**2. Intracellular barriers in gene therapy: Problems and potential solutions**

Ensuring the arrival of a plasmid at its site of action in a transcriptional state is the entire aim of gene delivery systems. However, plasmids face a constant threat of being degraded. The challenge begins as soon as they are introduced into the extracellular milieu (Figure 1). In most experimental setups, cells/tissues are maintained at 37°C in serum-supplemented medium where serum nucleases can extensively damage a naked plasmid. The plasmid therefore needs protection. Next, the plasmid needs to be internalised. However, both DNA (phosphate groups within the backbone) and plasma membrane (glycoproteins with their sialic acid groups, glycerophosphates with their phosphate groups and proteoglycans which contain sulphate groups) are negatively charged. Electrostatic repulsion then ensures that there is little chance of the plasmid being naturally taken up by a cell.

Peptides as Promising Non-Viral Vectors for Gene Therapy 617

by steric obstruction. This was shown by first exposing the complexes to DNase and then using gel electrophoresis to validate the physical integrity of the plasmid upon its release from the carrier. Control plasmids that were unprotected gave no bands in the subsequent

A problem with cationic carriers is that negatively charged serum protein can be nonspecifically bound. As a result, targeting signals on the carrier can become blocked or complexes can start to aggregate. Reducing or removing serum from the media during *in vitro* transfection can mitigate such effects and improve transfection (Moore et al., 2009; Moulton et al., 2004). However, this strategy fails during *in vivo* experiments where serum proteins are unavoidably present. Another approach is to mix DNA and carrier in precise stoichiometric ratio so as to result in electro-neutral complexes (Funhoff et al., 2005). Along similar lines, zwitterionic (McManus et al., 2004) or anionic (Liang et al., 2005) lipids have been proposed in which binding between carrier and plasmid is dependant on hydrophobic forces and the presence of divalent cations such as Ca2+, Zn2+ and Mg2+ to screen the disruptive repulsion between like charges. Neutral water-soluble polymers such as polyvinyl alcohol (PVA) and polyvinyl pyrolidone (PVP) similarly exploit non-electrostatic forces such as hydrogen bonding and van der Waal's attraction to condense plasmids (Park et al., 2006). These particles are however less popular because the lack of positive charge is expected to adversely impact cellular uptake. The PEGylation of carriers is another option. PEGylation refers to the surface decoration of carriers with flexible chains of polyethylene glycol (PEG), which is a highly hydrophilic polymer capable of making a surface less susceptible to protein adsorption (Duncan, 2006). Having said that, a major concern is that PEGylation obstructs the positive charges, resulting in carriers which are less efficient in DNA binding and transfection (Lin et al., 2008; Meyer et al., 2008). Moreover, while PEGylated drug formulations are already in clinical use, PEG is non-biodegradable and its

The positive charges on cationic carriers have also been implicated in the formation of pores on the membrane, leading to cytotoxicity (Rimann et al., 2008). The observation that carrier/DNA complexes are less toxic to the cells compared to the carrier alone can be interpreted as partial evidence that charge compensation on the carrier may be a reason, albeit a non-exclusive one, for the reduced toxicity (Niidome et al., 1997). However, the exact mechanism is not totally clear. It may be due to the high rate of uptake of cationic carriers – and not the positive charges *per se* – that is to blame for pore formation. Herein lies a dilemma of the gene therapist: a high rate of internalisation can increase transfection efficiency but is also frequently paralleled by toxicity (Gabrielson & Pack, 2006; Pouton & Seymour, 2001). Fortunately, cells do have membrane repair mechanisms. One example is the membrane repair response (MRR) where the influx of Ca2+ ions directs lysosomes to donate their vesicular membrane in a concerted effort to plug the hole (Palm-Apergi et al., 2009). As always, the challenge is to strike a fine balance between two counter-acting events. Assume the plasmid has been successfully taken up by a cell via endocytosis and now resides within an endosome. Another degradative fate awaits as endosomes eventually acidify into lysosomes and activate a broth of acid hydrolases capable of degrading nucleic acids. For this reason, a high rate of uptake may not necessarily translate into high transfection efficiency if, for example, most of the plasmids are degraded in lysosomes (Lundin et al., 2008). To avoid degradation, the plasmid/carrier complex will need to escape from the confines of the endosome into the cytosol. A popularly cited mechanism by which cationic carriers can achieve this is the proton-sponge hypothesis, so called because it relies

gel electrophoresis experiment.

chronic use may be a concern (Urakami & Guan, 2008).

