Loss of GSH, oxidative stress, and decrease of intracellular pH as sequential steps in viral infection.

Madin-Darby canine kidney cells infected with Sendai virus rapidly lose GSH without increase in the oxidized products. The reduced tripeptide was quantitatively recovered in the culture medium of the cells. Since the GSH loss in infected cells was not blocked by methionine, a known inhibitor of hepatocyte GSH transport, a nonspecific leakage through the plasma membrane is proposed. UV-irradiated Sendai virus gave the same results, confirming that the major loss of GSH was due to membrane perturbation upon virus fusion. Consequent to the loss of the tripeptide, an intracellular pH decrease occurred, which was due to a reversible impairment of the Na+/H+ antiporter, the main system responsible for maintaining unaltered pHi in those cells. At the end of the infection period, a rise in both pHi value and GSH content was observed, with a complete recovery in the activity of the antiporter. However, a secondary set up of oxidative stress was observed after 24 h from infection, which is the time necessary for virus budding from cells. In this case, the GSH decrease was partly due to preferential incorporation of the cysteine residue in the viral proteins and partly engaged in mixed disulfides with intracellular proteins. In conclusion, under our conditions of viral infection, oxidative stress is imposed by GSH depletion, occurring in two steps and following direct virus challenge of the cell membrane without the intervention of reactive oxygen species. These results provide a rationale for the reported, and often contradictory, mutual effects of GSH and viral infection.

Oxidative stress in biological systems is caused by overload of oxidants, in particular reactive oxygen species, with respect to the antioxidant defense system developed by cells to counteract oxidation (1,2). In fact, sustained oxidative insult tends to disrupt cell structures and functions, insofar as they are maintained and mediated by critical redox balance. Oxyradicals are suspected to be involved in several pathological processes such as inflammation, cancer, and neurodegenerative diseases (3), and, consequently, counteracting molecules are the object of intensive study. A major role in this respect has been assigned to the tripeptide glutathione, the most abundant non-protein thiol of the cell (4). It performs various functions ranging from cellular metabolism to transport as well as pro-tection against free radicals and reactive oxygen species. The antioxidant function of the tripeptide is related to oxidation of the thiol group of its cysteine residue with formation of a disulfide (GSSG), which is, in turn, catalytically reduced back to the thiol form (GSH) by glutathione reductase.
In the past few years, in addition to detrimental effects, a metabolic requirement for oxygen radical-induced oxidative stress has been proposed. Evidence has accumulated suggesting that redox mechanisms play a fundamental role in cellular events such as binding of nuclear transcription factors to DNA, protein binding to mRNA, and hormone-receptor interactions. The most striking example is the redox regulation of transcription factors such as NF-B and AP-1, which activate gene transcription in response to peroxide (5,6).
It has also been reported that the nuclear factor B is involved in the activation of HIV 1 gene transcription (7). Therefore, it seems conceivable that oxidative stress might be a control factor in HIV infection. Furthermore, an altered antioxidant status was observed in plasma and other tissues of AIDS patients (8). However, the role played by oxidative conditions in HIV infection is still debated, on the basis of contradictory results. In fact, glutathione levels have been reported to decrease upon infection with HIV (9). The decrease seems to affect preferentially GSH, since relatively normal levels of GSSG were found in tissues of HIV-infected patients. On the contrary, another study reported an increase of GSSG, as mediator of oxidative stress, in CD4 ϩ lymphocytes (10). Recently, it has also been suggested that GSH depletion is not directly related to infection of HIV virus in view of the lack of changes in the tripeptide content in circulating cells from AIDS patients (11).
