Mutagenicity studies of methyl-tert-butylether using the Ames tester strain TA102.

Methyl-tert-butylether (MTBE) is an oxygenate widely used in the United States as a motor vehicle fuel additive to reduce emissions and as an octane booster [National Research Council, Toxicological and Performance Aspects of Oxygenated Motor Vehicle Fules, National Academy Press, Washington, DC, 1996]. But it is the potential for MTBE to enter drinking water supplies that has become an area of public concern. MTBE has been shown to induce liver and kidney tumors in rodents but the biochemical process leading to carcinogenesis is unknown. MTBE was previously shown to be non-mutagenic in the standard Ames plate incorporation test with tester strains that detect frame shift (TA98) and point mutations (TA100) and in a suspension assay using TA104, a strain that detects oxidative damage, suggesting a non-genotoxic mechanism accounts for its carcinogenic potential. These strains are deficient in excision repair due to deletion of the uvrB gene. We hypothesized that the carcinogenic activity of MTBE may be dependent upon a functional excision repair system that attempts to remove alkyl adducts and/or oxidative base damage caused by direct interaction of MTBE with DNA or by its metabolites, formaldehyde and tert-butyl alcohol (TBA), established carcinogens that are mutagenic in some Ames strains. To test our hypothesis, the genotoxicity of MTBE-induced DNA alterations was assayed using the standard Ames test with TA102, a strain similar to TA104 in the damage it detects but uvrB + and, therefore, excision repair proficient. The assay was performed (1) with and without Aroclor-induced rat S-9, (2) with and without the addition of formaldehyde dehydrogenase (FDH), and (3) with human S-9 homogenate. MTBE was weakly mutagenic when tested directly and moderately mutagenic with S-9 activation producing between 80 and 200 TA102 revertants/mg of compound. Mutagenicity was inhibited 25%-30% by FDH. TA102 revertants were also induced by TBA and by MTBE when human S-9 was substituted for rat S-9. We conclude that MTBE and its metabolites induce a mutagenic pathway involving oxidation of DNA bases and an intact repair system. These data are significant in view of the controversy surrounding public safety and the environmental release of MTBE and similar fuel additives.

Mutation induction and RBE of low energy neutrons in V79 cells.

We have examined the neutron energy dependency of cell killing and mutation induction at the hprt locus in Chinese hamster V79 cells. Monoenergetic neutrons at 0.32, 0.57, and 1.2 MeV were generated at the Hiroshima University Radiobiological Research Accelerator (HIRRAC) Facility, and were used to irradiate cells. The variation in RBE with neutron energy for the end points of cell survival and hprt mutation induction was observed. When compared to 137Cs gamma-rays, all neutron energies were more effective at both cell killing and induction of mutation. Over the range of the neutron energies examined, we found that cytotoxicity increased as the energy decreased from 1.2 to 0.32 MeV. In comparison to gamma-rays, RBEs for cell lethality at 10% survival were 5.7, 6.7, and 7.6 for 1.2, 0.57, and 0.32 MeV, respectively. Mutation induction, on the other hand, was highest at 0.57 MeV with a gradual decrease at 1.2 and 0.32 MeV. RBEs for mutation induction were 9.7, 19.4, and 13.9 for 1.2, 0.57, and 0.32 MeV neutrons. We isolated independent V79 cell mutants at the hprt locus from untreated and neutron-exposed cells and determined the genetic changes underlying the mutation by multiplex polymerase chain reaction (PCR)-based exon deletion analysis. Preliminary results are suggestive of a specific relationship between deletion pattern and neutron energy.

Radiobiological studies using synchrotron-produced ultrasoft X-rays.

