114 (2006) 110–117www.elsevier.com/locate/jconrel

GENEDELIVERY

Journal of Controlled Release

Enhanced transfection of primary cortical cultures using arginine-graftedPAMAM dendrimer, PAMAM-Arg

Jung-Bin Kim a, Joon Sig Choi b, Kihoon Nam c, Minhyung Lee d,Jong-Sang Park c,⁎⁎, Ja-Kyeong Lee a,⁎

a Department of Anatomy and Center for Advanced Medical Education (BK21), Inha University School of Medicine, 7-241 Shinheung-dong,Jung-Gu Inchon, 400-712, South Korea

b Department of Biochemistry, Chungnam University, Daejeon, South Koreac School of Chemistry and Molecular Engineering, Seoul National University, Seoul, South Koread Department of Bioengineering, School of Engineering, Hanyang University, Seoul, South Korea

Received 12 July 2005; accepted 8 May 2006Available online 12 July 2006

Abstract

PAMAM-Arg is a cationic arginine-grafted polyamidoamine (PAMAM) dendrimer. In the previous study, we reported that PAMAM-Arg facilitatestransfection in a range of mammalian cell types. In the present study, we investigated the transfection efficiency of PAMAM-Arg in primary corticalcultures, which are known to be extremely vulnerable to exogenous gene transfection. PAMAM-Arg/DNA complexes showed particularly hightransfection efficiencies and low cytotoxicity in primary cortical cells, as compared to other gene carriers such as, native PAMAM, polyethylenimine(BPEI), and Lipofectamine. Efficient transfection was not limited to neurons but extended to all three glial cells, astrocytes, microglia, andoligodendrocytes, present in these primary cortical cultures. The potential use of PAMAM-Arg was demonstrated by efficient gene knock-down bytransfecting HMGB1 shRNA-expressing plasmid. The numbers of green fluorescent protein (GFP)-positive and HMGB1-negative cells indicated thatPAMAM-Arg/shRNA-expressing plasmid complex suppressed target gene expression in over 40% of cells, which is the highest level achieved to date inprimary cortical culture by any gene carrier. Here, we present evidence of the successful delivery and expression of both a reporter gene and of a shRNA-expressing plasmid in primary cortical cells, which demonstrates the potential of PAMAM-Arg for mediating gene delivery to primary neuronal cells.© 2006 Published by Elsevier B.V.

Keywords: PAMAM dendrimer; L-arginine; Primary cortical culture; Gene delivery

1. Introduction

The inefficient transfection of post-mitotic neurons hampersour ability to express exogenous genes in primary cortical cul-tures. Viral infections are predominantly used due to their highgene transfer efficiencies and neuronal tropism. Work in this areahas been mainly based on adenoviral or lentiviral vectors [1–5],adeno-associated virus (AAV) [6], and herpesvirus (HSV) [7,8].However, these viruses have certain inherent disadvantages whenused as gene delivery vectors, due to for example, a small insert

⁎ Corresponding author. Tel.: +82 32 890 0913; fax: +82 32 884 2105.⁎⁎ Co-corresponding author. Tel.: +82 2 871 5355; fax: +82 2 880 6660.

E-mail addresses: pfjspark@snu.ac.kr (J.-S. Park), jklee@inha.ac.kr(J.-K. Lee).

0168-3659/$ - see front matter © 2006 Published by Elsevier B.V.doi:10.1016/j.jconrel.2006.05.011

size for AAV [9] or the immune and inflammatory responseselicited by HSV [10] or adenoviruses [11].

Although other methods with relatively low toxicities andimmune responses have been used as alternatives to viralvectors, low transfection efficiency is a general problem. Thecalcium phosphate-DNA co-precipitation method [12–15],electroporation [16], and particle bombardment-mediated genetransfer [17] showed relatively low transfection efficiencies(0.1–7%, 2%, and 1–2%, respectively). More reproducibleresults have been obtained using lipid or polymer-based genedelivery systems, particularly with cationic lipid reagents.Among these, Lipofectamine and Lipofectamine Plus havetransfection efficiencies of 3%, and recently Lipofectamine 2000was found to produce a transfection rate of 25–27% in rat E18cortical cultures [18]. In addition, the cationic polymerpolyethylenimine (BPEI) was reported to have a transfection

GENEDELIVERY

111J.-B. Kim et al. / Journal of Controlled Release 114 (2006) 110–117

efficiency of 14% in post-mitotic rat fetal hypothalamic neurons[19]. Transfection efficiencies of transfection reagents men-tioned above were examined by reporter gene expression, forexample, β-galactosidase, under the microscope [15,16,18,19]or using biochemical assay [14,17].