In the search of a common redox event affecting virus replication, it has to be taken into account that the plasma membrane represents the earliest target of virus attack. It has been reported that oxidative stress is able to modify the activity of cell membrane ion transport systems, such as Ca 2ϩ movements, Na,K-pump, and sodium, potassium, and chlorine cotransport activity (16,17). These ion transport systems are essential for proper cellular function and viability, and their alterations may represent an early disturbance following oxidant exposure. Strictly connected to the previously mentioned ion transport mechanisms is the Na ϩ /H ϩ antiporter. This is a plasma membrane protein that exchanges sodium and hydrogen ions according to their concentration gradient, whose main function is the regulation of intracellular pH and cell volume (18). The Na ϩ /H ϩ antiporter is therefore strictly dependent on the sodium gradient, fueled by the Na-pump. Furthermore, the Na ϩ /H ϩ antiporter is able to modulate cell growth and proliferation, and its activation, with a consequent increase of intracellular pH, can represent a first response of the cell to hormones and growth factors (19). Recently, it has been reported that oxidative stress, associated with GSH decrease, produces changes in the Na ϩ /H ϩ antiporter activity, thus resulting in a significant decrease in the rate of recovery from an acid load (20).
In a previous study, we demonstrated that a loss in reduced glutathione occurs during the infection of epithelial cells with Sendai and herpes simplex type 1 (15,21) viruses at very early stages of replication. In the present report we investigated the replication of Sendai virus in Madin-Darby canine kidney (MDCK) cells. Adsorption of virus to the cell appeared to be responsible for a major loss of GSH; GSH loss oxidatively affected the function of the Na ϩ /H ϩ antiporter, leading to lower intracellular pH. Since acidic conditions have been demonstrated to favor early stages of viral replication (22,23), a more general relationship between virus replication and oxidative stress may be found in the events following GSH depletion.

EXPERIMENTAL PROCEDURES
Cells and Reagents-MDCK cells were purchased from the American Type Culture Collection and grown in RPMI 1640 supplemented with heat-inactivated 5% fetal calf serum, 0.3 mg/ml glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin in a humidified atmosphere containing 5% CO 2 at 37°C. Vitality was estimated by trypan blue exclusion, and cell density was determined using a Neubauer chamber.
Virus Infection and Titration-Sendai virus is a nonsegmented single-stranded RNA virus of negative polarity. Stocks of the virus were obtained by allantoic inoculation of 10-day-old embryonated eggs with 0.2 ml of a 10 Ϫ6 dilution of infected allantoic fluid. Confluent monolayers of cells were infected with 10 hemagglutinating units (HAU)/2 ϫ 10 5 cells. After incubation for 1 h at 37°C (adsorption period), unadsorbed virus was removed, and the monolayers were washed three times with phosphate-buffered saline (PBS; 20 mM sodium phosphate, 140 mM NaCl, pH 7.4) and then incubated with RPMI supplemented with 2% fetal calf serum. Virus production was determined by measuring the HAU present in the medium of infected cells at different times postinfection. Hemagglutinin titrations were done according to standard procedures, using human type O Rh ϩ erythrocytes (24). Mock infection was performed by using allantoic fluid of uninfected eggs at the same dilution.
Purified Sendai virus was obtained from allantoic fluid of eggs infected for 48 h (24). Specifically, the fluid was collected and clarified at 6000 ϫ g for 20 min. Supernatants were diluted in PBS and centrifuged at 12000 ϫ g for 2 h with a 45Ti rotor (Beckman). The virus pellet was resuspended in PBS and sonicated in a water bath sonicator to break up virus aggregates. Titration of purified virus was performed by measuring the number of hemagglutinin units present in the medium of infected monolayers of MDCK cells at different times postinfection.
In some experiments, cells were prewashed with ice-cold PBS and either mock-infected or infected with Sendai at 4°C in order to avoid the fusion of the virus with the cell membranes. In other experiments, we infected the cells with UV-inactivated virus, which is unable to replicate as the consequence of UV-induced RNA damage. Virus suspensions were exposed to UV light (40 W, 254 nm, 8 cm distance) for 10 min with constant stirring. The reduction of infectivity was determined by measuring HAU released by UV-inactivated Sendai-infected cells. Uptake of [ 35 S]Cysteine-Confluent cell monolayers were labeled for 24 h, starting just after mock-or Sendai infection, with 5 mCi/1 ϫ 10 5 cells. After this period, cells were washed three times with cold PBS and lysed with lysis buffer (10 mM Tris, 150 mM NaCl, 0.25% Nonidet P-40). 40% trichloroacetic acid was added to the cell lysates. After 30 min in ice, the material was centrifuged at 12,300 ϫ g for 15 min, and the radioactivity incorporated into acid-soluble material was determined. The radioactivity incorporated into acid-precipitated material was determined after filtration with sampling manifold (Millipore Corp.). To determine the radioactivity incorporated into viral proteins, cell lysates were immunoprecipitated with goat anti-parainfluenza 1 antibodies adsorbed to protein A-agarose beads. After extensive washing, the radioactivity incorporated was determined in trichloroacetic acid-precipitated material.
