QUARTERLY Communication

The purpose of this study was to investigate the effect of endotoxin presence in plasmid DNA preparations on the efficiency of transfection achieved in vivo with B16(F10) and Renca tumors and to determine transgene localization. Our data show that endotoxin markedly decreases the efficiency of transfection. Furthermore, the transgene transferred in vivo can be found in both neoplastic and normal (most likely myofibroblast) cells lying in proximity of the administration site.

that endotoxin markedly decreases the efficiency of transfection.Furthermore, the transgene transferred in vivo can be found in both neoplastic and normal (most likely myofibroblast) cells lying in proximity of the administration site.
In vivo DNA transfer to cells, both normal and neoplastic, can be achieved by a variety of methods.Most often, attempts to transfer therapeutic DNA in vivo make use of viral and synthetic carriers or physical methods such as electroporation and gene gun (for the latest review see [1]).Numerous data indicate that plasmid DNA alone ("naked" DNA) may be transferred in vivo to various normal and neoplastic cells without the use of any carriers [2,3].Such an approach appears simple, rela-tively inexpensive and has been adapted to the administration of so-called DNA vaccines (see [4]).Unexpectedly, the efficiency of naked DNA transfection into neoplastic cells in vivo is higher than that obtained with cationic lipids [5].Some data, also from our laboratory, indicate that this DNA transfer method may be successfully used in experimental gene therapy [5][6][7].An essential problem in genetic manipulations involving plasmid DNA is the degree of its purification from bacterial endotoxin (lipopolysaccharide, LPS) [8][9][10].Endotoxin derived from Gram-negative bacteria is strongly toxic (LPS is an extremely potent stimulator of the mammalian immune system).Hence, in order to be used in gene therapy, plasmid DNA preparations isolated from E. coli should be of greatest purity available (the so-called pharmacological grade).The purpose of this study was to investigate the influence of endotoxin contamination in plasmid DNA preparations upon the efficiency of in vivo transfection with plasmid DNA alone, without any carriers.The presence of bacterial endotoxin in plasmid DNA preparations has been otherwise known to exert a clear effect on in vitro transfection using cationic lipids, decreasing its efficiency [11].In this study we also attempted to investigate the localization of the inserted transgene in various types of cells that form tumors.