In a previous paper, we described a novel nonviral genedelivery vector, PAMAM-Arg, which is composed of a poly-amidoamine (PAMAM) dendrimer (generation 4) grafted withbasic L-arginine residues [20]. Approximately 58 molecules ofarginine were coupled to a single PAMAM dendrimer that con-tained 64 surface primary amines [20]. Reporter gene assay usingβ-galactosidase revealed that gene delivery efficacy of PAMAM-Arg was greatly increased to a level comparable to that of BPEI(25,000 Da) for HepG2, 293, Neuro2A, and primary rat vascularsmooth muscle cells [20]. The purpose of this study was toexamine gene delivery efficiency by PAMAM-Arg in primarycultures of rat fetal neuronal cells, which are extremely vulnerableto conventional nonviral transfection agents.

2. Materials and methods

2.1. Preparation of mixed neuronal-glia culture

Mixed cortical cells including neurons and astrocytes werecultured according to the method described by Kim et al. [21].Mixed cultures were prepared from embryonic day 15.5 (E15.5)mouse cortices. Dissociated cortical cells were plated at a den-sity of five hemispheres per plate (approximately 4×105 cellsper well) onto poly-D-lysine (100 μg/mL) and laminin (100 μg/mL) coated 24-well plates. The plates were placed in an in-cubator at 37 °C, 5% CO2, with humidified air. Cultures weremaintained without antibiotics in Neurobasal™ mediumsupplemented with B27 (GIBCO Invitrogen, Grand Island,NY, USA). At day 6 in vitro (DIV6), when the astrocytesreached confluence underneath the neurons, cytosine arabino-furanoside (ara-C) was added to produce a final concentration of10 μM and maintained for 2 days to halt microglia growth. Fetalbovine serum and glutamine were no longer supplemented inthe culture medium from day 6, and the medium was changedtwice a week after day 8. Cultures were used at DIV13-15.

2.2. Generation of the HMGB1 shRNA transgene

To generate shHMGB1-pU6, two 64-mer sense and anti-sense oligonucleotides containing an inverted repeat of 19-ntcorresponding to the high mobility group box 1 (HMGB1)coding region were chemically synthesized. The inverted motif,that also contains the 7 nucleotides spacer and five Ts, wassubcloned into the BamH1 and HindIII sites of the pU6 plasmid.A mutant plasmid, MshHMGB1-U6, was constructed inthe same way except 6 nucleotides within the target regionwere substituted. Sequences inserted immediately downstreamof the U6 promoter were as follows shHMGB1, 5′-GATCCCGAAGCACCCGGATGCTTCTTTCAAGAGAAGAAGCATCCGGGTGCTTCTTTTTTGGAAA-3′; andMshHMGB1,5′-GATCCCGAAGCACTACTGCGCTTCTTTCAAGAGAAGAAGCGCAGTAGTGCTTCTTTTTTGGAAA-3′.

2.3. Plasmid preparation

The firefly luciferase gene was used as a reporter gene tomonitor the result of gene transfection. The luciferase expres-sion plasmid (pCN-Luci) was constructed by subcloning cDNAof Photinus pyralis luciferase with 21-amino acid nuclearlocalization signal from SV40 large T antigen to pCN [22]. TheGFP expression plasmid (pEGFP-N1) was purchased fromClontech (Palo Alto, CA, USA). Plasmid DNAwas transformedinto E. coli TOP 10 competent cells and highly purified co-valently closed circular plasmid DNA was isolated by plasmidpurification Mega kits from Qiagen (Valencia, CA, USA) ac-cording to the manufacturer's instructions. Plasmid was pre-cipitated in isopropanol and further washed with 70% ethanoltwice and resuspended in distilled water. The optical densityratios at 260–280 nm of these plasmid preparations were in therange of 1.8–2.0. DNA was stored at - 20 °C until use.

2.4. Transient transfections

For the transfection assays, the cells were seeded at a density of4×105 cells per well in 24-well plates. All transfections wereperformed using PAMAM-Arg as a gene carrier unless indicatedotherwise. pCN-Luci (1 μg)/polymer complexes were prepared atvarious N/P (nitrogen of polymer/phosphate of DNA) ratios,ranging from 2.4 to 18.8 in Opti-MEM, and incubated for 30 minat room temperature. In Fig. 1B and C, BPEI (25,000 Da, N/Pratio 7.8) (Sigma, St Louis, MO, USA), PAMAM (14,215 Da, N/P ratio 6.0) (Sigma, St Louis, MO, USA), poly-L-arginine(15,000–70,000 Da, N/P ratio 4.0) (Sigma, St Louis, MO, USA),and Lipofectamine (16 μg/ml) were used as transfection reagentsor naked DNA (1 μg per well) was used. The cells were washedtwice with serum-free MEM medium, and then 500 μl of freshserum-free medium was added. The plasmid/polymer complexwas added to each dish. The cells were then incubated for 4 h at37 °C in a 5% CO2 incubator. After 4 h, the transfection mixtureswere removed and 600 μl of freshMEMmedium containing 10%horse serum (HS) was added to each well. The cells wereincubated for additional 48 h at 37 °C and luciferase assay wereperformed. The pEGFP-N1 (Clontech, Palo Alto, CA, USA)/PAMAM-Arg or shHMGB1-expressing plasmid/PAMAM-Argcomplexes were transfected at N/P ratio 9.4. The numbers ofGFP-positive cells counted in 20 randomly selected regions fromtwo independent experiments are presented.

