Novel PEI/Poly-γ-Gutamic Acid Nanoparticles for High Efficient siRNA and Plasmid DNA Co-Delivery

The efficient delivery of sufficient amounts of nucleic acids into target cells is critical for successful gene therapy and gene knockdown. The DNA/siRNA co-delivery system has been considered a promising approach for cancer therapy to simultaneously express and inhibit tumor suppressor genes and overexpressed oncogenes, respectively, triggering synergistic anti-cancer effects. Polyethylenimine (PEI) has been identified as an efficient non-viral vector for transgene expression. In this study, we created a very high efficient DNA/siRNA co-delivery system by incorporating a negatively-charged poly-γ-glutamic acid (γ-PGA) into PEI/nucleic acid complexes. Spherical nanoparticles with about 200 nm diameter were formed by mixing PEI/plasmid DNA/siRNA/γ-PGA (dual delivery nanoparticles; DDNPs) with specific ratio (N/P/C ratio) and the particles present positive surface charge under all manufacturing conditions. The gel retardation assay shows both nucleic acids were effectively condensed by PEI, even at low N/P ratios. The PEI-based DDNPs reveal excellent DNA/siRNA transfection efficiency in the human hepatoma cell line (Hep 3B) by simultaneously providing high transgene expression efficiency and high siRNA silencing effect. The results indicated that DDNP can be an effective tool for gene therapy against hepatoma.


Introduction
Gene therapy is a promising strategy to treat challenging diseases by correcting or silencing defective genes. The therapeutic effects come from introducing genetic materials (DNA or RNA) to encode the correct version of mRNA and proteins that is missed/mutated in abnormal cells [1][2][3]. In addition to transfect cells with plasmid DNA carrying a specific gene, introducing the small interference RNA (siRNA) is an alternative gene therapy which can knockdown the overexpressed defective genes [4][5][6]. The mechanism of the RNA interference (RNAi) technique is to induce gene-specific silencing via the cleavage of mRNA by RNA complexed with Dicer and other nucleases [7,8].
A pivotal challenge of gene-based therapy is to development safe and effective delivery vehicles. Gene delivery vehicles are categorized into viral and non-viral vectors [9,10]. Although viral vectors are efficient in gene delivery, there are also several drawbacks, e.g., limited DNA packaging capacity, difficulty of vector production, and the safety issues such as virus insertion caused mutagenesis,

Results
In this study, we utilize the high-transfection efficiency PEI and polyanionic γ-PGA to complex with DNA/siRNA to develop DDNPs or γ-DDNPs for hepatoma cancer cell therapy. In order to monitor the effects of transgene expression or gene knockdown, pDsRed plasmid and GFP-siRNA were used, respectively, in the nanoparticle preparation. The characteristics of DDNPs and γ-DDNPs were investigated an in vitro experiments were determined in this study.

