Mutation Research, 237 (1990) 95-106 Elsevier
95
MUTAGI 09047
Cytotoxic effects of expression of human superoxide dismutase in bovine adrenocortical cells K i m H. N o r r i s a n d P e t e r J. H o r n s b y Department of Cell and Molecular Biology, Medical Collegeof Georgia, Augusta, GA 30912 (U.S.A.) (Received 3 January 1990) (Revision received 7 February 1990) (Accepted 14 February 1990)
Keywords: Superoxide dismutase; Transfcction; Senescence; Adrenocortical cells; Cytotoxicity; Cell fusion
Suramary Oxygen radicals and the cellular antioxidant enzymes may play a role in cellular senescence. We studied the feasibility of altering oxygen radical metabolism in a normal differentiated cell that undergoes senescence in culture by transfection of an expression vector containing human CuZn-SOD eDNA. Plasmid pRSV2-cSOD was constructed to contain the eDNA for human CuZn-SOD under the regulation of the Rous sarcoma virus long terminal repeat. Early passage cultures of bovine adrenocortical cells were cotransfected with pRSV2-cSOD and a plasmid (pSV3neo) allowing initial selection and continued growth of transfectants. Three passages after isolation of the polyclonal population, as cells grew to confluence, cultures showed focal cell death that spread outward to affect neighboring cells, so that by 72 h most cells had detached from the culture dish. Long term growth of the polyclonal population of tran~fectants without extensive cell death was achieved by continuous maintenance of low cell density during growth. Southern blot analysis of DNA from the pooled polyclonal population of transfected cells showed the presence of the expected 625 bp band from human CuZn-SOD. However, the intensity of this band indicated that only a minority of cells in the population had integrated the SOD plasmid, and DNA isolated from cells after 25 passages at low cell density/showed plasmid sequences only of an altered form, suggesting that cells containing intact human SOD eDNA had been selectively lost from the population. When early passage low density transfectants were allowed to grow back to high cell density, cell death foci were again observed. Additionally, ceil fusion with the formation of giant cells with massive multinucleation was observed by fluorescence microscopy after staining cultures with a DNA binding dye. In later stages of this process, the large nuclear mass in such a giant cell became fragmented as the cell detached from the dish and formed the center of a focus of cell death in the surrounding cells. Because cell death prevented the growth of large numbers of transfected cells, it was not possible to demonstrate the involvement of CuZn-SOD in the cytotoxic effect by direct means, but the control plasmid without the CuZn-SOD eDNA insert had no cytotoxic effect. Thus, the introduction of a vector for human CuZn-SOD in a normal differentiated cell caused a cytotoxic effect involving cell death, cell fusion, and nuclear fragmentation.
Correspondence: Dr. P.J. Hornsby, Department of Cell and Molecular Biology, Medical College of Georgia, Augusta, GA 30912
(U.S.A.). 0921-8734/90/$03.50 © 1990 Elsevier Science Publishers B.V. (Biomedical Division)
96 Bovine adrenocortical cells undergo cellular senescence in culture, a process that involves losses of differentiated gene expression as well as the loss of the ability to replicate (Hornsby et al., 1989, 1990). We have been interested in the role of oxidative damage in these changes in cell behavior, since oxidative damage is a constant potential threat to cells in culture (Taylor, 1984). When we found that (i) oxidative damage is a principal cause of the lack of appropriate enzymatic activity of one of the steroid hydroxylases, llfl-hydroxylase, in adrenal cells in culture, via interaction of the enzyme with pseudosubstrate steroids (Hornsby, 1980); and that (ii) adrenal cells under normal culture growth conditions are deficient in both selenium and a-tocopherol (Hornsby et al., 1985b), it became evident that oxidative damage might have extensive effects on cell functions that could play a role in senescence generally. Supplementation of cultures with selenium or vitamin E greatly reduced single-strand breaks in cellular DNA resulting from incubation of cells with cumene hydroperoxide (Hornsby and Harris, 1987). However, effects of vitamin E and selenium on total replicative lifespan were minor (Hornsby and Harris, 1987). Thus, vitamin E/selenium-preventable damage did not appear to be a major factor in adrenocortical replicative senescence (Hornsby and Harris, 1987; Norris and Hornsby, 1988), but these experiments did not rule out oxidative damage as a factor in senescence because such damage could be of a form not preventable by vitamin E or selenium. In particular, cellular antioxidant enzymes other than glutathione peroxidase might be able to prevent different forms of DNA damage (Fridovich, 1983; Cerutti, 1985; Vuillaume, 1987; Imlay and Linn, 1988), and thus increases in the cellular antioxidant enzymes might alter the process of cellular senescence. In other work designed to alter the intracellular levels of one of the principal antioxidant enzymes, CuZn-superoxide dismutase (SOD), liposome-entrapped CuZn-SOD delivered to cultured endothelial cells prevented oxygen cytotoxicity (Freeman et al., 1983). Subsequently, transfection was used to create mouse cell lines overexpressing human CuZn-SOD (Elroy-Stein et al., 1986). These cells were more resistant to paraquat, a herbicide
known to increase intracellular production of superoxide. Here, we approached the question of whether it is feasible to alter oxygen radical metabolism in a normal differentiated cell, the bovine adrenocortical cell, by transfection of an expression vector containing human CuZn-SOD cDNA. However, the introduction of this vector caused a cytotoxic effect involving cell death, cell fusion, and nuclear fragmentation. Because cell death prevented the growth of large numbers of transfected cells, it was not possible to demonstrate the involvement of CuZn-SOD in the cytotoxic effect by direct means, but a control plasmid, identical with the exception that it lacked the CuZn-SOD cDNA insert, had no such effect. Materials and methods
Bovine adrenocortical cell culture preparation and transfection Adrenocortical cells were prepared from adrenal cortex tissue (zona fasciculata-reticularis) from 2year-old steers by collagenase/DNAase digestion, as previously described (Gospodarowicz et al., 1977). Cells were stored frozen in 5% dimethyl sulfoxide until required for experiments. Frozen cells were thawed and plated in culture dishes coated with fibronectin. Cells were grown in a 1 : 1 mixture of Dulbecco's Eagles's medium and Ham's F-12 medium with 10% fetal bovine serum (Armour Inc.), 100 n g / m l of partially purified brain fibroblast growth factor, 20 nM selenite, 1 /~M a-tocopherol, and antibiotics. The gas phase used was 5% 02, 90% N 2 and 5% CO 2. Subculturing was performed by incubation with Pronase E (neutral protease type XIV, Sigma) in serum-containing medium. For long term growth studies, adrenocortical cells were grown with a 1 : 5 split ratio at subcultures. Transfection was performed on tertiary (second passage) cultures. Construction of vector pRSV2-cSOD pRSV2-cSOD was constructed to place cDNA for human CuZn superoxide dismutase under the regulation of the Rous sarcoma virus long terminal repeat (LTR). pRSV is a variant of the eukaryotic expression vector pSV2 (Southern and Berg, 1982), utilizing the LTR from Rous sarcoma
virus (RSV), and was supplied by Dr. Cori Gorman, Genentech Inc., South San Francisco, CA. pRSV was chosen as an expression vector based on data showing the RSV-LTR to be an efficient promoter in many mammalian cell types (Gorman, 1985). In the pRSV plasmids a polylinker has been placed downstream of the LTR. pRSV2 contains the polylinker inserted in the orientation 5'-3' EcoRI-EcoRV. Additionally, pRSV contains the cDNA for mouse dihydrofolate reductase (DHFR) under the control of the SV40 promoter, thus providing for co-amplification, using the drug methotrexate, of D H F R and the gene introduced into the vector at the polylinker site (Wurm et al., 1986). Originally we had intended to use this feature to amplify the integrated DNA cSOD sequences in order to increase the expression of cSOD in the cell (Norris and Hornsby, 1988). However, because of the observed phenotype of transfected, non-amplified, cells, this feature could not be used in these experiments. cDNA for human SOD (Groner et al., 1985) was generously supplied by Dr. Yoram Groner, Weizmann Institute, Tel Aviv, in the pUC18 cloning vector, pUC18-cSOD was digested with the restriction endonucleases PvuII and BamHI and the 732 bp fragment containing the human SOD cDNA was inserted into the SmaI-BamHI sites of the polylinker of pRSV2 (Maniatis et al., 1982). Plasmids were purified through ethidium bromide-CsC1 gradient centdfugation before use for transfections. The purified plasmid DNA was dialyzed and reprecipitated with ethanol.
Transfection Tertiary cultures of bovine adrenocortical cells were used for the transfection experiments. One week prior to transfection, cells were incubated with protease (Pronase E) and transferred into 10 cm culture dishes at low density. Cultures were allowed to grow to - 7 5 % confluence before transfection. At this time, cultures are in vigorous growth. Cells ( 2 x 1 0 6 - 1 x lO 7) were removed from dishes with Pronase and resuspended in 0.6 ml transfection buffer (140 mM NaC1, 25 mM HEPES, 0.75 mM Na2HPO4, pH 7.2) containing 10 pg/ml linearized pRSV-cSOD plasmid DNA and 10 pg/nd pSV3neo (Neumann et al., 1982).
Plasmid pRSV2-cSOD or pRSV2 was linearized by digestion with NdeI and pSV3neo was linearized with EcoRI. Cells were incubated for 10 min at 0 ° C before transfer to the sample chamber of a Prototype Design Services Model ZA 1000 electroporation apparatus. They were electroporated with a 2.5 k V / c m electric field and kept on ice for an additional 10 min before being returned to culture medium. Cells were maintained in standard medium for 24 h before transfer into medium containing 200 # g / m l G418 (G-ibco Laboratories, Grand Island, NY) for selection of transfectants. Long term growth studies were all in medium containing 200 /~g/ml G418.
