Design and in vitro evaluation of branched peptide conjugates: Turning nonspecific cytotoxic drugs into tumor-selective agents

Article abstract of DOI:10.1002/cmdc.200900527

The use of peptide receptors as targets for tumor-selective therapies was envisaged years ago with the findings that receptors for different endogenous regulatory peptides are overexpressed in several primary and metastatic human tumors, and can be used as tumor antigens. Branched peptides can retain or even increase, through multivalent binding, the biological activity of a peptide and are very resistant to proteolysis, thus having a markedly higher in vivo activity compared with the corresponding monomeric peptides. Oligo-branched peptides, containing the human regulatory peptide neurotensin (NT) sequence, have been used as tumor-specific targeting agents. These peptides are able to selectively and specifically deliver effector units, for cell imaging or killing, to tumor cells that overexpress NT receptors. Results obtained with branched NT conjugated to different functional units for tumor imaging and therapy indicate that branched peptides are promising novel multifunctional targeting molecules. This study is focused on the role of the releasing pattern of drug-conjugated branched NT peptides. We present results obtained with oligobranched neurotensin peptides conjugated to 6-mercaptopurin (6-MP), combretastain A-4 (CA4) and monastrol (MON). Drugs were conjugated to oligo-branched neurotensin through different linkers, and the mode-of-release, together with cytotoxicity, was studied in different human cancer cell lines. The results show that branched peptides are very promising pharmacodelivery options. Among our drug-armed branched peptides, NT4-CA4 was identified as a candidate for further development and evaluation in preclinical pharmacokinetic and pharmacodynamic studies. This peptide-drug exhibits significant activity against pancreas and prostate human cancer cells. Consequently, this derivative is of considerable interest due to the high mortality rates of pancreas neuroendocrine tumors and the high incidence of prostate cancer.

Full text of DOI:10.1002/cmdc.200900527

DOI: 10.1002/cmdc.200900527
Design and In vitro Evaluation of Branched Peptide
Conjugates: Turning Nonspecific Cytotoxic Drugs into
Tumor-Selective Agents
Chiara Falciani,*[a] Jlenia Brunetti,[a] Chiara Pagliuca,[b] Stefano Menichetti,[b] Lucia Vitellozzi,[b]
Barbara Lelli,[a] Alessandro Pini,[a] and Luisa Bracci[a]
The use of peptide receptors as targets for tumor-selective
therapies was envisaged years ago with the findings that re-
ceptors for different endogenous regulatory peptides are over-
expressed in several primary and metastatic human tumors,
and can be used as tumor antigens. Branched peptides can
retain or even increase, through multivalent binding, the bio-
logical activity of a peptide and are very resistant to proteoly-
sis, thus having a markedly higher in vivo activity compared
with the corresponding monomeric peptides. Oligo-branched
peptides, containing the human regulatory peptide neuroten-
sin (NT) sequence, have been used as tumor-specific targeting
agents. These peptides are able to selectively and specifically
deliver effector units, for cell imaging or killing, to tumor cells
that overexpress NT receptors. Results obtained with branched
NT conjugated to different functional units for tumor imaging
and therapy indicate that branched peptides are promising
novel multifunctional targeting molecules. This study is fo-
cused on the role of the releasing pattern of drug-conjugated
branched NT peptides. We present results obtained with oligo-
branched neurotensin peptides conjugated to 6-mercaptopur-
in (6-MP), combretastain A-4 (CA4) and monastrol (MON).
Drugs were conjugated to oligo-branched neurotensin through
different linkers, and the mode-of-release, together with cyto-
toxicity, was studied in different human cancer cell lines. The
results show that branched peptides are very promising phar-
macodelivery options. Among our drug-armed branched pep-
tides, NT4–CA4 was identified as a candidate for further devel-
opment and evaluation in preclinical pharmacokinetic and
pharmacodynamic studies. This peptide–drug exhibits signifi-
cant activity against pancreas and prostate human cancer cells.
Consequently, this derivative is of considerable interest due to
the high mortality rates of pancreas neuroendocrine tumors
and the high incidence of prostate cancer.
Introduction
The selective targeting of tumor cells is the goal of modern
cancer therapy aimed at overcoming the nonspecific toxicity
of most anticancer agents against normal cells. The discovery
that receptors for different endogenous regulatory peptides
are overexpressed in several primary and metastatic human
tumors led to their use as tumor antigens.[1] The bottleneck for
development of peptides as drug has always been their ex-
tremely short half-life due to physiological degradation by
peptidases and proteases. Branched peptides can retain (or
even increase through multivalent binding) the biological ac-
tivity of a peptide and are very resistant to proteolysis.[2–8] Con-
sequently, these branched peptides have a markedly higher in
vivo activity when compared to the monomeric peptides.