Ultrasoft X-rays have been extensively used to explore radiobiological mechanisms surrounding cell killing. These studies for the most part have been linked to a small number of X-ray energies. Recently, this field of study has been broadened by the availability of synchrotron-produced ultrasoft X-rays which can be produced at any desired energy. We have taken advantage of the University of Wisconsin Synchrotron to reexamine two fundamental radiobiological questions: Dose RBE vary with different ultrasoft X-ray energies? Dose the fraction of the nuclear volume exposed to equal total X-ray energy modify cell cytotoxicity? The first study focuses on the survival of Chinese hamster V79 and mouse C3H10T1/2 cells irradiated with synchrotron-produced 273 eV and 860 eV ultrasoft X-rays. These two energies, which are available by multilayer monochromatization of the synchrotron output spectrum, exhibit equal attenuation within living cells. Such an isoattenuating energy pair allows the direct examination of how biological effectiveness varies with the energy of the ultrasoft X-rays. In comparing survival results, we find similar biological effectiveness of these two energies for both the C3H10T1/2 and the V79 cells. These results are no consistent with previous findings of increasing RBE with decreasing ultrasoft X-ray energies. In addition, after correcting for mean nuclear based on measurements of cell thickness obtained with confocal microscopy, we find no significant differences in survival between the two ultrasoft X-ray energies and 250 kVp X-rays. These results suggest that RBE does not increase with decreasing energy of ultrasoft X-ray between 860 eV and 273 eV. In a second study we introduced an method which allows partial-volume irradiation of live cells using synchrotron-produced ultrasoft X-rays and micro-fabricated irradiation masks. The masks were made by X-ray lithography at the University of Wisconsin Synchrotron Radiation Center, and they consist of 1.85-micron-wide stripes of gold 1.35 microns apart plated onto thin silicon nitrate membranes. When placed adjacent to mylar on which live cells are plated, these masks allow cells to be irradiated in a striped pattern with dimensions much smaller than the cell nuclei. Using 1340 eV synchrotron-produced X-rays, we compare the survival of cells subjected to uniform irradiation and cells subjected to partial-volume irradiation. Our results show that, at equal mean dose to the nucleus (i.e. equal total energies deposited), survival is not statistically different for the two treatments over a wide range of doses. Thus, imparting equal energies to smaller intranuclear volumes does not appear to modulate cell killing.

Neutron carcinogenesis: past, present, and future.

An interest in the possible cancer causing ability of neutrons began soon after their discovery. Early use of neutrons from radioactive sources and from cyclotrons led to a need to define risk for such exposures. This need was soon followed by a more tangible need to define risk to the general population of high LET radiation from nuclear fall out and use of the Atomic bomb and possible use of the H-bomb. Neutrons were soon found to be very effective cell killing agents compared to conventional ionizing radiation. However High LET radiation sources and neutrons in particular, come in many different energies and from many types of sources. I will survey the differences between different energy neutrons and conventional types of radiation, particularly with respect to the dose rate of exposures and the influence of repair or lack thereof and more recently the effect of cell cycle distribution on the carcinogenic outcome. I will illustrate these ideas with examples of carcinogenicity studies and mutation studies from my own laboratory and in some cases from the work of others. Lastly I will introduce some possible avenues for molecular studies of neutron effects that might answer such vexing questions as the real risk at very low doses, is repair error free or error prone, do neutrons cause genetic instability for many cell generations after exposure, and others? There remain many questions about the biology of neutron action that require answers if we are to protect the ever increasing number of people exposed to them because of their growing use in medicine, in the military and in commercial industry.

A comparison of cytotoxicity after whole- or partial-cell irradiation with synchrotron-produced ultrasoft X rays.

We introduce a method which allows partial-volume irradiation of live cells using synchrotron-produced ultrasoft X rays and micro-fabricated irradiation masks. The masks were made by X-ray lithography at the University of Wisconsin Synchrotron Radiation Center, and they consist of 1.85-microm-wide stripes of gold 1.35 microm apart plated onto thin silicon nitride membranes. When placed adjacent to Mylar on which live cells are plated, these masks allow cells to be irradiated in a striped pattern with dimensions much smaller than the cell nuclei. Using 1340 eV synchrotron-produced X rays, we compare the survival of cells subjected to uniform irradiation and cells subjected to partial-volume irradiation. Our results show that, at equal mean dose to the nucleus (i.e. equal total energies deposited), survival is not statistically different for the two treatments over a wide range of doses. Thus imparting equal energies to smaller intranuclear volumes does not appear to enhance cell killing.

Synchroton-produced ultrasoft X rays: equivalent cell survival at the isoattenuating energies 273 eV and 860 eV.