Plasmids.
In this study we used the following plasmids: pVR1255 [12] (obtained from Dr. R. Zaugg, Vical Inc., San Diego, U.S.A.) containing luciferase reporter gene under the control of CMV promoter and two plasmids containing the lacZ gene, i.e. pCMVlacZ (Clontech) with the Eschericha coli b-galactosidase gene under the control of CMV promoter and pCMVnlacZ showing nuclear localization obtained from Dr. L. Sadowska (Université de Génève, Faculté de Biologie Moléculaire, Génève, Switzerland).
Plasmid DNA isolation and purification from endotoxin.Plasmid DNA preparations were isolated from spheroplasts according to Wicks et al. [9].Plasmid DNA was separated from endotoxin and other contaminants using Sephacryl S-1000-filled column (Sephacryl S-1000 Superfine, Pharmacia Biotech, column height = 100 cm, diameter = 1.6 cm).Other details were described by Horn et al. [10].
Determining endotoxin concentration in plasmid DNA preparations.Fifty ml of LAL solution (Limulus Amebocyte Lysate, BioWhittaker) was added to 50 ml aliquots of DNA solutions of varying concentrations (0.005 mg/ml, 0.001 mg/ml, 0.0005 mg/ml).Samples were incubated for 10 min at 37°C.Then, 100 ml of Chromogenic Substrate solution (BioWhittaker) was added and the sample incubated for 6 min at 37°C.The reaction was stopped by adding 100 ml of 10% SDS solution (BDH Biochemicals).Endotoxin concentration in the DNA preparations was checked by measuring absorbance at l = 405 nm.The degree of DNA preparations' purification from endotoxin was 700-1000-fold.One mg of DNA contained from 0.03 to 0.04 LPS units.
Animals.In vivo transfection.Six-to eight-week-old BALB/c and C57BL6 mice from on-site Animal Facility were used throughout the experiments.Mice's left dorsal side was shaved and the animals were inoculated subcutaneously with either 5 ´10 5 Renca cells/100 ml PBS per mouse (PBS : standard PBS buffer, pH 7.2, without Ca 2+ and Mg 2+ ) or with 2.5 ´10 6 B16(F10) cells/100 ml PBS per mouse.Ten mg of luci- ferase gene-containing pVR1255 plasmid DNA alone, or 10 mg of pVR1255 plasmid DNA complexed to varying amounts of cationic liposomes DC-Chol/ DOPE [13] were injected into tumors approximately 5-10 mm in diameter using 200 ml Ringer's solution with 0.05% lactose added.DC-Chol/DOPE cationic lipids (1:1, w/w) were prepared as described previously [14].After 24 h mice were killed and tumor material collected in order to measure the activity of the reporter gene.
Measuring luciferase gene activity in tumor specimens.Tumors excised from animals treated with pVR1255 plasmid DNA were homogenized in 1. >-Galactosidase activity detection.pCMVnlacZ plasmid DNA (50 mg) was transferred directly, without any carriers, to the growing neoplastic tumors.After 24 h, tumors were excised and fixed at 4°C in 4% paraformaldehyde in PBS containing two types of detergent: NP-40 (final concentration = 0.02%) and sodium deoxycholate (final concentration = 0.01%).After 4 h incubation tumors were immersed overnight in staining solution containing X-gal (5-bromo-4-chloro-3-indolyl-b-D-galactopyranoside, Sigma) and the above detergents [6].Stained tumors were freezed in isopentane and liquid nitrogen and then sliced using cryostat (-20°C) into slices 10 mm-thick that were subsequently placed on gelatin slides.The preparations were stained with hematoxylin-eosin, fixed, covered and dried.Stained tumor cells were examined under a microscope (total magnification 850´).