We have been studying protease-resistant branched pep-
tides as tumor targeting agents by using tetra-branched pep-
tides (NT4) containing the human regulatory peptide neuro-
tensin (NT) sequence.[5–6] Neurotensin receptors are overex-
pressed in several human malignancies, such as colon, pancre-
atic, prostate and small-cell lung cancer. We have been using
NT4 conjugated to different functional units for tumor imaging
and therapy, and found that NT4 conjugated to methotrexate
produced 60% reduction of tumor growth in xenografted
mice.[6]
Results obtained with NT4 indicated that branched peptides
are promising novel multifunctional targeting molecules,
which might allow cancer detection and therapy by means of
the same molecule, with no modification in target binding, but
rather a simple exchange of functional units.[6] Since cancer
cells are very different from one another in terms of drug sen-
sibility, not only in different tumors but in different patients
and stages of the disease, this approach prefigures the synthe-
sis of a number of constructs conjugated with differently
acting chemotherapeutics.
The type of linkage between effector unit (drug/imaging
agent etc.) and peptide is obviously crucial for this type of ap-
proach. The choice must be driven by two issues: 1) The
nature of the drug functional groups available for coupling
[a] Dr. C. Falciani, Dr. J. Brunetti, Dr. B. Lelli, Dr. A. Pini, Prof. L. Bracci
Department of Molecular Biology, University of Siena
Via Fiorentina 1, 53100 Siena (Italy)
Fax: (+39)0577-234903
E-mail: chiarafalciani@unisi.it
[b] Dr. C. Pagliuca, Prof. S. Menichetti, L. Vitellozzi
Department of Chemistry ‘U. Schiff‘, University of Florence
Via della Lastruccia 3–13, 50019 Sesto Fiorentino, Florence (Italy)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cmdc.200900527.
ChemMedChem 2010, 5, 567 – 574
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
567

C. Falciani et al.
MED
with the peptide; and 2) the mechanism-of-action of the drug.
When a prodrug acts without being released from the carrier
unit, a strong linker is preferred. However, if a drug has to be
released in order to interact with the intracellular target, the
linker must be cleavable. In latter, the linker has to be chosen
properly—not too labile or leakage will occur during drug dis-
tribution, but not too robust or the pharmacological action
will be impaired.
not only very effective in disrupting tumor vasculature but also
causes cell death by inhibiting AKT activity.[14] The novel antitu-
mor vascular targeting agent CA4-P (disodium combretastatin
A4-3-O-phosphate), a water-soluble prodrug, is the first com-
bretastatin analogue to enter clinic trials.[15] Though the agent
achieved the pharmacodynamic endpoint in suppressing
tumor perfusion, it showed strong cardiotoxicity. This side
effect might be an obstacle in the clinical development of this
agent. Thereby, new CA4 derivatives are under investigation to
overcome the problem.[16] In our case, the free phenolic OH of
CA4 was modified to a carboxylic acid using an ester link (5) or
an ether bond (6) (Scheme 2).
Results and Discussion
The objective of the present study was to transform efficient
but nonspecific drugs into tumor-selective agents. For this pur-
pose, we chose to combine branched neurotensin (NT4) to
monastrol (MON), combretastatin A4 (CA4) and 6-mercaptopur-
ine (6-MP).
6-Mercaptopurine (6-MP) is used to treat some types of
cancer, primarily leukemia, and acts by inhibiting the de novo
synthesis of purines.[17] The clinical application of 6-MP is limit-
ed by its toxicity towards bone marrow and liver. A few at-
tempts have been made to design new 6-MP prodrugs that
are activated in the proper area so that general toxicity is re-
duced;[18] one of these is 3-(9H-purin-6-ylthio)acrylic acid (PTA;
8).[19]
Monastrol (MON) is the first specific small-molecule inhibitor
of HsEg5,[10] and causes collapse of bipolar spindles in dividing
cells.[11–12] Its analogue SB-715992 (ispinesib; Cytokinetics, San
Francisco, USA) is currently being studied in multiple phase II
clinical trials. To allow the cou-
pling of monastrol to NT4, the
phenolic OH of racemic MON
(1a) was transformed to a car-
boxylic acid group through an
activated ester link (2) or an
ether bond (3) (Scheme 1).
Combretastatin A4 (CA4) is a
naturally occurring stilbene that
shows
potent
cytotoxicity
against
a broad spectrum of
human cancer cell lines, includ-
ing multidrug resistant lines. It
belongs to a class of antimitotic
agents that bind to the colchi-
cine site leading to cell-cycle
arrest and apoptosis.[13] CA4 is
Scheme 2. Synthesis of acid modified combretastatin derivatives 5 and 6. Reagents and conditions: a) DIPEA,
TBDMSCl, DMF, RT, 4 h; b) BrCH2COOtBu, K2CO3, CH3CN, 608C (2 h) ! RT (o/n); c) 1. NaBH4, MeOH, 08C (1 h) ! RT
(o/n); 2. PBr3, CH2Cl2, ꢀ58C, 45 min; 3. PPh3, PhCH3, RT, 24 h; d) nBuLi, THF, ꢀ238C (4 h) ! RT (18 h); e) 1. 4a,
TBAF·3H2O, THF, RT, 20 min; 2. glutaric anhydride, CH2Cl2, RT (16 h) ! reflux (24 h); f) 4b, LiOH, THF/MeOH, RT,
2 h. See Supporting Information for further details.
