Anionic activators for differential sensing with cell-penetrating peptides

Article abstract of DOI:10.1002/asia.201000700

The design, synthesis, and evaluation of small peptides with one to three negative charges and one to three hydrazides as key components of membrane-based synthetic sensing systems are reported. Their spontaneous reaction with hydrophobic aldehydes or ket

Full text of DOI:10.1002/asia.201000700

DOI: 10.1002/asia.201000700
Anionic Activators for Differential Sensing with Cell-Penetrating Peptides
Javier Montenegro and Stefan Matile*[a]
Dedicated to Professor Eiichi Nakamura on the occasion of his 60th birthday
Abstract: The design, synthesis, and
evaluation of small peptides with one
to three negative charges and one to
three hydrazides as key components of
membrane-based synthetic sensing sys-
tems are reported. Their spontaneous
reaction with hydrophobic aldehydes
or ketones gives rapid access to small
collections of amphiphilic anions.
These anionic amphiphiles can activate
polycations as anion transporters in
lipid-bilayer membranes. Odorants are
used as representative hydrophobic al-
dehydes and ketones, and cell-penetrat-
ing peptides (CPPs) as polycationic
transporters in fluorogenic vesicles.
Different activities obtained with dif-
ferent counterion activators are used to
generate multidimensional patterns
that can be recognized by principal
component and hierarchical cluster
analysis to extract unique “finger-
prints” for individual analytes (includ-
ing enantiomers, cis–trans isomers or
perfumes as illustrative analyte mix-
tures). Comparison of the peptide acti-
vators reveals that carboxylates per-
form
better
than
phosphonates.
Gemini-like activators containing two
carboxylates and two hydrophobic hy-
drazone tails are best, whereas exces-
sive charges and tails give weaker ac-
tivities. This result differs from cationic
activators of polyanionic transporters
such as DNA, which worked best with
octopus amphiphiles with one cationic
head and four hydrophobic tentacles.
Keywords: amphiphiles
·
membranes · peptides · sensors ·
supramolecular systems
Introduction
the context of synthetic membrane-based sensing systems
because, as with olfactory receptors, little discriminatory
power is needed and many analytes are detectable with the
same system.
Recently, we have secured access to a synthetic sensing
system that can function by pattern generation in lipid bilay-
ers (Figure 1).[44,45] In this approach, hydrophobic analytes
are incubated with reactive countercations. The resulting
amphiphiles then activate the polyanionic transporters to
transport cationic (and/or anionic) fluorescent probes out of
vesicles and cause a fluorescent response.
This sensing scheme has been elaborated for calf-thymus
(ct) DNA as an ideal polyanion, odorants such as jasminal-
dehyde O1 as hydrophobic analytes, and peptides containing
reactive hydrazides and positive charges such as G1H3 as
reactive countercations (Scheme 1).[44,45] The resulting hy-
drazones, for example, G1H3O1, are cationic amphiphiles
that activate ctDNA in fluorogenic vesicles.
For differential sensing, G1H1–G1H4 and A1H1–A1H4
have been prepared to produce a small collection of reactive
countercations. Incubation with analytes gives cationic am-
phiphiles that activate DNA transporters with different effi-
ciencies and thus produce the collection of fluorescent re-
sponses that is needed to generate patterns. Single analytes
are then identified by pattern recognition. Namely, routine
Olfactory sensing systems function as differential sensors in
lipid-bilayer membranes.[1] This is not surprising because
only systems that operate by pattern recognition can sense
10000 odorants with only 350 receptors.[1] Successful and
popular with most other chemosensing systems,[2–20] the de-
velopment of synthetic sensing systems that work like the
differential olfactory sensing systems in lipid bilayers has
been difficult, particularly for hydrophobic analytes that are
not recognized in the hydrophilic interior of synthetic
pores.[21–49] Contrary to membrane-based sensing systems
that use enzymes[38–42] or aptamers[43] to generate analyte-
specific signals, differential sensors[2–9] act as groups that
weakly respond to many analytes and generate analyte-spe-
cific patterns. This promiscuous behavior is also attractive in
[a] Dr. J. Montenegro, Prof. S. Matile
Department of Organic Chemistry
University of Geneva
Geneva, Switzerland
Fax : (+41)22-379-3215
Homepage: www.unige.ch/sciences/chiorg/matile/
E-mail: stefan.matile@unige.ch
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/asia.201000700.
Chem. Asian J. 2011, 6, 681 – 689
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
681

FULL PAPERS
methods such as principal component analysis (PCA) or hi-
erarchical cluster analysis (HCA) are used to extract unique
composite responses or “fingerprints.”