RESULTS AND DISCUSSION
The efficiency of in vivo transfection with plasmid DNA has a significant implication for the efficacy of prospective gene therapy: the higher the percentage of transfected cells the greater the production of the therapeutic protein.The impact of plasmid DNA preparations' contamination with bacterial endotoxin on the efficiency of in vivo transfection has not been reported so far.Experiments performed on rodents showing their remarkable resistance to LPS [9] have, as a rule, involved plasmid DNA preparations not puri-fied from endotoxin.Nonetheless, a negative effect of the contamination of plasmid DNA with endotoxin was previously described in a report from in vitro studies performed on cell cultures transfected with cationic liposomes [11].Studies on in vivo transfection of B16(F10) neoplastic tumors with either endotoxin-contaminated or purified DNA show (Fig. 1) that such contamination markedly decreases transfection efficiency with both naked DNA and DNA complexed to cationic lipids (in this case DC-Chol/DOPE).This finding does not appear to depend on neoplastic cell type (a similar effect was seen with Renca cells, data not shown).So far, we are unable to determine at which stage of DNA transfer (e.g.cellular uptake or trans-Direct in vivo transfer of plasmid DNA into murine tumors  Ten mg of pVR1255 plasmid DNA (containing luciferase gene) was injected using Ringers solution supplemented with 0.05% lactose (200 ml total) into growing B16(F10) neoplastic tumors either as naked DNA without any carriers or using cationic liposomes DC-Chol/DOPE at various liposomes/DNA ratios (0.5:1, 1:1, 3:1, w/w).Plasmid DNA used in the experiments was either endotoxin-contaminated or endotoxin-free.Endotoxin-contaminated DNA contained 26.27 LPS units/mg DNA while purified DNA contained 0.03 LPS units/mg DNA.Luciferase activity was measured in B16(F10) tumor cell lysates 24 h post DNA transfer.Each data point represents average luciferase activity (±S.D.) in cell lysates obtained from 5 different tumors.Y-axis is logarithmic.
port to the nucleus) is the endotoxin actually involved.
We cannot exclude the possibility that endotoxin has some influence on the luciferase gene expression rather than on the transfection efficiency itself.However, we believe that it is the transfection efficiency that is affected by endotoxin.Weber et al. [11] have shown that addition of endotoxin to previously purified plasmid DNA preparations exerts a negative effect on the efficiency of in vitro transfection.Also, our own experiments in vivo have shown that when plasmid DNA was transferred into cells by electroporation, i.e. when transfer occurs via micropores formed and not due to endocytosis [15], no effect of endotoxin upon luciferase expression could be detected (Cichoñ et al., manuscript in preparation).Even if endotoxin gains entry to the cell together with plasmid DNA, its subsequent effect upon the reporter gene expression, following transfer to the nucleus, is negligible.
The second important issue brought up in this report concerns the localization of the transferred transgene.Some data indicate that transgene inserted into tumor as naked DNA is localized into neoplastic cells [3].Our earlier studies of transgene transfer using cationic lipids also corroborate such localization [16].An important observation made during this study was that transgene transferred into a tumor as naked plasmid DNA, without carriers, is found in the proximity of the injection site not only in cancerous cells but also in normal cells forming tumor mass (or infiltrating it) (Fig. 2A, B).The presence of transgene in normal cells does not depend, however, on the type of carrier used: the transgene was found in normal cells also when DNA- Ten mg pCMVnlacZ plasmid DNA containing b-galactosidase reporter gene was injected into growing B16(F10) tumor using Ringers solution supplemented with 0.05% lactose (100 ml total).After 24 h the tumor was removed, rinsed in PBS `, fixed for 4 h at 4°C in 4% formaldehyde with sodium deoxycholate (final concentration = 0.01%) and NP-40 (final concentration = 0.02%) added.Then, the tumor was rinsed three times in PBS `and immersed in staining solution (X-gal) for 20 h (overnight) at room temperature.Following staining and rinsing (3´) with PBS `the sample was frozen in isopentane and liquid nitrogen.The specimen was stored at 70°C and sliced at 20°C using a cryostat.Slices obtained were then hematoxylin-eosin stained, fixed in ethyl alcohol and xylene and covered with a cover slip.Transfected cells (blue) are indicated with arrows.Fifty mg pCMVlacZ plasmid DNA containing b-galactosidase reporter gene was injected 3 times (at 24 h intervals) into growing B16(F10) tumor using Ringers solution supplemented with 0.05% lactose (100 ml total).For details see Fig. 2, part A. cationic lipids complexes were used for in vivo transfection.Transgene presence in normal cells resembling muscle cells (smooth muscle-like cells or myofibroblasts) [17][18][19][20] is indeed intriguing (Fig. 2B).According to some authors, myofibroblasts may play a role in the development of tumors [6,18].Others claim that they play a major role in the inflammatory response [20].Since the transgene is present in myofibroblasts it is possible that these cells contain more nucleic acids "receptors" than other cells forming the tumor.The presence of such receptors for oligonucleotides and nucleic acids has long been postulated for various types of normal cells; see Budker et al. [21].
The method of transferring plasmid DNA directly into cancer cells may be exploited in experimental tumor gene therapy [5][6][7].In our opinion, better suited to such therapy are genes coding for immunomodulatory proteins (cytokines) or angiostatic proteins (inhibiting neoplastic angiogenesis) compared to genes coding for suicide or proapoptotic proteins.The latter two kill cells directly and thus their activity should rather be confined to neoplastic cells only.Actually, for transgenes, whose protein products take part in indirect killing of neoplastic cells, their localization is not critical; they may localize themselves in both normal and neoplastic cells.We used successfully the method of direct plasmid DNA transfer into murine tumors in vivo to treat such tumors with the IL-12 gene [5] or the endostatin gene [7].
We are grateful to Dr. A. Sochanik for preparing cationic liposomes and critical review of the manuscript.We also thank M. Romanowska and M. Krawczyk for their technical assistance.

Figure 1 .
Figure 1.Enzymatic activity of luciferase in B16(F10) tumors transfected in vivo using lipoplexes and naked DNA.