Glutathione determination by the Tietze enzymatic recycling assay and its relationship to cellular radiation response.

Large fluctuations in glutathione content were observed on a daily basis using the Tietze enzyme recycling assay in a panel of six human cell lines of varying radiosensitivity. Glutathione content tended to increase to a maximum during exponential cell proliferation, and then decreased at different rates as the cells approached plateau phase. By reference to high-performance liquid chromatography and flow cytometry of the fluorescent bimane derivative we were able to verify that these changes were real. However, the Tietze assay was occasionally unable to detect glutathione in two of our cell lines (MGH-U1 and AT5BIVA), although the other methods indicated its presence. The existence of an inhibitory activity responsible for these anomalies was confirmed through spiking our samples with known amounts of glutathione. We were unable to detect a direct relationship between cellular glutathione concentration and aerobic radiosensitivity in our panel of cell lines.

For at least the past decade the tripeptide sulphydryl compound glutathione (GSH) has been the subject of intensive investigation regarding its role in mediating both radiationand drug-induced responses in living cells. GSH has been proposed to act in a vast number of different cellular processes, amongst them the maintenance of a suitable intracellular reductive environment, protection against harmful xenobiotics, a catalyst of or reactant in several metabolic schemes and an intracellular store of cysteine (Nygaard and Simic, 1983;Mitchell and Russo, 1987). Its potential importance for radiobiology has been recognised ever since the development of the competition model for radiation cell killing, a mathematical consequence of the reaction rates for competing reactions (Alper and Howard-Flanders, 1956). DNA radicals produced by irradiation are considered to be subject to a competition between oxidising (or electronaffinic) agents, leading to damage fixation and ultimate cell death, and reducing species (such as sulphydryl compounds) which, through hydrogen atom donation would facilitate damage repair and so lead to continued cell viability, against a background of the intramolecular decay of DNA radicals to render them non-restorable (Koch, 1983). Since the rate of reaction between oxygen and DNA radicals is far greater than that between GSH and DNA radicals (Bump and Brown, 1990), the competition theory predicts that under normal aerobic conditions the former reaction would dominate, rendering unimportant any changes in GSH content. However GSH may nevertheless play a significant role in the aerobic radiation response, its protective effects mediated through other processes limiting radiation damage, for instance the detoxification of radiation-induced hydroperoxides (Dethmers and Meister, 1981;Biaglow et al., 1984).
Cells may be depleted of GSH by treatment with the specific enzyme inhibitor DL-buthionine-S,R-sulphoximine (BSO) which prevents GSH synthesis (Griffith and Meister, 1979). Increased aerobic radiosensitisation following BSO exposure has been reported for a human lung adenocarcinoma , HeLa (van der Schans et al., 1986), V79 cells (Astor et al., 1984), human lymphoid cells (Dethmers and Meister, 1981), and drug-resistant variants both of a human breast line (Lehnert et al., 1990) and of a human ovarian cell line (Britten et al., 1990), but this is a far Correspondence: J J Eady Received 14 March 1995; revised 16 June 1995; accepted 21 June 1995 from universal phenomenon. Some authors have observed no relationship between glutathione concentration and aerobic or hypoxic radiosensitivity (Debieu et al., 1985), whilst many others have found that GSH depletion resulted in radiosensitisation only under hypoxic conditions, though the degee of sensitisation produced is highly variable (Bump and Brown, 1990).
We have sought to clarify the relationship between GSH and radiosensitivity under aerobic conditions for a panel of human cell lines of varying radiosensitivity using three methods for GSH determination: the standard Tietze enzyme recycling assay and both high-performance liquid chromatography (HPLC) and fluorescence-activated cell sorting (FACS) of the bimane derivative of GSH. Within our department we have observed significant differences in initial DNA damage levels following ionising radiation which show a correlation with radiosensitivity (Kelland et al., 1988;Whitaker et al., 1995). These have been proposed to be due to differences in non-protein thiol levels (Malaise, 1983), but radiochemical considerations suggest that this is unlikely (Ward, 1990). Radiation hypersensitive AT5BIVA cells do not exhibit increased initial damage (Peacock et al., 1989); instead their radiation sensitivity is considered to be due to a recombination defect (Taylor et al., 1994). Their inclusion in this study was prompted by a desire to provide a contrast to the other cell lines, since the (presuimably) fundamentally different mechanism governing their radiosensitivity should provide an exception to any relationship with GSH we observe in the other cell lines.

Materials and methods
Cell lines Six human cell lines were used in this study, four tumour lines and two virally transformed fibroblasts, reflecting a range of radiosensitivity. The relatively radioresistant RT112 was derived from a transitional cell carcinoma of the bladder (Masters et al., 1986), as was MGH-U1 (Kato et al., 1977).