2.5. Luciferase assay

After growth media was removed, the cells were washedwith PBS and lysed for 30 min at room temperature using150 μl of Reporter lysis buffer (Promega, Madison, WI). Luci-ferase activity was measured using a LB 9507 luminometer(Berthold, Germany).

2.6. Cell viability assay

The cytotoxicity of PAMAM-Arg was compared with that ofLipofectamine, BPEI, PAMAM, and poly-L-arginine by the

Fig. 1. Transfection efficiency in primary cortical cultures and cytotoxicityassays. (A) Comparison of transfection efficiencies at various charge ratios inprimary cortical culture at 1.0 μg DNA/well in the absence of serum. (B)Luciferase expressions mediated by BPEI, native PAMAM, Lipofectamine,poly-L-arginine and PAMAM-Arg under optimum conditions (see Materials andMethods) were measured at 48 h after transfection. (C) Cells were incubatedwith each complex and DNA for 24 h and cytotoxicities were assessed by MTTassay. Results are expressed as averages±SEM. (n=5). ⁎ p<0.05, ⁎⁎ p<0.01.

GENEDELIVERY

112 J.-B. Kim et al. / Journal of Controlled Release 114 (2006) 110–117

MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazoliumbromide) assay. Cells were seeded at a density of 1×105

cells/well in a 24-well plate and grown in 300 μl of media for2 days prior to the incubation with polymer/DNA complexes.Polymer/DNA complexes were prepared at their optimaltransfection condition as described above. After treating cellswith each complex for 2 days, 75 μl of MTT stock solution(2 mg/ml) was added to each well and incubated further for 4 hin 37 °C incubator. The media was removed and the cells ineach well were lysed by the addition of 300 μl dimethyl sul-

foxide. The plate was placed on a shaker for 1 h at roomtemperature to complete the lysing process, then the opticaldensity of each well was measured by a 96-well plate readerwith a filter setting at 570 nm.

2.7. RNA preparation and reverse transcriptase PCR

Total RNA was prepared using the TRIzol reagent (GibcoBRL, Gaithersburg, MD). The concentration of RNA was mea-sured by the absorbance at 260 nm, and RNA integrity wasconfirmed by formaldehyde-formamide denatured agarose gelelectrophoresis. One microgram of RNA isolated from eachsample was used for cDNA synthesis. First-strand cDNA syn-thesis was primed with random hexamers and carried out ac-cording to the manufacturer's specifications (RT-PCR kit; Roche,Mannheim, Germany). The cDNA equivalent to 200 ng of totalRNA was subjected to subsequent PCR analysis following themanufacturer's protocol (PCR core kit; Roche). The primersequences of rat HMGB1 have been described previously [21].

2.8. Immunocytochemistry

Immunological staining was performed following previouslydescribed method [23]. Primary antibodies were diluted at1:500 for anti-HMGB1 (BD Bioscience, Franklin Lakes, NJ,USA), anti-Neu N (Chemicon, Temecula, CA, USA), anti-GFAP (Jackson ImmunoResearch Lab, West Grove, PA, USA),anti-RIP (Chemicon, Temecula, CA, USA), and anti-CD45(Chemicon, Temecula, CA, USA). For double immunostaining,rhodamine-labeled anti-mouse IgG (Jackson ImmunoResLaboratory, West Grove, PA, USA) was used as the secondaryantibody for monoclonal anti-Neu N,-GFAP,-RIP, and-CD45antibodies. Rhodamine-labeled anti-rabbit IgG (Jackson Immu-noRes Laboratory, West Grove, PA, USA) was used as thesecondary antibody for monoclonal anti-HMGB1. The numbersof Neu N-positive cells counted in 20 randomly selected regionsfrom two independent experiments are presented.

2.9. Statistical analysis

Statistical analysis was performed by analysis of variance(ANOVA) followed by the Newman–Keuls test. All data werepresented as average±SEM and a statistical difference wasaccepted at the 5% level.

