Green fluorescent protein (GFP) is employed as a selection marker for gene transduction and to track tumor cells. Transduction of enhanced GFP (eGFP) into human neuroblastoma cell lines via a lentiviral vector significantly sensitized CHLA-20 (wild-type and functional TP53), and to a lesser extent CHLA-90 cells (multidrug-resistant, mutant, and nonfunctional TP53) to carboplatin, doxorubicin, etoposide, or melphalan, relative to cells transduced using the cell surface antigen CD80 as a selection marker. Total glutathione (GSH) was significantly up-regulated (1.8- to 2.8-fold) after eGFP (but not CD80) transduction in cell lines with, but not in those lacking, functional p53. Cytotoxicity of GSH depletion by buthionine sulfoximine in CHLA-20 (but not in CHLA-20-eGFP) was diminished by hypoxia (2% O2). Thus, oxidative stress produced by GFP selects for cells with up-regulated GSH in a p53-dependent manner, and also enhanced the cytotoxicity of anticancer drugs in neuroblastoma cell lines. Our data suggest caution when employing GFP-transduced cells to assess drug sensitivity and that using a cell surface antigen as a selection marker for gene transduction may perturb cells less than GFP.

The green fluorescent protein (GFP), cloned from Aequorea victoria, is widely used as a selection marker (reporter) for gene transfection (1) and as a tumor cell marker for in vivo studies of tumor biology and preclinical therapeutics (2–8). Cytotoxicity assays for assessing antineoplastic agents in vitro have also been developed using GFP-transduced cells (9–11). Transduction of GFP was initially reported to be nontoxic to cells (12). However, few studies have been performed to clarify the biological effects of GFP transduction. Human hepatocellular carcinoma cell lines have been reported to undergo apoptosis due to enhanced GFP (eGFP) transduction-mediated caspase-3 activation (13), suggesting that GFP could be toxic.

Reactive oxygen species (ROS) are involved in a variety of cellular processes, and can enhance the cytotoxic activity of various anticancer drugs (14, 15). Because ROS can be generated by GFP (16), we hypothesized that transduction of GFP could enhance anticancer drug sensitivity of human cancer cell lines. In this study, we showed that transduction of GFP using a lentiviral vector significantly sensitized neuroblastoma cell lines to anticancer drugs, and in a p53-dependent fashion, selected for cells with up-regulated glutathione, apparently as a means to counter the oxidative stress produced by GFP.

Cell Culture

Human neuroblastoma cell lines employed were CHLA-20 (17), CHLA-90 (17), CHLA-171 (18), and SMS-SAN (17). The CHLA-20 and SMS-SAN cell lines have wild-type, functional TP53, while the CHLA-90 and CHLA-171 cell lines have lost p53 function and show high-level multidrug resistance (19). Cells were grown in complete medium consisting of Iscove's modified Dulbecco's medium (IMDM, Bio Whittaker, Walkersville, MD) supplemented with 3 mml-glutamine (Gemini Bioproducts, Inc., Calabasas, CA), insulin and transferrin 5 μg/ml each and 5 ng/ml of selenous acid (ITS Culture Supplement, Collaborative Biomedical Products, Bedford, MA), and 20% fetal bovine serum at 37°C in a humidified 5% CO2 atmosphere. Cell lines were subcultured by detaching without trypsin from culture plates using a modified Puck's Solution A plus EDTA (Puck's EDTA) which contains 140 mm NaCl, 5 mm KCl, 5.5 mm glucose, 4 mm NaHCO3, 0.8 mm EDTA, 13 μm phenol red, and 9 mm HEPES buffer (pH 7.3) (20).

Reduced Oxygen Conditions

For hypoxia assays, cells were seeded into plates or flasks and placed into a sealed, humidified, modular incubation chamber that was flushed for 90 s at 10 p.s.i. with a mixture of 2% O2, 5% CO2, and 93% nitrogen (referred to as 2% O2 mixture) and then incubated at 37°C. Under these conditions, medium in plates and flasks attains a pO2 of approximately 15 mm Hg, which is below the degree of hypoxia found in bone marrow and in the range of hypoxia found in tumor tissue (21). Drug stock solutions were diluted in whole medium that had been allowed to equilibrate overnight in a loosely capped flask in a modular chamber flushed with the 2% O2 mixture as described above. After drug addition, plates and flasks were reflushed with the 2% O2 mixture for 90 s, and the chamber was sealed at approximately 5 p.s.i. over pressure to reduce atmospheric leaks, incubated at 37°C, and reflushed with the 2% O2 mixture every other day until assayed.

Gene Transduction

Transduction of cells was carried out using a self-inactivating lentivirus vector SMPU (Fig. 1A). The SMPU vector is a new self-inactivating lentivirus vector,1

1

M. Barcova, A. Logan, and P. M. Cannon. Engineering the polyadenylation sequence of a minimal self-inactivating lentiviral vector reduces read-through transcription and enhances gene expression from integrated vectors, manuscript in preparation.

which has a large deletion of the long terminal repeats (LTR), so that the LTR do not initiate transcription, and a large deletion of the human immunodeficiency virus 1 (HIV-1) gag region to decrease the potential for generation of replication-competent lentivirus. The SMPU vector has the HIV-1 central polypurine tract/central terminal site sequences to improve gene transfer, and the MND enhancer element (myeloproliferative sarcoma virus enhancer, negative control region deleted, d1587rev primer-binding site substituted) to drive the gene delivered (22); initially this was Clonetech eGFP. Additionally, the SMPU vector has three copies of the USE3 motif from SV40 virus to enhance the activity of the polyadenylation signal in the LTR. The lentivirus was produced by transient co-transfection of 293T cells with the expression vector plasmid SMPU, the packaging plasmid pCMV-ΔR8.9, and the envelope plasmid pMD.G as described previously (23). The virus supernatant was collected after 12–36 h of plasmid transfection, and virus titer was determined by assessing expression of fluorescence in transduced 293T cells by flow cytometry.

Figure 1.

