Insight into the mechanism of the peptide-based gene delivery system MPG: implications for delivery of siRNA into mammalian cells

The improvement of non-viral-based gene delivery systems is of prime importance for the future of gene and antisense therapies. We have previously described a peptide-based gene delivery system, MPG, derived from the fusion peptide domain of HIV-1 gp41 protein and the nuclear localisation sequence (NLS) of SV40 large T antigen. MPG forms stable non-covalent complexes with nucleic acids and improves their delivery. In the present work, we have investigated the mechanism through which MPG promotes gene delivery. We demonstrate that cell entry is independent of the endosomal pathway and that the NLS of MPG is involved in both electrostatic interactions with DNA and nuclear targeting. MPG/DNA particles interact with the nuclear import machinery, however, a mutation which affects the NLS of MPG disrupts these interactions and pre-vents nuclear delivery of DNA. Nevertheless, we show that this mutation yields a variant of MPG which is a powerful tool for delivery of siRNA into mammalian cells, enabling rapid release of the siRNA into the cytoplasm and promoting robust down-regulation of target mRNA. Taken together, these results support the potential of MPG-like peptides for therapeutic applications and suggest that speciﬁc variations in the sequence may yield carriers with distinct targeting features.


INTRODUCTION
The development of non-viral-based gene delivery systems constitutes an essential challenge in therapeutics. Although they exhibit several advantages over viral systems, the interest of non-viral synthetic gene delivery systems for therapeutic applications remains limited by their poor ability to escape from the endosomal compartment and to translocate DNA into the nucleus (1). During the past 10 years, several peptidebased gene delivery systems that can overcome both extracellular and intracellular limitations have been proposed (2±4). Poor release from the endosomal compartment after cellular uptake constitutes one of the major limitations of nonviral gene delivery systems. Research has therefore been focused on the design of strategies to either facilitate release from the endosome or to bypass the endosomal pathway (2±4). Peptide carriers that combine DNA binding and membrane destabilising properties have been demonstrated to promote gene transfer into cultured cells (5,6) and living animals (7).
Another major limitation of non-viral gene delivery systems is the poor nuclear delivery of DNA, which is however essential for transfection of non-dividing cells as well as for in vivo applications. In order to improve nuclear delivery of DNA, synthetic peptides containing nuclear localisation sequences (NLS) have been extensively used (8,9). Protein transduction domains cross the cell membrane independently of the endosomal pathway and have been used to improve the delivery of DNA (10,11). The third helix of the Antennapedia homeodomain has been shown to form stable non-covalent complexes with small oligonucleotides and to facilitate their internalisation in a non-endosomal fashion (12). The NLS domain of Tat has been shown to promote nuclear targeting of proteins and DNA (10,11,13). Tat peptide covalently attached to liposomes promotes rapid delivery of DNA, independently of the endosomal pathway (14,15). In contrast, oligomers of the arginine-rich motif of HIV-1 Tat protein have been reported to form stable particles with DNA through noncovalent interactions, but promote their delivery into cells through the endosomal pathway (16).
We have recently described a new peptide-based gene delivery system, MPG, which forms stable non-covalent complexes with nucleic acids and promotes their delivery into a large panel of cell lines (17±19). MPG is a bipartite amphipathic peptide derived from both the fusion peptide domain of HIV-1 gp41 protein and the NLS of SV40 large T antigen. In the present work, we have investigated the mechanism through which MPG promotes gene delivery into cells and have demonstrated that it is independent of the endosomal pathway. We show that the NLS of MPG is required for both electrostatic interactions with DNA and nuclear targeting, and that MPG/DNA particles interact with the nuclear import machinery. Moreover we have used MPG and a variant characterised by a mutation in the NLS to deliver siRNA into mammalian cells. Our results demonstrate that delivery of siRNA by both MPG and its NLS mutant enable robust down-regulation of target mRNA. Nevertheless, the NLS mutant induces rapid release of the siRNA into the cytoplasm, and this correlates with a more signi®cant biological response. Taken together, these results support the potential of MPG for therapeutic applications.

