK peptides on a C18 column by RP-HPLC. (C) UV/Vis
spectroscopy (190–300 nm) of the peptides. Inset shows the nucleolipid-derived D-(KLAKLAK)2-AK, 12, absorbance at 260 nm. (D) CD spectra of peptides at 12.5 lM in H2O:
TFE (50:50, v/v).
Please cite this article in press as: Rana, N.; et al. Bioorg. Med. Chem. Lett. (2016), http://dx.doi.org/10.1016/j.bmcl.2016.06.020

N. Rana et al. / Bioorg. Med. Chem. Lett. xxx (2016) xxx–xxx
3
The synthesis of the D-(KLAKLAK)2-AK sequence, 8, was
accomplished by conventional Fmoc-solid phase peptide synthesis
on a Rink amide hydrophilic poly(ethylene glycol) resin (Nova PEG,
0.47 mmol/g).22 At the N-terminus of the peptide sequence an
aminohexanoic acid (AHx) linker was attached for the incorpora-
tion of fluorescein isothiocyanante (FITC).23 Following cleavage
and deprotection, the FITC-labeled peptide sequence, 9, was
isolated by RP-HPLC for characterization studies. At the C-termi-
nus, an orthogonally protected Lys(Dde) residue was selectively
deprotected on solid-phase using hydroxylamine hydrochloride
buffered with imidazole. These mild conditions have been shown
to cleave the Dde group without concomitant Fmoc-deprotection
which has been shown to occur when hydrazine is used as Dde
deblocking reagent.24 Bioconjugation was first attempted by cou-
pling palmitic acid at the C-terminus of the partially deprotected
peptide sequence, 10, bound to the solid support (Scheme 2). Opti-
mized conditions for peptide coupling reactions using HCTU as
coupling reagent and NMM as base were adopted for making the
desired bioconjugates on solid support. The palmitamide-derived
D-(KLAKLAK)2-AK sequence was Fmoc-deprotected and either
coupled with FITC at the N-terminus, followed by cleavage and
deprotection from solid-phase, or was directly cleaved and depro-
tected for LCMS analysis and purification. The palmitamide-
derived D-(KLAKLAK)2-AK sequence, 11, and the FITC-labeled
sequence, 13, were isolated by RP-HPLC and their identities were
confirmed by MS. These coupling conditions were adopted for
the synthesis of the nucleolipid-derived D-(KLAKLAK)2-AK
sequence, which also produced, respectively 12 and 14, without
and with the FITC label. In either cases, the peptide bioconjugates
were isolated in sufficient yields (10–33%) and purities (P96%)
(Table S1, see Supporting Information) for exploring biophysical,
structure and biological properties.
the hydrophobic side-chain residues of the peptides and the
long-chain alkyl groups found within the lipid bilayer.25 Compar-
ison of the retention times on RP-HPLC for the native,
D-(KLAKLAK)2-AK sequence, 8, RT = 6.5 min, with the palmita-
mide-derived D-(KLAKLAK)2-AK, 11, RT = 17.6 min, and the nucle-
olipid-derived D-(KLAKLAK)2-AK, 12, RT = 11.3 min, confirmed
that the palmitamide, followed by the nucleolipid produced the
greatest hydrophobic effects onto the native D-(KLAKLAK)2-AK
sequence (Fig. 1B).
The hydrophobicity of the peptide bioconjugates was also
confirmed by the DLS zeta potential measurements and the logP
values evaluating the octanol:water partition coefficients of the
FITC-labeled peptides at 490 nm (Table S2, see Supporting
Information). Furthermore, UV–Vis spectroscopy also confirmed
the characteristic absorption bands of the peptide bonds
(kmax ꢀ 220 nm) and the thymine base (kmax ꢀ 260 nm) corre-
sponding to the nucleolipid-derived D-(KLAKLAK)2-AK, 12,
(Fig. 1C). CD spectroscopy was then used to evaluate the secondary
structures (if any) of the peptides in H2O:TFE. The combination of
H2O:TFE has been especially useful in mimicking the amphiphilic
lipid bilayer microenvironment which has served to stabilize
peptide secondary structures in solution.26 In this study, the native,
D-(KLAKLAK)2-AK sequence, 8, displayed a stable,
a-helical sec-
ondary structure (65% helicity), while the palmitamide-derived
D-(KLAKLAK)2-AK, 11, displayed enhanced helicity (76%) and the
nucleolipid- derived D-(KLAKLAK)2-AK 12, demonstrated a less
stable (42%)
a-helical secondary structure (Fig. 1D and Table S2).
