Problems related to the safety and efficacy of gene therapies have checked the enthusiasm once surrounding this field, though it remains a promising approach for the treatment of numerous diseases. Despite the high transfection efficiencies attainable using viral vectors, manufacturing difficulties, safety concerns, and limitations related to targeting and plasmid size have prompted considerable research into the development of non-viral vectors. Non-viral vectors demonstrate low toxicity, low immunogenicity, and ease of manufacture. However, they have not yet achieved the transfection efficiencies displayed by viruses. The inability to explain or predict transfection efficiencies results, in part, from insufficient understanding of the intracellular processes involved in gene delivery. Increasingly, research has been undertaken to probe the processes involved in overcoming the major obstacles to vector-mediated transfection: (1) internalization, (2) intracellular trafficking, (3) escape to the cytosol, (4) nuclear translocation, and (5) gene transcription/expression. This paper reviews and compares the pathways and techniques involved in successful viral and non-viral transfection. In addition, this review provides evidence that non-viral vector development has been pursued successfully thus far, producing systems capable of evading almost all major obstacles to transfection. Evaluating the abilities of non-viral and viral vectors to overcome specific cellular barriers reveals that the greatest advantage of viral vectors may be related to viral DNA, which is transcribed considerably more efficiently than plasmid DNA. Further study in this area should enable the development of non-viral vectors that transfect as efficiently as viral vectors.
The development of safe and effective methods of gene delivery to cells has been aggressively pursued for nearly 30 years as the next generation of disease treatment. To date, clinical trials have seen the use of viruses manipulated to deliver genes for treatment of haemophilia, heart failure, arthritis, muscular dystrophy, and cancer, among others (1, 2). Although viruses have an inherent ability to transfer genetic material to a variety of cells, safety and manufacturing issues, as well as limited targeting ability and limited plasmid size, have prompted the search for alternative, non-viral gene delivery systems (3). The goal in designing non-viral vectors is to create pseudo-viruses that mimic selected viral properties while overcoming targeting, plasmid cargo, and adverse immunogenicity issues. Unfortunately, non-viral vectors have yet to consistently demonstrate transfection efficiency comparable to that of viruses, regardless of gene or target cell type. Successful non-viral transfection can be hampered by several obstacles: degradation of therapeutic DNA extra- and intracellularly, poor cellular penetration, ineffective intracellular trafficking, and inadequate nuclear localization. The future success of gene therapy depends on the development of delivery systems designed to overcome these obstacles by protecting and stabilizing the genetic cargo while simultaneously facilitating cellular penetration and appropriate trafficking.
Non-viral vectors are generally non-immunogenic and non-pathogenic, demonstrate low toxicity and ease of manufacture, and can transport genes of unlimited size. While they hold significant potential as non-immunogenic, biocompatible delivery vehicles, improving their transfection capabilities is paramount to their ultimate success. The reasons for poor and cell-line-dependent transfection remain unclear, with the methods used in evaluating non-viral vectors contributing to this confusion. Vectors are assessed principally through the expression of a reporter gene, evaluating transfection ability but providing little information as to why transfection is successful or unsuccessful. Specifically, it provides no information as to the ability of vectors to overcome specific barriers to transfection, such as the cell, endosomal, lysosomal, and nuclear membranes (4, 5). Initial studies on overcoming the barriers to transfection have focused on the characteristics of the vectors themselves, with reports of size, charge, and composition affecting transfection (6-11). However, these findings have proven insufficient to explain or predict the observed differences in transfection efficiency between cell lines or between similar vectors (12-14). For these reasons, studies into the intracellular mechanisms involved in transfection have been gaining importance (10, 15). By focusing on the ability of certain vectors to overcome specific intracellular barriers, several studies have provided critical insight into the mechanisms underlying efficient transfection. Though results often conflict, the various findings present sufficient information to begin elucidating the mechanisms behind efficient transfection by non-viral vectors. Recent reviews have consolidated these findings, detailing transfection capabilities for numerous non-viral vectors (16, 17). This review takes a stepwise approach to comparing the abilities of non-viral vectors to overcome specific transfection obstacles relative to viruses, correlating individual processes with demonstrated transfection efficiencies. In particular, this review focuses on factors affecting specific stages in the transfection process: (1) internalization (usually endocytosis), (2) intracellular trafficking, (3) escape to the cytosol, (4) nuclear translocation, and (5) gene transcription/expression. Non-viral vectors are categorized according to the groups listed in Table 1, where complex refers to any non-viral vector associated with DNA.
Table Table 1.. Categories of Non-viral Vectors Considered in This Review. Complex Refers to Any Non-viral Vector Associated with DNA
• made from lipids (neutral, anionic or cationic)
• may be pH sensitive
• usually cationic
polylactic acid (PLA)
• synthetic or natural
• generally hydrophilic
poly(ethylene glycol) (PEG)
• may be combined in block copolymers
lipid and polymer
• may be covalently or noncovalently associated
peptides or proteins
fusion proteins (HIV Tat-1)
• may be covalently or noncovalently associated to DNA
• generally used in conjunction with lipoplexes or polyplexes for targeting purposes
nuclear localization signals
Mechanisms of Complex Internalization
As the first obstacle to effective transfection, clarifying the mechanism of vector uptake is a prerequisite to understanding transfection efficiency. Despite the fact that this step in the transfection process has witnessed more investigation than any other, results remain contradictory and the process is still not fully understood.