Peptide synthesis and analysis
All peptides were synthesised by solid phase peptide synthesis using AEDI-expansin resin with a 9050 Pep Synthesizer (Millipore, Watford, UK) according to the Fmoc/tBoc method and puri®ed as described previously (18). MPG and MPGD NLS were acetylated at their N-terminus and synthesised with a cysteamide group at their C-terminus. MPG, GALFLGFL-GAAGSTMGAWSQPKKKRKV; MPGD NLS , GALFLGFLG-AAGSTMGAWSQPKSKRKV.
Cell culture and MPG-mediated transfection HS-68, Cos-7 and HeLa cell lines were cultured in Dulbecco's modi®ed Eagle's medium (DMEM) supplemented with 2 mM glutamine, 1% antibiotics (10 000 mg/ml streptomycin, 10 000 IU/ml penicillin) and 10% (w/v) foetal calf serum (FCS) at 37°C in a humidi®ed atmosphere containing 5% CO 2 as described previously (18). For MPG-or MPGD NLSmediated gene delivery, peptide carrier/DNA complexes were formed in DMEM or phosphate-buffered saline (500 ml of DMEM containing 100 ng of DNA complexed with MPG at a charge ratio of 5:1) and incubated for 30 min at 37°C. Cells grown to 60% con¯uence were then overlaid with these preformed complexes. After 30 min incubation at 37°C, 1 ml of fresh DMEM supplemented with 10% FCS was added to the cells, without removing the overlay of carrier/DNA, and cells were returned to the incubator. For cell cycle-dependent studies, HS-68 ®broblasts were synchronised by serum deprivation for 40 h, then restimulated to enter the cycle and grow into early G 1 for 4 h by addition of fresh DMEM supplemented with 20% FCS, and incubated in the presence of MPG/DNA complexes as described above. Luciferase activity was monitored as described previously (19) and control transfection experiments were performed using LipofectamineÔ. When transfected at 4°C, cells were incubated for 1 h at 4°C prior to transfection, then incubated with MPG/DNA complexes for another 2 h at 4°C, extensively washed and ®nally returned to incubation at 37°C in fresh medium supplemented with serum. When transfections were performed in the presence of ba®lomycin A (175 nM), cytochalasin B (50 mg/ml) or chloroquine (100 mM), cells were preincubated with these inhibitors of endocytosis for 1 h prior to transfection. Transfections were performed for 1 h, after which cells were extensively washed and either analysed by¯uorescence microscopy in order to determine the cellular localisation of the¯uorescently labelled oligonucleotide or returned to the incubator at 37°C in the presence of fresh medium supplemented with serum.

MPG-mediated siRNA delivery
siRNA targeting the 3¢-untranslated region of GAPDH were from the SilencerÔ GAPDH siRNA kit (Ambion). Fluorescent labelling of siRNA was performed using either the Fam or the Cy3 SilencerÔ labeling kit (Ambion) and modi®ed as described in the manufacturer's protocol. The pRL-Luc reporter gene was from Promega and siRNA targeting luciferase sense (5¢-CUUACGCUGAGUACUUCGATT-3¢) and antisense (3¢-TTGAAUGCGACUCAUGAAGCU-5¢) and mismatch sense (5¢-CGUACGCGGAAUACUUCGATT-3¢) and antisense (3¢-TTGCAUGCGCCUUAUGAAGCU-5¢) siRNAs were obtained from Genset Oligos. For siRNA delivery, peptide carrier/siRNA complexes were formed in DMEM at a charge ratio of 10:1 and incubated for 30 min at 37°C. Cells grown to 60% con¯uence were then overlaid with these preformed complexes. After 30 min incubation at 37°C, 1 ml of fresh DMEM supplemented with 10% FCS was added to the cells, without removing the overlay of carrier/siRNA.

Northern blotting
HS-68 or HeLa cells (5 Q 10 6 cells) incubated with complexes were collected and total RNAs were isolated from cells using TriReagentÔ (Sigma, St Louis, MO) according to the manufacturer's recommendations. RNAs were then puri®ed by phenol extraction followed by ethanol precipitation. RNA samples (10 mg) were separated by electrophoresis in formaldehyde agarose gels (1.2%), transferred onto nylon membranes (Hybond N + ; Amersham Pharmacia Biotech) and hybridized with 32 P-labelled GAPDH and actin probes (the latter used to normalise RNA loading). Signals were detected by Phosphorimaging (Molecular Dynamics) and quanti®ed using ImageQuant software.

