Intracellular Accumulation of Secreted Proteoglycans Inhibits Cationic Lipid-mediated Gene Transfer

Molecules secreted by potential target cells may interfere with cationic lipid-mediated gene transfer. This has been studied using human lung fibroblasts and human epidermoid lung cancer cells. Secreted cell medium components caused a substantial decrease both in the uptake of cationic lipid-DNA complexes (2–4-fold) and in reporter gene expression (100–1000-fold). Metabolic labeling of the cell medium showed that especially [35S]sulfate-labeled macromolecules competed with DNA for binding to the cationic lipid. Release of DNA from the cationic lipid by cell medium components was demonstrated by an ethidium bromide intercalation assay. In the presence of the cationic lipid, the secreted macromolecules were internalized by the cells. By enzymatic digestions, it was shown that the competing macromolecules consist of chondroitin/dermatan sulfate and heparan sulfate proteoglycans and that the effects on transfection were mediated by the polyanionic glycosaminoglycan portion of the proteoglycan. Accordingly, pretreatment of cell medium with the polycationic peptide protamine sulfate abrogated the inhibitory effects on gene transfer. Fluorescence microscopy studies revealed that heparan sulfate, internalized as a complex with cationic lipids, accumulated in the cell nuclei. These results support the view that the lack of specificity of this type of gene transfer vehicle is a major hindrance to efficient and safein vivo administration.

More than 300 protocols have been approved for clinical gene therapy studies, in which roughly 20% use nonviral delivery systems (1). The use of viral vectors has potentially many risks, including toxicity associated with expression of viral structural proteins, insertional mutagenesis, and induction of different immune responses (2)(3)(4). With the advantages of being nonimmunogenic and easy to reproduce (purify to homogeneity), synthetic plasmid-liposome complexes constitute the most important alternative to viral vector systems. In particular, substantial efforts have been focused on the development of cationic lipids (CLs), 1 which are efficient carriers for DNA (5)(6), RNA (7), and oligonucleotides (8) in vitro. A number of in vivo studies have also been performed using this type of gene delivery system (9 -11).
Binding of the CL-DNA complex to the cell membrane and entry into the cell are the initial steps of the transfection process. The ratio of CL to DNA is known to be important for optimal efficiency in vitro (5). The optimal ratio occurs when the number of positive charges incorporated into the CL exceeds the number of negative charges on the DNA, which reflects a necessary electrostatic association of the CL-DNA aggregates with the negatively charged cell surface (12)(13)(14). The main mechanism of delivery to mammalian cells is believed to be endocytosis (15)(16)(17)(18), although a membrane fusion mechanism also has been proposed (5,7,12).
The effectiveness of gene delivery in vitro does not correspond to the results obtained in in vivo studies (19 -20). There is some evidence that serum components inhibit gene transfer by altering the size and shape of CL-DNA complexes (21) and that binding of serum proteins, such as albumin, lipoproteins, and macroglobulin, to CL-oligonucleotide complexes interferes with cellular uptake of the complexes (22). It has also been hypothesized that heparin present in serum could be an inhibitor of gene transfer by releasing the DNA from the CL (22,23). Moreover, it has been suggested that gene delivery by synthetic CL could be limited by activation of the complement system (22,24).
Most components in serum with the potential of interacting with gene therapy vehicles are produced by hepatocytes. However, in vivo studies of CL-mediated gene transfer have shown that there are considerable variations in transfection efficiency between different tissues (25)(26), suggesting that tissue-specific factors are important regulators. A better understanding of the nature of the molecules that can potentially interfere with CL-DNA complex-based gene delivery may give rise to new strategies for efficient transfection of specific tissues and cell types in vivo. Molecules that are potential inhibitors of CL-mediated gene transfer include proteoglycans (PGs). Their glycosaminoglycan side-chains consist of repetitive disaccharide units with a wide variety of substitutions, which results in a substantial structural heterogeneity (27). The large number of carboxyl and sulfate groups give them a greater linear charge density than DNA (27). PGs are fundamental components of the pericellular matrix, basement membranes, and other extracellular matrices (28 -31). Cell surface-associated PGs participate in the regulation of cellular proliferation, differentiation, and migration via specific interactions with various growth factors (30 -34).
