Access of therapeutic biomolecules to cytoplasmic and nuclear targets is hampered by the inability of these molecules to cross biological membranes. Approaches to overcome this hurdle involve CPPs (cell-penetrating peptides) or protein transduction domains. Most of these require rather high concentrations to elicit cell-penetrating functionality, are non-human, pathogen-derived or synthetic entities, and may therefore not be tolerated or even immunogenic. We identified novel human-protein-derived CPPs by a combination of in silico and experimental analyses: polycationic CPP candidates were identified in an in silico library of all 30-mer peptides of the human proteome. Of these peptides, 60 derived from extracellular proteins were evaluated experimentally. Cell viability and siRNA (small interfering RNA) transfection assays revealed that 20 out of the 60 peptides were functional. Three of these showed CPP functionality without interfering with cell viability. A peptide derived from human NRTN (neurturin), which contains an α-helix, performed the best in our screen and was uniformly taken up by cultured cells. Examples for payloads that can be delivered to the cytosol by the NRTN peptide include complexed siRNAs and both N- and C-terminally fused pro-apoptotic peptides.

INTRODUCTION

The hallmark of all life is compartmentalization. Genetic information as well as macromolecules such as proteins, peptides and carbohydrates are contained inside membrane-bound compartments. This thermodynamic barrier [1] not only prevents their loss to the outside world, but also allows cells to build-up gradients and tightly control trafficking of molecules. Macromolecules such as proteins and nucleic acids that are too big to cross cellular membranes require specialized transport systems like pores or vesicles [2,3] to traverse membranes.

Although compartmentalization and impermeability of membranes is essential for life, it is the mayor obstacle for delivery of therapeutic macromolecules into the cytoplasm of cells. So far, the intracellular target space is solely addressed by low-molecular-mass compounds that cross the cell membrane. The application of therapeutic macromolecules to this target space offers vast possibilities for the development of novel therapeutics.

To deliver macromolecules to the cytoplasm various approaches such as lipids [4], viral carriers [5] or nanoparticles [6] are currently being evaluated. One class of substances that has gained interest for the delivery of biomolecules into cells in previous years are CPPs (cell-penetrating peptides) [7]. These peptides are taken up by cells, penetrate into their cytosol and have the ability to deliver payloads into these cells [8].

CPPs are a heterogeneous group of peptides. Some are derived from viral proteins such as Tat [9,10] and Rev [11], others are part of venoms or toxins such as crotamin [12] or melittin [13]. Others have been designed in silico, such as stapled [14] or amphiphilic [15] peptides. Owing to their heterogeneity in sequence and structure, no common mode of action has so far been shown to account for their translocation. Cationic peptides are believed to enter cells by their interaction with proteoglycans on the surface of cells [16], followed by macropinocytosis. Once taken up, the peptides are still inside membrane-bound compartments, essentially still ‘outside’ of the cell. How they cross this barrier is a matter of extensive debate and research. It has been proposed that they are either retrogradely transported to the ER (endoplasmic reticulum) where they escape via the translocation machinery [17] or that they directly translocate across membranes [18].

Most CPPs, with few exceptions [19], have one potential issue in common with regards to their therapeutic applications: they are not of human origin and thus might not be tolerated by the immune system [20,21]. To minimize potential immunogenicity issues and identify non-pathogen-derived CPP modules, we searched the human proteome for peptide sequences with CPP functionality.

We applied an in silico procedure to identify CPPs in the human proteome that was based on a library of all 30-mer peptides. Our screen identified 20 out of 60 human peptides with either cytotoxicity and/or CPP functionality in cell-based assays. Three peptides did not display significant toxicity at concentrations sufficient for cell penetration and payload delivery, showing the applicability of our combined screen. Further analyses then focused on a peptide derived from the protein neurturin (termed NRTN). This peptide bound to cells, was internalized and delivered payloads into the cytoplasm. Experiments aimed at identifying the features of the peptide that are responsible for its activity indicate that an α-helix in combination with an arginine-rich stretch are required for functionality.

MATERIALS AND METHODS

Peptide synthesis

Peptide syntheses were performed according to established protocols (FastMoc 0.25 mmol) in an automated Applied Biosystems ABI 433A peptide synthesizer using Fmoc (fluoren-9-ylmethoxycarbonyl) chemistry. Further information is provided in the Supplementary section (at http://www.BiochemJ.org/bj/442/bj4420583add.htm).

Mammalian cell culture

MCF7 cells were seeded at a density of 15000 cells per well in 96-well plates. The cells were incubated for 24 h at 37°C, 5% CO2 and 85% humidity in RPMI 1640 medium with 10% FBS (fetal bovine serum) and L-glutamine. For functional assays, the cells were washed and incubated in Opti-MEM® (Invitrogen). Complexes of siRNA (small interfering RNA) and peptides were formed by incubation of 100 nM siRNA with the indicated peptide concentrations in Opti-MEM® for 30 min at room temperature (21°C). The complexes were added to the cells and incubated for 3 h. Subsequently, the medium was replaced with normal growth medium. The cells were incubated for a further 21 h for the generation of bDNA (branched DNA) lysates or CellTiter-Glo® (Promega) assays. For analysis of toxic fusion peptides, the peptides were added to the cells in normal growth medium and incubated for 24 h prior to analysis with the CellTiter-Glo® assay.

siRNA transfection screening

Complexes of peptides and siRNA oligonucleotides {AHA1 [activator of 90 kDa Hsp (heat-shock protein) ATPase homologue 1], AGUCAAAAUCCCCACUUGUTT; Luc (luciferase) CUUACGCUGAGUACUUCGATT} were formed in Opti-MEM® and incubated for 15 min at room temperature. The cells were incubated in the presence of the complexes for 3 h in Opti-MEM®. After a wash step, the cells were incubated for a further 21 or 45 h in normal growth medium. Thereafter, the levels of AHA1 and GAPDH (glyceraldehyde-3-phosphate dehydrogenase) mRNA were measured with a bDNA assay (Affymetrix).

bDNA analysis

mRNA levels were quantified by bDNA assays according to the manufacturer's instructions (Affymetrix). To perform these assays, cells were seeded into 96-well plates and allowed to attach overnight. At 24 h after transfection, the cells were lysed and transferred on to a capture plate in the presence of a gene-specific probe set, followed by an overnight incubation at 53°C. The plates were then incubated at 53°C sequentially with an amplifier and an alkaline phosphatase-linked label probe. After a final wash step, the luminescent alkaline phosphatase substrate dioxitane was added for 30 min at 53°C. The luminescence was detected using an InfiniteF200 luminescence reader (Tecan Austria)

CellTiter-Glo® assays

Cells were incubated for 48 h in the presence of peptides and siRNA. To measure the number of living cells, the CellTiter-Glo® Luminescent Cell Viability assay (Promega) was used according to the manufacturer's instructions. The cells were lysed and a luminescent signal proportional to the amount of ATP was generated. The 96-well plates were then analysed in an InfiniteF200 luminescence reader (Tecan Austria).

FACS analysis

Cells were detached by incubation in accutase for 15 min. After washing in FACS buffer (PBS containing 5% FBS) the cells were seeded in a 96-well round-bottomed multi-well plate (Corning, catalogue number 3799) to a final density of 3×105 cells/ml. The cells were incubated in the presence of 1, 5 or 10 μM FITC-labelled peptides in Opti-MEM® for 3 h at 37°C. The cells were then washed in FACS buffer and incubated in proteinase K-containing (0.02 mg/ml) buffer for 30 min at 37°C. The cells were washed twice and analysed with the FACSCanto II (BD Biosciences).

Fluorescence microscopy

Cells were grown on glass coverslips to a density of 50–70%. They were treated with 10 μM FITC-labelled peptides in Opti-MEM® for 30 min at 37°C. The cells were then washed twice with ice-cold growth medium containing FBS. Subsequently, the cells were washed twice with ice-cold PBS and fixed with paraformaldehyde. The DNA was labelled with DAPI (4′,6-diamidino-2-phenylindole; Roche) at a concentration of 10 μg/ml. Samples were analysed with a Leica SP20 confocal microscope.