Among polyanions, ctDNA was ideal for several rea-
sons,[37,43,46] whereas more-basic polyanions such as polyglu-
tamate were naturally inactive.[46] Among countercations,
guanidinium cations G1H1–G1H4 were consistently more
active than ammonium cations A1H1–A1H4.[44,45] Activator
efficiencies generally increased with an increasing number
of tails, G1H4 with one guanidinium head and four hydra-
zide tails is the most efficient reactive countercation known
so far.[45]
Differential sensing with polyion–counterion complexes in
lipid-bilayer membranes has been developed with odorants
as analytes.[44] Pattern recognition was successful for all
tested analytes, including single-atom homologues, cis–trans
isomers and enantiomers. The same was true for all tested
perfumes, including Chanel N85, Calvin Kleinꢁs Euphoria, a
scent by Issey Miyake, Silver Shadow by Davidoff, and
more, thus demonstrating compatibility of the differential
sensing system with complex
Figure 1. Sensing scheme for differential sensing of hydrophobic analytes
with counterion-activated polyion transporters in fluorogenic vesicles.
For sensing with polycationic CPPs, hydrophobic analytes are incubated
with reactive hydrophilic anions. The resulting amphiphiles act as coun-
terion activators of CPP transporters, and their activity is recorded as
fluorescence recovery during the release of self-quenched anionic fluoro-
phores from the vesicle.
matrices from the supermar-
ket.[44]
The objective of this report
was to explore the possibility of
full-charge inversion[46] of poly-
ion–counterion systems in the
context of differential sensing.
This was of interest not only to
expand and unify the concept
of multifunctional polyion–
counterion complexes but also
because cell-penetrating pep-
tides (CPPs) could be used as
transporters and the best identi-
fied amphiphilic counteranions
could possibly mediate cellular
uptake of CPPs similarly
well.[36,37,46–49] Sensing of hydro-
phobic analytes with anionic
hydrazides to produce anionic
counterion activators of CPP
transporters has been used to
create a cholesterol sensor.[42]
In that approach, signal genera-
tion was achieved enzymatically
with cholesterol oxidase. To
shift attention from enzymatic
biosensors to differential sens-
ing by pattern recognition, we
herein report the synthesis of a
series of short peptide “mini-
dendrons”[50–53] equipped with
one to three anionic charges
Scheme 1. Previously reported reactive countercations for differential sensing systems with polyanionic DNA
transporters in fluorogenic vesicles.[44, 45] They are classified according to nature (G, guanidinium; A, ammoni-
um) and number (1) of charges followed by nature (H, hydrazide) and number (1–4) of reactive groups to
attach hydrophobic analytes (e.g., jasminaldehyde O1).
and one to three hydrazides
(Scheme 2).
Gemini-like[54,55]
amphiphilic counterion activa-
682
www.chemasianj.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Asian J. 2011, 6, 681 – 689

Cell-Penetrating Peptides
applications in practice. Phos-
phonates were prepared by at-
taching the protected 2-phos-
phonate acetate 1 to the corre-
sponding amines, a convenient
reaction referred to as “amido-
phosphorylation.” Carboxylates
were made analogously from
amine precursors by “amidocar-
boxylation” by using mono-
methyl oxalyl chloride 2.
For the simplest reactive
counterions, that is the mini-
malist P1H1 and C1H1 with
one hydrazide and one anion,
modules 1 and 2 were treated
with the tert-butoxycarbonyl
(Boc)-protected hydrazine 3,
and hydrazides, phosphonates,
and carboxylates were liberated
by deprotection with HCl.
Anionic dihydrazides P1H2
and C1H2 were prepared by
amidophosphorylation and ami-
docarboxylation with 1 and 2 of
the previously reported amine 4
followed by deprotection with
base and acid.[44] The same re-
actions were applied to peptide
dendron 5[44] to afford trihydra-
zides P1H3 and C1H3 as HCl
salts.
Dianionic
monohydrazide
C2H1 was prepared by hydrazi-
dylation of Z-protected lysine 6
with Boc-protected hydrazine 3.
Scheme 2. The reactive counteranions for differential sensing systems with charge-inverted polycationic CPP
transporters in the fluorogenic vesicles described herein. They are classified according to nature (C, carboxyl- Hydrogenolytic, chemoselective
ate; P, phosphonate) and number (1–3) of charges followed by nature (H, hydrazide) and number (1–3) of re-
active groups to attach hydrophobic analytes (e.g., octanal O2).
amine deprotection of 7 gave
diamine 8, which was amidocar-
boxylated with acyl chloride 2
and deprotected to give C2H1
as the HCl salt. Application of
tors such as C2H2O2 with two carboxylate (rather than
phosphonate) heads and two fragrant tails emerge as the
best activators for differential sensing with CPPs. This is in
contrast to the charge-inverted DNA systems, which worked
best with octopus amphiphiles with one guanidinium head
and as many tails as possible.[45] The discriminatory power of
differential CPP sensors is better than that of the charge-in-
verted DNA sensors.
the same procedure to amine 4 gave dipeptide 9. Attach-
ment of 6 and hydrogenation followed by amidocarboxyla-
tion of amine 10 and final deprotection afforded the gemini
precursor C2H2. The dianionic trihydrazide C2H3 was ob-
tained in the same way from amine 5 by coupling with
lysine 6, hydrogenation of tripeptide 11, amidocarboxylation
of amine 12, and final deprotection.
