Biochemical Society Transactions

Cell-Penetrating Peptide

Cell-penetrating-peptide-based delivery of oligonucleotides: an overview

R. Abes, A.A. Arzumanov, H.M. Moulton, S. Abes, G.D. Ivanova, P.L. Iversen, M.J. Gait, B. Lebleu


Cationic CPPs (cell-penetrating peptides) have been used largely for intracellular delivery of low-molecular-mass drugs, biomolecules and particles. Most cationic CPPs bind to cell-associated glycosaminoglycans and are internalized by endocytosis, although the detailed mechanisms involved remain controversial. Sequestration and degradation in endocytic vesicles severely limits the efficiency of cytoplasmic and/or nuclear delivery of CPP-conjugated material. Re-routing the splicing machinery by using steric-block ON (oligonucleotide) analogues, such as PNAs (peptide nucleic acids) or PMOs (phosphorodiamidate morpholino oligomers), has consequently been inefficient when ONs are conjugated with standard CPPs such as Tat (transactivator of transcription), R9 (nona-arginine), K8 (octalysine) or penetratin in the absence of endosomolytic agents. New arginine-rich CPPs such as (R-Ahx-R)4 (6-aminohexanoic acid-spaced oligo-arginine) or R6 (hexa-arginine)–penetratin conjugated to PMO or PNA resulted in efficient splicing correction at non-cytotoxic doses in the absence of chloroquine. SAR (structure–activity relationship) analyses are underway to optimize these peptide delivery vectors and to understand their mechanisms of cellular internalization.

  • cell-penetrating peptide (CPP)
  • endocytosis
  • intracellular delivery
  • oligonucleotide
  • 6-aminohexanoic acid spaced oligo-arginine [(R-Ahx-R)4]


Rationally designed RNaseH-competent antisense ONs (oligonucleotides), and more recently siRNAs (short interfering RNAs), allow specific and efficient down-regulation of gene expression. Chemical modifications have been proposed, giving rise to improved pharmacological properties (in terms of metabolic stability in particular) while maintaining specific hybridization to the targeted RNA sequence and allowing recruitment of RNaseH (for antisense ONs) or of the RISC (RNA-induced silencing complex)-associated nuclease (for siRNA) [1,2]. Transfection of these negatively charged ON analogues is most often achieved in vitro through conjugation with cationic delivery vehicles such as commercially available lipoplex formulations [3]. RNaseH-recruiting ONs and siRNAs have rapidly become precious and widely used tools for functional genomics analysis. Their development as a new class of drugs, allowing the specific down-regulation of RNA transcripts overexpressed or modified (through mutation or translocation) in human pathologies, has led to pre-clinical studies in animal models and has been the subject of several clinical trials [4]. Most clinical data to date have been rather disappointing, since naked ON analogues have poor access to intracellular RNA targets and currently available non-viral delivery vectors are both cytotoxic and poorly efficient in the presence of serum proteins [5].

Antisense ONs are also able to interfere with gene expression through a steric-block mechanism, upon which the binding of RNaseH-incompetent ON analogues to the RNA target masks the recognition site of a protein (involved in translation initation, in splicing or in viral transactivation) or of a regulatory RNA, such as miRNA (microRNA) [68]. Steric-block ON-based strategies have not yet received much interest, since a lower efficiency is expected than with the ON-based strategies harnessing degradation of the RNA target. However, they allow the use of the various ON analogues which do not permit RNaseH recruitment, such as LNAs (locked nucleic acids), PNAs (peptide nucleic acids) or PMOs (phosphorodiamidate morpholino oligomers). Another advantage of steric-block ONs is their established specificity compared with ONs that use an RNaseH-based mechanism. Finally, new applications aimed at redirecting the splicing machinery cannot be implemented by alternative strategies. Examples include correction of aberrant splicing caused by mutations (observed in some β-thalassaemias) [7], skipping of mutated exons (seen in Duchenne muscular dystrophy) [9], or regulation of the balance between alternative splicing variants (as observed in many cancers overexpressing the anti-apoptotic version of Bcl-x) [10].

