Gene Ther Mol Biol Vol 10, 207-222, 2006

Effects of spatial configuration on tumor cells transgene expression

Research Article

Cecilia C. Casais, Armando L. Karara, Gerardo C. Glikin, and Liliana M. E. Finocchiaro*

Unidad de Transferencia Genética, Instituto de Oncología "Ángel H. Roffo", Universidad de Buenos Aires, Argentina

__________________________________________________________________________________

*Correspondence: Liliana M. E. Finocchiaro, Ph.D, Unidad de Transferencia Genética, Instituto de Oncología "A. H. Roffo" UBA, Av. San Martín 5481, 1417 Buenos Aires, Argentina; Telephone/FAX: 54 (11) 4580-2813; Email: gglikin@bg.fcen.uba.ar

Key words: multicellular tumor spheroids, persistent gene expression, non-viral vectors

Abbreviations: b-galactosidase, (bgal); analysis of variance, (ANOVA); cytomegalovirus immediate early promoter, (CMVie); LM05e spheroids, (LM05e/S); monolayers, (/M); simian virus 40 early promoter/enhancer, (SV40e); spheroids, (/S); three-dimensional, (3D)

Received: 10 April 2006; Revised: 26 June 2006

Accepted: 10 July 2006 electronically published: August 2006

Summary

We investigated the impact of the multicellular architecture on transgene expression of LM05e and LM3 spontaneous Balb/c-mammary adenocarcinoma and HEp-2 human laryngeal squamous carcinoma cell lines. When transferred from monolayers to spheroids, tumor cells strongly enhanced transient transgene expression, which surprisingly was still detectable 75 days after lipofection. The cytomegalovirus immediate early promoter (CMVie) yielded a very high b-galactosidase (bgal) transgene expression, which resulted 8-, 6- and 3-fold greater in LM05e, LM3 and HEp-2 spheroids than the corresponding monolayers. The SV40 early promoter displayed both, a lower spheroids/monolayers ratio and about 10% of bgal expression driven by CMVie. Cis-addition of Epstein Barr virus EBNA-1/oriP cassette enhanced the CMVie-driven transgene expression only in human HEp-2. Deletion of a 325 bp 5’ fragment of the CMVie promoter dropped spheroids bgal expression to 5%. This effect was restored to 10-25% or 25-60% by the insertion of one KCS (18 bp) or four myc-max consensus sequences (67 bp) respectively. When spheroids disassembled and grew as monolayers, bgal activity dropped accordingly. Our results demonstrated that the spatial configuration determined the expression enhancement and persistence in spheroids, being an event: fully reversible, proportional to spheroid compactness and independent of plasmid integration into the host genome.

I. Introduction

Multicellular spheroids are tissue-like structures of cells, with no artificial substrate for cell attachment (Mueller-Klieser, 1997). These cell aggregates organized in vitro have a great potential for a number of clinical and biomedical applications (Sutherland, 1998; Santini and Rainaldi, 1999). This three-dimensional (3D) cell system has been widely used as a model for microenvironmental effects on basic biological mechanisms, such as the regulation of proliferation, metabolism, differentiation, cell death, invasion, angiogenesis or immune response (Bates et al, 2000; Fehlauer et al, 2004). Compared to conventional monolayer cultures, 3D-cell aggregates resemble more closely the in vivo situation with regard to cell shape and cell environment, which in turn can affect gene expression and biological behavior of the cells. These 3D-structures offer a versatile in vitro system of intermediate complexity between monolayer cultures in vitro and tumors in vivo. In brief, spheroids combine the relevance of organized tissues with the controlled environment of in vitro methodology (Mueller-Klieser, 1997; Bates et al, 2000). Furthermore, they mirror the radius and chemosensitivity of differentiating tumors in vivo more closely than conventional cell cultures (Olive and Durand, 1994; Kolchinsky and Roninson 1997; Fehlauer et al, 2004). Being highly complex systems, their cellular properties are dependent on the origin of the tumor cells, their transformation state, and medium and growth conditions.

Non-viral vectors such as cationic lipids have important safety advantages over viral approaches, including their reduced immunogenicity, low cytotoxicity and minimal capacity for insertional mutagenesis (Glover et al, 2005). Although the efficacy of new cationic lipids formulations is comparable to adenovirus vectors, it takes many more copies of transgene to achieve a comparable expression. Despite the relative in vivo efficacy and variability frequently associated to these non-viral vectors, that varies greatly depending on the targeted tissue, many groups have demonstrated clinical efficacy using intra-tumor cationic lipid mediated gene transfer (Gottesman 2003; Yoshida et al, 2004; O’Malley et al, 2005).

We have developed 3D-cell cultures established from LM05e and LM3 spontaneous Balb/c murine mammary adenocarcinoma cell lines (Karara et al, 2002; Finocchiaro et al, 2004) and from HEp-2, a well-established human derived laryngeal squamous carcinoma tumor cell line, as models to investigate how the spatial configuration of cells affects the expression level of a transfected gene.

In this work we present evidence showing that transiently lipofected tumor cells, when transferred from 2D- to 3D-cultures, displayed higher and prolonged expression achieved by non-viral plasmid-based vectors. This enhancement was reverted when the spheroids were disassembled and reorganized as monolayers, and would occur independently of vector structure or integration into the host genome.

II. Materials and Methods

A. Cell cultures and growth

Cell lines derived from M05 (LM05e), M3 (LM3) and M38 (LM38) spontaneous Balb/c murine mammary adenocarcinomas; B16-F10 C57 murine melanoma (ATCC #CRL-76475), and HEp-2 (human laryngeal squamous carcinoma, ATCC #CCL-23) were cultured as monolayers and multicellular spheroids as described (Karara et al, 2002, Finocchiaro et al, 2004). The size of growing spheroids was estimated during a period of 75 days as the average of two diameters and the results were expressed as mean (of a minimum of 20 spheroid diameters) ± s.e.m. (n=4 independent assays).

B. DNA synthesis determinations

DNA synthesis was evaluated in cells seeded as spheroids in 96-well plates (5x10⁴ cells/well) by ³H-thymidine (New England Nuclear, Boston, MA; 1 Ci/mmol) incorporation as described (Finocchiaro et al, 2004). ³H-thymidine (0.3 mCi/well) was added to the cultures at 8, 15, 30, 45 and 60 days and incubation lasted for 72 hours. Cells were harvested and radioactivity was measured in a b-scintillation counter.

C. Plasmids

Plasmids pCMVb (MacGregor and Caskey, 1989) and pCH110 (Hall et al, 1983) are commercial (Clontech, Mountain View, CA), carrying the E. Coli lacZ gene under CMVie and SV40e promoters respectively.

