Journal of Controlled Release 182 (2014) 13–21

Contents lists available at ScienceDirect

Journal of Controlled Release

journal homepage: www.elsevier.com/locate/jconrel

Delivering wasp venom for cancer therapy

Miguel Moreno ⁎, Esther Zurita, Ernest Giralt ⁎⁎
Institute for Research in Biomedicine (IRB Barcelona), Barcelona, Spain

a r t i c l e i n f o a b s t r a c t

Article history: Cytolytic peptides with potential therapeutic properties have appeared during the last three decades. However,
Received 10 February 2014 the use of these natural weapons is relatively narrow due to their non-specific cytolytic activity as well as their
Accepted 3 March 2014 rapid degradation and excretion when injected in blood. In order to rescue the use of these lytic peptides, we
Available online 12 March 2014
have designed pro-cytotoxic systems based on cytotoxic peptides conjugated to poly(L-glutamic acid) PGA poly-
mer through specific cleavage sequences that are sensitive over-expressed tumor proteases, such as the
Keywords:
Pro-cytotoxic systems
metalloproteinase-2 (MMP-2) or cathepsin B. The potent cytotoxic peptide tested here, Mitoparan, is inactive
Cytolytic peptides when conjugated to the polymer and then become active again once released through the tumor proteases. Fur-
Polyglutamic acid thermore, this pro-cytotoxic system was decorated by a particular targeting peptide which binds to HER2 recep-
Targeting peptides tors over-expressed in some types of breast tumor cells, thereby increasing the selective release of cytolytic
Overexpressed tumor proteases peptides inside tumor cell with exquisite spatiotemporal control. In this way, the system would improve the
HER2+ breast cancer maximum tolerated dose and the pharmacokinetic parameters of cytotoxic peptides in vivo.
© 2014 Elsevier B.V. All rights reserved.

1. Introduction
The use of venoms in cancer therapy continues to be a challenge.
These natural weapons have been widely studied for the treatment of
several immune-related diseases, and have recently entered preclinical
phases for cancer treatment [1]. However, the high toxicity of these po-
tential therapeutic peptides caused by non-specific lytic activity and
their rapid degradation in blood make them of limited use in cancer
therapy. These peptides are between 10 and 50 residues in length, and
they show amphipathic properties. They have a propensity to interact
with membranes, oligomerizing on the cell surface so as to form tran-
sient pores, thus causing cell death. Since free cytolytic peptides are
not able to elicit a therapeutic benefit at a safe dose, they have to be
targeted and delivered as pro-drugs. Previous studies report on how dif-
ferent types of cytolytic peptides have been successfully conjugated to
proteins through tumor-specific proteases, such as metalloproteinase-
2 (MMP-2) [2,3]. In addition, cytotoxic peptides have been carried by li-
pidic systems, as in Yamada's work [4], where the peptide mastoparan
was transported inside a transferrin-modified liposome. The most re-
cent case is the 26-mer cytotoxic peptide melittin, derived from bee
venom, which has been successfully targeted and delivered via fluoro-
carbon nanocarriers to tumor models in vivo [5,6]. Further improve-
ments in order to increase the efficacy of the systems are needed.
Alternative types of carriers, such polymers, with suitable features could
improve the efficacy of delivery cytolytic peptides, overcoming some
drawbacks related to low activity, elaborate complexes and non-scalable
production. In fact, an increasing number of polymer therapeutics have
started to move from basic science into clinical trials and, even into clini-
cal practice [7]. Not only should the polymer in the drug delivery system
be innocuous, hydrophilic, non-immunogenic and biodegradable, but also
should it have a long-circulating behavior and enhanced accumulation at
the tumor site. In particular, the well-characterized poly(L-glutamic acid)
(PGA), which has an attractive safety profile, has been widely used with
many kinds of cargo, such as low MW drugs [8] and, even, peptoids [9].
In this study, we present a peptide-polymer design strategy to
obtain pro-cytotoxic systems based on lytic peptides conjugated to
PGA polymer through specific cleavage sites that are sensitive over-
expressed tumor proteases, such as MMP-2 or cathepsin B (Fig. 1).
The potent cytotoxic peptides are inactive when conjugated to the poly-
mer, but once released through the tumor proteases, they recover their
activity. This strategy is thought to prevent the side effects that occur
in vivo. Furthermore, this pro-cytotoxic carrier was decorated with pep-
tides able to specifically target tumor cells. In this way, the system
would improve the maximum tolerated dose and the pharmacokinetic
parameters of cytotoxic peptides in vivo.

2. Materials and methods

⁎ Correspondence to: M. Moreno, Institute for Research in Biomedicine (IRB Barcelona),
C\Baldiri I Reixac, 10, 08028 Barcelona, Spain. Tel.: +34 934037127. 2.1. Materials
⁎⁎ Correspondence to: E. Giralt, University of Barcelona, Institute for Research in
Biomedicine (IRB Barcelona), C\Baldiri I Reixac, 10, 08028 Barcelona, Spain.
E-mail addresses: miguel.moreno@irbbarcelona.org (M. Moreno), Fmoc-Nα-protected amino acids and poly(L-glutamic acid) (Npt-
ernest.giralt@irbbarcelona.org (E. Giralt). PGA(100)Na) were obtained from IRIS Biotech GmbH (Marktredwitz,

http://dx.doi.org/10.1016/j.jconrel.2014.03.005
0168-3659/© 2014 Elsevier B.V. All rights reserved.
14 M. Moreno et al. / Journal of Controlled Release 182 (2014) 13–21

Fig. 1. a) Synthetic strategy of the targeted pro-cytotoxic conjugate based on modified PGA polymer (PGAmod). PGA decoration with different reagents is used to prepare a soluble scaffold
onto which the targeting and pro-apoptotic peptides (Mitoparan) are attached. b) Schematic illustration showing the interaction of targeted pro-cytotoxic conjugates with HER2 positive
tumor cells, which overexpress MMP-2 on the plasmatic membrane (A) or cathepsin B inside lysosomes (B). These two enzymes will promote the release of Mitoparan, cutting through
specific cleavage sequences, transforming this pro-drug system into a toxic drug.