In this paper we report on survival of Chinese hamster V79 and mouse C3H 10T1/2 cells after irradiation with synchrotron-produced 273 eV and 860 eV ultrasoft X rays. These two energies, which are available by multilayer monochromatization of the synchrotron output spectrum, exhibit equal attenuation within living cells. Such an isoattenuating energy pair allows the direct examination of how biological effectiveness varies with the energy of the ultrasoft X rays. In comparing survival results, we find similar biological effectiveness of these two energies for both the C3H 10T1/2 and the V79 cells. These results are not consistent with previous findings of increasing RBE with decreasing ultrasoft X-ray energies. In addition, after correcting for mean nuclear dose based on measurements of cell thickness obtained with confocal microscopy, we find no significant differences in survival between the two ultrasoft X-ray energies and 250 kVp X rays. These results suggest that RBE does not increase with decreasing energy of ultrasoft X rays between 860 eV and 273 eV. The possible impact of our results on past results for ultrasoft X rays is discussed.

Phase transitions in the growth of C3H 10T1/2 cells.

In systems used to express transformation using focus formation as the end point, nontransformed cells generally express a down-regulation of cell growth and division made evident by the formation of a monolayer of cells that completely covers the growth surface. In C3H 10T1/2 cells, down-regulation is thought to be progressively effected principally by cell-to-cell communication via gap junctions. Starting with a sparse population in asynchronous growth--e.g. containing cells in all phases of the growth cycle--as the area density increases, cells are progressively lost from the distribution in the order M phase, G2 phase, S phase and G1 phase, leading to the accumulation of viable cells out of cycle in so-called G0 phase. We have measured the progressive phase transitions as a function of inoculum size and time. The influence of a promoter and an antipromoter was also examined as well as the expression of the cyclin/cyclin-dependent kinase inhibitors p21Waf1/Cip1 and p27Kip1 as the cells grew into confluence. Using cells synchronized in mitosis, we found that with increasing cell density the expression of p27 increased and concomitantly p21 decreased.

Dependence of mutation induction on fast-neutron energy in a human epithelial teratocarcinoma cell line (P3).

To date, few studies have evaluated the magnitude of the risks of somatic effects in humans from low-dose or protracted radiation exposure to neutrons using in vitro or in vivo techniques (A. Kronenberg, Radiat. Res. 128, S87-S93, 1991). In earlier study a strong energy dependence was shown for neutron-induced mutations at both the hprt and the tk loci in a rodent fibroblast cell line (Zhu and Hill, Radiat. Res. 139, 300-306, 1994). Using fast neutrons produced by impinging protons on a beryllium target at the UCLA/VA cyclotron, we have been examining the energy dependence of mutation induction at the HPRT locus in a human epithelial cell line derived from solid tumor tissue. In the present study, human epithelial teratocarcinoma cells were exposed to neutron beams produced from protons with 46, 30, 20 and 14 MeV energy. We found that cytotoxicity increased by 50% as the neutron beams produced from 46 MeV to 14 MeV, confirming many earlier reports. But as with the Chinese hamster cells, the mutation frequency at the HPRT locus increased 2.5-4-fold with decreasing neutron energy. Additionally, although there was a strong energy dependence for mutation induction, we noted that the shape of the induction curves was curvilinear for the human cells compared to the linearity of the curves obtained for the Chinese hamster cells and some other non-solid tissue human cell lines. Calculations of the RBE, using gamma rays as the standard reflected these differences. The RBE for mutation at the HPRT locus was dependent not only upon energy but also on dose, giving rise to RBEs that were in some cases distinctly different from those found in the Chinese hamster cell line. In the low-dose region (doses below 75 cGy) the maximum RBE of about 5 resulted from irradiation by the lowest-energy neutron beam (14 MeV protons on beryllium).

Increased radiosensitivity of normal tissue fibroblasts in patients with acquired immunodeficiency syndrome (AIDS) and with Kaposi's sarcoma.

Fibroblasts cultured from skin biopsies of patients with AIDS and Kaposi's sarcoma were found to be more radiosensitive than fibroblasts from non-HIV-infected-sources. This supports clinical observations of overt sensitivity to radiotherapy in some AIDS patients with Kaposi's sarcoma.

Characterization of a mammalian cell line that exhibits spontaneous and ultraviolet light-induced hypermutability while retaining resistance to cell killing by ultraviolet light.

