Introduction

The cell membrane acts as a barrier and has been shown to have poor permeability to large molecules. In addition, living cells do not actively import most macromolecules (Gros et al., 2006). One of the main difficulties in the development of new drugs whose targets are located within host cells is membrane impermeability. This challenge is reflected in the fact that greater than 95% of all new therapeutics possesses poor pharmacokinetics. In general, the delivery of macromolecules utilizes lipids, electroporation, or viral vectors (Chae et al., 2005; Nenoi et al., 2006; Vinogradov, 2006; Cho et al.,2007). However, these tools exhibit several limitations, including: (1) the inability to deliver to primary and non-dividing cells, (2) the requirement for optimization with each cell type, (3) low transduction level, (4) high cellular toxicity (Carriere et al., 2003; Green et al., 2003; Bachran et al., 2005; Kuriyama et al., 2006).

Because of these limitations, many alternative methods of macromolecule delivery have been generated through the observation of protein transduction domains (PTD), particularly PTDs derived from HIV-1 transcriptional regulator Tat (Schwarze et al., 1999; Sihol et al., 2002) and the Drosophila transcription factor Antennapedia (Antp) (Joliot et al., 1991). These PTDs have been shown to mediate the rapid cellular delivery of proteins (Lundberg et al., 2003) and peptides (Eum et al., 2005) and have been observed to possess the ability to traverse the lipid bilayer of cells in a concentration dependent manner (Langedijk et al., 2004). The exact mechanism of delivery is not fully understood, although the PTDs are generally known to undergo endocytosis (Drin et al., 2003; Ignatovich et al., 2003; Richard et al., 2003; Trehin and Merkle, 2004). The importance of these proteins lies in their ability to uniformly transport large, biologically active molecules to a population of mammalian cells growing under standard culture conditions (Snyder and Dowdy, 2004).

Hoxc8 is one of many homeobox genes that encode a set of master transcription factors (Lei et al., 2007). It contains a similar homeodomain to that of Antp (Le Mouellic et al., 1992). Homeobox genes encode protein domains of 60 amino acids, which are known to play important roles in defining the body plan of vertebrates (McGinnis and Krumlauf, 1992; Waltregny et al., 2002). Homeodomain proteins belong to a class of transcription factors involved in multiple morphological processes. The homeodomain is composed of three α helices with one β turn between helices 2 and 3. The third helix is known to possess translocation properties (Derossi et al., 1998).

It has been found that the C-terminus of Hoxc8 (the acidic portion) contains many negatively charged cations such as glutamic acid and aspartic acid. Because DNA is highly negatively charged, when DNA and a construct that contains a homeodomain with an acidic C-terminus are combined in solution, a DNA-protein complex will not form due to the repelling effect between the molecules' negative charges (Sloots and Wels, 2005).

Although the entire homeoprotein can be used for delivering macromolecules, small poly-arginine (poly-R) and poly-lysine peptides have been shown to exhibit greater efficiency in delivering several peptides and proteins (Noguchi and Matsumoto, 2006).

Therefore, it was hypothesized that a homeodomain would be a more efficient delivery vector than a full sequence Hoxc8 protein. In this study, we designed three different constructs and examined their membrane transduction activity: Hoxc81-242, which contained a full-length protein, Hoxc8152-242, which contained the homeodomain with a partially deleted N-terminus and fully intact acidic C-terminus, and Hoxc8149-208, which contained the homeodomain in a glutathione S-transferase (GST) fused form. We also analyzed the possibility of our constructs as delivery vectors for nucleic acids since the cationic homeodomain can electrostatically interact with DNA.

Materials and Methods

Plasmid construction

In order to construct the plasmid pGEX:Hoxc81-242, a 780 bp Hoxc8 fragment was isolated from the plasmid pEGFP:C8 (Kwon et al., 2003) after digestion with BamHI and XhoI and then inserted into the same sites of the expression vector pGEX4T-1 (Amersham, Uppsala, Sweden). A 306 bp SalI fragment was isolated from pGEX:Hoxc81-242 after SalI digestion and then cloned into the SalI site in pGEX4T-1 to produce pGEX:Hoxc8152-242. For pGEX:Hoxc8149-208, PCR amplification was performed with Hoxc8 homeodomain-specific primers (forward primer, 5'-GAGGATCCC-GGCGCAGCGGTCG-3'; reverse primer, 5'-GGCTCGAGTTCTCCTTTTTCCAC-3'), which were designed to contain the BamHI and XhoI sites, respectively (Bioneer, Daejeon, Republic of Korea). Following the PCR reaction (first denaturation at 94℃ for 5 min; 30 cycles of denaturation at 94℃ for 30 s; annealing at 56℃ for 30 s; extension at 72℃ for 60 s; final extension at 72℃ for 5 min), the amplified fragment was cloned into the pGEM-T Easy vector (Promega, Madison, WI). Then, the BamHI and XhoI-containing portion of the 180 bp homeobox fragment was isolated and inserted into the same sites in the expression vector pGEX4T-1 to produce pGEX: Hoxc8149-208.