A. The SMPU vector is a SIN HIV-1-based lentiviral vector that contains a large deletion (ΔU3) of the LTR by self-inactivation, so that the LTR do not initiate transcription. The HIV-1 gag region (GA) is largely depleted to decrease the potential for generation of replication-competent lentivirus by homologous recombination during packaging. The packaging signal (ψ), the central polypurine tract/central terminal site (cPPT/CTS) sequence, three copies of the USE3 motif of SV40 virus (the polyadenylation enhancer), the MND enhancer/promoter, the internal ribosome entry site (IRES), and the cDNA for GFP (or CD80) are indicated (not to scale). B. Neuroblastoma cell lines, CHLA-20 and CHLA-90, were lentivirally transduced with eGFP. Single eGFP-positive cells were sorted by flow cytometry into microplates, and the subclones, CHLA-20-eGFP and CHLA-90-eGFP, were established. CHLA-20 was also lentivirally transduced with hrGFP, YFP, or human CD80 (CD80). The clones were analyzed for green (525 nm) or yellow (575 nm) fluorescence by flow cytometry. R-phycoerythrin-conjugated (PE) anti-human CD80 antibody was used for detection of CD80. Open curves, parental cell lines; shaded curves, subclones with GFP or CD80. C. Baseline and ETOP-induced apoptosis (5 μg/ml of ETOP for 24 h) was compared in CHLA-20, CHLA-20-eGFP, and CHLA-20-CD80. The % cells ± SD with sub-G0/G1 DNA contents are shown in each figure. The rate of apoptotic cells with a sub-G0/G1 DNA content was 48.9 ± 6.2% in CHLA-20-eGFP, which was significantly higher than 34.0 ± 4.7% of CHLA-20 (P < 0.001). The CHLA-20-CD80 clone showed a modest, but significant increase in ETOP-induced apoptosis (41.7 ± 5.9%, P = 0.031).

Figure 1.

A. The SMPU vector is a SIN HIV-1-based lentiviral vector that contains a large deletion (ΔU3) of the LTR by self-inactivation, so that the LTR do not initiate transcription. The HIV-1 gag region (GA) is largely depleted to decrease the potential for generation of replication-competent lentivirus by homologous recombination during packaging. The packaging signal (ψ), the central polypurine tract/central terminal site (cPPT/CTS) sequence, three copies of the USE3 motif of SV40 virus (the polyadenylation enhancer), the MND enhancer/promoter, the internal ribosome entry site (IRES), and the cDNA for GFP (or CD80) are indicated (not to scale). B. Neuroblastoma cell lines, CHLA-20 and CHLA-90, were lentivirally transduced with eGFP. Single eGFP-positive cells were sorted by flow cytometry into microplates, and the subclones, CHLA-20-eGFP and CHLA-90-eGFP, were established. CHLA-20 was also lentivirally transduced with hrGFP, YFP, or human CD80 (CD80). The clones were analyzed for green (525 nm) or yellow (575 nm) fluorescence by flow cytometry. R-phycoerythrin-conjugated (PE) anti-human CD80 antibody was used for detection of CD80. Open curves, parental cell lines; shaded curves, subclones with GFP or CD80. C. Baseline and ETOP-induced apoptosis (5 μg/ml of ETOP for 24 h) was compared in CHLA-20, CHLA-20-eGFP, and CHLA-20-CD80. The % cells ± SD with sub-G0/G1 DNA contents are shown in each figure. The rate of apoptotic cells with a sub-G0/G1 DNA content was 48.9 ± 6.2% in CHLA-20-eGFP, which was significantly higher than 34.0 ± 4.7% of CHLA-20 (P < 0.001). The CHLA-20-CD80 clone showed a modest, but significant increase in ETOP-induced apoptosis (41.7 ± 5.9%, P = 0.031).

Close modal

Cells to be transduced were seeded into six-well plates at a concentration of 1.0 × 105 cells/well a day before transduction. Virus supernatant was added into each well at 1.0 × 107 infective units/ml of virus titer, and after 4 h incubation at 37°C in 5% CO2, the medium was changed to fresh complete culture medium. Cells were then analyzed for green-fluorescence expression using a Coulter Elite flow cytometer (Beckman Coulter, Miami, FL) equipped with an argon laser at 488 nm and 525 nm ± 10 nm band pass filter. Highly fluorescent cells were sorted into 96-well plates (one cell/well) using the “autocloning” mode of the “EXPO2” software (Beckman Coulter) and representative clones were selected from each transduction for further study. CHLA-20 was also transduced with the Stratagene human recombinant green fluorescent protein (hrGFP) (24) or the Clonetech yellow fluorescent protein (YFP) (9) cDNA using the SMPU vector, sorted, and cloned.

As a control for the lentivirus vector, eGFP cDNA was released from the SMPU vector plasmid, and the human CD80 antigen cDNA (23) was cloned into SMPU as an alternative selection marker (SMPU-CD80). CHLA-20 was transduced with SMPU-CD80, and exogenous CD80 expression was detected with PE conjugated antihuman CD80 antibody (BD PharMingen, San Diego, CA). The cloning of the CD80-positive cells was carried out in 96-well plates by flow cytometry as described for eGFP transduction.

Cytotoxicity Assay

Cytotoxicity assays for carboplatin (CBDCA), doxorubicin (DOX), etoposide (ETOP), melphalan (l-PAM), and buthionine sulfoximine (BSO) were performed by a fluorescence microplate cytotoxicity digital image microscopy assay (DIMSCAN), as previously described (17, 25). Briefly, cells were seeded into 96-well plates at 2000 cells/well in 100 μl of complete medium. After 24 h preincubation, 100 μl of complete medium with each drug at the indicated concentration were added. Each condition was tested in replicates of 12 microwells. After incubation of cells with DOX or ETOP for 4 days, and with CBDCA, l-PAM, or BSO for 7 days, 50 μl of fluorescein diacetate (FDA) solution were added to wells at 8 μg/ml (17). Plates were incubated at 37°C for 30 min, and then 30 μl/well of 0.5% eosin Y was added to quench background fluorescence in nonviable cells. Total fluorescence (which correlated with the number of viable cells) was measured using a custom-designed semiautomated digital imaging microscopy system. Digital thresholding was used to eliminate low-level background fluorescence in medium or in eGFP-transduced cells (25). Due to the quenching of green fluorescence in nonviable cells by eosin Y (17), and the low amount of fluorescence of GFP-transduced cells relative to viable cells stained with FDA, eGFP transduction did not affect the DIMSCAN assay. The results were expressed as surviving fractions of treated cells compared with control cells. Statistical comparison of the dose-response curves was done by the ANOVA test and/or a comparison of relevant survival fractions by two-tailed Student's t test.