Chronic exposure of V79 cells to 80 daily doses of 150 J/M2, 290-330-nm ultraviolet light (UVB) produced a mixed cell population that was found to be generally more resistant to cell killing by both UVB and UVC (254 nm) than the wild-type cells. Several subclones from this population were studied for their survival and mutation responses and then one was chosen for further characterization based on this data. The studies carried out on this subclone, designated N806, show that its spontaneous HPRT mutation rate is approximately 10 times higher than that of wild-type V79 cells and it is almost three times more mutable than the wild-type cells when both are induced by UVB or UVC. The mutation responses of N806 and MI2G cells to 50-kVp X-rays are different, but the N806 cells do not appear to be hypermutable as they are with UV. N806 cells are also moderately more resistant to the cytotoxic effects of UV radiation but are more sensitive than MI2G cells when exposed to X-rays. Assays to measure the removal of cyclobutane pyrimidine dimers (CPDs) and the incision step of nucleotide excision repair have revealed no detectable difference in the repair capacities of N806 and parental V79 cells. These results suggest that chronic, protracted UV irradiation may be able to induce a 'mutator phenotype' in a subpopulation of the progenitor cells.

Are three forms of potentially lethal damage expressed after x irradiation by treatment with hypertonic solutions in chinese hamster V79 cells?

X rays have been shown to induce two forms of potentially lethal damage (PLD), "fast-repairing" PLD and "slowly repairing" PLD, whose repair is completed in 1 h and 4-6 h, respectively. In this study three modes of treatment with hypertonic solutions containing different NaCl concentrations for different durations (0.5 M for 30 min, 0.225 M for 4 h, 0.16 M for 16 h) were examined to determine which form of PLD is expressed under each condition. These three modes of treatment enhanced the cell-killing action of X rays on actively growing V79 cells due to fixation of PLD. The kinetics of recovery from PLD was assessed by delayed treatments with hypertonic solutions. Cells exposed to one of the three treatments (see above) had completed recovery times of 1, 4 and 8 h, respectively, suggesting the possibility that these three modes of treatment cause the expression of different forms of PLD. As has been reported, treatment with 0.5 M NaCl for 30 min expressed fast-repairing PLD. The independence of the PLD expressed by 0.225 M NaCl for 4 h from fast-repairing PLD expressed after 0.5 M NaCl for 30 min was shown by combined treatment with the two modes, which reduced survival to the level that would be reached if the two modes acted independently. The data on the recovery time and on the inhibition by 0.225 M NaCl of recovery from slowly repairing PLD in plateau-phase cells indicated that the PLD expressed after 0.225 M NaCl for 4 h may be related to slowly repairing PLD. The combined treatment of 0.16 M NaCl for 16 h with 0.225 M NaCl for 4 h indicated independent action, albeit incomplete, of the PLD expressed after 0.16 M NaCl for 16 h from slowly repairing PLD. We propose for the first time that "very slowly repairing" PLD is expressed after 0.16 M NaCl for 16 h in exponentially growing cells and that therefore three forms of PLD are expressed by hypertonic treatments after X irradiation.

Neutron-energy-dependent mutagenesis in V79 Chinese hamster cells.

There has been a keen interest in the past decade in both elucidating the mutation frequency for different energy radiations and determining if mutation frequencies vary from one gene to another. This interest is driven in part by the strong link between mutational events and subsequent development of the carcinogenic state. Using fast neutrons produced by impinging protons on a beryllium target at the UCLA/VA cyclotron, we have examined the energy dependence of the induction of mutants at the hprt and tk loci. In this paper, we present studies using V79 Chinese hamster cells exposed to beams of neutrons produced from protons with 46, 30, 20 and 14 MeV energy. There is a gradually increasing cytotoxic effect of the neutrons as the energy decreases. In a similar fashion, the mutation frequency also shows a strong energy dependence with the frequency increasing as the energy decreases. The results also show that the frequency of induced mutants at the tk gene is higher than at the hprt gene. Calculations of RBE using gamma rays as the standard radiation showed a maximum for 14 MeV neutrons of 5.4 for the hprt locus and 36.6 for TK normal-growth mutants (TKng). Most of the curves for induction are best fitted with a linear function in the low-dose region with a few becoming curvilinear at higher doses.

The enhancement of cell lethality and changes in production and repair of DNA damage by hypertonic treatment after exposure to X rays, bleomycin, and neocarzinostatin.

