Oligomers of the Arginine-rich Motif of the HIV-1 TAT Protein Are Capable of Transferring Plasmid DNA into Cells* 210

We constructed multimers of the TAT-(47–57) peptide. This polycationic peptide is known to be a protein and particle transduction domain and at the same time to comprise a nuclear localization function. Here we show that oligomers of the TAT-(47–57) peptide compact plasmid DNA to nanometric particles and stabilize DNA toward nuclease degradation. At optimized vector compositions, these peptides mediated gene delivery to cells in culture 6–8-fold more efficiently than poly-l-arginine or the mutant TAT2-M1. When DNA was precompacted with TAT peptides and polyethyleneimine (PEI), Superfect, or LipofectAMINE was added, transfection efficiency was enhanced up to 390-fold compared with the standard vectors. As early as after 4 h of transfection, reporter gene expression mediated by TAT-containing complexes was higher than the 24-h transfection level achieved with a standard PEI transfection. When cells were cell cycle-arrested by serum starvation or aphidicolin, TAT-mediated transfection was 3-fold more efficient than a standard PEI transfection in proliferating cells. In primary nasal epithelial cells and upon intratracheal instillation in vivo, TAT-containing complexes were superior to standard PEI vectors. These data together with confocal imaging of TAT-DNA complexes in cells support the hypothesis that the TAT nuclear localization sequence function is involved in enhancing gene transfer.

Poor escape of nonviral gene vectors from the endosomal compartment after cellular uptake and inefficient translocation into the nucleus substantially limit their efficiency (1,2). In this study, the arginine-rich motif of the HIV-1 1 TAT protein should be used to overcome these obstacles. The 101-amino acid HIV-1 TAT protein regulates transcriptional activation of the human immunodeficiency virus type 1 long terminal repeat promoter element by binding to a short nascent stem-bulgeloop leader RNA, trans-activation response (TAR), recruiting a positive transcription elongation complex (P-TEFb) (3). Binding of HIV-1 TAT protein to the TAR RNA is substantially mediated through the arginine-rich motif (amino acids 47-57), which represents a basic stretch of amino acids located to domain 4 of the HIV-1 TAT protein (3). Besides its importance for binding to the TAR RNA, the arginine-rich motif of the HIV-1 TAT protein (TAT peptide) has been shown to function as a protein transduction domain, penetrating cell membranes in a manner different from classic endocytosis (4 -6). This uptake mechanism was also observed for heterologous proteins when they were chemically coupled (7) or genetically fused (8) to the TAT peptide. In addition, the conjugation of the TAT peptide to various structures of nanometric size, such as superparamagnetic nanoparticles (9), liposomes (10), and phage (11), led to their uptake into cells in a manner apparently different from endocytosis. Besides this unique feature of the TAT-peptide, the TAT peptide represents a nuclear localization sequence (12). Upon binding of the TAT peptide itself or its conjugates to the nucleocytoplasmic shuttle protein importin ␤, nuclear import through the nuclear pore has been observed (4, 6 -10).
The objective of the present study was to examine whether the unique features of the TAT peptide (protein transduction domain and nuclear localization sequence) and its polycationic nature are suitable to enhance nonviral gene delivery. Chemical properties of cationic polymers, such as the degree of polymerization, the type of cationic groups present in the polymer, or the amino acid composition of cationic oligopeptides influence the biophysical properties of their polyelectrolyte complexes with plasmid DNA. DNA binding affinity, surface charge, and particle size have pronounced effects on their efficacy in gene delivery and on biodistribution in vivo (13)(14)(15). Since compaction of DNA into stable microparticulate structures by oligopeptides suitable for gene transfer requires a minimum chain length of 6 -10 cationic amino acids (13), we synthesized the TAT peptide as oligomers. The dimer, trimer, and tetramer of identical repeats with intervening glycine residues comprise 16,24, and 32 positive charges, respectively.
We attempted to find a correlation between the degree of oligomerization of the TAT peptide, the resulting biophysical characteristics of peptide-DNA complexes, and their gene delivery efficiency in context with the unique functions of the TAT peptide.

