Identification of short hydrophobic CPPs for cytosolic peptide delivery by rational design

Abstract

Cell-penetrating peptides (CPPs) enhance the cellular uptake of membrane-impermeable molecules. Most CPPs are highly cationic, potentially increasing the risk of toxic side effects and leading to accumulation in organs such as the liver. As a consequence, there is an unmet need for less cationic CPPs. However, design principles for effective CPPs are still missing. Here, we demonstrate a design principle based on a classification of peptides according to accumulated side-chain polarity and hydrophobicity. We show that in comparison to randomly selected peptides, CPPs cover a distinct parameter space. We designed peptides of only 6 to 9 amino acids with a maximum of three positive charges covering this property space. All peptides were tested for cellular uptake and subcellular distribution. Following an initial round of screening we enriched the collection with short and hydrophobic peptides and introduced D-amino acid substitutions and lactam bridges which increased cell-uptake, in particular for long-term incubation. Using a GFP complementation assay, for the most active peptides we demonstrate cytosolic delivery of a biologically active cargo peptide.

Full text

Identification of Short Hydrophobic Cell-Penetrating Peptides for
Cytosolic Peptide Delivery by Rational Design
Samuel Schmidt,
§
Merel J. W. Adjobo-Hermans,
§
Robin Kohze,
§
Thilo Enderle,
†
Roland Brock,
§
and Francesca Milletti*
,‡
§
Department of Biochemistry (286), Radboud Institute for Molecular Life Sciences, Radboud University Medical Center, Geert
Grooteplein 28, 6525 GA Nijmegen, The Netherlands
†
Roche Pharmaceutical Research and Early Development, Roche Innovation Center Basel, Grenzacherstrasse 124, 4070 Basel,
Switzerland
‡
Roche Pharmaceutical Research and Early Development, Roche Innovation Center New York, 430 East 29th Street, 10016 New
York, New York, United States
*
SSupporting Information
ABSTRACT: Cell-penetrating peptides (CPPs) enhance the
cellular uptake of membrane-impermeable molecules. Most
CPPs are highly cationic, potentially increasing the risk of toxic
side effects and leading to accumulation in organs such as the
liver. As a consequence, there is an unmet need for less cationic
CPPs. However, design principles for effective CPPs are still
missing. Here, we demonstrate a design principle based on a
classification of peptides according to accumulated side-chain
polarity and hydrophobicity. We show that in comparison to
randomly selected peptides, CPPs cover a distinct parameter
space. We designed peptides of only six to nine amino acids with a maximum of three positive charges covering this property
space. All peptides were tested for cellular uptake and subcellular distribution. Following an initial round of screening we
enriched the collection with short and hydrophobic peptides and introduced D-amino acid substitutions and lactam bridges which
increased cell uptake, in particular for long-term incubation. Using a GFP complementation assay, for the most active peptides we
demonstrate cytosolic delivery of a biologically active cargo peptide.
■INTRODUCTION
Cell-penetrating peptides (CPPs) hold great promise as vectors
to yield cellular entry of charged (macromolecular) molecules
that otherwise do not cross the plasma membrane. In most
cases conjugation has been achieved through covalent coupling,
but for charged oligonucleotides, noncovalent polyplexes are
used preferentially. CPPs are about 5 to 30 amino acids long,
and can be subdivided into different classes based on the
properties of the amino acid side chains.
1
Even though most
CPPs are positively charged, and arginines result in better
uptake than lysines, noncharged residues are also crucial for
peptide uptake. In addition, the specific sequence can also play
a role.
2
For penetratin, activity depends on the presence of
tryptophan residues that are believed to promote interactions
with lipid bilayers.
3,4
For arginine-rich CPPs, one long stretch
of arginine residues is more effective than two spatially
separated domains of the same total length.
5
The prototypic arginine-rich CPPs nonaarginine and Tat,
and the amphipathic penetratin peptide are among the most
effective and widely used CPPs. However, their large number of
positive charges poses toxicity concerns, particularly for CPP
development as therapeutics.
6,7
Highly hydrophobic peptides
such as the so-called membrane translocating sequence tend to
have poor solubility and to aggregate in aqueous solution.
8,9
Hallbrink et al. have shown that peptide bulk properties
based on amino acid descriptors can be used to predict CPPs.
10
This analysis employed a five-dimensional parameter set and
focused on a scanning of transmembrane proteins for domains
with CPP activity.
11
Here, we set out to identify short peptides with minimal total
charge. This search was based on a parametrization of peptides
according to accumulated side-chain properties with respect to
only two dimensions, polarity and hydrophobicity. We focused
on the application of peptides in the delivery of small molecules
and peptides. For delivery of negatively charged oligonucleo-
tides, the capacity to form noncovalent polyplexes constitutes
an additional structure−activity relationship, and therefore
these peptides should be considered as a class of their own.
12
As a model cargo we employed a 16-amino-acids peptide
corresponding to the 11th strand of the green fluorescent
protein β-barrel. In cells expressing the GFP1−10 fragment,
Received: September 17, 2016
Revised: November 22, 2016
Published: November 28, 2016
Article
pubs.acs.org/bc
© XXXX American Chemical Society ADOI: 10.1021/acs.bioconjchem.6b00535
Bioconjugate Chem. XXXX, XXX, XXX−XXX

cytosolic delivery of this peptide leads to formation of
fluorescent GFP following binding of the cargo to GFP1−10.