Determination of Glutathione and Mixed Disulfides-Intracellular glutathione was assayed upon formation of S-carboxymethyl derivatives of free thiols with iodoacetic acid followed by conversion of free amino groups to 2,4-dinitrophenyl derivatives by reaction with 1-fluoro-2,4-dinitrobenzene as described by Reed et al. (25). Monolayers of MDCK cells were carefully washed with a large volume of PBS. The cells were detached by gentle scraping and centrifuged at 700 ϫ g for 10 min. The supernatant was discarded, and the cell pellet was resuspended in 200 l of buffer. Cell lysate was obtained by repeated cycles of freezing and thawing under liquid nitrogen. Protein was then precipitated by adding metaphosphoric acid (5% final concentration). After centrifugation at 22,300 ϫ g for 30 min, the low molecular thiols in the supernatant were derivatized and measured by HPLC using a Bondapak NH 2 column (Waters). GSH and GSSG were used as external standards. Mixed disulfides were determined by HPLC after reduction of cell lysates with NaBH 4 for 30 min at 40°C (26). GSH and GSSG in the culture medium were determined after centrifugation of the collected media at 700 ϫ g for 10 min and subsequent acidification.
Aliquots of cell lysates were used for total protein determination by the method of Lowry et al. (27).
Lactic dehydrogenase enzyme activity (LDH) was determined as described (28) both in the culture medium and intracellularly. Control or infected cells were gently scraped in the culture medium. An aliquot of this suspension was utilized for total enzyme assay after cell lysis. Another aliquot was centrifuged at 700 ϫ g for 10 min. The supernatant was then used for extracellular enzyme assay. The extracellular enzyme was expressed as the percentage of the total activity.
Cell Treatments-Confluent monolayers were treated with 0.5 mM DTT or 1 mM DTNB for 24 h after virus infection. BSO (1 mM) was added to cells 18 h before virus infection, and/or for 24 h after. At the end of the treatments, culture medium was collected and utilized for virus titration, and cells were washed, scraped, and pelleted by centrifugation at 700 ϫ g for 10 min and treated for GSH determination. All substances were used at nontoxic concentrations, as evaluated by the trypan blue exclusion test.
In some experiments monolayers were treated with 1 mM methionine 1 h before virus infection and for 24 h after.
Intracellular pH Measurements-For the experiments of fluorescence assays, cells were grown in chamber slides and used at confluency. Intracellular pH was measured using a previously established technique based on the fluorescent intracellular pH indicator BCECF/AM (29). BCECF/AM (1 mg/ml) and EIPA (10 mM) were dissolved in Me 2 SO, which did not affect the fluorescent signal.
In order to rule out the contribution of HCO 3 Ϫ -dependent transport mechanisms (30), all experiments were carried out in bicarbonate-free buffer having the following composition: 135 mM NaCl, 5 mM KCl, 1 mM CaCl 2 , 1 mM MgCl 2 , 10 mM glucose, 20 mM HEPES, pH 7.3. This buffer was used for the incubation with the fluorescent probe and for the determination of intracellular pH, unless otherwise stated; the cells incubated in this buffer were considered virtually depleted of bicarbonate.
NH 3 /NH 4 -containing solutions were prepared by replacing, in the above buffer, 20 mM NaCl with 20 mM NH 4 Cl. Routinely, during the exposure to NH 4 Cl, external Na ϩ was replaced by equimolar choline chloride concentrations in order to keep quiescent the antiporter.