Fig. 1. Chronological sequence of events and challenges that a plasmid faces during its treacherous journey towards the nucleus. 1) A naked plasmid is susceptible to degradation by nucleases and is likely to be repelled from the plasma membrane. 2) Assuming successful endocytosis, the plasmid has to avoid trafficking into lysosomes where it will be degraded. 3) The plasmid has to escape into the cytoplasm. 4) It now has to diffuse through the viscous cytoplasm towards the nucleus while avoiding degradation and penetrate the nuclear membrane. 5) Transcription into mRNA can then occur if the plasmid is still intact.

For these reasons, gene delivery systems are frequently designed to be cationic in character and this fulfils several functions. First, the carrier can use its positive charges to mask the negative charges on the plasmid and package itself as a carrier/DNA complex with an overall positive charge. As expected, positively charged particles are internalised much more readily, as confirmed by an elegant study using PRINT (Particle Replication In Nonwetting Templates) technology to fabricate particles with exquisite control over their size, shape and surface charge (Gratton et al., 2008b). By keeping size and shape constant, positively charged particles were found in 84% of cells after an hour of incubation compared to the <5% uptake of negatively charged particles, proving that surface charge alone can influence uptake dramatically. Second, due to a charge screening effect, the macromolecular plasmid is collapsed (or condensed) into a compact structure more amendable for cellular uptake. This condensing process was clearly demonstrated using transmission electron microscopy which showed an elongated plasmid (long-axis diameter of ~470 nm) being compacted into tight, 80-100 nm toroid-shaped complexes by polylysine carriers (Wagner et al., 1991). Third, the carrier protects its cargo against degradation by nucleases, presumably

Cytoplasm Nucleus

5) Transcription into mRNA

External milieu

Plasmid

1) Degradation and repulsion

Plasma

Fig. 1. Chronological sequence of events and challenges that a plasmid faces during its treacherous journey towards the nucleus. 1) A naked plasmid is susceptible to degradation by nucleases and is likely to be repelled from the plasma membrane. 2) Assuming successful endocytosis, the plasmid has to avoid trafficking into lysosomes where it will be degraded. 3) The plasmid has to escape into the cytoplasm. 4) It now has to diffuse through the viscous cytoplasm towards the nucleus while avoiding degradation and penetrate the nuclear membrane. 5) Transcription into mRNA can then occur if the plasmid is still intact.

degradation

4) Diffusion and nuclear entry

3) Endosomal escape

membrane 2) Lysosomal

Endosome

For these reasons, gene delivery systems are frequently designed to be cationic in character and this fulfils several functions. First, the carrier can use its positive charges to mask the negative charges on the plasmid and package itself as a carrier/DNA complex with an overall positive charge. As expected, positively charged particles are internalised much more readily, as confirmed by an elegant study using PRINT (Particle Replication In Nonwetting Templates) technology to fabricate particles with exquisite control over their size, shape and surface charge (Gratton et al., 2008b). By keeping size and shape constant, positively charged particles were found in 84% of cells after an hour of incubation compared to the <5% uptake of negatively charged particles, proving that surface charge alone can influence uptake dramatically. Second, due to a charge screening effect, the macromolecular plasmid is collapsed (or condensed) into a compact structure more amendable for cellular uptake. This condensing process was clearly demonstrated using transmission electron microscopy which showed an elongated plasmid (long-axis diameter of ~470 nm) being compacted into tight, 80-100 nm toroid-shaped complexes by polylysine carriers (Wagner et al., 1991). Third, the carrier protects its cargo against degradation by nucleases, presumably by steric obstruction. This was shown by first exposing the complexes to DNase and then using gel electrophoresis to validate the physical integrity of the plasmid upon its release from the carrier. Control plasmids that were unprotected gave no bands in the subsequent gel electrophoresis experiment.

