Effect of Yeast Killer Toxin on Sensitive Cells of Saccharomyces cerevisiae *

Killer toxin from Saccharomyces cerevisiae inhibited the pumping of protons into the medium by metabol- ically active sensitive cells. Such inhibition coincided with that of the uptake of potassium ions which are thought to be accumulated by yeast cells in order to neutralize the membrane potential created because of the extrusion of protons. The consumption of glucose, however, was identical in killer-treated and untreated cells. These alterations can be explained by the ability of the toxin to reduce the chemical proton gradient across the plasma membrane as measured by the accumulation of the weak permeable [14C]propionic acid. With this method, an internal pH of 6.42 was calculated from normal cells (the external pH was 4.6) while that of toxin-treated cells was decreased as a function of time. The proton concentration gradient was reduced from 66- to 17“fold. It is shown that the toxin-induced alteration of the proton gradient is due to an enhanced proton permeability of the yeast plasma membrane upon binding of the toxin. It is suggested that killer toxin acts as a macromolecular proton conductor sim- ilar in some respects to the known proton conductors 2,l-dinitrophenol and carbonyl cyanide m-chlorophen- ylhydrazone, since all the described effects are also observed with these substances.

Killer strains of yeast contain a double-stranded RNA (1) which codes for an extracellular protein (2). This protein kius sensitive cells upon binding to the cell surface. Although the killer phenomenon in yeast was discovered as early as 1963, ' very little has been known about the mechanism of action until recently. It was known that killer toxin induced a leakage of ATP (3) and potassium ions (4). However, these effects as well as the toxin-induced inhibition of macromolecular processes were observed in the late phase of killing, by which time 7040% of the cells were already dead (5). For this reason it was suspected that the observed effects were really a final consequence rather than a primary effect of the toxin. Recent results from our laboratory (6) have shown what appear to be primary effects of yeast killer toxin, namely the inhibition of the proton-amino acids symport, as well as the inhibition of proton pumping by metabolically active cells. In this paper we extend these observations to show that the inhibition of proton efflux is coupled to inhibition of potassium influx, and * This work was supported in part by a grant from the Comision Asesora para la Investigacion Cientifica y Tecnica. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
To whom all correspondence should be sent.

1, 203.
we provide evidence that the toxin increases the proton permeability of yeast plasma membrane, thus partially disrupting the chemical gradient of protons across it. The results are discussed in the light of previously observed effects of the toxin, and with regard to the mechanism by which the toxin could act by disrupting an energized membrane state.

Materials
Yeast Strains-All the strains of Saccharomyces cerevisiae used were from the laboratory of Dr. G. Fink (Cornell University, Ithaca, NY), and were a generous gift of Dr. J. Conde (Cervezas Cruzcampo, Sevilla, Spain). Strain X 17/17 (a, his 1) was used as sensitive strain.
Strain A 8207 B was used as killer producer and the nonkiller strain derived from it, A 8207 B-NK 1 (both a, his 4) (7), was used as control.
Chemicals-["C]Propionic acid was obtained from The Radiochemical Centre, Amersham. C3H]H2O and ['4C]methoxy inulin were from New England Nuclear. Yeast extract and peptone were from Difco. DNP,' DCCD, and glucose were obtained from Merck. CCCP was from Calbiochem, and 2-deoxyglucose, antimycin A, and EDC from Sigma, CCCP, DNP, DCCD, and antimycin A were dissolved in 96% ethanol.