Incubation with the fluorescent dye was carried out as previously reported (31); cells were washed twice with Na ϩ buffer, and under these conditions cells were considered bicarbonate-free. Cells were then incubated in Na ϩ buffer with the fluorescent dye (1 mg/ml in Me 2 SO) at a final concentration of 1 g of BCECF/10 6 cells, for 30 min at 37°C in the dark. Then the medium containing the dye was eliminated, and the cells were washed twice with the same buffer.
The calibration curve was carried out as previously reported (32) using the nigericin (10 mM in ethanol) method in a medium containing high potassium with same composition as Na ϩ medium but equimolar KCl substituted for NaCl. The calibration curve was linear in the range of pH 6.5-7.5 (not shown).
Fluorescence was also measured under continuous magnetic stirring and with controlled temperature (37°C) in a Perkin-Elmer luminescence spectrometer LS-5 equipped with a chart recorder model R 100A, with excitation and emission wavelengths of 500 and 530 nm, using 5 and 10 nm slits, respectively, for the two light pathways. Fluorescence was also routinely measured at 450-nm excitation (at this wavelength the fluorescence is proportional to intracellular dye concentration but is relatively pH-insensitive), and the value did not change more than 5% during the experimental period.
where ␤ t is the total intracellular buffering capacity, but in the nominally HCO 3 Ϫ -free solutions used in this study, ␤ CO 2 was assumed to be negligible, and ␤ t was therefore taken to be equal to ␤ i . ␤ i was determined using the NH 4 ϩ pulse technique as described previously (33) according to the formula, taking into account that NH 3 equilibrates across the cell membrane (i.e. [NH 3 ] i ϭ [NH 3 ] o ) and that the pK a of NH 4 ϩ (8.92) is the same intra-and extracellularly. [NH 4 Cl] in the absence of NH 4 Cl was taken to be 0.
Statistical analyses were performed using Student's t test for unpaired data, and p values Ͻ0.05 were considered significant. Data are presented as mean Ϯ S.D.

RESULTS
Effect of Virus Adsorption on the Glutathione Status-MDCK cells infected with Sendai virus show a time-dependent decrease in the intracellular GSH content (Fig. 1). GSH decrease began after a few minutes of infection and reached its maximum value in 20 min during virus adsorption (p Ͻ Ͻ 0.001). At the end of the virus infection (1 h) GSH content tended to increase; however, the values were always significantly lower than those observed in control cells (Fig. 1). A further decrease was observed in the following 24 h (Fig. 1, inset). In particular, the loss of GSH 3 h after infection was highly significant (p Ͻ 0.001) with respect to 1 h. At the end of 24 h, virus particles were detectable in the supernatant of infected cells, indicating that the virus had completed its replication. The same results were obtained by infecting the cells with purified Sendai virus. On the contrary, changes in the GSH content were not observed when treating the cells with allantoic fluid from uninfected cells (mock infection) at the same virus dilution (data not shown).
Since GSH decrease was a two-step process, we performed the experiments at 20 min, in the early phase of virus infection, and at 24 h, at the late stage of virus replication. In particular, 20 min referred to monolayers infected with Sendai virus only for this length of time; 24 h referred to cells infected for 1 h and refed with fresh medium for an additional 24 h. GSH depletion is commonly observed when cells are oxidatively stressed. It can usually be detected by measurement of transient increases in the intracellular content of either GSSG or protein glutathione mixed disulfides. To assess whether a typical oxidative burst was taking place under our experimental conditions, we measured the intracellular GSSG content and the GSH equivalents bound to proteins.
However, despite a massive and rapid GSH decrease, we were unable to detect any change in the concentrations of the two oxidized GSH derivatives after 20 min of incubation of MDCK cells with Sendai virus. The GSSG value in infected  (Table I). Thus, glutathione disappearing from the GSH pool of the infected cells is not recovered in its further redox form within the cell.
A steady state level of GSH in the cell results from a balance between the rates of synthesis and loss of the tripeptide via oxidation or excretion. A declining in GSH levels in the absence of oxidation can potentially result either from an increased efflux out of the cell or a reduced rate of uptake of precursors and subsequent resynthesis. Under our conditions the fall in GSH was too rapid to be accounted for by an inhibition of precursor uptake or biosynthesis. Moreover, MDCK cells treated with 1 mM buthionine sulfoximine, a specific inhibitor of glutathione synthesis, show a 50% decrease of GSH only after 18 h of treatment.