The objective of this study was to identify secreted molecules that may have diverse negative effects on CL-mediated gene transfer. We present evidence that CL-mediated gene transfer is inhibited by secreted cellular proteins, in particular PGs. PGs form complexes with CLs, leading to the release of DNA and intracellular accumulation of CL-PG complexes. This results in inhibition of DNA uptake (2-4-fold) and a disproportionate inhibition of reporter gene expression (100 -1000-fold). It is also shown that the effects exerted by PGs on CL-mediated gene transfer are mediated by the glycosaminoglycan portion of the PGs, i.e. chondroitin sulfate (CS), dermatan sulfate (DS), and heparan sulfate (HS). Protamine sulfate, a polybasic peptide that neutralizes the anticoagulant activity of glycosaminoglycans (35), abrogated the inhibitory effects exerted by PGs. Finally, by fluorescence microscopy it was shown that glycosaminoglycans, internalized by a CL-dependent route, accumulate in the cell nuclei.
Preparation of Radiolabeled, Conditioned Cell Culture Medium-Subconfluent monolayers of HFL-1 or A 549 cells in 25-cm 2 culture flasks (approximately 1.5 ϫ 10 6 cells) were labeled with Na 2 35 SO 4 (50 Ci/ml) and L- [4, H]leucine or [ 35 S]methionine (20 Ci/ml) in 4 ml of fresh MEM for 24 h. The medium was aspirated, and cells were extensively washed with MEM to remove free radioisotope. This was followed by another incubation in 4 ml MEM for 24 h. The cell culture medium (CM) was reisolated for cellular uptake experiments. In some cases, the same procedure was performed without radioisotopes to obtain unlabeled CM used in DNA uptake and gene expression experiments, respectively.
Luciferase Gene Expression Assay-HFL-1 and A 549 cells were plated in 24-well plates at 1 ϫ 10 5 cells/well (approximately 50% confluency) in 0.5 ml of growth medium 24 h before transfection. To prepare CL-DNA complexes, pGL3 plasmid DNA (1 g/ml) and the desired amount of DOSPA/DOPE or DOGS were separately diluted in 100 l of MEM or CM, mixed by inversion, and incubated for 30 min at room temperature. The DNA-lipid mixture was diluted to a final volume of 250 l, vortexed, and then added to cells that had been rinsed twice with MEM. At the end of the 4-h incubation at 37°C in 5% CO 2 , the medium was aspirated and replaced with 0.5 ml of growth medium. After an additional period of incubation for 48 h, cells were washed twice with MEM and treated with 150 l of cell lysis reagent containing 25 mM Tris-HCl, pH 7.8, 2 mM CDTA, 2 mM dithiothreitol, 10% glycerol, 1% Triton X-100 (Promega) for 10 min at 4°C followed by scraping. Luciferase expression was quantified in 5 l of the cell lysate supernatant, using a luciferase assay kit. Light emission was measured by integration over 30 s at 25°C using an EG & G Berthold Microlumat LB 96P luminometer. Luciferase activity was normalized to the protein content of each sample, determined with the Pierce BCA protein assay.
Labeling of Plasmid DNA with [ 32 P]dCTP-Two g of plasmid DNA pGL3 was labeled with 32 P by nick translation using Dnase I (100 pg), DNA polymerase I (5 units), and 0.7 nmol of dATP, dGTP, and dTTP and 33 pmol of [ 32 P]dCTP for 2 h at 15°C in 50 mM Tris-HCl, pH 7.5, 10 mM MgCl 2 , 50 g/ml bovine serum albumin, 0.1 mM dithiothreitol (final volume of 110 l). The reaction was terminated by the addition of 10 l 0.2 M EDTA. Labeled DNA was purified from unincorporated nucleotides using Microspin S-200 HR columns according to the manufacturer's instructions, followed by precipitation with 0.1 volume of 3 M sodium acetate and 2.5 volumes of 95% EtOH. DNA was recovered by centrifugation at maximum speed (14,000 rpm) in a bench-top centrifuge for 4 min. After washing with 70% EtOH, isolated DNA was assessed for integrity by 0.8% agarose gel electrophoresis at 70 V in Tris acetate EDTA buffer for 45 min. [ 32 P]DNA plasmid bands were detected by EtBr staining.