Protein preparation
Plasmids for the expression of recombinant GST±importin a and GST±importin b fusion proteins were kindly provided by B. Cullen. Proteins were puri®ed as described previously (13). For af®nity chromatography experiments, GST±importin a and GST±importin b were coupled onto GST beads in 150 mM NaCl, 1 mM EDTA and 10 mM HEPES pH 7.5. Importincoupled beads were then incubated in batch with MPG/DNA complexes for 1 h, extensively washed and incubated with 50 nM glutathione to elute the complexes. DNA was then extracted from the complexes with phenol/chloroform, followed by ethanol precipitation, and ®nally analysed by agarose gel electrophoresis (0.8%).

MPG-mediated gene delivery is independent of the endosomal pathway
Transport of DNA into cells involves a number of selective steps, therefore understanding the mechanisms underlying each of these steps as well as the behaviour of carrier/DNA complexes within the cell is essential to develop more ef®cient carriers. We have already reported that MPG can deliver oligonucleotides into mammalian cells at low temperature with high ef®ciency (17). In order to understand the mechanism through which MPG mediates gene delivery, transfection experiments were performed in the presence of several inhibitors that interfere with the endosomal pathway, including cytochalasin B, ba®lomycin A and chloroquine. Cytochalasin B induces depolymerisation of the micro®laments involved in phagocytosis and macropinocytosis without affecting other endocytotic processes (20). Ba®lomycin A speci®cally prevents the acidi®cation of early endosomes, by inhibiting a proton pump know as vacuolar ATPase (21), and chloroquine is a weak base, which inhibits the maturation of the transport vesicles into late endosomes and neutralises the pH of the latter (22).
A plasmid encoding the reporter gene luciferase, pRL-SV40, was associated with the MPG carrier at a charge ratio of 5:1 (MPG/DNA) and overlaid onto cultured human ®broblasts (HS-68) or HeLa cells in the presence of either ba®lomycin A (150 nM), chloroquine (100 mM) or cytochalasin B (50 mg/ml). Transfection ef®ciency was monitored 30 h after transfection and LipofectamineÔ was used as a standard control method for transfection. As shown in Figure 1, the transfection ef®ciency of MPG was not affected by ba®lomycin A or cytochalasin B, suggesting that the cellular uptake of MPG/ DNA complexes is independent of the classical endosomal pathway. In contrast, the ef®ciency of the lipid-based carrier was dramatically reduced by these inhibitors of the endosomal pathway. Chloroquine barely increased the ef®ciency of MPG (by only 10%), whereas it enhanced delivery by LipofectamineÔ more than 20-fold, as already reported for different liposome-based gene delivery systems (23,24). When transfection experiments were performed at 4°C, the ef®ciency of MPG was reduced 12-fold, compared to that of LipofectamineÔ by more than 1000-fold. Based on these results, we believe that the decrease in the ef®ciency of MPG is associated with cellular stress induced by low temperature. To exclude the possibility that the MPG/DNA complexes remain associated with the cell membrane in the presence of inhibitors and only enter the cell upon dilution of the inhibitor, transfection experiments were performed in the presence of inhibitors, using a¯uorescently labelled oligonucleotide (17). As reported in Figure 1B, after 1 h most of the¯uorescently labelled oligonucleotides localised to the nucleus and cytoplasm of the cells, suggesting that MPG/DNA complexes are not associated with the membrane and indeed enter the cells in the presence of inhibitors. Interestingly, a small fraction of these complexes are associated with the membrane, which may correspond to large aggregates unable to enter the cells. Taken together, these data suggest that the cellular uptake mechanism of MPG/DNA complexes is essentially independent of the endosomal pathway.