On the other hand, large amounts of glutathione were detected in the media of infected cells (Table I), predominantly as GSH (90%). Since the medium derived from infected cells was not able to reduce externally added GSSG (data not shown), it was concluded that during virus infection, glutathione is released by the cell in its reduced form. Furthermore, changes in the GSH content of the culture medium were not observed upon incubation with 5 mM DTT for 1 h, before acidification.
Effect of Virus Replication on the Glutathione Status-On the contrary, during virus replication (24 h) a significant amount of intracellular GSH was found to be bound to proteins (Table II). However, the GSH equivalents present under this form represented only 10% of the total glutathione of infected cells and did not account for the GSH depletion observed. Furthermore, GSSG was not increased (Table II), indicating that the decrease in GSH was not due to oxidation. GSH was detected in the medium of either infected or control cells, also in this case accounting for 90% of the total value (Table II). These results demonstrate that GSH is efficiently extruded from cells during the 24 h considered and that infected cells do not differ, in this respect, from control cells. In fact, the increased GSH values obtained in the infected medium were less significant with respect to the differences observed at 20-min incubations (Tables I and II). In addition, the values of medium GSH were only 3-fold higher than the ones detected in the medium at 20 min. This indicates that either the rate of GSH extrusion is significantly different at the two times examined or that GSH leaves the cells with different modality at different times.
Effect of Virus Infection on GSH Loss-In order to address the question whether GSH efflux from cells occurs by leakage or by carrier-mediated transport, we measured the activity of lactic dehydrogenase enzyme in the media of infected cells as an indicator of cell membrane perturbation. The data reported in Fig. 2A clearly demonstrate that outside enzyme activity can be detected at 20 min. However, the values represent only a small amount of the total enzyme present intracellularly, 2-3%. As assessed by trypan blue exclusion, cells were 98 -100% viable, the same level observed in uninfected cells.
An increase in the activity of this enzyme was also observed in the following 24 h of infection in the media of both control and infected cells. In this case the enzyme activity assayed in infected media was not significantly different from control media (Fig. 2B).
Several GSH-specific transporters were identified on the cellular and mitochondrial membranes of different cellular systems. These are specifically inhibited by L-methionine or cystathionine. MDCK cells were treated with 1 mM methionine for 1 h before virus infection and for 24 h after infection. The rate of GSH efflux was decreased in uninfected cells; in fact, a 30 and 70% inhibition of the efflux rate was obtained at 80 min (1 h before and 20 min during infection) and 24 h, respectively (Fig. 3), with a slight increase (10 -15%) in the intracellular GSH content at 80 min. However, the inhibitory effect was not observed in the infected cells at either 20 min or 24 h (compare Tables I and II with Fig. 3), in line with the result of an unaltered virus production obtained at 24 h as reported in Fig.  3, inset.
The results obtained point to a carrier-independent GSH extrusion. To further investigate this phenomenon, two different approaches were adopted. One approach was to infect MDCK cells at 4°C, a condition under which the virus is adsorbed to the cells but is not able to fuse with them. The other approach was to infect cells with Sendai virus unable to replicate because inactivated by UV radiation. GSH loss at 20 min was not observed when cells were infected with Sendai virus at 4°C (data not shown). On the contrary, the same extent of loss, in terms of GSH content and LDH activity, was observed when cells were infected with Sendai virus previously  inactivated (Fig. 4). Moreover, the GSH value present in the medium at 24 h was the same as that of control cells, indicating that, as expected, the UV-irradiated virus was not able to replicate. This was confirmed by the absence of hemagglutinating activity in the media (data not shown).