Bcl-2 binding assays

Nur77 (nuclear receptor subfamily 4 group A member 1 isoform 1)-fused or unmodified NRTN peptide was coupled to iodoacetyl beads (Pierce) according to the manufacturer's instructions. The beads were subsequently incubated with recombinant Bcl-2 (Calbiochem, catalogue number PF126) and washed with PBS containing 0.5 M NaCl, 0.025 M sodium azide and 0.05% Tween 20 to remove unbound protein. Binding of Bcl-2 to the immobilized peptides was determined by eluting (pH 2.8) and transfer on to nitrocellulose membranes (Invitrogen), followed by detection with the MemCode Reversible Protein Stain kit (Pierce) according to the manufacturer's instructions.

CD spectroscopy

CD analysis was performed using a Jasco J715 Spectropolarimeter. The measurement was conducted from 260 to 195 nm with a data pitch of 0.1 nm and a bandwidth of 1 nm. The cell used had a length of 0.1 cm and the peptides measured were tested at a concentration of 0.1 mg/ml. To induce structure formation, TFE (trifluorethanol; Sigma) was added up to 50% (v/v).

RESULTS

Bioinformatic identification of potential CPPs derived from extracellular human proteins

To identify peptides of human origin with CPP functionality, an in silico library that covers all human-protein-derived peptides was generated. Subsequently, various ‘filtering’ steps were applied to that library to define CPP candidates. The library was compiled from overlapping peptides derived from the SwissProt database. A window size of 30 amino acids was chosen and applied to all human SwissProt entries (Figure 1). Peptides of this size have a higher probability of having secondary structure elements than 8–10-mers commonly used as CPPs. Secondary structures have been shown to be advantageous for cell penetration [22], therefore selecting 30-mers should increase the chance to identify novel CPPs. The in silico library contained 10459557 individual peptides. Each of these peptides was linked to both source {sequence, source protein ID, position within the protein, GO (Gene Ontology) [23] annotations of the source protein} and sequence information [amino acid composition, number of charges, IEP (isoelectric point) and hydrophobicity].

Overview of the bioinformatical screen

Figure 1
Overview of the bioinformatical screen

All human proteins were added with a description and GO annotation (A) and analysed with a sliding window of 30 amino acids (B). To identify putative CPPs, bioinformatic filters were applied (C): first, only peptides with ten or more positive charges were chosen; secondly, only extracellular proteins with low probability of immunogenicity were chosen.

Figure 1
Overview of the bioinformatical screen

All human proteins were added with a description and GO annotation (A) and analysed with a sliding window of 30 amino acids (B). To identify putative CPPs, bioinformatic filters were applied (C): first, only peptides with ten or more positive charges were chosen; secondly, only extracellular proteins with low probability of immunogenicity were chosen.

Many CPPs such as Tat, penetratin, Rev and poly-arginine are characterized by a high content of positively charged amino acids. Therefore a filter that removes all peptides with less than 30% positive charged residues (arginine, histidine and lysine) was applied to enrich for cationic peptides. The resulting sub-library contained 227307 individual cationic peptides.

A further filter was applied thereafter to select for peptides with a high probability of being immune-tolerated. This filter retained only peptides derived from secreted/extracellular proteins which are visible to the immune system. This step resulted in 8630 peptides from 583 individual proteins. From these 583 peptides from unique proteins, 60 were chosen for experimental evaluation (Supplementary Table S1 at http://www.BiochemJ.org/bj/442/bj4420583add.htm). Disulfide bridges were maintained in synthesis and single cysteine residues were changed to alanine.

Several control peptides were chosen for assay validation and benchmarking: Tat, Rev [11], 10xArg [24] as the bona fide cationic peptide, truncated protamine [25] and MTS [26] as examples for CPPs derived from human proteins. All candidate and control peptides that were applied to experimental evaluation are listed in Supplementary Table S1, a dot-plot of their IEP and hydrophobicity (compared with 500 randomly chosen human peptides) is shown in Supplementary Figure S1 at http://www.BiochemJ.org/bj/442/bj4420583add.htm.

CPP identification by siRNA transfection screening

The challenge in screening for potent non-toxic CPPs is to find sensitive and robust assays that can differentiate CPP-mediated delivery from non-specific uptake, and detect growth inhibition or cell death caused by the peptides. We applied a screen for siRNA transfection [27,28]. In contrast with tools frequently applied for CPP identification (fluorescent or cytotoxic entities with determination of cell-associated fluorescence or cytotoxicity), this screen met all requirements for CPP identification: RNAi (RNA interference) can only occur when the siRNA molecule is delivered into the cytoplasm and non-specific peptide-mediated toxicity can be identified given the knockdown of the target mRNA does not affect cell viability. Furthermore, mRNA knockdown can be determined by a sensitive and robust bDNA assay that allows measuring the mRNA of the targeted message against a reference mRNA to ensure specificity.

To screen the CPP candidates for their transfection ability, non-covalent charge-based complexation [29,30] was used. For complexation with siRNA, an increasing concentration of peptides (1–20 μM) was added to 100 nM siRNA and incubated for 15 min. When the ability of selected peptides to form complexes with the siRNA molecules was analysed by electrophoretic mobility-shift assays, they were found to retard the mobility, dependent on the peptide concentration (Supplementary Figure S2 at http://www.BiochemJ.org/bj/442/bj4420583add.htm). The siRNAs applied in the screen were targeting the Hsp90 co-chaperone Aha1 (siAha1) [31] and firefly luciferase (siLuc) as a control. Addition of these siRNAs to cells without transfection reagents does not cause mRNA knockdown. Transfection of either siRNA using standard lipid-based transfection reagents did not interfere with cell viability, whereas siAha1 led to a significant reduction of AHA1 mRNA compared with a reference mRNA (Supplementary Figure S3 at http://www.BiochemJ.org/bj/442/bj4420583add.htm). For screening, pre-formed siRNA–peptide complexes were added to cultured cells and mRNA levels were determined after 24 h of incubation.

As a measure for specific mRNA knockdown, AHA1 mRNA levels were determined relative to GAPDH in siAha1- and siLuc-transfected cells. The AHA1/GAPDH ratio of nontransfected control cells was set to 100% (for each assay) and expressed as relative units for siAha1 or siLuc-transfected cells. For this normalized data set, the difference between siLuc and siAha1-transfected cells provided a readout for specific knockdown. An increasing difference reflects increasing efficacy of CPP-mediated transfection. The assay was validated using the selected control peptides as shown for Tat and 10xArg as an example (Supplementary Figure S4a at http://www.BiochemJ.org/bj/442/bj4420583add.htm).

GAPDH mRNA levels were also used as a readout for cytotoxic or growth inhibitory effects of the peptides. The CellTiter-Glo® assay showed that cytotoxicity correlated well with the reduction of overall GAPDH mRNA (Supplementary Figure S4b). A reduction of GAPDH mRNA due to cytotoxicity furthermore influenced the AHA1/GAPDH ratio. Therefore these values were not informative for peptides with strong cytotoxic phenotypes, as indicated by dramatic decreases in GAPDH levels [shown, for example, for BPIL3 (BPI-fold-containing family B, member 6) in Figure 3b and Supplementary Figure S4b]. Minor influences of the peptides on cellular viability did not interfere with the determination of AHA1/GAPDH.

Classification of peptides by transfection potency and growth inhibition

For each peptide, data pairs of the transfection value (normalized AHA1/GAPDH difference siAha1 compared with siLuc) and the viability value (normalized overall GAPDH mRNA levels) were collected. This information is compiled in Figure 2, with transfection values indicated on the x-axis and viability values on the y-axis. To classify peptide candidates, transfection mediated by the Tat peptide (Supplementary Figure S4a) was used to define a threshold for transfection potency: peptides with less transfection activity were rated as non-transfecting. Average measurements of 100% to 75% of GAPDH mRNA relative to the medium control were considered as not or moderately toxic, values of less than 75% were considered toxic. On the basis of both cutoffs, four categories of peptides were defined. The ‘non-functional’ category (less transfection efficacy than Tat and no toxicity) was assigned to 41 out of the 61 screened human-protein-derived peptides (Figure 2). Two of the control peptides (crotamine and MTS) fell into this category. Figure 3(a) shows the phenotype of a peptide derived from the protein WNT16 (GenBank® accession number Q9UBV4); the corresponding viability assay is shown in Supplementary Figure S4(b). The ‘toxic’ category (less transfection activity than Tat and toxic) was assigned to 11 out of the 61 peptides (Figure 2). A peptide derived from BPIL3 (GenBank® accession number Q8NFQ5) serves as an example (Figure 3b); the corresponding viability assay is shown in Supplementary Figure S4(b). Both AHA1 and GAPDH mRNA levels are affected similarly, independent of the siRNA used, explaining the abnormal AHA1/GAPDH values. The ‘transfecting toxic’ category (equal or higher transfection potency than Tat and toxicity) was assigned to five out of the 61 human-protein-derived peptides (Figure 2). Figure 3(c) shows a peptide derived from the CU025 protein (GenBank® accession number Q9Y426) as an example for this peptide class.

