Decrease of H₂O₂ Plasma Membrane Permeability during Adaptation to H₂O₂ in Saccharomyces cerevisiae^*

In general, the first cellular defense against exogenous toxic agents is the plasma membrane, which limits the influx of these agents into cells. For lipophilic agents that cross the plasma membrane, the cells evolved exporting mechanisms to expel the agent, and acquired drug resistance is often associated with the expression of drug efflux pumps such as the P-glycoprotein and the multidrug resistance protein (1). Concerning H₂O₂, the most abundant cellular reactive oxygen species, it is widely believed that this agent crosses biomembranes freely, and so mechanisms that impose an extracellular/intracellular gradient for this species have been ignored so far, being implicitly assumed that such gradient does not occur.

However, in extracellular fluids concentrations up to 100 μm H₂O₂ have been reported (2), whereas the intracellular H₂O₂ concentration is estimated in the range 0.01-0.1 μm (3, 4), implying the existence of an outside/inside gradient. Supporting this, we showed that H₂O₂ does not permeate biomembranes freely in a human cell line and that upon exposure to external H₂O₂ the intracellular consumption of H₂O₂ catalyzed by antioxidant enzymes is able to generate a gradient of H₂O₂ across the plasma membrane, resulting in a lower H₂O₂ concentration in the intracellular milieu (5). These results were later confirmed in an Escherichia coli strain by others (6). Furthermore, if an extracellular/intracellular gradient is not formed, the intracellular enzymes are not able to protect individual cells against H₂O₂, and it was shown in an E. coli strain that does not show a gradient that the increase in catalase activity only protects high density or colonial E. coli and does not protect low density or individual E. coli (7). Further supporting the importance of the plasma membrane is the observation that after a nonlethal dose of H₂O₂, the levels of mRNA that are more repressed are those corresponding to the enzyme HMG-CoA reductase (8). This is the rate-limiting enzyme of the synthesis of sterols, which are key components of the plasma membrane controlling its fluidity (9) and consequently the permeability of this membrane to a species like H₂O₂ (10).

The possibility that the plasma membrane protects cells against external H₂O₂ is important because, in most cases, oxidative stress is expected to occur as a result of cells being exposed to an exogenous source of H₂O₂ (2) such as that produced from environmental factors (i.e. redox-cycling agents and radiation) in unicellular organisms or inflammation and injury responses, which in humans are associated with the etiology of several vascular diseases such as atherosclerosis, diabetes, neuronal disorders, and ischemia reperfusion injury (11). On the other hand, intracellular production of H₂O₂ is expected to have mainly a regulatory role (12) (e.g. regulation of cell proliferation (13)), although in some pathological situations, like aging, high levels of oxidative stress may be expected to be generated endogenously (14).

In this work, we tested the hypothesis that the plasma membrane is a key cellular site protecting cells against external H₂O₂. If the hypothesis is correct, then the adaptation to H₂O₂, i.e. the induction of resistance to a high level (usually lethal) of H₂O₂ by a preliminary low (adaptive) dose of H₂O₂, could involve alterations in the plasma membrane permeability to H₂O₂. Therefore, we carried out rigorous kinetic studies on the adaptation to H₂O₂ in Saccharomyces cerevisiae (Sc)¹ cells growing in exponential phase, a well known model for adaptation to oxidative stress (15). We further investigated the susceptibility to H₂O₂ in two mutant Sc strains with impaired ergosterol biosynthesis, which show higher membrane permeability to lipophilic agents. We concluded that the plasma membrane is important in the protection against H₂O₂ and that its properties are subjected to regulation leading to a decrease of the permeability coefficient to H₂O₂ during adaptation to this agent.

EXPERIMENTAL PROCEDURES

Materials—Sc strains used in this work are Y00000 (wild type, genotype BY4741 MATa; his3Δ1; leu2Δ0; met15Δ0; ura3Δ0), Y02667 (erg3Δ, isogenic to BY4741 with YLR056w::kanMX4), Y00568 (erg6Δ, isogenic to BY4741 with YML008c::kanMX4), and Y04718 (ctt1Δ, isogenic to BY4741 with YGR088w::kanMX4), and they were obtained from EUROSCARF (Frankfurt, Germany).