Uptake of Radiolabeled DNA by Cells-HFL-1 and A 549 cells were seeded in 24-well plates at 1 ϫ 10 5 cells/well in 0.5 ml of growth medium. After 24 h, cells were rinsed twice with 0.5 ml of MEM and then supplemented with 1 g/ml of pGL3 [ 32 P]DNA (specific activity, 10 7 cpm/g) and various amounts of CL, as described above. After 4 h of incubation at 37°C in 5% CO 2 , cells were transferred to an ice bath and washed with ice-cold MEM, and then detached by trypsin (0.5 mg/ml) treatment in 10 mM KH 2 PO 4 , pH 7.5, 0.15 M NaCl (PBS), for 1 min at 20°C, followed by suspension in 0.5 ml of growth medium. Cells were pelleted by centrifugation, followed by two washings with MEM. This treatment removed CL-DNA complexes that were unspecifically attached to the cells. Finally, cells were lysed in 0.5 ml of cell culture lysis reagent (defined above) for 10 min at 20°C, and [ 32 P]DNA radioactivity was determined in 250 l of the cell lysate by liquid scintillation counting using Readysafe scintillation mixture (Beckman) and a Wallac Rack Beta counter (Amersham Pharmacia Biotech). Cells from parallel cultures were counted in a Bü rker chamber, and cell-associated DNA was expressed as ng/10 6 cells.
Ethidium Bromide Intercalation Assay-Plasmid DNA (20 g/ml) was preincubated with EtBr (2.0 g/ml) either in MEM or CM (prepared as described above) in a 96-well Microfluor plate. The indicated amount of DOSPA/DOPE or DOGS was added to give a final volume of 100 l. EtBr fluorescence at 610 nm was continuously monitored using a Perkin-Elmer LS-5B spectrometer (excitation wavelength of 500 nm) and a Perkin-Elmer plate reader. The fluorescence signal stabilized in less than 5 min after the addition of the CL and was presented as the fraction in percent of the maximum fluorescence signal obtained without CL (control). Control values for the different media used were equal.
Cellular Uptake of Radiolabeled Components from Conditioned Cell Culture Medium-HFL-1 and A 549 cells were seeded in 24-well plates at 1 ϫ 10 5 cells/well in 0.5 ml of growth medium. After 24 h, cells were rinsed twice with 0.5 ml of MEM and then supplemented with a mixture of fresh radiolabeled CM, prepared as described above, 1 g/ml DNA plasmid, and the desired amount of either DOSPA/DOPE or DOGS. After preincubation for 30 min at room temperature, the mixture was diluted to a final volume of 0.5 ml, vortexed and then added to the cells. At the end of the 4-h incubation at 37°C in 5% CO 2 , the medium was aspirated, and cells were transferred to an ice bath and washed with ice-cold MEM, and then detached by trypsin treatment for 1 min, followed by suspension in 0.5 ml of growth medium. Cells were pelleted by centrifugation followed by washing with MEM. Then cells were lysed in 0.5 ml of cell lysis reagent for 10 min at 20°C, and cell-associated [ 35  S]methionine, 1 g/ml plasmid DNA, and various amounts of DOSPA/DOPE. DNA plasmid was included in order to mimic the conditions of the experiments described above. After incubation for 4 h at 37°C, the medium was removed, followed by extensive washing with PBS. Cell layers were extracted with 2.5 ml of 2% (v/v) Triton X-100 in PBS containing 10 mM EDTA, 10 mM Nethylmaleimide (NEM), and 1 mM di-isopropyl fluorophosphate for 10 min at 4°C. Extracts were then subjected to either of the following procedures: 1) precipitation with 8 volumes of 95% ethanol, or 2) dilu-tion with 1.3 volumes of 7 M urea, 10 mM Tris, pH 7.5, containing 0.1% (v/v) Triton X-100 and 10 mM NEM followed by chromatography on DEAE-cellulose columns (1 ml) equilibrated in the same urea buffer as above. After sample application, the columns were washed with 10 bed volumes of the urea buffer and 10 volumes of 50 mM Tris-HCl, pH 7.5, followed by elution with 3 bed volumes of 4 M guanidinium chloride, 50 mM NaOAc, pH 5.8, 0.2% (v/v) Triton X-100, 10 mM NEM, and 5 g/ml bovine serum albumin. Eluted material was recovered by precipitation with 8 volumes of ethanol. Samples from procedure 1 were dissolved in 150 l of 5% (w/v) SDS buffer and subjected to SDS electrophoresis on 3-12% polyacrylamide gradient gels with a 3% stacking gel, as described elsewhere (37).