Germany). The Fmoc-Rink-Amide AM polystyrene resin was pur-
chased from CBL-PATRAS (Patras, Greece). Coupling reagents: 2-
(1H-benzotriazol-1-yl)-1,1,3,3-tetramethyluronium tetrafluoroborate
(TBTU) was from Albatros Chem, Inc. (Montreal, Canada). Trifluoroacetic
acid (TFA) was purchased from Scharlab S.L. (Barcelona, Spain).
Piperidine, dimethylformamide (DMF), dichloromethane (DCM)
and acetonitrile (MeCN) were from SDS (Peypin, France). N,N-
Diisopropylethylamine (DIEA) was obtained from Merck (Darmstadt,
Germany). Tri-isopropylsilane (TIS) was from Fluka (Buchs,
Switzerland). Lissamine rhodamine B sulfonyl chloride (Sulrhodamine
B) was obtained from Acros (Somerville, NJ). Methoxy PEG Amine,
HCl Salt, MW 5000 was from JenKem Technology USA Inc. LysoTracker®
Green DND-26 and Red DND-99 were supplied by Invitrogen (Carlsbad,
CA). Cathepsin B from human liver, Sephadex® G-25, 5(6)-carboxyflu-
orescein (CF), tetrakis(triphenylphosphine)palladium(0), (Pd(PPh3)4),
phenylsilane (PhSiH3) and ethanolamine were obtained from Sigma-
Aldrich (St. Louis, USA). Recombinant Human Matrix Metalloproteinase
2 (MMP-2) was purchase from Sina Biological Inc. (Beijing, China).
2.2. Peptide synthesis
Peptides were synthesized by solid phase synthesis using the 9-
fluorenylmethoxycarbonyl/tert-butyl (Fmoc/tBu) strategy. Fmoc-
Rink-Amide AM polystyrene resin, Nα-Fmoc-protected amino acids
(3 eq)/TBTU(3 eq.), and DIEA(6 eq.) were used. The Fmoc protecting
group was cleaved by treatment with a solution of 20% piperidine in
DMF. For labeling with 5(6)-carboxyfluorescein (CF) of the dye-
peptides 2, 3, and 9, firstly, Nα-Fmoc-Lys(Alloc)-OH was incorporated
at the desired position. For the Alloc group removal, the resin was treat-
ed with Pd(PPh3)4 and PhSiH3 (3 × 15 min) and then washed with DMF.
Afterwards, 2 treatments of 30 min with a solution of sodium N,N′
diethyldithiocarbamate in DMF were performed. Finally, the CF was in-
corporated as it was a Nα-Fmoc-protected amino acid. Peptides were
cleaved from the resin by treatment with 95% TFA, 2.5% TIS, 2.5%
water for 3 h and detected by analytical RP-HPLC Waters 996 photodi-
ode array detector equipped with the Waters 2695 separation module,
the Symmetry column (C 18, 5 mm, 4.6P150 mm) and the Millennium
 M. Moreno et al. / Journal of Controlled Release 182 (2014) 13–21
software. Peptides were purified by semipreparative RP-HPLC (Waters
2487 Dual l Absorbance Detector equipped with a Waters 2700 Sample
Manager, a Waters 600 Controller, a Waters Fraction Collector, a Sym-
metry column [C 18, 5 mm, 30P100 mm] and Millennium chromatogra-
phy manager software) (see Supplementary data, Fig. S1). Peptides
were characterized by MALDI-TOF mass spectrometry (Voyager-DE RP
MALDI-TOF, PE Biosystems with a N2-laser of 337 nm) and high-
resolution ESI-MS model (LTQ-FT Ultra, Thermo Scientific) (see Supple-
mentary data, Table S1).
2.3. Synthesis of modified polyglutamic acid polymers (PGAmod)
The pH of a solution of poly(L-glutamic acid) (PGA) sodium was
lowered to obtain its proton form. After centrifugation, the precipitate
was lyophilized and dehydrated in a vacuum drying oven in the pres-
ence of phosphorus pentoxide. The protonated PGA form (50 mg), N-
(2-Aminoethyl)maleimide trifluoroacetate salt (19 mg), Methoxy PEG
Amine HCl Salt MW 5 kDa (74 mg), Lissamine RhoB sulfonyl-GK
amidated peptide (4 mg), and O-(7-azabenzotriazol-1-yl)-N,N,N′,N′-
tetramethyluronium hexafluorophosphate (HATU) (100 mg) were dis-
solved in anhydrous N,N-dimethylformamide (DMF) (5 ml). Finally,
ethanolamine (12 μl) and 4-Methylmorpholine (NMM) (100 μl) were
added. The stoichiometry of PGA:Maleimide:PEG5 kDa:RhoGK:Ethanol-
amine was 1:25:5:2:65 mol%. The reaction was left at RT for 24 h. The
reaction was dialyzed against distilled water using a dialysis bag
(cut-off 6–8 kDa) for 3 h. Immediately after, it was purified by SEC
(Sephadex® G-25). The lyophilized sample was analyzed by 1H-NMR
at 400 MHz (Varian Mercury) and 800 MHz (Bruker Digital Avance)
(see Supplementary data, Figs. S2 and S3 and Table S2).
2.4. Formation of PGA-peptide conjugates
The reaction between the modified PGA (PGAmod) and peptides with
cysteine at the N-terminus, took place in 1 ml of 100 mM Tris buffer at
pH 7.0 and RT for 2 h. An example of reaction would be mixing PGAmod
(6 mg) with Peptide 6 (1.5 mg) and Peptide 13 (0.6 mg), where the stoi-
chiometry was 1:4:3 (see Table 2 for all the conjugates accomplished).
Upon completion of the reaction, an excess of cysteamine was added
to ensure that no free reactive maleimides were present in the final
conjugate. The PGAmod-peptide(s) conjugates were purified using a
Sephadex® G-25 column.
2.5. Amino acid analysis
The conjugated amount of peptide along PGAmod was determined
through amino acid analysis. 0.5 mg of each sample was hydrolyzed
with 6 N HCl at 110 °C for 18 h. Later, the samples were evaporated
and suspended in 0.5 ml of 20 mM HCl with an internal standard of
AABA 2.5 mM standard (Pierce Amino Acid Standard H, Thermo Scien-
tific). Amino acid analysis was performed at the Scientific and Techno-
logical Centers of the University of Barcelona (CCiT UB), using the
AccQTag pre-column derivatization method. Reaction of amino acids
with 6-aminoquinolyl-N-hydroxysuccinimidyl carbamate yields deriva-
tives that are detected at 254 nm. 20 μl of the derivatized samples or stan-
dards was injected in a column of Nova-Pak C18 4 μm, 3.9 × 150 mm
(Waters). Data acquisition and treatment was done using the ChromPass
Chromatography Data System (Jasco).
2.6. Leakage measurement
Aliquots containing the appropriate amount of lipid in chloroform/