A slightly varied approach was needed to synthesize the
trianionic C3H1 with a single hydrazide. Z-protected lysines
6 and 13 were coupled to give dipeptide 14. Esterolysis with
base liberated the carboxylic acid 15, which was treated with
hydrazine 3 to yield the Boc-protected hydrazide 16. Hydro-
genolysis gave triamine 17 for amidocarboxylation with acti-
vated oxalate 2 and deprotection with base and acid afford-
ed the target molecule C3H1.
Results and Discussion
Synthesis: The synthesis of reactive counteranions was as
straightforward as that of their cationic counterparts
(Scheme 3).[44,45] This simplicity is important for potential
Chem. Asian J. 2011, 6, 681 – 689
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemasianj.org
683

S. Matile et al.
FULL PAPERS
Routine spectroscopic and
analytical characterization of all
new compounds was in agree-
ment with sample structure,
purity, and homogeneity. Chiral
compounds
were
optically
active, no indication of epimeri-
zation was visible in their
1H NMR spectra, results from
reverse-phase (RP)-HPLC of
the protected target molecules
gave single peaks.[56]
Characterization: The ability
of the reactive carboxylates
C1H1–C1H3 and phosphonates
P1H1–P1H3 to capture hydro-
phobic analytes on the one
hand and activate CPP trans-
porters on the other hand was
determined under routine con-
ditions[35–48] with octanal O2 as
the representative analyte[44,45]
and poly-l-arginine (pR, MW
13,300, DP 72) as the represen-
tative CPP transporter.[36,39,42,46]
Anionic amphiphiles C1H1O2-
C1H3O2
and
P1H1O2-
P1H3O2 were prepared indi-
vidually by incubation with 2
equivalents of O2 per hydrazide
in DMSO for 1 hour at 608C
(Scheme 2).[44,45]
Quantitative
hydrazone formation was con-
firmed by mass spectrometry.
The activation of pR as an
anion transporter was explored
in standard EYPC-LUVsꢀCF
(i.e., egg-yolk phosphatidylcho-
line (EYPC) large unilamellar
vesicles (LUVs) loaded with
the
anionic
fluorophore
5(6)-carboxyfluorescein (CF) at
concentrations high enough for
self-quenching
to
occur).[36,39,42,46] In this assay,
the ability of CF to leave the
vesicle is reported as fluores-
cence recovery caused by mini-
mized self-quenching in re-
sponse to local dilution. This re-
sponse is independent of the
mechanism of transport, al-
though extensive early studies
in bulk (U-tube) and lipid-bi-
layer membranes have con-
Scheme 3. a) 1. 2, DIEA, DMAP, DCM, 08C, 70%; 2. NaOH (1m), EtOH, 79%; 3. HCl (3m), 84%.
b) 1. HBTU, DIEA, DCM, RT, 92%; 2. Me3SiBr, DCM, 74%. c) 1. 2, DIEA, DMAP, DCM, 08C, 69%;
2. NaOH (1m), EtOH, 87%; 3. HCl (3m), 93%. d) 1. HBTU, DIEA, DCM, RT, 72%; 2. Me3SiBr, DCM,
58%. e) 1. 2, DIEA, DMAP, DCM, 08C, 70%; 2. NaOH (1m), EtOH, 94%; 3. HCl (3m), 81%. f) 1. HBTU,
DIEA, DCM, RT, 73%; 2. Me3SiBr, DCM, 44%. g) HBTU, DIEA, DCM, RT, 91%. h) Pd/C, H2, RT, quanti-
tative. i) 1. 2, DIEA, DMAP, DCM, 08C, 74%; 2. NaOH (1m), EtOH, 92%; 3. HCl (3m), 92%. j) HBTU,
DIEA, DCM, RT, 74%. k) Pd/C, H2, RT, quant. l) 1. 2, DIEA, DMAP, DCM, 08C, 60%; 2. NaOH (1m),
EtOH, 86%; 3. HCl (3m), 95%. m) HBTU, DIEA, DCM, RT, 74%. n) Pd/C, H2, RT, quant. o) 1. 2, DIEA,
DMAP, DCM, 08C, 62%; 2. NaOH (1m), EtOH, 89%; 3. HCl (3m), 81%. p) HBTU, DIEA, DCM, RT, 87%.
q) NaOH (1m), EtOH, 74%. r) NH2-NHBoc 3, HBTU, DIEA, DCM, RT, 57%. s) Pd/C, H2, RT, quant. t) 1. 2,
DIEA, DMAP, DCM, 08C, 50%; 2. NaOH (1m), EtOH, 70%; 3. HCl (3m), 86%. DCM=dichloromethane,
DIEA=diisopropylethylamine, DMAP=4-dimethylaminopyridine, HBTU= O-(1H-benzotriazol-1-yl)-
N,N,N’,N,’-tetramethyluronium hexafluorophosphate.