PTA (8) acts as a Michael acceptor, and, in the pres-
ence of nucleophiles, it undergoes addition–elimina-
tion processes releasing 6-MP. Biological activity of 8
is promoted by its glutathione (GSH)-mediated con-
version to 6-MP, thus PTA is a prodrug of 6-MP. PTA
was particularly interesting to us since it could be
used to link 6-MP to branched neurotensin through
its carboxylic group, and it can be easily released
once taken up by the tumor cell. In facts, reduced
GSH is particularly abundant in tumor cells where all
reducing pathways are activated due to oxidative
stress.
Since the efficiency of the linker in releasing the
drug is crucial for this kind of prodrug, we decided
Scheme 1. Synthesis of monastrol 1a (R=H) and acid derivatives 2 and 3. Reagents and
conditions: a) Yb(OTf)3·ꢁH2O, THF, N2, reflux, 7–12 h; b) 1a, glutaric anhydride, CH2Cl2, N2,
to prepare a few different types of 6-MP derivatives
to be conjugated to the carrier peptide. Due to the
presence of the a,b-unsaturated ketone, 6-(2-acetylvi-
RT (24 h) ! reflux (27 h); c) 1b, TFA, CH2Cl2, RT 1 h. See Supporting Information for fur-
,
ther details.
568
www.chemmedchem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2010, 5, 567 – 574

Branched Peptide Conjugates for Tumor-Cell-Selective Drug Delivery
nylthio)purine (AVTP) was shown to undergo an addition–elim-
ination reaction with GSH to yield 6-MP more easily than PTA.
We prepared derivative 9 as a suitable analogue of AVTP, with
a a,b-unsaturated carbonyl moiety linked to the carboxylic
acid (Scheme 3). We also designed the 6-MP disulfide deriva-
tive 10; in this case the release of 6-MP inside tumor cells
should take place through a simple thiol–disulfide exchange
with, for example, GSH (Scheme 4).
acid (Scheme 2) under Wittig conditions. Chromatographic
separation of E and Z olefins allowed the isolation of 4a (R=
SiMe2tBu). Desilylation with tetra-n-butylammonium fluoride
(TBAF) followed by reaction with glutaric anhydride afforded
compound 5 bearing the required carboxylic acid functionality.
The introduction of an ether linkage was carried out at the be-
ginning of the synthetic sequence by reacting isovanillin with
tert-butylbromoacetate. After Wittig olefination and separation
of the diastereoisomers, combretastatin analogue 4b (R=
CH2C(O)OtBu) was isolated. Subsequent hydrolysis under basic
conditions (LiOH in THF/MeOH) afforded compound 6, with
the acid moiety linked by a robust ether bridge to the combre-
tastatin core (Scheme 2).
The transformation of 6-mercaptopurine (7; 6-MP) into 3-
(9H-purin-6-ylthio)acrylic acid (8; PTA) was achieved using a
slightly modified literature procedure,[19] where equimolecular
amounts of sodium methanoate, propiolic acid and 6-MP
(Scheme 3).
Scheme 3. Synthesis of 3-(9H-purin-6-ylthio)acrylic (PTA) 8. Reagents and con-
ditions: MeOH, MeONa, H-CꢁCꢀCOOH, N2, reflux, o/n. See Supporting Infor-
mation for further details.
After several attempts, acid-modified AVTP analogue 9 was
obtained by desymmetrization of glutaric anhydride to the
mono-tert-butyl ester followed by preparation of the corre-
sponding Weinreb’s amide. Reaction of the amide with an
excess of ethynyl magnesium bromide gave the corresponding
alkynyl-g-ketoester, which subsequently underwent Michael ad-
dition to 6-MP and hydrolysis of the tert-butyl ester with TFA
to give the desired compound
(Scheme 4).
9 in reasonable yield
10-Mercaptoundecanoic acid was transformed into the cor-
responding sulfenyl chloride by reaction with SO2Cl2 in tolu-
ene. The crude sulfenyl chloride was reacted with 6-MP in dry
pyridine to obtain the acid disulfide 10 (Scheme 5).
Scheme 4. Synthesis of 6-mercaptopurine g-ketoacid derivative 9. Reagents
and conditions: a) NHS, DMAP, tBuOH, TEA, RT (30 min) ! reflux (24 h);
b) 1. DMF, HBTU, DIPEA, RT, 10 min; 2. DMF, Wienreb’s amine, RT, 3 h; c) THF,
ethynyl-MgBr, ꢀ788C (30 min) ! RT (3 h); d) 7, MeOH, MeONa, RT, o/n;
e) TFA, CH2Cl2, RT, 2 h. See Supporting Information for further details.
Scheme 5. Synthesis of 6-mercaptopurine disulfide derivative 10. Reagents
and conditions: a) SO2Cl2, toluene, RT, 15 min; b) 7, Py, 848C, 2 h. See Sup-
porting Information for further details.