While it was originally believed that most cationic lipoplexes were internalized through fusion, it has since been shown that only a few possess this capability; generally these are liposomes with an associated fusogenic peptide or viral protein (18, 19). Most non-viral vectors, including lipoplexes, are now accepted to be internalized through endocytosis, where binding to or association with surface receptors through specific or nonspecific interactions activates intracellular signaling pathways that promote endocytosis (17). Evidence for endocytosis rests largely on the dependence of internalization on temperature, time, concentration, and energy (20, 21). Confocal and electron microscopy studies confirm internalization via endocytosis, showing complexes in intracellular vesicles suggestive of endosomes (8, 18, 19, 22). Mammalian cells demonstrate a number of endocytic processes, including clathrin- and caveolin-mediated endocytosis, macropinocytosis, and phagocytosis. As well, there are clathrin- and caveolin-independent processes thought to involve cholesterol-mediated lipid rafts (Figure 1) (20, 23, 24). Detailed reviews of these processes are available (25, 26). Each process has been implicated in the internalization of various viral and non-viral vectors across a range of cell lines (Table 2).
Evidently, there is no single internalization pathway for viral or non-viral vectors. The results of studies investigating the internalization of specific non-viral vectors by different cell lines indicate considerable conflicting evidence regarding the endocytic processes employed. One reason for these inconsistencies is the range of vectors and cell lines being investigated. Although it is difficult to conclusively associate specific pathways with specific non-viral vectors, it underlines the fact that no unique pathway leads to efficient transfection for all vectors in all cell types. Rather, the size, charge, composition, and stability of complexes appear to affect their internalization and trafficking (28, 31, 51-55).
Complex size has long been known to impact the ability to cross cell membranes, with smaller complexes repeatedly shown to better penetrate cells (51, 56-58). Additionally, endocytic processes have been shown to be strongly size-dependent (31). However, other studies have concluded that internalization efficiency is better explained by the charge or ζ potential of complexes than by size (28, 59). Another factor that may affect complex internalization is the composition of the complex. A recent study quantifying polyplex uptake demonstrates 30- to 50-fold greater uptake of PEI polyplexes as compared to PLL polyplexes despite similar size distributions (60). Other results demonstrate that uptake of PEI polyplexes in two cell lines occurs by both clathrin- and caveolin-mediated processes, while DOTAP lipoplex internalization relies solely on clathrin-dependent endocytosis in the same cell lines (23). Unfortunately, details of complex size and charge were not provided, so their impact cannot be ruled out. As is the case with size and charge, the impact of composition on internalization is disputable. One study in particular found that uptake was not affected by size (∼100−250 nm), charge (∼10−25 mV), or composition, with both non-coated and PEGylated polyplexes demonstrating similar uptake in BHK-21 cells (61). Recently, it was established that DNA conformation (circular or linear) does not affect internalization when complexed with the same non-viral vector (62). Given the conflicting evidence regarding the impact of vector size, charge and composition on transfection, it must be considered that the physiology of the target cells may also influence internalization. Convincing evidence includes a study showing that human bronchial epithelial cells (BEC) internalized 2−3 times more polyplexes than A549 cells (63). A recent study also suggests cell line-dependent internalization pathways; chitosan polyplexes (∼150 nm) are internalized via clathrin- and caveolin-dependent endocytosis in 293T and COS-7 cells but via only the caveolin-dependent pathway in CHO cells (64). Further studies are required with a greater variety of complexes and cell lines to determine the effect of cell physiology on internalization mechanisms of non-viral vectors.
As with non-viral vectors, most viruses are believed to use either fusion or endocytosis to gain cellular entry. Direct penetration, or fusion, is thought to be reserved for non-enveloped viruses, although evidence is accumulating that even non-enveloped viruses enter cells through the same endocytic routes as enveloped viruses (34, 39). The endocytic pathways vary depending on the virus and are reported to include clathrin-dependent (37, 39, 40) routes, as well as macropinocytosis (48) (Table 2). Clathrin-mediated endocytosis is the predominant entry mechanism for viruses, including hepatitis B and C, coxsackievirus B3, and AAV (adeno-associated virus) (34, 39-41). The herpes simplex virus demonstrates cellular penetration through fusion in some cells, while various endocytosis processes are used in others (65). An increasing number of studies report that viruses previously believed to enter cells through fusion, such as HIV, actually require internalization by endocytosis for productive infection (36). The TAT fusion proteins of HIV have alternatively been reported to be internalized via lipid rafts and caveolin-dependent endocytosis in HeLa and COS-1 cells (43), clathrin-mediated processes in Jurkat human T cells (37). Clearly, as with non-viral vectors, viruses mediate entry through a variety of routes, with vectors initiating different routes in various cell lines (22, 39, 65). Interestingly, despite their evolutionary advantage over non-viral vectors, viruses may not demonstrate better cell penetration abilities (66). Studies demonstrate that co-incubation with lipoplexes improves viral internalization, presumably due to the nonspecific interactions between cells and lipoplexes, rather than the receptor-mediated interactions required for viral internalization. The effect is most pronounced in cells generally resistant to infection by Ad (adenovirus), such as CHO cells, indicating that limited surface receptor expression may limit viral penetration (67). With the availability of specific surface molecules affecting uptake for both viral and non-viral vectors, explaining these poorly understood processes becomes additionally complex.