methanol (1:1, v/v) were placed in a test tube, the solvents were re-
moved by evaporation under a stream of O2-free nitrogen, and finally
traces of solvents were eliminated under vacuum in the dark for more
than 3 h. After, 1 ml of buffer containing 10 mM HEPES, 100 mM NaCl,
0.1 mM EDTA, pH 7.4 buffer and CF at a concentration of 40 mM was
15
added, and multilamellar vesicles were obtained. Large unilamellar ves-
icles (LUVs) with a mean diameter of 200 nm were prepared from the
multilamellar vesicles by the LiposoFast device from Avestin, Inc.,
using polycarbonate filters with a pore size of 0.2 μm (Nuclepore
Corp., Cambridge, CA, USA). Breakdown of the vesicle membrane leads
to content leakage, i.e., CF fluorescence. Non-encapsulated CF was sepa-
rated from the vesicle suspension through a Sephadex G-25 filtration
column eluted with buffer containing 10 mM HEPES, 150 mM NaCl,
and 0.1 mM EDTA, pH 7.4. Leakage of intraliposomal CF was assayed
by treating the probe-loaded liposomes (final lipid concentration,
0.125 mM) with the appropriate amount of peptide or conjugate in Co-
star 3797 round-bottom 96-well plates, each well containing a final vol-
ume of 100 μl. Leakage was measured at various peptide-to-lipid ratios.
The micro titer plate was incubated at RT for 1 h to induce dye leakage.
Leakage was measured at various peptide-to-lipid ratios. Changes in
fluorescence intensity were recorded using the FL600 fluorescence mi-
croplate reader with excitation and emission wavelengths set at 492
and 517 nm, respectively. One hundred percent release was achieved
by adding Triton X-100 to a final concentration of 1% v/v to the microti-
ter plates. Fluorescence measurements were made initially with probe-
loaded liposomes, afterwards by adding peptide or conjugate solution
and, finally by adding Triton X-100 to obtain 100% leakage. The results
were expressed as percentage of CF released relative to the positive con-
trol (Triton X-100).
2.7. Red blood cell lysis assay
Human blood was collected in 10 ml EDTA Vacutainer tubes. A small
aliquot was assessed for evidence of hemolysis by centrifugation at 800 g
for 10 min, and non-hemolyzed samples were carried forward into the
assay. Red blood cells (RBCs) were washed three times in PBS pH 7.4 by
centrifugation at 800 g for 10 min and resuspending in the same buffer
to yield a 10× dilution. RBCs were then diluted in appropriate pH buffer
to yield approximately ± 15 × 107 cells/100 μl PBS in Costar 3797
round-bottom 96-well plates for the lysis assay. The micro titer plate
was covered with a low evaporation lid and incubated at 37 °C for 1 h
to induce hemolysis. Negative controls were PBS, while positive con-
trols were 1% v/v solution of Triton X-100 (100% lysis). The plate was
then centrifuged at 800 g for 10 min and 80 μl of supernatants were
transferred to a Costar 3632 clear bottom 96-well plate. Hemoglobin ab-
sorbance was read at 560 nm using the ELx800 absorbance microplate
reader. The results were expressed as percentage of hemoglobin re-
leased relative to the positive control (Triton X-100).
See the rest of Materials and methods in the Supplementary data.
3. Results and discussion
3.1. Design, synthesis and characterization of the pro-cytotoxic conjugates
The choice of lytic peptide was based on the cytolytic-activity/length
ratio of several cytotoxic peptides. We chose the 14-mer cytolytic pep-
tide Mitoparan, which exhibits mitochondriotoxic and apoptogenic
properties and does not induce cancer drug resistance [10]. Given that
the induction of apoptosis is the best way to prevent unwanted immune
responses [11], and to minimize damage and disruption of neighboring
cells, a method based on a selective activable Mitoparan (targeted pro-
Mitoparan) is an elegant way to efficiently kill tumor cells. Since an
enzyme-catalyzed degradation combines chemical stability and high
specificity, we used peptide sequence-based linkers to achieve cytolytic
peptide release. We compared the efficacy of two enzymatic cleavage
sites triggered by specific overexpressed tumor enzymes, namely:
MMP-2 and cathepsin B. MMP-2 is known to be overexpressed on the
cell surface in several cancers and in tumor microvascular endothelial
cells. In addition, it has been strongly associated with the progression
of malignancy in several types of carcinoma [12]. In cancer cells and
also tumor microvascular endothelial cells, cathepsin B is thought to
16 M. Moreno et al. / Journal of Controlled Release 182 (2014) 13–21
Table 1
Sequences of the peptides used to conjugate the polymer.
Peptides Names Sequence
1 Labeling peptide Rh-GK
2 Labeling mitoparan CGGGIPVSLRSKGKGINLKKLAKL(Aib)KKIL
3 Labeling mitoparan CGGGFL(PABC)GKGINLKKLAKL(Aib)KKIL
4 D-Mitoparan CGinlkklakl(Aib)kkil
5 D-Mitoparan CGGIPVSLRSKGinlkklakl(Aib)kkil
6 D-Mitoparan CGPPPPPPPPPGIPVSLRSKGinlkklakl(Aib)kkil
7 D-Mitoparan CGGFL(PABC)GGinlkklakl(Aib)kkil
8 D-Mitoparan CGPPPPPPPPPGFL(PABC)GGinlkklakl(Aib)kkil
9 Labeling D-mitoparan CGPPPPPPPPPGIPVSLRSKGKGinlkklakl(Aib)kkil
10 CGPPG-penetratin CGPPGRQIKIWFQNRRMKWKK
11 CGY-TAT CGYGRKKRRQRRR
12 C-SAP(E) CVELPPPVELPPPVELPPP
13 Targeting peptide CGPPPPPPPPPGLTVSPWY
[a] Global Charge at pH 7.4. All peptides are C-amided. The cleavage site is underlined: GIPVSLRSK for MMP-2 and GFL(PABC) for cathepsin B. Minor capital letters represent
D-amino acids.
Rh: Lissamine Rhodamine B sulfonyl Aib: α-aminoisobutyric acid
K: Lysine with 5(6)-carboxyfluorescein at its side chain PABC: 4-aminobenzoic acid
be overexpressed at the extra- and intracellular levels, being located
mainly in the endo/lysosomal compartments [13].