Dot plot analysis of the siRNA transfection screen

Figure 2
Dot plot analysis of the siRNA transfection screen

All peptides analysed for their ability to transfect siRNA and cellular toxicity are plotted for their average GAPDH mRNA levels at 20 μM siRNA (y-axis, ‘Toxicity’) and the difference of AHA1/GAPDH for siLuc and siAha1 at 20 μM (x-axis, ‘Transfection’). Threshold values for toxicity (<75% average GAPDH content) and transfection (>18%, Tat) are indicated by broken lines. Top left-hand section, ‘non-functional’ peptides; top right-hand section, ‘transfecting’ peptides; bottom left-hand section, ‘toxic’ peptides; bottom right-hand section, ‘transfecting toxic’ peptides. The inset shows toxic peptides outside of the range of the plotted area.

Figure 2
Dot plot analysis of the siRNA transfection screen

All peptides analysed for their ability to transfect siRNA and cellular toxicity are plotted for their average GAPDH mRNA levels at 20 μM siRNA (y-axis, ‘Toxicity’) and the difference of AHA1/GAPDH for siLuc and siAha1 at 20 μM (x-axis, ‘Transfection’). Threshold values for toxicity (<75% average GAPDH content) and transfection (>18%, Tat) are indicated by broken lines. Top left-hand section, ‘non-functional’ peptides; top right-hand section, ‘transfecting’ peptides; bottom left-hand section, ‘toxic’ peptides; bottom right-hand section, ‘transfecting toxic’ peptides. The inset shows toxic peptides outside of the range of the plotted area.

Detailed analysis of human-protein-derived peptides

Figure 3
Detailed analysis of human-protein-derived peptides

(a) Analysis of WNT16, a ‘non-functional’ peptide. (b) Analysis of BPIL3, a ‘toxic’ peptide. The AHA1/GAPDH ratio (left-hand panel), AHA1 values (middle panel) and GAPDH values (right-hand panel) are shown. (c) Analysis of CU025, a ‘transfecting toxic’ peptide. The AHA1/GAPDH ratio (left-hand panel) and viability (right-hand panel) are shown. (d) Analysis of the ‘transfecting’ peptides CPXM2 and ASM3B. The AHA1/GAPDH ratio is shown. (e) Analysis of NRTN, a ‘transfecting’ peptide. The The AHA1/GAPDH ratio (left-hand panel) and viability (right-hand panel) are shown. In all cases, viability is expressed as the fold change against the medium control. All transfection values are relative to the medium control. mRNA levels at 1 μM were set to 100% to show dose-dependent effects. siAha1 is shown as black filled squares and siLuc is shown as non-filled circles. (f) Analysis of NRTN as a transfecting peptide for an siRNA targeting Eg5. The EG5/GAPDH ratio is shown. All transfection values are relative to the medium control. mRNA values obtained at 1 μM were set to 100% to show dose-dependent effects. Eg5 siRNA is shown as black filled squares and siLuc is shown as non-filled circles.

Figure 3
Detailed analysis of human-protein-derived peptides

(a) Analysis of WNT16, a ‘non-functional’ peptide. (b) Analysis of BPIL3, a ‘toxic’ peptide. The AHA1/GAPDH ratio (left-hand panel), AHA1 values (middle panel) and GAPDH values (right-hand panel) are shown. (c) Analysis of CU025, a ‘transfecting toxic’ peptide. The AHA1/GAPDH ratio (left-hand panel) and viability (right-hand panel) are shown. (d) Analysis of the ‘transfecting’ peptides CPXM2 and ASM3B. The AHA1/GAPDH ratio is shown. (e) Analysis of NRTN, a ‘transfecting’ peptide. The The AHA1/GAPDH ratio (left-hand panel) and viability (right-hand panel) are shown. In all cases, viability is expressed as the fold change against the medium control. All transfection values are relative to the medium control. mRNA levels at 1 μM were set to 100% to show dose-dependent effects. siAha1 is shown as black filled squares and siLuc is shown as non-filled circles. (f) Analysis of NRTN as a transfecting peptide for an siRNA targeting Eg5. The EG5/GAPDH ratio is shown. All transfection values are relative to the medium control. mRNA values obtained at 1 μM were set to 100% to show dose-dependent effects. Eg5 siRNA is shown as black filled squares and siLuc is shown as non-filled circles.

The ‘transfecting’ category (equal or higher transfection potency than Tat without significant reduction of cell viability) was assigned to three out of the 61 peptides (Figure 2). Four control peptides fell into this class (Tat, Rev, 10xArg and protamine). The three peptides are derived from CPXM2 (GenBank® accession number Q8N436) (Figure 3d), ASM3B (GenBank® accession number Q92485) (Figure 3d) and NRTN (GenBank® accession number Q99748) (Figure 3e). The NRTN peptide mediated effective reduction of AHA1 mRNA levels relative to GAPDH mRNA levels with only minor effects on cell viability (Figure 3e). To examine the general applicability of the NRTN peptide for siRNA transfection and to exclude a sequence specific effect, a siRNA targeting Eg5 (kinesin-like protein KIF11) [32] was used analogous to siAha1. Using this siRNA, a specific reduction of the EG5 mRNA (Figure 3f) was observed. On the basis of the finding that the NRTN peptide was the most potent peptide in siRNA transfection, it was chosen for further characterization.

Cellular uptake of the NRTN peptide does not require siRNA complexation

To investigate whether the candidate peptides act as universal or as a siRNA-restricted CPPs, N-terminally FITC-labelled peptides were analysed by FACS and microscopy. For all subsequent analyses, the NRTN peptide was used as the CPP example, the WNT16 peptide as a non-functional peptide and the Tat peptide as a control.

For determination of cellular uptake in FACS experiments, MCF7 cells were incubated in the presence of fluorescent peptides for 3 h at 37°C and subsequently treated with proteinase K to remove surface-bound peptides. In agreement with previous studies, Tat bound to cells and was internalized (Figure 4a, left-hand panel). The ‘non-functional’ peptide WNT16 did not show significant uptake (Figure 4a, middle panel). The ‘transfecting’ peptide NRTN on the other hand was internalized into MCF7 cells (Figure 4a, right-hand panel) in a concentration-dependent manner.

Cellular uptake of the NRTN peptide

Figure 4
Cellular uptake of the NRTN peptide

(a) FACS analysis of Tat (left-hand panel), WNT16 (middle panel) and NRTN (right-hand panel) with N-terminal FITC label. MCF7 cells were incubated for 3 h in the presence of peptide, treated with proteinase K for 30 min and analysed by FACS for internalized peptides in the FITC channel. Red lines correspond to 1 μM, green lines correspond to 5 μM and blue lines correspond to 10 μM. (b) Microscopic analysis of the uptake of Tat (left-hand panel), WNT16 (middle panel) and NRTN (right-hand panel). MCF7 cells grown on glass coverslips were treated with 10 μM FITC-labelled peptides (green) for 30 min. The cells were then washed and fixed with paraformaldehyde. The DNA was labelled with DAPI (blue).