Many amphiphilic cell penetrating peptides adopt
a-helical sec-
ondary structures. In certain cases, this peptide motif functions
as a recognition site with other membrane spanning proteins
(such as the pore proteins) and facilitates cellular translocation
activity.27 The amphiphilic
a-helical peptides, 8, 11 and 12, were
The peptide bioconjugates were anticipated to display greater
hydrophobicity relative to the native D-(KLAKLAK)2-AK sequence.
The enhanced hydrophobic character of cell-penetrating peptides
has been shown to improve their membrane translocation, pre-
sumably due to the favorable non-polar interactions in between
subsequently analyzed by DLS and TEM to determine ionic charge,
size, shape and particle distributions (Fig. 2). The DLS data
(Table S2, see Supporting Information) confirmed positively
charged peptide ionic structures according to the zeta-potential
measurements obtained in H2O. The size distributions
Scheme 2. Synthesis of FITC and non-FITC tagged D-(KLAKLAK)2-AK conjugated with nucleolipid 6 or palmitamide.
Please cite this article in press as: Rana, N.; et al. Bioorg. Med. Chem. Lett. (2016), http://dx.doi.org/10.1016/j.bmcl.2016.06.020

4
N. Rana et al. / Bioorg. Med. Chem. Lett. xxx (2016) xxx–xxx
Figure 2. TEM images and DLS particle diameter size distributions. (A) The native D-(KLAKLAK)2-AK sequence, 8, (B) palmitamide-derived D-(KLAKLAK)2-AK sequence, 11,
and (C) the nucleolipid-derived D-(KLAKLAK)2-AK sequence, 12.
(60–300 nm) detected by DLS and TEM were found to be
comparable and indicated a single, major population of nanoparti-
cles for each formulation (Fig. 2).
Table 1
NCI-60 cancer cell line screen lethality (% cell death) data for a selection of human
Non-Small Cell Lung Cancer cell lines
For example, the native D-(KLAKLAK)2-AK sequence, 8,
displayed, uniform, square shaped particles ranging in sizes from
70 to 270 nm. The palmitamide-peptide bioconjugate, 11, pro-
duced smaller, spherical-shaped particle sizes (70–190 nm), while
the nucleolipid-peptide bioconjugate 12, demonstrated the largest
Lethality (% cell death) in NSCLC
Sequence
A549
NCI-H226
NCI-H522
Native, 8
Palmitamide, 11
Nucleolipid, 12
19
32
31
1
0
4
14
16
11
particle size distribution (60
– 310 nm). Considering the
self-assembly and nanoparticle formulation of similar amphiphilic
D-(KLAKLAK)2 sequences have led to significant anti-cancer
effects,28 the cancer cell line toxicities of the peptides were
subsequently evaluated.
A lethality assay was performed against a panel of 60 cancer cell
lines at National Cancer Institute (NCI).29 A single dose screen was
performed with the native, D-(KLAKLAK)2-AK sequence, 8,
formulation which enhanced the cytotoxicity of the killer peptide
sequence within the A549 lung cancer cells and within the drug-
resistant lung cancer A549/Taxol cell line.30 Thus, the amphiphilic
palmitamide or nucleolipid-derived D-(KLAKLAK)2-AK sequences
represents an interesting class of bioconjugates that have served
to potentiate the cancer cell death response of the native
D-(KLAKLAK)2-AK within the A549 lung cancer cell line.
Furthermore, the cytotoxic selectivity for cancer cell lines over
normal tissues for bioconjugates 11 and 12 is unanticipated but
may be introduced by the incorporation of cancer cell targeting
ligands.6,7,9,12–14 This is currently a focal point in our research
program geared towards the development of cancer-targeting
peptide bioconjugates.