Peptide Synthesis
The following peptides were synthesized. Peptide C(YGRK-KRRQRRRG) 2-4 (TAT 2-4 ) contained the arginine-rich motif of the human immunodeficiency virus type 1 TAT protein. Peptides were synthesized on an Applied Biosystems 431 A automatic synthesizer according to a standard Fmoc (N-(9-fluorenyl)methoxycarbonyl) proto-* This work was supported by Deutsche Forschungsgemeinschaft Grants Ro994/2-1. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.  1 The abbreviations used are: HIV-1, human immunodeficiency virus, type 1; TAR, trans-activation response; HBS, Hepes-buffered saline; FCS, fetal calf serum; pLa, poly-L-arginine; PEI, polyethyleneimine; RLU, relative light units.
Plasmid pCMV-Luc containing firefly luciferase cDNA driven by the cytomegalovirus promoter was generously provided by Dr. E. Wagner (Department of Pharmacy, Ludwig-Maximilians-University Munich) and amplified as described in Ref. 16.

Size and Potential Measurement
The particle sizes were determined by dynamic light scattering, and potentials were measured electrophoretically (Zetasizer 3000HS; Malvern Instruments, Herrenberg, Germany). Gene vector solutions in distilled water or HBS were generated at a DNA concentration of 10 g/ml (17).

Fluorescence Quenching Assay
Plasmid DNA (pCMVLuc) was labeled with TOTO-1 (dye/base pair ratio of 1:20). The TAT oligomers were serially diluted in HBS in a 96-well plate corresponding to the indicated charge ratios. 100 l of a solution of TOTO-1-labeled DNA (0.25 g of DNA in HBS) was pipetted to either 100 l of TAT oligomer solution or 100 l of HBS (100%) and thoroughly mixed. The excitation filter was set at 485 nm, and the emission filter was set at 535 nm (SPECTRAFluor Plus, Tecan, Germany). Measurements were performed in quadruples.

Cell Culture
Human bronchoepithelial cells (16HBE14o Ϫ ) were provided by Dieter C. Gruenert (University of Vermont, Burlington, VT). Cells were grown in FCS (10%)-supplemented minimum essential medium (Invitrogen GmbH, Karlsruhe, Germany) at 37°C in a 5% CO 2 humidified air atmosphere. COS-7 cells were cultivated in FCS (10%)-supplemented Dulbecco's modified Eagle's medium (Invitrogen). Fresh nasal respiratory epithelium was obtained from patients through surgery and immediately processed as follows. Briefly, epithelial cells were detached from connective tissue after 1-h incubation at 37°C in the presence of dispase II (1 unit/ml phosphate-buffered saline; Roche Diagnostics) and then filtered and centrifuged (150 ϫ g, 5 min, ambient temperature). The pellet was resuspended in 8 ml of RPMI plus 10% FCS containing antibiotics, and cells were seeded into 35-mm dishes. The next day, the medium was replaced by bronchial epithelial cell growth medium (catalog no. C21160; Promocell).

Preparation of Gene Vector Complexes
Gene vector complexes for one well were formulated as follows. 1 g of DNA and the corresponding amount of vector were diluted in HBS (150 mM NaCl, 10 mM HEPES, pH 7.4) to 75 l, respectively. The DNA solution was pipetted to the vector solution and mixed vigorously by pipetting up and down. The complexes were incubated for 20 min at ambient temperature before use. Ternary gene vector complexes for one well were generated in the same manner, but DNA, TAT oligomer, and standard cationic transfection agent (PEI, average molecular mass of 25 kDa; Aldrich, Deisenhofen, Germany; dialyzed against water, 12-14-kDa molecular mass cut-off and adjusted to pH 7; fractured dendrimers; SuperFect (Qiagen, Hilden, Germany) or LipofectAMINE (Invitrogen, GmbH) were diluted in HBS to 50 l, respectively. The DNA solution was pipetted to the TAT oligomer or poly-L-arginine (pLa; Sigma catalog no. P 4663; M r 5000 -15,000) solution, mixed vigorously, and incubated at ambient temperature for 10 min, and then the standard cationic vector solution was added, again incubated at ambient temperature for 10 min. Alternatively, DNA was first pipetted to the standard vector solution, and the TAT oligomer solution was then added.

