Targeting DNA to cells with basic fibroblast growth factor (FGF2).

Ligand-mediated targeting of DNA was validated by condensing a plasmid DNA encoding the β-galactosidase (β-gal) gene with a basic fibroblast growth factor (FGF2) that was first chemically conjugated to polylysine (K). The conditions that gave optimal binding of this FGF2 to DNA also generated the highest level of β-gal expression when added to FGF2 target cells like COS-1, 3T3, baby hamster kidney (BHK), or endothelial cells. This β-gal activity increased in a time- and dose-dependent manner and was dependent on the inclusion of FGF2 in the complex. FGF receptor specificity was demonstrated by competition of the complex with FGF2 and heparin, and by the failure of cytochrome c or histone H1 to mimic the gene-targeting effects of FGF2. The expression of β-gal was also endosome dependent because chloroquine increased β-gal expression 8-fold and endosome disruptive peptides increased expression of β-gal 26-fold. Taken together these findings establish that DNA can be introduced into cells through the high affinity FGF receptor complex, and while its efficiency will require significant enhancements to achieve sustained and elevated transgene expression, the possibility that the technique could be used to deliver DNAs encoding cytotoxic molecules is discussed.

We previously demonstrated that basic fibroblast growth factor (FGF2) 1 can specifically target cytotoxic molecules to cells that express high affinity receptors (1). For example, a mitotoxin engineered by expressing a DNA encoding FGF2 and saporin (SAP) is an effective and potent cytocidal agent for cells that express high affinity FGF receptors (2). This FGF2-saporin fusion protein inhibits smooth muscle cell proliferation in vitro and in experimental models of restenosis, blocks tumor cell growth in vitro and solid tumor growth in mouse xenografts, prevents the proliferation of lens epithelial cells in culture, and delays secondary lens clouding in vivo (3)(4)(5)(6)(7)(8). These findings led us to explore the possibility that FGF2 could also promote the entry of other macromolecules into cells, specifically DNA.
FGF2 was selected because it is associated with a wide variety of diseases including tumor growth and metastasis, arterial wall hyperplasia of smooth muscle cells, and various angiogenic diseases (e.g. diabetic retinopathy and rheumatoid arthritis). FGF2 binds to one of a family of four high affinity tyrosine kinase receptors of the Ig superfamily and to a class of lower affinity cell surface receptors that are related to heparan sulfate proteoglycans to form a trimolecular complex consisting of ligand and high and low affinity receptors that elicit signal transduction (9 -11). Because FGF2 is naturally translocated to the nucleus (12), we reasoned that it could serve as a vehicle to introduce DNA into cells and subsequently shepherd it to the nucleus if necessary. In this report, we describe FGF2's ability to specifically target plasmid DNAs to its target cells.

MATERIALS AND METHODS
FGF2 Preparation-In all of the experiments described here, we used a 155-amino acid FGF2, in which the cysteine at position 96 had been mutagenized to serine (13). FGF2 was prepared in Escherichia coli and purified to homogeneity by conventional chromatography techniques.
Conjugation of FGF2 to Polylysine-Polylysine polymers with average lengths of 13, 39, 84, 152, and 265 (K 13 , K 39 , K 84 , K 152 , and K 265 ) were purchased from Sigma and dissolved in 0.1 M NaPO 4 , 0.1 M NaCl, 1 mM EDTA, pH 7.5 (buffer A) at concentrations of 3-5 mg/ml. N-Succinimidyl-3(pyridyldithio)proprionate (Pierce) in anhydrous ethanol was added to polylysine at a molar ratio of 1.5 and incubated at room temperature for 30 min. The mixture was dialyzed against buffer A for 4 h at 4°C to remove excess N-succinimidyl-3(pyridyldithio)proprionate and then combined at a molar ratio of 1.5 with the FGF2 mutein (in buffer A). The mixture was incubated overnight at 4°C. The conjugation reaction was applied to a Resource S column (Pharmacia, Uppsala, Sweden) and eluted with a gradient of 0.15 to 2.1 M NaCl in 20 mM NaPO 4 , 1 mM EDTA, pH 8.0 (buffer B), over 24 column volumes. The fractions containing the conjugated FGF2-polylysine (FGF2-K) were concentrated and loaded onto a gel-filtration column (Sephacryl S100, Pharmacia, Uppsala, Sweden) in 20 mM HEPES, 0.15 M NaCl, pH 7.3. The conjugate was stored at Ϫ80°C. Histone H1 and cytochrome c were purchased from Sigma and conjugated to polylysine as described above. Endotoxin activity was determined by the LAL gel clot assay (Biowhittaker, Walkersville, MD) and was less than 0.01 endotoxin unit/g.