Isolation of DNA and Southern blotting DNA was prepared either by phenol/chloroform extraction or using a commercial anion exchange column (Extractor, Molecular Biosystems, San Diego, CA) designed for extracting nucleic acids from low numbers of cells. After restriction enzyme digestion, digests were fractionated on 0.8% agarose and DNA was transferred to nylon (Pall Biodyne Inc.) (Maniatis et al., 1982). Plasmid pUC18-cSOD was digested with PstI to create a 625 bp SOD cDNA probe, and pRSV2 was digested with StuI to create a 999 bp D H F R cDNA probe. The latter was used as a probe for pRSV2cSOD, outside of the SOD insert, that does not cross-react with other plasmid sequences in pSV3neo. Inserts were isolated using low melting temperature agarose (SeaKem, FMC Bioproducts) and labeled with 32p-dCTP using random oligonucleotide priming (Pharmacia kit). Membranes were hybridized with standard techniques (Maniatis et al., 1982). Following overnight hybridization, membranes were washed with 0.5 x SSC with 0.5% SDS and exposed to X-ray film for 6-14 days. Fluorescence observations of cells with DAPI (4,6diamidino-2 -phenylindole) For combined phase contrast/DAPI fluorescence observations, cells were plated in Petriperm dishes (Hereaus Inc., South Plainfield, N J) which have a thin, flexible, hydrophilic polytetrafluoroethylene bottom allowing for superior optics. After growth to the required cell density, cultures were fixed with 4% freshly made paraformaldehyde,
98
and were then dehydrated with acetone at - 2 0 °C for 5 rnin. Following the removal of the acetone, the dish was allowed to air dry. DAPI (Sigma Chemical Co.) was added as a 0.5 # g / m l solution in phosphate buffered saline with 5 mM Mg 2+ for 10 min. Cells were now visualized using an excitation wavelength of 365 um and an emission barher filter of 420 nm using a Nikon Diaphot epifluorescence microscope. Results Transfection of an expression vector for human CuZn-SOD into cultured bovine adrenocortical cells Plasmid pRSV2-cSOD, containing eDNA for human CuZn-SOD under the regulation of the Rous sarcoma virus LTR, was transfected into bovine adrenocortical cells by eleetroporation. In order to obtain selective growth of transfectants, cultures were cotransfected with pRSV2-cSOD and pSV3neo, a plasmid carrying the bacterial neo gene which confers resistance to the neomycin analog G418. pSV3neo also contains the early region of SV40 virus; SV40 T antigen has been shown in previous experiments to cause extended, vigorous growth of bovine adrenocortical cells (Cheng et al., 1989). In preliminary experiments, cotransfection with pSV2neo, which is similar to pSV3neo but lacks the SV40 early region encoding T antigen, produced cells that senesced soon after transfeetion and thus could not be used for these experiments. This indicates that T antigen may counteract residual G418 toxicity which may otherwise limit the growth of transfectants, since normal non-transfeeted clones have a much greater growth potential than pSV2neo transfectants (Cheng et al., 1989). After transfection, cells were replated at moderate cell density in 6 cm dishes in medium with G418. After 13 days, a large number of G418 resistant colonies were observed. All the colonies were pooled and were subcultured at a 1 : 2.5 split ratio into 10 cm dishes. When these cultures were confluent, 90% of the cells were frozen in liquid nitrogen and stored for future experiments. Southern blot analysis of DNA isolated from transfectants 2 passages after pooling and freezing of the polyclonal population was used to assess the state of integration of the plasmid sequences (Fig. 1). After growth of the cell population to a
A
A~" 0
0
> u) ~" ¢~.
> u~ ¢E ¢~.
o o
o
~ ¢~.
o
o
? ? O m I-
>,
~,
--23 --23 --9.4 --6.5 --4.3
--9.4 --6.5 --4.3
--2.3 --2.0
--2.3 --2.0
--1.3 --0.56
--1.0
Fig. 1. Detection of human SOD cDNA by Southern blot analysis in DNA from bovine adrenocortical cells transfected with pRSV2-cSOD. (A) DNA from the following32celis was isolated, restricted with PstI, and probed with P-labeled human SOD cDNA. TBAC: tertiary (second passage) cultured bovine adrenocortical cells; normal cells that were used for transfection, pSV3neo: bovine adrenocortical cells transfected with pSV3neo only. pRSV2-cSOD (2): bovine adrenocortical cells cotransfected with pRSV2-cSOD and pSV3neo, passage 2 after transfection, pRSV2-cSOD (25): bovine adrenocortical cells cotransfected with pRSV2-cSOD and pSV3neo, passage 25 after transfection. The arrow indicates the position of the expected 625 bp human SOD cDNA. Positions of HindIII digested ~, size markers are indicated. (B) The blots from A were reprobed with riP-labeled DHFR from pRSV2. Lane identifiers are the same as for the SOD hybridization. The arrow indicates the position of the expected 2.9 kb PstI fragment of pRSV2-cSOD. Exposure has been adjusted to equalize the intensity of the endogenous band at - 6.0 kb.
sufficient size, DNA was prepared and restricted with PstI, which excises human SOD cDNA from pRSV2-cSOD. Hybridization with the PstI cDNA insert from pUC18-cSOD showed the presence of the expected 625 bp band in the transfectants. Presumably because of the high degree of homology between human and bovine CuZn-SOD
(Steinman et al., 1974; Sherman et al., 1983) the human SOD cDNA probe hybridized strongly to endogenous bovine SOD sequences. Transfectants were grown at low cell density for 25 passages, using the protocol described later. At this point, DNA was isolated from cultures and reanalyzed for the presence of human SOD cDNA. The expected 625 bp band was not observed, but an additional larger fragment of ~ 5.5 kb, not detected either in control, non-transfected cells or in early passage transfectants, was observed to hybridize with the human SOD probe. Autoradiogram~ were analyzed by scanning densitometry. By using the endogenous bovine SOD gene sequences as internal standards, it was confirmed that the 625 bp band was completely absent in 25th passage ceils and that the new - 5 . 5 kb band was not detectable in passage 2 cells or in non-transfectants. Moreover, it was clear that, in both early passage and late passage transfectants, the SOD eDNA was present at less than 1 copy per cell, although the precise copy number cannot be determined because the degree of cross-reaction of the probe with endogenous SOD sequences is not known. Thus, although all ceils presumably successfufiy integrated pSV3neo, since cells were G418 resistant, only a minority of ceils received pRSV2-cSOD. This is consistent with observations on the population behavior of the transfectants, described below, which suggest that a minority of ceils differed in phenotype from the remainder of the population. The lack of the expected cSOD band in late passage transfectants was not due to a complete lack of plasmid sequences in these cells. When blots were rehybddized with 32p-labeled DHFR, to serve as a probe for a different region of the pRSV2 plasmid, the expected 2.9 kb plasmid band was present in pRSV2-cSOD/pSV3neo cotransfectants (Fig. 1). Endogenous bovine sequences cross-hybridized less strongly with the probe in this case. The 2.9 kb band was present in transfectant DNA from cells at both 2 and 25 passages, but not in the normal cells. Growth of transfectants and observation of loci of cell death The polyclonal population of pRSV2-cSOD/ pSV3neo cotransfected cells was grown in medium
with G418 at 1 : 5 sprit ratios. Three passages after isolation of the polyclonal population, as cells grew to confluence, loci of rounded, dying cells were observed. This phenomenon was noted reproducibly at this passage, when vials of cells frozen at the isolation of the polyclonal population were replated and grown in culture. Cell death began at a focal point and spread outward, affecting neighboring cells within 24 h. At 24 or 48 h after initiation of these foci, cells were seen in various stages of rounding up and detachment from the culture dish. By 72 h, < 10% of cells still remained attached to the culture dish (Fig. 2). In several experiments, conducted either separately or simultaneously with the transfection with pRSV2-cSOD, cells transfected with pSV3neo alone, or cells cotransfected with pSV3neo and pRSV2 (the parent plasmid lacking the SOD cDNA insert), never showed any evidence of focal cell death, at any passage level. The number of foci of dying cells per dish of transfectants suggested that the cell death was initiated by a minority of cells in the population. In order to determine what fraction of the cell population was capable of initiating foci, pRSV2cSOD/pSV3neo cotransfectants frozen from the polyclonal population were seeded at low density in a 96 well plate ( < 10 cells per well). The cells were allowed to grow in medium with G418. After 12 days, each well was observed for presence of foci of cell death. The numbers of wells with and without cell death were tabulated (Table 1). This experiment gives a value of - 24% for the upper limit of the fraction of cells that can initiate foci. The actual value is likely to be much less, since most wells were seeded with more than one cell. pRSV2-cSOD/pSV3neo cotransfectants were grown under various culture conditions in order to test possible mechanisms for the cell death phenomenon and to assess the feasibility of successful growth of pure cultures of these cells in order to allow for further analysis. Factors examined included omission or inclusion of G418; omission of serum; inclusion of a combination of growth factors (UltroSer G) previously found to enhance the growth of bovine adrenocortical cells (McAllister and Hornsby, 1987); and replacement of serum with ether-extracted serum (Hornsby et al., 1985a) to assess the possible role of serum lipids. None of
100
g~: .Z:,-.., ..... . . . . . . . . . . . . . . . .