This reporter assay works with micromolar sensitivity.
13,14
By testing two consecutive collections of peptides in three
different cells lines and for various times of incubation we
identified CPPs with only 6 to 8 amino acids in length and only
two positively charged residues that yield delivery and GFP
complementation. In particular for long-term incubation, the
incorporation of individual D-amino-acids and a lactam bridge
increased uptake efficiency.
■RESULTS
Physicochemical Properties of Cell-Penetrating Pep-
tides. The broad sequence diversity of cell-penetrating
peptides has thwarted efforts to understand cellular uptake
based on structure−activity relationships. However, CPPs differ
from most peptides based on their amino acid composition.
CPPs have a disproportionally larger number of positively
charged residuesespecially argininesand hydrophobic
residuessuch as leucines. To better understand how CPPs
differ from other peptides we used a data set of 109 CPPs
(average length, 18 amino acids; average net charge, +4.8) and
of 1000 15-mer peptides obtained by randomly selecting
peptide sequences from a collection of approximately 20 000
human protein sequences. For these peptides we calculated two
physical−chemical propertiespolarity (PP1) and hydro-
phobicity (PP2)which are scaled principal component scores
that summarize a broad set of descriptors calculated based on
the interaction of each amino acid residue with several chemical
groups (or “probes”), such as charged ions, methyl, hydroxyl
groups, and so forth, as described by Cruciani et al.
15
The
probes reflect a broad range of chemical groups that amino acid
residues may encounter in a biological system. PP1 and PP2
were defined for individual amino acid residues (Figure 1A, SI
Table 1). To calculate PP1 and PP2 of each peptide we
calculated the average PP1 and PP2 of all the amino acids in the
corresponding peptide sequence.
As shown in Figure 1B, randomly generated peptides and
CPPs occupy distinct regions of the PP1/PP2 space. CPPs
disproportionally occupy the area H colored in green (43% of
CPPs vs 1% of random peptides occupy this space), suggesting
that peptide sequences with an amino acid composition that
falls into this area have a significantly increased likelihood of
being cell-penetrating (Figure 1C). Using this classification
scheme, R9 possesses high hydrophobicity due to the long
carbon chain but also high polarity due to the positive charge.
As a further comparison, the well-established penetratin CPP is
relatively hydrophobic with intermediate polarity.
We designed a collection of cell-penetrating peptides starting
from peptide sequences of varying length (5−15 amino acids)
that fall into the region H of the PP1-PP2 space (Table 1,
Figure 2A). These peptide sequences were extracted from
proteins included in the UNIPROT database. MIILII (L1) was
derived from MILLII, which corresponds to one of the
transmembrane motifs of Aquaporin 9 (aa 170−175);
FLLIRRVL (L2) was derived from YLLIRRVL, which
corresponds to the transmembrane region of the Antigen
peptide transporter 2 protein (aa 376−383); PWPRVPWRW
(L3) corresponds to a fragment of the signal peptide of Emilin-
2 (aa 7−15)signal peptides direct nascent proteins into their
correct subcellular localization and due to their hydrophobic
nature they have been found as a source of CPPs; similarly,
FRWLFRLLFR (L5) is a fragment of a signal peptide of the
Cronobacter Bifunctional Protein Aas (aa 6−15); YLKFIPLK-
DAIWKIK (C2) was derived by cyclization of YLKFIPLKRAI-
WLIK, which was selected from a fragment of Saccharomyces
Figure 1. (A) Principal properties of amino acid side chains. The green area was defined based on the largest enrichment of CPPs versus random
peptides (43% vs 1%); the yellow and red areas correspond to medium and low prevalence of CPPs, respectively. (B) Coverage of the property space
by cell-penetrating versus random peptides and (C) fraction of peptides belonging to each class.
Table 1. Cell-Penetrating Peptides for the First Round of
Testing
a
abbreviation sequence type
L1 Fluo-MIILIIGSTSRDHMVLHEYVNAAGIT-NH2
b
linear
L2 Fluo-FLLIRRVLGSTSRDHMVLHEYVNAAGIT-
NH2
linear
L3 Fluo-
PWPRVPWRWGSTSRDHMVLHEYVNAAGIT-
NH2
linear
L4 Fluo-LKRAIWLIKGSTSRDHMVLHEYVNAAGIT-
NH2
linear
L5 Fluo-
FRWLFRLLFRGSTSRDHMVLHEYVNAAGIT-
NH2
linear
C1 Fluo-FIDLKRKIWLIK-NH2
c
cyclic
C2 Fluo-YLKFIPLKDAIWKIK-NH2cyclic
R9 Fluo-RRRRRRRRRGSTSRDH(Nle)
VLHEYVNAAGIT-Ado-NH2
d
linear
a
The CPP domains are printed in bold face. The short linear CPPs are
conjugated to the 11th strand of the GFP protein via a GSTS spacer.
b
Fluo refers to an N-terminal carboxyfluorescein, NH2to a C-terminal
amidation.
c
The side chains of the underlined residues are linked by a
lactam bridge.
d
Ado: 8-amino-3,6-dioxaoctanoic acid.