Characterization, Cellular Uptake, and Transfection Efficiency of DDNPs
Combining the advantages of branched PEI with poly-γ-glutamic acid, a dual nucleic acid co-delivery system (DDNPs or γ-DDNPs) encapsulating DNA and siRNA was developed in this study. The DDNPs was prepared first by mixing the PEI with siRNA, and then the resulting complexes were vortexed again with DNA-or DNA/γ-PGA-solution to form a dual nucleic acid co-delivery nanoparticle (DDNPs or γ-DDNPs) as shown in Figure 1. The characteristics of DDNPs were evaluated by different methods. The sizes of the prepared DDNPs at N/P ratio ranging from 3/1 to 10/1 were about 180-200 nm in diameter, with wide size distributions (i.e., a high polydispersity index; PDI). The zeta potential of the test nanoparticles at N/P ratio of 3/1-10/1 showed that the net charge on the particle surface was positive. With increasing N/P ratio (from 3/1 to 10/1), the zeta potentials (39.9 to 47.8 mV) increased in parallel as a result of presence of more cationic group-carrying PEI being incorporated into the nanoparticles (Table 1). The characteristics of DDNPs were evaluated by different methods. The sizes of the prepared DDNPs at N/P ratio ranging from 3/1 to 10/1 were about 180-200 nm in diameter, with wide size distributions (i.e., a high polydispersity index; PDI). The zeta potential of the test nanoparticles at N/P ratio of 3/1-10/1 showed that the net charge on the particle surface was positive. With increasing N/P ratio (from 3/1 to 10/1), the zeta potentials (39.9 to 47.8 mV) increased in parallel as a result of presence of more cationic group-carrying PEI being incorporated into the nanoparticles (Table 1). The binding capacity between branched PEI and nucleic acids (DNA and siRNA) was determined by a gel retardation assay. As shown in Figure 2A, the mobility of plasmid DNA and siRNA in all test nanoparticles was retarded irrespective of N/P ratio, indicating that the plasmid and siRNA can be stably complexed with the branched PEI.  The binding capacity between branched PEI and nucleic acids (DNA and siRNA) was determined by a gel retardation assay. As shown in Figure 2A, the mobility of plasmid DNA and siRNA in all test nanoparticles was retarded irrespective of N/P ratio, indicating that the plasmid and siRNA can be stably complexed with the branched PEI. The binding capacity and cellular uptake of DDNPs. The nanoparticles were prepared by using different amount of PEI, plasmid and siRNA. The binding capacity between PEI and nucleic acid were analyzed by electrophoresis (A); The cellular uptake diagram (B) and fluorescent intensity (C) of nanoparticles composed of Alexa Fluor-647 labeled siRNA were examined by flow cytometry. * p < 0.05, significant difference between the group with various N/P ratio and the group with N/P ratio of 3/1 as analyzed by Student's t test.
In order to quantify the amount of internalized nanoparticles, Alexa Fluor-647-labeled siRNA was used to prepare nanoparticles to transfect Hep 3B cells and then analyzed by flow cytometry. The mean fluorescence intensities (MFI) of cells with internalized Alexa Fluor-647-labeled nanoparticles were enhanced, in parallel with increases in N/P ratio of nanoparticles, indicating that nanoparticles with a higher N/P ratio can deliver more siRNA into cells ( Figure 2B,C). About 90% of cells uptook the nanoparticles ( Figure 2B).
Either the green or red fluorescent protein (GFP or DsRed) was used as indicators to evaluate the efficiency of the dual-delivery nanoparticles. The decreasing level of GFP or the increasing level of DsRed signal were monitored by flow cytometry after incubating the GFP-siRNA or DsRed gene containing nanovehicle with Hep 3B cells. As shown in Figure 3A, the fluorescent intensities of GFP in GFP-expressing Hep 3B cells were inhibited by 20%-50% after the incubation with GFP-siRNA carrying DDNPs and variable amount of PEI (N/P ratio = 3/1 to 10/1). The DsRed gene transfection efficiency was assessed by the percentage and fluorescence intensity of cells transfected with DDNPs. The Hep 3B cells transfected with DDNP at N/P ratio of 3/1 showed low cellular transfection efficiency (~12%) with low level of DsRed expression ( Figure 3B,C). Comparing the results of DDNPs at N/P ratio = 3/1, the nanoparticles at N/P ratios from 5/1 to 10/1 enhanced the DsRed gene transfection efficiency 2-to 3-fold (25%~35%) ( Figure 3B). The mean fluorescent intensities (MFI) of expressed DsRed protein increased drastically in the groups treated with DDNPs at N/P ratios of 5/1, 7/1, or 10/1 in Hep 3B cells ( Figure 3C). These results indicated that our DDNPs can exert high siRNA silencing and moderate foreign gene transfection performance simultaneously.

Figure 2.
The binding capacity and cellular uptake of DDNPs. The nanoparticles were prepared by using different amount of PEI, plasmid and siRNA. The binding capacity between PEI and nucleic acid were analyzed by electrophoresis (A); The cellular uptake diagram (B) and fluorescent intensity (C) of nanoparticles composed of Alexa Fluor-647 labeled siRNA were examined by flow cytometry. * p < 0.05, significant difference between the group with various N/P ratio and the group with N/P ratio of 3/1 as analyzed by Student's t test.
In order to quantify the amount of internalized nanoparticles, Alexa Fluor-647-labeled siRNA was used to prepare nanoparticles to transfect Hep 3B cells and then analyzed by flow cytometry. The mean fluorescence intensities (MFI) of cells with internalized Alexa Fluor-647-labeled nanoparticles were enhanced, in parallel with increases in N/P ratio of nanoparticles, indicating that nanoparticles with a higher N/P ratio can deliver more siRNA into cells ( Figure 2B,C). About 90% of cells uptook the nanoparticles ( Figure 2B).
Either the green or red fluorescent protein (GFP or DsRed) was used as indicators to evaluate the efficiency of the dual-delivery nanoparticles. The decreasing level of GFP or the increasing level of DsRed signal were monitored by flow cytometry after incubating the GFP-siRNA or DsRed gene containing nanovehicle with Hep 3B cells. As shown in Figure 3A, the fluorescent intensities of GFP in GFP-expressing Hep 3B cells were inhibited by 20%-50% after the incubation with GFP-siRNA carrying DDNPs and variable amount of PEI (N/P ratio = 3/1 to 10/1). The DsRed gene transfection efficiency was assessed by the percentage and fluorescence intensity of cells transfected with DDNPs. The Hep 3B cells transfected with DDNP at N/P ratio of 3/1 showed low cellular transfection efficiency (~12%) with low level of DsRed expression ( Figure 3B,C). Comparing the results of DDNPs at N/P ratio = 3/1, the nanoparticles at N/P ratios from 5/1 to 10/1 enhanced the DsRed gene transfection efficiency 2-to 3-fold (25%~35%) ( Figure 3B). The mean fluorescent intensities (MFI) of expressed DsRed protein increased drastically in the groups treated with DDNPs at N/P ratios of 5/1, 7/1, or 10/1 in Hep 3B cells ( Figure 3C). These results indicated that our DDNPs can exert high siRNA silencing and moderate foreign gene transfection performance simultaneously.  The above results showed that the gene silencing and transfection efficiency of DDNPs were dependent on the N/P ratios of nanoparticles. The cells transfected nanoparticles with higher N/P ratio resulted in better gene silencing and gene transfection efficiency outcomes. However, along with excellent efficiency comes higher cytotoxicity at high N/P ratios due to the presence of more cationic functional groups encapsulated inside the nanoparticles ( Figure 4A). DDNPs which were prepared with various N/P ratios. The control group denotes the group of cells without any treatment. * p < 0.05, significant difference between the group with various N/P ratio and the control group as analyzed by Student's t test.
The cells transfected with N/P ratio of 3/1, 5/1 or 7/1 exhibited acceptable cell viabilities in Hep 3B cells, showing little influence on cell number and morphological changes after DDNP treatment. When the N/P ratio reached 10/1, the cell viability decreased to 70%. Under the microscope, a trend of decreasing cell number and changing normal cell morphology was observed following the incubation with nanoparticles the increased N/P ratio (from 7/1 to 10/1), indicating that redundant PEI already The above results showed that the gene silencing and transfection efficiency of DDNPs were dependent on the N/P ratios of nanoparticles. The cells transfected nanoparticles with higher N/P ratio resulted in better gene silencing and gene transfection efficiency outcomes. However, along with excellent efficiency comes higher cytotoxicity at high N/P ratios due to the presence of more cationic functional groups encapsulated inside the nanoparticles ( Figure 4A). The above results showed that the gene silencing and transfection efficiency of DDNPs were dependent on the N/P ratios of nanoparticles. The cells transfected nanoparticles with higher N/P ratio resulted in better gene silencing and gene transfection efficiency outcomes. However, along with excellent efficiency comes higher cytotoxicity at high N/P ratios due to the presence of more cationic functional groups encapsulated inside the nanoparticles ( Figure 4A).