,,~,,~,~ ,~ ........ ~ , ~ . , ~ , , ~
Fig. 2, Cell death in cultures of bovine adrenocortical cells cotransfected with pRSV2-cSOD and pSV3neo. (A) Culture of cells 3 passages following transfection, just after reaching confluence, (B) Culture 24 h later, showing evidence of initiation of loci of cell death. (C) Culture 48 h later, showing widening of loci, some cells having detached from the substratum. (D) Culture 72 h later, showing effects of extensive spread of cell death and almost complete detachment of cells. Phase contrast micrographs × 50.
101 TABLE 1 ANALYSIS OF FRACTION OF pRSV2-cSOD/pSV3neo TRANSFECTED CELL POPULATION ABLE TO FORM FOCI OF CELL DEATH Exp. 1 Number of wells with cell death foci Confluent Non-confluent Number of wells without cell death foci Confluent Non-confluent Total number of wells Total number of wells with cell death loci U p p ~ limit to fraction of cells in ~hc population able to i~tia~c loci of cell d ~ t h :
Exp. 2
9 7
6 3
15 37 68
2 24 35
16
6
16 = 23.5% ~ =17.1% b~g
p RSV2-cSOD/pSV3neo transfected bovine adrenocortical cells were plated in 96 well plates at low density ( < I0 cells per well). After 10-12 days, when msny wells had reached confluence, cells were fixed, stained, and egamined for numbers of fO~i of cell death.
_
_ _ ~ , _ _
these conditions affected the incidence or severity of'cell death in the transfectant cultures. The observation that cell death began at a focal point and spread outward to neighboring cells suggested that the initiating cells might release a toxic factor or transmissible agent (e.g., virus) into the medium. In order to check for such a factor, medium was removed from a plate of transfectants with extensive cell death, filtered through a 0.22 # m filter, diluted 1 : 1 with fresh medium, and added to a - 80% confluent plate of control (pSV3neo transfected) cells. This addition was repeated for several days, without evidence of cell death in the recipient cells. These data appear to exclude the involvement of (i) a diffusible toxic factor, unless it is too labile for transfer to a target cell population, and (ii) a transmissible agent, or at least an agent that passes through a 0.22 #m filter. Became production of hydrogen peroxide might be elevated in cells expressing high levels of SOD, catalase (100 # g / m l ) was tested as a possible inhibitor of cell death, but was without effect. It
T - -
Fig. 3. Cell death and multinucleation in cultures of pRSV2-cSOD/pSV3neo cotransfected bovine adrenocortical cells during growth to higher cell density. Transfectants were grown with maintenance of < 20% confluence for 6 passages after isolation of the polyclonal cell population, and were then allowed to grow back from this low cell density to higher density over a period of 10 days. (Left) A group of cells in various stages of rounding and detachment from the dish. (Right) Several cells with massive multinucleation adjacent to a focus of cell death. Phase contrast micrographs × 50.
102
was also thought that intracellular production of hydrogen peroxide might occur. Because this would not be accessible to extracellular scavenging enzymes, we investigated the possible protective effects of dimethyl sulfoxide, which we had previously shown to be an effective radioprotectant in these cells (Hornsby et al., 1984). The radioprotectant effect results from scavenging of the hydroxyl radical, the probable final agent of toxicity of hydrogen peroxide (Floyd, 1981). We noted that 2% dimethyl sulfoxide caused a slowing of growth such that cultures never became completely confluent. Under these circumstances, it appeared that scattered individual cells died but spreading of death to neighboring cells was pre-
vented. Although it is possible that OH" scavenging by dimethyl sulfoxide was involved in prevention of the spread of cell death, it seemed more likely that its major effect was on cell growth, indicating that the spreading of cell death might be dependent on cell density. Combined with evidence that a transmissible toxic factor could not be demonstrated, this suggested that the factor may not be freely diffusible, requiring cell contact for the spread of cell death. Growth of pRSV2-cSOD/pSV3neo cotransfectants under conditions in which < 20% confluence was maintained by appropriate timing of subcultures, thus preventing cultures from achieving confluence, showed that extensive cell death
Fig. 4. Phase contrast appearance and DAPI fluorescence of p R S V 2 - c S O D / p S V 3 n e o cotransfectants during multinucleation and nuclear fragmentation. Brackets on the phase contrast images indicate the area of the corresponding fluorescence image. Sixth passage transfectants were grown as described in Fig. 3 and were fixed and stained with DAPI. ( A - C ) Examples of multinucleated cells. ( D - G ) Cells in various stages of nuclear fragmentation and cell detachment. Magnification × 600.
103 could in fact be avoided. Cells-were.grown for 25 passages ( - 63 population doublings) under these conditions. However, as shown in Fig. 1, Southern blot analysis of DNA from 25th passage transfectants showed an absence of cells carrying the expected 625 bp SOD cDNA, suggesting that such cells had been lost from the population despite the maintenance of low cell density. Apparently, growth at low cell density did not prevent the death of pRSV2-cSOD transfected cells, but
spreading of cell death to other cells in the population was avoided.
Multinucleation, nuclear fragmentation, and cell death During low density growth of pRSV2cSOD/pSV3neo cotransfectants, continuous proliferation was observed, although, especially in later passages, abnormal numbers of cells in metaphase were noted. Foci of cell death were not
Fig. 4 (continued).
104
observed. We examined whether cell death would reappear when low density cultures were allowed to grow back to high cell density. As low density cultures grew to confluence, foci of cell death appeared (Fig. 3). Associated with cell death foci were giant cells with massive multinucleation (Fig. 3). Cultures containing giant cells with various stages of multinucleation and cell death were fixed and stained with a fluorescent DNA binding dye, DAPI, for observation of nuclear morphology. Cells surrounding the giant cells had nuclei that were smooth in outline and regular in shape and size. In multinucleated cells, a variable number of nuclei were clustered in the center of the cell, sometimes appearing to be in the process of fusion into an irregularly shaped mass (Fig. 4A-C). Nuclear number varied continuously in the range of 2 to - 6 0 . In later stages of this process, the large nuclear mass became fragmented, with nuclear material becoming condensed into brightly fluorescent spots (Fig. 4D-G). The phase contrast appearance of such cells during nuclear fragmentation showed withdrawal of contact from the dish and eventual detachment. Adjacent to such multinucleated, detaching, cells were individual dying cells with fragmented and condensed nuclear material. Multinucleation and cell death were observed in cultures of transfectants at early passages following isolation of the transfected population. However, at later passages, cell death and mnltinucleation were not observed when cells were allowed to grow back to high cell density. This correlates with the absence of cells with the expected 625 bp SOD cDNA band observed by Southern blotting in later passage cells (Fig. 1). Discussion
Transfection of an expression vector containing human CuZn-SOD cDNA did not appear to increase the resistance of a normal differentiated cell type, the bovine adrenocortical cell, to oxidative damage. The introduction of this vector caused a cytotoxic effect involving cell death and nuclear fragmentation. Because cell death prevented the growth of large numbers of transfected cells, it was not possible to demonstrate the involvement
of CuZn-SOD in the cytotoxic effect by direct means, but a control plasmid, identical with the exception that it lacked the CuZn-SOD cDNA insert, had no such effect. This suggests that the CuZn-SOD cDNA may have contributed to increased, rather than decreased, oxidative damage in the adrenocortical cell. Mouse cells overexpressing human CuZn-SOD, although more resistant to paraquat, were reported to show an increase in thiobarbituric acid products in homogenates of the cultured cells, suggesting an increased potential for lipid peroxidation (ElroyStein et al., 1986). In extensions of the reported work on SOD transfected mouse cells, in transgenic mice, PC12 adrenomedullary cells, and NS20Y neuroblastoma cells, there were additional suggestions of toxic effects of excess CuZn-SOD (Epstein et al., 1987; Elroy-Stein and Groner, 1988; Cebailos et al., 1988). However, no direct cytotoxic effects of excess CuZn-SOD were reported. Cytotoxic effects would have been difficult to observe directly because cells suffering cytotoxic effects would not have survived the initial clonal cell selection. In fact, because transfectants with more than a 6-fold elevation of CuZn-SOD activity were not found, it was speculated that such transfectants might have been lost due to a toxic effect of overexpression (Elroy-Stein et al., 1986). In the experiments reported here, no attempt was made to clone out the SOD cDNA transfected cells from the initial mixed population, comprising both cells that received only the selectable plasmid, pSV3neo, and cells that were successfully cotransfected with pSV3neo and pRSV2-cSOD. Southern blot analysis indicated that the cotransfection efficiency, i.e., the fraction of cells that received both selectable and nonselectable plasmids, was low. This may have been due to the use of electroporation, which may produce less concatemerization of plasmids than calcium phosphate transfection methods (Andreason and Evans, 1988). A greater fraction of cotransfectants in the population would evidently have resulted in much more extensive cell death, perhaps to the extent that a transfected population would not have been isolatable. Cotransfecrants did survive long enough to exert their cytotoxic effect at the third passage after isolation of
105
the polyclonal population. It is not clear why there was a delay in the expression of the cytotoxic effect. Once the cytotoxic effect was expressed, however, it spread to neighboring cells at confluence, eventually resulting in the death of the entire culture. Even when the spread of cell death was prevented by growth of cells at low cell density, the transfectants carrying the expected 625 bp SOD cDNA fragment did not survive long term growth, as shown by Southern blotting after 25 passages. Presumably, these cells either died or were diluted out of the population as the cultures were passed at low density. This indicates that the presence of the 625 bp SOD insert did not give the cells a growth advantage. In late passage transfectants, only an abnormal larger band hybridizing with the SOD probe was observed, probably indicating that the subgroup of transfectants that were able to survive long term growth received only a truncated SOD cDNA sequence during chromosomal integration. However, these ceils still carried other portions of the pRSV2-cSOD plasmid. Presumably, only cells containing the cSOD 625 bp insert initiated the ceil death loci. The non-survival of transfectants carrying the 625 bp cDNA made it impossible to isolate sufficient of these cells for analysis of human CuZn-SOD expression. The large multinucleated cells observed here probably resulted from fusion of individual ceils to multinucleated giant cells. Infection with many viruses results in the creation of large multinucleated cells through a process of cell fusion (Poste, 1970). The subsequent fusion of nuclei to form the irregularly shaped masses observed here presumably results in the death of the multinucleated ceil, with the formation of a focus of cell death affecting adjacent ceils. The close resemblance of the cytotoxic effect to that of a viral infection raises the possibility that introduction of pRSV2-cSOD into cells activated a latent virus or rendered the ceils more susceptible to infection by viruses normally present in the environment, although the occurrence of the cytotoxic effect in a variety of different media argues against the latter possibility. Currently, the sequence of molecular events between integration of pRSV2-cSOD and ceil fusion/ceil death is unknown. Possibly, excess cellu-
lax SOD can act as a peroxidase; SOD can catalyze the peroxidation of linoleic acid in vitro (Hodgson and Fridovich, 1975). Transfected SOD could be expressed in a form that has an altered subcellulax distribution, which might cause a change in the properties of the enzyme. Alternatively, excessive hydrogen peroxide formation could cause increased cellular damage. Presumably, by processes that remain to be elucidated, peroxidative damage resulting from excess SOD expression then causes cell fusion and the subsequent propagation of cell death.