Material isolated by procedure 2 (polyanionic macromolecules) were dissolved in 0.4 ml of the 4 M guanidinium chloride-containing elution buffer (without bovine serum albumin) and chromatographed on Superose 6 HR 10/30 (flow rate, 0.4 ml/min), equilibrated in the same buffer. One-min fractions were analyzed for 35 S or 3 H radioactivity by scintillation counting. Pooled material was concentrated in Centriplus 30 concentrators, recovered by ethanol precipitation, and then applied to a Mono Q HR 5/5 column equilibrated in 7 M urea, 10 mM Tris, 0.1% Triton X-100, pH 8.0 (starting buffer). After a 5-ml wash with starting buffer, the column was eluted with a linear gradient of 0.3-1.2 M NaCl in the same buffer over 60 min. The flow rate was 0.5 ml/min, and 0.5 ml fractions were collected. Aliquots of the collected fractions were analyzed by scintillation counting and pooled as indicated for further analysis.
Enzymatic Degradation-Material collected from ion-exchange chromatography on Mono Q was recovered by ethanol precipitation and dissolved in the appropriate buffer. Digestions with chondroitin ABC lyase (50 milliunits/ml) were performed in 200 l of 10 mM Tris-HOAc, pH 7.3, 10 mM Na 2 EDTA, 10 mM NEM at 37°C for 10 h. Digestions with HS lyase (3 milliunits/ml) were conducted in 200 l of 3 mM Ca(OAc) 2 , 10 mM Hepes-NaOH, pH 7.0, 10 mM NEM at 37°C for 10 h. Digested samples were lyophilized, dissolved in 0.5 ml of the starting buffer, and reapplied to the Mono Q column for linear gradient elution as described above.
Fluorescence Microscopy-HFL-1 cells were seeded at a density of 50 ϫ 10 4 cells/well in 24-well plates in growth medium. After 24 h, cells were rinsed with MEM and then supplemented with 1 g/ml of AMAC-HS with or without 8 g/ml DOSPA/DOPE in MEM. At the end of the 4 h incubation at 37°C, medium was aspirated, and cells were trypsinized, suspended in growth medium, centrifuged, resuspended in growth medium, and replated on 4-well coverslips. With this procedure, unspecifically bound AMAC-HS was removed from the cells. After an additional incubation period of 24 h, cells were washed three times with PBS and fixed in 2% paraformaldehyde in PBS for 30 min, followed by extensive rinsing in PBS and distilled water. Cells were then viewed using a Nikon Diaphot 300 microscope and a Bio-Rad MRC-1024 confocal laser scanning microscope system. The optics used to separate fluorescence signal from noise was a dichroic mirror (562DCLP) and a band-pass filter (D540/30) from Chroma Technology, Inc. The light source used was a Kr ϩ -laser (Coherent Innova 300) supplying 2-mW, 413-nm continuous light. The images were digitized and transferred to a computer workstation for merging, annotation, and printing.
The following procedure was performed to assess the integrity of AMAC-HS isolated from cells: 1.5 ϫ 10 6 cells were seeded in a 25-cm 2 cell culture flask and grown for 24 h. Cells were then incubated with HS-AMAC (100 g) and CL (800 g) as described above. After the second incubation period of 24 h, cells were detached by scraping in PBS, centrifuged, and then lysed in the cell extraction buffer. The cell lysate was then passed through a DEAE-cellulose column (1 ml) as described above, in order to separate free AMAC from intact AMAC-HS. Eluted AMAC-HS was recovered by precipitation with ethanol and then dissolved in 200 l of destilled water. The recovered material from cell cultures and standard solutions of AMAC-HS were then analyzed by fluorescence spectroscopy using a Perkin-Elmer LS-5B spectrometer (excitation wavelength of 430 nm) and a Perkin-Elmer plate reader (emission wavelength of 520 nm).

Inhibition of Cationic Lipid-mediated Gene Transfer by Conditioned Cell Medium
Components-We examined the influence of secreted cell molecules on CL-mediated gene transfer in two different types of cells: HFL-1 (human lung fibroblasts) and A 549 (a human lung carcinoma cell line). The CL formulations used were DOSPER/DOPE and DOGS, which are commercially available under the trade names LipofectAMINE and Transfectam, respectively. They are both lipospermines, carrying spermine, a natural counterion of DNA in vivo, as the DNA-binding group. The former transfection reagent also contains the neutral colipid DOPE. Throughout this study, these transfection reagents behaved very similarly.