firmed
that
counteranion-
684
www.chemasianj.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Asian J. 2011, 6, 681 – 689

Cell-Penetrating Peptides
Table 1. Characteristics of reactive counteranions as activators of pR
transporters.[a]
activated pR can act as an anion carrier in intact vesi-
cles.[36,47,48]
Entry
Counterion[b]
YMAX [%][c]
EC50 [mM][d]
n[e]
Hydrazone C1H1O2 is a classical amphiphile with one
charged head and one hydrophobic tail. Addition of a con-
centrated stock solution of C1H1O2 in DMSO to a stirred
and thermostated aqueous suspension of CF vesicles caused
negligible fluorescence recovery up to a final concentration
of 90 mm (Figure 2 g, t=0–1 min). However, addition of pR
as a concentrated aqueous solution to give a final concentra-
[f]
[f]
[f]
1
2
3
4
5
6
7
8
C1H1
P1H1
C1H2
P1H2
C1H3
P1H3
C2H1
C2H2
C2H3
C3H1
–
–
–
–
–
–
[f]
[f]
[f]
66.6Æ5.2
25.0Æ2.3
4.3Æ1.6
[f]
[f]
[f]
–
–
–
–
–
–
–
–
–
[f]
[f]
[f]
[f]
[f]
[f]
82.2Æ5.0
58.4Æ3.2
47.7Æ1.8
18.5Æ2.6
3.8Æ0.4
23.5Æ2.2
1.7Æ0.5
4.0Æ1.3
1.1Æ0.8
9
10
[f]
[f]
[f]
–
–
–
[a] From dose–response curves for hydrazone activators obtained by in-
cubation with octanal O2, recorded in fluorogenic CF vesicles at constant
pR and vesicle concentration (see Figures 2 and 3 for original data).
[b] For structures, see Scheme 2. [c] Maximal activity at saturation with
counterions (data ÆSE). [d] Counterion concentration required for 50%
activity. [e] Hill coefficient. [f] Systems with YMAX <10% and/or EC50
100 mm were considered as inactive.
<
brane, particularly at high concentrations, and by the effi-
ciency of intervesicular transfer.
The effective concentration of C1H2O2 to reach 50% of
YMAX was determined by Hill analysis of the dose response
curve (Figure 3 (&). The half maximal effective concentra-
tion (EC50)=25.0Æ2.3 mm is a measure for the efficiency of
Figure 2. Changes in fractional CF emission IF during addition of
C2H1O2 (a) 3 mm, b) 10 mm, c) 20 mm, d) 30 mm, e) 100 mm, and f) 200 mm
final concentrations, t<0 min) or C1H1O2 (g) 90 mm final), pR
(1.25 mgmLÀ1 final concentration, t=1 min), and triton X-100 (0.024%
final concentration, t=5 min) to EYPC-LUVsꢀCF.
tion of 1.25 mgmLÀ1 caused no significant change in fluores-
cence either, also at high counterion concentrations (Fig-
ure 2 g, t=1–5 min). At saturation, the vesicles were lysed
for total fluorescence recovery (Figure 2 g, t>5 min).
This lack of responsiveness to pR addition demonstrated
that C1H1O2 is inactive. The same inactivity was obtained
when the carboxylate in C1H1O2 was replaced by the phos-
phonate in P1H1O2. This finding with single-tail monoan-
ions was the same as that with countercations. Simple, deter-
gent-like one-head-one-tail amphiphiles G1H1O1 and
A1H1O2 were also incapable of activating the complemen-
tary DNA transporters.[44]
Figure 3. Dose–response curves for pR activation by the divalent dianion
C2H2O2 (*), the monovalent dianion C2H1O2 (*), divalent monoan-
ion C1H2O2 (&), and trivalent dianion C2H3O2 (ꢂ); for details, com-
pare original data for amphiphile C2H1O2 in Figure 2 and the Support-
ing Information.
The introduction of
a second tail in the lipid-like
C1H2O2 completely changed the situation. Addition to CF
vesicles still caused negligible fluorescence recovery at
<100 mm (compare Figure 2, t=0–1 min). Subsequent pR
addition, however, produced a rapid increase in fluorescence
emission (as in Figure 2, t=1–5 min). The activity found for
polyion–counterion complexes pR–C1H2O2 was calibrated
between minimal fractional activity Y=0 before pR addi-
tion and maximal fractional activity Y=1 after lysis. This
fractional activity increased with increasing concentration of
C1H2O2 up to saturation at a maximal YMAX =66.6Æ5.2
(Table 1, entry 3). The YMAX of counterion activators de-
pends on experimental conditions and is mainly influenced
by competing precipitation during delivery to the mem-
the counterion activator C1H2O2 (Table 1, entry 3). Above
the general barrier of stoichiometric binding around 1–
10 mm, also this value depends strongly on the delivery of
the transporter to the membrane as well. The EC50 is thus
only very loosely related to the dissociation constant of the
active polyion-counterion complex pR-C1H2O2 in the
membrane. Hill analysis further delivered the Hill coeffi-
cient n=4.3Æ1.6 as third significant characteristics describ-
ing the pR–C1H2O2 complex (Figure 3 (&); Table 1,
entry 3).