Bioconjugate Chemistry Article
DOI: 10.1021/acs.bioconjchem.6b00535
Bioconjugate Chem. XXXX, XXX, XXX−XXX
B

cerevisiae mediator of RNA polymerase II transcription subunit
12 (aa 161−176); LKRAIWLIK (L4) is a derivative of
YLKFIPLKRAIWLIK; last, FIDLKRKIWLIK (C1) was derived
by cyclization of L4.
16
These peptides were selected to explore
how extreme ranges of properties might affect cellular uptake.
For example, we incorporated a very short CPP (L1), and
analogues of linear and cyclic CPPs (L4, C1, C2).
In order to validate their capacity as vectors for cellular
peptide import, the linear peptides were C-terminally elongated
with a 4-amino-acids GSTS-spacer followed by a 16-amino-
acids peptide corresponding to the 11th strand of the enhanced
GFP protein. This 16-mer peptide has a net charge of −1 and
little hydrophobicity, and therefore it should not promote
uptake. Elongation with this moiety shifts the peptides into the
less hydrophobic and more polar region of parameter space out
of the region populated by CPPs (SI Figure 1)
Cytosolic delivery of this peptide into cells transfected with a
GFP fragment comprising the first ten strands of the GFP beta-
barrel can be detected by a robust and sensitive reporter assay
basedonGFPcomplementationandformationofthe
fluorescent protein.
13,14
For reasons of complexity during
synthesis, the lactam-bridged cyclic peptides were tested
without C-terminal elongation.
Cellular Uptake and Distribution. Except for one peptide
(L5) all peptides were readily dissolvable in DMSO. The
cellular uptake of the fluorescein-labeled peptides was
investigated by flow cytometry and the subcellular distribution
by confocal microscopy using HeLa cells (Figures 3 and 4).
HeLa cells are a widely used cell model and therefore highly
suitable for cross-referencing of results. Uptake was assessed in
the absence and presence of serum. Serum has been recognized
as an important factor in the uptake of CPPs and delivery
vectors in general.
17
On one hand, it can sequester peptide, and
on the other hand, proteases present in serum can degrade
peptide. Both mechanisms reduce the effective peptide
concentration. We had observed that for nonaarginine for
short-term incubations of about 30 min the presence of serum
lowers the effective concentration by one-half. In the absence of
serum, the apolar and highly hydrophobic hexapeptide MIILII
showed the most effective uptake. However, it reached only
about 13% of the uptake measured for R9 (Figure 3A). Uptake
was detected also for the two lactam-bridged peptides, but to a
lower degree. Interestingly, peptides L3 and L4 showed even
less uptake, even though with the presence of arginine and
tryptophan residues they conformed more strongly with
structural characteristics of CPPs.
The sensitivity to serum differed greatly. Uptake of the linear
hexapeptide was nearly eliminated completely in the presence
of serum, while for the cyclopeptides serum had no impact at
all. In the absence of serum, an extension of incubation time to
2 h had little impact on relative uptake efficiencies (Figure 3B).
To validate that uptake was a characteristic of peptides falling
into the green area of parameter space we also tested uptake of
a collection of peptides from an earlier project addressing
intracellular peptide stability. We also included the GFP11
motif to demonstrate that this peptide alone does not have the
capacity to enter cells. These peptides corresponded to known
interaction motifs with signal transduction motifs and intra-
cellular stability required introduction by electroporation (SI
Figure 2. Properties of the designed CPPs of the first collection (A) and the second collection (B). Properties were calculated based the bold face
sequences in Tables 1 and 2. Residues involved in lactam bridge formation were replaced by glutamine and asparagine, norleucine by leucine.
Figure 3. Normalized intracellular fluorescence intensities of analyzed
carboxyfluorescein-labeled peptides. HeLa cells were incubated for (A)
30 min with 5 μM in the presence (open bars) or 2 μM CPP in the
absence (closed bars) of fetal calf serum and for (B) 2 h with 2 μM
CPP in medium without serum and analyzed by flow cytometry for
intracellular fluorescence. Before measuring, trypan blue (0.4% end
concentration) was administered to quench cell surface-bound
fluorescence and restrict the recorded signal to intracellular
fluorescence.
Bioconjugate Chemistry Article
DOI: 10.1021/acs.bioconjchem.6b00535
Bioconjugate Chem. XXXX, XXX, XXX−XXX
C

Table 2).
18
Except for one very polar peptide, these peptides
mapped into the non-CPP area of the parameter space (SI
Figure 2). At the concentration of 20 μM GFP11 had a cell
associated fluorescence of only 3.7% in comparison to R9. For
the collection of control peptides, the maximum cell-associated
fluorescence was 3% of the one of R9 (SI Figure 3).
To assess the subcellular distribution, confocal microscopy
was performed before washing of cells and after washing in the
presence of trypan blue. In this way we aimed to first capture
potential accumulation and aggregation of peptides at the
plasma membrane and second to restrict detection to only
intracellular fluorescence. Even for the very hydrophic
hexapeptide there was no indication of membrane enrichment.
All peptides showed primarily punctate intracellular fluores-
cence indicative of endocytic uptake (Figure 4).