Acknowledgement This work was supported by a grant from the Greenwall Foundation to Dr. J.E. Seegmiller, Institute for Research on Aging, University of California, San Diego, CA.
References Andreason, G.L., and G.A. Evans (1988) Introduction and expression of DNA molecules in eukaryotic cells by electroporatiun, BioTeclmiques, 6, 650-660. Ceballos, I., J.M. Delabar, A. Nicole, R.E. Lynch, R.A. Hallewell, P. Kamouss and P.M. Sinet (1988) Expression of transfected human Cu-Zn SOD gene in mouse L ceils and NS20Y neuroblastoma cells induces enhancement of glutathione peroxidase activity, Biochim. Binphys. Acta, 949, 5864-5870. Cerutti, P. (1985) Prooxidant states and tumor promotion, Sc/ence, 227, 375-381. Cheng, C.Y., R.F. Ryan, T.P. Vo and P.J. Hornsby (1989) Cellular senescence involves stochastic processes causing loss of expression of differentiated function genes: transfection with SV40 as a means for dissociating effects of senescence on growth and on differentiated function gene expression, Exp. Cell Res., 180, 49-62. Elroy-Stein, O., and Y. Groner (1988) Impaired neurotransmitter uptake in PC12 cells overexpressing human C u / Z n superoxide dismutase - impfication for gene dosage effects in Down syndrome, Cell, 52, 259-267. Elroy-Stein, O., Y. Bernstein and Y. Groner (1986) Overproduction of human Cu/Zn-superoxide dismutase in transfected cells: extenuation of paraquat-mediated cytotoxicity and enhancement of lipid peroxidation, EMBO J., 5, 615622. Epstein, C.J., K.B. Avrahan~ M. Lovett, S. Smith, O. ElroyStein, G. Rotman, C. Bry and Y. Groner (1987) Transgenic mice with increased Cu/Zn-superoxide dismutase activity: animal model of dosage effects in Down syndrome, Proc. Natl. Acad. Sci. (U.S.A.), 84, 8044-8048.
106 Floyd, R.A. (1981) DNA-ferrous iron catalyzed hydroxyl free radical formation from hydrogen peroxide, Biochem. Biophys. Res. Commun., 99, 1209-1215. Freeman, B.A., S.L. Young and J.D. Crapo (1983) Liposomemediated augmentation of superoxide dismutase in endothelial cells prevents oxygen injury, J. Biol. Chem., 258, 12534-12542. Fridovich, I. (1983) Superoxide radical: an endogenous toxicant, Annu. Rev. Pharmacol. Toxicol., 23, 239-257. Gorman, C. (1985) High efficiency gene transfer into mammarian cells, in: D.M. Glover (Ed.), DNA Cloning, Vol. 2, IRL Press, Oxford, pp. 143-190. Gospodarowicz, D., C.R. Ill, P.J. Hornsby and G.N. Gill (1977) Control of bovine adrenal cortical cell proliferation by fibroblast growth factor. Lack of effect of epidermal growth factor, Endocrinology, 100, 1080-1089. Groner, Y., J. Lieman-Hurwitz, N. Dafni, L. Sherman, D. Levanon, Y. Bernstein, E. Danciger and O. Elroy-Stein (1985) Molecular structure and expression of the gene locus on chromosome 21 encoding the C u / Z n superoxide dismutase and its relevance to Down syndrome, Ann. N.Y. Acad. Sci., 450, 133-156. Hodgson, E.K., and I. Fridovich (1975) The interaction of bovine erythrocyte superoxide dismutase with hydrogen peroxide: chemiluminescence and peroxidation, Biochemistry, 14, 5299-5303. Hornsby, P.J. (1980) Regulation of cytochrome P-450-supported llfl-hydroxylation of deoxycortisol by steroids, oxygen, and antioxidants in adrenocortical cell cultures, J. Biol. Chem., 255, 4020-4027. Hornsby, P.J., and S.E. Harris (1987) Oxidative damage to DNA and replicative life span in cultured adrenocortical cells, Exp. Cell Res., 168, 203-217. Hornsby, P.J., S.E. Harris and K.A. Aldern (1984) Mode of action of sulfoxides in preventing loss of activity of llfl-hydroxylase in cultured bovine adrenocortical cells, Chem.Biol. Interact., 51, 335-346. Hornsby, P.J., K.A. Aldern and S.E. Harris (1985a) Mode of action of butylated hydroxyanisole (BHA) and other phenols in preventing loss of 11fl-hydroxylase activity in cultured bovine adrenocortical cells, Biochem. Pharmacol., 34, 865-872. Hornsby, P.J., D.W. Pearson, A.P. Autor, K.A. Aldern and S.E. Harris (1985b) Selenium deficiency in cultured adrenocortical cells: restoration of glutathione peroxidase and resistance to hydroperoxides on addition of selenium, J. Cell. Physiol., 123, 33-38. Hornsby, P.J., R.F. Ryan and C.Y. Cheng (1989) Replicative
senescence and differentiated gene expression in cultured adrenocortical cells, Exp. Gerontol., 24, 539-558. Hornsby, P.J., C.Y. Cheng, R.F. Ryan and L. Yang (1990) Stochastic changes in gene expression in adrenal cell senescence, in: C.E. Finch and T.E. Johnson (Eds.), Molecular Biology of Aging, Wiley-Liss, New York, NY, pp. 249-263. Imlay, J.A., and S. Lima (1988) DNA damage and oxygen radical toxicity, Science, 240, 1302-1309. Maniatis, T., E.F. Fritsch and J. Sambrook (1982) Molecular Cloning: A Laboratory Manual, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY. McAilister, J.M., and P.J. Hornsby (1987) Improved clonal and non-clonal growth of human, rat, and bovine adrenocortical cells in culture, In Vitro Cell. Dev. Biol., 23, 677-685. Neumann, E., M. Schaefer-Ridder, Y. Wang and P.H. Hofschneider (1982) Gene transfer into mouse lyoma cells by electroporation in high electric fields, EMBO J., 1, 841-845. Norris, K., and P.J. Hornsby (1988) Oxidative damage to DNA in adrenocortical cells during senescence in culture, in: M.G. Simic, K.A. Taylor, J.F. Ward and C. von Sormtag (Eds.), Oxygen Radicals in Biology and Medicine (Basic Life Sciences, Vol. 49), Plenum, New York, NY, pp. 461466. Poste, G. (1970) Virus-induced polykaryocytosis and the mechanism of cell fusion, Adv. Virus Res., 16, 303-356. Sherman, L., N. Dafni, J. Lieman-Hurwitz and Y. Groner (1983) Nucleotide sequence and expression of human chromosome 21-encoded superoxide dismutase mRNA, Proc. Natl. Acad. Sei. (U.S.A.), 80, 5465-5469. Southern, P.J., and P. Berg (1982) Transformation of mammalian cells to antibiotic resistance with a bacterial gene under control of the SV40 early region promoter, J. Mol. Appl. Genet., 1, 327-341. Steinman, H.M., V.R. Naik, J.L. Abernethy and R.L. Hill (1974) Bovine erythrocyte superoxide dismutase. Complete amino acid sequence, J. Biol. Chem., 249, 7326-7338. Taylor, W.G. (1984) Toxicity and hazards to successful culture: cellular responses to damage induced by light, oxygen or heavy metals, in: M.K. Patterson Jr. (Ed.), Uses and Standardization of Vertebrate Cell Cultures, Tissue Culture Association, Galthersburg, MD, pp. 58-70. Vuillaume, M. (1987) Reduced oxygen species, mutation, induction and cancer initiation, Mutadon Res., 186, 43-72. Wurm, F.M., K.A. Gwirm and R.E. Kingston (1986) Inducible overproduction of the mouse c-myc protein in mammalian cells, Proc. Nat. Acad. Sei. (U.S.A.), 83, 5414-5418.