Apoptosis

To measure the rate of drug-induced apoptosis, cells (replicates of six) were stained with propidium iodide in a hypotonic lysis buffer and the portion of cells with a sub-G0/G1 DNA content was measured by a flow cytometer as previously described (21).

Glutathione Levels

Cells were mechanically harvested with Puck's EDTA, washed in PBS, and centrifuged. The pellet was acidified with 200 μl of 5% sulfasalicylic acid, flash frozen in liquid nitrogen, and analyzed for total glutathione (GSH) content within 48 h by the DTNB-GSSB reductase method (26, 27), using a THERMOmax microplate reader (Molecular Devices, Sunnyvale, CA), temperature controlled at 30°C, with results normalized to total protein. The experiments were done in triplicate.

Human neuroblastoma cell lines, CHLA-20 (a moderately drug-resistant cell line with functional p53) and CHLA-90 (a highly drug-resistant cell line with mutated and nonfunctional TP53), were transduced with eGFP via the SMPU vector (Fig. 1A). Fluorescence-positive cells of each cell line were sorted by flow cytometry, one cell per well, into 96-well plates to generate the transduced clones, CHLA-20-eGFP or CHLA-90-eGFP (Fig. 1B). Clones of CHLA-20 transduced with hrGFP or YFP were also established. As a control for lentivirus vector-mediated gene transduction, CHLA-20 was transduced with human CD80 cDNA using the identical lentivirus vector backbone. High-level surface CD80 expression was detected on transduced neuroblastoma cells with PE-conjugated antihuman CD80 antibody, clearly distinguishing them from nontransduced cells, and CHLA-20-CD80 was cloned by flow cytometry into 96-well plates (Fig. 1B).

Apoptosis in eGFP-transduced CHLA-20 cells induced by ETOP was measured by flow cytometry and propidium iodide staining of nuclei (Fig. 1C). Transduction of eGFP has been reported to induce apoptosis in mouse fibroblast or human hepatocellular carcinoma cell lines (13). However, CHLA-20-eGFP did not have a significant increase in baseline apoptosis (5.2%, P = 0.16 compared to 3.1% of parental cells). After incubation with 5 μg/ml of ETOP for 24 h, the fraction of cells with sub-G0/G1 DNA content (apoptosis) was 48.9 ± 6.2% in CHLA-20-eGFP, which was significantly higher than 34.0 ± 4.7% in parental CHLA-20 (P < 0.001). ETOP-induced apoptosis was also slightly enhanced by CD80 transduction (41.7 ± 5.9%, P = 0.03), consistent with the results obtained using the DIMSCAN cytotoxicity assay (Fig. 2).

Figure 2.

Cytotoxicity dose-response curves were determined for cells treated with DOX (at 0, 15, 30, 60, 120 ng/ml) or ETOP (at 0, 1.25, 2.5, 5, 10 μg/ml) for 4 days, CBDCA (at 0, 1.5, 3, 6, 12 μg/ml) or l-PAM (at 0, 1.25, 2.5, 5, 10 μg/ml) for 7 days. Cytotoxicity was analyzed by a fluorescence microplate cytotoxicity assay (DIMSCAN assay, see “Materials and Methods”). The fluorescence intensity of nontreated control was set to 100, and the fractional survival of cells at each drug concentration was expressed as the ratio of the fluorescence intensity. The fluorescence intensity of GFP was low compared with the fluorescence of viable cells stained with fluorescein diacetate, and did not affect the DIMSCAN assay. Relative to parental CHLA-20, transduction of eGFP significantly sensitized CHLA-20 to CBDCA, ETOP, l-PAM across all concentrations tested (P < 0.001), and DOX (P < 0.001 only at 30–120 ng/ml). CHLA-20-CD80 showed slightly higher sensitivity to CBDCA at 3–12 μg/ml (P = 0.043), ETOP at 10 μg/ml (P = 0.011), and l-PAM at all tested concentrations (P = 0.01), but not to DOX (P = 0.13). Relative to CHLA-20-CD80, CHLA-20-eGFP showed significantly higher sensitivity to CBDCA, ETOP, and DOX across the full range of tested concentrations (P < 0.01), but for l-PAM, the difference was only significant at 2.5 μg/ml (P < 0.025). CHLA-20 was also significantly sensitized by YFP and hrGFP to all drugs tested across all range of tested concentrations (P < 0.001). CHLA-90 was moderately sensitized by eGFP transduction to CBDCA at 3–12 μg/ml (P < 0.001), DOX at 15–60 ng/ml (P = 0.002), and ETOP at all concentrations (P = 0.037).

Figure 2.