Protein purification and SDS-PAGE

In order to purify the fusion proteins, each of the three plasmids - pGEX:Hoxc81-242, pGEX:Hoxc8152-242, and pGEX:Hoxc8149-208 - was transformed into the E.coli strain BL21 (DE3) and cultured in 2 × YT media containing ampicillin (100 µg/ml) at 37℃. When the OD600 reached approximately 0.8, IPTG (Isopropyl-b-D-thiogalactopyranoside) was added to a final concentration of 500 nM, and the cells were incubated for another 3 h at 28℃. The cells were collected through centrifugation, cracked with a French Press (SLMOAMINCO, SLM instrument Inc., Urbana, IL), and then centrifuged at 12 krpm for 60 min. The supernatant was removed, incubated with glutathione sepharose beads (Sigma, St. Louis, MO) overnight, and then centrifuged for 10 min at 2 krpm. The precipitant was washed three times with ice cold PBS, added to reduced glutathione elution buffer (pH 8.0) containing 50 mM Tris and 10 mM reduced glutathione, and then incubated as per the manufacturer's suggestion. After centrifugation for 5 min at 1 krpm, the supernatant was removed, and the concentration of the purified protein was measured using Eppendorf Biophotometer (Eppendorf, Hamburg, Germany). Purified proteins were denatured in sample buffers (50 mM Tris-HCl, pH 6.8, 10% glycerol, 2% SDS, 100 mM DTT, and 0.1% bromophenol blue) by boiling for 10 min, and then resolved on a 10% SDS-polyacrylamide gel. After electrophoresis, proteins were stained with Coomassie blue.

Labeling proteins with Oregon 488

In order to effectively label proteins with Oregon 488 (Invitrogen, Carlsbad, CA), it was advised according to the manufacturer's instructions to use a protein concentration of higher than 2.0 µg/µl. Therefore, purified proteins were concentrated to 4-10 µg/µl using a centrifugal filter device (Amicon Bioseparations Centricon, Milipore, Bedford, MA). To do this, purified proteins were loaded onto the column of the centrifugal filter device and centrifuged for 2,000 g until the desired concentration of protein was obtained. About 5 mg of the protein were dissolved in 0.5 ml of 0.1 M sodium bicarbonate buffer. 100 µl of dye solution (0.5 mg of Oregon 488 dye dissolved in 0.5 ml of DMSO) was then added slowly while vortexing the protein solution. This final solution was incubated for an hour at room temperature with continuous stirring. The dye solution was freshly prepared immediately prior to starting the reaction because reactive compounds are not very stable in solution. The labeled proteins were purified using Sephadex G25 (Sigma).

Cell culture and intracellular delivery of protein and nucleic acid

PPFF (pig primary fetal fibroblast), COS-7 (SV40-transformed African green monkey kidney cells), and T98G cells (Myeloma cells obtained from the brain) were cultured in DMEM (Welgene, Daegu, Korea) containing 10% FBS and 5% penicillin-streptomycine at 37℃ in a 5% CO2 atmosphere. To analyze protein transduction, cells were seeded on a 24-well plate at a density of 1 × 105 cells/well. After incubating for 24 h, the medium was replaced with fresh medium containing fusion proteins (10 nM, 100 nM, 500 nM, or 1 mM) and then incubated again (0, 5 min, 10 min, 15 min, 30 min, 1 h, or 2 h). A trypsin treatment was added, which removed proteins that were electro-statically bound to the cell surface. Protein transduction was analyzed under a fluorescent microscope (Olympus IX70, Olympus, Melille, NY). To analyze the transduction of nucleic acid complexed with recombinant Hoxc8 protein, cells were incubated in the presence of a medium containing the protein-DNA (pDsRed1-C1 plasmid) complex for 24 h. Fresh medium was added, and the cells were incubated 24 h further. After cell viability analysis with the light microscope, the cells were maintained for another 24 h to allow for protein expression. The cells were washed three times with PBS, fixed in 1.5% formaldehyde solution in PBS for 20 min and then permeabilized with 100% methanol, After washing with PBS, cells were stained with DAPI, and analyzed under the fluorescence microscope.