Despite studies demonstrating improved transfection when cells are exposed to higher concentrations of viral and non-viral complexes, numerous studies have established that internalization is generally efficient for many complexes and is therefore not the limiting step (68). Inconsistent findings highlight the complexity of internalization processes. Fortunately, the endocytic pathway for non-viral vectors can often be “forced” through receptor targeting, since receptor-mediated endocytosis is thought to occur through clathrin-mediated processes. Given that internalization is likely not the limiting factor in transfection, the specific endocytic process used becomes more significant amid mounting evidence that intracellular trafficking, and ultimately transfection, may be directly related to the internalization pathway (25, 30).
From Internalization to the Cytosol: Vesicle Trafficking
During clathrin-dependent and clathrin- and caveolin-independent endocytosis, the plasma membrane invaginates and pinches off to form vesicles. These vesicles, containing the endocytosed material, are then quickly uncoated and become early endosomes. Early endosomes are normally directed to sorting endosomes, where material is either moved to the surface and exocytosed or trafficked to the lysosomal pathway (Figure 1). Material destined for the lysosomal pathway is transferred to late endosomes, which undergo gradual acidification as they merge with lysosomes, where degradation occurs through low pH and various enzymes. In contrast to the classic endocytic process, internalization by caveolin-dependent processes leads to material being entrapped in caveosomes. Caveosomes lack signaling molecules required to interact with other cellular compartments and do not progress toward the endo-lysosomal pathway, thereby avoiding acidic and digestive processes. Similarly, material internalized through macropinocytosis and phagocytosis, entrapped in macropinosomes and phagosomes, respectively, is not believed to be transferred to the endo-lysosomal pathway (17). Exploitation of these pathways may enable better transfection if an escape method from the vesicles to the cytosol is identified (69).
In addition to internalization, there is evidence that vector size and composition affect their intracellular trafficking. One study found that poly-d-lysine (PDL) complexes are not transferred to lysosomes, whereas poly-l-lysine (PLL) complexes are, suggesting that the inability of enzymes to hydrolyze PDL results in larger particles that cannot be passed to lysosomes (54). Similarly, size-related restrictions are thought to prevent transfer of some lipoplexes to lysosomes, which are unable to accommodate the large vesicles formed during phagocytosis and macropinocytosis (70). While this presents an interesting possibility to prevent transfer to lysosomes, the poor internalization of larger complexes would likely nullify any subsequent benefit. Additionally, since phagocytosis is generally not a constitutive process in mammalian cells, its applicability would be limited, unless macropores are artificially induced in plasma membranes through electroporation or the use of ultrasound.
Table Table 2.. Summary of Internalization Pathways for Gene Delivery Vectors
Although internal trafficking mechanisms for many non-viral vectors remain unclear, accumulating evidence supports classic endosomal trafficking for many complexes. Confocal microscopy studies demonstrate internalized polyplexes in punctate patterns, indicative of vesicle trafficking and entrapment (8, 61, 64, 68). Complex-containing vesicles also show a gradual progression toward peri-nuclear locations, confirming their processing through the endo-lysosomal pathway (11, 61). It has also been noted that vesicles generally contain more complex aggregates as they approach the peri-nuclear region, suggesting that vesicle fusion may be occurring (61, 71). While evidence for polyplex trafficking through endosomes is mounting, evidence of trafficking to lysosomes is not conclusive. Studies using lysosomal markers have failed to indicate co-localization with some polyplexes or lipoplexes (32, 72). This absence suggests lysosomal avoidance or trafficking through an alternate pathway, such as via caveosomes or macropinosomes. Alternatively, it could result from quenching of the pH-sensitive lysosomal marker through the prevention of lysosomal acidification. The latter case seems likely, since co-localization has been shown to be dependent on vector composition, with PLL and PLGA polyplexes demonstrating co-localization with LysoTracker Red (73), while PEI, which exhibits a pH buffering capacity, does not (68). Additional support stems from antibody labeling of marker proteins, revealing the presence of PEI polyplexes in lysosomes despite an absence of co-localization using fluorescent markers (68).