Role of the NLS of MPG in nuclear targeting of plasmids MPG peptide is derived from the NLS of the SV40 large T antigen and from the fusion peptide domain of HIV-1 gp41 protein (17). We have demonstrated that lysine residues are required for the formation of stable non-covalent interactions with DNA. We have investigated to what extent the NLS sequence was involved in nuclear targeting of plasmids. Nuclear import of proteins is generally mediated by the interaction of the NLS with importin, which in turn favours the interaction with importin b (25,26). Two peptides, wild-type MPG and MPGD NLS , a variant characterised by a single mutation of the second lysine residue in the NLS motif to serine (KSKRKV), were used. This mutation has previously been reported to dramatically reduce import of NLS-containing proteins into the nucleus (25,26). To determine to what extent the NLS moiety of MPG was required for nuclear import of plasmids, we ®rst performed in vitro binding experiments using GST±importin fusion proteins. Importin a and b immobilized on beads were incubated with MPG/DNA or MPGD NLS /DNA complexes for 1 h at 25°C. Then, both bound and unbound fractions were analysed by agarose gel electrophoresis. As shown in Figure 2A, MPG/DNA complexes signi®cantly interacted speci®cally with importin a (lane 4) but not with importin b (lane 5). Moreover, binding of the complex to importin a was dramatically reduced when the plasmid was associated with MPGD NLS (lane 3), suggesting that mutations in the NLS sequence signi®cantly affect the interaction with importin a. As a control, we veri®ed that the plasmid alone did not interact with either importin a (lane 2) or b (data not shown).
One of the major limitations of most synthetic gene delivery systems, which signi®cantly hampers gene delivery in vivo and into non-dividing cells, is their poor ability to translocate plasmid DNA into the nucleus of cells (1). In order to study the impact of the NLS of MPG on nuclear translocation of plasmids, we performed transfection experiments on synchronised or growth-arrested cells with either MPG or its NLS mutant. Cells were synchronised in G 0 by serum starvation then released back into the cell cycle by addition of serum. Four hours after release, cells were transfected with MPG/ DNA or MPGD NLS /DNA complexes and transgene expression was monitored 12 h later, when most of the cells were in late G 1 , and 20 h later, in late G 2 prior to mitosis. Data were normalised with luciferase activity obtained 48 h after transfection and compared with asynchronous cells. As reported in Figure 2B, high levels of luciferase expression were observed with MPG after only 12 h, which corresponds to 65% of expression observed after 48 h. After 24 h, transfection ef®ciency reached 70%. In contrast, when transfection was performed with MPGD NLS , luciferase activity was negligible after 12 h (7%) and only reached 26% after 24 h. Nevertheless, after 48 h, the ef®ciency of MPGD NLS was 72% that of MPG. In comparison, lipid-based systems induced basal levels of luciferase activity after 12 h (10%), which may be explained by a fraction of unsynchronised, mitotic cells in the initial population. Similar experiments were performed on serum-starved, arrested cells and transgene expression monitored 24 h after transfection revealed that when transfected with MPG, these cells exhibited 50% luciferase activity in comparison with dividing cells (Fig. 2B). In contrast, no transfection was observed with either MPGD NLS or lipid-based formulations. These data demonstrate that MPG facilitates translocation of DNA into the nucleus of arrested cells and that the NLS plays a crucial role in its nuclear uptake. The low ef®ciency of transgene expression observed with MPGD NLS reveals that nuclear translocation of MPG/DNA particles is directly mediated by the import machinery through interaction with importin a.

Role of the cysteamide group in the mechanism of MPG-mediated gene delivery
Cysteamide was originally attached to the C-terminus of MPG, so as to enable covalent attachment of probes or drugs. We previously demonstrated that this modi®cation did not affect the ability of MPG to form stable complexes with DNA. In order to investigate the impact of this C-terminal cysteamide group on gene delivery, comparative experiments were performed with an MPG peptide lacking a C-terminal cysteamide group. As shown in Figure 1A (grey bars), luciferase expression levels measured following delivery of pRL-SV40 by MPG lacking a cysteamide group were 1000fold lower than those observed with MPG modi®ed with a cysteamide group. These data suggest that the cysteamide moiety is required to promote ef®cient transfection. One explanation for this observation is that the cysteamide stabilises the carrier/DNA particle, as similarly reported for gene delivery mediated by cysteine-containing peptides (27).