Figure 4
Cellular uptake of the NRTN peptide

(a) FACS analysis of Tat (left-hand panel), WNT16 (middle panel) and NRTN (right-hand panel) with N-terminal FITC label. MCF7 cells were incubated for 3 h in the presence of peptide, treated with proteinase K for 30 min and analysed by FACS for internalized peptides in the FITC channel. Red lines correspond to 1 μM, green lines correspond to 5 μM and blue lines correspond to 10 μM. (b) Microscopic analysis of the uptake of Tat (left-hand panel), WNT16 (middle panel) and NRTN (right-hand panel). MCF7 cells grown on glass coverslips were treated with 10 μM FITC-labelled peptides (green) for 30 min. The cells were then washed and fixed with paraformaldehyde. The DNA was labelled with DAPI (blue).

To further study the cellular uptake of these peptides, fluorescence microscopy was applied. MCF7 cells grown on glass coverslips were incubated in 10 μM FITC-labelled peptides for 30 min, washed and fixed with paraformaldehyde. Consistent with the FACS analysis and previous studies, the Tat peptide was internalized under these circumstances (Figure 4b, left-hand panel). Using concentrations above 10 μM, nuclear staining was observed in a subpopulation of cells (Supplementary Figure S5 at http://www.BiochemJ.org/bj/442/bj4420583add.htm). The WNT16 peptide did not give rise to a significant signal (Figure 4b, middle panel) as expected on the basis of its inability to transfect siRNA and give rise to a FACS signal. The NRTN peptide was efficiently taken up by the cells and localized to intracellular membrane-bound compartments (Figure 4b, right-hand panel). This peptide showed a homogeneous staining patter throughout the entire cell population and over a wide range of concentrations.

Taken together, these data show that the NRTN peptide binds to cells and is internalized. Furthermore, the NRTN peptide does not require siRNA complexation to be functional.

NRTN peptide mediated intracellular delivery of an N-terminalfused Nur77 peptide

The internalization of the NRTN peptide shown by FACS and microscopy indicated that this peptide functions not only as a transfection reagent for siRNA molecules, but also as a CPP. However, due to limited sensitivity, crossing into the cytosol cannot be shown using fluorescence-based methods. To investigate whether the peptide possess ‘classical’ CPP functionality and is capable of mediating cellular uptake of payloads, biologically active peptides were fused to NRTN.

One such peptidic payload was a Nur77-derived peptide that interacts with Bcl-2 and converts it into a pro-apoptotic factor [33]. The Nur77 peptide itself cannot cross cell membranes and is therefore not cytotoxic upon external administration. It has to be combined with a CPP to reach its mitochondrial membrane-associated target Bcl-2. A hybrid sequence containing Nur77 at the N-terminus that extended into the NRTN peptide at the C-terminus was designed (Nur–NRTN, Figure 5a). The point of fusion of the two peptides was defined within a stretch with sequence similarity between both peptides (Figure 5a). The fusion peptide retained Bcl-2 binding functionality. Bcl-2 pull-down experiments showed that Nur–NRTN, but not the NRTN peptide, was able to bind recombinant Bcl-2 (Figure 5b). Additional control peptides that were generated included a fusion of Nur77 to the WNT16 peptide, and a fusion of NRTN to an inactive version (Nur-In–NRTN) of the Nur77 peptide [33] (Figure 5a).

Cytosolic delivery of Nur77 by the NRTN peptide

Figure 5
Cytosolic delivery of Nur77 by the NRTN peptide

(a) The Nur77 sequence (italics) was fused to the NRTN peptide at a region of sequence similarity (grey box). Altered amino acids in inactive peptides are bold and underlined. (b) Nur–NRTN- and NRTN-coupled beads were incubated with recombinant Bcl-2 and washed. Both flow-through (FT) and eluted protein (PD) were collected and transferred on to nitrocellulose membranes and analysed. (c) MCF7 cells were incubated in the indicated concentration of NRTN (black squares, solid line), Nur77 (black circles, solid line) or Nur–NRTN (black triangles, solid line) peptides for 24 h (left-hand panel). MCF7 cells were incubated in the indicated concentration of Nur–NRTN (black triangles, solid line) or Nur-In–NRTN (black triangles, solid line) peptides for 24 h (middle panel). MCF7 cells were incubated in the indicated concentration of Nur77 (black triangles, solid line) or Nur–WNT16 (black rectangles, solid line) peptides for 24 h (right-hand panel). In all cases, the cells were then analysed using a CellTiter-Glo® assay. Viability is expressed as the fold change over the medium control.

Figure 5
Cytosolic delivery of Nur77 by the NRTN peptide

(a) The Nur77 sequence (italics) was fused to the NRTN peptide at a region of sequence similarity (grey box). Altered amino acids in inactive peptides are bold and underlined. (b) Nur–NRTN- and NRTN-coupled beads were incubated with recombinant Bcl-2 and washed. Both flow-through (FT) and eluted protein (PD) were collected and transferred on to nitrocellulose membranes and analysed. (c) MCF7 cells were incubated in the indicated concentration of NRTN (black squares, solid line), Nur77 (black circles, solid line) or Nur–NRTN (black triangles, solid line) peptides for 24 h (left-hand panel). MCF7 cells were incubated in the indicated concentration of Nur–NRTN (black triangles, solid line) or Nur-In–NRTN (black triangles, solid line) peptides for 24 h (middle panel). MCF7 cells were incubated in the indicated concentration of Nur77 (black triangles, solid line) or Nur–WNT16 (black rectangles, solid line) peptides for 24 h (right-hand panel). In all cases, the cells were then analysed using a CellTiter-Glo® assay. Viability is expressed as the fold change over the medium control.

When applied to MCF7 cells, neither the NRTN nor the Nur77 peptide significantly interfered with viability (Figure 5c, left-hand panel). In contrast, the Nur–NRTN fusion peptide led to a significant dose-dependent reduction of cell viability (Figure 5c, left-hand panel). This effect was not observed when the inactive Nur-In–NRTN peptide was used (Figure 5c, middle panel), pointing towards a specific effect of the Nur77 sequence. Neither the WNT16 peptide nor its fusions with the Nur77 peptide, Nur–WNT16, significantly interfered with the viability of the cells (Figure 5c, right-hand panel).

NRTN peptide-mediated intracellular delivery of a C-terminal fused 4EBP1 peptide

Another payload applied to evaluate NRTN-mediated cytoplasmic delivery was a 4E-BP1 [eIF4E (eukaryotic translation initiation factor 4E)-binding protein 1]-derived peptide called 4EBP1 [34,35]. This peptide interferes with the translation initiation by eIF4E and induces apoptosis in cancer cells. The 4EBP1 peptide itself cannot cross cell membranes and is therefore not cytotoxic upon external administration. It has to be combined with CPPs, such as Tat or penetratin, to cross cell membranes and induce cell death. To evaluate whether the fusion position is of importance for the CPP functionality of the NRTN peptide, the 4EBP1 peptide (Figure 6a) as well as an inactive version of 4EBP1 (4EBP1-In) [35], were added to its C-terminus (in contrast with Nur77 which replaced the N-terminal part of NRTN). A C-terminal fusion of 4EBP1 to the WNT16 peptide served as an additional control.

Cytosolic delivery of 4EBP1 by the NRTN peptide

Figure 6
Cytosolic delivery of 4EBP1 by the NRTN peptide

(a) The 4EBP1 sequence (italics) was fused to the C-terminus of the NRTN peptide. Altered amino acids in inactive peptides are bold and underlined. (b) MCF7 cells were incubated in the indicated concentration of NRTN (black squares, solid line), 4EBP1 (black circles, solid line) or NRTN–4EBP1 (black triangles, solid line) peptides for 24 h (left-hand panel). MCF7 cells were incubated in the indicated concentration of NRTN–4EBP1 (black triangles, solid line) or NRTN–4EBP1-In (black triangles, broken line) peptides for 24 h (middle panel). MCF7 cells were incubated in the indicated concentration of 4EBP1 (black triangles, solid line) or WNT16–4EBP1 (black rectangles, solid line) peptides for 24 h (right-hand panel). In all cases, the cells were analysed using a CellTiter-Glo® assay. Viability is expressed as the fold change over the medium control.