95
MUTAGI 09047
Cytotoxic effects of expression of human superoxide dismutase in bovine adrenocortical cells K i m H. N o r r i s a n d P e t e r J. H o r n s b y Department of Cell and Molecular Biology, Medical Collegeof Georgia, Augusta, GA 30912 (U.S.A.) (Received 3 January 1990) (Revision received 7 February 1990) (Accepted 14 February 1990)
Keywords: Superoxide dismutase; Transfcction; Senescence; Adrenocortical cells; Cytotoxicity; Cell fusion
Suramary Oxygen radicals and the cellular antioxidant enzymes may play a role in cellular senescence. We studied the feasibility of altering oxygen radical metabolism in a normal differentiated cell that undergoes senescence in culture by transfection of an expression vector containing human CuZn-SOD eDNA. Plasmid pRSV2-cSOD was constructed to contain the eDNA for human CuZn-SOD under the regulation of the Rous sarcoma virus long terminal repeat. Early passage cultures of bovine adrenocortical cells were cotransfected with pRSV2-cSOD and a plasmid (pSV3neo) allowing initial selection and continued growth of transfectants. Three passages after isolation of the polyclonal population, as cells grew to confluence, cultures showed focal cell death that spread outward to affect neighboring cells, so that by 72 h most cells had detached from the culture dish. Long term growth of the polyclonal population of tran~fectants without extensive cell death was achieved by continuous maintenance of low cell density during growth. Southern blot analysis of DNA from the pooled polyclonal population of transfected cells showed the presence of the expected 625 bp band from human CuZn-SOD. However, the intensity of this band indicated that only a minority of cells in the population had integrated the SOD plasmid, and DNA isolated from cells after 25 passages at low cell density/showed plasmid sequences only of an altered form, suggesting that cells containing intact human SOD eDNA had been selectively lost from the population. When early passage low density transfectants were allowed to grow back to high cell density, cell death foci were again observed. Additionally, ceil fusion with the formation of giant cells with massive multinucleation was observed by fluorescence microscopy after staining cultures with a DNA binding dye. In later stages of this process, the large nuclear mass in such a giant cell became fragmented as the cell detached from the dish and formed the center of a focus of cell death in the surrounding cells. Because cell death prevented the growth of large numbers of transfected cells, it was not possible to demonstrate the involvement of CuZn-SOD in the cytotoxic effect by direct means, but the control plasmid without the CuZn-SOD eDNA insert had no cytotoxic effect. Thus, the introduction of a vector for human CuZn-SOD in a normal differentiated cell caused a cytotoxic effect involving cell death, cell fusion, and nuclear fragmentation.
Correspondence: Dr. P.J. Hornsby, Department of Cell and Molecular Biology, Medical College of Georgia, Augusta, GA 30912
(U.S.A.). 0921-8734/90/$03.50 © 1990 Elsevier Science Publishers B.V. (Biomedical Division)
96 Bovine adrenocortical cells undergo cellular senescence in culture, a process that involves losses of differentiated gene expression as well as the loss of the ability to replicate (Hornsby et al., 1989, 1990). We have been interested in the role of oxidative damage in these changes in cell behavior, since oxidative damage is a constant potential threat to cells in culture (Taylor, 1984). When we found that (i) oxidative damage is a principal cause of the lack of appropriate enzymatic activity of one of the steroid hydroxylases, llfl-hydroxylase, in adrenal cells in culture, via interaction of the enzyme with pseudosubstrate steroids (Hornsby, 1980); and that (ii) adrenal cells under normal culture growth conditions are deficient in both selenium and a-tocopherol (Hornsby et al., 1985b), it became evident that oxidative damage might have extensive effects on cell functions that could play a role in senescence generally. Supplementation of cultures with selenium or vitamin E greatly reduced single-strand breaks in cellular DNA resulting from incubation of cells with cumene hydroperoxide (Hornsby and Harris, 1987). However, effects of vitamin E and selenium on total replicative lifespan were minor (Hornsby and Harris, 1987). Thus, vitamin E/selenium-preventable damage did not appear to be a major factor in adrenocortical replicative senescence (Hornsby and Harris, 1987; Norris and Hornsby, 1988), but these experiments did not rule out oxidative damage as a factor in senescence because such damage could be of a form not preventable by vitamin E or selenium. In particular, cellular antioxidant enzymes other than glutathione peroxidase might be able to prevent different forms of DNA damage (Fridovich, 1983; Cerutti, 1985; Vuillaume, 1987; Imlay and Linn, 1988), and thus increases in the cellular antioxidant enzymes might alter the process of cellular senescence. In other work designed to alter the intracellular levels of one of the principal antioxidant enzymes, CuZn-superoxide dismutase (SOD), liposome-entrapped CuZn-SOD delivered to cultured endothelial cells prevented oxygen cytotoxicity (Freeman et al., 1983). Subsequently, transfection was used to create mouse cell lines overexpressing human CuZn-SOD (Elroy-Stein et al., 1986). These cells were more resistant to paraquat, a herbicide
known to increase intracellular production of superoxide. Here, we approached the question of whether it is feasible to alter oxygen radical metabolism in a normal differentiated cell, the bovine adrenocortical cell, by transfection of an expression vector containing human CuZn-SOD cDNA. However, the introduction of this vector caused a cytotoxic effect involving cell death, cell fusion, and nuclear fragmentation. Because cell death prevented the growth of large numbers of transfected cells, it was not possible to demonstrate the involvement of CuZn-SOD in the cytotoxic effect by direct means, but a control plasmid, identical with the exception that it lacked the CuZn-SOD cDNA insert, had no such effect. Materials and methods
Bovine adrenocortical cell culture preparation and transfection Adrenocortical cells were prepared from adrenal cortex tissue (zona fasciculata-reticularis) from 2year-old steers by collagenase/DNAase digestion, as previously described (Gospodarowicz et al., 1977). Cells were stored frozen in 5% dimethyl sulfoxide until required for experiments. Frozen cells were thawed and plated in culture dishes coated with fibronectin. Cells were grown in a 1 : 1 mixture of Dulbecco's Eagles's medium and Ham's F-12 medium with 10% fetal bovine serum (Armour Inc.), 100 n g / m l of partially purified brain fibroblast growth factor, 20 nM selenite, 1 /~M a-tocopherol, and antibiotics. The gas phase used was 5% 02, 90% N 2 and 5% CO 2. Subculturing was performed by incubation with Pronase E (neutral protease type XIV, Sigma) in serum-containing medium. For long term growth studies, adrenocortical cells were grown with a 1 : 5 split ratio at subcultures. Transfection was performed on tertiary (second passage) cultures. Construction of vector pRSV2-cSOD pRSV2-cSOD was constructed to place cDNA for human CuZn superoxide dismutase under the regulation of the Rous sarcoma virus long terminal repeat (LTR). pRSV is a variant of the eukaryotic expression vector pSV2 (Southern and Berg, 1982), utilizing the LTR from Rous sarcoma
virus (RSV), and was supplied by Dr. Cori Gorman, Genentech Inc., South San Francisco, CA. pRSV was chosen as an expression vector based on data showing the RSV-LTR to be an efficient promoter in many mammalian cell types (Gorman, 1985). In the pRSV plasmids a polylinker has been placed downstream of the LTR. pRSV2 contains the polylinker inserted in the orientation 5'-3' EcoRI-EcoRV. Additionally, pRSV contains the cDNA for mouse dihydrofolate reductase (DHFR) under the control of the SV40 promoter, thus providing for co-amplification, using the drug methotrexate, of D H F R and the gene introduced into the vector at the polylinker site (Wurm et al., 1986). Originally we had intended to use this feature to amplify the integrated DNA cSOD sequences in order to increase the expression of cSOD in the cell (Norris and Hornsby, 1988). However, because of the observed phenotype of transfected, non-amplified, cells, this feature could not be used in these experiments. cDNA for human SOD (Groner et al., 1985) was generously supplied by Dr. Yoram Groner, Weizmann Institute, Tel Aviv, in the pUC18 cloning vector, pUC18-cSOD was digested with the restriction endonucleases PvuII and BamHI and the 732 bp fragment containing the human SOD cDNA was inserted into the SmaI-BamHI sites of the polylinker of pRSV2 (Maniatis et al., 1982). Plasmids were purified through ethidium bromide-CsC1 gradient centdfugation before use for transfections. The purified plasmid DNA was dialyzed and reprecipitated with ethanol.
Transfection Tertiary cultures of bovine adrenocortical cells were used for the transfection experiments. One week prior to transfection, cells were incubated with protease (Pronase E) and transferred into 10 cm culture dishes at low density. Cultures were allowed to grow to - 7 5 % confluence before transfection. At this time, cultures are in vigorous growth. Cells ( 2 x 1 0 6 - 1 x lO 7) were removed from dishes with Pronase and resuspended in 0.6 ml transfection buffer (140 mM NaC1, 25 mM HEPES, 0.75 mM Na2HPO4, pH 7.2) containing 10 pg/ml linearized pRSV-cSOD plasmid DNA and 10 pg/nd pSV3neo (Neumann et al., 1982).
Plasmid pRSV2-cSOD or pRSV2 was linearized by digestion with NdeI and pSV3neo was linearized with EcoRI. Cells were incubated for 10 min at 0 ° C before transfer to the sample chamber of a Prototype Design Services Model ZA 1000 electroporation apparatus. They were electroporated with a 2.5 k V / c m electric field and kept on ice for an additional 10 min before being returned to culture medium. Cells were maintained in standard medium for 24 h before transfer into medium containing 200 # g / m l G418 (G-ibco Laboratories, Grand Island, NY) for selection of transfectants. Long term growth studies were all in medium containing 200 /~g/ml G418.