Here we report on an easily synthesizable peptide platform based on
PGA polymer as a scaffold to carry cytolytic peptides. We attached a few
copies of the cytotoxic positively charged Mitoparan along with the
naked PGA by selective ligation methods such as Native Chemical Liga-
tion (NCL) and maleimide-thiol coupling chemistry. Although the yield
of the peptide-polymer conjugation was significant using both strate-
gies (~50%), the final products were insoluble. The positively charged
Mitoparan attached to the negatively charged PGA may not have been
the best combination. To gain insight into why the final product was ag-
gregated, we conjugated PGA with a negatively charged model peptide.
After the attachment of this peptide, the final peptide-PGA product was
completely soluble (data not shown). Given this result, we postulate
that the strength of the ionic interactions between the glutamic acids
of PGA and the lysine residues of Mitoparan is strong enough to induce
aggregation in the final product. Several easy and reproducible chemical
modifications were performed along the naked PGA (Fig. 1a) in order to
improve the solubility of the final products and to prevent aggregation.
Part of the negative charges on PGA was capped by ethanolamines, thus
changing carboxylic for hydroxyl groups. Moreover, molecules of poly-
ethylene glycol of 5 kDa (PEG5 kDa) were incorporated into the polymer.
Not only does the hydrophilic character of PEG make the system more
soluble and increase the blood circulation half-life, but also it prevents
the uptake of PEGylated macromolecules by the mononuclear phago-
cytic system [14], although this last issue is totally speculative because
it has not been checked in this work. Furthermore, the polymer showed
red fluorescence with peptide 1 (Table 1) for cellular uptake assays. Fi-
nally, N-(2-aminoethyl) maleimide was also incorporated along the
system to be ready for subsequent conjugation of Cys-peptides through
thiol-maleimide chemistry.
In order to characterize the molecular weight of PGAmod, we used
SEC-MALS-IR (Supplementary data, Figs. S4 and S5 and Table S3). Sig-
nals from the RI detector combined with MALS data were used to calcu-
late the molecular mass of PGAmod. The cumulative mass fractions were
plotted against molar mass in Fig. S5, depicting molecular weight distri-
butions. These measurements indicate that the PGAmod has a MW
around 60 kDa. In addition, a detailed 1H-NMR study to characterize
the PGAmod was carried out (Supplementary data, Figs. S2 and S3 and
Table S2). Unequivocal assignment of all the signals was not possible
at 400 MHz (Fig. S2B) due to overlapping. In contrast, at 800 MHz the
system showed a better chemical shift (Fig. S3C). The PGA derivatized
MW (Da) Charge[a]
742 +1
3361 +7
2859 +5
1766 +5
2761 +7
3634 +7
2260 +5
3133 +7
4177 +7
2655 +7
1718 +8
2018 −3
1954 0
only with ethanolamine and maleimide was used to assist the spectro-
scopic assignment of the totally functionalized system. All NMR signals
were assigned on the basis of chemical shift and intensity data and,
when required, confirmed by homonuclear double resonance experi-
ments (Fig. S3B). The relative number of bound groups was determined
from the relative integration of their 1H-NMR signals in relation to those
associated with the PGA protons (Table S2). Comparing the measured
integral intensities of the resonance signals at 6.85 ppm (2 aromatic
protons per maleimide unit), at 3.31 ppm (CH2-N protons of ethanol-
amine) and at 2.22–2.42 ppm (γ protons of the PGA backbone), we
could be estimated that around 23% and 67% of the glutamates were
functionalized with maleimide and ethanolamine, respectively. Similar-
ly, from the measured integral of the signal at 3.70 ppm and assuming
85 ethylene glycol moieties per PEG5 kDa group (4 methylene protons
per moiety), it estimated that PEG was incorporated in around 7%
of the glutamate residues. Combining the NMR and MALS data from
the 124 Glu side-chains available in the starting 16 kDa PGA (see
Table S3), 28 were found to be functionalized with maleimide, 8 with
PEG5 kDa and 83 capped with ethanolamine. Furthermore, we checked
the structural conformation by circular dichroism (CD) of PGAmod at a
several pH values. It was similar to naked PGA, having α-helical confor-
mation at low pH and random coil at neutral pH (Supplementary data,
Fig. S6). Below pH ~ 6, it tended to adopt an α-helical conformation
(two minimum peaks around 208 and 222 nm), because of protonated
side chains, and thus this uncharged form allowed amino acids to be
brought closer together. However, at higher pH values, the main confor-
mation tended to be random coil, because side chains are largely ionized
with negative charge, which repel each. Similar behavior was observed
for PGAmod, although the tendency to form α-helical at low pH was
hampered. At low pH, the capacity to form hydrogen bonds between
the carboxylic groups of Glu side chains is impede by PGA modifications.
Finally, several Cys-peptides were conjugated to the PGAmod
(Table 2), and cytotoxic activity was evaluated. As an example, the con-
jugate PGAmod:6:13 (see Section 3.4) was characterized by SEC-MALS-
IR to determine the MW which was 75 kDa (Table S3). This is an ideal
size for increasing the blood half-life and the passive tumor targeting
known as enhanced permeability and retention (EPR) [15].
3.2. In vitro cytotoxicity of Mitoparan and Mitoparan-conjugates
We first studied two modifications of Mitoparan peptide with differ-
ent cleavage sequences at the N-terminus (Table 1, Peptides 2 and 3) in
 M. Moreno et al. / Journal of Controlled Release 182 (2014) 13–21 17