(10
11, (7.25
sequence, 12, (2.13
l
M), the palmitamide-derived D-(KLAKLAK)2-AK sequence,
M), and the nucleolipid-derived D-(KLAKLAK)2-AK
M). The cell growth data (see Supporting Infor-
l
l
mation) indicated that the peptides were found to be most active
towards a panel of non-small cell lung carcinoma (NSCLC). Interest-
ingly, the palmitamide and nucleolipid-derived D-(KLAKLAK)2-AK
sequences, 11 and 12, showed the highest toxicity (ꢀ30%) within
the A549 NSCLC cell line, resulting in a 1.5-fold enhancement rela-
tive to the native D-(KLAKLAK)2-AK sequence, 8, (ꢀ20%) (Table 1).
These results correlate nicely with the recently reported anti-cancer
effects of the amphiphilic D-(KLAKLAK)2-phospholipid liposome
In conclusion, the synthesis, characterization and anti-cancer
activity of a new class of killer peptide bioconjugates are reported
in this study. The peptide bioconjugates were found to be more
hydrophobic relative to the native sequence. Moreover, the peptide
Please cite this article in press as: Rana, N.; et al. Bioorg. Med. Chem. Lett. (2016), http://dx.doi.org/10.1016/j.bmcl.2016.06.020

N. Rana et al. / Bioorg. Med. Chem. Lett. xxx (2016) xxx–xxx
5. Horton, K. L.; Kelley, S. O. J. Med. Chem. 2009, 52, 3293.
5
bioconjugates displayed
a-helical structures and cationic
6. Ellerby, H. M.; Arap, W.; Ellerby, L. M.; Kain, R.; Andrusiak, R.; Rio, G. D.;
Krajewski, S.; Lombardo, C. R.; Rao, R.; Ruoslahti, E.; Bredesen, D. E.; Pasqualini,
R. Nat. Med. 1999, 5, 1032.
7. Liu, Y.; Steiniger, S. C.; Kim, Y.; Kaufmann, G. F.; Felding-Habermann, B.; Janda,
K. D. Mol. Pharm. 2007, 4, 435.
8. Mai, J. C.; Mi, Z.; Kim, S. H.; Ng, B.; Robbins, P. D. Cancer Res. 2001, 61, 7709.
9. Marks, A. J.; Cooper, M. S.; Anderson, R. J.; Orchard, K. H.; Hale, G.; North, J. M.;
Ganeshaguru, K.; Steele, A. J.; Mehta, A. B.; Lowdell, M. W.; Wickremasinghe, R.
G. Cancer Res. 2005, 65, 2373.
10. Law, B.; Quinti, L.; Choi, Y.; Weissleder, R.; Tung, C. H. Mol. Cancer Ther. 1944,
2006, 5.
nanoparticle formulations that were contributive towards their
anti-cancer activities. Within a NCI-60 cancer cell line screen, the
peptide bioconjugates displayed enhanced cell death activity
relative to the native sequence within the human A549 lung cancer
cell line. These results validates the anti-cancer utility of the new
amphiphilic peptide bioconjugates and paves the way for a cancer
cell biology study that may ultimately unveil their mechanism of
action.
11. Zurita, A. J.; Troncoso, P.; Cardó-Vila, M.; Logothetis, C. J.; Pasqualini, R.; Arap,
W. Cancer Res. 2004, 64, 435.
12. Rege, K.; Patel, S. J.; Megeed, Z.; Yarmush, M. L. Cancer Res. 2007, 67, 6368.
13. Cai, H.; Yang, H.; Xiang, B.; Li, S.; Liu, S.; Wan, L.; Zhang, J.; Li, Y.; Cheng, J.; Lu, X.
Mol. Pharm. 2010, 7, 586.
14. Agemy, L.; Friedmann-Morvinski, D.; Kotamraju, V. R.; Roth, L.; Sugahara, K. N.;
Girard, O. M.; Mattrey, R. F.; Verma, I. M.; Ruoslahti, E. Proc. Natl. Acad. Sci. USA
2011, 108, 17450.
15. Simeone, L.; Irace, C.; Di Pascale, A.; Ciccarelli, D.; D’Errico, G.; Montesarchio, D.
Eur. J. Med. Chem. 2012, 57, 429.