Figure 6
Cytosolic delivery of 4EBP1 by the NRTN peptide

(a) The 4EBP1 sequence (italics) was fused to the C-terminus of the NRTN peptide. Altered amino acids in inactive peptides are bold and underlined. (b) MCF7 cells were incubated in the indicated concentration of NRTN (black squares, solid line), 4EBP1 (black circles, solid line) or NRTN–4EBP1 (black triangles, solid line) peptides for 24 h (left-hand panel). MCF7 cells were incubated in the indicated concentration of NRTN–4EBP1 (black triangles, solid line) or NRTN–4EBP1-In (black triangles, broken line) peptides for 24 h (middle panel). MCF7 cells were incubated in the indicated concentration of 4EBP1 (black triangles, solid line) or WNT16–4EBP1 (black rectangles, solid line) peptides for 24 h (right-hand panel). In all cases, the cells were analysed using a CellTiter-Glo® assay. Viability is expressed as the fold change over the medium control.

When applied to MCF7 cells, neither the NRTN peptide nor the 4EBP1 peptide showed significant cytotoxicity (Figure 6b, left-hand panel). In contrast, the NRTN–4EBP1 fusion peptide mediated a significant dose-dependent reduction of cell viability (Figure 6b, left-hand panel). This effect was not observed when the inactive NRTN–4EBP1-In peptide was applied (Figure 6b, middle panel), which points to a specific effect of the 4EBP1 peptide portion. Cytotoxicity was also not observed when the WNT16–4EBP1 control peptide was applied (Figure 6b, right-hand panel).

Taken together, these results show that non-cell-permeable peptides can enter cells and elicit intracellular activity when they are fused to the NRTN peptide. This finding demonstrates that the activity of the NRTN peptide is not restricted to the delivery of siRNA payload. Furthermore, the results indicate that the NRTN-derived sequences can be altered at both the N- and C-terminus and still retain CPP functionality.

The NRTN peptide has a helical structure element

CPPs frequently contain secondary structural elements, in particular α-helices [22,36]. The NRTN peptide is derived from neurturin [37], which is closely related to GDNF (glial-derived neurotrophic factor; GenBank® accession number CAG46721) and ARTN (artemin; GenBank® accession number AAD13110). The crystal structures of both of these proteins have been solved [38,39]. An alignment of neurturin, ARTN and GDNF shows the NRTN peptide covering stretch of neurturin holds an α-helix in both homologue sequences (Figure 7a and Supplementary Figure S6 at http://www.BiochemJ.org/bj/442/bj4420583add.htm).

Helical elements in NRTN

Figure 7
Helical elements in NRTN

(a) Sequence alignment of Rat (rn) GDNF (GenBank® accession number CH474048), human (hs) ARTN and human (hs) NRTN. The α-helical stretch of GDNF and ARTN is indicated by a grey box. The peptide derived from the NRTN protein is boxed. Arginine residues potentially involved in sulfated proteoglycan binding are indicated by an asterisk. An arrow indicates a cysteine residue that is changed to alanine in the peptides. (b) UV-CD spectroscopic analysis of the NRTN (left-hand panel) and WNT16 (right-hand panel) peptides. Analysis in water is indicated by a dotted line, 10% TFE by a dashed line, 25% TFE by a dashed and dotted line and 50% TFE by a solid line. (c) The 4EBP1 and the Nur77 sequences (underlined) were fused to the N- or C-termini of human (hs) GDNF or ARTN respectively. MCF7 cells were incubated in the indicated concentration of GDNF (left-hand panel) of ARTN (right-hand panel) fusion peptides for 24 h. In all cases, the cells were then analysed using a CellTiter-Glo® assay. Viability is expressed as the fold change over the medium control.

Figure 7
Helical elements in NRTN

(a) Sequence alignment of Rat (rn) GDNF (GenBank® accession number CH474048), human (hs) ARTN and human (hs) NRTN. The α-helical stretch of GDNF and ARTN is indicated by a grey box. The peptide derived from the NRTN protein is boxed. Arginine residues potentially involved in sulfated proteoglycan binding are indicated by an asterisk. An arrow indicates a cysteine residue that is changed to alanine in the peptides. (b) UV-CD spectroscopic analysis of the NRTN (left-hand panel) and WNT16 (right-hand panel) peptides. Analysis in water is indicated by a dotted line, 10% TFE by a dashed line, 25% TFE by a dashed and dotted line and 50% TFE by a solid line. (c) The 4EBP1 and the Nur77 sequences (underlined) were fused to the N- or C-termini of human (hs) GDNF or ARTN respectively. MCF7 cells were incubated in the indicated concentration of GDNF (left-hand panel) of ARTN (right-hand panel) fusion peptides for 24 h. In all cases, the cells were then analysed using a CellTiter-Glo® assay. Viability is expressed as the fold change over the medium control.

To examine the NRTN peptide for the presence of secondary structural elements, UV-CD spectroscopy was applied. In buffer containing TFE, which stabilizes and induces formation of secondary structures [40], helical elements were observed (minima at 208 and 222 nm) (Figure 7b, left-hand panel). The well-characterized α-helical FALL peptide (Supplementary Figure S4b) [41] showed a similar behaviour (Supplementary Figure S7 at http://www.BiochemJ.org/bj/442/bj4420583add.htm). In contrast, the ‘non-functional’ WNT16 peptide adopted a random coil conformation (Figure 7b, right-hand panel) under all conditions. These data seem to point to the fact that helical elements are required for CPP functionality.

To analyse whether the presence of a helical portion is sufficient for CPP activity, fusions of the sequence stretches of ARTN and GDNF (Figure 7a) with the toxic Nur77 and 4EBP1 peptides were generated. Application of these fusion peptides to MCF7 cells showed that neither the ARTN- nor GDNF-derived peptides had significant CPP functionality (Figure 7c). Thus CPP functionality is a specific feature of the NRTN peptide sequence.

Determinants of the CPP activity of NRTN

To further address the influence of helical components and of adjacent amino acids on CPP functionality, various Nur–NRTN or NTRN–4EBP1 derivatives were generated and analysed. In these assays, reduction of viability indicates CPP-mediated delivery of the toxic fusion peptides. In contrast, lack of cytotoxicity reflects loss of CPP functionality. C-terminal deletions were assessed with Nur77-fusion peptides (Figure 8a) and N-terminal deletions were characterized with 4EBP1 fusions (Figure 8b). Deletions of both the N- and C-terminus were examined using 4EBP1 fusions (Figure 8c).

Sequence determinants of NRTN peptide functionality

Figure 8
Sequence determinants of NRTN peptide functionality

(a) MCF7 cells were incubated in the indicated concentrations of C-terminal deletion mutants of Nur–NRTN for 24 h. A grey box indicates the helical stretch. (b) MCF7 cells were incubated in the indicated concentrations of N-terminal deletion mutants of NRTN–4EBP1 for 24 h. A grey box indicates the helical sequence. (c) MCF7 cells were incubated in the indicated concentrations of both N- and C-terminal deletion mutants of NRTN–4EBP1 for 24 h. A grey box indicates the helical sequence. (d) MCF7 cells were incubated in the indicated concentration of Nur–NRTN without heparin (black squares, solid line), or in the presence of 5 μg/ml (black circles, solid line) or 50 μg/ml (white circles, broken line) heparin for 24 h. (e) MCF7 cells were incubated in the indicated concentration of Nur–NRTN with one (black circles, solid line), two (black squares, solid line) or three (white circles, broken line) alanine residues (bold and underlined) in the proposed proteoglycan binding motifs (*) for 24 h. In all cases, the cells were then analysed using a CellTiter-Glo® assay. Viability is expressed as the fold change over the medium control.

Figure 8
Sequence determinants of NRTN peptide functionality

(a) MCF7 cells were incubated in the indicated concentrations of C-terminal deletion mutants of Nur–NRTN for 24 h. A grey box indicates the helical stretch. (b) MCF7 cells were incubated in the indicated concentrations of N-terminal deletion mutants of NRTN–4EBP1 for 24 h. A grey box indicates the helical sequence. (c) MCF7 cells were incubated in the indicated concentrations of both N- and C-terminal deletion mutants of NRTN–4EBP1 for 24 h. A grey box indicates the helical sequence. (d) MCF7 cells were incubated in the indicated concentration of Nur–NRTN without heparin (black squares, solid line), or in the presence of 5 μg/ml (black circles, solid line) or 50 μg/ml (white circles, broken line) heparin for 24 h. (e) MCF7 cells were incubated in the indicated concentration of Nur–NRTN with one (black circles, solid line), two (black squares, solid line) or three (white circles, broken line) alanine residues (bold and underlined) in the proposed proteoglycan binding motifs (*) for 24 h. In all cases, the cells were then analysed using a CellTiter-Glo® assay. Viability is expressed as the fold change over the medium control.

