Studies on the Depolarization of the Escherichia coti Cell Membrane by Colicin El*

When a small pulse of oxygen is added to an anaerobic suspension of logarithmic phase Escherichia coli cells, the subsequent acidification of the medium which is observed in the absence of permeant charged ions is slow (t,,, = 10 s) as is its relaxation (t,,z = at least several minutes). The number of protons extruded for each oxygen atom added (H+/O) is small, varying from about 0.4 to about 1.0 depending upon the carbon source used for growth and the growth phase of the cells. Treatment of the cells with colicin El causes a large increase in the amplitude of the proton extrusion elicited by an oxygen pulse, so that the H+/O ratio attains values >2.0 regardless of the cell growth conditions. In addition, the rate of proton efflux (t,,* -=z 1 s) and its relaxation (t,)* = 10 to 20 s) are greatly accelerated in colicin- treated cells. After addition of colicin El, the increase in the H+/O ratio has a time course which is similar to the El-induced loss of K+. Furthermore, the effect of colicin El on the kinetics and extent of H+ efflux is dependent upon the presence of K+, Na+, or Li+ in the medium, with an apparent K, for potassium of about 0.5 mu. The properties of the proton pulses measured in the presence of colicin El plus

When a small pulse of oxygen is added to an anaerobic suspension of logarithmic phase Escherichia coli cells, the subsequent acidification of the medium which is observed in the absence of permeant charged ions is slow (t,,, = 10 s) as is its relaxation (t,,z = at least several minutes). The number of protons extruded for each oxygen atom added (H+/O) is small, varying from about 0.4 to about 1.0 depending upon the carbon source used for growth and the growth phase of the cells. Treatment of the cells with colicin El causes a large increase in the amplitude of the proton extrusion elicited by an oxygen pulse, so that the H+/O ratio attains values >2.0 regardless of the cell growth conditions. In addition, the rate of proton efflux (t,,* -=z 1 s) and its relaxation (t,)* = 10 to 20 s) are greatly accelerated in colicintreated cells. After addition of colicin El, the increase in the H+/O ratio has a time course which is similar to the Elinduced loss of K+. Furthermore, the effect of colicin El on the kinetics and extent of H+ efflux is dependent upon the presence of K+, Na+, or Li+ in the medium, with an apparent K, for potassium of about 0.5 mu. The properties of the proton pulses measured in the presence of colicin El plus K+ are very much like those measured in cells treated with the permeant anion SCN-. Thus, these experiments provide direct evidence for a rapid, colicin El-induced depolarization of the bacterial membrane.
The actual pattern of ion movements leading to this depolarization is at present not known. Proton uptake occurring with the same time course as the El-induced potassium efflux can be detected, but the amount of H+ uptake is too small to balance the observed potassium efflux. The efflux of the organic anions pyruvate and glucose B-phosphate also seems to be much too small to provide adequate charge balance for the K+ which leaks out after colicin El treatment. Jacob et al. (1) reported that colicin El inhibited growth and nucleic acid synthesis in a sensitive Escherichia coli strain while not inhibiting respiration. It was also found that this colicin inhibits a range of active transport systems (2). These observations, and that of an oxygen requirement for the action of this colicin, led to the hypothesis that colicin El is an l This work wss supported by Grant GM18457 of the National Institutes of Health.
$ Recipient of National Research Service Award lF32GM99913 from the National Institute of General Medical Sciences, National Institutes of Health. Resent address, Department of Chemistry,. University of Notre Dame, Notre Dame, Ind. 46556. uncoupler of oxidative phosphorylation (Levinthal and Levinthal,quoted in Ref. 2). This hypothesis was consistent with further studies on colicin inhibition of active transport (3,4) and with the observation of a rapid decrease in intracellular ATP levels caused by colicin (5).
However, it was also found that the ef&iency of colicins El and K is not markedly decreased in anaerobic samples, as measured by the inhibition of active transport (5) and the colicin El-induced increase in fluorescence intensity of a lipophilic probe (6). These observations, together with the report that the effectiveness of coiicins K and El was not changed in hemin-less mutants (51, implied that the action of these colitins does not necessarily require respiratory activity. An energized membrane, however, may be required for the action of all colicins (7, 81, although the actual degree of required energization may be small (91, and this energization may be achieved through either ATP hydrolysis or respiration. It has heen proposed that the energy level of energy transducing membranes is controlled by the free energy stored in an electrochemical potential consisting of a proton gradient and/ or an electrical potential across the membrane (10, 11). According to this chemiosmotic hypothesis, charge separation generated through electron transport can generate a transmembrane potential, which can subsequently be relaxed by the movement of a proton through an energy transducing complex (e.g. ATP synthetase, transport proteins). At present, there is controversy over aspects of this model relating to the precise significance of the ion movements observed between the bulk aqueous phases separated by the membrane (12, 13), and some of these problems are discussed elsewhere (14). However, the data presented here is not inconsistent with the existence of a local or transmembrane potential in the energized membrane of E. coti. In intact E. coli, it appears that the dominant component of the electrochemical potential at pH 7 is the membrane electrical potential (15,16). It is reasonable to expect, therefore, that the de-energization of the membrane by colicin would involve dissipation of this potsntial. In a system in which energy is stored primarily as an electrical potential, the appropriate movement of any charged species, not just protons, can serve to de-energize the membrane. Indeed, an event which appears to be closely related in time to the first biochemical events caused by colicin is an increase in K+ permeability of the cell resulting in the leakage of intracellular K+ into the medium (6,(17)(18)(19). Fluorescence studies with the probe 3,3'-dihexyloxycarbocyanin indicated that colicin K partially depolarized the cytoplasmic membrane (20  Mitchell and Moyle (23,24) observed that addition of an oxygen pulse to an anaerobic suspension of mitochondria causes an increase in the acidity of the suspension. In those experiments, and in similar experiments done with anaerobic bacterial suspensions , the amplitude and rate of the proton eMux a&r an oxygen pulse was increased in the presence of mobile, charge-compensating counterions (e.g. SCN-, K+ plus valinomycin).