Isolation of DNA and Southern blotting DNA was prepared either by phenol/chloroform extraction or using a commercial anion exchange column (Extractor, Molecular Biosystems, San Diego, CA) designed for extracting nucleic acids from low numbers of cells. After restriction enzyme digestion, digests were fractionated on 0.8% agarose and DNA was transferred to nylon (Pall Biodyne Inc.) (Maniatis et al., 1982). Plasmid pUC18-cSOD was digested with PstI to create a 625 bp SOD cDNA probe, and pRSV2 was digested with StuI to create a 999 bp D H F R cDNA probe. The latter was used as a probe for pRSV2cSOD, outside of the SOD insert, that does not cross-react with other plasmid sequences in pSV3neo. Inserts were isolated using low melting temperature agarose (SeaKem, FMC Bioproducts) and labeled with 32p-dCTP using random oligonucleotide priming (Pharmacia kit). Membranes were hybridized with standard techniques (Maniatis et al., 1982). Following overnight hybridization, membranes were washed with 0.5 x SSC with 0.5% SDS and exposed to X-ray film for 6-14 days. Fluorescence observations of cells with DAPI (4,6diamidino-2 -phenylindole) For combined phase contrast/DAPI fluorescence observations, cells were plated in Petriperm dishes (Hereaus Inc., South Plainfield, N J) which have a thin, flexible, hydrophilic polytetrafluoroethylene bottom allowing for superior optics. After growth to the required cell density, cultures were fixed with 4% freshly made paraformaldehyde,
98
and were then dehydrated with acetone at - 2 0 °C for 5 rnin. Following the removal of the acetone, the dish was allowed to air dry. DAPI (Sigma Chemical Co.) was added as a 0.5 # g / m l solution in phosphate buffered saline with 5 mM Mg 2+ for 10 min. Cells were now visualized using an excitation wavelength of 365 um and an emission barher filter of 420 nm using a Nikon Diaphot epifluorescence microscope. Results Transfection of an expression vector for human CuZn-SOD into cultured bovine adrenocortical cells Plasmid pRSV2-cSOD, containing eDNA for human CuZn-SOD under the regulation of the Rous sarcoma virus LTR, was transfected into bovine adrenocortical cells by eleetroporation. In order to obtain selective growth of transfectants, cultures were cotransfected with pRSV2-cSOD and pSV3neo, a plasmid carrying the bacterial neo gene which confers resistance to the neomycin analog G418. pSV3neo also contains the early region of SV40 virus; SV40 T antigen has been shown in previous experiments to cause extended, vigorous growth of bovine adrenocortical cells (Cheng et al., 1989). In preliminary experiments, cotransfection with pSV2neo, which is similar to pSV3neo but lacks the SV40 early region encoding T antigen, produced cells that senesced soon after transfeetion and thus could not be used for these experiments. This indicates that T antigen may counteract residual G418 toxicity which may otherwise limit the growth of transfectants, since normal non-transfeeted clones have a much greater growth potential than pSV2neo transfectants (Cheng et al., 1989). After transfection, cells were replated at moderate cell density in 6 cm dishes in medium with G418. After 13 days, a large number of G418 resistant colonies were observed. All the colonies were pooled and were subcultured at a 1 : 2.5 split ratio into 10 cm dishes. When these cultures were confluent, 90% of the cells were frozen in liquid nitrogen and stored for future experiments. Southern blot analysis of DNA isolated from transfectants 2 passages after pooling and freezing of the polyclonal population was used to assess the state of integration of the plasmid sequences (Fig. 1). After growth of the cell population to a
A
A~" 0
0
> u) ~" ¢~.
> u~ ¢E ¢~.
o o
o
~ ¢~.
o
o
? ? O m I-
>,
~,
--23 --23 --9.4 --6.5 --4.3
--9.4 --6.5 --4.3
--2.3 --2.0
--2.3 --2.0
--1.3 --0.56
--1.0
Fig. 1. Detection of human SOD cDNA by Southern blot analysis in DNA from bovine adrenocortical cells transfected with pRSV2-cSOD. (A) DNA from the following32celis was isolated, restricted with PstI, and probed with P-labeled human SOD cDNA. TBAC: tertiary (second passage) cultured bovine adrenocortical cells; normal cells that were used for transfection, pSV3neo: bovine adrenocortical cells transfected with pSV3neo only. pRSV2-cSOD (2): bovine adrenocortical cells cotransfected with pRSV2-cSOD and pSV3neo, passage 2 after transfection, pRSV2-cSOD (25): bovine adrenocortical cells cotransfected with pRSV2-cSOD and pSV3neo, passage 25 after transfection. The arrow indicates the position of the expected 625 bp human SOD cDNA. Positions of HindIII digested ~, size markers are indicated. (B) The blots from A were reprobed with riP-labeled DHFR from pRSV2. Lane identifiers are the same as for the SOD hybridization. The arrow indicates the position of the expected 2.9 kb PstI fragment of pRSV2-cSOD. Exposure has been adjusted to equalize the intensity of the endogenous band at - 6.0 kb.
sufficient size, DNA was prepared and restricted with PstI, which excises human SOD cDNA from pRSV2-cSOD. Hybridization with the PstI cDNA insert from pUC18-cSOD showed the presence of the expected 625 bp band in the transfectants. Presumably because of the high degree of homology between human and bovine CuZn-SOD
(Steinman et al., 1974; Sherman et al., 1983) the human SOD cDNA probe hybridized strongly to endogenous bovine SOD sequences. Transfectants were grown at low cell density for 25 passages, using the protocol described later. At this point, DNA was isolated from cultures and reanalyzed for the presence of human SOD cDNA. The expected 625 bp band was not observed, but an additional larger fragment of ~ 5.5 kb, not detected either in control, non-transfected cells or in early passage transfectants, was observed to hybridize with the human SOD probe. Autoradiogram~ were analyzed by scanning densitometry. By using the endogenous bovine SOD gene sequences as internal standards, it was confirmed that the 625 bp band was completely absent in 25th passage ceils and that the new - 5 . 5 kb band was not detectable in passage 2 cells or in non-transfectants. Moreover, it was clear that, in both early passage and late passage transfectants, the SOD eDNA was present at less than 1 copy per cell, although the precise copy number cannot be determined because the degree of cross-reaction of the probe with endogenous SOD sequences is not known. Thus, although all ceils presumably successfufiy integrated pSV3neo, since cells were G418 resistant, only a minority of ceils received pRSV2-cSOD. This is consistent with observations on the population behavior of the transfectants, described below, which suggest that a minority of ceils differed in phenotype from the remainder of the population. The lack of the expected cSOD band in late passage transfectants was not due to a complete lack of plasmid sequences in these cells. When blots were rehybddized with 32p-labeled DHFR, to serve as a probe for a different region of the pRSV2 plasmid, the expected 2.9 kb plasmid band was present in pRSV2-cSOD/pSV3neo cotransfectants (Fig. 1). Endogenous bovine sequences cross-hybridized less strongly with the probe in this case. The 2.9 kb band was present in transfectant DNA from cells at both 2 and 25 passages, but not in the normal cells. Growth of transfectants and observation of loci of cell death The polyclonal population of pRSV2-cSOD/ pSV3neo cotransfected cells was grown in medium
with G418 at 1 : 5 sprit ratios. Three passages after isolation of the polyclonal population, as cells grew to confluence, loci of rounded, dying cells were observed. This phenomenon was noted reproducibly at this passage, when vials of cells frozen at the isolation of the polyclonal population were replated and grown in culture. Cell death began at a focal point and spread outward, affecting neighboring cells within 24 h. At 24 or 48 h after initiation of these foci, cells were seen in various stages of rounding up and detachment from the culture dish. By 72 h, < 10% of cells still remained attached to the culture dish (Fig. 2). In several experiments, conducted either separately or simultaneously with the transfection with pRSV2-cSOD, cells transfected with pSV3neo alone, or cells cotransfected with pSV3neo and pRSV2 (the parent plasmid lacking the SOD cDNA insert), never showed any evidence of focal cell death, at any passage level. The number of foci of dying cells per dish of transfectants suggested that the cell death was initiated by a minority of cells in the population. In order to determine what fraction of the cell population was capable of initiating foci, pRSV2cSOD/pSV3neo cotransfectants frozen from the polyclonal population were seeded at low density in a 96 well plate ( < 10 cells per well). The cells were allowed to grow in medium with G418. After 12 days, each well was observed for presence of foci of cell death. The numbers of wells with and without cell death were tabulated (Table 1). This experiment gives a value of - 24% for the upper limit of the fraction of cells that can initiate foci. The actual value is likely to be much less, since most wells were seeded with more than one cell. pRSV2-cSOD/pSV3neo cotransfectants were grown under various culture conditions in order to test possible mechanisms for the cell death phenomenon and to assess the feasibility of successful growth of pure cultures of these cells in order to allow for further analysis. Factors examined included omission or inclusion of G418; omission of serum; inclusion of a combination of growth factors (UltroSer G) previously found to enhance the growth of bovine adrenocortical cells (McAllister and Hornsby, 1987); and replacement of serum with ether-extracted serum (Hornsby et al., 1985a) to assess the possible role of serum lipids. None of
100
g~: .Z:,-.., ..... . . . . . . . . . . . . . . . .