An Eco RI fragment containing the human Epstein-Barr virus oriP and EBNA-1 gene (under its own promoter) from p205MTCAT (Yates et al, 1985) was cloned at the Eco RI site of pCMVb, yielding pEBCMVb.

A Sal I – Bst YI fragment containing the human Epstein-Barr oriP and EBNA-1 gene from pREP4 (Invitrogen, Carlsbad, CA) was cloned together with a Sal I - Bam HI fragment containing the CMVie promoter from pRc/CMV (Invitrogen) at the Sal I site of pCMVb, yielding pEB2CMVb. In this plasmid EBNA-1 is under the CMVie promoter.

We created a series of promoter constructs containing various lengths of the CMVie promoter upstream of bgal reporter gene. After deleting in CMVie the Eco RI – Nco I 5’- fragment (326 bp), oligodeoxynucleotides carrying (i) 4 copies of the myc-max consensus binding sequence (bold) (Sugaya et al, 1996): 5’-AATTCCCACCACGTGGTGCCTCCCACCACGTG GTGCCTCCCACCACGTGGTGCCTCCCACCACGTGGTGCCTC-3’ or (ii) one copy of the kinase consensus sequence (KCS, bold) (Kuhen et al, 1998): 5’-AATTCAGGGAAGG CGGAGTCCAAC-3’ were ligated to replace the removed fragment yielding pMYCCMVb and pKCSCMVb respectively. (iii) Fill-in and self-ligation of the Eco RI – NcoI sites yielded pD5´CMVb. On the other hand, the full-length CMVie promoter was deleted in pCMVb (between Eco RI and Sac I sites) and replaced by (iv) an oligodeoxynucleotide preserving the CMVie sequences TATA-BOX and Sp1-CS2, obtaining pTATAb. (v) By inserting in pTATAb the oligodeoxynucleotide with the 4 copies of the myc-max consensus sequence upstream of the TATA-BOX and Sp1-CS2 sequences, we obtained pMYCTATAb.

pCMVGM was obtained by replacing the lacZ gene in pCMVb by the hGM-CSF gene. A Not I - Not I fragment containing the lacZ gene was deleted from pCMVb and replaced by a suitable multiple cloning site, in which an Xho I - Hind III fragment containing the hGM-CSF gene was inserted. In a similar way, we replaced the lacZ gene in pCH110 (Kpn I - Bam HI fragment) by the hGM-CSF gene through an intermediate multiple cloning site, creating pSVGM.

Plasmids were amplified, grown and purified as described (Finocchiaro et al, 2004). Plasmid constructs used in this work are schematically depicted in Figure 1.

D. Liposome preparation and in vitro lipofection

DC-Chol (3b(N-(N',N'-dimethylaminoethane)-carbamoyl cholesterol) and DMRIE (1,2-dimyristyloxypropyl-3-dimethyl-hydroxyethilammonium bromide) were synthesized and kindly provided by BioSidus S.A. (Buenos Aires, Argentina). DOPE (1,2-dioleoyl-sn-glycero-3-phosphatidyl ethanolamine) was purchased from Sigma (St Louis, MO). Liposomes were prepared at lipid/co-lipid molar ratios of 1:1 (DMRIE:DOPE) or 3:2 (DC-Chol:DOPE) by sonication as described (Felgner et al, 1994; Gao and Huang, 1995). Optimal DNA:lipid ratios and lipid mixtures were determined for every cell line: LM05e and HEp-2 cells were transfected with a mixture of 3:1 DC-Chol:DOPE/ DMRIE:DOPE at 1:4 mg DNA/nmol lipid, and LM3 cells were transfected with an equimolar mixture of DC-Chol:DOPE/DMRIE:DOPE at 1:6 mg DNA/nmol lipid.

Lipoplexes (0.5 mg DNA/cm²) were prepared in 0.1 M Na₂HPO₄/NaH₂PO₄buffer (pH 7.3) and applied to cultured cells at a density of 3x10⁴cells/cm² (about 30% confluence) in a serum-free medium (OptiMEM, Gibco-BRL, Gaithersburg, MD). In co-lipofections, 0.25 mg DNA/cm² of each plasmid was used. After 6-8 hours, the lipofection mixture was removed and medium with serum was added. 12-18 hours later, lipofected cells were trypsinized and some of them were seeded on the top of solidified agar to form spheroids (2-3 x 10⁵cells/ml) while the remaining ones were kept in monolayer cultures on regular plates (2-3 x 10⁴cells/cm²). Cells were incubated in regular culture conditions. Twice a week, culture medium was totally (monolayers) or partially (spheroids) replaced.

For stable expression, cells were lipofected with hIL-2 or lacZ gene carried by pRc/CMV (Invitrogene) as described above. After 48 h cells were selected with medium containing 500-700 mg/ml geneticin (Gibco-BRL). Single clones were isolated and tested for their hIL-2 or bgal expression by ELISA or ONPG assays as described below.

E. b-Galactosidase assays

To measure gene transfer efficiency, lipofected cells were trypsinized, fixed in suspension, stained with 5-bromo-4-chloro-3-indolyl b-D-galactopyranoside (X-Gal, Sigma) by standard methods (Teifel and Friedl, 1995; Finocchiaro et al, 2004) and counted. The same fixation and staining procedure was performed onto spheroids in suspension for micrography (Finocchiaro et al, 2004).

For quantitative gene expression, trypsinized monolayers and untreated spheroids were collected, washed with PBS and divided in two fractions. One fraction was resuspended in hypotonic solution (10 mM HEPES, 10 mM KCl, 0.1 mM EDTA, 1 mM PMSF and 1 mM DTT) and sonicated for 5 seconds, and bgal activity was assayed with orthonitrophenyl 1-b-D-galactopyranoside (ONPG, Sigma) as described (Teifel and Friedl, 1995; Finocchiaro et al, 2004). The remaining fraction of each sample was resuspended in 0.1 N NaOH and total protein was measured as described (Bradford, 1976). Specific bgal activity was expressed as mU bgal/mg protein, as the mean ± s.e.m. of n independent assays measured by triplicate.

F. ELISA hGM-CSF assay

Human recombinant GM-CSF secreted to the culture medium was assayed by ELISA. Briefly, assays were performed in 96-well plates coated overnight at 4°C with 0.4 mg/well anti-hGM-CSF monoclonal antibody (R&D, Minneapolis, MN). Plates were subsequently blocked at room temperature with 2% BSA in PBS for 2 h. hGM-CSF samples and standards (purified recombinant hGM-CSF, R&D) were added and incubated overnight at 4°C. Then the samples were consecutively incubated with a biotinylated polyclonal anti-hGM-CSF antibody (20 ng/well) (R&D), streptavidin-peroxidase conjugate (Sigma) and a colorimetric substrate (OPD: o-phenylenediamine dihydrochloride, Gibco BRL). Absorbance was measured at 490 nm. Total protein was measured as described above. hGM-CSF levels were expressed as ng/mg protein/day, as the mean ± s.e.m. of n independent assays measured by triplicate.