Methods
Preparetion of Toxin-Partially purified killer toxin was obtained from cultures of strain A 8207 B grown at 22 "C in a medium containing 1% yeast extract, 2% peptone, and 2% glucose supplemented with 400 mg/liter of adenine, 30 mg/liter of histidine, and 30 mg/liter of leucine which had been previously filtered through a Diafio PM 30 membrane. The medium was buffered with 0.25 M citrate phosphate, pH 4.7. Culture supernatants were fdtered as above followed by precipitation with 80% ammonium sulfate. The precipitate was resuspended in 5 mM citrate phosphate, pH 4.7, or 10 mM e-aminocaproic-HC1, pH 4.6, as indicated. Identical preparations from culture supernatants of the nonkiller strain A 8207 B-NK 1 were used throughout the work. When necessary, killer toxin activity was assayed using the well test method of Woods and Bevan (8) as previously described ( 6 ) .
Determination of Glucose and ATP-Glucose was determined by added 2 m l of 0.1 M phosphate buffer, pH 7.5, containing 100 pg/ml of a glucose oxidase-peroxidase coupled assay. To 50-pl samples were glucose oxidase, 5 pg/ml of peroxidase, and 300 pg/ml of o-dianisidine and incubated at 37 "C for 15 min. The reaction was stopped with 2 ml of 6 N HCl and the absorbance at 540 nm was measured. ATP was determined by a luciferin-luciferase assay. The enzyme was activated according to Kimmich et a!. (9). To 0.9 ml of 5 mM arsenate buffer, pH 8.0, containing 4 mM MgS04 and 20 mM glycylglycine, were added 50 pl of sample and 50 pLl of activated luciferase and the mixture counted in a Beckman scintillation counter after 20 s for at least 30 s.