Testing of Conformationally Constrained Hydro-
phobic Peptides. Considering the activity of the hydrophobic
and conformationally constrained peptides, in a second round
we set out to explore combinations of both structural
characteristics. Lactam bridges were incorporated into highly
hydrophobic, nonpolar sequences. In addition, we incorporated
single D-amino-acid substitutions and nonproteinogenic amino
acids as a further means to prevent proteolysis (Table 2). A
peptide sequence (L1−6) which was reported as capable of
spontaneous cell membrane permeability was also included.
19
In comparison to the first collection, this second one was more
biased toward the nonpolar parameter space (Figure 2B, SI
Figure 1). All peptides were C-terminally extended with an
optimized GFP11 domain to detect cytosolic delivery using the
GFP complementation assay.
13
This peptide differs from the
original one by an isosteric methionine to norleucine
substitution and a C-terminal extension with a short poly-
ethylene glycol building block.
With the exception of one purely hydrophobic D-peptide
(L1−8) which could not be dissolved, peptides were tested for
uptake and intracellular distribution. For uptake in HeLa cells,
one lactam-bridged, fully hydrophobic peptide (L1−7) reached
about 40% of the uptake of R9 and two more about 15%
demonstrating an increase in overall activity (Figure 5).
This time we extended our analyses to two more cell lines.
Next to HeLa cells we also included HEK and Jurkat cells. For
arginine-rich CPPs HEK cells had shown a concentration-
independent homogeneous cytoplasmic distribution of fluo-
rescence indicating a direct translocation of the plasma
Figure 4. Live cell confocal microscopy for the determination of uptake and intracellular localization. HeLa cells were incubated with 2 μM CPP in
the absence of fetal calf serum and analyzed by confocal microscopy regarding uptake efficiency and localization after 2 h. After washing, trypan blue
was administered to quench extracellular and cell surface-bound fluorescence. Bar, 20 μm.
Table 2. Peptides to Test the Role of Conformationally
Constrained Hydrophobic Motifs for Cell-Penetrating
Capacity
abbreviations sequences
L1−1 Fluo-KIIIIDGSTSRDH(Nle)VLHEYVNAAGIT-Ado-NH2
a
L1−2 Fluo-KNleIILIIDGSTSRDH(Nle)VLHEYVNAAGIT-Ado-
NH2
L1−3 Fluo-NleIILIIGSTSRDH(Nle)VLHEYVNAAGIT-Ado-NH2
L1−4 Fluo-mIiLIIGSTSRDH(Nle)VLHEYVNAAGIT-Ado-NH2
L1−5 Fluo-MIILIIMGVADLIKKFESISKEE-NH2
L1−6 Fluo-PLILLRLLRGSTSRDH(Nle)VLHEYVNAAGIT-Ado-
NH2
L1−7 Fluo-FIDIIIKILLIGSTSRDH(Nle)VLHEYVNAAGIT-Ado-
NH2
L1−8 Fluo-fiiliiGSTSRDH(Nle)VLHEYVNAAGIT-Ado-NH2
R9 Fluo-RRRRRRRRRGSTSRDH(Nle)VLHEYVNAAGIT-Ado-
NH2
a
Fluo refers to an N-terminal carboxyfluorescein, NH2to a C-terminal
amidation, Ado to 8-amino-3,6-dioxo-octanoic acid. The side chains of
the underlined residues are linked by a lactam bridge. Peptide L1−5
was extended by a random sequence instead of GFP11.
Figure 5. Relative uptake efficiencies of the second peptide collection.
Uptake was normalized to uptake of the Fluo-R9 conjugate. HeLa cells
were incubated with the indicated peptides at a concentration of 2 μM
for 2 h in serum-free medium. Error bars represent standard errors of
the mean (SEM) of three independent experiments.
Bioconjugate Chemistry Article
DOI: 10.1021/acs.bioconjchem.6b00535
Bioconjugate Chem. XXXX, XXX, XXX−XXX
D

membrane as a mechanism of uptake.
13
Jurkat cells were
included as a suspension cell line.
In this case the most effective peptide in all three cell lines
was a linear hydrophobic peptide with two arginine residues at
a three residues spacing (L1−6). After 24 h long-term
incubation for this peptide more fluorescence was retained
inside the cell than for nonaarginine. The higher activity of L1−
7 was restricted to HeLa cells at low concentrations (Figure 6).
Since cells were coincubated with propidium iodide as a marker
for compromised membrane integrity the flow cytometry
measurements also enabled an assessment of toxicity. At a
concentration of 20 μM and 2 h incubation in the absence of
serum membrane integrity was maintained to larger than 90%
(SI Figure 4).
The intracellular distribution revealed a striking difference
between HeLa and HEK cells. While in HeLa cells fluorescence
was heterogenously distributed throughout the cells, indicative
of sequestration in endolysosomal compartments, in HEK cells
a major part of the fluorescence was distributed homogenously
throughout the cytoplasm and nucleus. These differences that
already became apparent after 2 h (SI Figure 5) were even
more prominent after 24 h (Figure 7). After 24 h the signal for
L1−6 was more intense than the one for R9.
Cytoplasmic Delivery of a Functional Peptide. Finally,
we were interested to learn whether the cytoplasmic
fluorescence reflected the presence of a functional peptide.
For this purpose, we employed the detection of GFP
complementation. We restricted ourselves to the CPPs L1−2
and L1−6 which had shown high uptake efficiency in
combination with cytosolic staining in HEK293 cells (Table 3).