Table 2
PGA, modified PGA (PGAmod), and peptide-PGAmod conjugates.

Samplesa Stoichiometry (feeding ratio) Stoichiometry (final ratio)b Conjugation yield (%) Zeta Potential (mV)

PGA – – – −30
PGAmod – – – −5
PGAmod:2 1:4 1:2.23 55.7 +1
PGAmod:3 1:4 1:2.14 53.5 +0.5
PGAmod:4 1:4 1:2.37 59.2 +1
PGAmod:5 1:4 1:2.19 54.7 −1
PGAmod:6 1:4 1:2.35 58.7 −0.5
PGAmod:7 1:4 1:2.27 56.7 −1
PGAmod:8 1:4 1:2.11 52.7 −0.5
PGAmod:10 1:4 1:2.53 63.2 +3
PGAmod:11 1:4 1:2.39 59.7 +3
PGAmod:12 1:4 1:3.2 80 −6
PGAmod:13 1:3 1:2.73 91 −5
PGAmod:6:10 1:4:4 1:2.11:1.88 52.7 (pept 6); 47 (pept 10) −0.5
PGAmod:6:13 1:4:3 1:2.23:2.34 55.7 (pept 6); 78 (pept 13) −1
PGAmod:8:10 1:4:4 1:1.98:1.89 49.5 (pept 8); 47.2 (pept 10) +1
PGAmod:8:13 1:4:3 1:2.33:2.45 58.2 (pept 8); 81.6 (pept 13) −0.5
PGAmod:9:13 1:4:3 1:1.86:2.32 46.5 (pept 9); 77.3 (pept 13) −1
a
Numbers represent peptides from Table 1.
b
The conjugated amount of peptide along PGAmod was determined through amino acid analysis.

order to compare the activity of the two peptides alone and when at-
tached to the PGAmod (Table 2, samples PGAmod:2 and PGAmod:3). A
non-cytotoxic effect was observed for both Mitoparan-PGAmod conju-
gates. However, a concentration-dependent decrease in SK-BR3 cell
viability was promoted by the free peptides (Supplementary data,
Fig. S7a). To further explore the cytotoxic effect live cell images were ex-
amined by confocal laser scanning microscopy (CLSM) (Supplementary
data, Fig. S7b), which revealed an intracellular distribution of free la-
beled peptides and location in the nuclei. As expected, nuclear DNA
fragmentation indicative of apoptosis was detected, as was damaged
the cell integrity (cell shrinkage and blebbing). However, in cells
treated with Mitoparan-PGAmod conjugates, which were not dam-
aged, colocalization of Mitoparan (green) and PGAmod (red) was ob-
served, indicating that they were engulfed in the same endosome,
remaining together until lysosomal enzymatic digestion. There are two
possible non-mutually exclusive explanations for the non-cytotoxic ef-
fect observed for the Mitoparan-conjugates: 1) MMP-2 and/or cathepsin
B were inaccessible to the cleavage sites, and therefore the Mitoparan
peptides remained attached to the polymer, which made it inactive; or
2) Mitoparan was digested at the same rate as the enzymatic cleavage
sites, thus losing cytotoxicity. To further address these two scenarios,
we performed enzymatic degradation assays of the two peptides (Sup-
plementary data, Fig. S8).
Analysis of the HPLC chromatograms and the cleavage fragments by
MALDI-TOF revealed that cathepsin B is capable of splitting not only the
N-terminal cleavage site but also the two additional amide positions
along the Mitoparan peptide per se. The peptide fragmentation along
the sequence thus renders Mitoparan inactive. We postulate that the
overexpressed cathepsin B, which is located at extra- and intracellular
levels in the tumor, would be the main protease to cleavage and
would subsequently inactivate Mitoparan. However, MMP-2 accurately
cleaved the peptide through the cleavage site, as has been previously
described [16,17].
3.3. In vitro cytotoxicity of D-Mitoparan and D-Mitoparan conjugates
An enantiomer of Mitoparan has recently been characterized [18],
revealing that it maintains similar cytotoxic properties with the addi-
tional protease-resistant feature. Taking advantage of this property,
we synthesized a battery of new enantiomer Mitoparan peptides to re-
place L-Mitoparan. For this purpose, the C-terminal peptide sequence,
corresponding to the 14 amino acids of Mitoparan per se, was synthe-
sized with D-amino acids (Table 1, peptides 4–9). Hereinafter this enan-
tiomer will be referred to as D-Mitoparan. In addition, we synthesized
D -Mitoparan peptides with 9 Pro as a spacer (Pro9) to place the
cleavage site far from the PGA scaffold (Table 1, peptides 6, 8 and
9), thus decreasing the possibility of the PGA polymer shielding the
N-terminal cleavage site from proteases.
We compared the pro-apoptotic ability of the new D-Mitoparan con-
jugates with or without the Pro9 spacer (Fig. 2). For free D-Mitoparan

Fig. 2. Comparative MTT viability measurements of SK-BR3 (HER2+) cells after a 24 h incubation with free D-peptides and PGA-D-peptide conjugates (see Tables 1 and 2). Data are
expressed as the mean ± SE (n = 3). IC50 values are shown in the right table. Similar cytotoxic effects to peptides 6 and 8 were observed for peptides 4, 5, 7 and 9.
18 M. Moreno et al. / Journal of Controlled Release 182 (2014) 13–21

Fig. 3. a) Confocal laser scanning microscopy of SK-BR3 cells after 1 h of co-incubation at 37 °C in CO2 with 25 μM of PGAmod conjugates and subsequent washing with medium. The red
fluorescence (maximum intensity projection) belongs to Rhodamine-labeled PGAmod. b) Flow cytometry results of the co-incubation of SK-BR3with 25 μM of PGAmod conjugates for 1 h
and subsequent washing with medium.