Transfection Procedure and Luciferase Activity Measurement
150 l of gene vector solution, corresponding to 1 g of DNA, were pipetted onto cells (COS-7, 30,000 cells/well; 16HBE, 100,000 cells/ well), which had been seeded into 24-well plates 1 day before, and then covered with 850 l of medium in the absence of FCS. After 4 h of incubation at 5% CO 2 and 37°C, the medium was replaced with 10% FCS-containing medium supplemented with 0.1% (v/v) penicillin/streptomycin and 0.5% (v/v) gentamycin (Invitrogen). When transfections were performed at 4°C, cells were incubated for 1 h at 4°C before transfection and incubation with the gene vectors was performed at 4°C. When transfections were performed in the presence of a mixture of endocytosis inhibitors, the transfection medium was supplemented with antimycin A (1 g/ml; Sigma), sodium fluoride (10 mM; Sigma), and sodium azide (0.1% (m/v); Sigma), respectively. After 4 h, the medium was replaced with 10% FCS-containing medium supplemented with antibiotics (see above). Transfections of growth-arrested cells were performed as follows. Cells were incubated for 24 h in the absence of FCS (serum starvation) or subjected to incubation followed by a 12-h incubation period with aphidicolin (25 M, with FCS) before transfection. Transfections were then performed either in the presence of aphidicolin (25 M) followed by further incubation of the cells for 24 h in aphidicolin-supplemented medium or in the absence of FCS. 24 h later, cells were lysed by the addition of 200 l of lysis buffer (250 mM Tris, 0.1% Triton X-100, pH 7.8) per well, and luciferase activity was measured (Promega kit, Lumat LB 9507; Berthold, Bad Wildbach, Germany). Protein content was determined by a standard Bio-Rad protein assay (Bradford method).

Fluorescence in Situ Hybridization and Confocal Laser-scanning Microscopy
A three-dimensional fluorescence in situ hybridization was done following the protocol of Solovei and Cremer (18) with several modifications for our specific needs.
Probe Generation-A DNA probe was generated from the pEGFP plasmid (Clontech) by nick translation. A digoxigenin hapten was inserted with Dig Nick Translation Mix (Roche Diagnostics) generating a probe with a 300-bp median length. The reaction mixtures were cleaned of small fragments and nucleotides with the Qiaquick PCR Purification Kit (Qiagen, Hilden, Germany). 1 g of probe was precipitated with 50 g of salmon testes DNA (Sigma). The dried pellet was reconstituted in 20 l of formamide at 37°C for 2 h and stored until use at Ϫ20°C.
Hybridization-The probe was brought to 37°C and mixed 1:1 with "Fish Mix" (4ϫ SSC, 40% dextran sulfate), resulting in a hybridization solution of 25 ng/l probe in 2ϫ SSC, 20% dextran sulfate, 50% formamide. The hybridization solution was denatured at 75°C for 5 min and   (nm) and potential (mV) in water and HBS Gene vector complexes were prepared in double-distilled water or HBS at the indicated charge ratios, and the particle size (nm) or potential (mV) was measured. briefly kept at 37°C prior to slide application. The slides were removed one at a time from the storage solution, 50 l of hybridization solution was placed on each slide, a coverslip was applied, and finally rubber cement was used as a sealant. The sealed slides were heated to 75°C for 3 min to denature the cellular DNA, also further denaturing the probe DNA. The slides were placed in a moist chamber within a 37°C oven and were hybridized for 2-3 days. Image Acquisition-Light optical sections were generated for each nucleus with a three-channel confocal laser-scanning microscope (TCS 4D; Leica Inc., Deerfield, IL) equipped with a Plan Apo 63ϫ/1.32 oil immersion lens. Using the 488 and 567 lines of an argon/krypton laser for visualization of the Sytox and rhodamine, signals respectively, stacks of 256 ϫ 256 equidistant 8-bit grayscale images were generated at an axial distance of 250 nm, pixel size of 1 nm. Each series consisted of ϳ20 -25 images. Lines were averaged eight times. The different fluorochromes were imaged sequentially in identical nuclear planes. Laser power and voltage for the rhodamine channel were maintained at the same level for all conditions, whereas the voltage of the Sytox channel was adjusted for variations in nuclear counterstain. Cells were randomly selected when a signal appeared to be in the cell, and the midnuclear sections were used to detect the distribution of plasmid DNA.