3. Results and discussion

3.1. Transfection efficiency of PAMAM-Arg in primary corticalcultures

Primary cortical cultures were derived from the cortex ofE15.5 rat embryos, and consisted mainly of neurons and glialcells. These cells have been used in various experiments re-garding neurogenesis, neurophysiology, and neuropathologystudies. Based on our previous observation that PAMAM-Argenables high transfection efficiencies in Neuro 2A cells, amouse neuroblastoma cell line [20], we investigated gene

GENEDELIVERY

113J.-B. Kim et al. / Journal of Controlled Release 114 (2006) 110–117

delivery by PAMAM-Arg in primary cortical cultures usingreporter gene expression analysis. Transfection efficiencies atvarious charge ratios of PAMAM-Arg (2.4, 4.7, 9.4, and 18.8)indicated that maximal reporter gene expression was obtained atcharge ratios of 9.4 and 18.8 (Fig. 1A). Transfection efficienciesin the presence or absence of serum were comparable at all N/Pratios tested (data not shown).

When the transfection efficiency of PAMAM-Arg in primarycortical cultures was compared to those of native PAMAM, BPEI(BPEI 25000), poly-L-arginine, Lipofectamine, and naked DNA,PAMAM-Arg (N/P ratio 9.4) was found to produce the highestgene expression levels. Optimum condition of each transfectionreagent in primary cortical cultures was obtained by testingtransfection efficiencies at various concentrations (data notshown). The transfection efficiency of PAMAM-Arg was morethan two orders of magnitude greater than Lipofectamine or poly-L-arginine, and more than one order of magnitude greater thanBPEI or native PAMAM(Fig. 1B). The result agreeswellwith ourprevious report on the high transfection efficiency of PAMAM-Arg in various cell lines, including Neuro 2A cells [20].

3.2. Cytotoxicity of PAMAM-Arg in primary cortical cultures

The cytotoxicities of cationic polymers have been reportedto be a function of the interaction between the polymer and thecell membrane and/or of cellular uptake efficiency [24,25]. Tocompare the cytotoxicity of PAMAM-Arg (N/P ratio 9.4) in

Fig. 2. Transfection of primary cortical cultures with GFP-expressing plasmid usingpEGFP-N1 (1 μg DNA per 4×105 cells/well) using PAMAM-Arg (A,B,C), PAMAM2% paraformaldehyde, and permeabilized with 0.1% Triton. Images were obtainefluorescence (B,E,H) are presented. The right panels are overlays of the two preceding20 randomly selected regions from two independent experiments are presented (J). D

primary cortical culture to those of native PAMAM dendrimer,BPEI, poly-L-arginine, and Lipofectamine, the cells were in-cubated with the transfection reagents for 48 h as described inFig. 1B, and then subjected toMTTassay. Lipofectamine/DNAand poly-L-arginine/DNA complexes were toxic to the cellswith mean cell viabilities of 50.6±6.6% and 53.1±6.1%, res-pectively, and all of the other polymer/DNA complexes, in-cluding PAMAM-Arg/DNA, showed low toxicity (Fig. 1C).This finding agrees well with our previous report on the lowcytotoxicity of PAMAM-Arg in Neuro 2A cells [20]. Takentogether these results indicate that PAMAM-Arg is an excellentcationic polymer for gene delivery in primary cortical culturesand that its gene delivery efficiency is superior to that ofPAMAM dendrimer, BPEI, or Lipofectamine.

3.3. Transfection efficiency of PAMAM-Arg in neurons inprimary cortical culture

To evaluate transfection efficiencies at the cell level, GFPexpression was examined by confocal microscopy 48 h after thetransfection with pEGFP-N1/PAMAM-Arg, pEGFP-N1/PAMAM or pEGFP-N1/BPEI complex. Neurons were stainedwith a monoclonal antibody reactive to Neu N (Fig. 2A, D, G)and GFP expression was visualized by fluorescent microscopy(Fig. 2B, E, H). In cells transfected with pEGFP-N1/PAMAM-Arg, GFP expression was detected in more than 35% of Neu N-positive cells (35.1±4.5%; Fig. 2C, J), which was significantly

PAMAM-Arg, PAMAM, or BPEI. Primary cortical cells were transfected with(D,E,F) or BPEI (G,H,I). At 48 h post-transfection, cells were washed, fixed ind by immunostaining neurons with anti-Neu N antibody (A,D,G) and EGFPimages. The numbers of Neu N-positive cells and GFP-positive cells counted inata represent averages±SEM. ⁎ p<0.05, ⁎⁎ p<0.01. Scale bar presents 50 μm.

GENEDELIVERY

114 J.-B. Kim et al. / Journal of Controlled Release 114 (2006) 110–117

higher than that obtained by PAMAM-or BPEI-mediated trans-fection (10.9±8.2% and 14.3±6.3%, respectively; Fig. 2F, I, J).There was almost no visible change in cell morphology. Theseresults indicate that PAMAM-Arg is a highly efficient DNAcarrier in primary cortical cultures.