Replacement of the carboxylate in C1H2O2 with the
phosphonate in P1H2O2 resulted in a total loss of activity
Chem. Asian J. 2011, 6, 681 – 689
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemasianj.org
685

S. Matile et al.
FULL PAPERS
(Table 1, entry 4). The same was true for the introduction of
a third tail in C1H3O2 (Table 1, entry 5). Also the trivalent
phosphonate analogue P1H3O2 showed little ability to acti-
vate pR transporters and suffered from detergent effects al-
ready at >30 mm (Table 1, entry 6). This trend was contrary
to countercation activators in which the ability to activate
the complementary DNA transporters increased with an in-
creasing number of tails up to octopus amphiphiles
G1H4O2 with one head and four tails.[45]
Based on these initial trends, studies on phosphonates and
counteranions with more than three tails were discontinued,
and attention was shifted to counteranions with more than
one carboxylate headgroup. In sharp contrast to the inactive
monoanion C1H1O2, the presence of a second carboxylate
head in dianion C2H1O2 gave excellent activities with a
very high YMAX =82.2Æ18.5 and a low EC50 =18.5Æ2.6 mm
(Figure 2 and 3 (*), Table 1, entry 7). Addition of another
tail led to “gemini”[54,55] amphiphile C2H2O2 with an extra-
ordinary EC50 =3.8Æ0.4 mm (Figure 3 (*), Table 1, entry 8).
However, as with monoanionic C1H3O2, a third tail was
one too much in the dianionic series, C2H3O2 was clearly
less active (Figure 3 (ꢂ), Table 1, entry 9).
with its characteristic “green” odor and the all-trans isomer
O5 are interesting to probe for the differentiation between
cis–trans isomers. (S)-(À)-Citronellal O6 and (R)-(+)-citro-
nellal O7, the insect-repellent lemon scent of citronella oil,
are classics in the field for enantiodiscrimination, the cycla-
men aldehyde O8 and jasminaldehyde O1 are examples of
branched and aromatic structures.
The protocol for differential sensing differs slightly from
the protocol for activator screening.[44] Increasing concentra-
tions of each odorant were incubated with each reactive
counteranion at constant concentration, and the correspond-
ing dose–response curves were recorded for pR activation in
CF vesicles. The resulting dose–response curves recorded
for varied odorant concentration are different from the
dose–response curves recorded for varied concentrations of
the hydrazone amphiphiles needed for activator screening
(Figure 3 vs Figure 4). However, as with octanal O2
The fine activity of C2H1O2 provided an additional in-
centive to synthesize counteranions with three carboxylate
heads and one hydrazone tail. However, the ability of
C3H1O2 to activate pR-transporters was insignificant
(Table 1, entry 10).
This limited counteranion screening revealed overall that,
for pR–counteranion complexes, amphiphiles with equal or
similar number of heads and tails are operational, whereas
excess of either headgroups or tails rather quickly cancels
the activity. Gemini-type dicarboxylates C2H2O2 are so far
clearly the best (Table 1, entry 8). The four functional reac-
tive counteranions C2H1, C2H2, C2H3, C1H2 were pre-
served to construct differential sensors while the rest was
not studied further.
Figure 4. Dose–response curves for pR activation by (R)-(+)-citronellal
O7 after incubation with C2H1 (&), C2H2 (*), C2H3 (*) and C1H2
(&) at constant concentration. For details, see the Supporting Informa-
tion.
For differential sensing with CPP transporters, a focused
collection of odorant analytes O1–O9 was selected. Octanal
O2, applied here as a standard to characterize reactive coun-
teranions, is a citrus odorant used in the food industry. The
longer and shorter homologues nonanal O3 and heptanal
O9, respectively, are of interest to explore homologue dis-
crimination in the alkyl series. The cucumber aldehyde O4
(Figure 3), the best activities for (R)-(+)-citronellal O7 were
found for the gemini amphiphile C2H2O7, followed by the
dianionic, single-tail amphiphile C2H1O7, the most com-
plex triple-tail dianion C2H3O7, and finally the lipid-like
C1H2O7 (Figure 4). However, with the branched and aro-
matic jasminaldehyde O1, clearly different trends were ob-
served (Figure 5). Excellent activities were found for the
lipid-like C1H2O1 followed by the constantly powerful
gemini amphiphile C2H2O1, whereas hydrophilic C2H1O7
gave poor EC50 values and triple-tail C2H3O1 gave surpris-
ingly poor YMAX values.
Differences as with O1 and O7 were observed for all
other odorants.[56] They were quantified by Hill analysis to
give EC50, YMAX and n for each analyte and each reactive
counteranion. Three variables for four reactive counteran-
ions gave a 12-dimensional (12D) pattern. The 12D patterns
obtained for O1–O9 were subjected to PCA.[8] PCA concen-
trates the most significant characteristics of n-dimensional
patterns into virtual 3D space (Figure 6). This is achieved by
calculating eigenvectors (or principal components, PC) in
the direction of maximal variances to obtain a single score
686
www.chemasianj.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Asian J. 2011, 6, 681 – 689

Cell-Penetrating Peptides
Figure 5. Dose–response curves for pR activation by jasminaldehyde O2
after incubation with C2H1 (&), C2H2 (*), C2H3 (*), and C1H2 (&)
at constant concentration. For details, see the Supporting Information.