A problem with cationic carriers is that negatively charged serum protein can be nonspecifically bound. As a result, targeting signals on the carrier can become blocked or complexes can start to aggregate. Reducing or removing serum from the media during *in vitro* transfection can mitigate such effects and improve transfection (Moore et al., 2009; Moulton et al., 2004). However, this strategy fails during *in vivo* experiments where serum proteins are unavoidably present. Another approach is to mix DNA and carrier in precise stoichiometric ratio so as to result in electro-neutral complexes (Funhoff et al., 2005). Along similar lines, zwitterionic (McManus et al., 2004) or anionic (Liang et al., 2005) lipids have been proposed in which binding between carrier and plasmid is dependant on hydrophobic forces and the presence of divalent cations such as Ca2+, Zn2+ and Mg2+ to screen the disruptive repulsion between like charges. Neutral water-soluble polymers such as polyvinyl alcohol (PVA) and polyvinyl pyrolidone (PVP) similarly exploit non-electrostatic forces such as hydrogen bonding and van der Waal's attraction to condense plasmids (Park et al., 2006). These particles are however less popular because the lack of positive charge is expected to adversely impact cellular uptake. The PEGylation of carriers is another option. PEGylation refers to the surface decoration of carriers with flexible chains of polyethylene glycol (PEG), which is a highly hydrophilic polymer capable of making a surface less susceptible to protein adsorption (Duncan, 2006). Having said that, a major concern is that PEGylation obstructs the positive charges, resulting in carriers which are less efficient in DNA binding and transfection (Lin et al., 2008; Meyer et al., 2008). Moreover, while PEGylated drug formulations are already in clinical use, PEG is non-biodegradable and its chronic use may be a concern (Urakami & Guan, 2008).

The positive charges on cationic carriers have also been implicated in the formation of pores on the membrane, leading to cytotoxicity (Rimann et al., 2008). The observation that carrier/DNA complexes are less toxic to the cells compared to the carrier alone can be interpreted as partial evidence that charge compensation on the carrier may be a reason, albeit a non-exclusive one, for the reduced toxicity (Niidome et al., 1997). However, the exact mechanism is not totally clear. It may be due to the high rate of uptake of cationic carriers – and not the positive charges *per se* – that is to blame for pore formation. Herein lies a dilemma of the gene therapist: a high rate of internalisation can increase transfection efficiency but is also frequently paralleled by toxicity (Gabrielson & Pack, 2006; Pouton & Seymour, 2001). Fortunately, cells do have membrane repair mechanisms. One example is the membrane repair response (MRR) where the influx of Ca2+ ions directs lysosomes to donate their vesicular membrane in a concerted effort to plug the hole (Palm-Apergi et al., 2009). As always, the challenge is to strike a fine balance between two counter-acting events. Assume the plasmid has been successfully taken up by a cell via endocytosis and now resides within an endosome. Another degradative fate awaits as endosomes eventually acidify into lysosomes and activate a broth of acid hydrolases capable of degrading nucleic acids. For this reason, a high rate of uptake may not necessarily translate into high transfection efficiency if, for example, most of the plasmids are degraded in lysosomes (Lundin et al., 2008). To avoid degradation, the plasmid/carrier complex will need to escape from the confines of the endosome into the cytosol. A popularly cited mechanism by which cationic carriers can achieve this is the proton-sponge hypothesis, so called because it relies

Peptides as Promising Non-Viral Vectors for Gene Therapy 619

and caveolae-mediated endocytosis (Lundin et al., 2008; Sahay et al., 2010) produce vesicles that do not undergo acidification, while others claim that macropinosomes do acidify (Räägel et al., 2009; Wadia et al., 2004). Furthermore, is it safe to assume that vesicles which do not acidify remain distinct from endo/lysosomes? On this, opinion is also polarised, with some claiming that caveosomes (Pelkmans & Helenius, 2002; Sahay et al., 2010) and macropinosomes (Wadia et al., 2004) remain distinct from endo/lysosomes; and others arguing that vesicles from different pathways can eventually interact (Sahay et al., 2010). Thus, the proton-sponge hypothesis even if true, may not always be relevant and on top of

In the cytosol, the plasmid continues its migration towards the nucleus. Current dogma suggests that this proceeds via passive diffusion and that nuclear localisation is a hit-or-miss event. The viscous cytosolic environment makes diffusion extremely inefficient. The diffusion coefficient of bovine serum albumin (BSA) in human fibroblasts, for example, is about 70× lower than in buffer (Wojcieszyn et al., 1981). Protecting the plasmid during migration is also important as cytosolic nucleases restrict the half life of naked DNA to