Heparin Chromatography-FGF2-K conjugates were loaded onto a 1-ml heparin Hi-Trap column preequilibrated with 10 mM NaPO 4 , 1 mM EDTA, pH 6.0 (buffer A). The FGF2-K conjugates were eluted with a gradient from 0.2 M to 2.0 M NaCl in buffer A. Protein fractions were analyzed by SDS-polyacrylamide gel electrophoresis under nonreducing and reducing conditions. Proliferation Assay-Bovine aortic endothelial cells were seeded at 1000 cells/well on a 24-well flat-bottom tissue culture plate in Dulbecco's modified Eagle's medium (Biowhittaker, Walkersville, MD), 10% fetal calf serum (Hyclone, Logan, UT), 50 g/ml gentamycin (Life Technologies, Inc.), and 2 mM L-glutamine (Biowhittaker). The following day serial dilutions of FGF2 and FGF2-K ranging from 1 ϫ 10 Ϫ9 to 5 ϫ 10 Ϫ13 M were added to wells in triplicate. After 48 h the medium was removed, and 1.5 ml of fresh medium containing the same concentrations of FGF2 or FGF2-K were added to the cells. Following another 72 h of incubation, the cells were washed with phosphate-buffered saline, treated with 0.25% trypsin, and counted using a Coulter counter (Coulter, Hialeah, FL).
Cells and Cell Culture-Throughout the initial experiments, the ability of FGF2 to introduce DNA into cells was evaluated using COS-1 and BHK cells. All cell lines were obtained from the ATCC (Rockville, MD) and were grown in Dulbecco's modified Eagle's medium containing 10% fetal calf serum, L-glutamine, nonessential amino acids, and gen-* 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.
Expression Vectors-Mammalian expression plasmids encoding ␤-gal (pSV-␤ and pNASS-␤) were obtained from Clontech (Palo Alto, CA). The pSV-␤ plasmid expresses ␤-gal from the SV40 early promoter. The pNASS-␤ plasmid was the equivalent mammalian reporter vector containing the ␤-gal gene but without a promoter.
Condensation of FGF2-K with DNA-DNA was isolated using Qiagen's (Chatsworth, CA) standard protocol. Each lot of DNA was analyzed by gel electrophoresis for plasmid integrity and the absence of genomic DNA, RNA, and protein. Complexes of FGF2-K with either pSV-␤ or pNASS-␤ were prepared by slowly mixing DNA solutions with FGF2-K conjugates in 20 mM HEPES (pH 7.3), 0.15 M NaCl. The complexes were incubated for 1 h at room temperature prior to addition to cells or further analysis. Unless specified otherwise, 5 g of the complex were added to cells.
Condensation of FGF2-K⅐DNA with Endosomal Disruptive Peptides-The endosomal disruptive peptide sequence GLF EAI EGFI ENGW EGMI DGWYGC was chosen for analysis. The peptide sequence was generated by Bio-synthesis, Inc. (Lewisville, TX), purified to greater than 95%, and analyzed by laser desorption mass spectrometry. The peptide was resuspended in 20 mM HEPES, pH 7.3, 150 mM NaCl, and 30 g were added to the FGF2-K 84 ⅐DNA ␤-gal complex (FGF2-K 84 : DNA ratio of 1:1). The mixture was added to COS cells and incubated for 48 h. Cells were assayed for ␤-galactosidase activity (A 405 ) and normalized to total protein (A 280 ).