CPP (cell-penetrating peptide)-based delivery of steric-block ONs is limited by endosome trapping

CPPs, and Tat (transactivator of transcription) (48–60) protein transduction domain in particular, have been used largely for delivery of a vast collection of biomolecules. Surprisingly, very few papers have described the successful delivery of antisense ONs, and our own attempts in this field have been disappointing [11]. In recent studies, we have made use of a splicing correction assay in which a mutated intron carrying an aberrant splice site has been inserted into the coding sequence of a reporter [luciferase or EGFP (enhanced green fluorescent protein)] gene [12]. The incomplete removal of this intron prevents the production of functional correctly spliced mRNA and, as a consequence, the expression of functional luciferase or EGFP proteins [6]. The intronic hybridization of the correcting steric-block ON re-orients the splicing machinery, which results in removal of the mutated intron, production of fully spliced mRNA, as monitored by RT (reverse transcription)–PCR, and expression of the reporter gene, as monitored by luciferase or EGFP assays. This functional assay is advantageous in providing a positive read-out over a low background and is increasingly used by researchers interested in ON delivery strategies.

Our recent studies (largely corroborated by other groups using the same assay) have indicated that CPP–PNA or –PMO conjugates, including Tat, penetratin, (Arg)n or (Lys)n conjugates, did not lead to significant splicing correction at low concentrations, although FACS analysis clearly indicated that cell uptake was occuring [6,8]. Fluorescence microscopy also demonstrated that none of the CPP-transported PNA or PMO could be found outside of endocytic vesicles, although a low level of endosomal release would have escaped detection [6,7]. Similarly, in another test of nuclear delivery, the same range of CPPs, when conjugated to a 16-mer PNA targeted to the HIV-1 TAR (transactivation-response region) RNA, were unable to show activity in a HeLa cell luciferase-reporter assay of Tat-dependent transactivation. Only two CPPs, transportan and R6 (hexa-arginine)-penetratin, which were disulfide-conjugated to PNA, showed some inhibition 24 h after delivery [8]. Co-administration of chloroquine, an endosomolytic agent, enabled inhibition to be observed only 6 h after delivery and, in some cases, caused inactive CPP–PNA conjugates to become active [6,8]. The most thorough explanation for all these results is that CPP–ON steric-block conjugates are internalized through an endocytic mechanism [13,14], but escape poorly from endocytic vesicles (Figure 1).

Figure 1 Splice-correcting conjugate internalization: endocytosis is the major mechanism used by CPPs to enter cells

HSPG (heparan sulfate proteoglycan) binding to CPPs is followed by endocytosis, and sequestration in vesicular compartments occurs. A proportion of the HSPG-bound CPPs escapes to reach cytosolic and nuclear compartments, which are involved in CPP internalization. E, endosome; LE, late endosome; LY, lysosome.

In keeping with this hypothesis, several drugs or experimental conditions known to foster endosome destabilization, such as chloroquine or high sucrose concentration, did significantly increase the biological activity of our CPP–ON conjugates in the splice-correction assay [6]. Although endosomolytic drugs cannot be envisaged as tools to ensure the nuclear or cytosolic delivery of CPP-conjugated material in vivo, they provide an additional argument in favour of endocytic mechanisms being involved in CPP–ON internalization. It is also worth noting that increased splicing correction was achieved when using an experimental protocol in which CPP–ON treatment and chloroquine treatment were performed sequentially. However, seemingly well-controlled experiments are still reported in other publications, which suggests that endocytic-independent mechanisms are also involved in the internalization of CPPs and CPP–drug conjugates [1517]. These discrepancies can be explained by subtle differences in experimental conditions, leading to various levels of membrane destabilization. As an example, we found out that a PNA–(Lys)n conjugate did promote splicing correction at 5–10 μM concentrations, in keeping with previously reported data [6]. However, under these conditions, membrane destabilization (as monitored by a sensitive propidium iodide FACS uptake assay) occurred and uptake of the conjugates became energy-independent.

(R-Ahx-R)4 (6-aminohexanoic acid-spaced oligo-arginine) and R6–penetratin are efficient delivery vectors for steric-block ON

Our goal is to modify CPPs in order to improve endosomal escape (or eventually following of another internalization pathway) and to allow efficient splice correction at a low dosage. It should be noted that our functional assays are rather demanding, since they require the delivery of steric-block ON acting in stoichiometric amounts. Assays of enzymatic proteins (such as Cre recombinase) or of RNaseH-recruiting ON obviously require the delivery of lower concentrations of material.