L1−6 yielded a clearly detectable GFP complementation and
also L1−2 yielded some GFP fluorescence (Figure 8). Still, the
R9-conjugated peptide exceeded both. Overall, the degree of
complementation fully reproduced the relative uptake
efficiencies observed for the fluorescein-labeled peptides
demonstrating that the cell-associated fluorescence directly
correlated to the amount of intact peptide present in the
cytosol.
■DISCUSSION
Here, we predicted and validated CPPs based on average side
chain polarity and hydrophobicity as descriptors. First we
showed that known CPPs preferably populate a region that is
distinct from the one populated by the majority of randomly
generated peptides in this two-dimensional parameter space.
For a collection of peptides that comprised short hydro-
phobic sequences as well as cationic amphipathic peptides, we
surprisingly identified a purely hydrophobic peptide of only six
amino acids as the most active peptide. We then generated a
collection of analogs from this initial sequence which yielded a
further increase in activity by up to a factor of 3. Next to
nonaarginine we included a further previously published, but so
far little explored sequence (L1−6) for further referencing. This
peptide was identified by screening peptides from a highly
biased random library, in which either arginine, lysine, or a
hydrophobic residue was permitted at the various positions.
The assay was designed in a way that only water-soluble
peptides with the ability to enter liposomes by passive diffusion
and that did not show membrane-disruptive behavior were
selected. To determine the degree of bias of this collection, we
Figure 6. Uptake of conformationally constrained and hydrophobic
CPPs in HEK, HeLa, and Jurkat cells. Uptake was assessed (A) after a
2 h incubation in the absence of serum and (B) after a 24 h incubation
in the presence of serum. The values are averages of two independent
experiments. Green reflects high, white low uptake.
Figure 7. Subcellular distribution of CPPs. HeLa cells and HEK cells were incubated with the indicated peptides at a concentration of 20 μM in the
presence of serum for 24 h. Following washing, cells were imaged in the presence of trypan blue to quench cell-associated fluorescence. Fluorescence
from trypan blue is displayed as a red signal outside the cells. Bars, 20 μm.
Table 3. Peptides to Test for GFP Complementation
abbreviation sequences
L1−2 Ac Ac-K(Nle)IILIIDGSTSRDH(Nle)VLHEYVNAAGIT-Ado-
NH2
a
L1−6 Ac Ac-PLILLRLLRGSTSRDH(Nle)VLHEYVNAAGIT-Ado-NH2
R9 Ac-RRRRRRRRRGSTSRDH(Nle)VLHEYVNAAGIT-Ado-
NH2
a
Ac refers to an N-terminal acetylation, NH2to a C-terminal
amidation, Nle to norleucine, and Ado to 8-amino-3,6-dioxo-octanoic
acid. The side chains of the underlined residues are linked by a lactam
bridge, Nle refers to a norleucine residue.
Bioconjugate Chemistry Article
DOI: 10.1021/acs.bioconjchem.6b00535
Bioconjugate Chem. XXXX, XXX, XXX−XXX
E

plotted all 10 368 peptides into the parameter space. All
peptides fell into the CPP-enriched region (SI Figure 6).
Uptake efficiencies were determined for peptides conjugated
to a 16-amino-acids cargo peptide which strongly affected the
bulk properties of the peptides (SI Figure 1). In most cases,
CPPs are tested with a fluorophore as cargo only, which
severely limits the relevance of the findings in contrast to our
approach.
Cyclization by lactam bridge formation and incorporation of
D-amino-acids benefited uptake in only some cases. Both
modifications should increase peptide stability. Nevertheless, in
our case, linear peptides were the most active ones, in contrast
to previous reports on beneficial aspects of cyclization.
5,20
For the second peptide collection, uptake was measured for
two adherent cell lines (HEK293T, HeLa) and one suspension
cell line (Jurkat). Two peptides, L1−6 (PLILLRLLR) and R9,
showed activity in all cell lines; the cyclic peptide L1−2 (CPP
K(Nle)IILIID) had activity in HEK293T and Jurkat cells while
L1−4 (mIiLII) only showed activity in Jurkat cells. Cell line
preferences of CPPs have been reported before.
21
However, the
molecular basis of these differences is not clear.
After a 24 h incubation L1−6 (PLILLRLLR) showed even a
higher cell-associated fluorescence than R9. Earlier we had
observed that for nonaarginine only little cell-associated
fluorescence remained after long-term incubation, likely due
to proteolytic degradation and cellular release of fluorescein
labeled fragments.
5
In HEK293T and Jurkat cells uptake of the cyclic peptide
L1−2(
KNleIILIID) was higher than that of its linear analog
L1−3 (NleIILII). However, there was no gain in relative uptake
efficiency over time, indicating that most likely this higher
uptake was not due to an increased stability but rather due to
structural characteristics. A difference that may well be
attributed to stability is the higher retention of L1−4 (mIiLII)
which incorporates two D-amino-acids in comparison to L1−5
(MIILII) in HEK293T and Jurkat cells.
Endosomal sequestration is a major concern in the
application of CPPs. In HeLa cells, staining was punctate in
all cases, independent of incubation time, in line with uptake by
endocytosis. Also, for L1−6 direct membrane permeation, as
published before, was only observed for HEK but not for HeLa
cells demonstrating that this route of import is a function of cell
type and not a general characteristic.