Figure 4. Cell viabilities (A) and cell morphological changes (B) of Hep 3B cells after incubated with
DDNPs which were prepared with various N/P ratios. The control group denotes the group of cells without any treatment. * p < 0.05, significant difference between the group with various N/P ratio and the control group as analyzed by Student's t test.
The cells transfected with N/P ratio of 3/1, 5/1 or 7/1 exhibited acceptable cell viabilities in Hep 3B cells, showing little influence on cell number and morphological changes after DDNP treatment. When the N/P ratio reached 10/1, the cell viability decreased to 70%. Under the microscope, a trend of decreasing cell number and changing normal cell morphology was observed following the incubation with nanoparticles the increased N/P ratio (from 7/1 to 10/1), indicating that redundant PEI already

. Cell viabilities (A) and cell morphological changes (B) of Hep 3B cells after incubated with
DDNPs which were prepared with various N/P ratios. The control group denotes the group of cells without any treatment. * p < 0.05, significant difference between the group with various N/P ratio and the control group as analyzed by Student's t test.
The cells transfected with N/P ratio of 3/1, 5/1 or 7/1 exhibited acceptable cell viabilities in Hep 3B cells, showing little influence on cell number and morphological changes after DDNP treatment. When the N/P ratio reached 10/1, the cell viability decreased to 70%. Under the microscope, a trend of decreasing cell number and changing normal cell morphology was observed following the incubation with nanoparticles the increased N/P ratio (from 7/1 to 10/1), indicating that redundant PEI already interferes with the proliferation and viability of Hep 3B cells ( Figure 4B). Therefore, the nanoparticles with N/P ratio equal to 5/1 were chosen to avoid cytotoxicity problems.