Yeast extract, bactopeptone, yeast nitrogen base, and agar were from Difco (Detroit, MI). Glucose oxidase (Aspergillus niger) and digitonin were from Aldrich. Bovine liver catalase, lyticase (Arthrobacter luteus), phenylmethylsulfonyl fluoride, and cytochrome c were from Sigma. Hydrogen peroxide was obtained from Merck, pyruvate was from Fluka (Buchs, Switzerland), and NADH was from Roche Applied Science.

Media and Growth Conditions—Sc cells were inoculated at an A₆₀₀ of 0.05 and cultured in synthetic complete medium (6.8% (w/v) yeast nitrogen base, 2% (w/v) glucose, 0.002% (w/v) arginine, 0,002% (w/v) methionine, 0.003% (w/v) tyrosine, 0.003% (w/v) isoleucine, 0.003% (w/v) lysine, 0.005% (w/v) phenylalanine, 0.01% (w/v) glutamic acid, 0.015% (w/v) valine, 0.01% (w/v) aspartic acid, 0.0025% (w/v) adenine, 0.04% (w/v) serine, 0.01% (w/v) leucine, 0.005% (w/v) tryptophan, 0.01% (w/v) histidine, 0.02% (w/v) threonine, and 0.0025% (w/v) uracil) at 30 °C with shaking at 160 rpm. For all experiments, the cells in the exponential phase were harvested at A₆₀₀ = 0.5 (1 A₆₀₀ = 2-3 × 10⁷ cells).

Cell Permeabilization—Cell membrane permeabilization was achieved by incubating cells with 0.01% (w/v) digitonin dissolved in dimethyl sulfoxide in 0.1 m potassium phosphate buffer, pH 6.5, for 5 min at 30 °C with shaking. Permeabilization was checked by means of lactate dehydrogenase activity measurement, according to Ref. 16.

Preparation of Spheroplasts—Spheroplasts were obtained by treating cells for 30 min at 30 °C with lyticase (40 units/ml) in 0.1 m potassium phosphate buffer, pH 6.5, containing 1 m sorbitol, 6.8% (w/v) yeast nitrogen base, and 2% (w/v) glucose. Spheroplast formation was checked by following the decrease of absorption at 660 nm, according to Ref. 17.

Exposure to and Measurement of H₂O₂ and Cell Survival—The cells were exposed to steady state hydrogen peroxide concentrations in synthetic complete medium at 30 °C and with shaking at 160 rpm, using glucose oxidase as described in Ref. 18. In brief, steady state levels of H₂O₂ were obtained by adding an initial amount of H₂O₂ together with some glucose oxidase that, by forming H₂O₂, compensated for the consumption of H₂O₂ by the cells. The consumption of H₂O₂ in Sc cells shows first order decay kinetics with a rate constant of 0.059 min^-1 A₆₀₀^-1 (see “Results”). By balancing the initial additions of H₂O₂ and glucose oxidase, a steady state concentration range of 0.15-1 mm was generated.

To establish the sublethal hydrogen peroxide dose, the cells were exposed to steady state concentrations of hydrogen peroxide (150-400 μm), up to 90 min at 30 °C and with shaking at 160 rpm. Cell survival was monitored by taking samples at 30-min intervals, diluting them, and plating aliquots on YPD plates (1% (w/v) yeast extract, 2% (w/v) bactopeptone, 2% (w/v) glucose with 2% (w/v) agar) and counting colonies after 48-72 h. The sublethal dose was defined as the dose that induced about 10% loss in cell viability when compared with control (150 μm under our conditions).

H₂O₂ was measured as O₂ release with an oxygen electrode (Hansatech Instruments Ltd., Norfolk, UK) following the addition of catalase (18). Glucose oxidase activity was measured by following O₂ consumption with the oxygen electrode.