G. Southern blot analysis

Cells were lipofected with pCMVb or pEBCMVb plasmids, cultured as spheroids over a 40-day period as described, and genomic (Maniatis et al, 1982) and episomal (Hirt, 1967) DNA was extracted by standard methods. Portions (8 to 15 mg) of DNA were digested withHind III, and fragments were electrophoresed on a 0.8% agarose geland subjected to standard Southern transfer onto positivelycharged nylon membranes (GeneScreen, New England Nuclear). Hybridization was performedwith a ³²P-radiolabeled Eco RV - Sac Ifragment (825-bp probe) of the lacZ gene contained in all bgal plasmids used.

H. Statistical analysis

Results were expressed as mean ± standard error of the mean (s.e.m.) (n: number of experiments corresponding to independent assays). Differences between groups were determined by analysis of variance (ANOVA).

III. Results and Discussion

A. Tumor cells grew in vitro as multicellular spheroids

LM3, LM05e and LM38 (murine mammary adenocarcinomas), B16 (murine melanoma), and HEp-2 (human laryngeal squamous carcinoma) cells readily formed spheroids when plated on the top of solidified agar. While LM05e and LM3 spheroid cells appeared intimately associated with each other and closely packed, HEp-2 formed more loosely associated cell aggregates in which single cells could be clearly distinguished (Figure 4). B16 initially formed lax aggregates, which became more compact beyond day 15 (Finocchiaro et al, 2004), and LM38 spheroids resulted similar to HEp-2 aggregates (data not shown).

Spheroids obtained from LM05e, LM3 and HEp-2 tumor cell lines revealed different growth potential (Figure 2). LM3 and HEp-2 aggregates showed extensive growth, increasing their diameter about 2.5-fold from day 4 to day 40, when they reached a plateau up to day 75. Conversely, compact LM05e spheres showed only a slight increase of 1.3-fold in diameter from day 4 to day 20, and then they reached a plateau up to day 75 (Figure 2a).

3D cell aggregates incorporate less ³H-thymidine than an equivalent amount of the corresponding monolayers (Finocchiaro et al, 2004). The rate of ³H-thymidine incorporation into DNA correlated with the diameter increase during the spheroid growing phase. Whereas LM05e spheroids (LM05e/S) displayed a very low ³H-thymidine incorporation rate over time, both LM3/S and HEp-2/S showed an initial higher rate at day 8 followed by a steady lower rate up to day 60 (Figure 2b). HEp-2/S doubled the LM3/S ³H-thymidine incorporation as total protein did, while both spheroids had similar diameters. Therefore, the higher ³H-thymidine incorporation by HEp-2/S should reflect a higher number of spheroids. LM05e/S showed low total amounts of protein, which correlates to their small size.

Total protein remained relatively constant over time in HEp-2/S and LM3/S, suggesting balanced growth and death rates. On the other hand, LM05e/S total protein decreased gradually over time, reaching 50% of the initial value at day 75. Considering that LM05e/S size did not decrease, this protein decay would be due to death of some small spheroids (Figure 2c).

B. Spheroids displayed enhanced and persistent transgene expression

In a previous study, we demonstrated that CMVie-driven transgene expression in LM05e, LM3 and B16 spheroids was considerably higher than in their respective monolayers (Finocchiaro et al, 2004). To address this issue in greater detail, we compared the temporal course of CMVie and simian virus 40 early promoter/enhancer (SV40e) driven bgal expression in cells grown as monolayers (/M) or spheroids (/S).

Before the development of long-lasting multicellular spheroid cultures generally, it was not possible to keep viable cells for more than two weeks in culture without active cell division, and transgene expression rapidly diluted over time. On the other hand, monolayers replating abolished most of transgene expression, which decreased between 10 to 100 times after two passages (data not shown). Therefore, we worked with monolayers that became mostly quiescent after reaching confluence, showing growth kinetics similar to spheroids: LM05e/M, LM3/M and HEp-2/M total protein increased 40, 50 and 70% respectively from day 8 to 15. A major advantage of spheroids is that they could be kept viable without replating for more than 75 days, while unreplated monolayer cultures started to detach and die beyond 15 days.

Although bgal activity in spheroids decreased over time, it is noteworthy that expression at day 75 was similar to monolayer expression at days 4-8 in all cell lines tested.

At day 4, bgal specific activity displayed by pCMVb resulted about 8-fold (LM05e), 6-fold (LM3) and 3-fold (HEp-2) greater in spheroids than the corresponding monolayers. In addition, pCMVb expression levels were longer standing in 3D- than 2D- cultures: at day 15, bgal activity was 109% (LM05e/S), 63% (LM3/S) and 117% (HEp-2/S) of that at day 4, while in monolayers, bgal activity relative to day 4 was 34, 21 and 45%, respectively. Taken together, these differences in levels and persistence of expression determined that, at day 15, bgal activity resulted 26-fold (LM05e), 14-fold (LM3) and 7-fold (HEp-2) greater in spheroids than in the corresponding monolayers. As it was the case in spheroids, monolayer maximal bgal activity in HEp-2 was lower than LM05e and LM3 (32 and 24% respectively).

The effect of spatial configuration resulted less dramatic when bgal was driven by SV40e promoter (pCH110), whose expression levels in spheroids were about 10% of pCMVb. Spheroid bgal expression was relatively constant over time in LM05e and HEp-2 cells, falling about 30% at day 15 in LM3.

In monolayers, differences between pCH110- and pCMVb- driven expression were smaller, with pCH110 displaying at day 4 after lipofection 26% of pCMVb activity in LM05e/M, 10% in LM3/M and 13% in HEp-2/M. SV40e-driven bgal activity was maximal in LM05e/M and LM3/M at day 4 followed by a 50% diminution at day 8 when a plateau was reached, while in HEp-2/M it remained constant from day 4 to day 15.

In both LM05e/S and LM3/S, SV40e-driven bgal activity was significantly higher, but in HEp-2/S was only slightly higher than their respective monolayers.

So, lower monolayers bgal expression with the two plasmids tested, was probably due to: (i) the decline in the percentage of transfected cells by transgene dilution during replication of the target population, and/or (ii) loss of the transgene by nuclease digestion or partitioning to non-nuclear compartments.

In general terms, cells growing as spheroids expressed significantly higher levels of bgal than the same cells in monolayers in all the assayed conditions, suggesting that 3D-configuration strongly enhanced transgene expression.