Chinese hamster ovary cells were treated with hypertonic 0.5 M NaCl solution after exposure to X rays or the radiomimetic drugs bleomycin or neocarzinostatin. The cytotoxicity of these agents was greatly enhanced by the hypertonic treatment. On the other hand, no effect was observed after exposure to ultraviolet light, and a significant effect was obtained with mitomycin C (MMC), adriamycin (ADR), and ethyl methanesulfonate (EMS). Assays by filter elution revealed that hypertonicity had various effects on the damage produced by the different agents. Strand breaks resulting from exposure to X rays and radiomimetic agents were sensitive to hypertonic treatment. Hypertonicity caused the production of new lesions and inhibited the rejoining of DNA strand breaks, both of which may be responsible for the enhanced cytotoxicity. On the other hand, the formation of crosslinks by MMC or protein-associated double-strand breaks by ADR, the major forms of damage by which these agents cause cytotoxicity, was not affected by hypertonic treatment. As strand breaks are known to be produced by MMC or ADR, they could account at least partly for the sensitization. However, various kinds of damage are produced by MMC, and any of these could be involved in the sensitization. To our knowledge EMS produces only base damage. Thus hypertonic treatment may have an effect on various types of damage.

Effects of WR-1065 and WR-151326 on survival and neoplastic transformation in C3H/10T1/2 cells exposed to TRIGA or JANUS fission neutrons.

We demonstrated the ability of aminothiols WR-1065 and WR-151326, each at concentration 1 mM, to protect C3H/10T1/2 cells against the transforming effects of fission neutrons under two distinct sets of experimental conditions. Experiments with WR-1065 were performed with stationary cultures of C3H/10T1/2 cells, and a TRIGA reactor-generated fission neutron field at the Armed Forces Radiobiology Research Institute (USA). Experiments with WR-151326 were performed with proliferating cultures of C3H/10T1/2 cells and a JANUS reactor-generated fission neutron field at the Argonne National Laboratory (USA). Radioprotectors were present before, during, and after irradiation for total-periods of 35 min (WR-151326; 10 min pre-incubation) or 1 h (WR-1065; 30 min pre-incubation). Bioavailability of WR-1065 and WR-151326 in extracellular medium under experimental conditions simulating those of the transformation experiments was studied by measuring oxidation rates in the presence of attached C3H/10T1/2 cells in plateau and exponential phase of growth for periods of up to 5 h. Estimated half-lives for autoxidation of WR-1065 or WR-151326 were approximately 8 min or 1 h regardless of the proliferative status of cells. In the absence of WR-compounds, dose-response data for transformation induction by neutrons from TRIGA and JANUS reactors were fitted to a common curve with a linear coefficient of about 7 x 10(-4)/Gy. WR-151326 and WR-1065 were found to provide significant radioprotection by factors of 1.79 +/- 0.08 and 3.23 +/- 0.19, respectively, against fission neutron-induced neoplastic transformation. No significant protection against neutron-induced cell lethality was observed.

Comparison of repair of DNA double-strand breaks caused by neutron or gamma radiation in cultured human cells.

The dose-response for the induction of initial double-strand breaks (dsb) in DNA of human epithelioid cells by JANUS 0.85 MeV fission-spectrum neutrons was parabolic as assayed by a calibrated neutral filter elution technique. The relative biological effectiveness (RBE) of these neutrons relative to 60Co gamma-rays was unity. The kinetics of repair after a 60 Gy gamma-ray exposure were biphasic. About 65% of these dsb were rapidly repaired (T 1/2 of approximately 2 min), and the remainder were almost completely removed after 150 min at a slower rate (T 1/2 = 30 min). After the same dose of JANUS neutrons, the rapid repair component was markedly reduced (possibly not a significant repair component), and the bulk of the dsb were sealed more slowly (T 1/2 = 90 min). After 150 min, 25% remained unsealed. Even after a lower neutron dose (20 Gy), a proportion of the dsb were refractory to repair. Thus, unrepaired (or irreparable) dsb induced by high energy neutrons might explain the high RBE of neutrons for cell killing.

Energy and dose-rate dependence of neoplastic transformation and mutations induced in mammalian cells by fast neutrons.