Enzyme Activities and H₂O₂ Consumption—Crude extracts were prepared by glass bead lysis as described in Ref. 19. Determinations of total protein were done according to Peterson (20). H₂O₂ catabolism in Sc cells has some differences when compared with higher eukaryotes, namely: two catalases are present, the cytosolic catalase T encoded by the gene CTT1 and the peroxisomal catalase A encoded by the gene CTA1 (15); in the mitochondrial intermembrane space a cytochrome c peroxidase, encoded by the gene CCP1, uses cytochrome c to reduce H₂O₂ (21); and the glutathione peroxidases are phospholipid hydroperoxide glutathione peroxidases (encoded by the genes GPX1, GPX2, and GPX3) and do not have selenium in the active center (22). We focused in catalase and cytochrome c peroxidase activities because these are the most important enzymes in the H₂O₂ catabolism for SC cells growing in exponential phase (23).

Cytochrome c peroxidase activity was measured in protein extracts by following spectrophotometrically the oxidation of cytochrome c at 550 nm (ϵ₅₅₀ = 29 mm^-1 cm^-1) at 25 °C for 4 min by adding 50 μm H₂O₂ (24). Ferrocytochrome c was prepared as described in Ref. 24. One unit is defined as the quantity of enzyme that catalyzes the oxidation of 1 μmol of cytochrome c/min at 25 °C and pH 7.4. Catalase activity in protein extracts was measured spectrophotometrically by following H₂O₂ consumption (initial concentration, 10 mm) at 240 nm at 25 °C for 2 min according to Ref. 25.

The assays performed to measure H₂O₂ consumption rate, either overall or only catalase-driven, were as follows: (a) For catalase activity in situ, H₂O₂ consumption (initial concentration, 100 μm) by permeabilized cells suspended in the permeabilization buffer at 30 °C was followed using the oxygen electrode. This assay can also be considered a measurement of the overall H₂O₂ consumption rate in permeabilized cells because catalase is the only active enzyme (see below). (b) For catalase activity in intact cells, the activity should be called apparent because it is partially limited by the plasma membrane, being thus lower than the real activity (see “Results”). A similar approach to the first assay was used with the following alterations: no digitonin was added to the buffer, and the cells were incubated with 150 μm H₂O₂ for 15 min before starting the measurements of H₂O₂ consumption, to oxidize internal pools of reducing equivalents. (c) For overall H₂O₂ consumption rate in intact cells, the rate was measured as in the second assay, but with cells suspended in synthetic complete medium (instead of buffer), and measurements were started immediately after adding 100 μm H₂O₂.

Because in the ctt1Δ strain no activity was detected with either the first or second approach but an H₂O₂ consumption rate was observed with the third assay, the first and second assays measured only cytosolic catalase activity, whereas in the third assay other enzymes were also active. Therefore, under our experimental conditions peroxisomal catalase did not represent an important H₂O₂ removing activity. The reason why enzymes other than catalase were not active in the first and second assays is probably the rapid depletion of internal pools of reducing equivalents necessary for these enzymes because of either the plasma membrane permeabilization (in the first assay) or the exposure to a 15-min preoxidation period with H₂O₂ (in the second assay). After being oxidized, they cannot be reduced back because of the lack of carbon sources in the buffer. In the third assay, the pools of reducing equivalents can be maintained in a quasi steady state, because incubation is done with intact cells in the presence of carbon sources.

In all cases, H₂O₂ concentrations were plotted semi-logarithmically against time, and catalase activity (or H₂O₂ consumption) was calculated as the slope of the linear fitting (i.e. as a first order rate constant).

Determination of Cellular H₂O₂ Gradient—The gradient generated by catalase was determined using the principle of enzyme latency (see “Results”). The ratio between apparent catalase activity in intact cells and catalase activity in permeabilized cells was used.

Statistical Analysis—The results presented are the means ± S.D. of independent experiments. Data statistical analysis was undertaken using either a two-tailed Student t test for comparison between means of two different groups or by using analysis of variance and the Tukey-Kramer multiple comparisons test for comparison of more than two different groups.