HX34 was established from a melanoma originally grown as a xenograft (Smith et al., 1978) and the radiosensitive HX142 was derived from a xenografted neuroblastoma (Deacon et al., 1985). These tumour lines were maintained in Ham's F12 medium with 10% fetal calf serum (FCS; Imperial Laboratories, Andover, UK) and were regularly passaged with 0.05% trypsin in 0.02% versene.
The two transformed fibroblast ines were grown in Dulbecco's modified Eagle medium with 10% FCS, buffered by 10 M Hepes (Sigma, Dorset, UK). MRC5-CVI is an immortalised fibroblast line originating from a normal patient, whilst AT5BIVA was derived from an individual with ataxia telangiectasia (Ariett et al., 1988).
Population growth kinetics for all the cell lnes was determined by plating 1-2 x 106 cells per 80CM2 tissu clture flask and trypsning monolayers at daily intervals threafter.
Cells were counted both by haemocytometer and by Coulter Counter, which was also used to size the cells. All cell lines reached plateau phase at 5-6 days of growth, when the numbers varied from 5 x 10' to 10' cells per flask (data not shown). The radiosensitivity of these cell lines has been described in a previous publication from this department and has been found not to vary with tume in exponential culture (Peacock et al., 1992). Cell extracts were prepared by washing cell monolayers in T80 flasks (Flow Laboratories) or trypsinisd cells twice in 10 ml of ice-cold (4C) phosphate-buffered saline (PBS), followed by lysis for 30 min in 2 ml of either 0.6% 5sulphosalicylic acid (5-SA; Tietze assay) or 5% metaphosphoric acid (HPLC: BDH Chemals, Lutterworth, UK), in the dark and on ice, occasionally gently shaking the tissue culture flask/tube to ensure that all the cells were covered by acid during the extraction. Samples were stored for up to 1 week at -2OC before being assayed, a process which we found not to alter the results obtained.
Tietze recycling assay GSH was determined using a slight variation of Griffith's (1980) modification of Tietze's (1969) assay, based on the prinaple that GSH can be measred by an enzymatic recyclmg procedure in which it is sequentially oxidised by 5,5'dithiobis-(2-nitrobenzoic acid; DTNB) and redced by NADPH in the presence of glutathione reductase. The rate of formation of 2-nitro-5-thiobenzoic acid (TNB) can be followed using a spectrophotometer and GSH quantitated by reference to a standard curve. A stock buffer of 143 mm sodium phosphate and 6.3 mm sodium-EDTA (pH 7.5) was made up in distlle water, and used to prepare separate solutions of 0.3 mm NADPH, 6 mm DTNB and 50 units ml-I GSH reductase (type HI, from Saccharomyces cerevisae, Sigma). For each lysate, a final tube was made up containing 700 pi NADPH solution, 100 p1 DTNB, 100 i1 of GSH standard or sample and 100 p1 of water. This mixture was warmed at 30-C for 10 min before being transferred to a cuvette containing 10 p1 of the GSH reductase, and the rate of absorbance at 412 nm measured on a spectrophotometer (Kontron Uvikon). Standards of known GSH content were made up by serial dilution in 0.6% 5-SA and the samples assayed by reference to a sandard curve, all points being repeated in tiplicate. For some cell extracts GSH was below the lmit of detection (<1 nmol 10' cells) and we further investigated this phenomenon by preparing standard curves in the presence of sample extracts to determin whether the cells really were deficient in GSH or whether there was some inhibitory atvity interfering with our assay.
In an analogous manner replcate cell extracts for both HIPLC and FACS analysis were set up to allow idependent comparison of GSH values determined from the three methods.

HPLC
Tlhe monobromobimane (mBBr) derivative (Fahey and Newton, 1987) of GSH in metaphosphoric acid (MPA) extracts of cell pellets was assayed by HPLC using a gradient reversephase ion-pairing technique with fluorescence detection. To a 400 p1 sample of the cell extract in 5% MPA was added 25 p1 100 #M mercaptoethanol 25 p1 10mm mBBr and 250 p1 2 M Tris/ I mm EDTA. After 15 min the reaction was terminated with 50 p1 6 M hydrogen chloride, the mixture extracted with 2 ml dichloromethane and an aliquot of the aqueous phase injected onto the HPLC (Stratford et al., manuscrpt in preparation).