Effect of the Modulation of Glutathione Status on Virus
Replication-To investigate the correlation between the change in intracellular redox state, consequent to GSH loss, and virus replication, we treated the cell with different agents, able to modulate the intracellular GSH status, and subsequently quantified the changes in virus yield. Data shown in Fig. 5 indicate that the treatment of cells with 1 mM BSO for 18 h before infection and 24 h after caused an 80% decrease in GSH. This decrease was reflected by a double yield in virus production by cells. The opposite effect has been already shown in a previous work (21); treatment of cells with exogenous GSH, which is able to maintain unaltered the intracellular GSH content, inhibited virus replication by 90%. Moreover, incubation of the monolayers with 0.5 mM DTT, a reducing agent, during virus replication lowered significantly the virus titer and increased the GSH content (Fig. 5). On the contrary, 1 mM DTNB, a sulfhydryl reagent, increased viral production without significant change in the GSH content of the cells (Fig. 5).
Effect of Viral Infection and Glutathione Depletion on the Steady-state pH i of MDCK Cells-Chemical depletion of intracellular GSH produced a decrease in cellular pH in Ehrlich tumor cells (34). Moreover, lower intracellular pH values are preferred by several viruses for efficient replication (22,23). We therefore evaluated whether changes in the cells' pH value were operative during Sendai virus infection.
In the standard HEPES-buffered bicarbonate-free solution, the mean steady-state pH i of MDCK cells was 7.25 Ϯ 0.07 (n ϭ 15). A short term infection (20 min) of MDCK cells gave rise to a decrease of intracellular pH of about 0.20 pH units, whereas an infection ranging from 1 h up to 24 h did not give any significant change in basal intracellular pH evaluated as reported by Thomas et al. (32) (Table III). GSH depletion achieved by 18-h treatment with 1 mM BSO was similar to that obtained by 20-min infection and at the same time gave rise to a significant decrease of intracellular pH, in the same range as a 20-min viral infection (Table III).
Recovery of pH i from an Acid Load: Effect of Viral Infection-In order to assess whether the decrease of intracellular pH observed either with virus treatment or with chemical GSH depletion was due to an effect on the Na ϩ /H ϩ antiporter, we studied the ion transport activity under conditions of maximum activation, after an acid load.
The exposure of MDCK cells to 20 mM NH 4 Cl-containing HEPES-buffered solution increased pH i rapidly to approximately 0.30 pH units over the basal value (Figs. 6A and 7A), via rapid diffusion of NH 3 . During the exposure to NH 4 Cl (about 4 min), pH i tended to decrease toward the base line due to slow inward diffusion of NH 4 ϩ . Abrupt removal of NH 4 Cl from the medium rapidly decreased pH i by approximately 1.00 pH unit, with respect to the base line, due to NH 3 leaving the cell. The extent of this acidification was the same in all monolayers during the NH 4 Cl washout.
The readdition of Na ϩ buffer caused a recovery of pH i within about 10 min, in agreement with previously published data (35) and with a time course normally fitted by a single exponential. Therefore, a plot of ⌬pH i /⌬t (dpH i /dt) versus pH i is a straight line, intersecting the pH i axis at the final steady-state pH i (Figs. 6B and 7B).
In MDCK cells infected for 20 min, the rate of recovery was decreased, and the final pH i did not reach the initial value (Fig.  6A), whereas a 24-h viral infection did not give rise to any significant change in the recovery of intracellular pH (not shown), but only a trend to a decrease in the set point was observed (Table III and Fig. 6B). The same panel also shows the lack of recovery both in control and in 20-min infected cells due to EIPA (20 M), a derivative of amiloride, a specific inhibitor of the Na ϩ /H ϩ antiporter. The sensitivity to the amiloride derivative confirms that we are dealing with the basolateral Na ϩ /H ϩ antiporter having a housekeeping function (36). The regression lines, in the range 6.2-7.4, computed from data of experiments similar to those reported in Fig. 6A and 7A are reported in Fig.  6B as mean ⌬pH (ϫ 10 Ϫ4 pH units)/⌬t (s) (dpH i /dt) versus pH i during recovery from NH 4 -NH 3 pulse in MDCK cells. The regression lines for control and infected cells were parallel, but the set point of pH i was 7.29, 7.04, and 7.24 for control, 20-min infected, and 24-h infected cells, respectively (Fig. 6B). The same type of result was achieved when cells were depleted of GSH by BSO treatment (Fig. 7). In the case of BSO-treated cells, the slope of the regression line was also affected, in addition to the set point, with respect to control (Fig. 7B). The set point of pH i was 7.28 and 7.01 for control and BSO-treated cells, respectively. The data of the regression lines are in good agreement with those evaluated at equilibrium for the same treatment reported in Table III.