FGF2 and FGF2-K Binding to DNA-Gel mobility shift assays were used to evaluate the interactions of FGF2 or FGF2-K with DNA. DNA digested with restriction endonuclease HindIII and the pSV-␤ plasmid digested with HincII (Boehringer Mannheim) were dephosphorylated with calf intestinal phosphatase prior to kinase labeling, according to standard techniques. Two g of treated and pSV-␤ DNA were labeled with 250 Ci of [␥-32 P]ATP. The kinase reactions were stopped with 0.5 M EDTA and extracted with a mixture of phenol/chloroform/isoamyl alcohol (25:24:1). The labeled DNAs were ethanol precipitated and resuspended in TE buffer (10 mM Tris, 1 mM EDTA, pH 8.0). The efficiency of the labeling reaction was determined for each of the reactions.
Thirty-five ng of the labeled DNA were added dropwise to the indicated concentrations of FGF2-K or the FGF2 in 20 mM HEPES, 0.15 M NaCl, pH 7.3. The mixtures were incubated for 1 h at room temperature, then fractionated on a 1% agarose gel. The gel was dried and exposed to Kodak XAR film for 30 min.
Sample Preparation for Electron Microscopy-FGF2-K⅐DNA complexes (20 -50 l) were prepared at the indicated protein to DNA ratios and placed on a sheet of Parafilm. After pick-up with a Porlodion-coated 300 mesh copper grid, the samples were stained for 30 s with 2% aqueous uranyl acetate. For contrast enhancement, samples were rotary shadowed with platinum/palladium (80/20) wrapped around a tungsten filament at an angle of 8°. Electron microscopy imaging was performed with a JOEL 1200 EX transmission electron microscope at 60 -80 kv.
Detection and Quantitation of ␤-Gal in Cells-Cells grown on plastic tissue chamber slides and treated with FGF2-K⅐DNA ␤-gal were incubated for 2 h at 37°C with 1 mg/ml 5-bromo-4-chloro-3-indoyl ␤-Dgalactoside substrate (Boehringer Mannheim) solution containing 2 mM MgCl 2 , 5 mM K 3 Fe(CN) 6 , and 0.3% Nonidet P-40 (Sigma) in phosphatebuffered saline, pH 7.4. Cells were rinsed, mounted, and counterstained with hematoxylin, rinsed, and protected with a coverslip. For quantification, 10 different fields were randomly selected. The number of ␤-galpositive cells and the total number of cells was determined using Image-Pro Plus (Media Cybernetics, Silver Spring, MD) image analysis program. The activity was confirmed by immunohistochemical staining for bacterial ␤-galactosidase using an antibody obtained from Oncogene Sciences (Manhasset, NY).
Assays to measure total ␤-gal activity in cell extracts were performed as described by Promega (Madison, WI). Briefly, the cells were washed twice at the indicated times with 1 ml of phosphate-buffered saline (Ca 2ϩ -and Mg 2ϩ -free) and then lysed with 400 l of lysis buffer. The lysate was collected and vortexed, and the cell debris pelleted by centrifugation. The supernatant was removed and stored at Ϫ70°C until analysis. The clarified lysate was assayed directly, or when necessary, diluted in 1 ϫ lysis buffer as recommended by the manufacturer. The clarified lysates were assayed for ␤-gal activity using an Emax plate reader (Molecular Devices, Sunnyvale, CA). Quantitation of ␤-gal activity was obtained using Soft Max Pro (Molecular Devices, Sunnyvale, CA).

Synthesis and Characterization of FGF2-K Conjugates-
The one remaining reactive cysteine (Cys 78 ) in the FGF2 mutein was conjugated to polylysine so that each FGF2 can react with only one polylysine. The molecular weights of the FGF2-K conjugates were determined by size exclusion HPLC (data not shown), and the conjugated molecules appear relatively homogeneous. For example, no free FGF is detected when the FGF2-K 152 conjugate was evaluated by SDS-polyacrylamide gel electrophoresis under nonreducing conditions (Fig. 1A, lane 1). Under reducing conditions, the FGF2 bound to polylysine is dissociated and a band migrating at the same molecular weight as FGF2 was readily observed (Fig. 1A, lane 3). Similar results were obtained with the other conjugates and reverse phase HPLC of the conjugate showed it to be substantially free of unconjugated FGF2.