Since material not transferred to the endo-lysosomal pathway has a greater chance of being exocytosed (22), trafficking from early endosomes to late endosomes and the lysosomal pathway may be necessary for efficient transfection. Studies comparing endocytosis with cytoplasmic microinjection of complexes indicate that DNA trafficking to late endo-lysosomes is necessary for efficient transfection using lipoplexes and lipopolyplexes (22, 32, 46, 74). Similar studies with polyplexes show that delivery of DNA to the cytoplasm and subsequent transfection only occurs when complexes are trafficked through the endo-lysosomal pathway (30). Additional studies using various chitosan, PEI, and pDMAEMA polyplexes demonstrate that although multiple endocytic pathways may be used, clathrin-mediated processes that traffic material to the endo-lysosomal pathway consistently lead to better transfection (64, 75, 76). These findings suggest that endo-lysosomal processing is critically implicated in the nuclear translocation of non-viral vectors, though this too may vary by cell line.
The processing of viral vectors is thought to be similar to that of non-viral vectors, with the internalization mechanism influencing intracellular trafficking and transduction capability. It has been suggested that trafficking that prevents functional processing may explain why only a small fraction of internalized viruses lead to transgene expression (77). As with non-viral vectors, trafficking of viral vectors to the endo-lysosomal pathway is essential for transfection. Many viruses, including AAV, Ad, coxsackievirus B, Semliki Forest, hepatitis C, and influenza, require trafficking to and acidification of the late endosomal compartment for effective transduction (34, 35, 37, 39, 40, 77-79). It is thought that processing of the viral capsid in an acidic environment may be necessary to activate viral components required for downstream processing, such as nuclear translocation (77, 78). Cell-line transduction dependence also correlates with intracellular trafficking: AAV viruses trafficked through the endo-lysosomal pathway mediate transduction in 293 cells; in contrast, they escape to the cytosol prior to acidification in NIH 3T3 cells, presumably through a different pathway, and are transfection incompetent (78). A separate study using an adenoviral protein as a non-viral vector demonstrates internalization through clathrin-dependent and independent processes (the latter likely caveolin-dependent), but vesicle escape and nuclear localization occurs only when the complexes are internalized through processes leading to endo-lysosomal trafficking (79). Although further research is required to elucidate the mechanisms involved, mounting evidence suggests that both viral and non-viral vectors require processing through the acidified endo-lysosomal pathway for efficient transgene expression.
Endosomal Escape: Escape to the Cytosol
Regardless of the mechanism of internalization, vectors must escape cellular organelles to gain access to the nucleus for transcription of the therapeutic DNA. Failure to escape endocytic vesicles results in complexes that become transfection-incompetent through vesicle-entrapment or lysosomal degradation. Although transfer to and degradation in lysosomes has been proposed as the most significant barrier to transfection (23), endosomal escape is a more accurate description of this obstacle. Efficient transfection by vectors internalized via macropinocytosis, phagocytosis or caveolin-dependent endocytosis, though less common, also requires escape of the vector to the cytosol for transfection. Several studies indicate that endosomal or lysosomal escape occurs efficiently. Using confocal microscopy, the gradual replacement of punctate fluorescence patterns, indicative of vesicular entrapment of labeled complexes, by diffuse homogeneous fluorescence throughout the cytoplasm has been interpreted as evidence of escape from endocytic vesicles (8, 64). Similarly, observations that co-localization of labeled complexes and the early endosomal marker Transferrin Red decreases over time suggests complexes escape from these vesicles (15, 73).
In the same manner that size and charge affect both internalization and trafficking, vector composition also influences endosomal escape. Studies show that different vectors internalized and trafficked via the same mechanisms demonstrate differing escape abilities, ultimately affecting transfection (15, 23, 64). For example, polyplexes prepared with linear- or branched-PEI efficiently escape endosomes in 293T cells, whereas PLL polyplexes remained entrapped, likely due to the inability of PLL to mediate a proton-sponge effect (15). Cellular phenotype has also been shown to influence complex escape to the cytoplasm. Fluorescently labeled chitosan-based polyplexes are observed to escape vesicles in COS-7 and 293T cells, where they are internalized via clathrin-mediated endocytosis; however, they remained entrapped in CHO cells, which internalize the complexes via caveolin-mediated endocytosis (64). Similarly, PEI and pDMAEMA polyplexes have been found to be internalized in COS-7 cells via clathrin- and caveolin-dependent routes, though only the clathrin-mediated route is transfection-competent (76). These observations do not suggest differing abilities to escape the same intracellular compartment in different cell lines but rather reflect the ability of complexes to escape from endosomes and lysosomes but not caveosomes. Furthermore, these results highlight the interdependency of the cellular processes involved in effective transfection.
Complexes demonstrate several endosomal escape mechanisms driven by different processes, including membrane destabilization and endosomolysis. Of note, the mechanisms involved in complex release from caveosomes or macropinosomes have not yet been determined (69).