Fow cytometry Freshly trypinisied cells were washed twice in ice-cold PBS and resuspended at 10' ml ' in serm-free medium containing 1 mM mBBr. The cells were incubated at 3TC for 20 min in the dark and then transferred on ice to an Ortho 50H flow cytometer. Stained cells were excited at 360 am and emission spectra measured at 420 am (Fahey and Newton, 1987) and the peaks of the distnbutions used as a relative measure of GSH between the cell samples (Rice et al., 1986).

Resis
Variation in GSH with age of culture Time-course plots for GSH determinations using the Tietze assay (both on monolayer and trypsnised cell extracts) and by HPLC (narily performed on trypsinised extracts) for each of the cell lines are shown in Figure 1. Both methods detected a decrease in GSH content per cell during the growth of the cultures, although there was not a close agreement between the methods. We have expressed these results in terms of GSH per cell in order to allow direct comparison between the different GSH measurement protocols examined here. Correction for cell volume to express GSH as a concentration has little effect on the pattern of variability observed; the differences are not due to differences in cell size. However we have found that we obtain consintly lower values for GSH content from trypsnised cell cxtracts (clod symbols in Figure 1) than from monolayer cell extracts (open squares) in the Tietze asy. This may be caused by a loss of GSH on washing the trypsinised cell pellets, or perhaps the prese of a trace contaminant in the trypsin interfering with the enzymatic processes of the Tietze assay. However this latter possibility seems extrmely unlikely as the cell pellets are washed twice in an excess of ice-cold PBS before assay. There are some apparently anomalous points within our data set, partularly for the RT1 12 day 2 and HX34 day 3 samples We find it difficult to account for these discrepancies, though they are partially due to very high values obtained from a single experiment, as indicated by the relaively large error bars for these points. These may inded be real values for GSH, but we cannot totally exclude the possibility that these apparently anomalous points could be caused by contamination of some sort.
Initially we suspected that fluctuations in GSH content were due to the progression of the cells from exponential growth to a plateau phase population, with subsequent depletion of the medium accounting for a reduction in GSH precursors and hence a gradual decrease in the levl of GSH measured. However, analysis of the growth curves for each of the cell lnes indicated that plateau phase cultures were only produced after at kast 5 days of growth (data not shown) and that this was not lily to be an explanation for the variation observed. lndeed, we began to suspect that these fluctuations may reprnt real variation in GSH content and sought to determine whether this was so by FACS, performed on cell extracts which were also used for HPLC and Tretze analysis at the same time.
Comparion between GSH estimations determinedfrom the Tietze enzymatic assay, HPLC andflow cytometry Tietze assay determinations of GSH content of trypsinised cells are compared with valus produced for the same cell samples by HPLC in Figure 2. While there is a good correlation (r = 0.85, slope = 0.96, P<0.0001) for most of the data, there are some exceptions, correspondming to MGH-Ul and AT5BIVA samples where GSH was greatly reduced or below the imits of detection in the Trez assay. There is a strong relationship between the peak of fluorescence produced by flow cytometry and the HPLC estimation of GSH content (Figure 3a). However, when values from the Tietze assay are compared with those produced from FACS ( Figure 3b) we find that there are anomalous points where there is considerable fluorescence but the Tietze assay indicates reduced or no GSH to be present, again corresponding to cell extracts prepared from MGH-Ul cells.
Identification of an enzymatic inhibitory activity interfering with the Tietze assay We investigated our anomalous results in the Tietze assay, where we were occasionally unable to detect any activity in our samples, by preparing standard curves in the presence of our samples using known concentrations of GSH.
The majority of cell extracts show a corresponding linear increase in the amount of thiol detected as GSH content in the standard is increased (Figure 4). However, AT5BIVA and MGH-Ul extracts produced unexpected results. The experimental points lie well below the 45°line, indicating that the assay underestimated GSH levels in these samples. This underestimation became more marked as the time from which the cells were last passaged increased. It appears that there is a build-up of an enzyme inhibitory activity which may render the Tietze assay unreliable in some cases. The nature of this inhibition is not known. In order to assess its reliability for any particular cell extract, we suggest that a standard curve be produced in the presence of the extract to determine whether there is any enzyme inhibition occurring.
The relationship between cellular radiosensitivity and glutathione content The day-to-day variability of GSH content has been confirmed by the internal consistency of our results using three separate measurement methods (Figures 2 and 3), with the exception of some samples assayed using the Tietze procedure. In general we have found that the trypsinisation process reduces the amount of GSH detected in the cell extracts, so we have used monolayer values for GSH concentration in 3-day-old exponentially growing cell cultures as our standard for comparison with radiation sensitivity of our cell lines, as suggested by other authors (Post et al., 1983;Batist et al., 1986). The majority of the 3 day cell extracts show no inhibition of enzyme activity, although slight enzyme inhibition is observed with AT5BIVA, and to a greater degree with MGH-Ul cell extracts (Figure 4). In this case we have used the value for 3-day-old cultures from extracts prepared for HPLC analysis.