The evaluation of the deletion variants revealed that the C-terminal arginine-rich sequence is required for the CPP function. Deletions of four or seven amino acids led to reduced activity, whereas deletions extending into the helical portion resulted in a complete loss of activity (Figure 8a). Complementary analyses of N-terminal deletions (Figure 8b) showed that up to 13 amino acids (only one arginine residue removed), even extending into the helical portion, are dispensable for CPP activity. The combination of both N- and C-terminal deletions (Figure 8c) showed a drastic reduction in the CPP activity, consistent with the effects observed upon deletions of the C-terminus. The helical portion alone did not show CPP activity when combined with the 4EBP1 peptide. These data confirm that the helical region by itself is not sufficient for CPP functionality. Flanking sequences, especially the cationic C-terminus, are also important.

Cationic CPPs bind sulfated proteoglycans [42,43], a feature that appears to be required for initial cell attachment. Neurturin requires binding to sulfated proteoglycans for signalling [44]. It has been proposed to interact with proteoglycans via the arginine-rich binding motif [45] which is contained within the NRTN peptide (Figure 7a, asterisks). To investigate the involvement of proteoglycan binding for the CPP function of the NRTN peptide, heparin was used to block proteoglycan binding. The addition of heparin interfered with the CPP-mediated cytotoxic activity of Nur–NRTN in a dose-dependent manner (Figure 8d), up to complete blockage at concentrations of 50 μg/ml. To further investigate proteoglycan binding, arginine residues in the proposed binding motifs were replaced by alanine residues (Figure 8e). Cell viability assays showed that the activity of Nur–NRTN was strongly reduced by introduction of these mutations (Figure 8e). Thus binding to sulfated proteoglycans appears to be a feature of the NRTN peptide which is important for its CPP functionality.

DISCUSSION

We applied an in silico procedure based on a library of all 30-mer peptides to identify CPPs in the human proteome. These peptides are larger than most CPPs that are commonly approximately 8–12 amino acids in length. The peptides used in the present study are derived from human proteins, meaning they have been selected by evolution to fold into structures. Therefore they have a higher probability to contain secondary structures than random peptides. Secondary structures have been shown to be advantageous for cell penetration [22]. A peptide of 30 amino acids can easily harbour an α-helix with the average length being approximately 12 amino acids [46]. Furthermore, the length of membrane-spanning helices is usually longer than 20 amino acids [47]. Peptides exceeding the minimum required length to span the lipid bilayer might therefore have superior membrane-penetrating properties.

Our screen identified 20 out of 60 human cationic extracellular peptides with either cytotoxicity and/or CPP functionality in cell-based assays. This functionality is not simply due to positive charges, as the 40 non-functional peptides were also cationic. Thus cell binding and uptake, as well as activity, appear to be sequence- and potentially structure-specific.

Most of the identified functional peptides were cytotoxic. Antimicrobial peptides (BPIL3 and FALL) also fell into this group. These peptides interfere with the membrane integrity of pathogens, and are toxic to mammalian cells at high concentrations. One explanation for the toxicity of these peptides might be that they are ‘too active’ CPPs that cause ‘holes’ in the plasma membrane. Plasma membrane repair mechanisms that might be involved in CPP uptake [48] could explain their partial functionality in siRNA transfection by secondary passive uptake of the nucleic acids. These peptides may be useful as toxic moieties for targeted tumour therapy. However, since our objective was the identification of peptides with CPP functionality at non-toxic doses, cytotoxic peptides were not further addressed.

The identification of three ‘functional’ peptides proves the applicability of our combined screen, in particular considering the fact that most control peptides fell into this category. These peptides showed evidence for CPP functionality (initially assayed via siRNA transfection) at non-toxic concentrations with some minor effects on cell viability at high concentrations exceeding 10 μM.

Further analyses then focused on the NRTN peptide which consistently showed the highest activity in siRNA transfection. This peptide not only transfects siRNA, it also shows other CPP features. This includes cell-surface binding, internalization and delivery of payloads into the cytoplasm of cells. Experiments aimed at identifying the features of the peptide that are responsible for its functionality point to an α-helix in combination with an arginine-rich stretch. Neurturin is closely related to GDNF and ARTN, both of which contain an α-helix in sequence sections that are homologous with the NRTN peptide. CD analyses confirmed the presence of a helical component in the NRTN peptide. The helix was observed after TFE addition, which leads to the formation of a more hydrophobic environment. Induction of helix formation upon interaction with the hydrophobic membrane might thus favour the membrane interaction/association of the peptide. On the basis of the primary sequence, the helix appears to be of both primary and secondary amphipathic nature. Amphipathic peptides were shown to bind to and disturb the integrity of model membranes at concentrations lower than only cationic ones [49]. Furthermore, they have been proposed to form transmembrane pores [50]. Therefore the presence of an amphipathic helix might help to explain the superior activity of the NRTN peptide. Deletion analyses revealed that the helix can be truncated without significant reduction of CPP activity. It has to be noted, however, that the NRTN–4EBP1 fusion peptide might harbour two helices, one in NRTN and the second within 4EBP1 (Supplementary Figure S8 at http://www.BiochemJ.org/bj/442/bj4420583add.htm). This second helix might compensate for the loss of helicity in the truncated NRTN sequence. On the other hand, deletions of the cationic sequences C-terminal of the helix gradually reduced the functionality of the peptide. This confirms that, in addition to helical composition, this arginine-rich sequence is important for CPP functionality. One contribution of the C-terminal sequence might be the interaction with sulfated proteoglycans, a feature that is relevant for the activity of cationic CPPs [42,43]. GDNF, ARTN and neurturin require sulfated proteoglycans for efficient signalling [51,52]. The proposed proteoglycan-binding motif of neurturin lies within the NRTN peptide's arginine-rich C-terminus [45]. The relevance of proteoglycan binding for the activity of the NRTN peptide is highlighted by the observation that it does not penetrate cells when it is either blocked by heparin competition or by the introduction of mutations into the binding site. Throughout the literature, there are three parameters that are associated with the activity of cationic CPPs: helicity, amphipathicity and the ability to bind to sulfated proteoglycans. The findings in the present study on the NRTN peptide are therefore in agreement with previous reports on the mechanism-of-action cationic CPPs.

Taken together, the results of the present study show that screening the human proteome with a filter as simplistic as 30% or more positively charged amino acids can be used to identify CPPs. A next step could be the incorporation of algorithms that are based on published CPP sequences [53] in a human proteome peptide library to potentially achieve a higher hit rate. The peptides that we have identified have a key advantage for application as part of therapeutically active biomolecules: as they are derived from human extracellular proteins, the immune system should be tolerant to them. They may therefore be less immunogenic than the pathogen-derived entities traditionally used. One potential disadvantage of our larger peptides with respect to their potential immunogenicity is the fact that smaller peptides can escape the presentation by MHC-II, which prefers peptides between 15 and 24 residues in length [54]. Therefore it will be useful to identify the minimal required sequences to avoid this potential pitfall.

Abbreviations

     
  • ARTN

    artemin

  •  
  • bDNA

    branched DNA

  •  
  • BPIL3

    BPI-fold-containing family B, member 6

  •  
  • CPP

    cell-penetrating peptide

  •  
  • DAPI

    4′,6-diamidino-2-phenylindole

  •  
  • Eg5

    kinesin-like protein KIF11

  •  
  • FBS

    fetal bovine serum

  •  
  • GAPDH

    glyceraldehyde-3-phosphate dehydrogenase

  •  
  • GDNF

    glial-derived neurotrophic factor

  •  
  • GO

    Gene Ontology

  •  
  • Hsp

    heat-shock protein

  •  
  • IEP

    isoelectric point

  •  
  • Nur77

    nuclear receptor subfamily 4 group A member 1 isoform 1

  •  
  • siRNA

    small interfering RNA

  •  
  • TFE

    trifluorethanol

AUTHOR CONTRIBUTION

Alexander Haas and Ulrich Brinkmann designed the research. Daniela Maisel did the bioinformatics. Alexander Haas, Juliane Adelmann, Christoffer von Schwerin and Ines Kahnt performed the research. Alexander Haas analysed the data. Alexander Haas, Juliane Adelmann and Ulrich Brinkmann wrote the paper.