MPG-mediated siRNA delivery
RNA interference (siRNA) constitutes a powerful tool to silence gene expression post-transcriptionally (28,29). In order to validate the potential of MPG-like carriers, we investigated the ability of MPG and MPGD NLS to promote cellular uptake of siRNA. Fluorescently labelled siRNA designed to silence GAPDH was complexed with both MPG and MPGD NLS at a molar ratio of 10:1, incubated on cultured HS-68 human ®broblasts in the presence of serum for 30 min, following which internalisation and subcellular localisation of siRNA were examined by¯uorescence microscopy. Two different¯uorescent probes (¯uorescein and Cy3) were used in order to avoid any artefacts which might have been associated with the nature of the probe. As ®xation procedures have sometimes been reported to cause artifactual uptake (30±32), experiments were performed on living cells. As shown in Figure 3, both peptide carriers were able to deliver siRNA into cultured cells with high ef®ciency (~90% cells). Interestingly, the ®nal subcellular localisation of the siRNA was dependent on the MPG carrier used. Fluorescently labelled siRNA localised rapidly to the nucleus when transfected with MPG, but remained mostly in the cytoplasm when transfected with MPGD NLS . Moreover, these experiments demonstrate that MPG-mediated siRNA delivery is independent of the presence of serum.
We next investigated the biological response of MPGmediated siRNA delivery. Two series of experiments were performed. First we used the protocol described by Elbashir et al. (28): HeLa and Cos-7 cells transfected with the plasmid encoding the reporter gene luciferase were incubated with siRNA/MPG or siRNA/MPGD NLS complexes in which the siRNA was designed to silence luciferase and luciferase activity was measured 48 h later. siRNA (50 nM) was associated with either MPG or MPGD NLS at a charge ratio of 10:1. As a control, siRNA transfection was performed with the commercially available lipid-based delivery system OligofectamineÔ. MPG-mediated delivery of siRNA yielded decreases of 78 and 85% in luciferase activity in Cos-7 and HeLa cells, respectively (Fig. 4A). This effect was further enhanced to 90 and 95% by MPGD NLS . Similar experiments performed with a mismatch siRNA complexed with MPG or MPGD NLS did not result in any change in levels of luciferase activity, suggesting that there is no side-effect associated with the presence of MPG. These experiments revealed that the ef®ciency of MPG-like peptides is similar to that of OligofectamineÔ, suggesting that MPG is able to release the siRNA rapidly and does not affect its biological effect upon cellular internalisation.
Implication of the NLS sequence of MPG in the kinetics of the response to siRNA To further understand the impact of the NLS in MPG, we examined the biological response of a siRNA targeting the GAPDH gene following transfection into human ®broblasts (HS-68). Data found with HS-68 human ®broblasts were essentially the same as with HeLa and Cos-7 cells (data not shown). Different concentrations of siRNA (25, 50 and 100 nM) were transfected with either MPG or MPGD NLS and the levels of GAPDH protein were analysed by western blotting 30 h post-transfection. As shown in Figure 4B, GAPDH protein levels were dramatically reduced when siRNA was transfected with either MPG carrier. Quanti®cation of the signals revealed that a concentration of 25 nM siRNA transfected with MPGD NLS reduced GAPDH protein levels by 80%, whereas 100 nM siRNA was required to reduce GAPDH levels by 60% when delivered by wild-type MPG, demonstrating that MPGD NLS is at least 2-fold more ef®cient. The kinetics of siRNA-induced degradation of GAPDH mRNA following transfection with either MPG or MPGD NLS were analysed by northern blotting. As shown in Figure 4C, MPG promoted a signi®cant reduction in GAPDH mRNA after 20 h, which was initiated after 10 h. In comparison, MPGD NLS induced a robust down-regulation of GAPDH mRNA after only 5 h. These data are comparable to those already reported for lipid-mediated siRNA delivery and suggest that the rapid release of siRNA into the cytoplasm signi®cantly improves the silencing response. Taken together, this set of experiments has led us to conclude that mutations which affect the NLS of MPG are essential to design a potent vector for speci®c delivery of siRNA.