CL-mediated transfection was inhibited by CM in both HFL-1 and A 549 cells (Table I). At the highest lipid concentration used (16 g/ml), transfection was inhibited by 99.5 and 99.9% in HFL-1 and A 549 cells, respectively. There was a substantial difference in transfection efficiency between the two cell lines (approximately 37-fold higher in A 549 compared with HFL-1). Conditioned CM obtained after shorter incubation periods was less inhibitory, but CM isolated already after 8 h significantly reduced gene expression. CM obtained after a 24 h-incubation was chosen for further studies. In a control experiment, plasmid DNA was incubated either in MEM or CM for 24 h at 37°C and then analyzed for integrity by agarose gel electrophoresis. There was no difference in the recovery of intact plasmid (approximately 87%), suggesting that the inhibition of gene expression by CM was not due to increased degradation of the reporter gene plasmid. Moreover, preexposure of HFL-1 cells to CM for 4 h, prior to the addition of CL-DNA complexes, did not affect gene transfer efficiency. However, preexposure of cells for 4 h to CM supplemented with 16 g/ml CL inhibited gene expression by approximately 85% compared with cells preexposed to MEM supplemented with CL (results not shown).
We then investigated whether the decrease in gene expression caused by CM originated from a reduced uptake of CL-DNA complexes. As shown in Fig. 1, there was a significant reduction of the amount of internalized DNA plasmid when the experiments were performed in CM as compared with MEM. The decrease in DNA uptake (approximately 60% at 16 g/ml CL) was not proportional to the inhibition of gene expression. Cells preexposed to CM supplemented with 16 g/ml CL internalized an equal amount of DNA plasmid as compared with TABLE I Influence of CM components on CL-mediated transfection efficiency HFL-1 and A 549 cells were plated in 24-well dishes at 1 ϫ 10 5 cells/well (24 h prior to the time of transfection) and then transfected with CL complexed with 1 g/ml of DNA plasmid at a 2:1 to 16:1 (w/w) ratio (corresponding to a theoretical charge ratio of 0.8 -6.4) either in CM or in MEM (control). After incubation for 4 h at 37°C, the medium was aspirated, and cells were rinsed twice with MEM. Growth medium was added, and the incubation was allowed to proceed for another 48 h. Then cells were washed with MEM and subjected to a luciferase assay as described under "Experimental Procedures." Luciferase activity is expressed as mean Ϯ S.E. (n ϭ 6). It is worth noting that the difference in luciferase activity between the cell lines did not correspond to the difference in DNA uptake (approximately 1.3 times higher in HFL-1 cells compared with A 549 cells at a CL concentration of 16 g/ml).
Effect of Cell Medium Components on Cationic Lipid-DNA Complex Formation-One possible explanation for the inhibitory activity of CM on the uptake of CL-DNA complexes is that the complexes dissociate upon interaction with CM components. We therefore studied the stability of CL-DNA complexes in CM using an EtBr intercalation assay. When EtBr is mixed with double-helical DNA in solution, it intercalates between the base pairs, emitting an intense fluorescence signal at 610 nm when excited at 500 nm. It has been reported that the addition of CL results in an immediate and substantial (approximately 90%) decrease of the fluorescence signal, as a result of the displacement of EtBr from the DNA (38). In the present study, a dose-dependent inhibition of the fluorescence signal was obtained with DOSPER/DOPE (Fig. 2, f). When the same experiment was performed in CM, a significant amount of the EtBr fluorescence was recovered, suggesting a reversal of CL-DNA complex formation and increased accessibility of the EtBr intercalation site (Fig. 2, Ⅺ). The detection limit of the EtBr intercalation assay (20 g/ml DNA) did not permit the use of the same concentration of CL and DNA as in the gene expression and DNA uptake experiments (1 g/ml DNA, see Table I and Fig. 1).
Cationic Lipid-dependent Cellular Uptake of Cell Medium Components-Altogether, the results described above suggest that CM inhibits CL-mediated uptake of DNA plasmid by interfering with the formation of CL-DNA complexes, resulting in reduced gene expression. However, the decrease in DNA uptake was not proportional to the inhibition of gene expression. Moreover, preexposure of cells to conditioned CM supplemented with CL significantly reduced gene expression, although subsequent uptake of DNA in MEM was equal as compared with control (cells preexposed to MEM supplemented with CL).