Animals and Delivery of Gene Vectors to the Lung
Ternary gene vector complexes used for in vivo experiments were generated as described above, but the TAT oligomer and DNA solution were diluted in double-distilled water (Fresenius AG, Bad Homburg, Germany), respectively. 60 l of gene vector solution containing 20 g of DNA were applied per mouse. The gene vector application was performed as described in Ref. 16. In brief, mice were anesthetized intraperitoneally with pentobarbital and directly intubated with a single 60-l bolus of the indicated gene vector using a 22-gauge intravascular cannula sleeve, needle removed (25 mm, 0.9-mm outer diameter, 0.6-mm inner diameter; Baxter, Germany). At 24 h post-transfection, mice were anesthetized intraperitoneally with pentobarbital, and mice peritoneum were opened by midline incision. In order to wash blood from the lungs and to avoid interference with the subsequent luciferase assay, a posterior vena cava exit was cut, and 1 ml of an isotonic sodium chloride solution was slowly perfused into the mice right cardiac ventricle. The lungs were dissected from animals, frozen in liquid nitrogen, and stored at Ϫ70°C. At assay time, the tissue was thawed on ice, and 500 l of ice-cold lysis buffer (25 mM glycylglycine, 15 mM magnesium sulfate, 4 mM EDTA, 0.1% Triton X-100 (m/V), 1 mM phenylmethylsulfonyl fluoride, 0.15 units/ml buffer aprotinin) was added to each sample and homogenized for 20 s using a Polytron Pt 2100 homogenizer (level 5 corresponding to 26,000 rpm; Kinematica, Litau/Luzern, Switzerland). Samples were centrifuged at 10,000 ϫ g at 4°C for 10 min, and 40 l of the supernatant were measured for luciferase activity in a Lumat LB 9507 instrument (Berthold) injecting 100 l of luciferase reagent (Promega) to each sample, and the light emitted over 10 s was measured. The background was subtracted from the reported values. 1 ϫ 10 6 relative luciferase units/10 s correspond to 1.25 ng of luciferase. All animal procedures were approved and controlled by the local ethics committee and carried out according to the guidelines of the German law of protection of animal life.

Statistical Analysis
Results are reported as means Ϯ S.D. The statistical analysis between different groups has been determined with a nonpaired t test. p Յ 0.05 was considered significant. All statistical analyses were performed using the program StatView 5.0. (SAS Institute Inc., Cary, NC).