Next to L1−6, in HEK293T cells also L1−2 yielded a diffuse
cytosolic distribution and absence of punctate staining
independent of incubation time. This observation suggests
that these peptides enter the cytoplasm by direct crossing of the
plasma membrane and not by endocytosis followed by
endosomal release. For both peptides, the presence of intact
peptide inside the cytosol was confirmed by the formation of
fluorescent GFP using a complementation assay. This finding
demonstrates that these eight/nine amino acid peptides had the
capacity to carry a nonhydrophobic peptide of twice the size
across the plasma membrane.
As a next step it will be highly interesting to investigate the in
vivo distribution of these peptides. Polycationic CPPs show a
propensity to accumulate in the liver and kidney.
22
If this was
not the case for these new CPPs, a further exploration in
combination with targeting ligands may be highly interesting.
23
In this case, the targeting ligand would guide the localization
while the CPP would drive uptake.
Following the work by Hä
llbrink et al., which was then also
applied to the prediction of CPPs with intrinsic biological
activity, this contribution is the second example of a successful
prediction of CPPs using a descriptor set of accumulated side
chain properties.
24
Nevertheless, our results also show the
limitations of this approach. Within the peptide collection the
most active one could only be identified by an empirical
approach. Also, the cell line dependence and uptake mechanism
Figure 8. GFP complementation and correlation to cytosolic concentration of the GFP11 variants. (A,D) Fluorescence for both mCherry and GFP
was detected by flow cytometry from morphologically intact cells gated based on forward versus sideward scatter after 2 h incubation with peptides
in FCS-free medium. The GFP fluorescence for cells was normalized to the background signal acquired from mock-treated cells. Curves indicate the
average of the normalized medians of (A) three and (D) two independent experiments. Error bars denote the standard error of the mean (SEM).
(B,E) Dose response curve for the fluorescein-labeled counterparts representing the concentration-dependent increase of the carboxyfluorescein
signal in the cyosol. (C,F) Correlation of the dose-dependent increase of GFP fluorescence and cell-associated carboxyfluorescein signal obtained
from the respective pairs of samples.
Bioconjugate Chemistry Article
DOI: 10.1021/acs.bioconjchem.6b00535
Bioconjugate Chem. XXXX, XXX, XXX−XXX
F

cannot be predicted at this point. Here, more research will be
needed to better understand the interplay of molecular and
cellular descriptors. Candidates for cellular descriptors are the
composition of the glycocalyx, the lipid composition, and the
presence of certain endosomal pathways.
25
■MATERIALS AND METHODS
Cell Culture. HeLa (DSMZ) and Jurkat E6.1 leukemia cells
(ATCC) were maintained in RPMI 1640 and HEK293T cells
(DSMZ) in DMEM (Gibco, Invitrogen, Eugene, OR, USA),
respectively, supplemented with 10% fetal calf serum (FCS;
PAN Biotech, Aidenbach, Germany). All cells were incubated at
37 °C in a 5% CO2-containing, humidified incubator. Cells
were passaged every 2 to 3 days. For detachment of adherent
cells phosphate-buffered saline (PBS) containing 2 mM EDTA
was used.
Peptides. Peptides extended by a nonaarginine moiety were
purchased as C-terminal peptide amides from EMC micro-
collections (Tübingen, Germany). The N-terminus was either
capped by acetylation or by a carboxyfluorescein moiety. All
other peptides were synthesized by Fmoc-strategy on an
automated CS336 synthesizer. Carboxyfluorescein was attached
on the resin using DIC/HOBt as coupling reagent. Peptides
were purified by reversed phase HPLC using water/ACN (0.1%
TFA) gradients to >95% purity. Lyophilization gave the final
peptide as TFA salt. Stock solutions were prepared in DMSO in
concentrations of 3−5 mM. Peptide concentrations were
determined by measuring the absorption of the fluorescein
moiety in 100 mM in Tris/HCl buffer pH 8.8, assuming a
molar extinction coefficient for fluorescein ε492 = 75 000 L/
(mol ×cm). For unlabeled peptides concentrations were
determined based on weight and molecular mass.
Transfection with GFP1−10. HEK293T cells were seeded
in 48-well plates for flow cytometry with 40 000 cells in DMEM
with FCS (10%) and cultured overnight at 37 °C, 5% CO2. The
next day, the culture medium was replaced and cells were
transfected with mCherry-GFP1−10 plasmid via lipofectamine
2000 (Invitrogen, Eugene, U.S.A.) under adherent condition.
13
After 24 h, the medium was replaced by fresh culture medium
containing 10% FCS and the cells were cultured overnight.
Flow Cytometry. For detection of GFP complementation
flow cytometry was performed on a CyAn ADP flow cytometer
(Beckman Coulter, Woerden, The Netherlands) using spectral
ranges 530/40 for GFP and 613/20 for mCherry. A BD FACS-
Calibur (BD Biosciences, Erembodegem, Belgium) flow
cytometer with a 488 nm argon ion laser was used for flow
cytometry of cells incubated with carboxyfluorescein-labeled
variants. FCS Express Version 5 Research Edition (De Novo
Software, Los Angeles, CA) was used for the analysis of the
generated data. Per sample, 10 000 morphologically intact cells
were gated based on forward versus sideward scatter and
analyzed.