peptides (6 and 8), the IC50 was similar to that of L-Mitoparan (~8 μM).
The IC50 values of D-Mitoparan-PGAmod conjugates (see Fig. 2) were 5
times higher than those of free D-Mitoparan. However, consistent with
our supposition, the conjugate PGAmod:4, which did not carry an enzy-
matic cleavage site, did not show any cytotoxic activity, thereby indicating
the need to release D-Mitoparan to promote cellular apoptosis. A compar-
ison of the activity of the D-Mitoparan-PGAmod conjugates (Table 2, conju-
gates PGAmod:4–8) revealed a cytotoxic improvement for conjugates with
a Pro9 spacer, namely conjugates PGAmod:6 and PGAmod:8 (Fig. 2). A good
correlation between activity and D-Mitoparan release was observed when

Fig. 4. Comparative MTT viability measurements of SK-BR3 (HER2+) cells after a 24 h incubation with free D-peptides and PGA-D-peptide conjugates (see Tables 1 and 2). Data are
expressed as the mean ± SE (n = 3). IC50 are shown in the right table, these values refer to peptides 6 or 8. Similar cytotoxic effects to peptides 6 and 8 were observed for peptides 4,
5, 7 and 9.
 M. Moreno et al. / Journal of Controlled Release 182 (2014) 13–21 19

Fig. 5. Comparative MTT viability measurements of MDA-231 (HER2-) cells after a 24 h incubation with PGA-D-peptide conjugates (see Table 2). Data are expressed as the mean ± SE (n = 3).
IC50 values are shown in the right table, these values refer to peptides 6 or 8.

comparing the cytotoxicity of PGAmod:6 and PGAmod:8 (Fig. 2) with their
enzymatic digestion (Supplementary data, Fig. S9).
3.4. Improving the efficacy of pro-cytotoxic conjugates by enhancing cellular
uptake
To continue further in the amendment of our cytotoxic conjugates,
we enhanced the cellular uptake of the conjugates in order to improve
their activity. Several studies have reported the efficiency of cell-
penetrating peptides (CPPs) in facilitating the translocation of cargoes
across the plasma membrane [19]. However, although CPPs are highly
efficient in mediating the cell uptake of cargoes into most cell lines,
the lack of cell specificity could be a serious problem in vivo. For this rea-
son, we performed parallel studies with a targeting peptide to compare
the specificity and affinity of our modified systems (Table 1, peptide 13).
Peptide 13, a selected peptide (LTVSPWY) discovered by phage display
technology [20], to which we introduced a Pro9 spacer and a Cysteine at
its N-terminus, targets the over-expressed epidermal growth factor
receptor 2 (HER2). This tyrosine kinase receptor promotes the growth
of certain breast cancer cells, which tend to be more aggressive than
other types of breast cancer [21]. For the development of this proof-
of-concept we used the SK-BR3 cell line, which is HER2+ [20].
Three well-known CCPs, two positively charged (Penetratin [22] and
TAT [23]) and one negatively charged (SAP(E) [24]) were selected to de-
termine the best option for our system (Table 1, peptides 10–12). Cellu-
lar uptake was quantified by fluorescence measurements using flow
cytometry and CLSM (Fig. 3). Penetratin (see Table 1, peptide 10) was
significantly more efficient than the other CPPs at internalizing the sys-
tem into the cell (Fig. 3). A closer look at the system distribution re-
vealed that the conjugates were within endosomes and lysosomes.
Therefore, all the CPPs checked internalized PGAmod via the endosomal
pathway, the lysosome being the final fate of these conjugates. Similar
results were seen when the targeting peptide was conjugated to our
system (Fig. 3, PGAmod:13). Our system attached to the cellular surface,
where the HER2 receptor is located. After this binding, cells internalized
the conjugate by receptor-mediated endocytosis, thus promoting the
cellular uptake of our system, which showed a satisfactory rate of inter-
nalization (Fig. 3b).
Having improved the cellular uptake, we designed systems with two
different peptides in the same carrier: D-Mitoparan and Penetratin or
targeting peptide (see the combinations in Table 2). The presence of
Penetratin increases the internalization of our system, thus allowing
the maximum amount taken up into lysosomes, where cathepsin B
could then release D-Mitoparan. The cytotoxicity observed after the con-
jugation of Penetratin or the targeting peptide on the D-Mitopara-
PGAmod conjugates was doubled that with the other conjugates
(Fig. 4). Systems with Penetratin showed a slightly higher — but not sig-
nificant cytotoxic effect than with targeting peptide. Surprisingly, a sim-
ilar cytotoxic effect was detected for systems with different cleavage
sites (MMP-2 and cathepsin B). Performing a digestion of peptide 6,
which holds the MMP-2 cleavage site (GIPVSLRSK), we found that
cathepsin B was capable to cut it at two sites, thereby promoting the re-
lease of D-Mitoparan (Supplementary data, Table S4). This observation
implies that the sequence designed to be cut by MMP-2 can also be
digested by cathepsin B, thus increasing the probability of D-Mitoparan
to be released either on the cell surface by MMP-2 or in lysosomes inside
the cell by cathepsin B. In addition, in order to study the specificity of the
targeting peptide, we checked our pro-cytotoxic system with or without
this targeting peptide in the presence of a non-overexpressing HER2
breast cancer cell line, namely MDA-231 cells [20]. No significant im-
provement in cytotoxic activity was observed when targeting peptide
was present, thus validating the specificity of this peptide against the
HER2 receptor (Fig. 5).
CLSM was used to verify the cleavage and subsequent release of
mod
D-Mitoparan from the PGA . Fig. 6 shows the localization of green la-
beled D-Mitoparan (peptide 9 in Table 1) and red labeled polymer in the