During migration, another feature of the carrier becomes important – the release of its plasmid cargo. To undergo transcription, the plasmid must first be unpackaged from its carrier and the trick here, is one of timing. A plasmid that gets released too early will risk degradation while one that binds too strongly is not accessible for transcription. For this reason, lower molecular weight chitosan transfects better because their higher molecular weight counterpart interacts too strongly with their plasmid cargo (Koping-Hoggard et al., 2004). Another example involves PEI, where acetylation of its polymeric chains (which removes the amines) reduced both its buffering capacity and binding strength, but improved its transfection (Gabrielson & Pack, 2006). This suggests that vector unpackaging can indeed be a rate-limiting step and a carrier that binds in moderation is ideal. Fluorescence resonance energy transfer (FRET) is a powerful technique to study the process of vector unpackaging. It depends on the excitation of an acceptor dye by a donor dye in close physical proximity, which is the case when the plasmid is being tightly condensed. Using FRET, plasmid-release in the perinuclear region has been observed; alternatively, the

plasmid/carrier can enter the nucleus as an associated complex (Seow et al., 2009).

Nuclear entry is believed to be achieved in two ways: either via the ~10 nm wide nuclear pores or during mitosis when the nuclear envelope momentarily disintegrates (Luo & Saltzman, 2000a). The latter appears to be a more efficient method for the large-scale accumulation of complexes. It further provides a possible explanation for why amitotic cells or primary cells which proliferate slower are more difficult to transfect than cancer-derived cell lines. A dilution effect after mitotic cell division also accounts for the transient nature of gene expression mediated by non-integrating vectors. Having said that, mitosis is not a prerequisite (Won et al., 2009) as amitotic cells have been successfully transfected – it merely provides a convenient window of opportunity for nuclear entry. A proposed method to improve nuclear penetration is to attach a nuclear localisation signal (NLS) to the carrier. The quintessential example of a NLS is the short peptide sequence corresponding to the Simian virus 40 (SV40) T antigen. However, responses regarding the benefits of including a NLS have been mixed, with some investigators (Trentin et al., 2005) more convinced than others (Zauner et al., 1998). A key issue pointed out was that studies involving the use of NLS failed to examine the effect of including a NLS on nuclear import *per se* (Lam & Dean, 2010). Instead, reporter gene expression was frequently used as a proxy and an improved

that, definitive trafficking studies of the various modes of uptake are required.

about 90 minutes (Belting et al., 2005).

on the buffering capability of the carrier to absorb H+ ions and thus function as a proton sink. This model assumes that as the H+-ATPase endosomal membrane pump injects protons into the vesicle during acidification, a build-up of positive charges will result due to the ability of the carrier to protonate and absorb the protons. This then triggers a concomitant influx of compensatory negative ions (e.g., Cl-) and water, leading to the osmotic swelling of the vesicle and its eventual rupture. Consistent with this line of argument, carriers must thus contain chemical groups that are capable of undergoing protonation within the pH range of the endo/lysosomal transition, i.e., a pKa of 7 to 4. This has motivated investigators to design carriers with a large buffering capacity. The polymer polyethylenimine (PEI), for instance, has a combination of protonated amines to bind plasmids at pH 7 and a stockpile of unprotonated amines that can still undergo protonation during the endo/lysosomal transition (Boussif et al., 1995). As such, PEI has a large buffering capacity and this feature is frequently cited as a main reason for PEI's status as one of the most efficient non-viral vector commercially available for *in vitro* transfection (Putnam, 2006).

Another common strategy to promote endosomal escape is to coincubate cells with a lysosomotropic agent such as chloroquine. Traditionally used as an anti-malaria drug, chloroquine is also a weak base capable of buffering the acidification of endosomes. In reality, however, chloroquine is pleiotropic in nature – besides its lysosomotropic property, chloroquine has been reported to be able to aid in the release of plasmid from its carrier and also to inhibit DNase activity (Yang et al., 2009) – and its actual mechanism of action remains controversial. Nevertheless, chloroquine does generally improve the transfecting capability of many carriers (Pouton & Seymour, 2001). A caveat, though, is that chloroquine at the dose normally used (100 μM) is toxic to cells (Wadia et al., 2004; Zauner et al., 1998). Glycerol is another agent reported to augment transfection due to its ability to weaken and make vesicular membrane more susceptible to disruption (Zauner et al., 1997). Interestingly, the more obvious effect of adding glycerol – its osmotic property – was ruled out as the main cause of vesicle escape. Finally, carriers can be functionalised with membranedisruptive peptides, a strategy which will be reviewed in later sections.