Cytotoxicity dose-response curves were determined for cells treated with DOX (at 0, 15, 30, 60, 120 ng/ml) or ETOP (at 0, 1.25, 2.5, 5, 10 μg/ml) for 4 days, CBDCA (at 0, 1.5, 3, 6, 12 μg/ml) or l-PAM (at 0, 1.25, 2.5, 5, 10 μg/ml) for 7 days. Cytotoxicity was analyzed by a fluorescence microplate cytotoxicity assay (DIMSCAN assay, see “Materials and Methods”). The fluorescence intensity of nontreated control was set to 100, and the fractional survival of cells at each drug concentration was expressed as the ratio of the fluorescence intensity. The fluorescence intensity of GFP was low compared with the fluorescence of viable cells stained with fluorescein diacetate, and did not affect the DIMSCAN assay. Relative to parental CHLA-20, transduction of eGFP significantly sensitized CHLA-20 to CBDCA, ETOP, l-PAM across all concentrations tested (P < 0.001), and DOX (P < 0.001 only at 30–120 ng/ml). CHLA-20-CD80 showed slightly higher sensitivity to CBDCA at 3–12 μg/ml (P = 0.043), ETOP at 10 μg/ml (P = 0.011), and l-PAM at all tested concentrations (P = 0.01), but not to DOX (P = 0.13). Relative to CHLA-20-CD80, CHLA-20-eGFP showed significantly higher sensitivity to CBDCA, ETOP, and DOX across the full range of tested concentrations (P < 0.01), but for l-PAM, the difference was only significant at 2.5 μg/ml (P < 0.025). CHLA-20 was also significantly sensitized by YFP and hrGFP to all drugs tested across all range of tested concentrations (P < 0.001). CHLA-90 was moderately sensitized by eGFP transduction to CBDCA at 3–12 μg/ml (P < 0.001), DOX at 15–60 ng/ml (P = 0.002), and ETOP at all concentrations (P = 0.037).

Close modal

To determine if GFP affected the sensitivity of cells to anticancer drugs, cytotoxicity dose-response curves were generated for CBDCA, DOX, ETOP, and l-PAM using a 96-well fluorescence plate reader (DIMSCAN) which measures the number of viable cells stained with FDA. The intensity of GFP fluorescence was substantially lower than the fluorescence of FDA in viable cells, enabling the elimination of background GFP fluorescence by digital thresholding. Fluorescence of nonviable GFP cells was quenched by eosin Y as used in the DIMSCAN assay. Thus, GFP fluorescence did not interfere with measuring cytotoxicity using DIMSCAN.

Compared to the parental CHLA-20 cells, CHLA-20-eGFP showed significantly higher sensitivity to CBDCA, ETOP, and l-PAM across the full range of tested concentrations (P < 0.001 for all drugs, Fig. 2A). A comparison of CHLA-20-eGFP to parental CHLA-20 showed that logs of cell kill were 4.5 versus 2.3 for CBDCA (12 μg/ml for 7 days), 3.6 versus 2.2 for ETOP (10 μg/ml for 7 days), or 4.1 versus 3.2 for l-PAM (10 μg/ml for 7 days). Higher drug concentrations of DOX (30–120 ng/ml) were significantly more toxic in CHLA-20-eGFP than in CHLA-20 (P < 0.001), causing 3.5 logs of cell kill at 120 ng/ml with eGFP versus 2.4 without eGFP, but no significant difference occurred at 15 ng/ml (P = 0.32). Also shown in Fig. 2A, CHLA-20-CD80 showed a modest increase in drug sensitivity to CBDCA only at 3–12 μg/ml (P < 0.001), ETOP only at 10 μg/ml (P = 0.01), and l-PAM at all tested concentrations (P = 0.01). Cytotoxicity of DOX was not altered by CD80 transduction (P = 0.13). These data suggest that exogenous gene integration by the lentiviral vector, CD80 expression, random selection of a clone from the heterogeneous CHLA-20 cell line, or all three, can cause a modest alteration in drug sensitivity. However, relative to CHLA-20-CD80, CHLA-20-eGFP showed significantly higher sensitivity to CBDCA, ETOP, and DOX across the full range of tested concentrations (P < 0.01), but for l-PAM, the difference was only significant at 2.5 μg/ml (P < 0.025). Thus, while some increase in drug sensitivity occurs during the transduction process with either CD80 or eGFP as the selection marker, eGFP significantly increased drug sensitivity relative to transduction with CD80.

Transduction of CHLA-20 with hrGFP and YFP also markedly increased the sensitivity to all drugs tested across all concentrations when compared to parental CHLA-20 (P < 0.001, Fig. 2B), and to DOX, ETOP, and CBDCA across all concentrations when compared to CHLA-20-CD80 (P < 0.01, Fig. 2B). Increase in sensitivity to l-PAM by hrGFP or YFP were both concentration dependent, with the hrGFP- and YFP-transduced cells significantly more sensitive relative to both parental cells or CHLA-20-CD80 at 1.25 and 2.5 μg/ml (P < 0.04).

The transduction of eGFP caused a modest, but significant increase in sensitivity (relative to parental cells) of the multidrug-resistant, TP53-mutated cell line CHLA-90 to CBDCA at 3–12 μg/ml (P < 0.001), DOX at 15–60 ng/ml (P = 0.03), and ETOP at all tested concentrations (P = 0.04, Fig. 2C). However, the degree that drug sensitivity was enhanced by eGFP transduction was considerably less in CHLA-90 compared to CHLA-20 for DOX and ETOP, and cyototoxicity of l-PAM was not altered by eGFP transduction in CHLA-90 (P = 0.27), suggesting that the drug-resistant phenotype of CHLA-90 helped overcome the effects of eGFP.

We measured GSH, a major cellular antioxidant that is necessary for neuroblastoma cell survival in vitro (26), before and after eGFP transduction (Fig. 3A). CHLA-20-eGFP expressed 1288 ± 85 nmol/mg protein of GSH, which was more than two times higher than the 535 ± 23 nmol/mg protein seen in parental CHLA-20 (P < 0.001), and GSH levels in CHLA-20-eGFP were also significantly (P < 0.01) higher than the 459 ± 10 nmol/mg protein observed in CHLA-20-CD80. The level of GSH in CHLA-20-CD80 was moderately lower (P=0.006) in comparison with parental CHLA-20. CHLA-90 has nonfunctional p53 due to a TP53 mutation. In contrast to CHLA-20 (functional p53), the GSH level of the CHLA-90-eGFP (502.4 ± 56.6 nmol/mg protein) was not significantly different from that of parental CHLA-90 (571 ± 27.7 nmol/mg protein, P = 0.11).

Figure 3.