Protein-nucleic acid complex analysis using agarose gel retardation

The complex formation between the recombinant Hoxc8 protein and the nucleic acid (a reporter plasmid DNA) was investigated using a gel retardation on agarose gel. Increasing amounts of protein (0.1 to 2 µg) were incubated with 0.1 µg of nucleic acid (pEGFP-C1 plasmid), not exceeding 20 µl in total volume. Glycerol was then added to 30%. The mixture was left at room temperature for 1 h, and then the electrophoretic mobility of the resulting protein-DNA complexes was analyzed on 1% agarose gel containing ethidium bromide.

Results

Production of Hoxc8 fusion protein and its fusion derivatives

In order to examine whether the Hoxc8 protein can deliver double-stranded DNA into mammalian cells, several Hoxc8-derived fusion proteins were prepared by cloning into the plasmid pGEX4T-1 (Figure 1). The plasmids pGEX:Hoxc81-242, pGEX:Hoxc8152-242, and pGEX:Hoxc8149-208 were constructed as described in Materials and Methods (Figure 1A). E.coli cells containing each plasmid were cultured in 2×YT media in the presence of 0.5 mM IPTG at 28℃ in order to produce the fusion proteins. To confirm whether the resultant fusion proteins matched the original design, each protein was partially purified using GST-agarose beads and analyzed on a polyacrylamide gel containing SDS. As shown in Figure 1C, the fusion proteins GST-Hoxc81-242, GST-Hoxc8152-242, and GST-Hoxc8149-208 were 54-, 38- and 35 kDa, respectively, matching their original design.

Figure 1
figure 1

Schematic drawing of plasmids and SDS-PAGE gel showing partially purified proteins of GST:Hoxc81-242, GST:Hoxc8152-242 and GST:Hoxc8149-208. (A) The full length Hoxc8 gene was inserted into the BamHI and XhoI sites located in the multi cloning sites (MCS) of the pGEX4T-1 vector, creating pGEX:Hoxc81-242, in which the Hoxc8 protein was expressed as a glutathione-S-transferase (GST)-fused form under the control of the tac promoter (ptac) and operator in the presence of an inducer (IPTG). The construct pGEX:Hoxc8152-242 was obtained by digesting pGEX:Hoxc81-242 with SalI endonuclease and ligating into the same site in the pGEX4T-1 vector. With pGEX:Hox8149-208, a homeodomain fragment was amplified using PCR in the presence of a homeodomain specific primer set that contained BamHI and XhoI sites at each end (BamHI, 5'-GAGGATCCCGGCGCAGCGGTCG; XhoI, 5'-GGCTCGAGTTCTCCTTTT-TCCAC). The fragment was then cloned into the same sites in the pGEX4T-1 vector after digestion with BamHI and XhoI. Each restriction site (BamHI, XhoI, and SalI) that was used for cloning, the replication origin (ori), the ampicillin resistance gene (Ampr), and the Lac repressor gene (LacIq) are marked on the map. The calculated PI of each GST-fusion protein was written on the right. (B) Schematic illustration of the Hoxc8 protein indicating the hexapeptide, the homeodomain, and the acidic C-terminus with the amino acid position in numbers. At the bottom, the amino acid residues from AA218 through AA242 are shown along with the stop codon, TAA. (C) SDS-PAGE gel showing partially purified proteins of GST:Hoxc81-242, GST:Hoxc8152-242 and GST:Hoxc8149-208. Each GST-fused protein was partially purified using glutathione agarose beads, analyzed on the 10% polyacrylamide gel containing SDS, and then detected by Coomassie staining. Lanes 1 to 4 indicate GST, GST:Hoxc8149-208, GST:Hoxc8152-242, and GST:Hoxc81-242, respectively. The molecular weights of marker (M) and partially purified proteins are written on the left and right in kDal, respectively.