Effect of Yeast Killer
Toxin iment, to 40 ml of a cell suspension (8 X 10' cells/ml) in 25 rn Tris-HCl buffer, pH 7.5, 2 mM /3-mercaptoethanol, 0.2 m~ EDTA were added 80 ml of glass beads (0.5 mm in diameter). The mixture was shaken for 4 min in a Vibrogen cell mill and the homogenate filtered through a sintered glass funnel which was washed subsequently with an additional 40 ml of Tris buffer. The extract thus obtained was centrifuged at 400 X g for IO min to discard the cell debris. The supernatant was then centrifuged at 35,000 rpm for 30 min in a 75 Ti rotor (Beckman). The pellet of this centrifugation was resuspended with 4 ml of Tris buffer and homogenized by 6 strokes in a Dounce homogenizer. The homogenate was placed at the top of a 2040% sucrose gradient which was centrifuged for 10 h in a SW 25 rotor. Plasma membrane fractions were identified by the sodium azideresistant ATPase activity, pooled, and used for the experiments. ATPase Assay-ATPase was assayed at pH 4.6 (pH at which the enzyme has about 50% of maximal activity) as follows. To 800 pl of 30 mM c-aminocaproic-HC1 buffer, pH 4.6, containing 10 m~ MgCL (with and without sodium azide) were added 100 p1 of membranes (about 50 pg of protein) and 50 pl containing the inhibitors or the appropriate amount of killer toxin in 50 p1 of 10 mM c-aminocaproic-HC1 buffer, pH 4.6. The reaction was started by addition of 50 p! of 20 mM ATP and incubated for 30 min at room temperature. The reaction was stopped by addition of 2 ml of a mixture containing 2% sulfuric acid, 0.5% SDS, and 0.5% ammonium molybdate (the first reagent for the determination of inorganic phosphate). After 5 min, 20 pl of 10% ascorbic acid were added and after 10 min further, the absorbance at 750 nm was determined. When required, sodium azide was added to a final concentration of 0.9 mM.
Determination of Cellular Volume-The cellular volume was determined according to Rottenberg (11). Typically, a cell suspension (2 X 10' cells/ml) in 10 rn r-aminocaproic-HC1 buffer, pH 4.6, was incubated for 30 min with 0.1 M glucose. After the preincubation, to 0.5 ml of the cell suspension were added 1 pl of [3H]Hz0 (1 mCi/ml), 5 pl of ["C]methoxy inulin (0.1 mCi/ml, 9 mg/ml), and 0.5 ml of caminocaproic-HC1 buffer, pH 4.6. After 10 min of incubation at room temperature, two 400-p1 aliquots were withdrawn and centrifuged for 5 min in a Beckman microfuge. To 20 pl of the supernatant and to the pellets (after the supernatants have been removed by aspiration) was added 1 ml of 1% SDS and incubated overnight at room temperature. After this incubation 800 pl of each sample were counted for 'H and I4C in a Beckman liquid scintillation counter. The internal water volume was determined as where p stands for pellet, s for supernatant, and V, is the volume of the sample used. An internal volume of 1.36 pl/108 cells was obtained.
Determination of ApH-ApH was calculated as the difference between the external and the internal pH values. The internal pH was determined based on the distribution of the permeable ["C] propionic acid according to the equation: where C, and C,, are the internal and the external concentrations of propionic acid, respectively, and Kt the dissociation constant of propionic acid whose value is 1.35 X IO-'. C, and C, were determined by two independent methods, i.e. centrifugation and fitration. When determined by centrifugation, cells were processed exactly as described for determination of the internal volume, except that 5 pl of 1 mM [I4C]propionic acid was substituted for ["Clmethoxy inulin. The determination by filtration was carried out as follows. A cell suspension (2 x 10' cells/ml) in 10 mM e-aminocaproic-HC1 buffer, pH 4.6, was incubated for 30 min with 0.1 M glucose. After the incubation, to 50-pl aliquots of cell suspension were added 250 pl of 10 mM r-aminocaproic-HC1 buffer and 1 p1 of 1 mM [I4C]propionic acid (0.1 mCi/ml). At intervals, samples were filtered through GF/C Whatman fiters and washed twice with 2 ml of cold 5 mM €-aminocaproic-HC1 buffer. After drying, the radioactivity in the filters was determined in a Beckman scintillation counter. When yeast killer toxin or proton conductors were added, a previous incubation of 10 min was carried out before the addition of ['4C]propionic acid. When ethanolic solutions were used, controls contained the alcohol as well.
Potentiometric Measurement of pH and Potassium-pH and K' concentration were continuously and simultaneously monitored in cell suspensions containing 8 X IO" cells in a total volume of 3.6 ml of 5 mM citrate phosphate buffer, pH 4.7, plus 5 mM KC1. The cell suspensions were placed in a water-jacketed vessel at 25 "C. pH was measured with an electrode GK 2401 C (Radiometer) and K' with an electrode F 2002 (Radiometer). Both were coupled to a pH meter PHM 64 from Radiometer and the results recorded in a two-channel LKB recorder.
Measurement of Proton Permeability-Cell suspensions in 10 mM c-aminocaproic-HC1 buffer, pH 4.6, were treated as follows. To 3 ml of cell suspension (2 X IO9 cells/ml) was added killer toxin or preton conductors, which was then incubated for 15 min at room temperature. After the incubation the cells were washed twice with distilled water and resuspended in 3 ml of 20 m~ KC1 containing 60 pg/ml of antimycin A. After 2 min of incubation at room temperature, 2deoxyglucose was added to a final concentration of 50 m~ and the incubation continued. At intervals, 10 m~ HC1 was added and the pH monitored as indicated above.

Killer Effect on Proton Efflux and Potassium Influx-
Yeast cells acidify the culture medium as they metabolize glucose. Simultaneously, the cells take up and accumulate potassium, possibly as a way of neutralizing the electrical potential generated because of the extrusion of protons. Both processes can be visualized with selective electrodes immediately after glucose is added to a cell suspension. When a preparation of yeast killer toxin was added to a suspension of sensitive cells an immediate inhibition of both proton efflux and potassium influx was observed (Fig. lA). Interestingly, when the rate of glucose consumption was measured a t early stages of toxin action in which no dead cells can be detected ( 6 ) , no difference was observed between the control and killertreated cells (Fig. 1B). At late stages of killing, however, the utilization of the sugar was clearly diminished in toxin-treated cells (not shown). When the protonophore CCCP was added to identical cell suspensions two different effects were observed depending on the relative Concentration of the ionophore. At low concentrations (12.5 nmol/108 cells) both the proton efflux and the potassium uptake were inhibited (Fig.  2 A ) while at higher concentrations (500 nmol/lOh cells) both ion fluxes were reversed (Fig. 2B). In the former conditions glucose consumption was unaffected, while in the latter it was  Cell suspensions of S. cereuisiae strain X 17/17 were incubated as in Fig. 1 and proton and potassium movements monitored. A, the cell suspension contained 8 X lo8 cells and CCCP was added to a final concentration of 33 p~. B, the cell suspension contained 2 X IO8 cells and CCCP was added to a final concentration of 0.33 mM. Glucose concentration was 6 mM.
inhibited (not shown). Killer toxin and CCCP were thus acting in a similar way; both were causing the uncoupling between the oxidation of glucose and the establishment of a proton gradient (via ATP) across the cell membrane. This action was observed at low concentrations of CCCP and at early stages of killer action.