Because the interaction between FGF2 and heparin is critical to stabilize the growth factor, to protect it from degradation, and is required for high affinity receptor binding (9, 10, 14, 15), we examined whether conjugation of FGF2 to polylysine interferes with this important interaction. For example, when FGF2-K 152 was loaded onto a heparin column, the complex eluted at 1.8 -2.0 M NaCl (Fig. 1B, closed arrow). This salt concentration is significantly greater than that required to elute unconjugated FGF2 or polylysine alone (Fig. 1B, open arrow), indicating that the complex has a higher affinity for heparan sulfate proteoglycans. Similar results were obtained with the other conjugates.
An endothelial cell proliferation assay was performed to determine if the FGF2-K conjugates have similar activity to unconjugated FGF2. As shown in Fig. 1C, FGF2-K 152 is equivalent to FGF2 in stimulating proliferation of bovine aortic endothelial cells. In order to determine if DNA interferes with FGF2's ability to bind to its receptor and elicit a proliferative response, various concentrations of pSV-␤ DNA were complexed with FGF2-K 152 and 2 ng of each of the complexes were evaluated for their ability to stimulate target cells. As shown in Fig. 1D, the presence of DNA in concentrations of up to 100 g/ml had no effect on the ability of FGF2 to stimulate endothelial cell proliferation when condensed with FGF2-K 152 . Similarly, the simple addition of the DNA and polylysine to FGF2 had no effect on the proliferative action of FGF2 (Fig. 1E). In both instances, however, higher concentration of DNAs (Ͼ100 g) appear inhibitory.
Complex Formation between FGF2-K and DNA-FGF2 has the ability to bind directly to DNA (12), presumably because of its high isoelectric point (pI ϳ9.56). We compared the relative affinity of this interaction to FGF2-K in an effort to enhance the ionic interaction between the protein and nucleic acid. DNA binding to FGF2-K and to FGF2 was analyzed with gel mobility shift assays using 32 P-labeled DNA digested with the restriction endonuclease HindIII or pSV-␤ DNA digested with HincII. We evaluated the binding of FGF2 and FGF2-K to DNA fragments ranging from 0.077 kb to over 9.4 kb (Fig. 2). The FGF2 mutein binds to large and small fragments of DNA, as demonstrated by the ability of 0.1 g to prevent the DNA from entering the gel (Fig. 2, panel A, lane 7). When the FGF2 mutein was conjugated to polylysine of 13 or 39 residues long, there is little change in DNA binding (Fig. 2, panels C and F). In contrast, FGF2-K complexes containing polylysine of 84 amino acids or greater are efficient at binding DNA; concentrations as low as 10 ng/ml of FGF2-K 84 , FGF2-K 152 , and FGF2-K 267 block DNA entry into the gel and elicit shifts in DNA mobility (Fig. 2, panels B, D, and E). While the FGF2 mutein, FGF2-K 13 , and FGF2-K 39 are capable of binding DNA, higher concentrations (Ͼ35 ng) are required to completely shift the fragmented DNA. Because of these results, all targeting of DNA was performed with FGF-K 84 and FGF-K 152 as described in the text.
A second approach to monitor the ability of the FGF2-K complexes to bind DNA involved measuring the growth factor's ability to promote the uptake of plasmid DNA into target cells. The FGF2-K conjugates were evaluated at different protein to DNA ratios for their ability to deliver pSV-␤ DNA to cells, thus leading to ␤-gal expression (Fig. 3A). The ␤-gal activity in lysates from treated cells showed that, as anticipated from the studies above, complexes containing 84 lysine residues gave the highest level of ␤-gal expression and that a protein to DNA ratio of 2:1 (w:w) was optimal for gene expression (Fig. 3A,  lane 3).
One of the critical functions played by cationic linkers is presumed to lie in their ability to promote the condensation of the DNA from long circular strands into highly condensed particles (16,17). Sufficient quantities of polylysine are thus required in the complex to promote the condensation of the DNA into toroids, a form amenable to receptor mediated endocytosis. These compact circular structures are formed when Ͼ90% of the negative charges along the DNA phosphate backbone are neutralized by positively charged polymers like polylysine (18). Toroid formation is thought to enhance DNA sta-bility, increases the concentration of DNA that can be internalized into the cell, and correlates directly with higher gene expression (16). Because toroid formation can be used to qualitatively assess the condensation of DNA, we used electron microscopy to analyze the condensates obtained with different ratios of FGF2-K and DNA. Toroids were generated using protein to DNA ratios of 2:1 (Fig. 3B), but were absent or partially formed at other protein to DNA ratios (Fig. 3C). Maximal ␤-gal activity was observed with the same ratios of FGF2-K and DNA that generate toroids.