N/P/C Ratio
The gel retardation results indicated that the anionic γ-PGA did not affect the nucleic acid binding capacity of PEI, as the results showed no free DNA or siRNA molecules were associated with nanoparticles under the electric field with N/P/C ratios from 5/1/0 to 5/1/3. This results indicate γ-DDNPs can form stable conformations on the nanometer scale ( Figure 5A). interferes with the proliferation and viability of Hep 3B cells ( Figure 4B). Therefore, the nanoparticles with N/P ratio equal to 5/1 were chosen to avoid cytotoxicity problems.
The gel retardation results indicated that the anionic γ-PGA did not affect the nucleic acid binding capacity of PEI, as the results showed no free DNA or siRNA molecules were associated with nanoparticles under the electric field with N/P/C ratios from 5/1/0 to 5/1/3. This results indicate γ-DDNPs can form stable conformations on the nanometer scale ( Figure 5A). The nanoparticles were prepared by using different amount of PEI, plasmid DNA, siRNA, and γ-PGA. The stability of different N/P/C ration were analyzed under electrophoresis (A). Total cellular uptake of γ-DDNPs (FITC-or Alexa Fluor-647 labeled) with varying N/P/C ratio from 5/1/0 to 5/1/3 were analyzed by flow cytometry (B). After incubated with γ-DDNPs of different N/P/C ratios associated with varying N/P/C ratio (5/1/0 to 5/1/3), the fluorescence images of cells were photographed with fluorescence microscope (C). The control group denotes the group of cells without any treatment. * p < 0.05, significant difference between the test groups and the group with N/P/C ratio of 5/1/0 as analyzed by Student's t test. The nanoparticles were prepared by using different amount of PEI, plasmid DNA, siRNA, and γ-PGA. The stability of different N/P/C ration were analyzed under electrophoresis (A). Total cellular uptake of γ-DDNPs (FITC-or Alexa Fluor-647 labeled) with varying N/P/C ratio from 5/1/0 to 5/1/3 were analyzed by flow cytometry (B). After incubated with γ-DDNPs of different N/P/C ratios associated with varying N/P/C ratio (5/1/0 to 5/1/3), the fluorescence images of cells were photographed with fluorescence microscope (C). The control group denotes the group of cells without any treatment. * p < 0.05, significant difference between the test groups and the group with N/P/C ratio of 5/1/0 as analyzed by Student's t test.
The cellular uptake of γ-DDNPs were investigated by flow cytometry and CLSM. After 2-h post-transfection, the fluorescence intensity of the internalized FITC-and Alexa Fluor-647-labeled nanoparticles were quantified using flow cytometry. As shown in Figure 5B, the FITC-PEI fluorescence signal indicated that the total amount of γ-DDNPs internalized into cells were significantly enhanced at the N/P/C ratio of 5/1/2 compared to that of 5/1/0 (p < 0.05). The fluorescence intensity of the internalized Alexa Fluor-647 siRNA signal was significantly decreased with an increase in γ-PGA ratio (p < 0.05) ( Figure 5B).
CLSM was used to visualize the cellular uptake of Alexa Fluor-647-labeled γ-DDNPs. The cell fluorescence images after incubation with nanoparticles of different N/P/C ratios were shown in Figure 5C. After 2 h post-transfection, the Alexa Fluor-647 fluorescence intensity decreased when the γ-PGA ratio increased.
As shown in Figure 6A, the DsRed expression levels of cells transfected with γ-DDNPs with various γ-PGA ratios (γ-DDNPs) were significantly higher than those without γ-PGA (DDNPs). Flow cytometry was used to determine the exact transfection efficiency after 48 h of transfection with γ-DDNPs containing DsRed gene. As shown in Figure 6B, only up to 20% of Hep 3B cells expressed DsRed gene when transfected with the nanoparticles without γ-PGA (DDNP) (N/P/C ratio of 5/1/0). Incorporating γ-PGA into nanoparticles (γ-DDNPs, N/P/C ratios from 5/1/1 to 5/1/3), increased the transfection efficiency (35%-45%) of DsRed (p < 0.05). This result was confirmed by flow cytometry where the DsRed fluorescent intensity of Hep 3B cells transfected with nanoparticles containing γ-PGA (γ-DDNPs) showed significantly higher fluorescent intensity than those transfected with the nanoparticles without γ-PGA (DDNP) (p < 0.05) ( Figure 6C). The cellular uptake of γ-DDNPs were investigated by flow cytometry and CLSM. After 2-h post-transfection, the fluorescence intensity of the internalized FITC-and Alexa Fluor-647-labeled nanoparticles were quantified using flow cytometry. As shown in Figure 5B, the FITC-PEI fluorescence signal indicated that the total amount of γ-DDNPs internalized into cells were significantly enhanced at the N/P/C ratio of 5/1/2 compared to that of 5/1/0 (p < 0.05). The fluorescence intensity of the internalized Alexa Fluor-647 siRNA signal was significantly decreased with an increase in γ-PGA ratio (p < 0.05) ( Figure 5B).
CLSM was used to visualize the cellular uptake of Alexa Fluor-647-labeled γ-DDNPs. The cell fluorescence images after incubation with nanoparticles of different N/P/C ratios were shown in Figure 5C. After 2 h post-transfection, the Alexa Fluor-647 fluorescence intensity decreased when the γ-PGA ratio increased.
As shown in Figure 6A, the DsRed expression levels of cells transfected with γ-DDNPs with various γ-PGA ratios (γ-DDNPs) were significantly higher than those without γ-PGA (DDNPs). Flow cytometry was used to determine the exact transfection efficiency after 48 h of transfection with γ-DDNPs containing DsRed gene. As shown in Figure 6B, only up to 20% of Hep 3B cells expressed DsRed gene when transfected with the nanoparticles without γ-PGA (DDNP) (N/P/C ratio of 5/1/0). Incorporating γ-PGA into nanoparticles (γ-DDNPs, N/P/C ratios from 5/1/1 to 5/1/3), increased the transfection efficiency (35%-45%) of DsRed (p < 0.05). This result was confirmed by flow cytometry where the DsRed fluorescent intensity of Hep 3B cells transfected with nanoparticles containing γ-PGA (γ-DDNPs) showed significantly higher fluorescent intensity than those transfected with the nanoparticles without γ-PGA (DDNP) (p < 0.05) ( Figure 6C). The siRNA delivery and gene knockdown ability of γ-DDNPs were also examined by the following experiments. The Hep 3B-GFP cells stably expressed GFP proteins were used as model cell The siRNA delivery and gene knockdown ability of γ-DDNPs were also examined by the following experiments. The Hep 3B-GFP cells stably expressed GFP proteins were used as model cell to test the silencing effect of γ-DDNPs containing GFP-siRNA. The gene silencing effects were investigated by fluorescent microscope and flow cytometry. The gene silencing abilities of DDNP or γ-DDNPs are shown in Figure 7A. The GFP expression levels were obviously inhibited (>40%) after treatment with γ-DDNPs of N/P/C ratios from 5/1/0 to 5/1/2. However, a high amount of γ-PGA incorporation (N/P/C ratio of 5/1/3) resulted in poor inhibition of GFP expression. We used Lipofectamine 2000, a commercially available tranfection reagent, as positive control to compare with our γ-DDNPs. It is evident that our γ-DDNPs were outperformed by the Lipofectamine 2000 with greater GFP knockdown efficiency (Figure 7). to test the silencing effect of γ-DDNPs containing GFP-siRNA. The gene silencing effects were investigated by fluorescent microscope and flow cytometry. The gene silencing abilities of DDNP or γ-DDNPs are shown in Figure 7A. The GFP expression levels were obviously inhibited (>40%) after treatment with γ-DDNPs of N/P/C ratios from 5/1/0 to 5/1/2. However, a high amount of γ-PGA incorporation (N/P/C ratio of 5/1/3) resulted in poor inhibition of GFP expression. We used Lipofectamine 2000, a commercially available tranfection reagent, as positive control to compare with our γ-DDNPs. It is evident that our γ-DDNPs were outperformed by the Lipofectamine 2000 with greater GFP knockdown efficiency (Figure 7).