FUNDING

This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

References

References
1
Davis
B. D.
On the importance of being ionized
Arch. Biochem. Biophys.
1958
, vol. 
78
 (pg. 
497
-
509
)
2
Kreil
G.
Transfer of proteins across membranes
Annu. Rev. Biochem.
1981
, vol. 
50
 (pg. 
317
-
348
)
3
Cook
A.
Bono
F.
Jinek
M.
Conti
E.
Structural biology of nucleocytoplasmic transport
Annu. Rev. Biochem.
2007
, vol. 
76
 (pg. 
647
-
671
)
4
Schroeder
A.
Levins
C. G.
Cortez
C.
Langer
R.
Anderson
D. G.
Lipid-based nanotherapeutics for siRNA delivery
J. Intern. Med.
2010
, vol. 
267
 (pg. 
9
-
21
)
5
Liu
Y. P.
Berkhout
B.
Lentiviral delivery of RNAi effectors against HIV-1
Curr. Top. Med. Chem.
2009
, vol. 
9
 (pg. 
1130
-
1143
)
6
Howard
K. A.
Delivery of RNA interference therapeutics using polycation-based nanoparticles
Adv. Drug Delivery Rev.
2009
, vol. 
61
 (pg. 
710
-
720
)
7
Fonseca
S. B.
Pereira
M. P.
Kelley
S. O.
Recent advances in the use of cell-penetrating peptides for medical and biological applications
Adv. Drug Delivery Rev.
2009
, vol. 
61
 (pg. 
953
-
964
)
8
Mäe
M.
Langel
Ü.
Cell-penetrating peptides as vectors for peptide, protein and oligonucleotide delivery
Curr. Opin. Pharmacol.
2006
, vol. 
6
 (pg. 
509
-
514
)
9
Frankel
A. D.
Pabo
C. O.
Cellular uptake of the tat protein from human immunodeficiency virus
Cell
1988
, vol. 
55
 (pg. 
1189
-
1193
)
10
Vives
E.
Brodin
P.
Lebleu
B.
A truncated HIV-1 Tat protein basic domain rapidly translocates through the plasma membrane and accumulates in the cell nucleus
J. Biol. Chem.
1997
, vol. 
272
 (pg. 
16010
-
16017
)
11
Futaki
S.
Suzuki
T.
Ohashi
W.
Yagami
T.
Tanaka
S.
Ueda
K.
Sugiura
Y.
Arginine-rich peptides. An abundant source of membrane-permeable peptides having potential as carriers for intracellular protein delivery
J. Biol. Chem.
2001
, vol. 
276
 (pg. 
5836
-
5840
)
12
Nascimento
F. D.
Hayashi
M. A.
Kerkis
A.
Oliveira
V.
Oliveira
E. B.
Radis-Baptista
G.
Nader
H. B.
Yamane
T.
Tersariol
I. L.
Kerkis
I.
Crotamine mediates gene delivery into cells through the binding to heparan sulfate proteoglycans
J. Biol. Chem.
2007
, vol. 
282
 (pg. 
21349
-
21360
)
13
Bettinger
T.
Carlisle
R. C.
Read
M. L.
Ogris
M.
Seymour
L. W.
Peptide-mediated RNA delivery: a novel approach for enhanced transfection of primary and post-mitotic cells
Nucleic Acids Res.
2001
, vol. 
29
 (pg. 
3882
-
3891
)
14
Walensky
L. D.
Kung
A. L.
Escher
I.
Malia
T. J.
Barbuto
S.
Wright
R. D.
Wagner
G.
Verdine
G. L.
Korsmeyer
S. J.
Activation of apoptosis in vivo by a hydrocarbon-stapled BH3 helix
Science
2004
, vol. 
305
 (pg. 
1466
-
1470
)
15
Oehlke
J.
Scheller
A.
Wiesner
B.
Krause
E.
Beyermann
M.
Klauschenz
E.
Melzig
M.
Bienert
M.
Cellular uptake of an α-helical amphipathic model peptide with the potential to deliver polar compounds into the cell interior non-endocytically
Biochim. Biophys. Acta
1998
, vol. 
1414
 (pg. 
127
-
139
)
16
Sandgren
S.
Cheng
F.
Belting
M.
Nuclear targeting of macromolecular polyanions by an HIV-Tat derived peptide. Role for cell-surface proteoglycans
J. Biol. Chem.
2002
, vol. 
277
 (pg. 
38877
-
38883
)
17
Fischer
R.
Kohler
K.
Fotin-Mleczek
M.
Brock
R.
A stepwise dissection of the intracellular fate of cationic cell-penetrating peptides
J. Biol. Chem.
2004
, vol. 
279
 (pg. 
12625
-
12635
)
18
Rothbard
J. B.
Jessop
T. C.
Wender
P. A.
Adaptive translocation: the role of hydrogen bonding and membrane potential in the uptake of guanidinium-rich transporters into cells
Adv. Drug Delivery Rev.
2005
, vol. 
57
 (pg. 
495
-
504
)
19
Duchardt
F.
Ruttekolk
I. R.
Verdurmen
W. P.
Lortat-Jacob
H.
Burck
J.
Hufnagel
H.
Fischer
R.
van den Heuvel
M.
Lowik
D. W.
Vuister
G. W.
, et al. 
A cell-penetrating peptide derived from human lactoferrin with conformation-dependent uptake efficiency
J. Biol. Chem.
2009
, vol. 
284
 (pg. 
36099
-
36108
)
20
Herve
F.
Ghinea
N.
Scherrmann
J. M.
CNS delivery via adsorptive transcytosis
AAPS J.
2008
, vol. 
10
 (pg. 
455
-
472
)
21
Trehin
R.
Merkle
H. P.
Chances and pitfalls of cell penetrating peptides for cellular drug delivery
Eur. J. Pharm. Biopharm.
2004
, vol. 
58
 (pg. 
209
-
223
)
22
Deshayes
S.
Plenat
T.
drian-Herrada
G.
Divita
G.
Le Grimellec
C.
Heitz
F.
Primary amphipathic cell-penetrating peptides: structural requirements and interactions with model membranes
Biochemistry
2004
, vol. 
43
 (pg. 
7698
-
7706
)
23
Ashburner
M.
Ball
C. A.
Blake
J. A.
Botstein
D.
Butler
H.
Cherry
J. M.
Davis
A. P.
Dolinski
K.
Dwight
S. S.
Eppig
J. T.
, et al. 
Gene ontology: tool for the unification of biology. The Gene Ontology Consortium
Nat. Genet.
2000
, vol. 
25
 (pg. 
25
-
29
)
24
Kim
W. J.
Christensen
L. V.
Jo
S.
Yockman
J. W.
Jeong
J. H.
Kim
Y. H.
Kim
S. W.
Cholesteryl oligoarginine delivering vascular endothelial growth factor siRNA effectively inhibits tumor growth in colon adenocarcinoma
Mol. Ther.
2006
, vol. 
14
 (pg. 
343
-
350
)
25
Song
E.
Zhu
P.
Lee
S. K.
Chowdhury
D.
Kussman
S.
Dykxhoorn
D. M.
Feng
Y.
Palliser
D.
Weiner
D. B.
Shankar
P.
, et al. 
Antibody mediated in vivo delivery of small interfering RNAs via cell-surface receptors
Nat. Biotechnol.
2005
, vol. 
23
 (pg. 
709
-
717
)
26
Lin
Y. Z.
Yao
S. Y.
Veach
R. A.
Torgerson
T. R.
Hawiger
J.
Inhibition of nuclear translocation of transcription factor NF-κB by a synthetic peptide containing a cell membrane-permeable motif and nuclear localization sequence
J. Biol. Chem.
1995
, vol. 
270
 (pg. 
14255
-
14258
)
27
Elbashir
S. M.
Harborth
J.
Lendeckel
W.
Yalcin
A.
Weber
K.
Tuschl
T.