FGF2-K Can Target DNA into Cells-Having shown that the FGF2-K conjugate is biologically active and can bind DNA with relatively high affinity, we further examined the ability of the conjugate to deliver DNA into target cells that have FGF receptors. The mammalian expression plasmid encoding the ␤-gal gene (pSV-␤) was condensed with FGF2-K 84 and used as a reporter gene to monitor gene transfer and expression into target cells. ␤-gal expression was demonstrated in a variety of cell types including COS (Fig. 4A, lane 2), B16FO mouse melanoma cells (lane 3), and NIH 3T3 cells (lane 4), all of which are known to express FGF receptors.
We then established the dependence of the promoter in controlling protein expression (Fig. 4B). We directed two types of DNA encoding ␤-gal into cells: the first has the ␤-gal gene driven by the SV40 promoter (pSV-␤), and the second has the FIG. 1. Synthesis and characterization of FGF2-K conjugates. Panel A, SDS-polyacrylamide gel electrophoresis of FGF2-K 152 under nonreducing and reducing conditions. Approximately 7 g of FGF2-K 152 , lanes 1 and 3, were prepared and electrophoresed as described in the text. Approximately 8 g of the FGF2 mutein, lanes 2 and 4, were electrophoresed and used as a standard (lanes 1 and 2,  nonreducing conditions; lanes 3 and 4, reducing conditions). The open arrow identifies the material that was unable to enter the gel. The closed arrow identifies a protein band corresponding to the FGF2 mutein. Panel B, heparin-Sepharose affinity chromatography of FGF2-K 152 . The FGF2-K 152 conjugate was purified by heparin-Sepharose chromatography, as described in the text, and eluted with a linear gradient of NaCl (0.1 M to 2.0 M). The expected peak elution point for FGF2 alone (open arrow) was compared to the peak elution point of the conjugate (closed arrow). Panel C, stimulation of endothelial cell proliferation by FGF2 and FGF2-K 152 . Bovine aortic endothelial cells were treated at the concentrations indicated with either the FGF2 mutein (fOOf) or with FGF2-K 152 (EOOE) as described in the text, and the cell number was determined after 7 days of culture. Panel D, the effect of DNA on the mitogenic activity of FGF2-K 152 . A total of 10 g of FGF2-K 152 were mixed with the indicated concentrations of plasmid DNA, and 2 ng were immediately added to bovine aortic endothelial cells to determine whether DNA blocks FGF2 activities. Cell number is determined after 7 days. Panel E, mitogenic activity of FGF2, DNA, and polylysine mixtures. Bovine aortic endothelial cells were treated with 2 ng of a mixture of FGF2 (10 g), K 152 (10 g), and varying amounts of plamid DNA, and cell counts were made 7 days later. All assays were performed in triplicate.
␤-gal gene without a promoter (pNASS-␤). Extracts of treated cells were then evaluated for ␤-gal activity. When either DNA was added alone to cells, background ␤-gal expression was detected (Fig. 4B, lanes 3 and 4). When DNA was introduced into cells by FGF2-K 84 , specific ␤-gal activity was observed, but only when the ␤-gal gene was linked to a functional promoter (Fig. 4B, lanes 1 and 2).
In the studies described here, the expression of the reporter gene was normally detected 48 h after treatment of the cells with FGF2-K 84 ⅐DNA. The appearance of this ␤-gal activity is time-dependent (Fig. 4C), and when increasing concentrations of the complex are incubated with cells, there is a concentration dependent increase in ␤-gal enzyme activity (Fig. 4D).