The proton-sponge hypothesis is the most commonly cited explanation to account for the positive correlation between increased buffering capacity and transfection ability. It is hence easy to overlook that no study has provided any convincing evidence in direct support of its mechanism (Won et al., 2009). In fact, discrediting observations exist. For instance, it was reported that ammonium sulphate, also a weak base and should theoretically be able to provide buffering effects, does not boost transfection (Pouton & Seymour, 2001). Recent calculations have also revealed that the amount of strain that lipid vesicles can withstand before rupture is significantly larger than that which can be induced by endosomal buffering (Won et al., 2009). This suggests that the proton-sponge hypothesis can at best contribute, but cannot be the only cause of endosomal escape. Moreover, increased buffering and endosomal escaping properties do not always produce an accompanying increase in transfection (Akita et al., 2010; Moore et al., 2009). This implies that, while important, cytosolic access is not the only bottleneck of the transfection process. Further, it has to be pointed out that of the multiple pathways which a cell can use for internalisation, only the clathrin-mediated one is widely-accepted to involve vesicle acidification (Won et al., 2009; Zauner et al., 1997). Vesicle acidification is, of course, an inherent requirement of the protonsponge hypothesis but whether vesicles from other pathways acidify is equivocal. Some researchers believe that macropinocytosis (Akita et al., 2010; Pelkmans & Helenius, 2002)

on the buffering capability of the carrier to absorb H+ ions and thus function as a proton sink. This model assumes that as the H+-ATPase endosomal membrane pump injects protons into the vesicle during acidification, a build-up of positive charges will result due to the ability of the carrier to protonate and absorb the protons. This then triggers a concomitant influx of compensatory negative ions (e.g., Cl-) and water, leading to the osmotic swelling of the vesicle and its eventual rupture. Consistent with this line of argument, carriers must thus contain chemical groups that are capable of undergoing protonation within the pH range of the endo/lysosomal transition, i.e., a pKa of 7 to 4. This has motivated investigators to design carriers with a large buffering capacity. The polymer polyethylenimine (PEI), for instance, has a combination of protonated amines to bind plasmids at pH 7 and a stockpile of unprotonated amines that can still undergo protonation during the endo/lysosomal transition (Boussif et al., 1995). As such, PEI has a large buffering capacity and this feature is frequently cited as a main reason for PEI's status as one of the most efficient non-viral vector commercially available for *in vitro* transfection

Another common strategy to promote endosomal escape is to coincubate cells with a lysosomotropic agent such as chloroquine. Traditionally used as an anti-malaria drug, chloroquine is also a weak base capable of buffering the acidification of endosomes. In reality, however, chloroquine is pleiotropic in nature – besides its lysosomotropic property, chloroquine has been reported to be able to aid in the release of plasmid from its carrier and also to inhibit DNase activity (Yang et al., 2009) – and its actual mechanism of action remains controversial. Nevertheless, chloroquine does generally improve the transfecting capability of many carriers (Pouton & Seymour, 2001). A caveat, though, is that chloroquine at the dose normally used (100 μM) is toxic to cells (Wadia et al., 2004; Zauner et al., 1998). Glycerol is another agent reported to augment transfection due to its ability to weaken and make vesicular membrane more susceptible to disruption (Zauner et al., 1997). Interestingly, the more obvious effect of adding glycerol – its osmotic property – was ruled out as the main cause of vesicle escape. Finally, carriers can be functionalised with membrane-

The proton-sponge hypothesis is the most commonly cited explanation to account for the positive correlation between increased buffering capacity and transfection ability. It is hence easy to overlook that no study has provided any convincing evidence in direct support of its mechanism (Won et al., 2009). In fact, discrediting observations exist. For instance, it was reported that ammonium sulphate, also a weak base and should theoretically be able to provide buffering effects, does not boost transfection (Pouton & Seymour, 2001). Recent calculations have also revealed that the amount of strain that lipid vesicles can withstand before rupture is significantly larger than that which can be induced by endosomal buffering (Won et al., 2009). This suggests that the proton-sponge hypothesis can at best contribute, but cannot be the only cause of endosomal escape. Moreover, increased buffering and endosomal escaping properties do not always produce an accompanying increase in transfection (Akita et al., 2010; Moore et al., 2009). This implies that, while important, cytosolic access is not the only bottleneck of the transfection process. Further, it has to be pointed out that of the multiple pathways which a cell can use for internalisation, only the clathrin-mediated one is widely-accepted to involve vesicle acidification (Won et al., 2009; Zauner et al., 1997). Vesicle acidification is, of course, an inherent requirement of the protonsponge hypothesis but whether vesicles from other pathways acidify is equivocal. Some researchers believe that macropinocytosis (Akita et al., 2010; Pelkmans & Helenius, 2002)