A first lead has been provided by SAR (structure–activity relationship) studies on arginine-rich CPPs cellular uptake by J. Rothbard and co-workers [1820]. Molecular modelling revealed that all guanidinium side chains of arginine residues (known to play a key role in basic CPPs) do not point in the same direction and are therefore not all available for membrane binding [1820]. Membrane-associated receptors could be negatively charged lipid, as initially thought, or heparan sulfate chains of membrane glycosaminoglycans, as revealed in more recent studies [21]. If this is the case, not all guanidinium groups would be required, and some arginine residues could be replaced by other amino acids, including non-natural ones. This strategy would be advantageous in decreasing CPP cytotoxicity (which is largely due to the density of cationic charges) and in increasing metabolic stability [22]. The studies by Rothbard and co-workers had pointed to (R-Ahx-R)4 as the most efficient CPP analogue in terms of cellular uptake, but no data concerning efficiency in terms of delivery of conjugated cargo were reported [1820]. In line with these studies, conjugates of (R-Ahx-R)4 to PMO or PNA did lead to significant and sequence-specific splice correction at low concentrations in the absence of endosomolytic agents. As illustrated in Figure 2, the incubation of HeLa pLuc705 cells with (R-Ahx-R)4–PMO conjugates for 4 h gave rise to a dosedependent increase in luciferase activity with a plateau around 2.5 μM. In keeping with these observations, RT–PCR analysis allows the detection of the correctly spliced transcript at low concentrations with an EC50 value in the micromolar range. The exquisite sequence selectivity of the correction is confirmed in both assays by the use of a scrambled version of the correcting steric-block ON (Figure 2).

Figure 2 Sequence-specific splicing correction by (R-Ahx-R)4–PMO conjugates

(A) Luciferase activity assay. HeLa pLuc 705 cells were incubated for 4 h in OptiMEM® in the absence (control) or in the presence of (R-Ahx-R)4–PMO or (R-Ahx-R)4–PMO sc (scrambled conjugate) at the indicated concentrations. Luciferase expression was quantified 20 h later and expressed as RLU (relative luciferase units)/μg of protein. Each experiment was performed in triplicate and results are means±S.E.M. (B) RT–PCR detection of the correctly spliced transcript. HeLa pLuc 705 cells were untreated (lane 1), incubated for 4 h with (R-Ahx-R)4–PMO conjugate (lanes 3–5), or with (R-Ahx-R)4–PMO sc (lanes 6–8) at the indicated concentrations. The upper band corresponds to the aberrantly spliced luciferase mRNA and the lower band to the correctly spliced mRNA.

Similar data have been obtained with (R-Ahx-R)4–PNA conjugates [11]. Importantly for potential clinical applications (as reviewed by Moulton et al. [23] on pp. 826–828 of this issue of Biochemical Society Transactions), these (R-Ahx-R)4–ON conjugates were not cytotoxic at concentrations up to 50 μM in in vitro studies.

These (R-Ahx-R)4 conjugates are internalized by an energy-dependent mechanism (Figure 3), but no clear explanation for their increased efficiency in splice correction can be provided. Affinity for heparan sulfate should be a key element for their efficiency and SAR studies are currently underway to validate this hypothesis. Indeed, affinity for heparan sulfate might be sufficient to provide binding to cell-surface proteoglycans, but should not be too high in order to allow dissociation after endocytosis.

Figure 3 (R-Ahx-R)4–PMO conjugates are internalized by an energy–dependent mechanism

(A) Effect of low temperature. Cells were pre-incubated in OptiMEM® at either 4°C or 37°C for 30 min and then incubated for 2 h with the PMO conjugates (at the concentrations indicated) at either 4°C or 37°C. Cells were then washed and the incubation continued for 20 h in DMEM (Dulbecco's modified Eagle's medium). (B) Effect of ATP depletion. Cells were pre-incubated at 37°C for 30 min in OptiMEM® supplemented with 10 mM sodium azide and 6 mM 2-deoxy-D-glucose. Cells were then incubated with the PMO conjugates (at the concentrations indicated) for 2 h under the same conditions. Cells were washed, and the incubation was continued for 20 h in DMEM. Luciferase expression was quantified 20 h later and expressed as RLU (relative luciferase units)/μg of protein. Each experiment was performed in triplicate and results are means±S.E.M.