Specificity of Targeting to the FGF Receptor-To determine if the FGF2-K 84 ⅐DNA complex is specifically delivered to cells by FGF receptor-mediated endocytosis, we performed several experiments. First, we showed that the DNA must associate with a covalently linked FGF2-K 84 complex to observe activity (Fig.  5A, lane 1). Treatment of the cells with individual components, or with mixtures of the components (uncondensed) failed to elicit the appearance of ␤-gal activity in treated cells (Fig. 5A,  lanes 2-8). Furthermore, a competition for receptor binding with either unconjugated FGF2 or heparin attenuated the signal generated by the FGF2-K 84 ⅐DNA complex (Fig. 5, B and C). The inhibition of signal was dose dependent with lower concentrations of FGF2 having smaller effects on the inhibition of or not mixed with any conjugate (f). The mixture (10 g/well) was then added to COS cells in triplicate as indicated in the text, and cell lysates were prepared at 48 h to assay for ␤-gal activity (A 405 ). Data are normalized to total protein (A 280 ). Panels B and C, ultrastructural analysis. FGF2-K 84 ⅐DNA complexes were prepared at the various protein: DNA ratios described above and evaluated by EM analysis as described in the text. Panel B shows a toroid formed at a protein:DNA ratio of 2:1. Panel C shows incomplete toroid formation at a protein: DNA ratio of 0.5:1.
reporter gene expression (results not shown). The specificity is further supported by experiments in which cytochrome c or histone H1 was conjugated to polylysine and condensed with DNA. These proteins have no high affinity receptors and no ␤-gal activity is detected in treated cells (Fig. 5D, lanes 3 and  4). Taken together, the findings support the hypothesis that the targeted DNA is introduced into cells via FGF receptors. Because histone binds immobilized heparin and heparan sulfate (the FGF low affinity receptor) and fails to elicit ␤-gal activity, the introduction of DNA by FGF2-K does not appear to be mediated by the low affinity FGF receptor. Accordingly, internalization of the FGF2-K⅐DNA complex likely requires the same trimolecular complex formed by FGF2, heparan sulfate proteoglycans and FGFRs.
Targeting Is Mediated by Passage of the Complex through Endosomes-In order to study how the complex is introduced into cells, we investigated whether the expression of the introduced gene is dependent on the internalized endosome. Treating cells with chloroquine increased expression 8-fold (Fig. 6A,  lane 2) and the inclusion of endosome disruptive peptides (19) in the conjugate increased expression 26-fold (Fig. 6B, lane 3). Concomitant with the changes in the total ␤-gal activity measured in cell extracts, there was a detectable increase in the number of cells that express detectable ␤-gal activity (Fig. 6D). These results further support the involvement of cell surface receptor-specific endocytosis, endosome internalization and release in the cell as the pathway of FGF2-mediated gene delivery.

DISCUSSION
The experiments described here demonstrate the feasibility of introducing DNA into cells via FGF receptors. The specificity of the process for FGF2 is demonstrated in several ways. First, in the absence of FGF2, no gene expression is detected. Second, if the FGF2 binding moiety of the complex is blocked with excess FGF2 or heparin, the response is attenuated. Finally, if FGF2 is replaced by molecules of similar size and charge but with no extracellular receptor (cytochrome c) or no high affinity receptor (histone H1), no gene expression is detected. The introduction of DNA thus appears to be mediated by high affinity receptors, and not by low affinity receptors alone. Finally, modulating the fate of the DNA that is internalized by endosomes (a process characterized by high affinity binding) significantly alters the detectable levels of foreign gene expression. Remarkably, the concentrations of ligand that are used to target DNA into cells are significantly higher than the amounts required for peak mitogenic activity. We attribute this difference to the unoptimized linking of protein to nucleic acid, the large capacity of binding that is required for charge neutralization, and that during condensation and toroid formation only a minor fraction of the FGF2 used is actually available within the complex for high affinity receptor binding. This hypothesis is supported by the inability of DNA alone to inhibit the activity of FGF2 and the equipotency of the conjugate and unconjugated FGF2 (see Fig. 1) Several investigators have attempted to devise strategies to specifically target, transcribe, and translate DNA in specific target cell types. Like here, one approach has been to introduce DNA into cells by attaching it to an antibody or ligand and exploiting the natural specificity of receptor mediated endocytosis. Indeed this mechanism may be shared by viruses: the herpes simplex virus has been reported to bind to the heparan sulfate component of the FGF receptor complex, rhinovirus binds to intercellular adhesion molecule-1, human immunodeficiency virus binds to the CD4 receptor, and Epstein-Barr virus binds to the C3d complement receptor (20 -25), although these reports are controversial (26). More recently, the polymeric immunoglobulin receptor has been used to directly target DNA to respiratory epithelial cells by attaching it to IgA and IgM (27). Alternatively, antibodies to the EGF receptor have been used to target DNA to A549 cells (28), while transferrin (16,29,30) and insulin (31) target human leukemic or epithelial and hepatoma cells, respectively. Although integrin recep- tor binding peptides containing the RGD sequence can also direct DNA (32, 33), very little is known regarding cell-specific ligand carriers of DNA or the different plasmids that they can transport. In at least one instance, a genetic immunotoxin has been described (34) to eliminate the inherent immunogenicity that characterizes immunotoxins.