Figure 6. PCA score plot for representative odorants. PC 1–PC 3 describe
the first three principal components, covering the corresponding percent-
age of data variance, in this case, 81% in total. Data points represent in-
dependent experiments with CPP sensors and counteranions C2H1,
C2H2, C2H3, and C1H2, usually three per analyte, to determine repro-
ducibility and experimental error (see Table S1 in the Supporting Infor-
mation).[56]
Figure 7. HCA dendrogram for representative odorants, showing the Eu-
clidian distances between average values from three experiments with
a) CPP sensors and counteranions C2H1, C2H2, C2H3, and C1H2,
b) DNA sensors and countercations G1H2, G1H3, A1H2, and A1H3,
and c) both data sets combined. Data used for (b) are from refer-
ence [44].
that can be plotted in the new PC space. Sorted in descend-
ing order, the greatest amount of data variance, here 40%,
is described by the first PC. PC 2 describes the greatest
amount of data variance in a direction orthogonal to PC 1,
here 24%. The first two PCs thus accounted for 64% of
data variance, thus indicating that expansion to a 3D score
plot was appropriate. With the inclusion PC 3, a satisfactory
81% of the data variance were accounted for. The 3D score
plot calculated for the odorants O1–O9 showed no overlap,
thereby demonstrating full discrimination of all analytes, in-
cluding cis–trans isomers O4 and O5 or enantiomers O6 and
O7 (Figure 6). This result suggested that differential CPP
sensors are as good as the charge-inverted DNA sensors.[44]
More quantitative comparison became possible with
HCA.[8] This routine multivariant analysis method converts
interpoint Euclidean distances between all samples in the
nD space into 2D dendrograms. The HCA dendrogram ob-
tained for differential CPP sensors showed clustering of cis–
trans isomers and enantiomers O4–O7 around approximate-
ly 12 Euclidian distance units (E.u.; Figure 7a). Initial clus-
tering occurred between the unrelated O4 and O6 at ap-
proximately 7 E. u. and the complementary O5 and O7 at
around 10 E. u. Initial clustering of structurally unrelated
analytes could suggest that the patterns produced by stereo-
isomers are perfectly recognized. Average analyte clustering
occurred around 20 E.u., and the best discrimination was in-
dicated for jasminaldehyde O1 at around 30 E. u. and octa-/
nonanal O2/O3 above 45 E. u.
For comparison, the previously reported[44] 12D patterns
obtained for the same odorants O1–O9 with differential
DNA sensors were subjected to HCA as well (Figure 7b).
Contrary to CPP sensors, cis–trans isomers O4 and O5 clus-
tered first at approximately 5 E.u., followed by enantiomers
O6 and O7 at around 6 E.u. Longer Euclidian distances and
second-generation stereoisomer clustering suggested that
cis–trans and enantiodiscrimination are better with CPP sen-
Chem. Asian J. 2011, 6, 681 – 689
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemasianj.org
687

S. Matile et al.
FULL PAPERS
sors than with DNA sensors. Similar improvements, which
are in part quite remarkable, could be observed for most
couples such as octanal O2 versus nonanal O3 or cyclamen
aldehyde O8 versus jasminaldehyde O1. Only heptanal O9
gave better Euclidian distances with DNA sensors.
agents can be prepared in situ and eventually release their
membrane-active and/or fluorescent tails while entering a
cell.
HCA dendrograms computed with combined data from
both CPP and DNA sensors showed roughly additive behav-
ior with a slight dominance of the more powerful CPP sen-
sors (Figure 7c). With a few exceptions, HCA results were
thus best with CPP data only (Figure 7a).
Acknowledgements
We thank D. Jeannerat, A. Pinto, and S. Grass for NMR spectroscopic
measurements, the Sciences Mass Spectrometry (SMS) platform at the
Faculty of Sciences, University of Geneva, for mass spectrometry serv-
ices, and the University of Geneva and the Swiss NSF for financial sup-
port.
Conclusions
[1] L. B. Buck, Angew. Chem. 2005, 117, 6283–6296; Angew. Chem. Int.
Ed. 2005, 44, 6128–6140.
[2] J. J. Lavigne, E. V. Anslyn, Angew. Chem. 2001, 113, 3212–3225;
Angew. Chem. Int. Ed. 2001, 40, 3118–3130.
[3] M. A. Palacios, R. Nishiyabu, M. Marquez, P. Anzenbacher, Jr., J.
Am. Chem. Soc. 2007, 129, 7538–7544.
[4] C. Zhang, K. S. Suslick, J. Am. Chem. Soc. 2005, 127, 11548–11549.
[5] H. Zhou, L. Baldini, J. Hong, A. J. Wilson, A. D. Hamilton, J. Am.
Chem. Soc. 2006, 128, 2421–2425.
[6] A. Buryak, K. Severin, J. Am. Chem. Soc. 2005, 127, 3700–3701.
[7] S. H. Shabbir, L. A. Joyce, G. M. da Cruz, V. M. Lynch, E. Sorey,
E. V. Anslyn, J. Am. Chem. Soc. 2009, 131, 13125–13131.
[8] J. C. Jurs, G. A. Bakken, H. E. McClelland, Chem. Rev. 2000, 100,
2649–2678.