According to Scholes and Mitchell (271, these counterions allow a larger and faster proton movement by dissipating the membrane electrical potential which is formed by the charge separation accompanying respiration-catalyzed proton efBux. In this paper, we report that colicin El, in the presence of K+, Na+, or Li+, causes an increase in the rate and extent of proton efilux after an oxygen pulse in a manner similar to SCN-, and that this effect closely parallels in time the loss of K+ by the cells.

Effects of Colicin El and FCCP on Oxygen
Pulse-dependent pH Changes -The addition of a small pulse of oxygen to an anaerobic suspension ofEscherichia coli B/1,5 cells results in a brief period of electron transport which is accompanied by an acidification of the medium. The efficiency of coupling between this proton emux into the external medium and the reduction of oxygen (H+/O ratio) is low in the absence of any additions to the resuspension medium, and is dependent upon both the growth phase of the cells and the carbon source used for growth ( Fig. 1 and Ref. 14). Cells grown on succinam and harvested in midlogarithmic phase growth typically exhibit H+/O ratios of 0.5 or less, whereas succinate-grown cells harvested in stationary phase or glycerol-grown cells harvested in either log phase or stationary phase exhibit H+/O ratios around 1.0 ( Fig. 1). In all cases, the kinetics of proton eftlux are rather slow (tliz = 5 to 15 s). The pH change decays very slowly, with only a small fraction of the change reversed in 10 min. West and Mitchell (25) 1). Furthermore, these effects of colicin El are only observed when colicin-sensitive cell strains are used (Table I). The H+/O ratios obtained with cells of the colicin-tolerant strain A,,, and the colicinogenic strain JC411 are unaffected by 1 pg/ml of colicin El, whereas cells of the colicin-sensitive strains B/1,5 and K12 1100 (cell survival <O.l%) show the effects described above and shown in Fig. 1. Strain ML 308-225, which is relatively insensitive to colicin El (survival = 50%) also showed an increase in the H+/O ratio in the presence of 1 PgI ml of colicin El; although the increase was not as large as that seen with the colicin-sensitive strains and is predictable on the basis of a 50% survival level in this experiment.
The effects of the uncoupler FCCP on the oxygen-dependent proton pulse are very different from those described above for colicin El. With concentrations of FCCP in the minimal range needed to inhibit active transport under aerobic conditions (  to 5 j.& (211, the proton extrusion following an oxygen pulse is almost completely eliminated by FCCP in both normal cells and in cells treated with colicin El (Fig. 2). At concentrations of FCCP which give a partial inhibition of the extent of the proton extrusion, FCCP causes a marked increase in the decay of the pH change, due to the re-entry of extruded protons into the cell (Table II). Thus, 1 to 5 pM FCCP appears to act by increasing the permeability of the membrane to H+ ions, a conclusion consistent with much data already in the literature (e.g. Ref. 32). Colicin El, over a multiplicity range of 1 to 100, causes an increase in the extent of the proton ef&x even though it also causes an increase in the rate at which extruded protons re-enter the cell (Fig. 1) (28). These findings, therefore, are consistent with the idea that colicin El causes an increase in the permeability of the bacterial membranes to charge-compensating counterion( This notion is further supported by the finding that the effects of colicin El on the proton pulse can be mimicked by including the permeant anion SCNin the reaction medium. In the presence of SCN-, the kinetics of H+ eMux, H+ influx, and the change in the H+/O ratio are increased (Fig. 31, as they are in the presence of colicin El (Fig. 1 increase in the H+/O ratio for a suspension of cells grown in glycerol and resuspended in 150 mM KC1 is presented in Fig. ha. The initial H+/O ratio in the glycerol-grown cells is -1.0 as in Fig. 1. After an initial lag of about 1 min, the H+/O ratio increases over a period of about 5 to 6 min (t,,* = 3V2 min) to a value approximately twice that measured before the colicin addition. The half-time for the increase in the H+/O ratio was generally about 3 min. This time course is similar to that observed previqusly under anaerobic conditions for the colicininduced change in the fluorescence intensity of the probe Nphenyl-1-naphthylamine (6). More importantly, the time course fOF the change in the H+/O ratio is also similar to the time course for K+ etIlux after colicin addition (t,,% = 2l/3 min) measured under the same conditions (Fig. 46 1 Fig. 6, where it is clear that in order to observe the effects of colicin El on the kinetics of proton efflux, and on the H+/O ratio, KC ion must be present in the medium. The smaller increases in the H+/O ratio observed in Figs. 5 and 6 