Cell Cytotoxicity of γ-DDNPs
The cytotoxicity of γ-DDNPs prepared with various N/P/C ratios was evaluated by MTT assays. The results are given in Figure 8. The cytotoxicity of raw materials (naked siRNA, naked plasmid DNA, γ-PGA and Lipofectamine 2000) was less than 10%. Similarly, minor cytotoxicity reduction was observed in all tested γ-DDNPs, ranging from 85% to 95%. There were only slight differences in cell viabilities with and without incorporated γ-PGA in our dual-delivery nanoparticles. These results indicated that the PEI/nucleic acid ratio of 5/1 is only slightly toxic to Hep 3B cells, which allows the γ-DDNPs to be used as a promising delivery system for hepatocellular carcinoma gene therapy.

Cell Cytotoxicity of γ-DDNPs
The cytotoxicity of γ-DDNPs prepared with various N/P/C ratios was evaluated by MTT assays. The results are given in Figure 8. The cytotoxicity of raw materials (naked siRNA, naked plasmid DNA, γ-PGA and Lipofectamine 2000) was less than 10%. Similarly, minor cytotoxicity reduction was observed in all tested γ-DDNPs, ranging from 85% to 95%. There were only slight differences in cell viabilities with and without incorporated γ-PGA in our dual-delivery nanoparticles. These results indicated that the PEI/nucleic acid ratio of 5/1 is only slightly toxic to Hep 3B cells, which allows the γ-DDNPs to be used as a promising delivery system for hepatocellular carcinoma gene therapy.

Discussion
Gene therapy is a promising next generation medicine strategy which offers great potential to overcome diseases caused by genetic disorders or uncontrolled expression of mutant genes, and even infectious diseases caused by viruses. There are two approaches to achieve effective manipulation of gene expression, one is to transfer the entire gene sequences into cells and replace the defective version; the other way is using technologies like RNAi or antisense DNA/RNA to knock down the expression of defective or abnormally regulated genes. Both approaches need effective nucleic acid delivery vehicles to successfully transfer giant DNA/RNA molecules which accompany with strong negative charges to across the native barriers like plasma membranes and to escape from endocytic organelles.
Non-viral gene delivery vectors are potential gene vehicles such as polymeric nanoparticles and liposomes [30,[38][39][40]. Polymeric nanoparticles usually consist of cationic polymers and anionic DNA/RNA sequences, where electrostatic interactions will condense the cationic polymers with their anionic cargos to form stable polymeric nanocomplexes. The DNA cargo will be released due to environmental pH changes after the polymeric nanoparticle enters into the endocytic pathways.
Polyamines, including PEI, poly-L-lysine, spermidine, and spermine, are positively charged polymers under physiological conditions. They can complex with negatively charged nucleic acid (DNA or RNA) by electrostatic interaction for gene delivery [41,42]. PEI is one of the potential polyamine-based transfection agents and it provides excellent pH-buffering capacity. When nanovehicles composed of PEI are trapped in the endosomal organelles, PEI will absorb the protons inside the organelles and cause the accumulation of water molecules that eventually disrupt the organelles' membrane releasing the nanovehicles and their cargos into the cytoplasm (the "proton sponge effect") [43,44]. Due to the advantages of PEI, PEI-based nanovehicles have been developed for cancer therapy [45][46][47].
DNA or RNA can electrostatically complex with cationic polymers to form nanovehicles for intracellular delivery of genetic materials. However, many non-viral gene vectors based on cationic polymer have been reported to cause cytotoxicity due to the strong electrostatic interaction of positively charged materials with cell membrane [48,49]. The cytotoxcity of PEIs are dependent on their structure and molecular weight. In general, low molecular-weight PEI reveals less cytotoxicity than high molecular-weight PEIs. Linear PEI shows low cytotoxicity than branched PEI [50]. Therefore, polyplexes composed of PEI show high gene expression in vitro and in vivo because they strongly interact with and are quickly taken up by cell surfaces due to their strong cationic properties. Furthermore, PEI provides the ability to trigger the endosomal escape which helps the nanoparticles translocate to