3.4. Transfection efficiency of PAMAM-Arg in glial cells inprimary cortical culture

Primary cortical cultures are a neuron-enriched mixed cul-ture consisting of 70–85% neurons, 15–30% astrocytes, 1–2%microglia, and 0.1–1% oligodendroglia [26], which makesthem useful for studying the characteristics of neurons in a glialcell background. Hence, efficient transfection in constitutingglial cells is essential. To evaluate the transfection efficiency toglial cell, GFP expression in glial cells was examined byfluorescent microscopy 48 h after transfection with pEGFP-N1/PAMAM-Arg complexes. Cell types were identified usingantibodies for cell type-specific markers, i.e., GFAP forastrocytes (Fig. 3B), CD45 for microglia (Fig. 3E), and RIPfor oligodendrocytes (Fig. 3H). PAMAM-Arg-mediated GFPexpression was detected in astrocytes, which are grown as afeeder layer in primary cortical cultures (Fig. 3A, B, C). GFPexpression was also observed in microglia (Fig. 3D, E, F) andoligodendrocytes (Fig. 3G, H, I), although they representedonly a minor population in primary cortical cultures. Theseresults demonstrate that PAMAM-Arg is an effective genecarrier not only to neurons but to all three glial cells present inprimary cortical culture.

Fig. 3. Transfection of glial cells in primary cortical cultures with a GFP-expressing pN1 (1 μg DNA per 4×105 cells/well) using PAMAM-Arg. At 48 h post-transfection,Triton. Images obtained by immunostaining astrocytes, microglia, and oligodendrocytdue to EGFP fluorescence (A,D,G) are presented. The right panels are overlays of t

3.5. Down-regulation of gene expression using a shRNA-expressing plasmid/PAMAM-Arg complex in primary corticalculture

Recently, RNAi has emerged as a powerful genetic tool forsilencing gene expression in multiple organisms, and it has beenwidely used to study gene function and to investigate the effects ofvarious treatments [27–31]. Successful application of RNAi tomammalian cells is dependent on efficient siRNA delivery. This isa problem particularly for primary neuronal cells, since it isdifficult to transfect these cells using cationic polymers. siRNAscan be delivered directly or indirectly via plasmids designed tosynthesize intermediates, such as short-hairpin RNAs (shRNAs),which are subsequently processed into siRNAs.

In order to investigate the potential of PAMAM-Arg-mediatedtransfection in primary cortical culture, shRNA-mediated geneknock-down experiments were performed using HMGB1 as amodel target gene. HMGB1 is a nonhistone DNA-binding protein[32], and widely expressed in various tissues, including brain. Aplasmid expressing the shRNA of the HMGB1 gene was con-structed. The hairpin-forming 64 bp insert in the plasmid harborsthe inverted repeat of 19-bp corresponding to the HMGB1 codingregion, which is separated by a 7 bp spacer (shHMGB1-pU6) (Fig.4A). A mutant shHMGB1-expressing plasmid containing the sub-stitution of 6 nucleotides within the 19-bp region was also con-structed (MshHMGB1-pU6). When shHMGB1-pU6 was co-transfected into primary neuronal cells with pEGFP-N1 plasmid,which was used as a transfection control, the marked HMGB1suppression was detected in the GFP-positive cells. Over 75% of

lasmid using PAMAM-Arg. Primary cortical cells were transfected with pEGFP-cells were washed, fixed in 2% paraformaldehyde, and permeabilized with 0.1%es with anti-GFAP, anti-CD45 and anti-RIP antibodies, respectively (B,E,H), andhe two preceding images. Scale bar presents 50 μm.

Fig. 4. HMGB1 shRNA transgene and PAMAM-Arg-mediated transfection in primary cortical cultures. (A) Schematic diagram of HMGB1 shRNA transgene showingthe sense and antisense regions that target the HMGB1 gene. (B) MTT assays were performed 48 h after transfection with shHMGB1-pU6 or MshHMGB1-pU6 usingPAMAM-Arg. Fluorescence photographs prepared 48 h after co-transfection with pd2EGFP-N1 and shHMGB1-pU6 (C–E). The images obtained afterimmunostaining with anti-HMGB1 antibody (C,F) and of GFP fluorescence (D,G) are presented. The right panels are overlays of the two preceding images.Percentages of GFP-positive and shHMGB1-pU6-positive cells were determined by count 20 photographic fields in three independent experiments (K). Arrowsindicate GFP-positive cells showing reduced HMGB1 expression (D). Data represent averages±SEM. ⁎ p<0.05, ⁎⁎ p<0.01. Scale bar presents 50 μm.