Biophysical Properties of the TAT Oligomer Gene Vectors-
The condensation of DNA into particulate structures is a prerequisite for gene delivery. For this reason, we first examined the capability of the TAT oligomers to condense DNA and to form particulate complexes. Laser light scattering showed that all of the TAT oligomer gene vectors had a diameter of ϳ70 nm in water, which increased to 700 nm in HBS at charge ratios of Ϯ1 and 5, respectively (Table I). pLa (control) gene vectors showed a similar diameter. TAT oligomer and pLa gene vectors had a positive potential of approximately ϩ40 mV in water and negative potential of approximately Ϫ20 mV at Ϯ1 in HBS, which changed into a positive potential of 20 mV at Ϯ 5 and slightly increased with the degree of oligomerization (Table I). The extent of complexation was independent of the degree of polymerization as examined by electrophoretic mobility shift assay and DNase I protection assay (see Supplementary Material). The only difference among TAT oligomer gene vectors depending on the oligomer size could be observed when complexes were analyzed by a fluorescence quenching assay (Fig. 1). Fluorescence quenching strongly increased at low charge ratios, reaching a plateau at approximately Ϯ2. Higher charge ratios only slightly increased further DNA condensation. Fluorescence quenching induced by pLa was shifted to a higher charge ratio compared with the TAT peptides. Fluorescence quenching at a given charge ratio increased with the degree of oligomerization, indicating that the DNA was more tightly packed in the complex when the size of the TAT oligomer increased.
Transfection Efficiency of TAT Oligomer Gene Vectors-Plasmid DNA was complexed with each of the TAT oligomers at different charge ratios and transfection efficiency was examined in vitro. In addition, pLa gene vectors were prepared under the same conditions as control to assess the possibility of a sequence-dependent process ( Fig. 2A). For each charge ratio tested, a significant improvement of transfection by TAT 2 and TAT 3 gene vectors compared with pLa gene vectors was observed. No statistical difference between TAT 4 and pLa gene vectors was observed. The transfection rates mediated by each of the oligomers did not correlate with the degree of oligomerization of the TAT peptide. Intermediate length of the TAT 3oligomer mediated the highest level of transgene expression. To further characterize sequence dependence, gene transfer efficiency mediated by the nuclear transport-deficient TAT 2 -M1 peptide (12) was examined. Gene transfer efficiency mediated by the TAT 2 -M1 peptide (Ϯ charge ϭ 10) was 6-fold lower as compared with the TAT 2 peptide (Ϯ charge ϭ 10, Fig. 2C, p Ͻ 0.01).

The Influence of Endocytosis on Transfection Efficiency of TAT Oligomer Gene Vectors and Comparison with Standard
Cationic Transfection Agents-To examine endocytosis as an uptake mechanism, transfections were performed in the presence of chloroquin (19), at 4°C (6,10,11) or in the presence of metabolic inhibitors (antimycin A, sodium azide, and sodium fluoride) (20 -22). For each of the TAT oligomer gene vectors, transfection efficiencies in the presence of chloroquine strongly increased (Fig. 2B; compare with Fig. 2A). Gene transfer mediated by TAT 2 and TAT 3 gene vectors was significantly higher as compared with pLa gene vectors. This was not the case for TAT 4 vectors. Overall transfection efficiency increased in the following order: TAT 2 Ͼ TAT 3 Ͼ TAT 4 . In the presence of chloroquine, the transfection efficiency of TAT 2 gene vectors was enhanced 40-fold, whereas the transfection efficiency of pLa gene vectors was only enhanced 3-fold. Transfections performed at 4°C led to a 20 -90-fold decrease of transgene expression (Fig. 2D). Transfection rates in the presence of the endocytosis inhibitors decreased even more (Fig. 2E). Transgene expression mediated by TAT 3 -DNA complexes in the presence of chloroquine was 3-and 4-fold higher as compared with PEI and SuperFect polyplexes and at the same level as Lipo-fectAMINE lipoplexes (Fig. 2F). These data suggest that the cellular uptake mechanism of the TAT oligomer gene vector complexes was predominated by the endosomal pathway and could not be precisely differentiated from endocytosis. However, due to the much stronger increase of gene transfer of the TAT oligomer complexes in the presence of chloroquin, as compared with pLa complexes, one could suggest that gene transfer of the TAT oligomers, in particular TAT 2 , was peptide sequencedependent. This is supported by the higher gene transfer efficiency of the TAT 2 gene vectors as compared with the TAT 2 -M1 gene vectors (Fig. 2C).
Intracellular Localization of Plasmid DNA after Transfection with TAT Oligomer Gene Vectors-To further investigate the localization of plasmid DNA electrostatically bound to each of the TAT oligomers within the cell, COS-7 cells were transfected with TAT oligomer polyplexes or with pLa polyplexes, and plasmid DNA distribution was visualized by fluorescence in situ hybridization. Transfections were stopped after 4 h, and slides were analyzed by confocal laser-scanning microscopy. Fig. 3 shows a single light optical section. Numerous rounded red spots of DNA were seen in the cytoplasm, and a few were located in the cell nucleus and at the nuclear membrane for TAT 2 gene vectors. The same distribution pattern could be observed for TAT 3 gene vectors, but the total number of intracellular DNA spots seemed to be reduced. There were no DNA spots found in the nucleus of cells transfected with TAT 4 gene vectors. In this case, DNA was rather found as rounded and irregular structures restricted to only the outer side of the nuclear membrane. A similar distribution pattern was ob- FIG. 3. Intracellular localization of plasmid DNA (pEGFP). COS-7 cells were transfected with TAT oligomer or pLa gene vectors, and transfections were stopped after 4 h. Images were generated with a 63ϫ objective by confocal laser-scanning microscopy. Green emission signal represents the cell nuclei stained with Sytox 16; red emission signal shows the distribution of pEGFP in the same microscope field using a digoxigenin-labeled DNA probe. The probe was detected with anti-digoxigenin rhodamine antibody. served for pLa gene vectors.