Both viral infection and BSO treatment did not affect the buffering capacity ␤ i , which was calculated to be at pH 6.8 about 86, 90, and 82 mM/pH unit for control, 20-min infected, and BSO-treated cells, respectively, in quite good agreement with previously published literature for the same cells (35), whereas an acute infection of MDCK cells resulted in an impairment of the Na ϩ /H ϩ antiporter function as shown by the lack of recovery from an acid pulse.
Cysteine Uptake-The preceding results demonstrate that GSH decreases during Sendai virus replication but with a less significant increase in the culture medium than that observed in the 20-min infection. Therefore, GSH metabolism in MDCK cells during virus replication was further investigated by labeling the intracellular GSH pool with [ 35 S]cysteine. During the 24-h labeling period, control and infected cells took up the same amount of 35 S, with significant changes in the trichloroacetic acid-soluble material; indeed, a decrease was observed in infected cells (Ϫ34%) (Table IV), in line with the intracellular GSH determinations. Furthermore, a significant amount of 35 S was found to be associated with viral proteins, as expected from the knowledge that Sendai envelope proteins contain a large number of cysteine residues. Interestingly, the loss of label in the acid-soluble material of infected cells was almost quantitatively recovered in the immunoprecipitated viral proteins (Table IV). DISCUSSION The results reported above demonstrate that GSH depletion is a direct consequence of viral infection.
It is usually assumed that GSH depletion reflects an intracellular oxidation; however, the results presented here suggest an alternative mechanism. GSH was lost from cells undergoing viral infection in a two-step process, it was mainly leaked out the plasma membrane rather than depleted by an oxidative process, as confirmed by unaltered intracellular GSSG content. A large amount of the tripeptide in its reduced form was, in fact, detected in the cell culture medium a few minutes after infection. The very rapid GSH loss from cells implied that the consequent intracellular decrease was not due to an inhibition in the uptake of GSH precursors and subsequent impairment of the synthesis. Moreover, chemical inhibition of the GSH synthesis by BSO led to very slow GSH decrease (50% reduction was reached by at least 18 h of treatment).
GSH decrease at the very early stages of virus infection was also observed during the infection of VERO cells with clinically isolated herpes simplex type 1 virus (15), indicating a more general relationship between oxidative stress and viral replication.
Sinusoidal efflux of GSH in the liver is markedly inhibited by low concentrations of methionine, and such inhibition is taken as evidence for the presence of a specific GSH carrier (37). We found that uninfected MDCK cells also extruded GSH, as shown by the increased thiol concentration in the medium of 24 h cultured cells, by a methionine-sensitive mechanism. At variance with this, no inhibition by methionine was observed in infected cells, ruling out an active extrusion process. Further, inhibiting the fusion process by infecting the cells at 4°C did not result in changes in GSH content. On the contrary, inhibiting viral replication by a preinactivation of Sendai virus by UV irradiation gave rise to changes, at 20 min, comparable with those observed in cells infected by the active virus, confirming that the major loss of GSH was associated with the fusion process.
The GSH loss observed at 20 min imposes an oxidative stress that can impair other cellular functions, in particular those regulated by the redox mechanism. We found that GSH leak-age was associated with a decrease in intracellular pH, consistent with a previous study that demonstrated lower intracellular pH value in Ehrlich-ascites tumor cells chemically depleted of GSH (34). Moreover, under our conditions, an 18-h treatment of monolayers with BSO reduced the intracellular GSH to values similar to those mediated by a 20-min virus infection, leading to a concomitant decrease in the pH.