Membrane Destabilization. Considerable evidence supports the notion that lipoplexes mediate DNA escape from endosomes through destabilization of the endosomal membrane, which some have termed “fusion”. The mechanism of lipoplex fusion is proposed as follows (Figure 2): (1) increasing cationic lipoplex charge during acidification causes anionic phospholipids in the membrane to flip from the cytoplasmic face to the intra-endosomal face; (2) the formation of charge neutral pairs between the membrane and lipids destabilizes the membrane, leading to (3) a reorganization of the membrane that allows DNA release (80). With lipoplex-mediated delivery, DNA is released to the cytosol “naked” according to this proposed mechanism. [For a review see Hoekstra et al. (69)]
Some polyplexes also interact with endosomal membranes at low pH, though the process is better described as membranolytic than fusogenic. Destabilization has been proposed for the escape of poly-l-histidine, PLL, PLGA, PAMAM dendrimers, PEI, and chitosan (10, 21, 24, 71, 81, 82). During acidification, polymers scavenge protons and become increasingly cationic. In a mechanism similar to that observed with lipoplexes, highly cationic polyplexes destabilize the endosomal membrane bilayer through phospholipid flipping, creating imperfections that allow complex release (Figure 3) (81, 83). Electron microscopy observations support this theory by revealing polyplexes in contact with the endosomal membrane prior to release, clearly showing membrane damage of varying size (68, 73).
Viruses are believed to mediate endosomal escape through fusogenic peptides that constitute part of the viral capsid. Following acid-induced exposure of a fusogenic peptide, many enveloped viruses escape endosomes following fusion of their coat with the endosomal membrane (55). Alternatively, the enveloped hepatitis B virus is trafficked to lysosomes, where proteolytic cleavage of surface proteins destabilizes the membrane to release the virus to the cytosol (41). Similarly, non-enveloped capsid proteins undergo a conformational change upon acidification that exposes hydrophobic domains to the endosomal membrane, causing membrane destabilization and viral release (40, 84).
Endosomolysis. Rather than inducing changes at the endosomal membrane, other processes are thought to permit release through osmotic swelling and vesicle lysis. Principle among these processes, the “proton-sponge” effect is based on proton absorption during lysosomal acidification, primarily by amino groups in polyplexes. Sequestration of protons from their counter-ions leads to increased osmotic pressure, swelling, and ultimate rupture of the vesicle (Figure 4) (85). Supporting evidence for endosomolysis, as the process is called, includes the increased transfection observed in the presence of excess cationic polymer and with materials having lower pKa values (10, 61, 86, 87). This escape method has been proposed for various cationic polyplexes, including those composed of PEI and chitosan (9, 88); however, other studies suggest different escape mechanisms for complexes of the same composition (see above).
Biomimetic Escape Mechanisms. Not all non-viral vectors possess the characteristics necessary for endosomal escape. Without an inherent feature to assist intracellular trafficking, they remain entrapped in vesicles and become transfection incompetent (15, 49). To overcome this barrier, vectors have been prepared incorporating appendages designed to enhance their cytosolic escape. Where popular vector materials, such as PLL, lack an inherent ability to escape endosomes, enhancement with fusogenic peptides (artificial or derived from viral coats), viruses, or mediating material (e.g., PEI) can increase transfer to the cytosol and transfection (70, 87). Membrane lytic peptides, synthetic pore-forming peptides, and endogenous lipids, implicated in protein trafficking through lipid-based plasma and sub-cellular membranes, improve lipoplex- and polyplex-mediated transfection (50, 89-91), but only when viruses are present in the same vesicles, further supporting the notion that endosomal escape is a critical barrier for most complexes.
It is interesting to note that all proposed escape mechanisms rely on the acidification of late endosomes. Through coincidence or design, many non-viral vectors have inherent pH-sensitive features that allow them to escape. Whether complexes escape through fusion, membrane disruption, or the proton-sponge mechanism, inhibition of the vacuolar proton pump used to acidify late endosomes and lysosomes significantly reduces transfection (10, 24, 50, 88, 93). Realizing that transfer of non-viral and viral vectors alike to the cytosol generally occurs following acidification lends further support to the notion that endo-lysosomal trafficking is critical for efficient transfection. While viruses demonstrate efficient escape mechanisms, they are not superior by orders of magnitude to non-viral vectors that mediate the proton-sponge effect or membrane disruption. A study using A549 cells revealed that adenoviral vectors demonstrate 70% escape efficiency, whereas lipoplexes demonstrate 53% efficiency (66). This represents the first transfection stage in which viral vectors may be superior to non-viral vectors, though this is dependent on complex composition. Unfortunately, no similar quantitative studies comparing viral and polyplex transfection have been reported. Despite differing vesicular escape efficiencies, the trafficking of viral and non-viral vectors to this stage in the transfection process are remarkably similar.
Regardless of the pathways involved for internalization, trafficking, and escape, vectors and/or associated DNA must translocate to the nucleus (involving both trafficking to the nucleus and penetration of the nuclear membrane) for DNA transcription to occur. Limited transfer of complexes and associated DNA into the nucleus is reflective of the nuclear membrane as the final barrier to efficient transfection. Using confocal and electron microscopy and gene expression analysis, one study revealed that although all cells demonstrated PEI polyplexes in intracellular vesicles, only 20% expressed the reporter gene (68). Similarly, a recent study revealed that while approximately 75% of bone marrow stromal cells demonstrate PEI or PLL vector fluorescence in the nucleus, labeled DNA is observed in less than 10% of cells (60).