RESULTS

Adaptation to H₂O₂ Does Not Increase H₂O₂ Removal—Fig. 1A shows that a 150 μm steady state H₂O₂ concentration triggered an adaptation in Sc cells. In accordance with the data in the literature (23), during this adaptation there were approximate 2- and 3-fold increases in two antioxidant enzymes activities, catalase and cytochrome c peroxidase, respectively (Table I). These enzymes are accountable for the removal of H₂O₂ in the cell (23), and so it could be expected that their induction increases the capacity of Sc cells to remove H₂O₂. Surprisingly, this was not observed (Fig. 1B), and the pseudo-first order rate constant describing the consumption of H₂O₂ by Sc was similar in control and adapted cells (Table I). Therefore, contrary to what is widely believed, an increased removal of external H₂O₂ is not part of the mechanism leading to resistance to external H₂O₂ in Sc.

H₂O₂ Does Not Diffuse Freely into Sc Cells—One possible explanation for the lack of increase in H₂O₂ consumption rate constant by Sc H₂O₂-adapted cells is that H₂O₂ does not freely diffuse into Sc cells, because of a permeability barrier either at the level of the plasma membrane or the cell wall. If this is the case, then the overall H₂O₂ consumption rate in Sc cells is not only determined by the intracellular capacity to remove H₂O₂, but it is also dependent on the permeability of Sc cells to H₂O₂.

To test the hypothesis that the diffusion of H₂O₂ into Sc cells limits the overall H₂O₂ consumption rate, the role of both the cell wall and the plasma membrane as barriers to the diffusion of H₂O₂ were studied. Firstly, spheroplasts (i.e. cells with removed cell wall) from H₂O₂-adapted and control cells were obtained. Despite the increased catalase and cytochrome c peroxidase activities in spheroplasts produced from adapted cells, the H₂O₂ consumption rate consumption was similar in spheroplasts produced from control and from adapted cells (not shown). Therefore, the cell wall does not limit H₂O₂ diffusion into Sc cells, as could be expected because the cell wall has pores that are permeable to low molecular weight molecules (26). Our observation is also consistent with the fact that Sc cell susceptibility to H₂O₂ is not altered by deleting genes ECM25, ECM33, and YOR275c, which are involved in cell wall integrity (27).

Next, the hypothetical role of the plasma membrane in limiting the diffusion of H₂O₂ was analyzed by selective permeabilization of the plasma membrane with a low dose of digitonin. The H₂O₂ consumption rate constant in digitoninpermeabilized adapted cells (0.097 ± 0.009 min^-1 A₆₀₀^-1, n = 5) was ∼2-fold higher than in permeabilized control cells (0.048 ± 0.004 min^-1 A₆₀₀^-1, n = 8). Therefore, it can be concluded that in fact the diffusion of H₂O₂ across the plasma membrane limits, at least partially, the overall H₂O₂ consumption rate in Sc cells.

H₂O₂ Gradients Are Higher in H₂O₂-adapted Sc than in Control Sc Cells—According to the principle of enzyme latency, a direct consequence of the limited diffusion of H₂O₂ across the plasma membrane is the formation of gradients between the extracellular and the intracellular milieu when cells are exposed to external H₂O₂ (5), i.e. the intracellular concentration is lower than the extracellular concentration. This principle states that an enzyme entrapped in a compartment shows a lower activity than when it is free in solution because of the permeability barrier made up by the compartment that limits the diffusion of the substrate to the enzyme (28), and from it Equation 1 is derived (28, 29), in which k_perm and k_catabolism refer to the first order rate constants for the permeation of H₂O₂ across the plasma membrane and to the intracellular catabolism of H₂O₂, respectively, and R refers to the ratio between the overall H₂O₂ consumption rate constant in intact cells over the consumption rate constant in permeabilized cells.