Fig. 6. Confocal laser scanning microscopy (CLSM) of SK-BR3 cells after 6 h of co-incubation at 37 °C in CO2 with PGAmod:9:13 conjugate, where the concentration of peptide 9 was 25 μM.
The cells were washed with media, before taking the pictures. The fluorescence is depicted as the maximum intensity projection, where red fluorescence is due to Rhodamine-labeled
PGAmod and green fluorescence to carboxyfluorescein-labeled peptide 9 and, finally, yellow shows the colocalization. A video related to the effect of PGAmod:9:13 conjugate in SK-BR3
cells is attached in Supplementary Information (Video S1).
20 M. Moreno et al. / Journal of Controlled Release 182 (2014) 13–21
Fig. 7. Dye leakage and red blood cell hemolysis of free D-peptides and PGAmod-D-peptide conjugates. (a) Effect on membrane rupture, i.e., leakage, of Large Unilamellar Vesicles (LUVs)
containing EPC/Chol at a molar ratio of 5:1 for 1 h at RT. (b) Effect on the hemoglobin release from red blood cells for 1 h at 37 ºC. Data are expressed as the mean ± SE (n = 3). For the
conjugates, the effect on membrane rupture and hemoglobin release refers to peptide 6 or 8. Similar dye leakage and hemolysis effects to peptide 4 were observed for peptides 5, 6, 7, 8 and 9.
SK-BR3 cells after 6 h of incubation with PGAmod:9:13. We observed yel-
low spots, which indicate colocalization of the two labels inside dotted
vesicles, corresponding to endosomes and lysosomes, but also the diffu-
sion of green-labeled D-Mitoparan in the cytoplasm and even in the cel-
lular nucleus.
A video was recorded to observe the behavior of our system (Supple-
mentary Video S1). Images were acquired every 20 min for 9 h and
20 min (the time in hours is shown in the top left of the video). Dur-
ing this time, green (carboxyfluorescein-labeled peptide 9) and red
(rhodamine-labeled PGAmod) fluorescence were observed both inside
(dotted color) and outside the cells. Inside the cells, red dotting was
more abundant than green. However, outside the cells, green showed
greater fluorescence than red. These observations are explained by the
fact that the two probes are sensitive to changes in pH (fluorescence
of carboxyfluorescein decreases at acidic pH, whereas rhodamine does
so at neutral-basic pH). Significant changes in cell morphology occurred
after 3 h of incubation. Cells started to shrink when a cytotoxic concen-
tration of peptide 9 was present in the cytoplasm. Peptide 9 was cleaved
from the PGAmod and released from lysosomes to reach the cytoplasm,
promoting apoptosis. From 6 h of incubation, we observed a significant
increase in green fluorescence on the cytoplasm, presumably because of
the irreversible and accumulated damage of lysosomes. At this time
point, to facilitate monitoring of intracellular fluorescence, we added
fresh cellular medium, thus diluting the fluorescence outside cells.
This approach allowed us to see the spread of peptide 9 through the cy-
toplasm and even in the nucleus.
3.5. Lipid perturbation of liposomal membrane and in vitro
hemocompatibility testing
The soluble systems designed in this study are neutral at physiolog-
ical pH (Table 2), having electrical charges of opposite signs (zwitter-
ionic). This neutral charge could be a positive feature because highly
positive particles interact with serum proteins, thus increasing the ag-
gregation of the protein and affecting its biodistribution in vivo [25],
while negatively charged ones are impeded to promote cellular uptake.
To predict the compatibility of our systems with biological environ-
ments, we characterized their capacity to perturb the bilayer of lipo-
somes and to cause cell hemolysis. Dye leakage results (Fig. 7a) show
that the free D-Mitoparan (peptide 4) perturbed membrane phospho-
lipids, inducing leakage at very low peptide/lipid ratios, reaching a maxi-
mum around the ratio of 0.01. Comparing this effect with the conjugates,
we observed that conjugates released only 50% of dye at the maximum
ratio tested (0.04), the leakage being slightly higher when Penetratin
was present. Complementary to this, robust hemolysis was observed for
peptide 4 (Fig. 7b), but moderate hemoglobin release was promoted by
the conjugates.
4. Conclusions
Cytolytic peptides are promising drugs for cancer treatment due to
their high cytotoxic efficacy and because they do not induce cancer
drug resistance. However, they have the drawback that they are non-
specific and easily degradable in blood. Here we report an example of
a smart, targeted, pro-cytotoxic system which carries Mitoparan in a
safe way, inhibiting its cytotoxic effect when it is attached to the modi-
fied polymer. We demonstrate the innocuousness of our pro-cytotoxic
system in the first hours of contact with tumor cells. After only 3 h,
when a significant amount of the pro-cytotoxic system had been
endocyted and the Mitoparan had been released by proteases, the cyto-
toxic effects could be observed (see Video S1 attached). To overcome
solubility issues, it was necessary to modify the PGA by incorporating
PEG and capping negative charges with ethanolamines. In addition,
the enantiomer form of Mitoparan was used to maximize the cytotoxic
effect. Not only does the system include targeting peptide to favor accu-
mulation in the tumor, but also a cleavage site designed to be cut specif-
ically by proteases overexpressed only in tumor cells, such as MMP-2
and cathepsin B. This strategy allows the cytotoxic peptide to be
released with exquisite spatiotemporal control. From the pharmacolog-
ical viewpoint, our large targeted pro-cytotoxic system of 75 kDa, which
carries only 2 cytotoxic peptides with a cytotoxic effect in the micromo-
lar range, could be considered insufficiently loaded. However, we con-
sider that the combined effect of targeting to HER2 + breast tumors
via the targeting peptide and its large MW would increase its blood
half-life and favor the well-documented EPR effect enhancing the local