Chemistry
Monastrol (1a, R=H) can be efficiently prepared in high yield
via a Yb(OTf)3-catalyzed, three-components Biginelli reaction.[20]
The reaction of 1a with glutaric anhydride in dry dichlorome-
thane allowed the preparation of derivative 2 with the re-
quired carboxylic acid functionality. The reaction of 3-hydroxy-
benzaldehyde with tert-butylbromoacetate was used to intro-
duce the ether link. The condensation reaction, under the
same conditions used above, allowed the isolation of the
modified Biginelli adduct 1b, which was transformed into the
ether-linked acid derivative 3 by hydrolysis with trifluoroacetic
acid (TFA) in dichloromethane (Scheme 1).
Eventually, products 2, 3, 5, 6, 8, and 9 were conjugated to
NT4 on solid phase (Scheme 6). Peptide 11 was synthesized on
solid phase using Fmoc-Lys(Dde)-OH as the first amino acid on
TGS RAM resin, then the second amino acid was Fmoc-PEG12-
OH. Two coupling steps with Fmoc-Lys(Fmoc)-OH were used to
build the core and the tetramer was synthesized using Boc-
Arg(Pbf)-OH as the final amino acid of the neurotensin se-
quence. The Dde protecting group was removed using hydra-
zine and the free amino group was coupled with Fmoc-PEG12-
OH. This spacer was chosen by considering the following syn-
thetic and biological issues: 1) The hindrance of the branched
peptide might impair the access of the activated drug and
therefore give poor reaction yields; 2) the drug moiety, if close-
ly linked to the branched peptide, might jeopardize receptor
Combretastatin derivatives 5 and 6 were obtained using an
adapted literature procedure,[21] involving the reaction of isova-
nillin, protected as the tert-butyldimethylsilyl (TBDMS) ether,
with a phosphonium salt prepared from 3,4,5-trimethoxy gallic
ChemMedChem 2010, 5, 567 – 574
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
569

C. Falciani et al.
MED
Scheme 6. Synthesis of NT4 derivatives 14–19. Reagents and conditions: a) 1. N2H4 (2%), DMF, 10 min, RT; 2. Fmoc-PEG12-OH, HBTU, DIPEA, 2 h; b) 1. Piperidine
(20%), 10 min; 2. 2, 3, 5, 6, 8, and 9, HBTU, DIPEA, 2 h; c) TFA (95%), 2 h. See Experimental Section for further details.
recognition; 3) for slow-/non-releasing compounds, the prox-
imity of the branched peptide might compromise drug–target
interaction within the cell. Subsequently, the Fmoc group was
removed to obtain primary amine 12, enabling the coupling
with 2, 3, 5, 6, 8, and 9 and formation of the peptide conju-
gates 13. Acid hydrolysis (TFA) concurrently cleaved the side
chain protecting groups and released the conjugates from the
resin allowing the isolation of derivatives 14–19 (Scheme 6).
Prodrugs 8, 9 and 10 were incubated in the presence of 1, 5
and 15 equivalents of GSH. The release of 6-MP at different
time intervals was measured by HPLC. 6-MP disulfide 10 was
extremely reactive, and 77% of 6-MP was already released
after 5 min (Table 1). We therefore gave up with the develop-
ment of this prodrug.
Conversely, PTA (8) was found to be too stable since only
10% of 6-MP was released after 18 h in the presence of 15
equivalents of GSH (Table 1). The most interesting releasing
profile was shown by ketone 9, which released 86% of 6-MP
after 135 min in the presence of 1 equivalent of GSH and
100% release of the drug after 30 min with 5 equivalents of
GSH (Figure 1). The possibility to modulate the release of the
drug as a function of the endothiol concentration inside the
cell is an important feature of the prodrug that was then car-
ried forward for further characterization.
Drug release from conjugates 14–19.
The rate of drug release from the branched peptides was eval-
uated in the following experimental models. Prodrugs such as
PTA (8), 9 and 10 can release 6-MP in the presence of nucleo-
philes, like GSH, which promotes an addition–elimination reac-
tion on the electron-poor double bond of 8 and 9 or a thiol–
disulfide interchange for 10. Since solid tumors are often in an
hypoxic environment where GSH is normally abundant,[19] this
mechanism of drug release should provide some accumulation
of the drug in the tumor cells.
MON and CA4 have never been proved to be inactivated by
conjugation, so the ability of conjugates 14, 15, 16 and 17 to
act as prodrugs was not known. It was necessary to know the
570
www.chemmedchem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2010, 5, 567 – 574

Branched Peptide Conjugates for Tumor-Cell-Selective Drug Delivery
ure 2d–f) and with that of an
unrelated tetra-branched pep-
tide (U4) conjugated to the
same drug (Figure 2g–i).
Table 1. Release of 6-MP, MON and CA4.