On the other hand, total spheroid bgal activity (mU) displayed a similar pattern to bgal specific activity (mU/mg protein). Maximal bgal total activity driven by CMVie promoter was comparable in LM05e/S and LM3/S (about 23 and 17 mU respectively) and much lower (about 6 mU) in HEp-2/S that displayed steady values from day 4 to 45 followed by a slow decay up to 50% on day 75. Nevertheless, total activity levels in the three assayed cell lines converged beyond day 45 (Figure 3d). It is worth to note that the relative values of maximum spheroid specific activity (mU/mg protein) among cell lines were maintained when expressed as total bgal activity (mU), demonstrating that they were not artificially produced by the differences in protein levels and that could be attributed to actual variations of transgene expression. Therefore, we might suppose that the high expression in LM05e/S is a consequence of their low growth rate, slow plasmid loss kinetics and/or to the availability of the transcription/translation cellular machinery in quiescent cells. However, LM3/S have a growth pattern similar to HEp-2/S, but LM3/S maximum expression levels are about 6-fold higher than HEp-2/S and only 40% lower than LM05e/S, suggesting that a high expression rate is not in direct correlation with slow growth kinetics. On the other hand, taking into account that LM05e/S and LM3/S are clearly more compact than HEp-2/S, it can be suggested that the high expression correlates with the degree of compactness. Indeed, B16 (Finocchiaro et al, 2004) and LM38 (data not shown) spheroids, which are initially poorly compacted, display low initial expression levels.

The effects of spatial configuration on bgal reporter gene expression were confirmed by X-Gal staining of bgal-lipofected cells (Figure 4). The amount of X-Gal stained cells, clustered in defined regions throughout the spheroid, increased from day 1 to 15 after lipofection, and then displayed a first fast diminution from day 15 to 30 followed by a slow decay from day 30 to 75.

C. The EBNA-1/oriP cassette increased the CMVie-driven bgal long-term expression in human cells

Since persistent gene expression is required for some applications of gene therapy, we assayed the effect of some persistence elements and factors. We constructed pEBCMVb, an Epstein-Barr virus (EBV)-based vector carrying the EBV latent origin of replication for episomal persistence, oriP (about 2200 bp) and a replication initiation factor, EBNA-1 (EBV-encoded nuclear antigen 1). By binding to the cis-acting viral DNA element oriP in the Epstein-Barr virus genome, EBNA-1 enables plasmids to persist as multicopy episomes that attach to chromosomes during mitosis and enhances transcription from these EBV episomes (Yates et al, 1985; Kaneda et al, 2000; Tu et al, 2000).

In HEp-2 human cells, when equipping the plasmid with this cassette (pEBCMVb), there was a significant expression increase both in monolayers and spheroids from day 4 to 15. In HEp-2/S, bgal activity increased about 2-fold from day 4 to 15 after lipofection; then it reached a steady state up to day 30 when it started a slow decrease up to day 75 (about 70%). In murine LM3/S and LM05e/S, the cis-addition of the EBNA-1/oriP sequences not only did not modify pCMVb bgal expression in LM3/S but resulted in about 32% diminution with respect to pCMVb in LM05e/S, probably because the expression of EBNA-1 gene was employing an important fraction of the spheroid cellular machinery involved in gene expression and/or because of larger plasmids lower lipofection efficiency (Figure 5d). After high initial levels from day 4 to 15, bgal activity promptly decreased (about 90%) between day 15 and 75 in LM50e and LM3 spheroids since mouse genomes do not possess elements that allow replication and further segregation of the

Figure 4. Distribution of long-term bgal expression in spheroids. Representative micrographs of X-Gal stained LM05e, LM3 and HEp-2 spheroids at 4; 8; 15; 30; 45 and 60 days post-lipofection with pCMVb. Cells were transfected in vitro with lipoplexes containing pCMVb, harvested 24 h later and seeded on coated plates as multicellular spheroids. At each time point, specimens were fixed in suspension and stained with X-Gal, as described in Materials and Methods. The dark spheroid areas indicate b-galactosidase activity.

Figure 5. Effect of EBNA1/oriP persistence elements on bgal expression. (a-c) Time course of specific bgal reporter activity following lipofection with pCMVb (˜,¡), pEBCMVb (¢,£), pEB2CMVb (p,r) or pCMVb+pCMVGM (pCMVb/2) (®,¯) plasmids in LM05e (a), LM3 (b) and HEp-2 (c) cells cultured as spheroids (main plot, black symbols) or monolayers (inserted plot, open symbols). At the indicated times, cells were homogenized and assayed for bgal activity as described in Materials and Methods. Results were expressed as mU of bgal activity/mg protein ± s.e.m. of (n) independent assays after correction for background (pCMVb: n=14; pEBCMVb: n=9; pEB2CMVb: n=8).

Showing the P-values obtained by ANOVA test

PLASMID \ CELLS	LM05e		LM3		HEp-2
pCMVb vs.	Spheroids	Monolayer	Spheroids	Monolayer	Spheroids	Monolayer
pEBCMVb	n.s.	n.s.	n.s	n.s.	p<0.05 (days 8-60)	p<0.05 (day 15)
pEB2CMVb	p<0.05 (days 4-30)	n.s.	p<0.05 (days 4-45)	n.s.	n.s	p<0.05 (day 4)
pCMVb/2	p<0.01 (days 4-45)	p<0.05 (days 4-8)	p<0.01 (days 4-45)	p<0.05 (days 4-8)	p<0.01	p<0.01 (days 4-8)
pCMVb/2 vs.
pEBCMVb	p<0.01 (days 4-15)	p<0.05 (days 4-8)	p<0.01 (days 4-15)	p<0.01 (days 4-8)	p<0.01	p<0.01
pEB2CMVb	p<0.01 (days 4-15)	p<0.05 (day 8)	p<0.05 (day 8)	p<0.05 (days 4-8)	p<0.01	n.s.

(d) Effect of EBNA1/oriP cassette on gene transfer efficiency: LM05e (gray bars), LM3 (white bars) and HEp-2 (light gray bars) cells transfected with pCMVb, pEBCMVb or pEB2CMVb lipoplexes were stained with X-Gal 48 h later and counted as described in Materials and Methods. The results were expressed as % of X-Gal blue staining cells ± s.e.m. of (n) independent experiments (pCMVb: n=16; pEBCMVb: n=9; pEB2CMVb: n=8). ⁺ p < 0.05 and ⁺⁺ p<0.01: with respect to pCMVb in the same cell line.^o p < 0.05 and ^oo p<0.01: with respect to LM05e/S respective plasmid.

replicated EBV oriP plasmids to daughter cells upon cell division (Yates et al, 1985; Tu et al, 2000). Despite the differences observed between pCMVb and pEBCMVb expression in spheroids at earlier times after lipofection, in the three cell lines values tended to converge on day 75.

On the other hand, in monolayers from day 4 to 15, pEBCMVb bgal activity decreased about 50% in LM05e and 70% in LM3 cells, while remained constant in HEp-2 cells.