Determination of Internal p H and Effect of Killer Toxin on ApH-The determination of internal pH by studying ion
distribution is based upon the use of acids or amines in which the neutral species diffuses across the membrane while the ion is impermeable. When a permeable acid such as propionic acid has reached an equilibrium distribution, [AHJ, = [AHlOut. Since the acid dissociates on both sides, assuming the dissociation constant is not changed, it follows that: It should be demonstrated, however, that the accumulation of the acid does not collapse ApH (that is, the CJC, should be the same over a wide range of concentrations), that the internalized product is not metabolized, and that it can be extruded with proton conductors which collapse ApH. Fig. 3A shows that when either DNP or CCCP was added to cells which were accumulating [14C]propionic acid, the acid was immediately lost from the cells. The same effect was observed when unlabeled propionic acid was added to a concentration of 20 mM. It was demonstrated by silica gel chromatography that the internal propionic acid was not modified and that the same C,/C, was obtained with concentration of propionic acid ranging from 3 pM to 10 mM (not shown). Considering an internal volume of 1.36 p1/106 cells, a C,/C,, of approximately 24 was obtained by filtration, which gives an internal pH of When S. cereuisiae X 17/17 was incubated with killer toxin and the distribution of [14C]propionic acid monitored by fiitration, it was observed that the accumulation of the acid was reduced by about 50% after 10 min of treatment (Fig. 3B); when DNP was present no accumulation was observed. Given the partial purity of the killer preparation used and to insure that the observed effects were killer specific, an identical amount of material from the nonkiller isogenic strain A 8207 B-NK1 was added to a separate series. In this case the accumulation of [14C]propionic acid ran parallel to that of the control series. Table I shows the values of internal pH for control cells, killer toxin-treated cells, and proton conductortreated cells as calculated from the steady state distribution of [14C]propionic acid in experiments such as those in Fig. 3. It is shown that killer toxin reduced significantly the concentration gradient of protons. It should be emphasized that the described effects are observed 10 min after addition of the toxin, when dead cells are still not detected, although the concentration gradient of protons was reduced further when longer incubations with the toxin were performed. At this ' I / P/

TABLE I
Effect of yeast killer toxin on proton gradient across the plasma membrane of S. cerevisiae S. cereuisiae strain X 17/17 was treated with killer toxin or ionophores as indicated in Fig. 3, and the accumulation of ["Clpropionic acid calculated at the steady state. The treatment with ionophores was for 10 min. The data of C,/C, are the mean f S.D. of four experiments. The data of [H+],", pH,,, and proton concentration eradient are calculated from the mean. The external DH was 4.6. latter stage a high percentage of cells were not viable (6). Fig.  4 shows the time course of the effect of the toxin and DNP on the proton concentration gradient. It can be observed that both toxin and the ionophore collapse the gradient. However, while DNP require a short period of time to complete its effect, the toxin takes longer.