,,~,,~,~ ,~ ........ ~ , ~ . , ~ , , ~
Fig. 2, Cell death in cultures of bovine adrenocortical cells cotransfected with pRSV2-cSOD and pSV3neo. (A) Culture of cells 3 passages following transfection, just after reaching confluence, (B) Culture 24 h later, showing evidence of initiation of loci of cell death. (C) Culture 48 h later, showing widening of loci, some cells having detached from the substratum. (D) Culture 72 h later, showing effects of extensive spread of cell death and almost complete detachment of cells. Phase contrast micrographs × 50.
101 TABLE 1 ANALYSIS OF FRACTION OF pRSV2-cSOD/pSV3neo TRANSFECTED CELL POPULATION ABLE TO FORM FOCI OF CELL DEATH Exp. 1 Number of wells with cell death foci Confluent Non-confluent Number of wells without cell death foci Confluent Non-confluent Total number of wells Total number of wells with cell death loci U p p ~ limit to fraction of cells in ~hc population able to i~tia~c loci of cell d ~ t h :
Exp. 2
9 7
6 3
15 37 68
2 24 35
16
6
16 = 23.5% ~ =17.1% b~g
p RSV2-cSOD/pSV3neo transfected bovine adrenocortical cells were plated in 96 well plates at low density ( < I0 cells per well). After 10-12 days, when msny wells had reached confluence, cells were fixed, stained, and egamined for numbers of fO~i of cell death.
_
_ _ ~ , _ _
these conditions affected the incidence or severity of'cell death in the transfectant cultures. The observation that cell death began at a focal point and spread outward to neighboring cells suggested that the initiating cells might release a toxic factor or transmissible agent (e.g., virus) into the medium. In order to check for such a factor, medium was removed from a plate of transfectants with extensive cell death, filtered through a 0.22 # m filter, diluted 1 : 1 with fresh medium, and added to a - 80% confluent plate of control (pSV3neo transfected) cells. This addition was repeated for several days, without evidence of cell death in the recipient cells. These data appear to exclude the involvement of (i) a diffusible toxic factor, unless it is too labile for transfer to a target cell population, and (ii) a transmissible agent, or at least an agent that passes through a 0.22 #m filter. Became production of hydrogen peroxide might be elevated in cells expressing high levels of SOD, catalase (100 # g / m l ) was tested as a possible inhibitor of cell death, but was without effect. It
T - -
Fig. 3. Cell death and multinucleation in cultures of pRSV2-cSOD/pSV3neo cotransfected bovine adrenocortical cells during growth to higher cell density. Transfectants were grown with maintenance of < 20% confluence for 6 passages after isolation of the polyclonal cell population, and were then allowed to grow back from this low cell density to higher density over a period of 10 days. (Left) A group of cells in various stages of rounding and detachment from the dish. (Right) Several cells with massive multinucleation adjacent to a focus of cell death. Phase contrast micrographs × 50.
102
was also thought that intracellular production of hydrogen peroxide might occur. Because this would not be accessible to extracellular scavenging enzymes, we investigated the possible protective effects of dimethyl sulfoxide, which we had previously shown to be an effective radioprotectant in these cells (Hornsby et al., 1984). The radioprotectant effect results from scavenging of the hydroxyl radical, the probable final agent of toxicity of hydrogen peroxide (Floyd, 1981). We noted that 2% dimethyl sulfoxide caused a slowing of growth such that cultures never became completely confluent. Under these circumstances, it appeared that scattered individual cells died but spreading of death to neighboring cells was pre-
vented. Although it is possible that OH" scavenging by dimethyl sulfoxide was involved in prevention of the spread of cell death, it seemed more likely that its major effect was on cell growth, indicating that the spreading of cell death might be dependent on cell density. Combined with evidence that a transmissible toxic factor could not be demonstrated, this suggested that the factor may not be freely diffusible, requiring cell contact for the spread of cell death. Growth of pRSV2-cSOD/pSV3neo cotransfectants under conditions in which < 20% confluence was maintained by appropriate timing of subcultures, thus preventing cultures from achieving confluence, showed that extensive cell death
Fig. 4. Phase contrast appearance and DAPI fluorescence of p R S V 2 - c S O D / p S V 3 n e o cotransfectants during multinucleation and nuclear fragmentation. Brackets on the phase contrast images indicate the area of the corresponding fluorescence image. Sixth passage transfectants were grown as described in Fig. 3 and were fixed and stained with DAPI. ( A - C ) Examples of multinucleated cells. ( D - G ) Cells in various stages of nuclear fragmentation and cell detachment. Magnification × 600.
103 could in fact be avoided. Cells-were.grown for 25 passages ( - 63 population doublings) under these conditions. However, as shown in Fig. 1, Southern blot analysis of DNA from 25th passage transfectants showed an absence of cells carrying the expected 625 bp SOD cDNA, suggesting that such cells had been lost from the population despite the maintenance of low cell density. Apparently, growth at low cell density did not prevent the death of pRSV2-cSOD transfected cells, but
spreading of cell death to other cells in the population was avoided.
Multinucleation, nuclear fragmentation, and cell death During low density growth of pRSV2cSOD/pSV3neo cotransfectants, continuous proliferation was observed, although, especially in later passages, abnormal numbers of cells in metaphase were noted. Foci of cell death were not
Fig. 4 (continued).
104
observed. We examined whether cell death would reappear when low density cultures were allowed to grow back to high cell density. As low density cultures grew to confluence, foci of cell death appeared (Fig. 3). Associated with cell death foci were giant cells with massive multinucleation (Fig. 3). Cultures containing giant cells with various stages of multinucleation and cell death were fixed and stained with a fluorescent DNA binding dye, DAPI, for observation of nuclear morphology. Cells surrounding the giant cells had nuclei that were smooth in outline and regular in shape and size. In multinucleated cells, a variable number of nuclei were clustered in the center of the cell, sometimes appearing to be in the process of fusion into an irregularly shaped mass (Fig. 4A-C). Nuclear number varied continuously in the range of 2 to - 6 0 . In later stages of this process, the large nuclear mass became fragmented, with nuclear material becoming condensed into brightly fluorescent spots (Fig. 4D-G). The phase contrast appearance of such cells during nuclear fragmentation showed withdrawal of contact from the dish and eventual detachment. Adjacent to such multinucleated, detaching, cells were individual dying cells with fragmented and condensed nuclear material. Multinucleation and cell death were observed in cultures of transfectants at early passages following isolation of the transfected population. However, at later passages, cell death and mnltinucleation were not observed when cells were allowed to grow back to high cell density. This correlates with the absence of cells with the expected 625 bp SOD cDNA band observed by Southern blotting in later passage cells (Fig. 1). Discussion
Transfection of an expression vector containing human CuZn-SOD cDNA did not appear to increase the resistance of a normal differentiated cell type, the bovine adrenocortical cell, to oxidative damage. The introduction of this vector caused a cytotoxic effect involving cell death and nuclear fragmentation. Because cell death prevented the growth of large numbers of transfected cells, it was not possible to demonstrate the involvement
of CuZn-SOD in the cytotoxic effect by direct means, but a control plasmid, identical with the exception that it lacked the CuZn-SOD cDNA insert, had no such effect. This suggests that the CuZn-SOD cDNA may have contributed to increased, rather than decreased, oxidative damage in the adrenocortical cell. Mouse cells overexpressing human CuZn-SOD, although more resistant to paraquat, were reported to show an increase in thiobarbituric acid products in homogenates of the cultured cells, suggesting an increased potential for lipid peroxidation (ElroyStein et al., 1986). In extensions of the reported work on SOD transfected mouse cells, in transgenic mice, PC12 adrenomedullary cells, and NS20Y neuroblastoma cells, there were additional suggestions of toxic effects of excess CuZn-SOD (Epstein et al., 1987; Elroy-Stein and Groner, 1988; Cebailos et al., 1988). However, no direct cytotoxic effects of excess CuZn-SOD were reported. Cytotoxic effects would have been difficult to observe directly because cells suffering cytotoxic effects would not have survived the initial clonal cell selection. In fact, because transfectants with more than a 6-fold elevation of CuZn-SOD activity were not found, it was speculated that such transfectants might have been lost due to a toxic effect of overexpression (Elroy-Stein et al., 1986). In the experiments reported here, no attempt was made to clone out the SOD cDNA transfected cells from the initial mixed population, comprising both cells that received only the selectable plasmid, pSV3neo, and cells that were successfully cotransfected with pSV3neo and pRSV2-cSOD. Southern blot analysis indicated that the cotransfection efficiency, i.e., the fraction of cells that received both selectable and nonselectable plasmids, was low. This may have been due to the use of electroporation, which may produce less concatemerization of plasmids than calcium phosphate transfection methods (Andreason and Evans, 1988). A greater fraction of cotransfectants in the population would evidently have resulted in much more extensive cell death, perhaps to the extent that a transfected population would not have been isolatable. Cotransfecrants did survive long enough to exert their cytotoxic effect at the third passage after isolation of
105
the polyclonal population. It is not clear why there was a delay in the expression of the cytotoxic effect. Once the cytotoxic effect was expressed, however, it spread to neighboring cells at confluence, eventually resulting in the death of the entire culture. Even when the spread of cell death was prevented by growth of cells at low cell density, the transfectants carrying the expected 625 bp SOD cDNA fragment did not survive long term growth, as shown by Southern blotting after 25 passages. Presumably, these cells either died or were diluted out of the population as the cultures were passed at low density. This indicates that the presence of the 625 bp SOD insert did not give the cells a growth advantage. In late passage transfectants, only an abnormal larger band hybridizing with the SOD probe was observed, probably indicating that the subgroup of transfectants that were able to survive long term growth received only a truncated SOD cDNA sequence during chromosomal integration. However, these ceils still carried other portions of the pRSV2-cSOD plasmid. Presumably, only cells containing the cSOD 625 bp insert initiated the ceil death loci. The non-survival of transfectants carrying the 625 bp cDNA made it impossible to isolate sufficient of these cells for analysis of human CuZn-SOD expression. The large multinucleated cells observed here probably resulted from fusion of individual ceils to multinucleated giant cells. Infection with many viruses results in the creation of large multinucleated cells through a process of cell fusion (Poste, 1970). The subsequent fusion of nuclei to form the irregularly shaped masses observed here presumably results in the death of the multinucleated ceil, with the formation of a focus of cell death affecting adjacent ceils. The close resemblance of the cytotoxic effect to that of a viral infection raises the possibility that introduction of pRSV2-cSOD into cells activated a latent virus or rendered the ceils more susceptible to infection by viruses normally present in the environment, although the occurrence of the cytotoxic effect in a variety of different media argues against the latter possibility. Currently, the sequence of molecular events between integration of pRSV2-cSOD and ceil fusion/ceil death is unknown. Possibly, excess cellu-
lax SOD can act as a peroxidase; SOD can catalyze the peroxidation of linoleic acid in vitro (Hodgson and Fridovich, 1975). Transfected SOD could be expressed in a form that has an altered subcellulax distribution, which might cause a change in the properties of the enzyme. Alternatively, excessive hydrogen peroxide formation could cause increased cellular damage. Presumably, by processes that remain to be elucidated, peroxidative damage resulting from excess SOD expression then causes cell fusion and the subsequent propagation of cell death.