Since our original observation that low-dose, low-dose-rate JANUS neutrons produced more transformation in C3H 10T1/2 cells than equivalent doses at high dose rate (Hill, Buonaguro, Myers, Han, and Elkind, Nature 298, 67-69, 1982), there have been many reports on the dose-rate dependence of a variety of high-LET radiations. Some of these have qualitatively supported our original and subsequent observations, and some have found no evidence for a dose-rate effect. Thus there remain questions about the generality, size, tissue specificity, and reproducibility of these observations. Furthermore, there are no well-established mechanistic descriptions or observations to account for such effects. In this report studies are presented using a range of neutron energies produced by the UCLA cyclotron from 12- to 46-MeV protons on beryllium. In particular, emphasis is placed on comparing results between energies for neoplastic transformation and mutation end points, and preliminary data are presented on the molecular and mechanistic aspects of neutron-induced damage.

Repair of potentially mutagenic damage and radiation quality.

The induction of potentially lethal damage and potentially mutagenic damage expressed by hypertonic salt treatment in late S-phase V79 cells after exposure to 46-MeV fast neutrons and 50-kVp X rays has been studied. Resistance to 6-thioguanine was used as the mutagenic end point. When cells in late S phase were treated with hypertonic salt solution immediately after fast neutrons or X irradiation, both cell killing and mutation induction were enhanced compared to fast neutrons or X irradiation alone. These results suggest that neutron as well as X irradiation of cells in late S phase induces both potentially lethal damage and potentially mutagenic damage. However, some differences in mutation frequency were observed between the two radiations after salt treatments. These results are discussed in terms of repair of mutagenic damage by an error-free or error-prone process.

The lethal effects of Fermilab fast neutrons vary with the depth of cells in a water phantom.

Using V79 Chinese hamster lung fibroblast cells grown in culture we have examined the lethal effects of fast neutrons from the Fermilab therapy facility as a function of depth in a water phantom. Exposures and dosimetry were performed from 3 to 24 cm deep in the phantom along the central axis of the neutron beam, using various collimator configurations. The results indicate that the relative biological effect (RBE), using 60Co gamma rays as the standard radiation, varies with depth in the neutron beam. Starting from d-max (approx. 3 cm), the RBE appears to decrease continuously with depth. At 24 cm deep, the relative effectiveness is 10-15% lower than at 3 cm deep. There appears to be no systematic variation of relative effectiveness with shape or size of collimator. If a hydrogenous filter is added before the beam passes into the water phantom, the variation with depth is greatly reduced.

A uv-sensitive Chinese hamster lung fibroblast cell line (V79/UC) with a possible defect in DNA polymerase activity is deficient in DNA repair.

Studies of repair enzyme activities in a uv-sensitive cell line (V79/UC) derived from Chinese hamster V79 cells have revealed levels of total DNA polymerase that are about 50% of the levels in the parental cell line. There are a number of DNA polymerase inhibitors available which allow us to distinguish between the major forms of DNA polymerase (alpha, beta, gamma, and delta) identified in mammalian cells. Enzyme assays with these inhibitors indicate that the aphidicolin-sensitive DNA polymerase is defective in the V79/UC cell line. This could be either polymerase alpha or delta, or both. The V79/UC cells do not express resistance to aphidicolin in standard toxicity studies. However, when aphidicolin is added postirradiation in survival assays designed to measure the extent of inhibitable repair, V79/UC cells do not respond with the further decrease in survival seen in the parental line. Further evidence of a polymerase-dependent repair defect is evident from alkaline elution data. In this case the V79/UC cells show the appearance of single-strand breaks following uv irradiation in the absence of any added inhibitor. Cells of the V79/M12G parental line, on the other hand, show the appearance of single-strand breaks only when aphidicolin is present.

Fission-neutron-induced expression of a tumour-associated antigen in human cell hybrids (HeLa x skin fibroblasts): evidence for increased expression at low dose rate.

The induction of a tumour-associated antigen in a human cell hybrid line (HeLa x skin fibroblast) following exposure to fission neutrons of average energy 0.85 MeV (Janus reactor, Argonne National Laboratory) at two dose rates, 0.086 and 10.3 cGy/min, has been examined. The dose-response data obtained indicate the lower dose rate to be 2.9-fold more effective than the higher in inducing expression of the tumour-associated antigen, while there was no significant dose-rate effect in terms of cell killing. These results are qualitatively in agreement with previous observations using neutrons from the Janus reactor for the neoplastic transformation of C3H10T1/2 cells and Syrian hamster embryo cells.