DISCUSSION
The two major barriers to the development of ef®cient nonviral gene delivery systems are the cell membrane and the nuclear membrane. In order to overcome these limitations we have designed a single peptide-based gene delivery system, MPG, which combines a hydrophobic domain derived for the fusion sequence of the HIV-1 gp41 protein with the NLS of the large T antigen of SV40 (17±19). MPG can cross cell membranes in a receptor-independent fashion and can therefore be associated with the class of peptides termed cellpenetrating peptides (for a review see 33). In the present work, we have investigated the mechanism through which MPG promotes gene delivery and have demonstrated that it is independent of the endosomal pathway. In this respect the mechanism through which MPG promotes gene delivery differs from that of other peptide-based gene delivery systems described so far (2,3) and is more related to that proposed for cell-penetrating peptides such as penetratin, transportan and TAT (9±11). However, recently some controversy concerning the mechanism of cell-penetrating peptides has arisen and divergent interpretations have been ascribed to differences in experimental procedures or in cell lines used (12,30,32). Recent progress in live cell imaging microscopy should help to avoid misinterpretations. Several studies have revealed that in the case of the TAT-based gene delivery system, the size of the peptide/DNA complexes plays an essential role in the uptake mechanism. Particles smaller than 300 nm do not enter the cell through the endosomal pathway (14,15), in contrast to particles of 500±700 nm, which are taken up by endocytosis (16). Along these lines, an explanation for the cellular uptake mechanism of MPG may therefore be found within the size of the MPG/DNA particles. Indeed, we previously determined by dynamic light scattering that the size of MPG/DNA particles was~200±300 nm (18). In this study, we also report that small MPG/DNA particles are able to enter cells in the presence of inhibitors of the endosomal pathway, in contrast to larger MPG/DNA aggregates. One part of the controversy concerning the mechanism of cell-penetrating peptides may be due to the size of the particles, which plays a crucial role in the internalisation pathway.
Most transfection systems described so far require the nuclear membrane to break down during mitosis for DNA to get into the nucleus, and are therefore dependent on active cell cycle progression (1). As the nucleus constitutes one of the major cellular compartments to be targeted, several strategies have been proposed to improve nuclear uptake of DNA. NLSs have been associated with gene delivery formulations, either covalently attached to DNA (34,35), fused with a peptide nucleic acid (36) or incorporated into lipid-based formulations (37). The nuclear translocation property of TAT has been used to promote nuclear targeting of plasmids (13,16). Here we have demonstrated that the NLS of MPG facilitates nuclear MPG/siRNA or MPGD NLS /DNA complexes were formed in DMEM at a charge ratio of 10:1 and incubated for 30 min at 37°C, then overlaid onto Cos-7 or HeLa cells grown to 60% con¯uence and previously transfected with the pRL-SV40 plasmid encoding the reporter gene luciferase. After 30 min incubation at 37°C, 1 ml of fresh DMEM supplemented with 10% FCS was added to the cells. Luciferase activity was measured 48 h after transfection. Transfections were also performed with OligofectamineÔ as a standard. Control experiments were performed with either naked siRNA or using a mismatch siRNA (dark grey bars). (B) MPG-mediated delivery of siRNA targeting the GAPDH gene: western blot analysis. Different concentrations of siRNA (25, 50 and 100 nM) were transfected with either MPG or MPGD NLS and the levels of GAPDH protein were analysed by western blotting 30 h post-transfection. Actin was used as a control to normalise protein loading. (C) MPG-mediated delivery of siRNA targeting GAPDH gene: northern blot analysis. The kinetics of siRNA (50 nM)induced degradation of GAPDH mRNA following transfection with either MPG or MPGD NLS were analysed by northern blotting. Actin was used as a control to normalise mRNA levels in each sample.
translocation of DNA and that MPG/DNA complexes interact directly with the nuclear import machinery though importin a. Hence, the NLS constitutes an essential and unique feature of MPG in comparison with other gene delivery systems. Moreover, we have shown that a speci®c mutation in the NLS impairs translocation of DNA into the nucleus and have used this property to promote delivery of siRNA into the cytoplasm, where it was more ef®cient in its silencing function than in the nucleus, consistent with the recently reported active site of siRNA (38).
Our investigation of the mechanism underlying the ability of MPG to deliver nucleic acids into cells has revealed that MPG-mediated delivery is independent of the endosomal pathway, and in part dependent on the NLS, which is involved in electrostatic interactions with DNA as well as speci®c interactions with the nuclear import machinery. Moreover, we have shown that the NLS is essential to promote delivery of nucleic acids to the nucleus, but that it is not required for cytoplasmic targeting. This latter feature suggests that speci®c variations in the sequence of MPG may yield carriers with distinct targeting features. In conclusion, the ef®ciency together with the targeting diversity of MPG-like peptides supports their potential for a wide range of therapeutic applications.