Acknowledgement This work was supported by a grant from the Greenwall Foundation to Dr. J.E. Seegmiller, Institute for Research on Aging, University of California, San Diego, CA.
References Andreason, G.L., and G.A. Evans (1988) Introduction and expression of DNA molecules in eukaryotic cells by electroporatiun, BioTeclmiques, 6, 650-660. Ceballos, I., J.M. Delabar, A. Nicole, R.E. Lynch, R.A. Hallewell, P. Kamouss and P.M. Sinet (1988) Expression of transfected human Cu-Zn SOD gene in mouse L ceils and NS20Y neuroblastoma cells induces enhancement of glutathione peroxidase activity, Biochim. Binphys. Acta, 949, 5864-5870. Cerutti, P. (1985) Prooxidant states and tumor promotion, Sc/ence, 227, 375-381. Cheng, C.Y., R.F. Ryan, T.P. Vo and P.J. Hornsby (1989) Cellular senescence involves stochastic processes causing loss of expression of differentiated function genes: transfection with SV40 as a means for dissociating effects of senescence on growth and on differentiated function gene expression, Exp. Cell Res., 180, 49-62. Elroy-Stein, O., and Y. Groner (1988) Impaired neurotransmitter uptake in PC12 cells overexpressing human C u / Z n superoxide dismutase - impfication for gene dosage effects in Down syndrome, Cell, 52, 259-267. Elroy-Stein, O., Y. Bernstein and Y. Groner (1986) Overproduction of human Cu/Zn-superoxide dismutase in transfected cells: extenuation of paraquat-mediated cytotoxicity and enhancement of lipid peroxidation, EMBO J., 5, 615622. Epstein, C.J., K.B. Avrahan~ M. Lovett, S. Smith, O. ElroyStein, G. Rotman, C. Bry and Y. Groner (1987) Transgenic mice with increased Cu/Zn-superoxide dismutase activity: animal model of dosage effects in Down syndrome, Proc. Natl. Acad. Sci. (U.S.A.), 84, 8044-8048.
106 Floyd, R.A. (1981) DNA-ferrous iron catalyzed hydroxyl free radical formation from hydrogen peroxide, Biochem. Biophys. Res. Commun., 99, 1209-1215. Freeman, B.A., S.L. Young and J.D. Crapo (1983) Liposomemediated augmentation of superoxide dismutase in endothelial cells prevents oxygen injury, J. Biol. Chem., 258, 12534-12542. Fridovich, I. (1983) Superoxide radical: an endogenous toxicant, Annu. Rev. Pharmacol. Toxicol., 23, 239-257. Gorman, C. (1985) High efficiency gene transfer into mammarian cells, in: D.M. Glover (Ed.), DNA Cloning, Vol. 2, IRL Press, Oxford, pp. 143-190. Gospodarowicz, D., C.R. Ill, P.J. Hornsby and G.N. Gill (1977) Control of bovine adrenal cortical cell proliferation by fibroblast growth factor. Lack of effect of epidermal growth factor, Endocrinology, 100, 1080-1089. Groner, Y., J. Lieman-Hurwitz, N. Dafni, L. Sherman, D. Levanon, Y. Bernstein, E. Danciger and O. Elroy-Stein (1985) Molecular structure and expression of the gene locus on chromosome 21 encoding the C u / Z n superoxide dismutase and its relevance to Down syndrome, Ann. N.Y. Acad. Sci., 450, 133-156. Hodgson, E.K., and I. Fridovich (1975) The interaction of bovine erythrocyte superoxide dismutase with hydrogen peroxide: chemiluminescence and peroxidation, Biochemistry, 14, 5299-5303. Hornsby, P.J. (1980) Regulation of cytochrome P-450-supported llfl-hydroxylation of deoxycortisol by steroids, oxygen, and antioxidants in adrenocortical cell cultures, J. Biol. Chem., 255, 4020-4027. Hornsby, P.J., and S.E. Harris (1987) Oxidative damage to DNA and replicative life span in cultured adrenocortical cells, Exp. Cell Res., 168, 203-217. Hornsby, P.J., S.E. Harris and K.A. Aldern (1984) Mode of action of sulfoxides in preventing loss of activity of llfl-hydroxylase in cultured bovine adrenocortical cells, Chem.Biol. Interact., 51, 335-346. Hornsby, P.J., K.A. Aldern and S.E. Harris (1985a) Mode of action of butylated hydroxyanisole (BHA) and other phenols in preventing loss of 11fl-hydroxylase activity in cultured bovine adrenocortical cells, Biochem. Pharmacol., 34, 865-872. Hornsby, P.J., D.W. Pearson, A.P. Autor, K.A. Aldern and S.E. Harris (1985b) Selenium deficiency in cultured adrenocortical cells: restoration of glutathione peroxidase and resistance to hydroperoxides on addition of selenium, J. Cell. Physiol., 123, 33-38. Hornsby, P.J., R.F. Ryan and C.Y. Cheng (1989) Replicative
senescence and differentiated gene expression in cultured adrenocortical cells, Exp. Gerontol., 24, 539-558. Hornsby, P.J., C.Y. Cheng, R.F. Ryan and L. Yang (1990) Stochastic changes in gene expression in adrenal cell senescence, in: C.E. Finch and T.E. Johnson (Eds.), Molecular Biology of Aging, Wiley-Liss, New York, NY, pp. 249-263. Imlay, J.A., and S. Lima (1988) DNA damage and oxygen radical toxicity, Science, 240, 1302-1309. Maniatis, T., E.F. Fritsch and J. Sambrook (1982) Molecular Cloning: A Laboratory Manual, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY. McAilister, J.M., and P.J. Hornsby (1987) Improved clonal and non-clonal growth of human, rat, and bovine adrenocortical cells in culture, In Vitro Cell. Dev. Biol., 23, 677-685. Neumann, E., M. Schaefer-Ridder, Y. Wang and P.H. Hofschneider (1982) Gene transfer into mouse lyoma cells by electroporation in high electric fields, EMBO J., 1, 841-845. Norris, K., and P.J. Hornsby (1988) Oxidative damage to DNA in adrenocortical cells during senescence in culture, in: M.G. Simic, K.A. Taylor, J.F. Ward and C. von Sormtag (Eds.), Oxygen Radicals in Biology and Medicine (Basic Life Sciences, Vol. 49), Plenum, New York, NY, pp. 461466. Poste, G. (1970) Virus-induced polykaryocytosis and the mechanism of cell fusion, Adv. Virus Res., 16, 303-356. Sherman, L., N. Dafni, J. Lieman-Hurwitz and Y. Groner (1983) Nucleotide sequence and expression of human chromosome 21-encoded superoxide dismutase mRNA, Proc. Natl. Acad. Sei. (U.S.A.), 80, 5465-5469. Southern, P.J., and P. Berg (1982) Transformation of mammalian cells to antibiotic resistance with a bacterial gene under control of the SV40 early region promoter, J. Mol. Appl. Genet., 1, 327-341. Steinman, H.M., V.R. Naik, J.L. Abernethy and R.L. Hill (1974) Bovine erythrocyte superoxide dismutase. Complete amino acid sequence, J. Biol. Chem., 249, 7326-7338. Taylor, W.G. (1984) Toxicity and hazards to successful culture: cellular responses to damage induced by light, oxygen or heavy metals, in: M.K. Patterson Jr. (Ed.), Uses and Standardization of Vertebrate Cell Cultures, Tissue Culture Association, Galthersburg, MD, pp. 58-70. Vuillaume, M. (1987) Reduced oxygen species, mutation, induction and cancer initiation, Mutadon Res., 186, 43-72. Wurm, F.M., K.A. Gwirm and R.E. Kingston (1986) Inducible overproduction of the mouse c-myc protein in mammalian cells, Proc. Nat. Acad. Sei. (U.S.A.), 83, 5414-5418.