As it was the case with pCMVb, pEBCMVb also displayed a remarkable increase of specific activity in spheroids with respect to monolayers: about 7-fold for LM05e, 5-fold for LM3 and 4-fold for HEp-2 at day 8. In an EBNA-1/oriP construct, the replacement of EBNA-1 promoter by the stronger CMVie promoter resulted in a 20-fold increase in EBNA-1 expression (Kaneda et al, 2000; Tu et al, 2000). So, to investigate if a higher amount of EBNA-1 could induce a greater enhancement of transgene expression, we constructed pEB2CMVb, a plasmid similar to pEBCMVb but with EBNA-1 under CMVie promoter. However, this construct resulted in less efficient bgal expression in the three cell lines (Figure 5a-c), suggesting that (i) the amount of this regulating protein driven by its own original promoter was already enough for maximal bgal activity driven by CMVie; (ii) an excessive amount of EBNA-1 bound to oriP might inhibit nuclear retention and/or migration of the plasmid, presumably because of the formation of large complexes that cannot pass through the nuclear pore (Kaneda et al, 2000), (iii) the presence of this second CMVie promoter, competing for the same factors and (iv) of this CMVie-driven gene competing for the transcription/translation machinery, had a significant inhibitory effect on bgal expression. A similar effect was observed with pCMVb bgal expression, when co-transfected with a second plasmid carrying the human granulocyte-macrophage colony stimulating factor (hGM-CSF) gene under CMVie promoter. As shown in Figure 5a-c, co-expression of hGM-CSF under CMVie promoter caused a dramatic inhibition of bgal activity (about 90% inhibition in LM05e/S and LM3/S (day 8), and 70% in HEp-2/S (day 15)). This exceeded the expected diminution in expression levels due to half amount of plasmid used in co-lipofection experiments. However, pEB2CMVb-driven bgal expression in spheroids and monolayers was higher than bgal expression in pCMVb+pCMVGM co-lipofection (Figure 5a-c). In spheroids, these differences were about 6-fold in LM05e, 2-fold in LM3 and 3-fold in HEp-2 at day 8, while beyond 45 days values tended to converge. In monolayers this effect was weaker: pEB2CMVb-driven bgal expression was about 2-fold (LM05e), 3-fold (LM3) and 1.5-fold (HEp-2) higher than bgal expression from pCMVb+pCMVGM at day 8.

On the other hand, in each cell line, lipofection efficiency measured as X-Gal stained cells at day 1 partially correlated with bgal specific activity measured by the ONPG method (Figure 5d). Despite the fact that pCMVb displayed the highest efficiencies, larger pEB2CMVb and pEBCMVb plasmids resulted about 55-60% of pCMVb. The relative strengths of the constructs in different cell lines were approximately the same, with LM05e being the most efficient for transgene expression followed by HEp-2 and LM3 (about 30 % of LM05e). It is worth to note that while LM3 and HEp-2 cells displayed similar lipofection efficiencies, the significantly higher total and specific bgal expression in LM3/S with respect to HEp-2/S would be related to the degree of spheroid compactness.

D. Persistent reporter activity was due to sustained transgene expression

To evaluate if bgal activity persistence was due to sustained transgene expression in addition to slow foreign protein turnover in the cytoplasm, we also analyzed the long-term expression of a secreting gene product such as hGM-CSF. By co-lipofection of pCMVb and pCMVGM, intracellular bgal expression was paralleled to extracellularly secreted cytokine produced by the hGM-CSF gene. As it occurred for bgal, the maximal hGM-CSF production in LM3 and LM05e spheroids appeared between days 4 and 15 with a fast decay up to day 30 followed by a slower decay up to day 75 (Figure 6a). Since 24 h hGM-CSF secretion after renewing the culture medium reflects the actual transgene expression rate, the equivalent kinetics of both transgenes in LM3 and LM05e spheroids confirmed that persistence was mainly due to continuous gene expression. But HEp-2 spheroids, whose expression levels were markedly lower than those of LM05e (about 10%), showed a maximal hGM-CSF production at day 4 followed by a continuous decay that dropped the expression to 5% of the initial level at day 40. When comparing the expression patterns of both transgenes, we can see that in HEp-2 cells hGM-CSF production dropped faster than bgal activity. Since the half-life of the b-galactosidase enzyme in some cell lines could reach several days (Klunder and Hulser, 1993), we can assume that persistence of HEp-2 bgal activity was partially due to its stability in cytoplasm. On the other hand, the continuous and long-term exposure of spheroid cells to high levels of secreted hGM-CSF could display unspecific mild toxic effect leading to down regulate its own expression or to hGM-CSF degradation. This result obtained with in vitro cultured HEp-2 spheroids strikingly paralleled in vivo G-CSF expression as measured in serum after i.v. injections of the G-CSF gene containing lipoplexes specially devised for long-term expression (Tu et al, 2000).

At day 8, monolayers displayed lower hGM-CSF production than spheroids in LM05e (about 3-fold) and in LM3 (about 20-fold), as occurred with bgal activity. Conversely, at day 4 the hGM-CSF production resulted equivalent in HEp-2 spheroids and monolayers. But in all cell lines monolayers production immediately dropped, while spheroid hGM-CSF production did it smoothly. This gave rise to greater differences between spheroids and monolayers at day 15: at this time, S/M production ratios were 118 for LM3, 20 for HEp-2 and 7 for LM05e.

As it was the case with the bgal gene, SV40e promoter drove a significantly lower hGM-CSF production than CMVie promoter with similar decay kinetics in all cell lines), and this production was lower in monolayers than in spheroids, except in LM05e cells, where both S and M displayed similar production levels (Figure 6d-f).

class=Section14>

Figure 6. Expression analysis of secreting human GM-CSF gene product. (a-c) Time course of specific b-galactosidase activity and hGM-CSF production after co-lipofection with pCMVb (circles) + pCMVGM (triangles) plasmids in LM05e, LM3 and HEp-2 cells cultured as spheroids (/S, black symbols) or monolayers (/M, open symbols). Data are expressed as a percentage over the b-gal activity or hGM-CSF production in spheroids at day 4. Each value represents mean ± s.e.m. of (n) independent assays (pCMVb: n=20; pCMVGM: n=9). Maximal b-gal activities (mU/mg protein): 64 (LM05e/S), 67 (LM3/S) 23 (HEp-2/S). Maximal hGM-CSF production (ng/10⁶cells/day): 1568 (LM05e/S), 1185 (LM3/S), 783 (HEp-2/S). pCMVb vs. pCMVGM: p<0.01 at days 15 to 75 in HEp-2/S. pCMVb vs. pCMVGM: p<0.01 at days 8 to 15 in HEp-2/M. (d-f) Time course of hGM-CSF specific production after co-lipofection with pCMVb + pCMVGM (triangles) or pCH110 + pSVGM (squares) plasmids in LM05e, LM3 and HEp-2 cells cultured as spheroids (/S, black symbols) or monolayers (/M, open symbols). Each value represents mean ± s.e.m. of (n) independent assays (pCMVGM: n=9; pSVGM: n=4). pCMVGM vs. pSVGM: p<0.01 in the 3 lines.