16. Arigon, J.; Prata, C. A.; Grinstaff, M. W.; Barthélémy, P. Bioconjug. Chem. 2005,
16, 864.
17. Patil, S. P.; Yi, J. W.; Bang, E. K.; Jeon, E. M.; Kim, B. H. Med. Chem. Commun.
2011, 2, 505.
18. Yang, H. W.; Yi, J. W.; Bang, E. K.; Jeon, E. M.; Kim, B. H. Org. Biomol. Chem. 2011,
9, 291.
Acknowledgements
The authors would like to thank Seton Hall University for
continued support, the Core Facility for Imaging, Cellular and
Molecular Biology at Queens College (CUNY) and support from
Undergraduate Research and Mentoring Education (URME) pro-
gram at Queens College. The biological testing of peptides was per-
formed by the Developmental Therapeutics Program, Division of
Cancer Treatment and Diagnosis, National Cancer Institute and
the URL to the Program’s website is http://dtp.cancer.gov.
Supplementary data
19. Ceballos, C.; Prata, C. A. H.; Giorgio, S.; Garzino, F.; Payet, D.; Barthélémy, P.;
Grinstaff, M.; Camplo, M. Bioconjug. Chem. 2009, 20, 193.
20. Patel, P.; Hanawa, E.; Yadav, R.; Samuni, U.; Marzabadi, C.; Sabatino, D. Bioorg.
Med. Chem. Lett. 2013, 23, 5086.
21. Hart, P. A.; Davis, J. P. J. Am. Chem. Soc. 1969, 91, 512.
22. Chantell, C. A.; Onaiyekan, M. A.; Menakuru, M. J. Pept. Sci. 2012, 18, 88.
23. Jullian, M.; Hernandez, A.; Maurras, A.; Puget, K.; Amblard, M.; Martinez, J.;
Subra, G. Tetrahedron Lett. 2009, 50, 260.
24. Diaz-Mochon, J. J.; Bialy, L.; Bradley, M. Org. Lett. 2004, 6, 1127.
25. Avrahami, D.; Oren, Z.; Shai, Y. Biochemistry 2001, 40, 12591.
26. Ozdirekcan, S.; Nyholm, T. K.; Raja, M.; Rijkers, D. T.; Liskamp, R. M.; Killian, J.
A. Biophys. J. 2008, 94, 1315.
27. Epand, R.; Shai, Y.; Segrest, J. P.; Anantharamaiah, G. M. Biopolym. Pept. Sci.
1995, 37, 319.
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.bmcl.2016.06.
020.
References and notes
1. Javadpour, M. M.; Juban, M. M.; Lo, W. C.; Bishop, S. M.; Alberty, J. B.; Cowell, S.
M.; Becker, C. L.; McLaughlin, M. L. J. Med. Chem. 1996, 39, 3107.
2. Glukhov, E.; Stark, M.; Burrows, L. L.; Deber, C. M. J. Biol. Chem. 2005, 280,
33960.
28. Standley, S. M.; Toft, D. J.; Cheng, H.; Soukasene, S.; Chen, J.; Raja, S. M.; Stupp,
S. I. Cancer Res. 2010, 70, 3020.
29. Showmaker, R. H. Nat. Rev. Cancer 2006, 6, 813.
30. Jiang, L.; Li, L.; He, X.; Yi, Q.; He, B.; Cao, J.; Pan, W.; Gu, Z. Biomaterials 2015, 52,
126.
3. McGrath, D. M.; Barbu, E. M.; Driessen, W. H.; Lasco, T. M.; Tarrand, J. J.;
Okhuysen, P. C.; Kontoyiannis, D. P.; Sidman, R. L.; Pasqualini, R.; Arap, W. Proc.
Natl. Acad. Sci. USA 2013, 110, 3477.
4. Costantini, P.; Jacotot, E.; Decaudin, D.; Kroemer, G. J. Natl Cancer Inst. 2000, 92,
1042.
Please cite this article in press as: Rana, N.; et al. Bioorg. Med. Chem. Lett. (2016), http://dx.doi.org/10.1016/j.bmcl.2016.06.020