GENEDELIVERY

115J.-B. Kim et al. / Journal of Controlled Release 114 (2006) 110–117

GFP-positive cells were HMGB1-negative, indicating thatshHMGB1-pU6 transfection had suppressed HMGB1 expression(Fig. 4C, D, E and I). In contrast, no suppression of HMGB1 wasdetected in theGFP-positive cells co-transfectedwithMshHMGB1(Fig. 4F, G, H and I), although its transfection efficiency asevidenced by GFP-positive cell numbers was comparable toshHMGB1-pU6 (Fig. 4D, G and I). The above demonstratesthat the PAMAM-Arg-mediated transfection of shRNA-expressing plasmid effectively and specifically inhibited targetgene expression.

Fig. 5. Suppression of HMGB1 expression mediated by shHMGB1-pU6/PAMAM-AHMGB1 RNA in cells transfected with shHMGB1-pU6 or MshHMGB1-pU6 usingData represent averages±SEM. ⁎ p<0.05, ⁎⁎ p<0.01.

To determine the level of target gene suppression, shHMGB1-pU6 was transfected into primary neuronal cells using PAMAM-Arg, PAMAMor BPEI, and theHMGB1mRNAwas quantified byRT-PCR 48 h after transfection. Transfection of shHMGB1-pU6resulted in the suppression of the HMGB1 mRNA (61.5± 3.7% ofthe control level), whereas transfection with MshHMGB1-pU6showed no suppressive effect (98.5±5.3%) (Fig. 5A, B). Whennative PAMAM or BPEI was used as a gene carrier, shHMGB1-pU6 also reduced theHMGB1mRNA level but to only 87.7±4.7%or 86.6±5.5% of the control, respectively (Fig. 5C, D). In all cases,

rg complex in primary cortical cultures. The expression levels of endogenousPAMA-Arg (A,B), PAMAM (C,D), or BPEI (E,F) were evaluated by RT-PCR.

GENEDELIVERY

116 J.-B. Kim et al. / Journal of Controlled Release 114 (2006) 110–117

however, GAPDH expression was not changed by the shHMGB1-pU6/PAMAM-Arg complex. These results indicate that thePAMAM-Arg-mediated expression of shHMGB1-pU6 suppressedendogenous HMGB1 expression, and that this suppression wasefficient and specific. The results of this study provide a basis forfuture research upon the use of RNAi in primary neuronal cells.

The present study demonstrates that PAMAM-Arg has ahigher transfection efficiency in primary cortical cultures than avariety of other widely used methods. Moreover, this level ofenhancement in primary cortical cultures was significantlyhigher than those observed in other cell lines, such as inHepG2, 293 cells, or primary rat aorta smooth muscle cells,wherein PAMAM-Arg and BPEI exhibited comparable trans-fection efficiencies [20]. Interestingly, PAMAM-Arg was foundto increase transfection efficiency particularly in Neuro 2A[20], implying that PAMAM-Arg is more effective in neuronalcells. In addition, efficient transfection in all three glial cellspresent in primary cortical cultures indicated that PAMAM-Argshould be widely applicable for various studies in primarycortical cultures.

Although the mechanism underlying the higher efficacy ofPAMAM-Arg in primary cortical cultures needs to be explored,we speculate that the high transfection efficiency of PAMAM-Argin brain cells might be related to its multiple arginine residues.Peptide signals rich in basic amino acids like arginine, which arefound in protein transduction domains (PTD) or membranetranslocational signals (MTS), such asTAT-derived peptides, havebeen known to enhance the translocation efficiencies ofbiologically active materials [33,34]. However, significantlylower transfection efficiency of poly-L-arginine alone indicatedthat increased gene expression might be attributed to thelocalization of the arginine residues on the surface of PAMAM-Arg/DNA complexes, which presumably facilitate uptake ornuclear localization. It is of interest that efficient exogenous genedelivery was achieved in rat brain after the localized administra-tion of PAMAM-Arg/DNA complex (Kim et al., submitted).Furthermore, efficient target gene silencing was also detected inrat brain due to PAMAM-Arg-mediated shRNA expression (Kimet al., submitted). Together with its low in vivo cytotoxicity, theseresults imply that PAMAM-Argmay prove to be an excellent genecarrier in the central nervous system (CNS).

4. Conclusion

PAMAM-Arg, an arginine-grafted PAMAM dendrimer,achieved transfection levels of 35–40% cells in primary corticalcultures, which is significantly higher than the levels achievedwith commercially available reagents, such as Lipofectamine,BPEI, and PAMAM. Efficient transfection was obtained inneurons and also in all CNS glial cells: astrocytes, microglia,and oligodendrocytes. Moreover, transfection of a shRNA-ex-pressing plasmid with PAMAM-Arg efficiently suppressedtarget gene expression, and PAMAM-Arg had no observablecytotoxic effect on primary neuronal cells. Thus, PAMAM-Arg,a novel non-viral gene carrier, presents a means of transfectingprimary neuronal cells. Further in vivo experiments are nowunderway, in which transfection of shRNA-expressing plasmid

complexed with PAMAM-Arg suppressed target gene expres-sion by 30–40% (Kim et al., unpublished observation), indi-cating efficient transfection rate of PAMAM-Arg in vivo.