The nuclear membrane consists of an impermeable bilayer with intermittent aqueous channels called nuclear pore complexes (NPC). These pores regulate the passage of all material into and out of the nucleus; molecules smaller than ∼40 kDa can diffuse passively, while larger molecules displaying specific nuclear localization signals (NLS) are actively transported by increasing the NPC aperture size (16). Without associated NLS, or if NPC-dependent active transport is inhibited, exogenous DNA can be excluded from the nucleus (94). Microinjection of DNA into the cytoplasm demonstrates that nuclear entry is an inefficient process, with some studies estimating that a maximum of 0.1% of cytosolically injected DNA is transcribed (46, 95). However, DNA size critically determines admission to or exclusion from the nucleus; while oligonucleotides are suspected of exhibiting a natural nuclear tropism (6), only a small portion of plasmid DNA is transported to the nucleus (15, 44). Further, single-stranded oligonucleotides translocate with near 10-fold greater efficiency than double stranded fragments (96). Although the nuclear membrane itself presents an obstacle to nuclear translocation, size-dependent mobility of DNA in the cytoplasm is a contributing factor. There are indications that the mesh-like cytoskeleton inhibits DNA mobility (97, 98); nucleic acids greater than 2 kb have limited mobility, whereas oligonucleotides diffuse rapidly throughout the cell (95, 99). This immobility impairs transfection, as confirmed by observations that perinuclear microinjection or the use of smaller DNA increases transfection (16). Additionally, the conformation of plasmid DNA has been found to affect cytosolic diffusion and subsequent transfection, with super-coiled and linearized DNA demonstrating the highest and lowest transfection levels, respectively (100). Many polymeric or peptide-based vectors condense DNA, improving its stability and mobility in the cytoplasm, as well as its ability to penetrate the nucleus (68, 101-103).
Though the processes involved in nuclear translocation of complexes are not clearly understood, confocal microscopy studies provide considerable evidence that they reach the nucleus (6, 8, 11, 15, 64, 88). Some studies observe DNA in the nucleus with no evidence of carrier co-localization, suggesting that complexes release DNA prior to nuclear transport (6, 8), while others demonstrate entire complexes entering the nucleus (9, 11, 15, 64, 68, 71, 88, 104). Studies reporting transgene expression without evidence of either complex or DNA nuclear translocation are thought to reflect the very small portion of internalized complexes or DNA that actually penetrate the nucleus, quantities likely below current analytical detection limits (15, 68).
The ability to mediate DNA translocation to the nucleus is carrier-dependent. One study demonstrated chitosan polyplexes in the nucleus as early as 4 h post-treatment, while PLL complexes never reached the nucleus (88); in a similar study, gelatin polyplexes penetrated the nucleus, while PEGylated polyplexes only reached the perinuclear region after 12 h (11). A comparison of commercial lipoplexes had similar findings, with only one mediating nuclear localization of DNA (6). It is important to note that these observed differences could be related to endosomal escape capabilities rather than nuclear translocation mechanisms. Studies on the effect of carrier composition on nuclear translocation using microinjection may further clarify this issue, but are currently rare.
It has been suggested that nuclear localization of exogenous DNA only occurs following mitosis. While a number of studies report gene expression in undividing cells (95, 105), the relatively higher transfectability of dividing cells indicates that mitosis may aid transfection. Several experiments have shown that transfection and gene expression decrease drastically in arrested cell populations and increase dramatically as cells pass through mitosis, regardless of vector composition (46, 100, 106-109). Other studies demonstrate minimal cell cycle effects on transfection, although this is vector- and cell line-dependent. The effects of cell cycle are generally less significant with polyplexes such as linear PEI and PLL (101, 108, 110); in contrast, lipoplexes and branched PEI polyplexes demonstrate transfection efficiencies that are strongly dependent on cell cycle (101, 108). Confounding the issue is the fact that most exogenous DNA is not incorporated into reformed nuclei following mitosis, even when injected into the nucleus, disputing the notion that dissolution of the nuclear membrane allows complexes or exogenous DNA to enter the nucleus (105). There remains the possibility that cells exhibit increased nuclear transport post-division as they strive to re-establish the nuclear environment through the import of NLS-bearing proteins, with passive diffusion of small particles also increasing. However, since this would preclude nuclear translocation of the comparatively large plasmid DNA used in most non-viral transfection experiments, the issue remains unresolved.
To improve nuclear translocation, NLS signals have been found to improve transfection, particularly in quiescent cells. Attachment of NLS to plasmid DNA, generally excluded due to its large size, increases nuclear accumulation and reporter gene expression, demonstrating that even large nucleic acids can gain entry to the nucleus through NPC (111-114). Various synthetic and natural NLS, including viral coat proteins, increase transfection with non-viral vectors, whether or not they are covalently conjugated (22, 91, 115-119). Nuclear translocation is affected by several variables, including DNA conformation, the specific NLS, the number of NLS peptides conjugated to each DNA fragment, and the manner in which they are conjugated. In one study, transfection with linear DNA containing a NLS peptide failed to increase expression over plasmid DNA or linear DNA without an attached NLS (76). The ability of an NLS sequence to mediate efficient nuclear translocation also depends on the vector; while lipoplexes release DNA to the cytosol, the ability of polyplexes to be released whole to the cytoplasm may render a DNA-conjugated NLS inactive due to steric, electrostatic, or size restrictions (120). Cationic polymers, such as PLL and PEI, may mimic some of the specific lysine- and arginine-rich motifs commonly found in NLS, potentially obviating the need for additional NLS signals in such polyplexes.