Ternary Gene Vector Complexes: Combination of TAT Oligomers with Various Standard Cationic Transfection Agents-
The previous experiments suggest that the TAT oligomers could facilitate the transport of DNA into the nucleus. Next we examined whether the TAT oligomers were able to enhance gene transfer efficiency of poly-or lipoplexes that lack efficient nuclear plasmid transport (2,23,24). Ternary gene vector complexes consisting of TAT oligomer, standard cationic transfection agent, and DNA were formulated in two different manners. Either TAT oligomer was added to preformed gene vectors consisting of DNA and standard cationic transfection agent, or complexes were generated vice versa through the addition of standard cationic transfection agent to preformed TAT oligomer gene vector complexes. When one charge equiv- alent of each of the TAT oligomers or pLa was added to preformed standard cationic gene vector complexes, the level of transgene expression remained approximately at the same level as of the standard cationic gene vectors (Fig. 4). In contrast, when the ternary gene vector complexes were generated vice versa (i.e. the DNA was first complexed with one charge equivalent of each of the TAT oligomers, and then the standard transfection agents were added), a strong increase of transgene expression depending on the degree of oligomerization and the type of standard transfection agent was observed (Fig. 4). The presence of TAT 2 , TAT 3 , and TAT 4 enhanced transgene expression of PEI 130-, 80-, and 40-fold, respectively. pLa did not enhance PEI-mediated gene transfer. An analog behavior could be observed when the TAT oligomers were combined with either fractured dendrimers or LipofectAMINE.
Kinetics of Transgene Expression Mediated by Ternary TAT n -PEI-DNA Complexes-To characterize in more detail the transfection mechanism of the ternary TAT n -PEI-DNA complexes, the kinetics of transgene expression was examined on 16HBE cells (Fig. 5). At the 4-h time point, the transgene expression of transfected 16HBE cells was 220-, 170-, and 100-fold higher for the TAT 2 -, TAT 3 -, and TAT 4 -derived ternary gene vectors as compared with PEI polyplexes, respectively. At the 8-h time point, the enhancement in transfection efficiency increased to 390-, 240-, and 140-fold, respectively. After 24 h, the level of transfection efficiency mediated by the ternary TAT 2-4 gene vectors was 104-, 56-, and 45-fold higher as compared with PEI polyplexes, respectively. The transfection efficiency of the TAT 2 -PEI-DNA complex at as early as the 4-h time point was 4-fold higher as compared with the transfection efficiency mediated by PEI polyplexes after 24 h (p Ͻ 0.01). These data suggest rapid accumulation of the DNA in the cell nucleus when TAT oligomers were incorporated into the PEI polyplexes.
Transfection Efficiency of Ternary TAT n -PEI-DNA on Growth-arrested Cells-To further assess whether the enhancement of transfection efficiency mediated by the TAT oligomers could be due to facilitated nuclear transport of the transgene, transfection experiments were performed on growth-arrested cells. Transfection experiments were performed on 16HBE cells either arrested by aphidicolin incubation (25) or through serum starvation (26). Under both conditions, the transfection efficiency of the ternary gene vector complexes was significantly higher as compared with PEI polyplexes (Fig. 6). The incorporation of TAT 2 , TAT 3 , and TAT 4 led to a 35-, 20-, and 12-fold higher transfection rate on aphidicolin-treated cells and a 12-, 10-, and 8-fold higher transfection rate on cells treated under conditions of serum starvation, respectively. In both cases, transgene expression mediated by the ternary complexes (e.g. TAT 2 -PEI-DNA) was 2-3-fold higher as compared with transgene expression mediated by PEI polyplexes under standard conditions (i.e. no growth-arrested cells). pLa improved PEI-mediated transfection 13-and 3.5-fold.
Transfection Efficiency of Ternary TAT n -PEI-DNA on Primary Cells and in Vivo-To better mimic the in vivo situation, transfection with the ternary TAT 2 or pLa gene vector complexes has been performed on primary nasal respiratory epithelium (Fig. 7A). Transgene expression mediated by the ternary TAT 2 -PEI-DNA complexes was 10-fold higher as compared with transgene expression mediated by PEI. The incorporation of pLa improved PEI-mediated transfection as well, but only 1.6-fold. Gene transfer efficiency of ternary TAT 2 gene vectors has been examined in vivo and compared with ternary pLa and standard PEI gene vectors. Mice were intratracheally instilled with gene vector solutions, and luciferase gene expression was measured after 24 h (Fig. 7B). Whereas luciferase expression mediated by ternary TAT 2 gene vectors was 4-fold higher as compared with PEI polyplexes, pLa induced only a 2-fold increase in luciferase gene expression. These data indicate that TAT 2 can improve gene delivery efficiency of PEI polyplexes in vivo.