In order to gain more insight into the mechanism of these effects, CM that had been metabolically labeled with [ 35 S]sulfate and [ 3 H]leucine was mixed with CL and then added to unlabeled HFL-1 and A 549 cells, respectively (Fig. 3). DNA plasmid (1 g/ml) was included in the respective media in order to mimic the experimental conditions in previous experiments.

Control cell cultures (no CL added) internalized a significant amount of 3 H-labeled protein, whereas the uptake of [ 35 S]sulfate-labeled compounds was very limited in both cell types. Addition of CL caused a dose-dependent increase in the uptake of both [ 3 H]leucine-and [ 35 S
]sulfate-labeled material (up to 8 and 16 g/ml DOSPER/DOPE in HFL-1 and A 549 cells, respectively). In particular, internalization of components labeled with [ 35 S]sulfate was favored by the presence of CL. At the CL concentrations indicated above, approximately 85 and 79% of total [ 35 S]sulfate-labeled material in CM from HFL-1 and A 549 cells, respectively, was associated with the cells. These results suggest that CL-DNA complex formation was inhibited primarily by [ 35 S]sulfate-labeled components in CM and that these components were internalized as a complex with the CL.
Identification of Internalized Cell Medium Components and Their Effects on Gene Transfer-To study the nature of the internalized CM components, detergent extracts of cells incubated with radiolabeled CM either in the absence or in the presence of various amounts of CL were analyzed by SDSpolyacrylamide gel electrophoresis. As shown in Fig. 4, CL caused a dose-dependent increase in the uptake of [ 35 S]methionine-labeled proteins (corresponding to the results presented in Fig. 3), in particular high molecular weight components.
In a similar set of experiments, [ 35 S]sulfate-labeled CM compounds, taken up by the cells in the presence of CL, were recovered by ion exchange chromatography on DEAE-cellulose (approximately 90% recovery of the starting material) and then analyzed by gel filtration chromatography, revealing the existence of one major component eluting with the void volume and one minor more retarded component (Fig. 5A). These components were further studied by ion-exchange chromatography on Mono Q eluted with a linear gradient (Fig. 5B). This procedure separated the material into two distinct populations, eluting at approximately 0.83 (pool I) and 1.0 M NaCl (pool II). A series of degradation experiments were then performed to specifically identify the sulfated, polyanionic macromolecules. The differential susceptibility of the material in pool I and pool II to HS lyase and chondroitin ABC lyase, showed that pool I consisted entirely of HSPG (Fig. 5C), whereas pool II contained a mixture of HSPG and CS/DSPG (Fig. 5D). Treatment of the material in pool II with both lyases resulted in a complete degradation (data not shown). By gel filtration chromatography on Superose 6, it was confirmed that the [ 35 S]sulfate-labeled CM compounds consisted of intact PGs rather than free glycosaminoglycan chains, as the latter (released from the protein core by alkali treatment) exhibited a significantly more retarded elution position on Superose 6 (data not shown).
When the experiments presented in Fig. 5 were conducted with CM labeled with [ 3 H]GlcN, the same results were obtained, indicating that other glycoproteins and hyaluronic acid were not internalized as a complex with CL, under the condi-tions described here (data not shown).
PGs purified from CM, as shown in Fig. 5, were tested for inhibitory activity on CL-mediated gene transfer. The inhibition of gene expression by PGs almost mirrored that of CM as a whole, indicating that PG was the major inhibitory component in CM (Table II). Pretreatment of PG with HS lyase or ABC lyase diminished the inhibitory activity by approximately 75 and 25%, respectively, suggesting that HSPG played a more important role than CS/DSPG. Moreover, exogenous addition of either HS or DS at concentrations as low as 5 g/ml completely abolished CL-mediated gene transfer, suggesting that the glycosaminoglycan portion of the PG was mediating the inhibition of gene transfer. It was not possible to digest the PG with the corresponding lyases directly in CM.