Full size image

Cellular translocation of GST-fused Hoxc8 derivatives; GST-Hoxc81-242, GST-Hoxc8152-242, and GST-Hoxc8149-208

In order to analyze the translocation efficiency of GST-fused Hoxc8 derivatives, each protein, in addition to GST alone, was labeled with Oregon 488 as described in the Materials and Methods. When the labeled GST-Hoxc81-242 was added to the culture media, green fluorescence was detected inside of all PPFF cells as shown in Figure 2 (A and B), indicating that the labeled protein entered the cell through the lipid bilayer; this was expected based on previous results (Lee et al., 2006). In contrast, no fluorescence or very low levels of fluorescence were detected in the cells treated with Oregon 488 dye (Figure 2C) and GST protein (Figure 2D). Although a small amount of GST was found to enter the cell, the translocation efficiency of the GST-Hoxc81-242 was far greater than that of GST alone.

Figure 2
figure 2

Cellular transduction of Oregon 488-labeled GST:Hoxc81-242 in PPFF cells. PPFF cells were cultured in the presence of Oregon 488-labeled GST:Hoxc81-242 protein (500 nM) as described in Materials and Methods and analyzed under the fluorescence (A) and light microscope (B). Oregon 488 alone (C) and Oregon 488-labeled GST protein (D) were included as controls. (E) Time course of protein (Oregon 488-labeled GST:Hoxc81-242) transduction. After adding the fusion protein GST:Hoxc81-242 labeled with Oregon 488 for 5 min, 10 min, 15 min, 30 min, 1 h, or 2 h, cells were washed with PBS and then fluorescence microscopy was taken.

Full size image

The cellular internalization of proteins was tested using 500 nM of Oregon 488-labeled GST-Hoxc81-242 protein following different incubation times (5 min, 10 min, 15 min, 30 min, 1 h, and 2 h). The green fluorescence intensity in the cell gradually increased with time until reaching a plateau at 1 h of incubation (Figure 2E).

In order to test the optimum fusion protein concentration for cellular transduction, 10 nM, 100 nM, 500 nM and 1 µM of Oregon 488-labeled full length Hoxc8 fusion proteins (GST-Hoxc81-242) were added to the T98G cells. As shown in Figure 3 (A-D), the intensity seemed to reach a peak at 500 nM of proteins after 1 h of incubation. When the transduction efficiency of the other two fusion proteins, GST-Hoxc8152-242 and GST-Hoxc8149-208, were analyzed in the same way, both showed similar results (Figure 3E-L), indicating that all three proteins have similar transduction efficiencies.

Figure 3
figure 3

Fluorescence microscopy showing the cellular transduction of each recombinant fusion protein labeled with Oregon 488. Each labeled protein of GST:Hoxc81-242 (A-D), GST:Hoxc8152-242 (E-H), or GST:Hoxc8149-208 (I-L) was added to the media containing T98G cells at different concentrations; 10 nM (A, E, and I), 100 nM (B, F, and J), 500 nM (C, G, and K) and 1 µM (D, H, and L). The protein transduction was then analyzed under the fluorescent microscope.

Full size image

Formation of DNA-protein complex

In order to test whether the Hoxc8 protein coul0d be applicable to the development of delivery vector for nucleic acid, the protein and DNA interaction was investigated using a gel retardation experiment. As described in Materials and Methods, different concentrations of GST-Hoxc8 fusion proteins were incubated with the same amount of DNA. GST-Hoxc81-242 seemed to bind well with DNA (Figure 4); most of the DNA complexed with the proteins in a ratio of 1 : 15 (mass ratio of DNA : protein). In the case of GST-Hoxc8152-242 having PI value of 5.8, the formation of DNA-protein complexes was inhibited. In contrast, GST-Hoxc8149-208 containing only the homeodomain (PI, 9.0) regained the ability to form DNA-protein complexes. Even with small amounts of GST-Hoxc8149-208 protein (at the mass ratio of 1 : 1), high amounts of protein-DNA complex were formed; about 50% of DNA formed complex with this protein at this mass ratio (Figure 4, lane 2). As controls, BSA and GST proteins did not form complexes with DNA at any concentration, even when 20 × excess proteins were added.

Figure 4
figure 4

Gel retardation experiment for DNA-protein complex. Different amounts of each protein (GST:Hoxc81-242, GST:Hoxc8152-242, GST:Hoxc8149-208, BSA, and GST) were mixed with 0.1 µg of pEGFP-C1 plasmid in 20 µl of reaction volume and were kept at room temperature for 1 h. After electrophoresis of the reaction mixtures on a 1% agarose gel containing ethidium bromide, photography was taken under the UV lamp.