We have found no apparent relationship between aerobic radiosensitivity (as measured by the surviving fraction at 2 Gy, SF2, determined from fitting the dose-survival points with the linear-quadratic model of radiation-induced cell kill; Peacock et al., 1992) and GSH concentration within our cell lines ( Figure 5).

Dioai
Large fluctuations in the day-to-day glutathione content of cells is a well-recorded though seldom addressed phenomenon. As early as 1969 it was recognised that HeLa cells showed a great variation in sulphydryl content during the cell cycle, with a minimum value at the end of G1 increasing 30-fold by the end of S-phase (Mauro et al., 1969). Chinese hamster ovary cells were shown to exhibit a similar variation, plateau phase G1 cells containing only 25% of the concentration of non-protein sulphydryls in cycling GI cells (Harris and Teng, 1973). However, these authors also reported the influence of culture conditions on sulphydryl content, for replacement of fresh medium was found to rapidly increase the concentration of non-protein sulphydryls so that they reached the same level as cycling cells within 4 h. A similar finding has been observed in the only widely studied human tumour cell line, the lung adenocarcinoma A549. GSH content varied by a factor of 3.5 over a period of 7 days (Oleinick et al., 1988), while changing the medium on 9-day-old plateau phase cells resulted in a similar increase in GSH content over the first 6 h . The importance of culture conditions has been emphasised by the findings that in A549 cells GSH content increases sharply for the first 24 h following passage and then decreases thereafter (with up to a 13-fold difference in the maximum and minimum values), and that serum content of the medium is important. As the serum concentration is increased, so is the level of GSH, and this increase is also mirrored with time after passage (Post et al., 1983). However, these changes are not due to depletion of GSH precursors in the medium, for similar results were obtained when the medium was replaced daily. Significant fluctuations in GSH during cell growth in vitro have also been described for ovarian cell lines (Batist et al., 1986), leading these authors to propose that the optimal time point for comparison between cell lines is mid-log phase growth, a protocol which we have followed here. Allied to these variations in GSH has been the realisation ..^...o 300 -F r that one of the major methods for GSH estimation, the enzymatic recycling assay developed by Frank Tietze (1969), is prone to perturbation by inhibitors of glutathione reductase (GR) the enzyme central to the recyding assay. Discrepancies between Tietze and mBBr HPLC results have been ascribed to the presence of acid-soluble sulphydryl proteins in the extracts, giving erroneous vahls for the enzymatic assay (Loh et al., 1990). Unidentified native inhibitors of GR have been recognised m ssue extracts (Oshino and Chance, 1977), and xenobiotics such as nitrobenzen and nitrofuran compounds have been shown to be enzyme inhibitors (Buzard and Kopko, 1963). In addition, it has been recognised that some reagents (e.g. N-ethylmalimide) will interfere with the reductase activity (Griffith, 1980), and such considerations have inspired comprehensive analysis of the Tietze assay, leading to the proposal that the assay be performed at pH 6.0, conditions under which the enzymatic reaction is not rate lmiting (Eyer and Podhradsky, 1986). However, even this suggestion is not entirely satisfactory because under these modified conditions the reaction mixture is not well buffered by phosphate, and small changes in pH can have a marked influen on the results. There is even a precedent for our spking of the samples with known concentrations of GSH, when a reduction in standard slope was observed in the presence of acid extracts from liver samples, indicative of an inhibitory activity (Brigelius et al., 1983). We have verified our GSH determinations from the Tietze assay with reference to standard curves produced in the presence of sample extracts and have found that only MGH-Ul exhibits enzyme inhibition to such a degree that the Tietze assay is unreliable for this cell ine. Hence we have in this case used the GSH value from HPLC analysis of the bimane derivative, a method which is not influenced by enzyme inhibition and which has here produced very simila results to the Tetze assay for the other cell lines studied. Disrepancies between the Tietze assay results and the use of monochlorobimane as a fluorescent probe for GSH have been reported for human (but not hamster) cells (Cook et al., 1989). Since the chloroderivative reacts much more slowly with GSH than does the bromoderivative (Rice et al., 1986) these discrepancies have been put down to insufficient levels of the GSH conjugating enzyme glutathione S-transferase, and so we have used the bromoderivative for HPLC to circumvent these problems. We have found that there is little correlation between the GSH concentration within exponentally growing cells plated 3 days previously and aerobic radiosensitivity.