Duplexes of 21-nucleotide RNAs mediate RNA interference in cultured mammalian cells
Nature
2001
, vol. 
411
 (pg. 
494
-
498
)
28
Fire
A.
Xu
S.
Montgomery
M. K.
Kostas
S. A.
Driver
S. E.
Mello
C. C.
Potent and specific genetic interference by double-stranded RNA in Caenorhabditis elegans
Nature
1998
, vol. 
391
 (pg. 
806
-
811
)
29
Crombez
L.
Charnet
A.
Morris
M. C.
drian-Herrada
G.
Heitz
F.
Divita
G.
A non-covalent peptide-based strategy for siRNA delivery
Biochem. Soc. Trans.
2007
, vol. 
35
 (pg. 
44
-
46
)
30
Meade
B. R.
Dowdy
S. F.
Enhancing the cellular uptake of siRNA duplexes following noncovalent packaging with protein transduction domain peptides
Adv. Drug Delivery Rev.
2008
, vol. 
60
 (pg. 
530
-
536
)
31
Panaretou
B.
Siligardi
G.
Meyer
P.
Maloney
A.
Sullivan
J. K.
Singh
S.
Millson
S. H.
Clarke
P. A.
Naaby-Hansen
S.
Stein
R.
, et al. 
Activation of the ATPase activity of hsp90 by the stress-regulated cochaperone aha1
Mol. Cell
2002
, vol. 
10
 (pg. 
1307
-
1318
)
32
Blangy
A.
Lane
H. A.
d'Herin
P.
Harper
M.
Kress
M.
Nigg
E. A.
Phosphorylation by p34cdc2 regulates spindle association of human Eg5, a kinesinrelated motor essential for bipolar spindle formation in vivo
Cell
1995
, vol. 
83
 (pg. 
1159
-
1169
)
33
Kolluri
S. K.
Zhu
X.
Zhou
X.
Lin
B.
Chen
Y.
Sun
K.
Tian
X.
Town
J.
Cao
X.
Lin
F.
, et al. 
A short Nur77-derived peptide converts Bcl-2 from a protector to a killer
Cancer Cell
2008
, vol. 
14
 (pg. 
285
-
298
)
34
Ko
S. Y.
Guo
H.
Barengo
N.
Naora
H.
Inhibition of ovarian cancer growth by a tumor-targeting peptide that binds eukaryotic translation initiation factor 4E
Clin. Cancer Res.
2009
, vol. 
15
 (pg. 
4336
-
4347
)
35
Herbert
T. P.
Fahraeus
R.
Prescott
A.
Lane
D. P.
Proud
C. G.
Rapid induction of apoptosis mediated by peptides that bind initiation factor eIF4E
Curr. Biol.
2000
, vol. 
10
 (pg. 
793
-
796
)
36
Niidome
T.
Takaji
K.
Urakawa
M.
Ohmori
N.
Wada
A.
Hirayama
T.
Aoyagi
H.
Chain length of cationic alpha-helical peptide sufficient for gene delivery into cells
Bioconjugate Chem.
1999
, vol. 
10
 (pg. 
773
-
780
)
37
Kotzbauer
P. T.
Lampe
P. A.
Heuckeroth
R. O.
Golden
J. P.
Creedon
D. J.
Johnson
E. M.
Jr
Milbrandt
J.
Neurturin, a relative of glial-cell-line-derived neurotrophic factor
Nature
1996
, vol. 
384
 (pg. 
467
-
470
)
38
Eigenbrot
C.
Gerber
N.
X-ray structure of glial cell-derived neurotrophic factor at 1.9 Å resolution and implications for receptor binding
Nat. Struct. Biol.
1997
, vol. 
4
 (pg. 
435
-
438
)
39
Wang
X.
Baloh
R. H.
Milbrandt
J.
Garcia
K. C.
Structure of artemin complexed with its receptor GFRα3: convergent recognition of glial cell line-derived neurotrophic factors
Structure
2006
, vol. 
14
 (pg. 
1083
-
1092
)
40
Buck
M.
Trifluoroethanol and colleagues: cosolvents come of age. Recent studies with peptides and proteins
Q. Rev. Biophys.
1998
, vol. 
31
 (pg. 
297
-
355
)
41
Agerberth
B.
Gunne
H.
Odeberg
J.
Kogner
P.
Boman
H. G.
Gudmundsson
G. H.
FALL-39, a putative human peptide antibiotic, is cysteine-free and expressed in bone marrow and testis
Proc. Natl. Acad. Sci. U.S.A.
1995
, vol. 
92
 (pg. 
195
-
199
)
42
Duchardt
F.
Fotin-Mleczek
M.
Schwarz
H.
Fischer
R.
Brock
R.
A comprehensive model for the cellular uptake of cationic cell-penetrating peptides
Traffic
2007
, vol. 
8
 (pg. 
848
-
866
)
43
Gump
J. M.
June
R. K.
Dowdy
S. F.
Revised role of glycosaminoglycans in TAT protein transduction domain-mediated cellular transduction
J. Biol. Chem.
2010
, vol. 
285
 (pg. 
1500
-
1507
)
44
Bespalov
M. M.
Sidorova
Y. A.
Tumova
S.
Ahonen-Bishopp
A.
Magalhaes
A. C.
Kulesskiy
E.
Paveliev
M.
Rivera
C.
Rauvala
H.
Saarma
M.
Heparan sulfate proteoglycan syndecan-3 is a novel receptor for GDNF, neurturin, and artemin
J. Cell Biol.
2011
, vol. 
192
 (pg. 
153
-
169
)
45
Silvian
L.
Jin
P.
Carmillo
P.
Boriack-Sjodin
P. A.
Pelletier
C.
Rushe
M.
Gong
B.
Sah
D.
Pepinsky
B.
Rossomando
A.
Artemin crystal structure reveals insights into heparan sulfate binding
Biochemistry
2006
, vol. 
45
 (pg. 
6801
-
6812
)
46
Barlow
D. J.
Thornton
J. M.
Helix geometry in proteins
J. Mol. Biol.
1988
, vol. 
201
 (pg. 
601
-
619
)
47
Hildebrand
P. W.
Preissner
R.
Frommel
C.
Structural features of transmembrane helices
FEBS Lett.
2004
, vol. 
559
 (pg. 
145
-
151
)
48
Palm-Apergi
C.
Lorents
A.
Padari
K.
Pooga
M.
Hallbrink
M.
The membrane repair response masks membrane disturbances caused by cell-penetrating peptide uptake
FASEB J.
2009
, vol. 
23
 (pg. 
214
-
223
)
49
Sauder
R.
Seelig
J.
Ziegler
A.
Thermodynamics of lipid interactions with cell-penetrating peptides
Methods Mol. Biol.
2011
, vol. 
683
 (pg. 
129
-
155
)
50
Deshayes
S.
Plenat
T.
Charnet
P.
Divita
G.
Molle
G.
Heitz
F.
Formation of transmembrane ionic channels of primary amphipathic cell-penetrating peptides. Consequences on the mechanism of cell penetration
Biochim. Biophys. Acta
2006
, vol. 
1758
 (pg. 
1846
-
1851
)
51
Barnett
M. W.
Fisher
C. E.
Perona-Wright
G.
Davies
J. A.
Signalling by glial cell line-derived neurotrophic factor (GDNF) requires heparan sulphate glycosaminoglycan
J. Cell Sci.
2002
, vol. 
115
 (pg. 
4495
-
4503
)
52
Tanaka
M.
Xiao
H.
Kiuchi
K.
Heparin facilitates glial cell line-derived neurotrophic factor signal transduction
NeuroReport
2002
, vol. 
13
 (pg. 
1913
-
1916
)
53
Hansen
M.
Kilk
K.
Langel
Ü.
Predicting cell-penetrating peptides
Adv. Drug Delivery Rev.
2008
, vol. 
60
 (pg. 
572
-
579
)
54
Fairchild
P. J.
Presentation of antigenic peptides by products of the major histocompatibility complex
J. Pept. Sci.
1998
, vol. 
4
 (pg. 
182
-
194
)

Supplementary data