The Mechanism by Which Killer Toxin Collapses &H-
Two mechanisms could account for the observed reduction of ApH. (i) the inhibition of a proton-translocating ATPase, and (ii) the enhancement of membrane permeability to protons which would result in the masking, or even the reversal, of the proton movements due to the hydrolysis of ATP. Table I1 shows that no difference was observed in the ATP content of killer toxin-treated cells compared to control cells. Moreover, when plasma membrane ATPase was measured in the presence and in the absence of the toxin no difference was observed, although proton pumping by intact cells was inhibited by 70% in the presence of the toxin. Table I1

TABLE I1
Efiect ofyeast killer toxin, proton conductors, and ATPase inhibitors on ATP content of S. cereuisiae, membrane ATPase, and proton pumping For the measurement of ATP and proton pumping, S. cereuisiae x 17/17 was incubated as in Fig. 1. Proton pumping was estimated from the slope obtained in the pH meter after addition of 2.5 mM glucose to cells which had been previously incubated for 10 min with the indicated addition. ATP was determined in aliquots obtained 10 min after the addition of glucose. The aliquots were boiled for 5 min immediately after withdrawal, and after centrifugation for 5 min, ATP was determined in the supernatant. For ATPase assay, membranes were prepared as described under "Experimental Procedures," and the enzymatic activity measured in the presence of the toxin or inhibitors. The reaction was started by addition of ATP after a preincubation period of 10 min. when cells were treated with the protonophore DNP, no inhibition of membrane ATPase was observed, even though proton pumping by intact cells was inhibited by 80%. On the other hand, upon treatment with DCCD both membrane ATPase (about 50% inhibition) and proton pumping (80% inhibition) were affected. The fact that in the presence of the carbodiimide EDC plasma membrane ATPase was not inhibited although inhibition of proton pumping was found will be discussed below. Although these results do not definitively prove that the proton ATPase is not affected by killer toxin, they strongly suggest it, especially when considered together with the data on changes in proton permeability (see below).
Proton permeability can be measured by looking at the equilibration of the ion when acid is added to a cell suspension which contains the appropriate number of cells. In order to be able to observe the proton influx due to the change in membrane permeability it is necessary that this be the only proton movement across the cell membrane. This condition is fulfiied when antimycin A and 2-deoxyglucose are added, depleting cells of ATP and therefore suppressing any energy-dependent proton movement (12). Fig. 5A shows that when HCl was added to a control cell suspension in which the pH was continuously monitored, a drop in the pH was observed which subsequently remained unchanged. On the other hand, if the cells had been previously treated with the proton conductor CCCP the exogenously added protons quickly equilibrated across the cell membrane (Fig. 5C). When cells from the killersensitive strain X 17/17 were treated with the toxin for 15 min previous to the experiment the pattern was similar to that obtained with CCCP (Fig. 5B); that is, the cell membrane was now permeable to protons. When the preparation from the nonkiller isogenic strain was used (not shown) the pattern was identical with that of Fig. 3A.
Interestingly, when potassium was monitored simultaneously in these experiments it was observed that as the protons moved to the cell interior to equilibrate, an efflux of potassium was detected when both CCCP and killer toxin were added. This, along with the experiments depicted in Figs. 1 and 2, proves that the movement of both ions are closely connected.