Discussion
Gene therapy is a promising next generation medicine strategy which offers great potential to overcome diseases caused by genetic disorders or uncontrolled expression of mutant genes, and even infectious diseases caused by viruses. There are two approaches to achieve effective manipulation of gene expression, one is to transfer the entire gene sequences into cells and replace the defective version; the other way is using technologies like RNAi or antisense DNA/RNA to knock down the expression of defective or abnormally regulated genes. Both approaches need effective nucleic acid delivery vehicles to successfully transfer giant DNA/RNA molecules which accompany with strong negative charges to across the native barriers like plasma membranes and to escape from endocytic organelles.
Non-viral gene delivery vectors are potential gene vehicles such as polymeric nanoparticles and liposomes [30,[38][39][40]. Polymeric nanoparticles usually consist of cationic polymers and anionic DNA/RNA sequences, where electrostatic interactions will condense the cationic polymers with their anionic cargos to form stable polymeric nanocomplexes. The DNA cargo will be released due to environmental pH changes after the polymeric nanoparticle enters into the endocytic pathways.
Polyamines, including PEI, poly-L-lysine, spermidine, and spermine, are positively charged polymers under physiological conditions. They can complex with negatively charged nucleic acid (DNA or RNA) by electrostatic interaction for gene delivery [41,42]. PEI is one of the potential polyamine-based transfection agents and it provides excellent pH-buffering capacity. When nanovehicles composed of PEI are trapped in the endosomal organelles, PEI will absorb the protons inside the organelles and cause the accumulation of water molecules that eventually disrupt the organelles' membrane releasing the nanovehicles and their cargos into the cytoplasm (the "proton sponge effect") [43,44]. Due to the advantages of PEI, PEI-based nanovehicles have been developed for cancer therapy [45][46][47].
DNA or RNA can electrostatically complex with cationic polymers to form nanovehicles for intracellular delivery of genetic materials. However, many non-viral gene vectors based on cationic polymer have been reported to cause cytotoxicity due to the strong electrostatic interaction of positively charged materials with cell membrane [48,49]. The cytotoxcity of PEIs are dependent on their structure and molecular weight. In general, low molecular-weight PEI reveals less cytotoxicity than high molecular-weight PEIs. Linear PEI shows low cytotoxicity than branched PEI [50]. Therefore, polyplexes composed of PEI show high gene expression in vitro and in vivo because they strongly interact with and are quickly taken up by cell surfaces due to their strong cationic properties. Furthermore, PEI provides the ability to trigger the endosomal escape which helps the nanoparticles translocate to the cytoplasm via the pH-buffering effect. However, the cationic properties of PEI will induce cytotoxicity like other cationic polymers [43,51,52]. In this study, a high amount of PEI was used to prepare nanoparticles which showed high transfection efficiency ( Figure 3B,C) but also exhibited severe cytotoxicity (Figure 4A,B). One promising approach to reduce the cytotoxicity is to incorporate or recharge anionic polymers into the cationic PEI complexes [26,29,30,53]. Therefore, in this study, γ-PGA was incorporated into γ-DDNPs and resulted in the high gene/siRNA transfection efficiency with minor cytotoxicity (Figures 6 and 7).
The major limitation of gene therapy is the inability to deliver the cargo successfully to the target site before or after entering cells. PEI can condense nucleic acids and form compact nanoparticles resulting in protection of nucleic acids from nuclease cleavage. The literature reports that PEI-based nanoparticles enter cells by endocytosis. The endocytosis pathway of PEI-based nanoparticles involved in cellular uptake and transgene expression is the caveolar-and clathrin-mediated pathway in HeLa cells [54][55][56]. Once taken up into cells, PEI/nucleic acid nanoparticles revealed strong buffering capacity to escape from the endosome and resulted in the release of nucleic acids (siRNA or DNA) into the cytoplasm. Finally, the siRNA directly affects the specific gene silencing and the DNA enters the nucleus and then expresses the transgene [57,58].
Several researchers have used γ-PGA as raw materials to form drug carriers for sustained release usage, or form biodegradable fibers [59][60][61]. In our study, the addition of γ-PGA to DDNPs decreased the zeta potential in a concentration-dependent manner ( Table 2). The particle sizes of γ-DDNPs were not larger than DDNP; therefore, γ-PGA might have changed the internal structure of the DDNPs [29,30]. The internal nanostructure of this system will be further analyzed in our next study. As shown in Figure 5B, the polyanionic γ-PGA has an effect on enhancing cellular uptake of γ-DDNPs which is consistent with our previous studies [29,30]. As shown in Figure 5B, the fluorescence intensity of Alexa Fluor-647-siRNA fluorescent positive cells upon internalization of nanoparticles showed a trend of decreasing signal with increasing amount of γ-PGA incorporation. These results could contribute to the anionic competition between γ-PGA and siRNA during complexing with cationic PEI, resulting in an unstable siRNA binding inside the nanoparticles when incorporating large amount of γ-PGA during nanoparticle preparation. The mechanism of how γ-PGA affects the cellular uptake is not well known. However, we previously found that a γ-PGA-incorporated or coated polymer complex was taken up via a receptor-mediated endocytosis pathway, the γ-glutamyl transpeptidase-mediated pathway [29,31]. After incorporating γ-PGA into our γ-DDNPs, an enhancement of the cellular uptake of the nanoparticles was observed ( Figure 6).
The ultimate goal of our research is to create an efficient tool with dual control ability which can simultaneously up-and down-regulate two specific target genes inside one cell. In this study, our dual delivery nanoparticle with low molecular-weight PEI (25 kDa) and low N/P ratio (5/1) (DDNP) can perform as an efficient dual nucleic acid delivery vehicle that shows low cytotoxicity in comparison with other nanoparticles made from high molecular-weight PEI. The introduction of γ-PGA into the DDNP system (γ-DDNPs) with N/P/C ratio of 5/1/1 exhibited improved DsRed gene transfection efficiency and high GFP knockdown effects in a hepatocellular carcinoma cell line (Hep 3B). These results indicate that the γ-DDNPs could serve as effective nucleic acid drug delivery vehicles for the future treatment of hepatoma. and antisense-5'GAACUUCAGGGUCAGCUUGCCG-3') and AllStarts Neg. siRNA AF 647 were obtained from Qiagen (Valencia, CA, USA). Plasmid peGFP-N2 (4.7 kb) and pDsRed-Monomer-N1 (4.7 kb) that encoded green and red fluorescent protein, respectively, was purchased from Clontech (Mountain View, CA, USA). Poly-γ-glutamic acid (γ-PGA) was purchased from Challenge Bioproducts (Taichung, Taiwan).