While nuclear transport of vectors or their DNA cargo seems a considerable obstacle to non-viral transfection, it is important to consider that viral vectors are only marginally better adapted to overcome this hurdle. Many viruses display capsid proteins that act as NLS and are thought to enter the nucleus through NPCs (34, 66, 77, 116). Though it remains an efficient process for viruses compared to naked DNA and many non-viral vectors, nuclear translocation is the rate-limiting process for several viruses, including AAV, Ad, and CPV (77, 121). Compared to lipoplex-delivered DNA, Ad demonstrate nuclear translocation efficiency only 2.5 times greater (66). Similarly, like many non-viral vectors, nuclear transport of AAV is a slow and inefficient process in several cell lines, with only a small fraction of AAV penetrating the nucleus (34, 77). One key difference between viral and non-viral transfection is that viral transduction is not affected by cell cycle, reflecting their ability to penetrate the nucleus via NPCs (108).
Viral vectors generally demonstrate greater nuclear translocation, owing predominantly to their incorporated NLS domains. However, their ability to penetrate the nucleus is not orders of magnitude greater than that of non-viral vectors. The use of biomimetic pathways, through coupling of an NLS to non-viral vectors or their DNA cargo, can significantly increase nuclear translocation, permitting non-viral vectors to match the nuclear translocation efficiency of their viral counterparts. This strategy may be beneficial for some types of vectors and specific target cell populations; it is not a strategy that will universally increase transfection in all applications.
Once a vector and/or its associated DNA overcome the nuclear membrane, transgene expression should be straightforward, relying on normal cellular processes. However, it remains unclear if DNA must dissociate from the complex to be transcribed. Mounting evidence suggests that DNA transcription only occurs following decondensation from the vector, whether it occurs prior to or following transport to the nucleus. Evidence of the need for free DNA has been demonstrated by studies indicating that transfection efficiency of linear PEI polyplexes is directly related to DNA dissociation, while polyplexes of branched PEI, which hinder dissociation, lead to lower transfection (15). Additionally, higher expression levels were recently demonstrated with chitosan polyplexes through pre-delivery of chitosanase to the cells. Presumably, this promoted DNA release through chitosan degradation, thereby facilitating expression (122). It has also been suggested that higher transfection by lower molecular weight polyplexes stems from their ability to dissociate more easily (9, 81). Similarly, higher lipoplex charge ratios reduce expression despite increased internalization and an improved ability to destabilize endosomes; the higher charge density is thought to prevent dissociation, thereby rendering DNA unavailable for transcription (123, 124). While the mechanisms of intracellular dissociation remain unknown, they are thought to involve displacement of DNA from the carrier by anionic lipids or proteins in the cytoplasm or to result from the large excess of DNA and RNA in the nucleus (15, 74, 125). The potential role of RNA in DNA release from polyplexes is supported by observations that PEI accumulates near nucleoli and does not interact with chromosomal DNA (68, 126).
Despite the obstacle of cargo dissociation from non-viral vectors, matching viral transfection efficiency is possible under optimum conditions, but requires significantly higher levels of DNA. Relative to nuclear plasmid copy numbers, Ad-mediated delivery results in 100-fold greater expression over polyplexes, though it yields 1000 times fewer plasmid copies in the nucleus (121). In terms of exogenous protein produced per plasmid copy, the efficiency of viral transfection is 5 orders of magnitude more effective than non-viral transfection. Another study found that transfection with lipoplexes requires 1000 times the amount of nuclear plasmid DNA to match Ad-transfected protein production; Ad transcription is over 7,000-fold higher in several cell lines (66). While the Ad genome evidently undergoes multiple transcriptions, it is unclear whether some lipoplex-delivered DNA is never transcribed, or if each plasmid is only transcribed a few times. Considering that the same promoter and reporter genes were used in these studies, a separate factor is clearly involved.
The lack of substantial difference between viral and non-viral vectors in overcoming the major obstacles to transfection is surprising. Although developments have progressed in the right direction, it becomes difficult to devise new strategies to improve transfection efficiency and gene expression using non-viral vectors. Acknowledging the similarities discussed, it could reasonably be surmised that smaller DNA fragments, of the order that can be accommodated in viral capsids, would improve transfection and subsequent expression through facilitated nuclear translocation. Recent studies examining transcription processes have shed light on this issue. Suspected causes of the disparity between non-viral and viral transfection efficiencies are numerous. One possible explanation for the stark contrast in transfection efficiency is that non-viral vectors compact DNA so efficiently that transcription is hindered. It has also been proposed that slow unpacking of viruses allows viral proteins to recruit transcription factors (121). The presence of other promoters, proteins, the genome structure, and the terminal repeat sequence and terminal protein of the viral construct, is also thought to affect transgene expression (66). As well, chaperone proteins play an important role in the remodeling of Ad DNA/core protein complexes during Ad transfection (127). Ad terminal proteins also bind to the nuclear matrix, which is thought to assist transcription by anchoring the complexes at particular sites within the nucleus (128).