E. The full-length CMVie promoter mediated maximal transgene expression in spheroids

The control of transgene expression is a complex process, dependent in part on the availability and/or activity of cellular factors and proximal sequences necessary for promoter function. The full-length CMVie promoter mediated a very high spheroid transgene expression of plasmid DNA for prolonged periods. To characterize some properties of CMVie promoter (533 bp), we designed a series of constructs derived from pCMVb (Figure 1) containing various lengths of the CMVie promoter upstream of bgal reporter gene: (i) pΔ5'CMVb: a construct containing the 3’ region of CMVie promoter that goes from Nco I to Sac I sites (208 bp), where the 5’ region between EcoR I and NcoI sites (325 bp) was deleted. This deleted region was substituted by (ii) four tandem repeats containing the myc-max consensus binding sequence (Sugaya et al, 1996), yielding pMYCCMVb, or (iii) 1 copy of the KCS sequence (Kuhen et al, 1998) (which binds factors released in presence of b-IFN), yielding pKCSCMVb. On the other hand, (iv) the full-length CMVie promoter was deleted and replaced by a minimal promoter containing the 3´CMVie sequences TATA-BOX and Sp1-CS2, obtaining pTATAb; and then (v) four tandem repeats of myc-max consensus binding sequence were added upstream, yielding pMYCTATAb.

The reporter gene activity of all these constructs was evaluated in monolayers and spheroids over a 75-day period (Figure 7).

Deletion of a 325 bp Eco RI - Nco I fragment (pΔ5'CMVb) strongly dropped the expression of the reporter gene driven by CMVie promoter in the three cell lines, either cultured as spheroids (more than 95% inhibition) or monolayers (about 80-85% inhibition). The insertion of 4 myc-max consensus sequences (67 bp) partially restored the CMVie promoter strength: 25% in LM05e/S and 50-60% in LM3/S and HEp-2/S. Since myc-max levels arise with proliferation and apoptosis, the lower activity of this construct in LM05e could be due to the lower growth rate of these cells as spheroids. Conversely, in monolayers this restoration was nearly total at day 4 in LM05e and HEp-2. Probably these cells express higher levels of myc-max proteins while proliferating.

The insertion of only 18 bp of the KCS sequence restored about 10-25% (spheroids) and 25-60% (monolayers) of the CMVie promoter activity. This specific behavior would be due to different levels of regulatory factors binding to promoters in 2D- and 3D-cultured cells.

On the other hand, because of the lack of enough regulatory elements, pTATAb could support only 10% of the pCMVb expression even after the insertion of 4 myc-max sequences (pMYCTATAb).

Four important conclusions may be drawn from these data: (i) the composition of the expression cassette was a major determinant of the levels of transgene expression, but did not affect its time extent; (ii) the full-length CMVie promoter mediated the best transgene expression of plasmid DNA; (iii) transgene expression was dependent on the promoter and the number of regulating sequences; and (iv) spheroids always displayed higher transgene activity than the corresponding monolayers.

Here, we demonstrated that cells assembled as spheroids strongly enhanced transgene expression of all the tested plasmids, but perhaps the most surprising finding was that reporter expression was still detectable 75 days after lipofection. As far as we know, such in vitro persistent transgene expression from non-viral vectors has not been reported previously.

F. The effects of culture configuration on transgene expression were reversible

When transferred from non-adhesive to regular cell culture plates, spheroids tended to disassemble and grow as monolayers. The ability to form these monolayers was inversely correlated to the degree of compactness of spheroids: HEp-2 spheroids formed these monolayers more readily than LM05e or LM3, and this ability decreased in the three cell lines over the time, when spheroids became more compact.

Spheroids lipofected with pCMVb were transferred to regular plates at different times (4 to 37 days post-lipofection), and 7 days later, specific bgal activity was measured in both spheroids and the resulting monolayers (removing previously the remaining spheroids). As it can be seen in Figure 8, bgal activity in these monolayers dropped to similar values than control monolayers in all cell lines. At every time point, monolayers bgal activities were more than 90% lower than the parental spheroids from which they derived 7 days before, while if they continued as spheroids expression only dropped 5 to 50% in LM05e/S and HEp-2/S, and 15 to 75% in LM3/S from day 15 to 45. These results demonstrated that the expression enhancement tightly depends on spatial configuration and that it can be reversible. These findings were confirmed by microscopy (Figure 8, right panel). Eight days after lipofection spheroids were transferred to regular plates, and 2 to 4 days later, the remaining spheroids and the radially growing monolayers were X-Gal stained for bgal expression and photographed. As expected, intense staining can be seen in the remaining assembled spheroids, while monolayers showed few or no stained cells.

G. Long-term transgene expression occurred independently of plasmid integration into the host genome

Genomic and episomal DNA of spheroids at day 40 post-lipofection with pCMVb and pEBCMVbwere prepared and subjected to Southern blot analysis with a lacZ probe (as described in Materials and methods). The Southern transfer could not reveal any integrationof plasmid vectors into the host genome and episomal plasmid was detected 40 days post-lipofection demonstrating that most of these lipofected plasmids remained as episomes (Figure 9).

On the other hand, pCMVhIL2 transiently lipofected LM3 cells produced at day 8: 36.5 ± 4.5 or 332.1 ± 47.8 ng hIL-2/mg protein/day as monolayers or spheroids respectively (n=7). Conversely, pRc/CMVhIL2 stably transfected LM3 and LM38 monolayers, expressed at day 4: 1.0 ± 0.4 and 2.3 ± 0.7 ng hIL-2/mg protein/day respectively (n=4). When transferred from monolayers to spheroids, the same stably transfected cells produced undetectable hIL-2 levels (<0.1 ng mg/mg protein/day). This opposite effect of spatial configuration on integrated transgenes was confirmed by pRc/CMVb stably transfected LM3 cells. Whereas as monolayers bgal activity remained mostly constant (146±18 U/mg protein) from day 4 to 15 respectively, the same stably lipofected cells growing as spheroids presented similar levels from day 4 to 8 (133±19 U/mg protein), dropping to 42 % of the

class=Section16>

Showing the P-values obtained by ANOVA test

PLASMID/CELLS	LM05e		LM3		HEp-2
pCMVb vs.	Spheroids	Monolayer	Spheroids	Monolayer	Spheroids	Monolayer
pMYCCMVb	p<0.01	n.s.	p<0.05 (days 45-75)	n.s.	p<0.05 (days 8-15)	p<0.05 (days 8-15)
pKCSCMVb	p<0.01	n.s.	p<0.05	n.s.	p<0.05	p<0.05 (days 8-15)
pD5'CMVb, pTATAb pMYCTATAb	p<0.01	p<0.05	p<0.01	p<0.05	p<0.01	p<0.05
pD5'CMVb, pTATAb pMYCTATAb vs.
pMYCCMVb	p<0.01	p<0.05	p<0.01 (days 4-30)	p<0.05	p<0.01 (days 4-60)	p<0.05 (days 4-8)
pKCSCMVb	p<0.01	p<0.05	p<0.01 (days 4-15)	p<0.05	p<0.01 (days 4-60)	p<0.05 (days 4-8)