Despite its simplicity, three important biological implications can be obtained from the analysis of Equation 1: (a) If k_perm >> k_catabolism, then [H₂O₂]_in = [H₂O₂]_out, i.e. there is no gradient; but if k_perm is limiting the consumption of H₂O₂ (either partially or totally), as observed in Sc cells, a H₂O₂ gradient is formed. (b) If k_catabolism increases, the gradient is also increased; so we can expect that in H₂O₂-adapted Sc cells, which showed an increase in the activities of catalase and cytochrome c peroxidase (Table I), there is a gradient that is larger than in control cells. 3) The determination of this gradient can be based on the experimental measurement of R (5), which is very helpful because a direct measurement of H₂O₂ inside the cells is not available.

To determine R, the overall H₂O₂ consumption rate constants both in intact cells and in permeabilized cells have to be measured. Experimentally, it was difficult to measure the overall rate of consumption of H₂O₂ in disrupted cells because after a short period where a rapid and changing consumption rate was observed, the rate of consumption decreased and then remained constant (not shown). Catalase is responsible for this constant rate because this consumption was not observed in the ctt1Δ mutant (not shown). Therefore, only the gradient generated by the action of catalase could be measured accurately. In this case, Equation 1 is transformed in Equation 2, where k_catalase refers to the first order rate constant describing the intracellular consumption of H₂O₂ by catalase (i.e. catalase activity) and R_catalase refers to the ratio between the apparent activity of catalase in intact cells and the activity in permeabilized cells. Because k_catalase < k_catabolism, the gradient caused by catalase is lower than the gradient when all antioxidant enzymes removing H₂O₂ are active, but it is a good indication for relative values of the overall gradient when comparing different cells.

As can be seen in Table II, the gradient is significantly higher in H₂O₂-adapted Sc cells than in control cells. Therefore, when exposed to the same external H₂O₂ concentration, adapted cells will endure a lower intracellular H₂O₂ concentration than control cells. Thus, the increase in catalase and cytochrome c peroxidase activities does not cause a higher rate of H₂O₂ disposal but the formation of a steeper gradient of H₂O₂ across the plasma membrane.

H₂O₂ Adaptation Decreases Plasma Membrane Permeability to H₂O₂ in Sc Cells—Having established the rate-limiting role of the plasma membrane for the overall H₂O₂ consumption and the formation of H₂O₂ gradients across this membrane, we next investigated whether the permeability to H₂O₂ is changed during adaptation. If the existence of this permeability barrier is important to provide protection against H₂O₂, it could be expected that during H₂O₂ adaptation changes in the plasma membrane occur to decrease H₂O₂ diffusion into the cell. To test this hypothesis we used Equation 2 to calculate k_perm in control and H₂O₂-adapted cells (Table II). k_perm in H₂O₂-adapted cells decreased ∼2-fold when compared with control cells, indicating that H₂O₂-adapted cells are less permeable to H₂O₂.

Therefore, in addition to the well known increase in antioxidant enzyme activities during H₂O₂ adaptation, we showed for the first time that in Sc cells, plasma membrane properties are altered, making it less permeable to H₂O₂ and thus protecting Sc cells against H₂O₂. Two important biological consequences of the combined effect of increased catalase and cytochrome c peroxidase activities and decreased plasma membrane permeability to H₂O₂ are: (a) a significant increase in the gradient between extracellular and intracellular H₂O₂ during adaptation to H₂O₂, because k_catabolism is increased and k_perm is decreased in Equation 1 and (b) unaltered H₂O₂ overall consumption rate in adapted cells, because the increased intracellular capacity for H₂O₂ removal is compensated by the decreased diffusion of H₂O₂ into the cell, as observed experimentally (Table I).