To assess the role of the DNA construct in transcription, a study involving nuclear microinjection of a viral genome versus plasmid DNA indicated that the genome led to 40% increased protein expression (66). Though the authors hypothesized that nuclear decondensation is a greater factor in transcription than the DNA construct, the current understanding of lipoplex-mediated transfection disputes this, since DNA is believed to be released alone to the cytosol following lipoplex-mediated membranolysis. Furthermore, the observed differences in nuclear translocation between viral- and lipoplex- or polyplex-delivered DNA is not sufficient to account for the documented discrepancy in gene expression levels. In light of this, the role of the viral capsid and/or viral proteins in transcription needs to be further investigated. Clearly, the role played by the DNA construct in efficient transfection merits considerable study, given that transgene expression is the ultimate goal in non-viral transfection.
Summary and Future Directions
The majority of published investigations on non-viral vectors focus on vector development and transfection efficiency. As the search for efficient transfection using non-viral vectors continues, studies are beginning to focus on the intracellular fate of non-viral vectors, as researchers realize that the design of better vectors requires a greater understanding of the associated obstacles. Through luck or design, many non-viral vectors mediate reasonable transfection through their abilities to overcome most of the identified hurdles. In fact, most non-viral vectors demonstrate equivalent or superior abilities compared to viruses to penetrate cells and escape endosomes following appropriate trafficking. While internalization through a nonideal pathway can render non-viral vectors transfection incompetent, the same is equally true for viral vectors. While not greater by orders of magnitude, one clear advantage viruses display is their ability to translocate to the nucleus.
Of the common barriers to transfection, internalization, intracellular trafficking, and endosomal escape can largely be considered to be a single hurdle since endosomal escape depends on trafficking through the right pathway, which in turn depends on the internalization mechanism. For this reason, non-viral vector design should begin with a prescribed mechanism for endosomal escape. Fortunately, most popular vector materials, including various liposomes and polyplexes of PEI and chitosan, have an inherent ability to escape to the cytosol. Those lacking this ability can be modified through the incorporation of one of these materials or a variety of fusogenic peptides or proteins. Intracellular trafficking depends on the internalization mechanism and is independent of external determinants, increasing the importance of the endocytic pathway; generally, this is not a limiting factor for non-viral vectors. Since most vectors require endosomal acidification for escape to the cytosol, trafficking through the endo-lysosomal pathway, reached through clathrin-dependent endocytosis, is critical. Incorporation of receptor-specific ligands can target specific cell populations and encourage clathrin-dependent endocytosis.
Following endosomal escape, the subsequent hurdle in transfection is nuclear translocation. Of the physical barriers to transfection, this is the only step in which viruses demonstrate a clear advantage over non-viral vectors. This ability can largely be attributed to the presence of NLS on viral capsids that allow them to be shuttled quickly to and through NPCs. To improve non-viral transfection, NLS can be incorporated to mimic this behavior and increase nuclear translocation. Unfortunately, while transfer to the nucleus represents the last membranous obstacle to transfection, it is not the last barrier. Mounting evidence suggests that dissociation of DNA from the carrier, either in the cytoplasm or nucleus, is essential for transcription. Thus, the choice of vector material is equally critical for this process since it affects the ease of DNA dissociation. Whereas lipoplexes are thought to dissociate upon endosomal escape, polyplexes demonstrate co-localization in the nucleus and require dissociation prior to transcription.
With the right choice of material, non-viral vectors can be designed to mimic the abilities of viral vectors to overcome cellular obstacles to transfection. Modeling studies based on quantitative observations of vector-mediated transfection, such as those by Varga et al., can be used to identify the rate-limiting step in the transfection process for any given vector, thereby providing insight as to which modifications could improve transfection (121). In addition, such studies can quantify the effect of vector modification on transfection efficiency. However, it is increasingly clear that delivery of intact exogenous plasmid DNA to the nucleus is insufficient to engender efficient gene expression. Recent studies indicate that the DNA construct plays a critical role, though the mechanisms involved are presently unknown. Clearly, the development of non-viral vectors with transfection capabilities matching viral vectors requires further research into the transcription events of viral DNA versus plasmid DNA. With due consideration and design, the transfection efficiency of non-viral vectors may soon match that of viral vectors. Development of an ideal system depends on the characteristics of the targeted cell(s) and the gene being delivered; it is unlikely that one system will work ideally for all applications or all cell types. Rather, each therapy involving a particular cell type will demand the optimization of its own delivery system. It will be important to continue to monitor successes and failures of gene delivery attempts using various vectors and in different cell lines to increase knowledge of the processes involved in successful gene delivery, for application to other relevant systems.
The author thanks Dr. Shawn D. Carrigan for expert assistance with manuscript preparation and Dr. Hasan Uludag for advice and assistance in its preparation.