Plasmid DNA Preparation
The plasmid peGFP-N2 and pDsRed-Monmer N1 used in this study were transformed into and amplified by DH5α. The plasmid DNAs were purified by Genopure Plasmid Midi Kit (Roche, Mannheim, Germany) according to the manufacturer's instructions. The purity and integrity of plasmids were analyzed by gel electrophoresis (0.8% agarose), while their concentration was measured by UV absorption at 260 nm.

Preparation of Test Nanoparticles
The charge ratio (N/P/C) of test nanoparticles was expressed as the mole ratio of the amine groups (N) on PEI to the total phosphate groups (P) on DNA and siRNA, and the carboxyl groups (C) on γ-PGA. Test nanoparticles at various N/P molar ratios (3/1, 5/1, 7/1, or 10/1) were prepared by an ionic gelation method and illustrated in Figure 1. Briefly, an aqueous siRNA (1 µg) was complexed with different amounts (5.4, 9.0, 12.6 or 18.0 µg) of PEI with a final volume of 100 µL followed by mixing with 100 µL plasmid DNA (4 µg) by vortexing for 40 s. DDNPs with a molar ratio of 5/1 were selected for incorporating γ-PGA and were prepared with two separate steps. First, the PEI/siRNA complexes were prepared by using PEI solution (9.0 µg) mixed with siRNA (1 µg) by vortexing for 30 s. Second, the PEI/siRNA solutions were added into the solution containing plasmid DNA (4 µg) and varying molar ratios (0, 1, 2, or 3) of γ-PGA (25 kDa, 2, 4 or 6 µg), then thoroughly mixed for 30-60 s by vortexer to form γ-DDNPs, and left for at least 1 h at room temperature.

Particle Size and Zeta Potential Measurements
The size distribution and zeta potential of test nanoparticles were measured by using a Zetasizer Nano ZS (Malvern Instruments Ltd., Worcestershire, UK).

Gel Retardation Assay
The test nanoparticles prepared at various N/P/C ratios were examined for their ability to bind DNA and siRNA through gel electrophoresis. The prepared nanoparticles containing 0.4 µg DNA and 0.05 µg siRNA were mixed with 6× DNA loading dye and then were analyzed on a 1.0% agarose gel. After electrophoresis, the gel was stained in ethidium bromide solution and the DNA or siRNA bands were visualized by a UV box.

Nanoparticle Morphology Observation
An aliquot (10 µL) of the prepared nanoparticles were dropped on a carbon-coated grid and the grid was stained with 0.5% uranyl acetate and the grid was dried by air dry overnight at room temperature. The morphology and the size of the test nanoparticles were examined under a Philips CM20 transmission electron microscope (Philips Electronic Instruments, Mahwah, NJ, USA) at 120 KV.