A second lead CPP has come from our previously reported studies of HIV-1 Tat-dependent transactivation [8]. We found recently that both thioether-linked and disulfide-linked conjugates of 18-mer PNA705 with R6–penetratin have high splice-correction activity (EC50 1.0±0.3 and 0.7±0.3 μM respectively) when assayed in HeLa cells, as judged by both luciferase up-regulation and RT–PCR [24].

Preliminary SAR studies have shown that splice-correction activity is higher for R6–penetratin compared with R3–penetratin or R9–penetratin–PNA conjugates. Furthermore, a Trp→Leu mutation in the hydrophobic domain of the penetratin sequence gave rise to enhanced splice-correction activity, even though the same mutation in the penetratin peptide alone had previously been shown to abolish translocation of the membrane [25]. The cationic–hydrophobic–cationic domain nature of the peptide sequence in R6–penetratin seems to be a key feature in this new CPP paradigm. Studies are in progress to find more active sequences that maintain such a domain structure, while also introducing improved serum stability that is necessary for its in vivo use.

Conclusions and perspectives

We aimed to discover CPPs that allow the nuclear (or cytoplasmic) delivery of ONs in a controlled assay that monitors sequence-specific biological responses accurately. We also aimed to achieve this goal at low concentrations (≤1 μM) to avoid membrane destabilization and asociated cytotoxicity, and to remain within reasonable costs in the perspective of further clinical applications.

Two lead CPPs have been designed, namely (R-Ahx-R)4 and R6–penetratin. PNA or PMO conjugation allows sequence-specific splicing correction with an EC50 in the micromolar to sub-micromolar range. Intracellular distribution studies using fluorescence microscopy have revealed that most conjugated PNA or PMO material was still trapped in cytoplasmic vesicles, even under conditions which allow the complete restoration of normal splicing. Accordingly, chloroquine treatment improved the efficiency of splicing correction (R. Abes and P. Prevot, unpublished work). SAR studies of both CPPs are ongoing, in order to identify analogues which have improved splicing-correction efficiency and the capacity to be released from endosomes effectively. We are also exploring other potential steric-block applications of CPP–PNAs, such as inhibition of miRNA activity.

Several in vivo models have recently been used to determine the efficacy of (R-Ahx-R)4–PMO conjugates including treatment of Duchenne muscular dystrophy [9] and also in inhibition of coronavirus replication in mice [22]. The first clinical trial using CPP–ON technology has started in Europe, a safety and efficacy study of (R-Ahx-R)4 conjugated to a PMO targeted to human c-Myc (AVI-5126) to prevent blockage of a transplanted vein after cardiovascular bypass surgery. After the vein is excised, it is immersed in a solution containing 10 μM AVI-5126 and then implanted as a bypass graft.


The research in our groups has been funded by EC FP5 (European Commission Framework programme 5) (B.L. and M.J.G.) and by IFCPAR (Indo–French Association for the Promotion of Advanced Research) (B.L.). S.A. is supported by the Ligue Regionale contre le Cancer and R.A. by Region Languedoc Roussillon. We thank Donna Williams and David Owen for synthetic PNA and peptides, Paul Prevot for fluorescence microscopy, and Philippe Clair and Martin Fabani for discussions.


  • Cell-Penetrating Peptides: A Biochemical Society Focused Meeting held at University of Wolverhampton, Telford, U.K., 9–11 May 2007. Organized and Edited by B. Austen (St. George's, University of London, U.K.), J. Howl (Wolverhampton, U.K.), S. Jones (Wolverhampton, U.K.) and B. Lebleu (Montpellier, France).

Abbreviations: CPP, cell-penetrating peptide; EGFP, enhanced green fluorescent protein; miRNA, microRNA; ON, oligonucleotide; PMO, phosphorodiamidate morpholino oligomer; PNA, peptide nucleic acid; R6, hexa-arginine; (R-Ahx-R)4, 6-aminohexanoic acid-spaced oligo-arginine; RT, reverse transcription; SAR, structure–activity relationship; siRNA, short interfeing RNA; Tat, transactivator of transcription


View Abstract