DISCUSSION
The results presented in this paper strongly suggest that killer toxin acts on sensitive cells by disrupting the normal state of electrochemical ion gradients. It is known that yeast cells depleted of ATP spontaneously lose potassium and that this loss is accompanied by the gain of an equivalent amount of protons (13). It seems therefore logical that if proton pumping is inhibited when killer toxin is added to cells, a concomitant reduction in the uptake of potassium is also observed. Even though the inhibition of the proton ATPase cannot be ruled out completely, the most likely explanation of the observed inhibition of ion movements is that as the toxin enhances the permeability of the membrane to protons, a flux to the cell interior begins which, as a function of the amount of toxin, neutralizes partially or totally the observable efflux due to the ATPase (6). This mechanism could explain the efflux of potassium ions which occurs in the late phase of killer action. If a massive inflow of protons occurs in this phase, it should be accompanied by a potassium efflux. Moreover, when the cell is finally de-energized, any metabolite or ion which is accumulated by the cell against its concentration gradient will tend to equilibrate and an efflux would be observed. This could be the case for potassium or ATP, both of which have been reported to leak out of killer toxin-treated cells (3, 4).
It is of interest to compare the action of killer toxin and that of CCCP (or DNP). While the toxin can just inhibit proton efflux, with CCCP a reversion of proton flux can also be obtained provided that the concentration of the ionophore is sufficiently high. Moreover, while CCCP or DNP collapses ApH in a few minutes, killer toxin apparently requires more than 3 h to do it. These differences can be explained because of the greater mobility and the lack of specificity of the ionophores so that even intracellular membranes would become permeable to protons. It is interesting that glucose consumption was not inhibited by the toxin a t early stages of toxin action, or by CCCP at low concentrations, while proton efflux was inhibited (because of the influx of protons from the medium). This can be interpreted as glucose being futilely used by cells in their attempt to establish a proton gradient across the cell membrane. Under these circumstances the decrease of internal pH would not be enough to inhibit glycolytic enzymes. At late stages of killer action or with high concentrations of CCCP the drop of internal pH could explain by itself the inhibition of glycolysis.
According to the hypothesis of Mitchell (14, 15), oxidation of electron donors by a membrane-bound respiratory chain and a (Ca'+ and Mg2+) ATPase are responsible for generating a A;H+ that is the immediate driving force for active transport in bacteria (16, 17). Much less is known in yeasts, where it has been suggested that the creation of a ApH is due to a proton ATPase (18,19) and evidence to this effect has been reported (20,21). Moreover, it is known that some amino acids are cotransported with protons (22), indicating that in yeast too, A;H+ is an immediate driving force for active transport even though no measurements of this parameter are still available in yeast. Since killer toxin acts by reducing ApH, it is to be expected that as the driving force is reduced, active transport is affected in the same way. Indeed, it has been shown that killer toxin inhibits the uptake of leucine and histidine, as well as that of protons which are cotransported with these amino acids (6). Fig. 6 depicts the changes which occur after a sensitive cell has bound the toxin. In normal conditions the hydrolysis of ATP by the proton ATPase is accompanied by the extrusion of protons into the external medium, originating a ApH. Two consequences are that a ApK+ (more concentrated inside) can be created and that certain amino acids will be accumulated against their concentration gradient. When killer toxin has affected the cell, independently of whether the ATPase is active or not, the ApH is partially collapsed since protons can now cross the membrane. As a consequence, t,he ApK' is also collapsed and the cotransport of protons and amino acids inhibited. Interestingly, all these effects can also be observed when DNP or CCCP are added to the cells. As shown in Table I, killer toxin does not completely collapse ApH. Whether this partial collapse is enough to de-energize the cell completely remains to be determined, but the possibility that the toxin exerts some other as yet unknown effects should be considered. In this regard the mechanism by which killer toxin de-energizes sensitive cells merits discussion. The experiments shown in Fig. 5 clearly demonstrate that the proton permeability has been enhanced in toxin-treated cells. The question remains as to whether the proton ATPase is affected. The results in Table I1 suggest that the enzyme is in toxin-treated cells. Moreover, when compared to DCCD and DNP the toxin behaved like the proton conductor, since both proton pumping and ATPase activity were inhibited by DCCD while only proton pumping, but not ATPase, was inhibited by DNP and killer toxin. However, ATP levels can be maintained by other homeostatic mechanisms even if an ATPase is inhibited. In addition, even when inhibition of proton pumping is observed in intact cells, changes may not be detected in membrane ATPase activity (note that EDC, a more hydrophilic carbodiimide, inhibited proton pumping but did not affect ATPase activity of the membranes). Therefore, a definitive answer about the action of killer toxin on the proton ATPase will not be obtained until purified preparations of the enzyme are available. Regardless of some possible additional effects, the results presented here indicate that killer toxin acts by disrupting an energized membrane state in a way which could be comparable to that of colicins. Thus, it has been shown that colicin Ia, colicin K, and colicin El can alter the permeability of Escherichia coli membranes to different ions (23-25). Although much remains to be done it is now possible to speculate on whether killer toxin is to yeast what colicins are to bacteria.