In the only other study which has s aerobic radiosensitivity in relation to the glutathione content of a series of human tumour cell lines, Carmichael et al. (1988) found 15-fold variation in GSH content in a panel of 13 colorectal carcinoma lines, and that this variation was unrelated to cellular radiosensitivity. V79 clones exhibiting enhanced levels of non-protein sulphydryls have also been shown to have the same radiosensitivity as control V79 cells, implying that radioresponsiveness may not be crtically determined by thiols (Hogis, 1990). No relationship between GSH concentration and either aerobic or hypoxic radiosensitivity has been observed in glutathione-deficient fibroblasts derived from patients with 5-oxoprolinuria. However, under hypoxic conditions a strong correlation existed with glutathione synthetase activity, g that GSH synthesis is required after irradiation (Debieu et al., 1985). An intriguing finding is that cells sensitised by extreme GSH depletion can have their resistance restored by the addition of a very low level of extracellular GSH. This GSH has been found not to enter the cell and its mechanism of action remains unknown (Clark et al., 1986). Glutathione protection is considered to occur principally through hydrogen atom donation to restore damaged macromoeules, particularly DNA radicals. Radical scavenging in aerobic cells requmres a greater GSH concentration than has nd rats in -1 J J Eady eti a 1093 ever been achieved in any cell, while other protective mechanisms need only a low GSH concentration so that they are operating close to maximally under normal conditions (Bump and Brown, 1990). However, extreme GSH depletion may reduce the influence of such mechanisms, particularly enzymatic protection against hydroperoxides, and it is the reduction of this function that is considered responsible for icreased aerobic radiosensitivity reported following extreme GSH depletion (Dethmers and Meister, 1981;Astor et al., 1984;Biaglow et al., 1984;van der Schans et al., 1986;Britten et al., 1990;Lehnert et al., 1990). Radiation-produced peroxide is reduced, and so detoxified, principally by GSH peroxidase, with catalase accounting for the remaining inactivation, so that peroxide is not thought to be responsible for any increased radiosensitivity under aerobic conditions (Biaglow et al., 1992).
The relationship which we have presented between glutathione and cellular radiosensitivity is to some degree an artificial onethere is no a priori reason why we should take the GSH content of 3-day-old exponentially growing cultures as a gold standard (except that other authors have suggested comparisons to be made between different cell lines at mid-log phase growth, e.g. Batist et al., 1986;Post et al., 1983). The large day-to-day variations in GSH content displayed by our cell lines make the choice of any one time point for comparison with a separate parameter (e.g. radiosensitivity) a largely arbitrary choice. Indeed this is highlighted by the fact that both AT5BIVA and MRC5-CV1 have been published to have a level of 42 nmol GSH IO-7 cells (Dean, 1987) in contrast to the values reported here which varied widely depending on the time at which the measurements were made.
Non-protein sulphydryls account for only a small proportion of total cellular thiol content, though the more abundant protein thiols were thought not to act as efficient radical scavengers owing to steric hindrance and their low diffusion coefficients, and so were considered unlikely to play a major role in protection against radiation damage (Biaglow et al., 1983). However their importance in radiation protection is increasingly being recognised (Held et al., 1991;Ljungman et al., 1991). Glutathione is not the most efficient radioprotector but its versatility as a substrate or co-factor for protective enzymes and relative abundance makes it effectively the most important non-protein sulphydryl in the cell. Other thiols, such as DTT (Held et al., 1984), cysteamine and WR-1065 (Fahey et al., 1991) are much better radioprotectors under aerobic conditions, and this is considered to be due to electrostatic interactions in close proximity to DNA (Aguikra et al., 1992).
The variation in GSH content we have described here contrasts with our experience of clonogenic stem cell survival assays where the radiobiological parameters describing the survival curves remain almost constant over time (provided the cultures are maintained in exponential growth), and little account is rouinely taken of the duration since the tumour or transformed fibroblast cells were last passaged (Peacock et al., 1992). These facts alone would tend to suggest a limited role for glutathione in the aerobic radiation response, and given the multifarious nature of thiol-mediated processes within the cell it would seem unlikely that the level of glutathione were to relate strongly to aerobic radiosensitivity compared between a varety of different cell lines.

ckm iedgs
The authors wish to express their gratitude to Professor GG Steel for his critical aIsessment of this manuscript, to Mrs J. Titley for her assstance with the flow cytometry, and to Sylvia Stockbridge and Rosemary Couch for their expert secrtari asistanc in the preparation of this paper. This work was supported by the Cancer Research Campaign.