[9] N. T. Greene, K. D. Shimizu, J. Am. Chem. Soc. 2005, 127, 5695–
5700.
[10] Creative Chemical Sensing Systems. Topics in Current Chemistry
(Ed.: T. Schrader) Springer, Heidelberg, 2007.
The objective of this study was to explore full charge inver-
sion in polyion–counterion transport systems in the context
of differential sensing. For this purpose, a small collection of
reactive counteranions was prepared and their ability to cap-
ture analytes, activate CPP transporters and generate ana-
lyte-specific patterns was evaluated. The result is that differ-
ential polyion–counterion sensing systems are compatible
with charge inversion. HCA dendrograms indicate that the
discriminatory power of counteranion-activated CPP sensors
is even slightly better than with the previous countercation-
activated DNA sensors. This is particularly true and impor-
tant for structurally challenging analytes such as cis–trans
isomers, which are illustrated here with cucumber aldehydes,
and enantiomers, here illustrated with citronellals.
Concerning the nature of the counteranion activators, car-
boxylates are clearly better than phosphonates. This differ-
ence has been explained previously with the better preorga-
nization of the hydrogen-bond-assisted carboxylate–guanidi-
nium ion pairing.[36,47,48] Gemini amphiphiles[54,55] with two
carboxylate heads and two hydrophobic tails are best, while
related cylindrical amphiphiles are almost as good. Cone-
shaped amphiphiles, normal or inverted, with a large excess
of either tails or heads, are inactive. This result differs from
countercation-activated DNA sensors. In this case, activities
increased with an increasing number of tails up to amphi-
philes with one guanidinium head and four hydrophobic
tails.[45] This significant difference originates either from dif-
ferent hydrophobicity and counterion affinity of the ex-
plored polyion transporters or from different mechanisms of
transport (and cellular uptake) of charge-inverted polyion–
counterion transporters (i.e., endovesiculation with polyan-
ion–countercation complexes[57,58] and micellar pores with
polycation–counteranion complexes).[21,22,46]
[11] C. Guarise, L. Pasquato, V. De Filippis, P. Scrimin, Proc. Natl. Acad.
Sci. USA 2006, 103, 3978–3982.
[12] A. Wada, S. Tamaru, M. Ikeda, I. Hamachi, J. Am. Chem. Soc. 2009,
131, 5321–5330.
[13] T. D. James, Top. Curr. Chem. 2007, 277, 107–154.
[14] R. Jelinek, S. Kolusheva, Top. Curr. Chem. 2007, 277, 155–180.
[15] L. Vial, P. Dumy, New J. Chem. 2009, 33, 939–946.
[16] M. Florea, W. M. Nau, Org. Biomol. Chem. 2010, 8, 1033–1039.
[17] J.-L. Reymond, V. S. Fluxa, N. Maillard, Chem. Commun. 2009, 34–
46.
[18] G. Das, S. Matile, Chem. Eur. J. 2006, 12, 2936–2944.
[19] S. Matile, H. Tanaka, S. Litvinchuk, Top. Curr. Chem. 2007, 277,
219–250.
[20] N. Sakai, J. Mareda, S. Matile, Acc. Chem. Res. 2008, 41, 1354–1365.
[21] S. Matile, A. Som, N. Sordꢃ, Tetrahedron 2004, 60, 6405–6435.
[22] A. L. Sisson, M. R. Shah, S. Bhosale, S. Matile, Chem. Soc. Rev.
2006, 35, 1269–1286.
[23] A. P. Davis, D. N. Sheppard, B. D. Smith, Chem. Soc. Rev. 2007, 36,
348–357.
[24] G. W. Gokel, N. Barkey, New J. Chem. 2009, 33, 947–963.
[25] J. T. Davis, P. A. Gale, O. A. Okunola, P. Prados, J. C. Iglesias-San-
chez, T. Torroba, R. Quesada, Nat. Chem. 2009, 1, 138–144.
[26] R. E. Dawson, A. Hennig, D. P. Weimann, D. Emery, V. Ravikumar,
J. Montenegro, T. Takeuchi, S. Gabutti, M. Mayor, J. Mareda, C. A.
Schalley, S. Matile, Nat. Chem. 2010, 2, 533–538.
[27] J. K. W. Chui, T. M. Fyles, Chem. Commun. 2010, 46, 4169–4171.
[28] C. P. Wilson, S. J. Webb, Chem. Commun. 2008, 4007–4009.
[29] A. Satake, M. Yamamura, M. Oda, Y. Kobuke, J. Am. Chem. Soc.
2008, 130, 6314–6315.
Operational charge inversion in differential polyion–coun-
terion sensors is significant for several reasons. The result
strengthens the importance of the unified concept of multi-
functional polyion–counterion systems in chemistry and
biology.[46] This includes implications and possible applica-
[30] X. Li, B. Shen, X. Q. Yao, D. Yang, J. Am. Chem. Soc. 2009, 131,
13676–13680.
[31] M. Jung, H. Kim, K. Baek, K. Kim, Angew. Chem. 2008, 120, 5839–
5841; Angew. Chem. Int. Ed. 2008, 47, 5755–5757.