Cells regulate their pH through a variety of ion transport mechanisms, including Na ϩ /H ϩ antiporter and several bicarbonate exchangers (38). The Na ϩ /H ϩ antiporter is a major H ϩ extruding mechanism, which is driven by the inwardly directed Na ϩ gradient (18). This antiporter has a high affinity for H ϩ at  pH 6. A wide variety of external signals enhances the affinity of the antiporter for H ϩ at neutral pH, leading to alkalinization of the cells. MDCK cells were found to regulate their pH i primarily via the Na ϩ /H ϩ antiporter, as demonstrated by the ability of EIPA to inhibit recovery from an acid load (Fig. 6A). Such recovery was not observed in infected cells at 20 min, suggesting that a modification of the Na ϩ /H ϩ antiporter was respon-sible for the acidification occurring at the early stage of infection. Interestingly, cells treated with BSO showed the same impairment in the function of the antiporter. In this context it has to be mentioned that oxidative stress decreases the Na ϩ /H ϩ antiporter activity in bovine pulmonary artery endothelial cells (20).
It has been demonstrated that acidic conditions favor many viral infections by accelerating the fusion process, thus enhancing viral replication (22,23). Sendai virus (hemagglutinating virus of Japan) belongs to the Paramyxovirus family and is believed to infect its host cells directly by fusion of its envelope with the cell plasma membrane (39). The rate of fusion of Sendai with several cellular systems was found to be optimal at pH 7.0 (40). This was in good agreement with the drop of pH i observed at 20 min of infection. Furthermore, by decreasing intracellular GSH and pH i upon BSO treatment we obtained 2 times higher virus production in 24 h. These results confirm that a lower pH i value in comparison with that observed in control cells favors Sendai virus replication.
Despite the large loss of GSH and the consequent severe impairment in the Na ϩ /H ϩ antiporter, cell death was not ob-  served in infected cells, even if there is convincing evidence that T lymphocytes infected with HIV exhibit an enhanced susceptibility to undergo apoptosis (programmed cell death) (41). Under our conditions MDCK cells are still viable after 48 -72 h of infection. A recovery from the initial oxidative conditions was observed at 1 h of infection as evidenced by the rise in both intracellular pH and GSH content. In fact, 1 h corresponds to the end of the viral adsorption period, thus confirming the strict association of early infection steps with GSH loss and pH i drop.
A completely different scenario is operative during virus replication. In this second phase of the virus growth program, prooxidative conditions are still in action as shown by a further GSH decrease. However, in this case GSH loss was not associated with significant GSH leakage from cells, changes of LDH activity, impairment of the Na ϩ /H ϩ antiporter, and, finally and most importantly, significant changes in the pH i value. Moreover, mixed disulfides were formed in significant amounts in infected cells. In other words, except for the presence of mixed disulfides, cells at 24 h after infection were more similar to control cells with respect to cells at 20 min after infection.
The major cause of GSH loss in the second phase was shown to be the preferential cysteine incorporation into viral proteins, which are very rich in this amino acid (42,43). In fact, it was found that the decrease of labeled thiols in the trichloroacetic acid-soluble material was almost totally compensated for by those found in immunoprecipitated viral proteins. In the second phase, viral replication and GSH content could be finely modulated by treating the cells with agents known to alter the sulfhydryl status. DTT, a reducing thiol, was able to markedly decrease virus production by maintaining higher GSH concentrations. The higher GSH content may derive from the inhibition of the formation of mixed disulfides, which in turn may produce inactive virus by inhibiting the folding process of viral proteins. On the other hand, DTNB, a mixed disulfide-forming agent, gave rise to higher virus yield with a slight decrease in GSH content. These results were in good agreement with our previous data on the antiviral effect exerted by externally added GSH (21).
In conclusion, the data presented in this report may give a clue to interpreting some critical points of the complex interrelationship between GSH status and viral replication, in particular the contradictory results reported in the literature about HIV replication. The rapid biphasic drop in GSH content, leading to changes in the redox state in a very short time, may easily lead to different cell responses. Our data also give a rationale for the oxidative stress mediated by the Tat protein, which is secreted from HIV-infected cells. Tat protein is able to permeate into uninfected lymphocytes, depleting the cells of GSH and inhibiting the manganese-superoxide dismutase expression and activity (44), thus behaving as a "minivirus" with fusion modality analogous to that observed for Sendai virus.