A, GSH was up-regulated after eGFP transduction in neuroblastoma cell lines with functional p53 (CHLA-20 and SMS-SAN). CHLA-20-eGFP expressed 1288 ± 85 nmol/mg protein of GSH which was significantly higher than the 535 ± 23 nmol/mg protein seen in parental CHLA-20 (P < 0.001) and significantly higher (P = 0.006) than the GSH level observed in CHLA-20-CD80 (459 ± 10 nmol/mg protein). GSH levels in CHLA-20-eGFP were also significantly (P < 0.01) higher the 459 ± 10 nmol/mg protein observed in CHLA-20-CD80. The GSH level in SMS-SAN before versus after eGFP transduction was 435 ± 18 versus 732 ± 38 nmol/mg protein (P < 0.001). By contrast, the GSH level was not altered by eGFP in those cell lines with nonfunctional p53 (CHLA-90 and CHLA-171). The GSH level before versus after eGFP transduction was 571 ± 27.7 versus 502.4 ± 56.6 nmol/mg protein in CHLA-90 (P = 0.11), and 539 ± 15 versus 488 ± 54 nmol/mg protein in CHLA-171 (P = 0.19). B, to determine the effect of glutathione depletion by BSO, cytotoxicity of BSO was measured by DIMSCAN assay both under standard (20% O2) or hypoxic (2% O2) culture conditions. Cells were treated with BSO (0–1000 μm) for 7 days. BSO was toxic to CHLA-20, CHLA-20-CD80, and CHLA-20-eGFP at 20% O2, but in 2% O2 only CHLA-20-eGFP showed significant cytotoxicity in response to BSO (P < 0.001 for CHLA-20-eGFP versus CHLA-20 or CHLA-20-CD80). ○, parental CHLA-20; ▪, CHLA-20-Egfp; ▴ CHLA-20-CD80 (in B and C). C, the ETOP sensitivity of CHLA-20-eGFP was antagonized by the antioxidant, N-acetyl-l-cysteine (NAC). Cells were preincubated with 250 μm of NAC for 3 h, then treated with ETOP. NAC antagonized the cytotoxicity of ETOP in CHLA-20-eGFP, but not in parental CHLA-20 or CHLA-20-CD80 (P = 0.007, 0.065, or 0.096, respectively, compared to the dose-response curve for each to ETOP without NAC).

Figure 3.

A, GSH was up-regulated after eGFP transduction in neuroblastoma cell lines with functional p53 (CHLA-20 and SMS-SAN). CHLA-20-eGFP expressed 1288 ± 85 nmol/mg protein of GSH which was significantly higher than the 535 ± 23 nmol/mg protein seen in parental CHLA-20 (P < 0.001) and significantly higher (P = 0.006) than the GSH level observed in CHLA-20-CD80 (459 ± 10 nmol/mg protein). GSH levels in CHLA-20-eGFP were also significantly (P < 0.01) higher the 459 ± 10 nmol/mg protein observed in CHLA-20-CD80. The GSH level in SMS-SAN before versus after eGFP transduction was 435 ± 18 versus 732 ± 38 nmol/mg protein (P < 0.001). By contrast, the GSH level was not altered by eGFP in those cell lines with nonfunctional p53 (CHLA-90 and CHLA-171). The GSH level before versus after eGFP transduction was 571 ± 27.7 versus 502.4 ± 56.6 nmol/mg protein in CHLA-90 (P = 0.11), and 539 ± 15 versus 488 ± 54 nmol/mg protein in CHLA-171 (P = 0.19). B, to determine the effect of glutathione depletion by BSO, cytotoxicity of BSO was measured by DIMSCAN assay both under standard (20% O2) or hypoxic (2% O2) culture conditions. Cells were treated with BSO (0–1000 μm) for 7 days. BSO was toxic to CHLA-20, CHLA-20-CD80, and CHLA-20-eGFP at 20% O2, but in 2% O2 only CHLA-20-eGFP showed significant cytotoxicity in response to BSO (P < 0.001 for CHLA-20-eGFP versus CHLA-20 or CHLA-20-CD80). ○, parental CHLA-20; ▪, CHLA-20-Egfp; ▴ CHLA-20-CD80 (in B and C). C, the ETOP sensitivity of CHLA-20-eGFP was antagonized by the antioxidant, N-acetyl-l-cysteine (NAC). Cells were preincubated with 250 μm of NAC for 3 h, then treated with ETOP. NAC antagonized the cytotoxicity of ETOP in CHLA-20-eGFP, but not in parental CHLA-20 or CHLA-20-CD80 (P = 0.007, 0.065, or 0.096, respectively, compared to the dose-response curve for each to ETOP without NAC).

Close modal

To provide further insight into the role of p53 function in mediating the effects of GFP, we transduced eGFP into two other neuroblastoma cell lines, a p53 functional cell line SMS-SAN (sensitive to all drugs), and a p53 nonfunctional, multidrug-resistant cell line CHLA-171 (19), and measured GSH levels. As expected, SMS-SAN-eGFP had a significantly higher GSH level than parental SMS-SAN (732 ± 38 nmol/mg protein with eGFP versus 435 ± 18 nmol/mg protein without eGFP, P < 0.001), while no significant difference occurred in GSH levels between CHLA-171-eGFP and parental CHLA-171 (488 ± 54 nmol/mg protein with eGFP versus 539 ± 15 nmol/mg protein without eGFP, P = 0.19). GSH levels of parental and transduced clones are shown in Fig. 3A.

We have shown in human neuroblastoma cell lines that increased ROS after GSH depletion by BSO is cytotoxic to neuroblastoma cell lines (26) in standard culture conditions (atmospheric oxygen), but that physiological hypoxia significantly reduces the cytotoxicity of BSO as a single agent (28). To provide further evidence that GSH modulates increased oxidative stress in the eGFP-transduced cells, BSO cytotoxicity was measured by DIMSCAN assay both in standard culture conditions (20% O2) and in hypoxic conditions (2% O2) (Fig. 3B). In 20% O2, BSO was cytotoxic to parental CHLA-20, CHLA-20-CD80, and to CHLA-20-eGFP. However, hypoxia (2% O2) markedly reduced BSO toxicity to CHLA-20 and CHLA-20-CD80, while CHLA-20-eGFP cells remained sensitive to BSO in hypoxia across all tested concentrations (P < 0.001 compared with parental CHLA-20 or CHLA-20-CD80).