Cycloheximide Causes H₂O₂ Adaptation and Decreases Plasma Membrane Permeability to H₂O₂ in Sc Cells—If the decrease in the plasma membrane permeability to H₂O₂ is an important regulatory response during adaptation to H₂O₂, then other agents that induce resistance to H₂O₂ could also decrease cell permeability to H₂O₂. Cycloheximide, a protein synthesis inhibitor, increases Sc cells resistance to H₂O₂ (30). The molecular mechanism involved is unknown, and the observation has been difficult to interpret because upon incubation with cycloheximide, catalase and cytochrome c peroxidase, as well as other proteins thought necessary for H₂O₂ adaptation, are probably not induced because of the inhibition of protein synthesis. Therefore, we tested whether cycloheximide-induced plasma membrane changes could explain the observed adaptive effect to H₂O₂. In fact, under conditions where a strong adaptive effect was observed (Fig. 1), plasma membrane permeability to H₂O₂ also decreased from 0.083 ± 0.028 min^-1 A₆₀₀^-1 to 0.042 ± 0.013 min^-1 A₆₀₀^-1 (n ≥ 8). This decrease is similar to that observed when cells are exposed to an adaptive dose of H₂O₂, and because cycloheximide is not an oxidant, this result is particularly relevant to support the possibility that the regulation of the plasma membrane permeability is a general mechanism by which cells acquire resistance to H₂O₂.

Changes in Ergosterol Composition Make Sc Cells More Susceptible to H₂O₂—To further test the importance of the plasma membrane in the protection against H₂O₂, we compared the susceptibility to H₂O₂ of two Sc strains, which have defective membranes, with that of the wild type Sc strain. The two strains used (erg3Δ and erg6Δ) have a defect in the pathway of ergosterol biosynthesis: erg3Δ mutants lack a C-5 desaturase producing ergosta-7,22-dienol instead of ergosterol (31), and erg6Δ mutants lack a C-24 methyltransferase producing zymosterol and colesta-5,7,24-trienol instead of ergosterol (32). In both mutants the membrane biophysical properties are changed, resulting in increased permeability to lipophilic compounds (9). As can be observed in Table III, these mutants show a higher H₂O₂ consumption rate constant when compared with wild type cells, despite a similar cytochrome c peroxidase activity and similar or lower catalase activity in permeabilized cells. This can be interpreted by increased membrane permeability to H₂O₂. To further confirm this, the H₂O₂ gradient in these mutants was calculated by a similar approach to that used for wild type cells. As shown in Table III, in erg3Δ and erg6Δ cells catalase activities both in intact and in permeabilized cells are identical, i.e. catalase-driven gradients are not produced; thus, the plasma membrane permeability to H₂O₂ is increased in these mutants. k_perm could not be calculated because in the absence of a gradient the only information that could be obtained was that k_perm >> k_catalase.

If the gradients are in fact important, then it could be expected that both mutants have a similar susceptibility to H₂O₂ and that this susceptibility is higher than that shown by the wild type strain. As can be seen in Fig. 2, this was the observed behavior when cells were subjected to H₂O₂ steady state incubations. This confirms the role of the plasma membrane for the protection against H₂O₂.

DISCUSSION

Fundamental biochemical aspects regarding cellular protection against H₂O₂ were made clear in this work and changed our view on the role of the plasma membrane as a protection barrier against H₂O₂. Contrary to the commonly accepted concept that H₂O₂ diffuses freely across biomembranes, we showed that this is not the case in Sc cells and that as consequence gradients are formed across the plasma membrane when Sc cells are exposed to exogenous H₂O₂. Most importantly, the role of the plasma membrane in the generation of gradients, by constituting a barrier against the diffusion of H₂O₂, is not a passive one because its permeability to H₂O₂ is subjected to regulation, as shown by our studies of adaptation to H₂O₂. In fact, upon acquisition of H₂O₂ resistance, by exposure either to a nonlethal adaptive dose of H₂O₂ or to cycloheximide, permeability toward H₂O₂ is decreased, increasing the plasma membrane protective role against H₂O₂.