Cell Culture and In Vitro Transfection
Hep 3B (human hepatoma) cells obtained from BCRC (Hsinchu, Taiwan) and Hep 3B cells expressing GFP were cultured in DMEM medium supplemented with 2.2 g/L sodium bicarbonate, 100 units/mL penicillin and 100 µg/mL streptomycin and 10% fetal bovine serum (FBS) in humidified air containing 5% CO 2 at 37 • C [62]. Cells were sub-cultured according to BCRC recommendations.
For transfection, cells were seeded on 6-well plates at 2 × 10 5 cells/well and transfected at next day at 70%-80% confluency. Prior to transfection, the culture medium was removed and cells were rinsed twice with transfection medium (DMEM medium without FBS). Cells were incubated in 1.2 mL transfection medium containing various N/P/C ratios of test nanoparticles or Lipofectamine 2000 (as a positive control), which carrying 2 µg DNA and 0.5 µg siRNA/well. After 2 h transfection, the transfection media containing nanoparticles were removed, the cells rinsed twice with transfection media and refilled with FBS-containing media until analysis at 48 h after transfection. Cells were then observed under a fluorescence microscope (Carl Zeiss Optical, Chester, VA, USA) to monitor the morphological changes and to obtain the transfection efficiency.

Fluorescent Nanoparticle Preparation, CLSM Visualization and Flow-Cytometry Analysis
FITC-labeled (PEI) and Alexa Fluor-647 labeled (siRNA) nanoparticles were then prepared as described in Section 2.2 to track the internalization of nanoparticles by CLSM and to quantify their cellular uptake by flow cytometry, respectively. FITC-labeled PEI (FITC-PEI) were synthesized according to the methods described in the literature.
To quantify the cellular uptake of nanoparticles, cells were plated on 6-well plates and transfected with FITC-and Alexa Fluor-647-labeled nanoparticles at a concentration of 2 µg DNA and 0.5 µg siRNA/well for 2 h. After transfection, cells were trypinized and transferred to microtubes. Subsequently, cells were washed with PBS and fixed in 4% paraformaldehyde. Finally, the cells were transfer to FACS tube and analyzed by a flow cytometer.
To track the cellular uptake of nanoparticles, cells were seeded on 12-well plates with a sterile glass coverslip at 1 × 10 5 cells/well and incubated overnight. Subsequently, cells were rinsed twice with transfection media and transfected with FITC-and Alexa-647-labeled nanoparticles. After incubation for 2 h, test samples were aspirated. Cells were then washed twice with PBS twice before they were fixed in 4% paraformaldehyde. Finally, the fixed cells were examined under a CLSM (TCS SL, Leica, Germany).

Percentage and Gene Expression Level of Cells Transfected
The percentage and gene expression levels of DsRed transfected cells were quantitatively assessed at 48 h after transfection by flow cytometry. Briefly, cells were treated with test nanoparticles encapsulated with pDsRed-N1. After 48 h, cells were trypsinized as described in Section 4.8. For each sample, 10,000 events were collected and fluorescence was detected. The percentage and mean fluorescence intensity of DsRed expressed cells was calculated as the events within the gate divided by the total number of events, excluding cell debris. And the DeRed expressed cells were also observed under fluorescence microscope.

In Vitro Gene Silencing
The gene silencing was evaluated by quantifying the silencing of GFP-expressing Hep 3B (GFP-Hep 3B) cells. For transfection, GFP-Hep 3B Cells were seeded on 6-well plates at a density of 5 × 10 5 cells/well and transfected at next day at 60%-80% confluency. Prior to transfection, the culture medium was removed and cells were rinsed twice with transfection medium (DMEM without FBS). Cells were incubated in 1.2 mL transfection medium containing test nanoparticles or Lipofectamine 2000 (as a positive control) at 2 µg pDsRed DNA and 0.5 µg GFP siRNA /well.
The gene silencing level of GFP expressing cells were quantitatively assessed at 48 h after transfection by flow cytometry. Cells were trypsinized and then transferred to microtubes, fixed by 4% paraformaldehyde and determined the transfection efficiency by flow cytometry (BD FACSCanto flow cytometer, San Jose, CA, USA). For each sample, 10,000 events were collected and fluorescence was detected.

Cell Viability Assay
The cytotoxicity of test nanoparticles was evaluated in vitro using the MTT assay. Hep3B cells were seeded on 24-well plates at 3 × 10 4 cells/well, allowed to adhere overnight and transfected by test nanoparticles containing 0.5 µg DNA and 0.125 µg siRNA. After 2 h post-transfection in transfection medium without FBS, test samples were aspirated and cells were incubated in DMEM medium with 10% FBS for another 46 h. Subsequently, cells were incubated in a DMEM medium containing 0.5 mg/mL MTT reagent for an additional 4 h. The MTT reagent was aspirated, refilled 500 µL of dimethyl sulfoxide (DMSO), and then measured in a microplate reader (Tecan Group Ltd., Männedorf, Switzerland) at a wavelength of 595 nm.

Statistical Analysis
Comparison between groups was analyzed by the one-tailed Student's t-test (SPSS, Chicago, IL, USA). All data are presented as a mean value with its standard deviation indicated (mean ± SD). Differences were considered to be statistically significant when the p values were less than 0.05.

Conclusions
A PEI-based nanoparticle system incorporating γ-PGA was developed in the study as an efficient vector for dual nucleic acid delivery. The use of such a nanoparticle composed of PEI, DNA, siRNA, and γ-PGA can simultaneously enhance the expression of a delivered gene and knockdown the overexpressed gene in hepatoma Hep 3B cells. The results obtained in the study may be used for the antitumor applications.