It is important to note that the number of cells that detectably transcribe and translate the foreign gene in these studies of endosome disruptive peptide (30 g) before their addition to COS cells. Cell extracts were prepared 48 h later, assayed for ␤-gal (A 405 ) activity and normalized to total protein (A 280 ). Panels C and D, staining for ␤-gal activity. FGF2-K 84 ⅐DNA ␤-gal was added in the absence (panel C) or presence (panel D) of the endosome disruptive peptide (as described above), to COS cells. After 48 h, cells were fixed and stained for ␤-gal activity and counterstained with hematoxylin. Similar results were obtained when cells were stained for ␤-gal immunoreactivity (not shown). is low. But the addition of endosome disruptive peptides to the condensate increases the apparent number of cells expressing the foreign gene. This increase is not likely to be due to more DNA entering the cell, but rather to more efficient processing of the DNA once it has been internalized. Therefore, in the absence of endosome-disruptive peptides, the apparent number of cells that internalize DNA is underestimated. We attribute this observation to the physical limit of ␤-gal detection (10 9 enzymes/cell) (35) and the ensuing underestimation of FGF2 receptor-dependent DNA uptake efficiency. Stronger promoters, replicating plasmids, and enhanced delivery of the foreign DNA to the nucleus and cytoplasm will be required for the introduction of therapeutic genes.
The findings described here also extend the scope of previous studies where ligands such as FGF2 have been used to introduce proteins including cytotoxic enzymes like ribosome inactivating proteins into cells (2, 13, 22, 36 -38). In some ways, the ligand-linker-DNA complex described here with FGF-K⅐DNA is similar to a virus; viral particles have coat proteins that are ligands which, as with the use of FGF2 described here, confer cellular tropism to the virus. Selected virion proteins also serve to compact, condense, and contain the viral DNA as with our use of polylysine. In our hands, polylysine and FGF serve multiple purposes. Both bind DNA and help neutralize its inherent negative charge. This effect is what leads to condensation into the compact toroid structures that are compatible with ligand mediated delivery. Optimizing the ratios of polylysine, FGF, and DNA will determine the best conditions for condensation and decrease the amounts of ligand that are currently used in the preparation of ligand-linker-DNA complex.
It is difficult to reconcile all of the findings described here with the overall goal of using ligand mediated gene delivery for sustained, elevated foreign gene expression in target cells. Based on these same studies however and our previous work with a FGF2-saporin mitotoxin (2,13), it is intriguing to speculate that a gene-based mitotoxin could be created that is analogous to protein immunotoxins and mitotoxins. In this instance FGF2 could be used to introduce a gene encoding a molecule like saporin, an enzyme with ribosome inactivating activity that is cytotoxic if introduced into cells. The expression of such a DNA may be feasible but only so long as the mammalian ribosome can synthesize the enzyme prior to its inactivation by the protein synthesized. Because few of these enzymes need be present in the cytoplasm to kill a cell (39), the amount of gene expression required to elicit a biological cytotoxic response should be significantly smaller than the amount of gene transcription and translation that is required for colorimetric detection of ␤-gal (10 9 enzymes/cell) or for that matter expression of a therapeutic protein. Therefore, ligand-mediated gene delivery may be much more amenable to cytotoxic rather than gene replacement therapy. Furthermore, the inclusion of tissue-specific promoters to restrict gene transcription could be used to specifically target cells and provide a layer of control and specificity that is currently not available to mitotoxins or immunotoxins.