disruptive peptides, a strategy which will be reviewed in later sections.

(Putnam, 2006).

and caveolae-mediated endocytosis (Lundin et al., 2008; Sahay et al., 2010) produce vesicles that do not undergo acidification, while others claim that macropinosomes do acidify (Räägel et al., 2009; Wadia et al., 2004). Furthermore, is it safe to assume that vesicles which do not acidify remain distinct from endo/lysosomes? On this, opinion is also polarised, with some claiming that caveosomes (Pelkmans & Helenius, 2002; Sahay et al., 2010) and macropinosomes (Wadia et al., 2004) remain distinct from endo/lysosomes; and others arguing that vesicles from different pathways can eventually interact (Sahay et al., 2010). Thus, the proton-sponge hypothesis even if true, may not always be relevant and on top of that, definitive trafficking studies of the various modes of uptake are required.

In the cytosol, the plasmid continues its migration towards the nucleus. Current dogma suggests that this proceeds via passive diffusion and that nuclear localisation is a hit-or-miss event. The viscous cytosolic environment makes diffusion extremely inefficient. The diffusion coefficient of bovine serum albumin (BSA) in human fibroblasts, for example, is about 70× lower than in buffer (Wojcieszyn et al., 1981). Protecting the plasmid during migration is also important as cytosolic nucleases restrict the half life of naked DNA to about 90 minutes (Belting et al., 2005).

During migration, another feature of the carrier becomes important – the release of its plasmid cargo. To undergo transcription, the plasmid must first be unpackaged from its carrier and the trick here, is one of timing. A plasmid that gets released too early will risk degradation while one that binds too strongly is not accessible for transcription. For this reason, lower molecular weight chitosan transfects better because their higher molecular weight counterpart interacts too strongly with their plasmid cargo (Koping-Hoggard et al., 2004). Another example involves PEI, where acetylation of its polymeric chains (which removes the amines) reduced both its buffering capacity and binding strength, but improved its transfection (Gabrielson & Pack, 2006). This suggests that vector unpackaging can indeed be a rate-limiting step and a carrier that binds in moderation is ideal. Fluorescence resonance energy transfer (FRET) is a powerful technique to study the process of vector unpackaging. It depends on the excitation of an acceptor dye by a donor dye in close physical proximity, which is the case when the plasmid is being tightly condensed. Using FRET, plasmid-release in the perinuclear region has been observed; alternatively, the plasmid/carrier can enter the nucleus as an associated complex (Seow et al., 2009).

Nuclear entry is believed to be achieved in two ways: either via the ~10 nm wide nuclear pores or during mitosis when the nuclear envelope momentarily disintegrates (Luo & Saltzman, 2000a). The latter appears to be a more efficient method for the large-scale accumulation of complexes. It further provides a possible explanation for why amitotic cells or primary cells which proliferate slower are more difficult to transfect than cancer-derived cell lines. A dilution effect after mitotic cell division also accounts for the transient nature of gene expression mediated by non-integrating vectors. Having said that, mitosis is not a prerequisite (Won et al., 2009) as amitotic cells have been successfully transfected – it merely provides a convenient window of opportunity for nuclear entry. A proposed method to improve nuclear penetration is to attach a nuclear localisation signal (NLS) to the carrier. The quintessential example of a NLS is the short peptide sequence corresponding to the Simian virus 40 (SV40) T antigen. However, responses regarding the benefits of including a NLS have been mixed, with some investigators (Trentin et al., 2005) more convinced than others (Zauner et al., 1998). A key issue pointed out was that studies involving the use of NLS failed to examine the effect of including a NLS on nuclear import *per se* (Lam & Dean, 2010). Instead, reporter gene expression was frequently used as a proxy and an improved