Acknowledgment

This work was financially supported by the Ministry ofScience and Technology in Republic of Korea (2005-01096) forJ-K Lee.

References

[1] M. Berry, L. Barrett, L. Seymour, A. Baird, A. Logan, Gene therapy forcentral nervous system repair, Curr. Opin. Mol. Ther. 3 (2001) 338–349.

[2] K.R. Clark, T.J. Sferra, W. Lo, G. Qu, R.J. Chen, P.R. Johnson, Genetransfer into the CNS using recombinant adeno-associated virus: analysisof vector DNA forms resulting in sustained expression, J. Drug Target. 7(1999) 269–283.

[3] S.M. Hughes, H. Moussavi, S.L. Sauter, B.L. Davidson, Viral-mediatedgene transfer to mouse primary neural progenitor cells, Mol. Ther. 5 (2002)516–524.

[4] L.A. King, K.A. Mitrophanous, L.A. Clark, V.N. Kim, J.B. Rohll, A.J.Kingsman, R.J. Colello, Growth factor enhanced retroviral gene transfer tothe adult central nervous system, Gene Ther. 7 (2000) 1103–1111.

[5] G.Martino, R. Furlan, F. Galbiati, P.L. Poliani, A. Bergami, L.M.E.Grimaldi,L.Adorini, G. Comi,A gene therapy approach to treat demyelinating diseasesusing nonreplicative herpetic vectors engineered to produce cytokines, Mult.Scler. 4 (1998) 222–227.

[6] B.K. Kaspar, D. Erickson, D. Schaffer, L. Hinh, F.H. Gage, D.A. Peterson,Targeted retrograde gene delivery for neuronal protection, Mol. Ther. 5(2002) 50–56.

[7] X.O. Breakefield, N.A. DeLuca, Herpes simplex virus for gene delivery toneurons, New Biol. 3 (1991) 203–218.

[8] B.K. Jin, M. Belloni, B. Conti, H.J. Federoff, R. Starr, J.H. Son, H. Baker,T.H. Joh, Prolonged in vivo gene expression driven by a tyrosinehydroxylase promoter in a defective herpes simplex virus amplicon vector,Hum. Gene Ther. 7 (1996) 2015–2024.

[9] K. Jooss, N. Chirmule, Immunity to adenovirus and adeno-associated viralvectors: implications for gene therapy, Gene Ther. 10 (2003) 955–963.

[10] M.S. Lawrence, H.G. Foellmer, J.D. Elsworth, J.H. Kim, C. Leranth, D.A.Kozlowski, A.L. Bothwell, B.L. Davidson, M.C. Bohn, D.E. Redmond,Inflammatory responses and their impact on beta-galactosidase transgeneexpression following adenovirus vector delivery to the primate caudatenucleus, Gene Ther. 6 (1999) 1368–1379.

[11] M.M. McMenamin, A.P. Byrnes, H.M. Charlton, R.S. Coffin, D.S.Latchman, M.J. Wood, A gamma34.5 mutant of herpes simplex 1 causessevere inflammation in the brain, Neuroscience 83 (1998) 1225–1237.

[12] M. Meichsner, T. Doll, D. Reddy, B. Weisshaar, A. Matus, The lowmolecular weight form of microtubule-associated protein 2 is transportedinto both axons and dendrites, Neuroscience 54 (1993) 873–880.

[13] Z. Xia, H. Dudek, C.K. Miranti, M.E. Greenberg, Calcium influx via theNMDA receptor induces immediate early gene transcription by a MAPkinase/ERK-dependent mechanism, J. Neurosci. 16 (1996) 5425–5436.

[14] A. Tabuchi, K. Sano, R. Nakaoka, C. Nakatani, M. Tsuda, Inducibility ofBDNF gene promoter I detected by calcium-phosphate mediated DNAtransfection is confined to neuronal but not to glial cells, Biochem.Biophys. Res. Commun. 253 (1998) 818–823.

[15] S.Y. Watanabe, A.M. Albsoul-Younes, T. Kawano, H. Itoh, Y. Kaziro, S.Nakajima,Y.Nakajima,Calciumphosphate-mediated transfection of primarycultured brain neurons using GFP expression as a marker: application forsingle neuron electrophysiology, Neurosci. Res. 33 (1999) 71–78.

[16] H. Li, S.T.H. Chan, F. Tang, Transfection of rat brain by electroporation,J. Neurosci. Methods 75 (1997) 29–32.