Prodrug
Drug release data
Drug
Mode
Amount [%]
Time
18 h
All branched peptides were
tested in an experimental proto-
col involving six days incubation
of the cells. In the case of slow-/
non-releasing adducts (15, 17
and 18), the cell medium was
unchanged for the full six days
after treatment. However, when
the fast releasing peptides were
tested in the same way, overlap-
8 (PTA)
6-MP
6-MP
6-MP
MON
MON
CA4
GSH-mediated retro-Michael
GSH-mediated retro-Michael
GSH-mediated thiol–disulfide interchange
Ester hydrolysis
Amide hydrolysis
Ester hydrolysis
10
9 (AVTP derivative)
10 (6-MP disulfide)
14 (NT4–MON ester)
15 (NT4–MON ether)
16 (NT4–CA4 ester)
17 (NT4–CA4 ether)
see Figure 1
77
50
undetectable
50
undetectable
5 min
2 h
2 h
CA4
Amide hydrolysis
Compounds were incubated at 378C in a phosphate buffer solution (pH 7.4) in the presence of 1, 5 and 15
equivalents of GSH. The release of 6-MP at different time intervals was measured by HPLC.
ping cytotoxicity results were observed for the NT4 (14, 16
and 19) and U4 conjugated peptides (data not shown). This
clearly indicated that the chemotherapeutic moiety was re-
leased into the cell medium and the drug was internalized into
the cells, probably by membrane diffusion. This became evi-
dent when observing the behavior of the unrelated conjugat-
ed peptides. These conjugates exhibited a cytotoxic effect that
could only be ascribed to the release of free drug. Unrelated
peptides cannot be internalized by the cells, as we demon-
strated in a previous paper,[6] and furthermore, Figure 2i shows
that the U4-CA4 ether, which only slowly releases CA4 or not
at all, had absolutely no activity in all the three cell lines.
Therefore the experimental protocol was modified for fast re-
leasing compounds. An additional cell washing one hour after
treatment of the cells with the test compound was introduced.
This procedure was essential to follow the activity of fast re-
leasing agents as it allowed the exclusion of any additive
effect of the released free drug. As expected, this procedure
lowered sensibly the IC50 values, and this had to be taken into
account when considering the overall worth of the tumor tar-
geting peptide.
~
Figure 1. 6-MP release by conjugate 9 in the presence of GSH (1 mm,
;
&
5 mm, ).
mode-of-action by which the drug is released to identify the
actual cytotoxic species.
MON and CA4 derivatives 14 and 16 were conjugated to
NT4 via a bifunctional linker that bears an ester bond at one
end and an amide at the other end, therefore drug release can
occur by ester hydrolysis. Drug release from derivatives 15 and
17, which have an ether bond at one end and an amide link at
the other, most likely proceeds via amide hydrolysis. Indeed,
when 14, 15, 16 and 17 were incubated with PANC-1 cells for
different lengths of time, analysis showed that 50% release of
MON and CA4 by hydrolysis of ester derivatives 14 and 16 was
observed after 2 h, while ether conjugated drugs 15 and 17
were found intact either in the supernate or in lysate even
after 12 h incubation. These results and those of 6-MP pro-
drugs 8 and 9 allowed us to classify the functionalized
branched peptides as fast releasing (14, 16 and 19) or slow-/
non-releasing (15, 17 and 18).
To assess the loss of initial drug concentration by means of
the additional washing, the behavior of free CA4 was assessed
under both assay conditions. Free CA4 exhibited an IC50 value
of 2.8ꢁ10ꢀ9 and 2.4ꢁ10ꢀ9 when assayed in PANC-1 and PC-3
cells without the additional wash and change of cell medium
after one hour (Figure 2 f). The IC50 values of CA4 increased to
micromolar range (PANC-1, 6ꢁ10ꢀ7 m; PC-3, 2.7ꢁ10ꢀ7 m) when
cells were washed after one hour and re-incubated for six days
in fresh medium, and to complete inefficiency for HT29 (Fig-
ure 2e).
Monastrol (MON) was not particularly efficacious as a free
drug in our model (Figure 2d). However, when bound to the
carrier peptide, it gained efficiency (Figure 2a and g). The best
peptide conjugate that deserves interest for future develop-
ment is NT4-CA4 ester (16) with an IC50 value of 5ꢁ10ꢀ7 m in
PANC-1 cells and 8.5ꢁ10ꢀ7 m in PC-3 cells. For compound 16,
the activity of the drug was increased by conjugation to NT4
through a fast releasing linker (Figure 2b), and this was evident
though the unfavorable conditions of the additional washing
protocol.
Cytotoxicity
Tetra-branched NT (NT4) conjugates with MON (14 and 15),
CA4 (16 and 17) and 6-MP (18 and 19) were tested on HT-29,
PANC-1 and PC-3 tumor cell lines, which express NT recep-
tors.[22–24] Cellular toxicity was measure using these three cell
lines for all the drug–NT4 conjugates (Figure 2a–c) and com-
pared with cytotoxicity of the corresponding free drugs (Fig-
Comparison of results obtained with CA4 in slow and fast re-
leasing conjugates indicated that greater efficacy was observed
ChemMedChem 2010, 5, 567 – 574
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
571

C. Falciani et al.
MED
Figure 2. Panels a, b and c: drug-conjugated peptides; panels d, e and f: free drugs; panels g, h and i: drug conjugated to an unrelated sequence. Cytotoxici-
&
*
*
ty of 15, 16 and 17 was tested on NT-receptor-expressing tumor cell lines: HT-29 ( ), PANC-1 ( ) and PC-3 ( ). All cells were incubated for 6 days at 378C
with no changing of the medium, with the exception of those treated with 14 and 16, which were incubated for 1 h at 378C, then washed and kept for
6 days at 378C.