Figure 8. Effects of culture configuration reversion on transgene expression. Left panel: Specific b-galactosidase activity from LM05e, LM3 and HEp-2 spheroids (gray bars) and monolayers derived from the respective spheroids (white bars) at different times after lipofection with pCMVb. At each time point, the monolayers derived from disassembling spheroids seeded in regular culture plates 7 days before. Each value represents mean ± s.e.m. of 6 independent assays. Right panel: Representative micrographs of X-Gal stained LM05e, LM3 and HEp-2 disassembling spheroids and the radially growing monolayers at 11 days post-lipofection with pCMVb. (Spheroids were transferred to regular culture plates at day 8 post-lipofection). Dark spheroid areas indicate b-galactosidase activity.

Figure 9. Southern blot analysis of spheroid episomal DNA. Forty days post-lipofection with pCMVb or pEBCMVb; LM05e, LM3 and HEp-2 spheroids DNA was extracted, electrophoresed, blotted and hybridized as described in Materials and Methods. Cell lines and plasmids are indicated on the picture. M: Hind III digested plasmids as size markers.

monolayers activity on day 15 (S: 69±11 M: 165±11 U/mg protein; p<0.001, n=4). These results agree with those reporting a reduced portion of producing cells in stably transfected spheroids with respect to the same cells growing as monolayers (Klunder and Hulser, 1993).

All these data support the hypothesis that the high transgene expression in spheroids was driven by episomal plasmids, since in the case of any plasmid integration; its contribution to transgene expression would be negligible.

IV. Conclusion

The results presented in this paper suggest that monolayer cultures and 3D- spheroids represent two very different experimental tumor models. The most surprising finding was that tumor cells assembled as spheroids provide an approach for achieving strongly enhanced and persistent transgene expression. As far as we know, such in vitro persistent transient transgene expression from non-viral vectors has not been reported previously. All the plasmids so far tested showed an improved transgene expression in spheroids that correlated with their degree of compactness. Then, the major reason for enhanced expression of a heterologous transgene should be searched on specific cellular properties that appear to be optimized when growing in three-dimensional aggregates with respect to flattened monolayer cells as: (i) spherical cell and nuclear shape, (ii) the cellular environment, (iii) the DNA conformation and packing, (iv) the accessibility and composition of transcription factors, (v) the transcriptional/post-transcriptional activation, (vi) the increased protein synthesis, and (vii) cell cycle times that can affect gene expression and biological behavior.

An exciting property of spheroids was that the reporter gene expression was maintained during all the spheroid life span and seemed to occur independently of plasmid integration into the host genome. The significant differences in the activities driven by different constructs observed at day 8 converged to similar low values after 30-60 days of spheroids incubation, indicating that beyond the promoter used, the 3D-configuration is the main responsible for long-term gene expression. It is noteworthy that spheroids transgene expression at day 75 not only was detectable but it was similar to monolayer expression at day 8 in all cell lines tested. At least four processes seem to be critical for spheroid efficient and sustained expression of a heterologous transgene. First, the ability of spheroid cells to retain transfected DNA. Second, a low decline in the percentage of transfected cells by transgene dilution during replication of slowly proliferating spheroids. Third, a low loss of the transgene by nuclease destruction or partitioning to non-nuclear compartments. Fourth, a low attenuation of promoter function leading to silencing of transgene expression.

Two questions arise from our data: How significant would be the spatial configuration effect on transgene expression in vivo where 3D-assembled differentiated cells present low replication rates and can be metabolically active for very long times? Could non-integrative non-viral gene transfer be useful for particular gene therapy applications that need long-term transgene expression?

Although the search for new vectors (viral and non-viral) continues, cationic liposomes are among the most interesting vectors for cancer gene therapy because they are non-infective, have low immunogenicity, low toxicity and high stability, as well as low cost and ease of production (Yoshida et al, 2004; Glover et al, 2005). In addition, cationic lipids demonstrated to be sufficiently effective in some cancer gene therapy approaches to be used in veterinary (Dow et al, 1998; Finocchiaro et al, 2005) and human (Bergen et al, 2003; Yoshida et al, 2004; O’Malley et al, 2005) clinical trials.

The most positive message emerging from this article is that the 3D-configuration is the main responsible for long-term gene expression. Multicellular tumor spheroids, which mimic more closely in vivo solid tumors and micrometastases, are realistic experimental models to investigate many aspects of tumor biology (Mueller-Klieser, 2000; Finocchiaro et al, 2004). It is therefore plausible to speculate that non-viral plasmid transfer of in vivo tumors can achieve enhanced long-term transgene expression. This was confirmed by the fact that early passages cultured cell lines derived from five spontaneous canine melanomas formed spheroids that expressed pCMVb 3- to 6-fold more efficiently than their respective monolayers during the first 15 days after transient lipofection. Conversely, preliminary results suggest that the expression enhancement observed in tumor spheroids did not occur in the non-tumor monkey kidney VERO cell line (ATCC #CCL 81), that displayed similar levels of bgal activity in spheroids and monolayers (19.5 ± 3.3 and 25.3 ± 4.5 mU/mg protein, respectively, n=13), during the first 15 days following transient lipofection.

The biological and clinical significance of these observations remains to be determined. Therefore, the next step is to evaluate how broad this effect is in human non-tumor and tumor cells of various histologies. If enhanced long-term spheroids transgene expression is characteristic of tumor spheroids, the possibility of a targeted gene therapy where tumor cells express higher levels of the delivered gene than normal tissue is open. In addition, whether after gene transfer a low probability event of plasmid integration occurs, it would not significantly contribute to transgene expression. All these observations encourage the implementation of non-viral gene therapy strategies for the delivery of therapeutic genes to tumors where high-level and fairly long-lasting gene expression is required.