[17] S. Jiao, L. Cheng, J.A. Wolff, N.S. Yang, Particle bombardment-mediatedgene transfer and expression in rat brain tissues, Biotechnology (NY) 11(1993) 497–502.

GENEDELIVERY

117J.-B. Kim et al. / Journal of Controlled Release 114 (2006) 110–117

[18] E.C. Ohki, M.L. Tilkins, V.C. Ciccarone, P.J. Price, Improving thetransfection efficiency of post-mitotic neurons, J. Neurosci. Methods 112(2001) 95–99.

[19] M. Guerra-Crespo, J.L. Charli, V.H. Rosales-Garcia, G. Pedraza-Alva, L.Perez-Martinez, Polyethylenimine improves the transfection efficiency ofprimary cultures of post-mitotic rat fetal hypothalamic neurons, J. Neurosci.Methods 127 (2003) 179–192.

[20] J.S. Choi, K. Nam, J.Y. Park, J.B. Kim, J.K. Lee, J.S. Park, Enhancedtransfection efficiency of PAMAMdendrimer by surface modification withL-arginine, J. Control. Release 99 (2004) 445–456.

[21] J.-B. Kim, C.S. Piao, K.-W. Lee, P.-L. Han, J.I. Ahn, Y.S. Lee, J.-K. Lee,Delayed genomic responses to transient middle cerebral artery occlusion inthe rat, J. Neurochem. 89 (2004) 1271–1282.

[22] M.J. Lee, S.S. Cho, J.R. You, Y. Lee, B.D. Kang, J.S. Choi, J.W. Park, Y.L.Suh, J.A. Kim, D.K. Kim, J.S. Park, Intraperitoneal gene delivery mediatedby a novel cationic liposome in a peritoneal disseminated ovarian cancermodel, Gene Ther. 9 (2002) 859–866.

[23] Y.-M. Yu, P.-L. Han, J.-K. Lee, JNK pathway is required for retinoic acid-induced neurite outgrowth of human neuroblastoma, SH-SY5Y, NeuroRe-port 14 (2003) 941–945.

[24] D. Fischer, Y. Li, B. Ahlemeyer, J. Krieglstein, T. Kissel, In vitrocytotoxicity testing of polycations: influence of polymer structure on cellviability and hemolysis, Biomaterials 24 (2003) 1121–1131.

[25] R. Jevprasesphant, J. Penny, R. Jalal, D. Attwood, N.B. McKeown, A.D'Emanuele, The influence of surface modification on the cytotoxicity ofPAMAM dendrimers, Int. J. Pharm. 252 (2003) 263–266.

[26] C. Sanfeliu, J.M. Wright, S.U. Kim, Neurotoxicity of isoniazid and itsmetabolites in cultures of mouse dorsal root ganglion neurons and hybridneuronal cell line, Neurotoxicology 20 (1999) 935–944.

[27] H. Aoki, M. Satoh, K. Mitsuzuka, A. Ito, S. Saito, T. Funato, M. Endoh, T.Takahashi, Y. Arai, Inhibition of motility and invasiveness of renal cellcarcinoma induced by short interfering RNA transfection of beta1,4GalNAc transferase, FEBS Lett. 567 (2004) 203–208.

[28] Y.Y. Cho, A.M. Bode, H. Mizuno, B.Y. Choi, H.S. Choi, Z. Dong, A novelrole for mixed-lineage kinase-like mitogen-activated protein triple kinasealpha in neoplastic cell transformation and tumor development, CancerRes. 64 (2004) 3855–3864.

[29] Y. Takei, K. Kadomatsu, Y. Yuzawa, S. Matsuo, T. Muramatsu, A smallinterfering RNA targeting vascular endothelial growth factor as cancertherapeutics, Cancer Res. 64 (2004) 3365–3370.

[30] J.M. Jacque, K. Triques, M. Stevenson, Modulation of HIV-1 replicationby RNA interference, Nature 418 (2002) 435–438.

[31] A.P. McCaffrey, H. Nakai, K. Pandey, Z. Huang, F.H. Salazar, H. Xu, S.F.Wieland, P.L. Marion, M.A. Kay, Inhibition of hepatitis B virus in mice byRNA interference, Nat. Biotechnol. 21 (2003) 639–644.

[32] D. Landsman, M. Bustin, A signature for the HMG-1 box DNA-bindingproteins, Bioessays 15 (1993) 539–546.

[33] S.R. Schwarze, A. Ho, A. Vocero-Akbani, S.F. Dowdy, In vivo proteintransduction: delivery of a biologically active protein into the mouse,Science 285 (1999) 1569–1572.

[34] C.H. Tung, R. Weissleder, Arginine containing peptides as deliveryvectors, Adv. Drug Deliv. Rev. 55 (2003) 281–294.