CPP-Mediated Import of GFP11 Variants. The medium
was replaced by RPMI 1640 for adherent HeLa and Jurkat E6.1
leukemia suspension cells or by 150 μL DMEM for adherent
HEK293T cells. For incubation over 24 h, the medium was
supplemented with 10% FCS. Peptides were administered at
the indicated concentrations by adding a 150 μL premix. Cells
were detached using PBS containing 2 mM EDTA (HeLa and
HEK293T), transferred into tubes, and flow cytometry was
performed.
Confocal Microscopy. Cells were seeded in 8-well
microscopy chambers (IBIDI, München, Germany) with
40 000 cells per well in DMEM with FCS (10%) and cultured
overnight at 37 °C, 5% CO2. The next day, medium of the cells
was replaced by 150 μL RPMI 1640 (HeLa) or DMEM
(HEK293T) without phenol red and FCS. Cell culture medium
containing the indicated concentrations of peptide was
prepared as premix and transferred to the IBIDI 8-well
microscopy chambers. Confocal images were acquired 2 h
after peptide administration. To study the peptide mediated
import of the CPPs of the first round, confocal microscopy was
performed on a TCS SP5 confocal microscope (Leica
Microsystems, Mannheim, Germany) equipped with an HCX
PL APO 63×1.2 N.A. water immersion lens. Carboxyfluor-
escein fluorescence was excited using the 488 nm line of an
argon ion laser and fluorescence detected at 500−550 nm.
Confocal pictures for the study of CPPs of the second round
were taken on a TCS SP8 confocal microscope (Leica
Microsystems, Mannheim, Germany) equipped with an HCX
PL APO 63×1.2 N.A. water immersion lens after 2 or 24 h of
incubation with peptides. Carboxyfluorescein fluorescence was
excited using the 488 nm line of a white laser (WLL) and
fluorescence detected at 500−550 nm. Trypan blue was excited
using the 568 nm line of the WLL and fluorescence detected at
600−660 nm.
■ASSOCIATED CONTENT
*
SSupporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.bioconj-
chem.6b00535.
Supplemental data providing further information on the
principal properties of amino acids (Table 1), properties
of the designed CPPs including GFP11 (Figure 1) and of
the control peptides (Figure 2), uptake of control
peptides (Figure 3), maintenance of membrane integrity
for peptides of the second collection (Figure 4), uptake
of the peptides of the second collection at 10 μM after 2
h (Figure 5), and properties of the biased peptide
collection through which peptide L1−6 was identified
(Figure 6) (PDF)
R-code to calculate peptide properties (ZIP)
■AUTHOR INFORMATION
Corresponding Author
*E-mail: francesca.milletti@roche.com.
ORCID
Roland Brock: 0000-0003-1395-6127
Notes
The authors declare no competing financial interest.
■ACKNOWLEDGMENTS
This work was supported by the Roche postdoc programme
(RPF 272).
■REFERENCES
(1) Milletti, F. (2012) Cell-penetrating peptides: classes, origin, and
current landscape. Drug Discovery Today 17, 850−860.
(2) Mitchell, D. J., Kim, D. T., Steinman, L., Fathman, C. G., and
Rothbard, J. B. (2000) Polyarginine enters cells more efficiently than
other polycationic homopolymers. J. Pept. Res. 56, 318−325.
(3) Derossi, D., Joliot, A. H., Chassaing, G., and Prochiantz, A.
(1994) The third helix of the Antennapedia homeodomain trans-
Bioconjugate Chemistry Article
DOI: 10.1021/acs.bioconjchem.6b00535
Bioconjugate Chem. XXXX, XXX, XXX−XXX
G

locates through biological membranes. J. Biol. Chem. 269, 10444−
10450.
(4) Prochiantz, A. (1996) Getting hydrophilic compounds into cells:
lessons from homeopeptides. Curr. Opin. Neurobiol. 6, 629−634.
(5) Wallbrecher, R., Depre, L., Verdurmen, W. P., Bovee-Geurts, P.
H., van Duinkerken, R. H., Zekveld, M. J., Timmerman, P., and Brock,
R. (2014) Exploration of the design principles of a cell-penetrating
bicylic peptide scaffold. Bioconjugate Chem. 25, 955−964.
(6) Saar, K., Lindgren, M., Hansen, M., Eiriksdottir, E., Jiang, Y.,
Rosenthal-Aizman, K., Sassian, M., and Langel, U. (2005) Cell-
penetrating peptides: a comparative membrane toxicity study. Anal.
Biochem. 345,55−65.
(7) Fischer, D., Li, Y., Ahlemeyer, B., Krieglstein, J., and Kissel, T.
(2003) In vitro cytotoxicity testing of polycations: influence of
polymer structure on cell viability and hemolysis. Biomaterials 24,
1121−1131.
(8) Lin, Y.-Z., Yao, S., Veach, R. A., Torgerson, T. R., and Hawiger, J.
(1995) Inhibition of nuclear translocation of transcripion factor NF-kB
by a synthetic peptide containing a cell membrane-permeable motif
and nuclear localization sequence. J. Biol. Chem. 270, 14255−14258.