Acknowledgments

We thank Ana Bihary for technical assistance, Dr. Gabriel Fiszman for hIL-2 stably transfected LM3 and LM38 and Dr. Alejandro Urtreger for bgal stably transfected LM3. This work was partially supported by a grant from FONCYT: BID1201/OC-AR - PICT 2002 -12084, and a grant from BioSidus S.A. A.L.K., G.C.G. and L.M.E.F. are members, and C.C.C. is a fellow of the Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET, Argentina).

References

Bates RC, Edwards NS, Yates JD (2000) Spheroids and cell survival. Critical Rev Oncol Hematol 36, 61-74.

Bergen M, Chen R, Gonzalez R (2003) Efficacy and safety of HLA-B7/beta-2 microglobulin plasmid DNA/lipid complex (Allovectin-7) in patients with metastatic melanoma. Expert Opin Biol Ther 3, 377-384.

Bradford MM (1976) A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal Biochem 72, 248-254.

Dow SW, Elmslie RE, Willson AP, Roche L, Gorman C, Potter TA (1998) In vivo tumor transfection with superantigen plus cytokine genes induces tumor regression and prolongs survival in dogs with malignant melanoma. J Clin Invest 101, 2406–2414

Fehlauer F, Stalpers LJ, Panayiotides J, Kaaijk P, Gonzalez Gonzalez D, Leenstra S, van der Valk P, Sminia P (2004) Effect of single dose irradiation on human glioblastoma spheroids in vitro. Oncol Rep 11, 477-485.

Felgner JH, Kumar R, Sridhar CN, Wheeler CJ, Tsai YJ, Border R, Ramsey P, Martin M, Felgner PL (1994) Enhanced gene delivery and mechanism studies with a novel series of cationic lipid formulations. J Biol Chem 269, 2550-2561.

Finocchiaro LME, Bumaschny VF., Karara AL, Fiszman GL., Casais CC, Glikin GC (2004) Herpes simplex virus thymidine kinase/ganciclovir system in multicellular tumor spheroids. Cancer Gene Ther 11, 333-345.

Finocchiaro LME, Maminska ME, Castillo PJJ, Karara AL, Fiszman GL, Riveros MD, Glikin GC. (2005) Tumor vaccine combined with cytokines and suicide gene therapy for canine spontaneous melanoma. Mol Ter 11 (Suppl. 1), S270.

Gao X, Huang L (1995) Cationic liposome-mediated gene transfer. Gene Ther 2, 710-722.

Glover DJ, Lipps HJ, Jans DA (2005) Towards safe, non-viral therapeutic gene expression in humans. Nat Rev Gen 6,299-310.

Gottesman MM (2003) Cancer gene therapy: an akward adolescence. Cancer gene Ther 10, 501-508.

Hall CV, Jacob PE, Ringold GM, Lee F (1983) Expression and regulation of Escherichia coli lacZ gene fusions in mammalian cells. J Mol Appl Genet 2, 101-109.

Hirt B (1967) Selective extraction of polyoma DNA from infected mouse cell cultures. J Mol Biol 26, 365-369.

Kaneda Y, Saeki Y, Nakabayashi M, Zhou WZ, Kaneda MW, Morishita R (2000) Enhancement of transgene expression by cotransfection of OriP plasmid with EBNA-1 expression vector. Hum Gene Ther 11, 471-479.

Karara AL, Bumaschny VF, Fiszman GL, Casais CC, Glikin GC, Finocchiaro LME. (2001) Lipofection of early passages of cell cultures derived from murine adenocarcinomas: in vitro and ex vivo testing of the thymidine kinase/ganciclovir system. Cancer Gene Ther 8, 96-99.

Klunder I, Hulser DF (1993) Beta-galactosidase activity in transfected Ltk- cells is differentially regulated in monolayer and in spheroid cultures. Exp Cell Res 207, 155-162.

Kolchinsky A, Roninson IB (1997) Drug resistance conferred by MDR1 expression in spheroids formed by glioblastoma cell lines. Anticancer Res 17, 3321-3328.

Kuhen KL, Vessey JW, Samuel CE (1998) Mechanism of Interferon Action: Identification of Essential Positions within the Novel 15-Base-Pair KCS Element Required for Transcriptional Activation of the RNA-Dependent Protein Kinase pkr Gene. J Virol 72, 9934–9939.

MacGregor GR, Caskey T (1989) Construction of plasmids that express E coli b-galactosidase in mammalian cells. Nucleic Acids Res 17, 2365-2365.

Maniatis T, Fritsch EF, Sambrook J (1982) Molecular Cloning, a laboratory manual.; Cold Spring Harbor Laboratory, 280-281.

Mueller-Klieser W (1997) Three-dimensional cell cultures: from molecular mechanisms to clinical applications. Am J Physiol 273, C1109-C1123.

Mueller-Klieser W (2000) Tumor biology and experimental therapeutics. Critical Rev. Oncol Hematol 36, 123-139.

Olive PL, Durand RE (1994) Drug and radiation resistance in spheroids: cell contact and kinetics. Cancer Metastasis Rev 13, 121-138.

O’Malley BW Jr, Li D, McQuone SJ, Ralston R (2005) Combination Nonviral Interleukin-2 Gene Immunotherapy For Head and Neck Cancer: From Bench Top to Bedside. Laryngoscope 115,391-414.

Santini MT, Rainaldi G (1999) Three-dimensional spheroid model in tumor biology. Pathobiology 67, 48-157.

Sugaya S, Fujita K, Kikuchi A, Ueda H, Takakuwa K, Kodama S, Tanaka K (1996) Inhibition of tumor growth by direct intratumoral gene transfer of herpes simplex virus thymidine kinase gene with DNA-liposome complex. Hum Gene Ther 7, 223-230.

Sutherland R, Buchegger F, Schreyer M, Vacca A, Mach JP (1987) Penetration and binding of radiolabeled anti-carcinoembryonic antigen monoclonal antibodies and their antigen binding fragments in human colon multicellular tumor spheroids. Cancer Res 47, 1627-1633.

Sutherland RM (1998) Cell and environment interactions in tumor microregions: the multicell spheroid model. Science 240, 177-184.

Teifel M, Friedl P (1995) New lipid mixture for efficient lipid-mediated transfection of BHK cells. Biotechniques 19, 79-82.

Tu G, Kirchmaier AL, Liggitt D, Liu Y, Liu S, Yu WH, Heath TD, Thor A, Debs RJ (2000) Non-replicating Epstein-Barr virus-based plasmids extend gene expression and can improve gene therapy in vivo. J Biol Chem 39, 30408-30416.

Yates JL, Warren N, Sugden B (1985) Stable replication of plasmids derived from Epstein-Barr virus in various mammalian cells. Nature 313, 812-815.

Yoshida J, Mizuno M, Wakabayashi T (2004) Interferon-b gene therapy for cancer: Basic research to clinical application. Cancer Sci 95, 858-865.