concentration of the cytolytic peptide at the tumor site. Our results
highlight the potential of a simple, targeted, pro-cytotoxic carrier for
clinical use in cancer treatment. Further studies are ongoing to test the
innocuous nature and efficacy of our system in vivo.
Supplementary data to this article can be found online at http://dx.
doi.org/10.1016/j.jconrel.2014.03.005.
Acknowledgement
Thanks go to Dr. Lidia Bardia for the technical support with cell im-
aging and useful discussions, the IRB Advanced Digital Microscopy
(ADM) Core Facility, and Dr. Jesús García for the technical support
with NMR data and useful discussions. We also acknowledge the Scien-
tific and Technological Centres of the University of Barcelona (CCiT UB)
Separation Techniques Analysis Facility, the NMR Facility, and the Flow
Cytometry Facility. We also thank Laura Nevola (IRB Barcelona) for
 M. Moreno et al. / Journal of Controlled Release 182 (2014) 13–21
fruitful scientific discussions. We acknowledge financial support from
MICINN-FEDER (BIO2008-799) and the Generalitat de Catalunya (XRB
and 2009SGR-1005).
References
[1] C. Leuschner, W. Hansel, Membrane disrupting lytic peptides for cancer treatments,
Curr. Pharm. Des. 10 (2004) 2299–2310.
[2] L. Holle, W. Song, E. Holle, Y. Wei, J. Li, T.E. Wagner, X. Yu, In vitro- and in vivo-
targeted tumor lysis by an MMP-2 cleavable melittin-LAP fusion protein, Int. J.
Oncol. 35 (2009) 829–835.
[3] L. Holle, W. Song, E. Holle, Y. Wei, T. Wagner, X. Yu, A matrix metalloproteinase 2
cleavable melittin/avidin conjugate specifically targets tumor cells in vitro and
in vivo, Int. J. Oncol. 22 (2003) 93–98.
[4] Y. Yamada, Y. Shinohara, T. Kakudo, S. Chaki, S. Futaki, H. Kamiya, H. Harashima, Mi-
tochondrial delivery of mastoparan with transferrin liposomes equipped with a pH-
sensitive fusogenic peptide for selective cancer therapy, Int. J. Pharm. 303 (2005)
1–7.
[5] N.R. Soman, G.M. Lanza, J.M. Heuser, P.H. Schlesinger, S.A. Wickline, Synthesis and
characterization of stable fluorocarbon nanostructures as drug delivery vehicles
for cytolytic peptides, Nano Lett. 8 (2008) 1131–1136.
[6] N.R. Soman, S.L. Baldwin, G. Hu, J.N. Marsh, G.M. Lanza, J.E. Heuser, J.M. Arbeit, S.A.
Wickline, P.H. Schlesinger, Molecularly targeted nanocarriers deliver the cytolytic
peptide melittin specifically to tumor cells in mice, reducing tumor growth, J. Clin.
Invest. 119 (2009) 2830–2842.
[7] R. Duncan, Polymer therapeutics as nanomedicines: new perspectives, Curr. Opin.
Biotechnol. 22 (2011) 492–501.
[8] C. Li, Poly(L-glutamic acid)—anticancer drug conjugates, Adv. Drug Deliv. Rev. 54
(2002) 695–713.
[9] L. Mondragon, M. Orzaez, G. Sanclimens, A. Moure, A. Arminan, P. Sepulveda, A.
Messeguer, M.J. Vicent, E. Perez-Paya, Modulation of cellular apoptosis with apoptotic
protease-activating factor 1 (Apaf-1) inhibitors, J. Med. Chem. 51 (2008) 521–529.
[10] S. Jones, C. Martel, A.S. Belzacq-Casagrande, C. Brenner, J. Howl, Mitoparan and
target-selective chimeric analogues: membrane translocation and intracellular re-
distribution induces mitochondrial apoptosis, Biochim. Biophys. Acta 1783 (2008)
849–863.
21
[11] R.C. Taylor, S.P. Cullen, S.J. Martin, Apoptosis: controlled demolition at the cellular
level, Nat. Rev. Mol. Cell Biol. 9 (2008) 231–241.
[12] A. Talvensaari-Mattila, P. Paakko, T. Turpeenniemi-Hujanen, Matrix metalloproteinase-
2 (MMP-2) is associated with survival in breast carcinoma, Br. J. Cancer 89 (2003)
1270–1275.
[13] K. Brix, A. Dunkhorst, K. Mayer, S. Jordans, Cysteine cathepsins: cellular roadmap to
different functions, Biochimie 90 (2008) 194–207.
[14] D.E. Owens III, N.A. Peppas, Opsonization, biodistribution, and pharmacokinetics of
polymeric nanoparticles, Int. J. Pharm. 307 (2006) 93–102.
[15] H. Maeda, Macromolecular therapeutics in cancer treatment: the EPR effect and be-
yond, J. Control. Release 164 (2012) 138–144.
[16] N. Yamamoto, N.S. Bryce, N. Metzler-Nolte, T.W. Hambley, Effects of enzymatic acti-
vation on the distribution of fluorescently tagged MMP-2 cleavable peptides in can-
cer cells and spheroids, Bioconjug. Chem. 23 (2012) 1110–1118.
[17] B.E. Turk, L.L. Huang, E.T. Piro, L.C. Cantley, Determination of protease cleavage site
motifs using mixture-based oriented peptide libraries, Nat. Biotechnol. 19 (2001)
661–667.
[18] S. Jones, J. Howl, Enantiomer-specific bioactivities of peptidomimetic analogues of
mastoparan and mitoparan: characterization of inverso mastoparan as a highly effi-
cient cell penetrating peptide, Bioconjug. Chem. 23 (2012) 47–56.
[19] I. Martin, M. Teixido, E. Giralt, Intracellular fate of peptide-mediated delivered
cargoes, Curr. Pharm. Des. 19 (2013) 2924–2942.
[20] M. Shadidi, M. Sioud, Identification of novel carrier peptides for the specific delivery
of therapeutics into cancer cells, FASEB J. 17 (2003) 256–258.
[21] J. Kao, K. Salari, M. Bocanegra, Y.L. Choi, L. Girard, J. Gandhi, K.A. Kwei, T. Hernandez-
Boussard, P. Wang, A.F. Gazdar, J.D. Minna, J.R. Pollack, Molecular profiling of breast
cancer cell lines defines relevant tumor models and provides a resource for cancer
gene discovery, PLoS One 4 (2009) e6146.
[22] L. Theodore, D. Derossi, G. Chassaing, B. Llirbat, M. Kubes, P. Jordan, H. Chneiweiss, P.
Godement, A. Prochiantz, Intraneuronal delivery of protein kinase C pseudosubstrate
leads to growth cone collapse, J. Neurosci. 15 (1995) 7158–7167.
[23] G. Tunnemann, R.M. Martin, S. Haupt, C. Patsch, F. Edenhofer, M.C. Cardoso, Cargo-
dependent mode of uptake and bioavailability of TAT-containing proteins and pep-
tides in living cells, FASEB J. 20 (2006) 1775–1784.
[24] I. Martin, M. Teixido, E. Giralt, Design, synthesis and characterization of a new anion-
ic cell-penetrating peptide: SAP(E), Chembiochem 12 (2011) 896–903.
[25] F. Alexis, E. Pridgen, L.K. Molnar, O.C. Farokhzad, Factors affecting the clearance and
biodistribution of polymeric nanoparticles, Mol. Pharm. 5 (2008) 505–515