[32] P. L. Boudreault, N. Voyer, Org. Biomol. Chem. 2007, 5, 1459–1465.
[33] L. Fischer, G. Guichard, Org. Biomol. Chem. 2010, 8, 3101–3117.
tions
toward
counterion-mediated
cellular
uptake,[46,49,52,53,55,59,60] counterion-mediated voltage gating of
ion channels,[46] sensing, and so on. Dynamic counterion acti-
vators are particularly attractive with regard to cellular
uptake because large numbers of non-viral transfection
688
www.chemasianj.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Asian J. 2011, 6, 681 – 689

Cell-Penetrating Peptides
[34] R. J. Brea, C. Reiriz, J. R. Granja, Chem. Soc. Rev. 2010, 39, 1448–
1456.
[35] N. Sakai, B. Baumeister, S. Matile, ChemBioChem 2000, 1, 123–125.
[36] N. Sakai, S. Matile, J. Am. Chem. Soc. 2003, 125, 14348–14356.
[37] T. Takeuchi, V. Bagnacani, F. Sansone, S. Matile, ChemBioChem
2009, 10, 2793–2799.
[38] G. Das, P. Talukdar, S. Matile, Science 2002, 298, 1600–1602.
[39] T. Miyatake, M. Nishihara, S. Matile, J. Am. Chem. Soc. 2006, 128,
12420–12421.
[49] T. Takeuchi, M. Kosuge, A. Tadokoro, Y. Sugiura, M. Nishi, M.
Kawata, N. Sakai, S. Matile, S. Futaki, ACS Chem. Biol. 2006, 1,
299–303.
[50] K. Sadler, J. P. Tam, Rev. Mol. Biotechnol. 2002, 90, 195–229.
[51] T. Darbre, J.-L. Reymond, Acc. Chem. Res. 2006, 39, 925–934.
[52] D. A. Tomalia, A. M. Naylor, W. A. Goddard, Angew. Chem. 1990,
102, 119–157; Angew. Chem. Int. Ed. Engl. 1990, 29, 138–175.
[53] M. Liu, J. M. J. Frechet, Pharm. Sci. Technol. Today 1999, 2, 393–
401.
[40] S. Litvinchuk, N. Sordꢃ, S. Matile, J. Am. Chem. Soc. 2005, 127,
9316–9317.
[54] F. M. Menger, J. S. Keiper, Angew. Chem. 2000, 112, 1980–1996;
Angew. Chem. Int. Ed. 2000, 39, 1906–1920.
[41] S. Litvinchuk, H. Tanaka, T. Miyatake, D. Pasini, T. Tanaka, G.
Bollot, J. Mareda, S. Matile, Nat. Mater. 2007, 6, 576–580.
[42] S. M. Butterfield, T. Miyatake, S. Matile, Angew. Chem. 2009, 121,
331–334; Angew. Chem. Int. Ed. 2009, 48, 325–328.
[43] T. Takeuchi, S. Matile, J. Am. Chem. Soc. 2009, 131, 18048–18049.
[44] T. Takeuchi, J. Montenegro, A. Hennig, S. Matile, Chem. Sci., DOI:
10.1039/c0sc00386g.
[45] J. Montenegro, P. Bonvin, T. Takeuchi, S. Matile, Chem. Eur. J. 2010,
16, 14159–14166.
[46] T. Takeuchi, N. Sakai, S. Matile, Faraday Discuss. 2009, 143, 187–
203.
[55] A. J. Kirby, P. Camilleri, J. B. F. N. Engberts, M. C. Feiters, R. J. M.
Nolte, O. Sçderman, M. Bergsma, P. C. Bell, M. L. Fielden, C. L.
Garcꢄa Rodrꢄguez, P. Guꢃdat, A. Kremer, C. McGregor, C. Perrin,
G. Ronsin, M. C. P. van Eijk, Angew. Chem. 2003, 115, 1486–1496;
Angew. Chem. Int. Ed. 2003, 42, 1448–1457.
[56] See Supplementary Information.
[57] M. M. Tedesco, S. Matile, Bioorg. Med. Chem. 1999, 7, 1373–1379.
[58] H. Matsuo, J. Chevallier, F. Vilbois, R. Sadoul, J. Faurꢃ, S. Matile, N.
Sartori Blanc, J. Dubochet, J. Gruenberg, Science 2004, 303, 531–
534.
[59] S. Bhattacharya, A. Bajaj, Chem. Commun. 2009, 4632–4656.
[60] H. Isobe, W. Nakanishi, N. Tomita, S. Jinno, H. Okayama, E. Naka-
mura, Chem. Asian J. 2006, 1, 167–175.
[47] M. Nishihara, F. Perret, T. Takeuchi, S. Futaki, A. N. Lazar, A. W.
Coleman, N. Sakai, S. Matile, Org. Biomol. Chem. 2005, 3, 1659–
1669.
[48] N. Sakai, T. Takeuchi, S. Futaki, S. Matile, ChemBioChem 2005, 6,
114–122.
Received: September 27, 2010
Published online: December 22, 2010
Chem. Asian J. 2011, 6, 681 – 689
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemasianj.org
689