To further assess the role of oxidative stress in the enhanced ETOP sensitivity caused by eGFP transduction, CHLA-20 cells were preincubated for 3 h with the thiol antioxidant, NAC (29), then treated with ETOP (Fig. 3C). NAC (250 μm) significantly antagonized ETOP cytotoxicity in CHLA-20-eGFP (P = 0.007), but NAC did not significantly effect ETOP cytotoxicity in parental CHLA-20 (P = 0.065) or CHLA-20-CD80 (P = 0.096), compared with each dose-response curve for ETOP without NAC.

Because these results pointed to oxidative stress generated by eGFP as a major cause for the increased sensitivity to cytotoxic drugs, cytotoxicity of ETOP or l-PAM was tested in hypoxic (2% O2) culture conditions (Fig. 4). Compared with normal culture conditions (20% O2), hypoxia (2% O2) significantly reduced the toxicity of ETOP for CHLA-20-eGFP (P = 0.004). However, neither hypoxia (Fig. 4) nor NAC (Fig. 3) significantly altered the response to ETOP in parental CHLA-20 or CHLA-20-CD80 (P > 0.159). In contrast to ETOP, the cytotoxic response to l-PAM (2.5–10 μg/ml) of parental CHLA-20 (P < 0.001) and CHLA-20-CD80 (P = 0.041) was significantly decreased in 2% O2 relative to 20% O2, while l-PAM cytotoxicity was not significantly decreased by hypoxia in CHLA-20-eGFP (P = 0.505). Thus, ETOP cytotoxicity is not decreased by hypoxia and oxidative stress contributed by eGFP appears to be sufficient to enhance ETOP cytotoxicity only under standard oxygen conditions. By contrast, l-PAM cytotoxicity is significantly decreased by hypoxia, but this effect is overcome by eGFP-mediated oxidative stress.

Figure 4.

Cytotoxicity of ETOP or l-PAM was tested in atmospheric oxygen (♦, 20% O2) or hypoxic (⋄, 2% O2) culture conditions. ETOP toxicity was significantly reduced by hypoxia in CHLA-20-eGFP (P = 0.004), but not in parental CHLA-20 and CHLA-20-CD80 (P > 0.159). In contrast to ETOP, l-PAM cytotoxicity was significantly decreased by hypoxia in CHLA-20-CD80 (P = 0.041), but not in CHLA-20-eGFP (P = 0.505). For parental CHLA-20, the fractional survival at 2.5–10 μg/ml of l-PAM was significantly higher in 2% O2 compared to the results in 20% O2 (P < 0.001), although no difference occurred at 1.25 μg/ml (P = 0.074).

Figure 4.

Cytotoxicity of ETOP or l-PAM was tested in atmospheric oxygen (♦, 20% O2) or hypoxic (⋄, 2% O2) culture conditions. ETOP toxicity was significantly reduced by hypoxia in CHLA-20-eGFP (P = 0.004), but not in parental CHLA-20 and CHLA-20-CD80 (P > 0.159). In contrast to ETOP, l-PAM cytotoxicity was significantly decreased by hypoxia in CHLA-20-CD80 (P = 0.041), but not in CHLA-20-eGFP (P = 0.505). For parental CHLA-20, the fractional survival at 2.5–10 μg/ml of l-PAM was significantly higher in 2% O2 compared to the results in 20% O2 (P < 0.001), although no difference occurred at 1.25 μg/ml (P = 0.074).

Close modal

Green fluorescent protein (GFP) provides an attractive and widely used selection marker for gene transduction (1) and is also employed as a marker for detecting tumor cells in vivo and in vitro for biological and therapeutic studies (2–11). Transduced GFP was shown to be nontoxic to cells in some studies (12), but able to promote caspase-3 activation and apoptosis in others (13). We hypothesized that GFP transduction could cause cellular stress that would enhance the cytotoxic effects of chemotherapeutic agents, potentially effecting assays testing chemotherapy against GFP-transduced cells. We tested the effect of GFP and derivatives of GFP (eGFP and YFP) for the ability to sensitize human neuroblastoma cell lines to various cytotoxic agents, using a highly efficient lentiviral vector to enhance the speed and efficiency of gene transduction (22). As a control for these experiments, and as a possible alternate selection marker for gene transduction, we substituted the cDNA for the human CD80 antigen, which is expressed by antigen-presenting cells as a co-stimulation factor for T cells (30), and was not expressed on neuroblastoma cell lines. We showed that relative to both parental and CD80-transduced cell lines, the transduction of any of the tested fluorescence proteins caused a significant increase in the sensitivity to cytotoxic agents of neuroblastoma cell lines.

ROS are involved in a variety of cellular processes, and can enhance the cytotoxic activity of various anticancer drugs (14, 15). Because ROS can be generated by GFP (16), we hypothesized that the enhanced anticancer drug sensitivity we observed in GFP-transduced neuroblastoma cell lines was due to ROS. Consistent with this hypothesis, we observed that clones of cells after GFP transduction appear to be selected for cells with up-regulated glutathione in a p53-dependent fashion, apparently as a means to counter the oxidative stress produced by GFP. Photoactivated GFP generates singlet oxygen (16), but whether or not it generates ROS, with or without ambient light stimulation, is unknown. Unfortunately, green fluorescence from GFP precluded us from measuring ROS using 2′,7′-dichlorofluorescein diacetate (26). However, the effect of GFP transduction on drug sensitivity was altered by both hypoxia and the thiol antioxidant NAC, implying that oxidative stress in neuroblastoma cell lines is normally environmental oxygen dependent, while eGFP generates toxic oxidative stress even under hypoxia.

Our results suggest that the toxicity of ETOP is not greatly affected by oxidative stress, but that ROS generated by eGFP can enhance the ETOP toxicity only in cooperation with endogenous cellular ROS, which is decreased in neuroblastoma cell lines by hypoxia or by the antioxidant NAC (28). Consistent with previous studies of alkylating agents (14) and similar to our results with BSO (28), l-PAM cytotoxicity is diminished in hypoxia, and the oxidative stress from eGFP overcomes the effect of hypoxia on l-PAM cytotoxicity.