Pretreatment of Cell Culture Medium with Protamine Sulfate Restores Transfection Efficiency-Protamine sulfate is a United States Food and Drug Administration-approved compound with a known neutralizing effect on the anticoagulant activity of heparin and DS in vivo (35), due to its polycationic nature. DNA plasmid uptake and gene expression were measured either in CM or MEM, pretreated with increasing concentrations of protamine sulfate. An optimal ratio of DNA to DOSPER/DOPE, as determined in the experiments described above (see Table I and Fig. 1) was used. As shown in Fig. 6A, protamine sulfate specifically potentiated DNA uptake in cells incubated with CM. At the highest protamine sulfate concentration used (100 g/ml), DNA uptake was almost 3-fold greater as compared with the control (no protamine sulfate added). Addition of protamine sulfate to MEM only had a limited effect on DNA uptake by HFL-1 cells. Accordingly, pretreatment of CM with 10 g/ml protamine sulfate restored gene expression to the same level as in MEM (Fig. 6B). Furthermore, pretreatment of CM with protamine sulfate reduced the CL-mediated intracellular accumulation of [ 35 S]PG (data not shown).
Nuclear Accumulation of Glycosaminoglycans-Confocal microscopy studies with rhodamine-labeled DNA plasmid confirmed the results presented above. The uptake of DNA was significantly reduced by CM components and exogenous glycosaminoglycans. However, we were not able to observe an altered intracellular distribution of rhodamine-DNA as compared with cells incubated with MEM (results not shown).
We then studied the effect of CL on the cellular uptake of DS  and HS, using 125 I-labeled polysaccharide chains. Fig. 7A shows that the addition of CL increased the uptake of both HS and DS chains in a dose-dependent manner, corresponding to the results presented in Fig. 3. As the nucleus is the intracellular target for transduced DNA, it was of interest to study the subcellular localization of internalized glycosaminoglycan chains. HS, end-labeled with the fluorophore AMAC, was added to cells either with or without CL (8 g/ml) for 4 h, followed by an incubation period of 24 h in growth medium. In the absence of CL, the amount of intracellular HS was low with a diffuse extranuclear localization (Fig. 7B). On the contrary, in the presence of CL, there was a substantial nuclear accumulation of HS and a relatively low amount of HS in the extranuclear compartment (Fig. 7C). Altogether, CL caused an increase in the amount of internalized glycosaminoglycans and nuclear accumulation of HS. This pattern was obtained after a growth period of approximately 24 h. Analysis of cells exposed to HS-CL complexes after shorter growth periods showed a less pronounced nuclear accumulation and a stronger extranuclear fluorescence (results not shown). Intracellular AMAC-HS was assessed for integrity by reisolation of the material from cell cultures as described under "Experimental Procedures," showing that all of the AMAC-HS remained intact (results not shown).

DISCUSSION
This study provides the first evidence that molecules secreted by cells are inhibitory to CL-mediated DNA uptake and gene expression. Our results indicate a dramatic inhibitory effect on gene expression caused by intracellular accumulation of CM components, in particular sulfated PGs. PGs isolated from CM were found to inhibit CL-mediated gene transfer almost to the same degree as whole CM, and this effect could be reversed by enzymatic treatment of PGs with glycosaminoglycan lyases. Exogenous glycosaminoglycans also abolished CLmediated gene transfer. CM components interfered with CL-DNA complex formation, resulting in reduced cellular uptake of DNA plasmids with a concomitant increase in the uptake of sulfated PGs. Low amounts of CM-derived PGs were internalized in the absence of CL. Cells preexposed to CM supplemented with CL exhibited a substantial decrease in gene expression, although DNA uptake activity was unaffected, suggesting that intracellular accumulation of CM components per se results in inhibition of gene expression. Preexposure of cells to CM without CL affected neither gene expression nor DNA uptake. Zabner et al. (16) presented evidence that the most important barriers to CL-mediated transfection are the intracellular dissociation of DNA from CL and the movement of DNA from endosomes into the nucleus, respectively. One possible explanation of our results is that intracellular accumulation of  PGs/glycosaminoglycans interferes with these processes.
It has been demonstrated by fluorescence resonance energy transfer and gel shift assay studies that heparin displaces oligonucleotides from another type of CL (22). However, under physiological conditions in vivo, it is likely that CS/DS and HSPG/glycosaminoglycans, rather than heparin (which is mainly confined to the mast cell granules), will act as barriers to CL-mediated gene transfer. Spermine, the cationic group of DOSPA and DOGS, has been shown to interact with DS (39) and HS (40) with similar or higher affinity, respectively, than with DNA.