Peptides as Promising Non-Viral Vectors for Gene Therapy 621

directly shoots DNA-coated metal particles (frequently gold) into cells (Merediz et al., 2000). Expectedly, such aggressive methods can irreversibly damage the cell membrane and cause widespread cell death. Scrape loading, first described in 1984, requires the forceful scraping of cells attached to their culture dishes, thereby creating pores on their membranes for plasmid entry (McNeil et al., 1984). This technique, however, is only applicable to adherent cells. The osmotic lysis of pinosomes was proposed in 1982 and requires that cells capable of pinocytosis be initially exposed to a hypertonic medium of sucrose, PEG and the plasmid of interest (Okada & Rechsteiner, 1982). Subsequent exchange to a hypotonic medium then released the pinosomal content. Although cells are constitutively capable of pinocytosis, such wild fluctuations in osmotic conditions can cause cell death. Finally, microinjection refers to the piecemeal injection of plasmids directly into the cell. This technique, while relatively gentle, is extremely laborious. As such, it is prone to failure and the number of

There are systems that function as particulate carriers by ferrying nucleic acids into or near to their site of action. Such systems can be viral or non-viral in nature. Viral vectors such as adenovirus, adeno-asociated virus (AAV), human immunodeficiency virus (HIV), equine infectious anaemia virus (EIAV), herpes simplex virus (HSV) and murine leukaemia virus (MLV) have been used and comprehensively reviewed elsewhere (Kay et al., 2001). The prime advantage of viral vectors is their transfection efficiency which has benefitted from centuries of selective evolutionary pressure. As a result, viruses are several orders of magnitude more efficient than non-viral vectors (Kircheis et al., 1997) and account for about 70% of all clinical trials involving gene therapy so far (Won et al., 2009). A recent success for viral gene therapy involved the use of lentiviruses to deliver a correct copy of a therapeutic gene to an adult patient suffering from β-thalassaemia (Cavazzana-Calvo et al., 2010). Such patients suffer from defective haemoglobin production and require chronic blood transfusion for survival. Upon reinfusing the patient with his own bone-marrow haematopoietic stem cells that had earlier been transduced *ex vivo*, the patient became transfusion free 1 year after treatment and has been doing well, according to the most recent report at 33 months after treatment. The longer-term outcome, of course, remains to be seen. Having said that, investigators have not abandoned all forms of non-viral research as there are limitations in the use of viral vectors. For example, the technical difficulty of scaling up virus production compliant to good manufacturing practices (GMP) may prevent such therapy from being cheaply accessible (Sheridan, 2011). The size of the construct that can be delivered is also limited. Above all, toxicity and immunogenicity (reviewed by Nayak & Herzog, 2010) provoked by the viral vectors can and have brought clinical trials to premature ends. Repeated administration is also not possible if the body has mounted a systemic immune response. The danger of viral gene therapy was first illustrated about a decade ago by the death of Jesse Gelsinger (Hollon, 2000). Researchers were using experimental adenoviruses to correct his partial ornithine transcarbamylase deficiency affecting the ability of his body to metabolise nitrogen. Unfortunately, the systemic inflammatory response syndrome was triggered and he succumbed, rather rapidly, to multiple organ failure. In another high-profile example, stem cells transduced *ex vivo* with MLV were used to treat children suffering from X-linked severe combined immunodeficiency (Hacein-Bey-Abina et al., 2003). However, MLV is an integrating retrovirus and can cause insertional mutagenesis, which is a phenomenon where random

cells that can be processed is limited.

**3.2 Viral vectors** 

expression was simply accepted to be due to the inclusion of NLS. Given that most NLS are cationic, it is debatable if the observed increase in transfection is strctly the result of improved nuclear import *per se*, or due to other non-specific effects such as enhanced plasmid association and uptake. Moreover, the fact that a NLS can be sterically hindered by plasmids upon binding also contradicts the requirement of NLS to be freely accessible for interaction with importins, the nuclear entry regulating proteins.

In light of the many obstacles that nucleic acids and their carriers have to surmount, it is perhaps understandable that less than 10% of the pool of plasmids that made it into a cell will go on to accumulate in the nucleus (Lam & Dean, 2010). The challenge is to design a multifunctional vector that can address the issues highlighted above and yet, remain safe to use in a human body.