when the active agent was released inside the cell compared
to when CA4 remains bound to the carrier peptide. Nonethe-
less, NT4–MON ether 15 and NT4–CA4 ether 17 showed a
moderate cytotoxic activity in all the three cell lines (Figure 2a
and c, respectively), and this observation suggests that they
might interact with their cellular targets despite the branched
peptide burden. We consider this finding very important be-
cause it indicates that these two drugs might be used as non-
releasable prodrugs with improvement to their nonselective
cytotoxicity.
gate 19 did not give the expected results; in fact, it showed
extremely poor cytotoxic activity (data not shown). The mode-
of-release of 6-MP from prodrug 9 described in Table 1 was
ideal but probably did not match the actual cell internalization
and activation of prodrug 19, which did not produce the ex-
pected cytotoxicity.
The most prominent feature of the branched peptide carrier
NT4 is its target specificity. This was demonstrated by the loss
of activity, in part or in full, of the active drugs conjugated to
the unrelated branched peptide U4 when tested in the NT-re-
ceptor expressing cell lines (Figure 2g–i). Changing the mecha-
nism of membrane transport by switching to a peptide recep-
tor-mediated mechanism can significantly modify drug trans-
port from outside to inside the cell. Moreover, conjugation to
branched peptides might also impair mechanisms of drug
export from inside the cell, trapping the conjugated drug in
the target cell. This might be extremely important for the ther-
apy of tumors that overexpress NT receptors and do not re-
spond to classical chemotherapy.
As described above, prodrugs such as PTA (8), 9 and 10
were tested for their ability to release the active drug before
being conjugated to the branched NT peptide. Disulfide 10
was discarded because of its extreme fragility. However, com-
pounds 8 and 9 were used to prepare 18 and 19, respectively.
As expected, 6-MP derivative 18, a very slow releasing adduct,
was completely inactive in this in vitro model (data not
shown). Furthermore, in spite of the promising results ob-
served for the drug release from prodrug 9, the peptide conju-
572
www.chemmedchem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2010, 5, 567 – 574

Branched Peptide Conjugates for Tumor-Cell-Selective Drug Delivery
enable coupling with 2, 3, 5, 6, 8 and 9. Peptides were cleaved
from the resin and deprotected. Unrelated analogues (Asp-Asp-His-
Ser-Val-Ala) were synthesized analogously and the N terminus was
acetylated before Dde removal. All peptides are released as amides
from the resin. HPLC purification was performed using a C18 Vydac
column on a Shimadzu apparatus (Milan, Italy). Solvent system:
A) H2O+0.1% TFA and B) methanol; gradients of B were run for
30 min at flow rates of 0.8 mLminꢀ1 and 4 mLminꢀ1 for analytical
and preparatory procedures, respectively. All compounds were also
characterized on an Ettan Maldi-ToF mass spectrometer (Amersham
Biosciences, Buckinghamshire, UK). Stability of the products to pro-
teases was tested as previously described.[2]
Conclusions
Tumor targeting agents currently in the clinics are either small
molecules or antibodies. The former are mainly designed to
impair specific tumor pathways, the latter are designed to se-
lectively bind to tumor antigens. As a consequence, the main
drawback of the small molecules is poor selectivity over
healthy cells due to shared biological pathways (e.g., tyrosine
kinase inhibitors). The major drawback for tumor-specific anti-
bodies is their potency; an antibody alone is usually not suffi-
ciently cytotoxic and must rather be used as a carrier of cyto-
toxic moieties. So far, the most promising molecules are those
where an effector unit (drug/imaging agent) is delivered spe-
cifically to the tumor cell via antigen recognition by a bound
selective ligand, combining in a single molecule efficacy and
strict specificity.[25]
NT4-MON ester 14: HPLC: tR =26.51 min (5!80% solvent B); tR =
22.20 min (5!70% B); MALDI-ToF m/z: [M+H]+ calcd: 4785.82,
found: 4787.74.
NT4-MON ether 15. HPLC: tR =25.40 min (1!80% solvent B); tR =
19.80 min (5!70% B); MALDI-ToF m/z: [M+H]+ calcd: 4729.44,
found: 4729.44.
The results reported herein indicate that branched, drug-
armed peptides are very promising as drug delivery agents.
The main benefit of using this kind of carrier over administra-
tion of the free drug is improved selectivity, efficacy, and also
the circumvention of drug-resistance mechanisms. Drug target-
ing, an approach by which normal tissues are spared the toxic
effect of the free drugs, is successfully being applied to already
existing chemotherapeutics, which are all strongly toxic. How-
ever, those agents discarded from development due to poor
cell uptake could benefit from this approach. Moreover, tumor
cells often develop drug resistance via different mechanisms,
including decreased uptake or increased export of the drug by
the cells; pharmacodelivery of branched peptide–drug conju-
gates would circumvent this mechanisms.
NT4-CA4 ester 16. HPLC: tR =25.13 min (1!80% B), tR =26.12 min
(20!70% B); MALDI-ToF m/z: [M+H]+ calcd: 4809.89, found:
4810.89.