Fibroblasts have a major influence on the production and assembly of the extracellular matrix, which makes HFL-1 cells suitable for the aim of this study. As a comparison, we selected a cancer cell line (A 549) from the same source. The difference in gene transfer efficiency between HFL-1 and A 549 cells was probably due to the superior ability of A 549 cells to replicate pGL3 plasmid DNA, which is in accordance with a previous report, in which human and murine melanoma cells internalized equivalent amounts of CL-DNA complexes, yet the level of transgene expression differed considerably between the cell lines studied (41).
By confocal microscopy studies, we also demonstrated that exogenous glycosaminoglycans, such as HS, accumulate in the cell nuclei, a process that seems to be dependent on CL. It has been suggested that nuclear HS may play a role in regulating nuclear activities (42). In the same report, exogenous HS was reisolated from hepatocytes, in which approximately 10% of the internalized material was confined to the nuclei. In our study, an insignificant amount of HS was internalized in the absence of CL, with no localization to the cell nuclei, which partly may be explained by the relatively short period of incubation. It is noteworthy that HS variants exhibiting strong binding to spermine inhibit the proliferation of human lung fibroblasts by up to 63% (40).
A number of investigations have been directed at studying the production of PGs in fibroblasts, showing that the major forms secreted to the extracellular matrix are the CS/DSPGs decorin and versican. Fibroblasts also secrete a HSPG with a 250 -400-kDa protein, probably perlecan (37,43,44). Conditioned media may also contain shed cell-surface HSPGs, such as syndecan, glypican, or betaglycan. Syndecans can be hybrid PGs carrying both HS and CS (30). As one of the major challenges of future gene therapy studies is to treat diseases such as cancer, it is noteworthy that elevated quantities of PGs and glycosaminoglycans have been found in tumor tissue (45,46).
The ability of serum components to inhibit CL-mediated gene transfer has been studied previously (21,22). The amount of internalized CL was not affected by the presence of 10% serum, and the reduction in DNA uptake (2-3-fold) was disproportionate to the 100-fold decrease in gene expression (21). In our study, DNA uptake was reduced by 2-4-fold by CM, whereas gene expression was reduced by 100 -1000-fold. Our results would suggest that the unaltered uptake of CL in their study was due to the internalization of CL as a complex with serum components in place of the DNA plasmid and that the substantial inhibition of gene expression was caused by the intracellular accumulation of PGs and other proteins found in serum.
In a previous report, it was suggested that HSPGs are involved in the internalization of polylysine-DNA complexes (14). Polylysine-mediated gene transfer was substantially reduced in Chinese hamster ovary cells deficient in PG biosynthesis compared with wild-type cells. In a recent report, the role of PGs in CL-mediated gene transfer in vivo was investigated (47). Pretreatment of mice with a HS-degrading enzyme prior to intravenous administration of CL-DNA complexes resulted in reduced levels of reporter gene expression. The authors present two possible mechanisms for the inhibition by HS lyase pretreatment: either 1) destruction of cell-surface HSPG necessary for the internalization of CL-DNA complexes, or 2) enzymatic release of glycosaminoglycan chains in the tissues, preventing the CL-DNA complexes from reaching their target cells. Either of these mechanisms indicates a role for PGs in CL-mediated gene transfer in vivo.
The use of synthetic polycationic lipids for the delivery of negatively charged nucleic acids has been proven to be an important alternative to viral gene transfer. Although in vivo gene delivery mediated by CL does occur (9 -11), a synthetic compound relying on electrostatic interactions with the gene to be delivered is likely to exhibit unwanted interactions with other negatively charged components residing in the extracellular matrix and in the serum. Potential strategies to overcome the PG-mediated inhibition of gene transfer in vivo are suggested by our study. Pretreatment with the polycationic peptide protamine sulfate neutralized the inhibition of gene transfer by CM. Because protamine sulfate has been proven to be nontoxic and only weakly immunogenic in humans, this compound may be useful for gene therapy in vivo. Indeed, the addition of protamine sulfate to CL-DNA complexes has been shown to increase gene transfer efficiency in vivo (48). Preinjection of polybasic compounds such as protamine sulfate could potentially block the inhibitory action of PGs and other polyanionic compounds in serum and the extracellular compartment.
In conclusion, our results support the view that the lack of specificity of this class of gene transfer vehicles is a major hindrance to efficient and safe in vivo administration. Future studies should be aimed at developing gene delivery vehicles based on structures with higher binding specificities for DNA. Another possibility would be to manipulate the production and distribution of PGs and other components capable of interfering with CL-mediated gene transfer.