Concerning the detailed molecular mechanisms that are responsible for the change in the plasma membrane permeability properties, they were not addressed in this work. The plasma membrane of Sc cells is a complex biological site, and a multitude of factors can change its permeability properties. Four observations, however, point to a repression of the ergosterol biosynthesis pathway as a possible mechanism: (a) for a non-lethal dose of H₂O₂, the levels of mRNA that are more repressed are those corresponding to the enzyme HMG-CoA reductase (8), the rate-limiting enzyme in the ergosterol biosynthesis pathway; (b)H₂O₂ inhibits the enzyme HMG-CoA reductase (33); (c) cycloheximide, a protein synthesis inhibitor that can possibly lead to a lower steady state level of HMG-CoA reductase, decreases plasma membrane permeability to H₂O₂ (this work); and (d) erg3Δ and erg6Δ, which accumulate altered sterols (9) and have the ergosterol biosynthesis pathway up-regulated (31, 34), show increased permeability to H₂O₂ (this work).

Our results conciliate the contradictory findings in the literature concerning whether catalase is important in the adaptive response of cells at low densities; it was shown in an E. coli strain that an increase in catalase activity does not protect low density or individual E. coli (7), which was attributed to the inability of this strain to form an extracellular/intracellular H₂O₂ gradient; however, in Sc cells catalase was found to be important even at low density conditions, a discrepancy that was not explained (23). If the gradient concept is taken into account this behavior is expected because the imposition of a gradient is much more difficult in E. coli than in Sc cells because the surface to volume ratio is much higher in the relative smaller E. coli cells than in the relative larger Sc cells. The H₂O₂ gradient in Sc cells found in the present work supports this explanation.

Concerning biomembranes other than the plasma membrane, it is expectable that they also constitute permeability barriers to H₂O₂, and in fact we have estimated a gradient of 3 across the peroxisomal membrane in Jurkat T-cells (5). The concept of intracellular gradients of H₂O₂ has important implications for cellular redox compartmentalization (35). For example, it helps to understand how it is possible to achieve simultaneously oxidative conditions in cytosol and reductive conditions in the nucleus, which are necessary for activation of NF-κB by H₂O₂ (36, 37), or how a mild oxidative environment in the endoplasmic reticulum necessary for a correct protein folding (38) can be maintained in a overall reductive cellular environment.

In addition to the well recognized pathological relevance of extracellular H₂O₂ in inflammation and injury responses (11), another situation in which extracellular H₂O₂ was also a potential major hazard to the cells, constituting a possible evolutionary pressure for the origin of the eukaryotic cell, was during the pre-Cambrian Earth. Approximately between 2 and 3 billion years ago two events occurred: (a) the slow conversion of a small prokaryote ancestor into a large phagocytic cell possessing most of the characteristics of the modern eukaryotes and able to acquire endosymbionts (39, 40) and (b) massive generation of H₂O₂ resulting from the oxidation of the abundant Fe²⁺ ions to Fe³⁺ by the O₂ produced by cyanobacteria, which lead to the formation of ferric oxide deposits (41). H₂O₂ concentrations in the aqueous environment during this period have been estimated in the millimolar range (7). Taking into consideration that during this period cellular iron-chelating systems would not be yet fully optimized, H₂O₂ would be extremely hazardous to cells, because upon diffusion into the cell it would easily react with Fe²⁺ producing hydroxyl radicals. Supporting this view, catalase has been described as an ancient enzyme, and many of the primitive cells would already have this enzyme (42). However, in the absence of a gradient, catalase does not protect individual cells (7), and therefore smaller cells, where the gradient is either nonexistent or smaller, would be more susceptible to H₂O₂. The driving forces for the increase in the cell volume during the slow conversion of prokaryotes to eukaryotes remain largely unknown, and we speculate, that the protective effect of the formation of a steeper H₂O₂ gradient in larger cells may have been a driving force for the increase in the cell volume and consequently the appearance of eukaryotic cells.

In conclusion, there are now several studies, performed in a range of organisms, E. coli (6), Sc cells (this work), and human cell lines (Jurkat T-cells (5); MCF-7),² that clearly show that H₂O₂ does not diffuse freely across plasma membranes. Furthermore, we showed here for the first time that the plasma membrane is not a passive actor in the defense against H₂O₂ but is subjected to regulation to fulfill its protective role. Therefore, more attention should be paid to the role of biomembranes when studying the biological actions of H₂O₂, and the common assumption that H₂O₂ diffuses freely across biomembranes should be avoided.