(9) Fischer, R., Waizenegger, T., Köhler, K., and Brock, R. (2002) A
quantitative validation of fluorophore-labelled cell-permeable peptide
conjugates: Fluorophore and cargo dependence of import. Biochim.
Biophys. Acta, Biomembr. 1564, 365−374.
(10) Hä
llbrink, M., Kilk, K., Elmquist, A., Lundberg, P., Lindgren, M.,
Jiang, Y., Pooga, M., Soomets, U., and Langel, Ü. (2005) Prediction of
cell-penetrating peptides. Int. J. Pept. Res. Ther. 11, 249−259.
(11) Sandberg, M., Eriksson, L., Jonsson, J., Sjostrom, M., and Wold,
S. (1998) New chemical descriptors relevant for the design of
biologically active peptides. A multivariate characterization of 87
amino acids. J. Med. Chem. 41, 2481−2491.
(12) van Asbeck, A. H., Beyerle, A., McNeill, H., Bovee-Geurts, P. H.,
Lindberg, S., Verdurmen, W. P., Hallbrink, M., Langel, U.,
Heidenreich, O., and Brock, R. (2013) Molecular Parameters of
siRNA-Cell Penetrating Peptide Nanocomplexes for Efficient Cellular
Delivery. ACS Nano 7, 3797−3807.
(13) Schmidt, S., Adjobo-Hermans, M. J., Wallbrecher, R.,
Verdurmen, W. P., Bovee-Geurts, P. H., van Oostrum, J., Milletti, F.,
Enderle, T., and Brock, R. (2015) Detecting Cytosolic Peptide
Delivery with the GFP Complementation Assay in the Low
Micromolar Range. Angew. Chem., Int. Ed. 54, 15105−15108.
(14) Milech, N., Longville, B. A., Cunningham, P. T., Scobie, M. N.,
Bogdawa, H. M., Winslow, S., Anastasas, M., Connor, T., Ong, F.,
Stone, S. R., et al. (2015) GFP-complementation assay to detect
functional CPP and protein delivery into living cells. Sci. Rep. 5, 18329.
(15) Cruciani, G., Baroni, M., Carosati, E., Clementi, M., Valigi, R.,
and Clementi, S. (2004) Peptide studies by means of principal
properties of amino acids derived from MIF descriptors. J. Chemom.
18, 146−155.
(16) Magzoub, M., Oglecka, K., Pramanik, A., Goran Eriksson, L. E.,
and Graslund, A. (2005) Membrane perturbation effects of peptides
derived from the N-termini of unprocessed prion proteins. Biochim.
Biophys. Acta, Biomembr. 1716, 126−136.
(17) Kosuge, M., Takeuchi, T., Nakase, I., Jones, A. T., and Futaki, S.
(2008) Cellular internalization and distribution of arginine-rich
peptides as a function of extracellular peptide concentration, serum,
and plasma membrane associated proteoglycans. Bioconjugate Chem.
19, 656−664.
(18) Ruttekolk, I. R., Witsenburg, J. J., Glauner, H., Bovee-Geurts, P.
H., Ferro, E. S., Verdurmen, W. P., and Brock, R. (2012) The
intracellular pharmacokinetics of terminally capped peptides. Mol.
Pharmaceutics 9, 1077−1086.
(19) Marks, J. R., Placone, J., Hristova, K., and Wimley, W. C. (2011)
Spontaneous membrane-translocating peptides by orthogonal high-
throughput screening. J. Am. Chem. Soc. 133, 8995−9004.
(20) Lattig-Tunnemann, G., Prinz, M., Hoffmann, D., Behlke, J.,
Palm-Apergi, C., Morano, I., Herce, H. D., and Cardoso, M. C. (2011)
Backbone rigidity and static presentation of guanidinium groups
increases cellular uptake of arginine-rich cell-penetrating peptides. Nat.
Commun. 2, 453.
(21) Mueller, J., Kretzschmar, I., Volkmer, R., and Boisguerin, P.
(2008) Comparison of cellular uptake using 22 CPPs in 4 different cell
lines. Bioconjugate Chem. 19, 2363−2374.
(22) Sarko, D., Beijer, B., Garcia Boy, R., Nothelfer, E. M., Leotta, K.,
Eisenhut, M., Altmann, A., Haberkorn, U., and Mier, W. (2010) The
pharmacokinetics of cell-penetrating peptides. Mol. Pharmaceutics 7,
2224−2231.
(23) Cheng, C. J., and Saltzman, W. M. (2011) Enhanced siRNA
delivery into cells by exploiting the synergy between targeting ligands
and cell-penetrating peptides. Biomaterials 32, 6194−6203.
(24) Jones, S., Holm, T., Mager, I., Langel, U., and Howl, J. (2010)
Characterization of bioactive cell penetrating peptides from human
cytochrome c: protein mimicry and the development of a novel
apoptogenic agent. Chem. Biol. 17, 735−744.
(25) Favretto, M. E., Wallbrecher, R., Schmidt, S., van de Putte, R.,
and Brock, R. (2014) Glycosaminoglycans in the cellular uptake of
drug delivery vectors - Bystanders or active players? J. Controlled
Release 180,81−90.
Bioconjugate Chemistry Article
DOI: 10.1021/acs.bioconjchem.6b00535
Bioconjugate Chem. XXXX, XXX, XXX−XXX
H