We have previously shown that p53 is nonfunctional in CHLA-90 due to a TP53 mutation, whereas CHLA-20 has wild-type and functional TP53 (19). Sensitization of neuroblastoma cell lines to cytotoxic drugs by GFP occurred to a lesser extent in the multidrug-resistant, TP53-mutated cell line CHLA-90. We observed that both p53 functional neuroblastoma cell lines (CHLA-20 and SMS-SAN) showed increased GSH in eGFP-transduced clones, while the p53 nonfunctional cell lines CHLA-90 (TP53 mutated) and CHLA-171 (wild-type but nonfunctional TP53) neuroblastoma cell lines did not show increased GSH when transduced with eGFP. Because p53 is involved in ROS-mediated apoptosis (31), cells with nonfunctional p53 may tolerate ROS produced by eGFP, allowing survival of eGFP-transduced cells without up-regulation of GSH.

In conclusion, eGFP transduction results in oxidative stress that enhanced the sensitivity of neuroblastoma cell lines to a variety of anticancer drugs. These data suggest that extreme caution must be used when employing eGFP-transduced cells to test the response of tumor cells to anticancer drugs. This same caution should extend to the use of GFP-transduced cells for studying the cell biology and behavior of cells in vitro and in vivo. Some experimental settings allow comparison of gene-transduced cells to GFP-only transduced cells as a control to account for the GFP-only effects on cell behavior. However, many of the uses of GFP, especially in testing anti-cancer agents, cannot employ such controls because the GFP-only cells are the primary cells used in the assay.

In this study, we used CD80 as an alternative selection marker for gene transduction. We showed that while CD80 was not detected by an anti-CD80 monoclonal antibody on neuroblastoma cells, it could be readily detected in neuroblastoma cells after CD80 lentiviral transduction. CD80 transduction caused far less enhancement in drug sensitivity and did not select for cells with increased GSH levels. Thus, genes such as the gene-encoding CD80, a cell surface antigen that is nonfunctional in the target cells, might provide a superior selection marker for gene transduction than does the currently widely employed green fluorescent protein.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Grant support: Neil Bogart Memorial Laboratories of the T.J. Martell Foundation for Leukemia, Cancer, and AIDS Research; National Cancer Institute Grants CA82830 and CA60104; and a fellowship from Hope Street Kids (to H. Goto).

The authors thank DeAnna Reed for outstanding editorial assistance in the preparation of this manuscript.