Full size image

Intracellular delivery of DNA complexed with murine Hoxc8 homeoprotein

In order to test whether the Hoxc8 protein could act as a nucleic acid delivery vector for future therapeutic purposes, the protein-DNA complexes were added to COS-7 cells in fresh DMEM medium containing 10% FBS as described in Materials and Methods. For easy observation of delivered DNA, pDsRed1-C1 plasmid encoding red fluorescent protein was used. And the fusion protein GST-Hoxc8149-208 showing the most efficient complex formation with DNA was chosen as cargo protein of DNA. pDsRed1-C1 DNA (1 µg) complexed with GST-Hoxc8149-208 at a mass ratio of 1 : 16 (DNA : protein) was applied to a well (12-well plate), incubated for 24 h, and then analyzed under the fluorescence microscope for red fluorescence. As shown in Figure 5, the red fluorescence protein turned out to be expressed and detected in the cells indicating that the murine Hoxc8 homeoprotein GST-Hoxc8149-208 delivered the nucleic acid (pDsRed1-C1 plasmid) into COS-7 cells.

Figure 5
figure 5

Intracellular delivery of DNA complexed with the recombinant murine Hoxc8 homeoprotein. The recombinant fusion protein GST:Hoxc8149-208 complexed with pDsRed1-C1 plasmid was transfected into the COS-7 cells and then analyzed the red fluorescence under the fluorescence microscope as described in the Materials and Methods. (A) Red color indicates the expression of red protein encoded in the pDsRed1-C1 DNA transfected into the cells along with the GST:Hoxc8149-208 protein. (B) and (C) are the DAPI staining of (A) and merged photo of (A) and (B), respectively.

Full size image

Discussion

Protein transduction domains (PTD) have received much attention since being reported to traverse biological membranes. The homeodomain of Antennapedia, which acts as a PTD, is well known and well studied. In this study, we show that the homeodomain of Hoxc8 can traverse mammalian cell membranes. According to Derossi et al. (1996), the third helix of the Antennapedia homeodomain is essential for cellular internalization of Antennapedia. Derossi et al. (1996) found that one mutant, which had two hydrophobic residues (tryptophan48 and a phenylalanine49) deleted from the third helix of the homeodomain, lost its translocating ability. Using synthesized peptides, they found that a peptide of 16 amino acid residues, which corresponded to amino acids 43-58 of the homeodomain (penetratin-1 or 'third helix'), possessed translocation properties comparable to that of the entire homeodomain.

We have previously reported that murine Hoxc8 homeoprotein can traverse the cellular membrane efficiently (Lee et al., 2006). We have found that the Hoxc8 protein contains many glutamic acid and aspartic acid residues at the C-terminus (Figure 1B), which gives the protein a negatively charged property. When PTDs are constituted with basic amino acids such as poly-Arg or poly-Lys, they seem to be more effectively transduced into cells. Therefore, in this study, we constructed several Hoxc8 deletion derivatives that contained homeodomains and analyzed their transduction activity alone or with nucleic acids. We generated three types of GST-Hoxc8 fusion proteins: a protein that contained a full length Hoxc8 (GST-Hoxc81-242), a protein that contained a homeodomain, which possessed a negatively charged C-term (GST-Hoxc8152-242), and a protein that contained a homeodomain with a deleted C-term (GST-Hoxc8149-208) (Figure 1). With GST-Hoxc8152-242, 3 amino acids were missing at the beginning of the homeodomain. These missing amino acids were not considered to be of serious importance because it has been suggested that the first few amino acids of the homeodomain are not important, while the last helix is crucial (Roy and Sen, 2005).

In order to investigate protein transduction with the fluorescence microscope, the proteins were labeled with Oregon 488. The Oregon 488 dye alone could not cross the cell membrane (Figure 2C). As expected, the GST-Hoxc81-242 protein traversed the cellular membrane efficiently (Figure 2A and B). The GST alone showed a very weak signal (Figure 2D), which was not surprising because GST itself has been reported to be a PTD (Namiki et al., 2003). Our previous work has also shown GST to possess low cell traversing activity (Lee et al., 2006). Although more research is needed to prove that the Hoxc8 protein alone is an effective penetrating protein, the protein itself could be as effective as GST-Hoxc8. In our laboratory, a 16 amino acid derivative of the Hoxc8 homeodomain has been shown to have PTD activity (Data not shown).