DISCUSSION
In this study, we examined whether the unique features of the arginine-rich motif of the HIV-1 TAT protein (TAT peptide) (i.e. protein transduction domain and nuclear localization sequence) would enhance nonviral gene transfer. We examined these unique features in the context of biophysical parameters of the gene vectors. The biophysical parameters of synthetic vectors have been shown, on the one hand, to influence their gene transfer efficiency (13,27,28); on the other hand, they are influenced by the molecular weight of the cationic polymer (13,14). Thus, to control the biophysical parameters (i.e. also to control in parts the gene transfer efficiency), the TAT peptide was synthesized as oligomers of different molecular weights. Interestingly, the biophysical complex parameters of the TAT 2-4 gene vectors were similar. The degree of oligomerization only correlated with the degree of DNA condensation as examined by fluorescence quenching assay and increased with the higher degree of oligomerization. Gene vectors formulated with pLa exhibited similar biophysical parameters. These data show that the size and surface characteristics of both the TAT oligomer and pLa-derived gene vectors were similar, whereas DNA compaction of the gene vectors slightly varied depending on the degree of oligomerization. From this we conclude that differences in transfection efficiency among the TAT oligomers and when compared with the pLa can be primarily attributed to the TAT sequence itself rather than to different biophysical gene vector parameters. The results obtained from transfection experiments were further confirmed by confocal laser-scanning microscopy, which showed higher accumulation of DNA in the cell nucleus mediated by TAT 2 and TAT 3 than by TAT 4 and pLa. In addition, these data could suggest that the affinity of the TAT oligomers to the nuclear import machinery decreases with the higher degree of oligomerization and is minimal for pLa.
The presence of chloroquine during transfection has been shown to increase the level of transgene expression of several nonviral gene vectors due to reduced endolysosomal entrapment and degradation of the gene vector complexes (19). Since transgene expression strongly increased in the presence of chloroquine, this indicated that endosomes were involved in cellular uptake of the gene vectors. However, it is conceivable that two separate uptake mechanisms (i.e. endosomal uptake and direct membrane penetration) take place independently in parallel at the same time. To examine this possibility, the transfection process has been analyzed when endocytosis was blocked (i.e. at 4°C (6, 10, 11)) or in the presence of metabolic inhibitors (20 -22). Both transfections performed at 4°C and in the presence of endocytosis inhibitors led to a strong reduction of transgene expression. Therefore, the main uptake mechanism of TAT oligomer gene vectors was apparently through endocytosis and not mediated by a proposed protein transduction domain. A possible explanation could be due to the size of the gene vectors (ϳ700 nm). Constructs that have been delivered into cells via the conjugation of the TAT peptide so far have been much smaller in size (45-200 nm) (9, 10). Thus, simple electrostatic binding of DNA by the TAT peptide might lead to particles inappropriate for membrane penetration.
The higher gene transfer efficiency mediated by the TAT 2 peptide as compared with the nuclear transport-deficient TAT 2 -M1 peptide (12) and the strong increase of gene transfer of the TAT oligomer complexes in the presence of chloroquine as compared with pLa complexes suggests that gene transfer of the TAT oligomers was peptide sequence-dependent, which could be due to facilitated nuclear translocation mediated by the nuclear localization sequence function. This concept is supported by the behavior of ternary gene vector constructs. Transfection efficiency of PEI in the presence of the TAT oligomers was strongly