1
Wahlfors, J., Loimas, S., Pasanen, T., and Hakkarainen, T. Green fluorescent protein (GFP) fusion constructs in gene therapy research.
Histochem. Cell Biol.
,
115
:
59
–65, 
2001
.
2
Yang, M., Baranov, E., Li, X. M., Wang, J. W., Jiang, P., Li, L., Moossa, A. R., Penman, S., and Hoffman, R. M. Whole-body and intravital optical imaging of angiogenesis in orthotopically implanted tumors.
Proc. Natl. Acad. Sci. USA
,
98
:
2616
–2621, 
2001
.
3
Lee, N. C., Bouvet, M., Nardin, S., Jiang, P., Baranov, E., Rashidi, B., Yang, M., Wang, X., Moossa, A. R., and Hoffma, R. M. Antimetastatic efficacy of adjuvant gemcitabine in a pancreatic cancer orthotopic model.
Clin. & Exp. Metastasis
,
18
:
379
–384, 
2000
.
4
Lin, X., Gately, D. P., Hom, D., Mishima, M., Los, G., and Howell, S. B. Quantification of tumor cell injury in vitro and in vivo using expression of green fluorescent protein under the control of the GADD153 promoter.
Int. J. Cancer
,
91
:
555
–562, 
2001
.
5
Udagawa, T., Fernandez, A., Achilles, E. G., Folkman, J., and D'Amato, R. J. Persistence of microscopic human cancers in mice: alterations in the angiogenic balance accompanies loss of tumor dormancy.
FASEB J.
,
16
:
1361
–1370, 
2002
.
6
Huang, M. S., Wang, T. J., Liang, C. L., Huang, H. M., Yang, I. C., Yi-Jan, H., and Hsiao, M. Establishment of fluorescent lung carcinoma metastasis model and its real-time microscopic detection in SCID mice.
Clin. & Exp. Metastasis
,
19
:
359
–368, 
2002
.
7
Yang, M., Baranov, E., Wang, J. W., Jiang, P., Wang, X., Sun, F. X., Bouvet, M., Moossa, A. R., Penman, S., and Hoffman, R. M. Direct external imaging of nascent cancer, tumor progression, angiogenesis, and metastasis on internal organs in the fluorescent orthotopic model.
Proc. Natl. Acad. Sci. USA
,
99
:
3824
–3829, 
2002
.
8
Bouvet, M., Wang, J., Nardin, S. R., Nassirpour, R., Yang, M., Baranov, E., Jiang, P., Moossa, A. R., and Hoffman, R. M. Real-time optical imaging of primary tumor growth and multiple metastatic events in a pancreatic cancer orthotopic model.
Cancer Res.
,
62
:
1534
–1540, 
2002
.
9
Torrance, C. J., Agrawal, V., Vogelstein, B., and Kinzler, K. W. Use of isogenic human cancer cells for high-throughput screening and drug discovery.
Nat. Biotechnol.
,
19
:
940
–945, 
2001
.
10
Izycki, D., Gryska, K., Grabarczyk, P., Wysocki, P. J., Jarosinska, A., Nawrocki, S., Kowalczyk, D. W., and Mackiewicz, A. Flow cytometric cytotoxicity assay with GFP gene modified target cells.
Adv. Exp. Med. Biol.
,
495
:
429
–434, 
2001
.
11
Steff, A. M., Fortin, M., Arguin, C., and Hugo, P. Detection of a decrease in green fluorescent protein fluorescence for the monitoring of cell death: an assay amenable to high-throughput screening technologies.
Cytometry
,
45
:
237
–243, 
2001
.
12
Marshall, J., Molloy, R., Moss, G. W., Howe, J. R., and Hughes, T. E. The jellyfish green fluorescent protein: a new tool for studying ion channel expression and function.
Neuron
,
14
:
211
–215, 
1995
.
13
Liu, H. S., Jan, M. S., Chou, C. K., Chen, P. H., and Ke, N. J. Is green fluorescent protein toxic to the living cells?
Biochem. Biophys. Res. Commun.
,
260
:
712
–717, 
1999
.
14
Troyano, A., Fernandez, C., Sancho, P., de Blas, E., and Aller, P. Effect of glutathione depletion on antitumor drug toxicity (apoptosis and necrosis) in U-937 human promonocytic cells. The role of intracellular oxidation.
J. Biol. Chem.
,
276
:
47107
–47115, 
2001
.
15
Kovacic, P. and Osuna, J. A., Jr. Mechanisms of anti-cancer agents: emphasis on oxidative stress and electron transfer.
Curr. Pharm. Des.
,
6
:
277
–309, 
2000
.
16
Greenbaum, L., Rothmann, C., Lavie, R., and Malik, Z. Green fluorescent protein photobleaching: a model for protein damage by endogenous and exogenous singlet oxygen.
Biol. Chem.
,
381
:
1251
–1258, 
2000
.
17
Keshelava, N., Seeger, R. C., Groshen, S., and Reynolds, C. P. Drug resistance patterns of human neuroblastoma cell lines derived from patients at different phases of therapy.
Cancer Res.
,
58
:
5396
–5405, 
1998
.
18
Anderson, C. P., Seeger, R. C., Statake, N., Monforte-Munoz, H. L., Keshelava, N., Bailey, H.H., and Reynolds, C. P. Buthionine sulfoximine and myeloablative concentrations of melphalan overcome resistance in a melphalan-resistant neuroblastoma cell line.
J. Pediatr. Hematol. Oncol.
,
23
:
500
–505, 
2001
.
19
Keshelava, N., Zuo, J. J., Chen, P., Waidyaratne, S. N., Luna, M. C., Gomer, C. J., Triche, T. J., and Reynolds, C. P. Loss of p53 function confers high-level multidrug resistance in neuroblastoma cell lines.
Cancer Res.
,
61
:
6185
–6193, 
2001
.
20
Maurer, B. J., Melton, L., Billups, C., Cabot, M. C., and Reynolds, C. P. Synergistic cytotoxicity in solid tumor cell lines between N-(4-hydroxyphenyl)retinamide and modulators of ceramide metabolism.
J. Natl. Cancer Inst.
,
92
:
1897
–1909, 
2000
.
21
Maurer, B. J., Metelitsa, L. S., Seeger, R. C., Cabot, M. C., and Reynolds, C. P. Increase of ceramide and induction of mixed apoptosis/necrosis by N-(4-hydroxyphenyl)-retinamide in neuroblastoma cell lines.
J. Natl. Cancer Inst.
,
91
:
1138
–1146, 
1999
.
22
Halene, S., Wang, L., Cooper, R. M., Bockstoce, D. C., Robbins, P. B., and Kohn, D. B. Improved expression in hematopoietic and lymphoid cells in mice after transplantation of bone marrow transduced with a modified retroviral vector.
Blood
,
94
:
3349
–3357, 
1999
.
23
Stripecke, R., Cardoso, A. A., Pepper, K. A., Skelton, D. C., Yu, X. J., Mascarenhas, L., Weinberg, K. I., Nadler, L. M., and Kohn, D. B. Lentiviral vectors for efficient delivery of CD80 and granulocyte-macrophage-colony-stimulating factor in human acute lymphoblastic leukemia and acute myeloid leukemia cells to induce antileukemic immune responses.
Blood
,
96
:
1317
–1326, 
2000
.
24
Levy, J. P., Muldoon, R. R., Zolotukhin, S., and Link, C. J., Jr. Retroviral transfer and expression of a humanized, red-shifted green fluorescent protein gene into human tumor cells [see comments].
Nat. Biotechnol.
,
14
:
610
–614, 
1996
.
25
Proffitt, R. T., Tran, J. V., and Reynolds, C. P. A fluorescence digital image microscopy system for quantifying relative cell numbers in tissue culture plates.
Cytometry
,
24
:
204
–213, 
1996
.
26
Anderson, C. P., Tsai, J. M., Meek, W. E., Liu, R. M., Tang, Y., Forman, H. J., and Reynolds, C. P. Depletion of glutathione by buthionine sulfoxine is cytotoxic for human neuroblastoma cell lines via apoptosis.
Exp. Cell Res.
,
246
:
183
–192, 
1999
.
27
Vandeputte, C., Guizon, I., Genestie-Denis, I., Vannier, B., and Lorenzon, G. A microtiter plate assay for total glutathione and glutathione disulfide contents in cultured/isolated cells: performance study of a new miniaturized protocol.
Cell Biol. Toxicol.
,
10
:
415
–421, 
1994
.
28
Yang, B, Keshelava, N., Anderson, C. P., and Reynolds, C. P. Antagonism of buthionine sulfoximine cytotoxicity for human neuroblastoma cell lines by hypoxia is reversed by the bioreductive agent tirapazamine.
Cancer Res.
,
63
:
1520
–1526, 
2003
.
29
De Flora, S., Cesarone, C. F., Balansky, R. M., Albini, A., D'Agostini, F., Bennicelli, C., Bagnasco, M., Camoirano, A., Scatolini, L., and Rovida, A. Chemopreventive properties and mechanisms of N-acetylcysteine. The experimental background.
J. Cell Biochem. Suppl.
,
22
:
33
–41, 
1995
.
30
Salazar-Fontana, L. I. and Bierer, B. E. T-lymphocyte coactivator molecules.
Curr. Opin. Hematol.
,
8
:
5
–11, 
2001
.
31
Gansauge, S., Gansauge, F., Gause, H., Poch, B., Schoenberg, M. H., and Beger, H. G. The induction of apoptosis in proliferating human fibroblasts by oxygen radicals is associated with a p53- and p21WAF1CIP1 induction.
FEBS Lett.
,
404
:
6
–10, 
1997
.