NT4-CA4 ether 17. HPLC: tR =22.22 min (2!80% B), tR =22.40 min
(30!70% B); MALDI-ToF m/z: [M+H]+ calcd: 4553.76, found:
4556.96.
NT4-PTA 18. HPLC: tR =28.67 min (20!80% B); tR =14.70 min (2!
60% B); MALDI-ToF m/z: [M+H]+ calcd: 4600.61, found: 4602.17.
NT4–6MP 19. HPLC: tR =23.02 min (5!80% B), tR =17.63 min (1!
70% B); MALDI-ToF m/z: [M+H]+ calcd: 4656.67, found: 4658.12.
Among our drug-armed branched peptides, we identified a
new candidate for preclinical development in pharmacokinetic
and pharmacodynamic studies. NT4-CA4 ester exhibits signifi-
cant activity against pancreatic and prostate human cancer
cells. Consequently, this derivative is of considerable interest
due to the high mortality rates of pancreas neuroendocrine
tumors and the high incidence of prostate cancer.
Biology
Cell cultures: HT-29 human colon adenocarcinoma, PANC-1 human
pancreatic epitheloid carcinoma and PC-3 human prostate carcino-
ma cells were grown in McCoy’s 5A, RPMI 1640 and Ham’s F-12
medium, respectively, (supplemented with 10% fetal bovine
serum, 200 mgmLꢀ1 glutamine, 100 mgmLꢀ1 streptomycin,
60 mgmLꢀ1 penicillin) and maintained at 378C, 5% CO2. Cell lines
were purchased from Istituto Zooprofilattico Sperimentale (Brescia,
Italy).
Cytotoxic unit release: model compounds PTA (8), a,b-unsaturated
ketone 9 and 6-MP disulfide 10 were incubated at 378C in a phos-
phate buffer solution (pH 7.4) in the presence of 1, 5 and 15 equiv-
alents of GSH. The 6-MP release was measured at different time in-
tervals by HPLC of the crude mixture.
Experimental Section
Chemistry
Details on the preparation, isolating and characterization of com-
pounds 1–10 are reported in the Supporting Information. Peptide
synthesis was performed using a MultiSynTech Syro, automated,
multiple peptide synthesizer (Witten, Germany) by standard Fmoc
chemistry. Protected l-amino acids, coupling reagents (DIPEA and
HBTU) and TGS RAM resin were purchased from Iris Biotech (Markt-
redwitz, Germany).
The released amount of MON (1a) and CA4 (4a) was obtained by
incubating PANC-1 cells (1ꢁ106) with peptides 14, 15, 16 and 17
(100 mm) at 378C for different time intervals. Cells were centri-
fuged, washed and then lysed in water after freezing and thawing.
Supernate and lysed cells were analyzed by mass spectrometry,
after addition of NT4 as an internal standard, using a Ettan Maldi-
ToF mass spectrometer in reflection mode with an accelerating
voltage of 20 kV.
Peptides 14, 15, 16, 17, 18 and 19 were synthesized using Fmoc-
Lys(Dde)-OH as the first amino acid on TGS RAM resin, and the
second amino acid was Fmoc-bAla-OH. Two coupling steps with
Fmoc-Lys(Fmoc)-OH were used to build the core, and the tetramer
was synthesized using Boc-Arg(Pbf)-OH as the final amino acid of
the neurotensin sequence. The Dde protecting group was removed
using hydrazine in DMF (2% v/v) for 10 min at RT. The free amino
group was coupled with Fmoc-NH-PEG(12)-COOH (Iris Biotech,
Marktredwitz, Germany). The Fmoc group was then removed to
Cytotoxicity of drug-conjugated branched NT: HT-29, PANC-1 or PC-3
cells were plated at a density of 2.5ꢁ104 per well in 96-well micro-
plates. Different concentrations of free or NT4-conjugated drugs
(0.15–30 mmolLꢀ1), were added 24 h after plating. Cells were
grown without changing the medium for 6 days. Cells treated with
14 and 16 were washed after 1 h incubation and then left 6 days
ChemMedChem 2010, 5, 567 – 574
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
573

C. Falciani et al.
MED
[11] L. Luo, J. D. Carson, D. Dhanak, J. R. Jackson, P. S. Huang, Y. Lee, R. Sako-
wicz, R. A. Copeland, Biochemistry 2004, 43, 15258–15266.
[12] Y. Yan, V. Sardana, B. Xu, C. Homnick, W. Halczenko, C. A. Buser, M.
Schaber, G. D. Hartman, H. E. Huber, L. C. Kou, J. Mol. Biol. 2004, 335,
547–543.
at 378C with the same medium. Growth inhibition was assessed
using
a 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bro-
mide (MTT) test. IC50 values were calculated by nonlinear regres-
sion analysis using GraphPaD prism 3.02 software.
[13] S. L. Gwaltney II, H. M. Imade, K. J. Barr, Q. Li, L. Gehrke, R. B. Credo, R. B.
Warner, J. Y. Lee, P. Kovar, J. Wang, M. A. Nukkala, N. A. Zielinski, D.
Frost, S. C. Ng, H. L Sham, Bioorg. Med. Chem. Lett. 2001, 11, 871–874.