When the three different GST-Hoxc8 fusion proteins were tested to find the optimum concentration for transduction activity, all three proteins (GST-Hoxc81-242, GST-Hoxc8152-242, and GST-Hoxc8149-208) seemed similar, showing the highest efficiency at 500 nM concentration after 1 h incubation. Furthermore, all of them were internalized into almost 100% of the cells, showing even more transduction efficiency similarities. The sub-cellular localization seemed to be in both the cytoplasm and the nucleus in all cases (Figure 2 and 3).

When complexed to DNA, each of the three proteins had very different complex formation efficiencies. GST-Hoxc8152-242 did not form a complex with DNA, like the control proteins BSA and GST both did. This could be due to the negatively charged C-term of Hoxc8, which could have repelled the acidic DNA. In fact, GST-Hoxc8152-242 is an acidic protein with a theoretical PI of 5.8; BSA has a theoretical PI of 5.82. In contrast, both GST-Hoxc81-242 and GST-Hoxc8149-208 formed complexes with DNA (Figure 4). The PI of GST-Hoxc81-242 and GST-Hoxc8149-208 is 6.37 and 9.0, respectively. Although the GST-Hoxc81-242 contained an acidic C-term, it was still able to complex with DNA. GST-Hoxc81-242 contained a full length Hoxc8 protein, which is a well known transcription factor that can bind to DNA through a helix-turn-helix DNA binding motif. Regional specificity for binding seemed to be enough to form a complex with DNA even though GST-Hoxc81-242 is acidic as a whole. This could be partially explained by the fact that the PI of GST itself is 10.84. However, the GST alone did not form a complex with DNA, meaning that the positive charge itself is not sufficient for DNA binding. The basic protein GST-Hoxc8149-208, which contained only the homeodomain, made a complex with acidic DNA more efficiently than the other fusion proteins, as was expected from another report's results (Unamalai et al., 2004).

In this study, we found that GST-Hoxc8 fusion proteins not only traversed the mammalian cell membrane, but also delivered nucleic acids into cells (Figure 5), although the transfection efficiency was rather low; less than 5% of the cells expressed the red fluorescence protein (Figure 5). This result is still interesting since it is the first time to show that Hoxc8 PTD can deliver DNA into the cell. In the case of Tat PTD, it also formed a complex with DNA very well. However, the efficiency of gene transfer was reported to be extremely low, almost null unless poly lysine was fused to the Tat PTD (Tat-pK), in which case about 5% of the cells expressed foreign gene (Hashida et al., 2004). In order to express foreign genes, DNA must be translocated across the cell membrane and then transported into the nucleus. Recently Doyle and Chan (2007) have reported that the intracellular localization of DNA influenced the exogenous gene expression in mammalian cells: DNAs complexed with PEI (polyethylenimine) were found to be located in the close proximity to the nucleus and expressed highly, whereas DNAs complexed with PEI conjugated with polyarginine PTD were found to be widely distribute throughout the cytoplasm and negligible gene expression was noted although polyarginine PTD enhanced the membrane transductional properties of PEI. Although the intracellular localization of the plasmid DNA (pDsRed1-C1) was not examined here, the DNA could be located in somewhere unfavorable for expression. Harada et al. (2006) also reported that the transfection efficiency of Tat-pK was easily influenced by several factors, such as pH or temperature of medium, preincubation period, and quantity of DNA. Not only these we can not exclude the possibility of DNA degradation before and/or after entering the cell. Thus far, there have been many reports on PTD as a delivery vector for proteins though, not many cases have been described for nucleic acid delivery (Morris et al., 2000, 2004; Mukherjee et al., 2005). Although the transfection efficiency was not as effective as commercialized lipofectamine or viral vectors (Harada et al., 2006), using PTDs in combination with other gene delivery systems could enhance the transfection efficiency of PTDs in the future.

In conclusion, the homeodomain of Hoxc8 has high membrane traversing ability in mammalian cells. Viral vectors are mostly used in gene therapy, though they have been reported to have high toxicities and cause immune reactions. However, no alternatives are yet available. PTDs like Hoxc8 could be a promising tool for delivering macromolecular nucleic acids such as oligonucleotides or DNA/siRNA in research or in clinical use. The results presented here could be a step in the development of safe and tissue-specific vectors for gene therapy.