enhanced on growth-arrested and primary cells as well as in vivo and at early time points after transfection. In all of these cases, only a minor fraction of cells underwent mitosis (i.e. breakdown of the nuclear membrane could be excluded as the major mechanism for nuclear DNA localization). Consequently, these experiments suggest that the TAT oligomers could promote nuclear translocation of the DNA. An increase was observed when pLa was combined with Lipo-fectAMINE and when experiments were performed on growtharrested cells and in vivo. These observations could be due to partial activity of pLa as a nuclear localization sequence, which has been reported recently (29).
The formulation order of vector complexes was found to have an enormous effect on the gene delivery efficiency. A 390-fold increase of transgene expression was only observed when the DNA was first complexed with the TAT oligomer and PEI was added afterward. When the complexes were generated vice versa, transgene expression remained on the same level as of PEI polyplexes. These differences could be explained by analysis of the structure of the resulting gene vectors. When DNA was first complexed with the TAT oligomers at a charge ratio of Ϯ1, the resulting intermediate complexes had a potential of Ϫ20 mV (Table I), whereas DNA complexed with PEI at a nitrogen of PEI per phosphate of the DNA ratio of 10 resulted in a potential of ϩ32 mV. Thus, preformed negatively charged TAT gene vectors allow the binding of positively charged PEI to their surface through electrostatic interaction, which was indicated by a change of the potential from negative to positive (ϩ40 mV) and by size reduction to 600 nm. In contrast, such changes were not observed when gene vectors were formulated vice versa. Thus, the formulation method could result in a shell-like ternary complex with the TAT oligomers bound to the DNA in the core of the complex and a layer of PEI at the periphery of the complex (see Supplementary Material Fig.  3A). In contrast, binding of the TAT oligomers on the surface of positively charged PEI polyplexes should not be possible due to electrostatic repulsion. In this case, TAT oligomers could rather be homogeneously distributed in the solution (Supplementary Material Fig. 3B). From these data, we infer the following model, which could explain the exceptional efficiency of the ternary gene vector complexes. The ternary complexes, which apparently look like a plain PEI complex, could behave like a plain PEI polyplex concerning the first steps that are involved in gene transfer (i.e. the complexes are apparently taken up into the cell via receptor-mediated endocytosis (heperan sulfate proteoglycan receptor) (30) and are located to the endosomal compartment. In the endolysosomes, the "protonsponge" effect of PEI (31) could induce their disruption and release the complexes into the cytosol. Within the next steps, the PEI shell could be released, probably due to nonspecific interaction with cytosolic components, and the core complex could slowly diffuse toward the cell nucleus. Binding of the TAT oligomer to the nucleocytoplasmic shuttle protein importin ␤ could then facilitate DNA translocation into the nucleus, resulting in high transgene expression. In this model, PEI would function as an endosomal disrupting agent such as chloroquin but incorporated into the complex itself.
In conclusion, we showed that oligomers of the TAT peptide mediate efficient gene delivery. In particular, their combination with powerful standard cationic transfection reagents (e.g. PEI) resulted in very efficient gene transfer. This effect correlated inversely with the degree of oligomerization, and we suggest that facilitated nuclear localization could be involved. Further studies will focus on improved formulation methods of TAT-DNA complexes to overcome size restrictions that could limit the functionality of the protein transduction domain and nuclear localization sequence so far.