[14] H. L. Lin, S. H. Chiou, C. W. Wu, W. B. Lin, L. H. Chen, Y. P. Yang, M. L. Tsai,
Y. H. Uen, J. P. Liou, C. W. Chi, J. Pharmacol. Exp. Ther. 2007, 323, 365–
373.
[15] A. Dowlati, K. Robertson, M. Cooney, W. P. Petros, M. Stratford, J. Jes-
berger, N. Rafie, B. Overmoyer, V. Makkar, B. Stambler, A. Taylor, J. Waas,
J. S. Lewin, K. R. McCrae, S. C. Remick, Cancer Res. 2002, 62, 3408–3416.
[16] D. E. Durrant, J. Richards, A. Tripathi, G. E. Kellogg, P. Marchetti, M. Eleo-
pra, G. Grisolia, D. Simoni, R. M. Lee, Invest. New Drugs 2009, 27, 41–52.
[17] S. Coulthard, L. Hogarth, Invest. New Drugs 2005, 23, 523–532.
[18] T. Miron, F. Arditti, L. Konstantinovski, A. Rabinkov, D. Mirelman, A. Ber-
rebi, M. Wilchek, Eur. J. Med. Chem. 2009, 44, 541–550.
Acknowledgements
We thank Silvia Scali her valuable contribution in the peptide
synthesis. .This work was supported by grants from the Italian
Association for Cancer Research (AIRC) and the Istituto Toscano
Tumori and Fondazione Monte dei Paschi di Siena.
Keywords: antitumor agents
selectivity · peptides
·
cancer
·
cytotoxicity
·
[19] S. Gunnarsdottir, A. A. Elfarra, J. Pharmacol. Exp. Ther. 1999, 290, 950–
957.
[20] A. Dondoni, A. Massi, S. Sabbatini, Tetrahedron Lett. 2002, 43, 5913–
5916.
[21] G. R. Pettit, S. B. Singh, M. R. Boyd, E. Hamel, R. K. Pettit, J. M. Schmidt, F.
Hogan, J. Med. Chem. 1995, 38, 1666–1672.
[1] J.C Reubi, Endocr. Rev. 2003, 24, 389–427.
[2] L. Bracci, C. Falciani, B. Lelli, L. Lozzi, Y. Runci, A. Pini, M. G. De Montis, A.
Tagliamonte, P. Neri, J. Biol. Chem. 2003, 278, 46590–46595.
[3] A. Pini, Y. Runci, C. Falciani, B. Lelli, J. Brunetti, S. Pileri, M. Fabbrini, L.
Lozzi, C. Ricci, A. Bernini, F. Tonello, F. Dal Molin, P. Neri, N. Niccolai, L.
Bracci, Biochem. J. 2006, 395, 157–163.
[22] V. Navarro, S. Martin, J. Mazella, Peptides 2006, 27, 2502–2507.
[23] S. Guha, O. Rey, E. Rozengurt, Cancer Res. 2002, 62, 1632–1640.
[24] L. Seethalakshmi, S. P. Mitra, P. R. Dobner, M. Menon, R. E. Carraway,
Prostate 1997, 31, 183–192.
[25] G. D. Lewis Phillips, G. Li, D. L. Dugger, L. M. Crocker, K. L. Parsons, E.
Mai, W. A. Blꢂttler, J. M. Lambert, R. V. Chari, R. J. Lutz, W. L. Wong, F. S.
Jacobson, H. Koeppen, R. H. Schwall, S. R. Kenkare-Mitra, S. D. Spencer,
M. X. Sliwkowski, Cancer Res. 2008, 68, 9280–9290.
[4] A. Pini, A. Giuliani, C. Falciani, Y. Runci, C. Ricci, B. Lelli, M. Malossi, P.
Neri, G. M. Rossolini, L. Bracci, Antimicrob. Agents Chemother. 2005, 49,
2665–2672.
[5] C. Falciani, L. Lozzi, A. Pini, F. Corti, M. Fabbrini, A. Bernini, B. Lelli, N.
Niccolai, L. Bracci, Chem. Biol. Drug Des. 2007, 69, 216–221.
[6] C. Falciani, M. Fabbrini, A. Pini, L. Lozzi, B. Lelli, S. Pileri, J. Brunetti, S.
Bindi, S. Scali, L. Bracci, Mol. Cancer Ther. 2007, 6, 2441–2448.
[7] C. Falciani, A. Pini, L. Bracci, Expert Opin. Biol. Ther. 2009, 9, 171–178.
[8] A. Pini, C. Falciani, L. Bracci, Curr Protein Pept Sci. 2008, 9, 468–477.
[9] I. U. Khan, A. G. Beck-Sickinger, Anti-Cancer Agents Med. Chem. 2008, 28,
186–199.
Received: December 22, 2009
Revised: February 2, 2010
[10] L. Lad, L. Luo, J.D Carson, . K. W. Wood, J. J. Hartman, R. A. Copeland, R.
Sakowicz, Biochemistry 2008, 47, 3576–3585.
Published online on March 10, 2010
574
www.chemmedchem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2010, 5, 567 – 574