before the addition of potassium can be attributed to the colicin El-induced loss of K+ from the cells (see below), which results in a final K+ concentration in the medium of approximately 0.3 to Q-5 mM. This concentration of K+ increases the H+/O ratio to approximately one-half of the maximum value which can be obtained at higher external K+ concentrations, indicating that the K,, for potassium required for colicin El effects is approximately 0.3 to 0.5 mM. In a similar experiment, the effects of some other cations on the rates of H+ efllux and H+ intlux were tested (Fig. 7). The final concentration of K+ in the medium after colicin addition was evidently s&X- cient to allow a larger increase (Fig. 76) in the rates of H+ efthx (t,12 = 3 s), H+ influx (tllZ = 60 s), and the H+/O ratio in this experiment compared with the experiment shown in Fig.  6b. The addition of 5 mu Na+ (Fig. 7~) caused a further increase in the rate of H+ efflux (& 5 1 s) and H+ influx (t,!z = 12 s). The addition of 5 mu Li+ (Fig. 7d) also caused an increase in the rates of H+ movement, although Li+ was less effective in stimulating H+ influx (tt,l = 25 to 30 s). The addition of 5 mM Mgz+ (Fig. 7e) caused very little further increase in H+ efflux or H+ influx rates above those in the control (Fig. 7b). Fluxes of Other Ions -As described earlier, the addition of colicin El to a suspension of sensitive cells results in the loss of cellular potassium with a time course similar to that observed for the change in the H+/O ratio (Fig. 4b). Colicin-induced K+ leakage with similar kinetics has also been reported for colicin K (18) and for colicin El (6). It can be calculated that such a potassium efJlux, if electrically uncompensated by simultaneous cation influx or anion eflux, would hyperpolarize the membrane to a potential in excess of 100 V, an event which cannot occur. Indeed, as the cell is losing internal K+, the membrane is simultaneously lusing ita ability to become electrically polarized by proton translocation. This suggests that, in fact, the efflux of internal K+ is electrically compensated and gives no net polarization to the membrane. At the moment, however, there is little data to indicate the nature of the ion or ions which compensate K+ ef%x. In the absence of oxygen, a colicin-induced H+ influx with kinetics similar to those of the K+ efflux can be observed (Fig. E), but to date the highest H+/K+ ratio observed is -0.1. Although it has previously been suggested that sodium can act as a counterion for potassium leakage caused by colicin K (331, the kinetic data of K+ efflux a&r the addition of colicin El were found to be the same in the presence or absence (Fig. 4b) of 10 mM NaCl. It has been suggested (20) that chloride efTlux, or the efflux of organic anions originally observed by Fields and Luria (5), could account for the apparently electroneutral K+ efflux. However, the intracellular Cl-levels are much less than the intracellular K+ levels (33a), and the amounts of pyruvate and glucose B-phosphate lost by the cells during the time course of K+ efllux appear to be very small. In a typical experiment, the concentration of K+ in the medium afbzr a 36-min incubation  (Table III). DISCUSSION The increase in H+/O ratio caused by colicin El has been shown previously to occur at colicin multiplicities as low as one (26) and to require the presence of potassium ion. The time course for the increase in the H+/O ratio (Fig. 4o) is similar to the time course for other early biochemical events initiated by colicin El, such as the leakage of intracellular K+ (Fig. 4b1, the decrease in intracellular ATP levels, and the structural changes in the cell envelope monitored by fluorescence probes (6). In colicin-treated cells, the time course of the change in H+/O ratio (Fig. 4~1, the dependence of the amplitude of the H+/O ratio on externally added monovalent cations (Figs. 5 to 71, and the similarity of the proton pulses obtained in the presence of colicin El to those obtained in the presence of SCN- (Fig. 31 or valinomycin plus potassium (251, imply that the larger H+/O ratios measured in these colicin-treated cells result from the fact that the efflux of H+, which would normally generate a membrane electrical potential, can now be balanced by a counter flow of potassium or other ions. That is, the cell membrane seems to be freely permeable to potassium movement in either direction in the presence of colicin El, and to be permeable as well to external sodium and lithium ions. We know little at this time about the mechanism of this increase in ionic conductance across the membrane, whether it is due to (a) an ion channel created by the colicin itself or (b) induced in the inner membrane through structural changes. While the former possibility seems somewhat easier to visualize, it has been inferred from experiments with colicin immobilized on Sephadex beads that colicin El may not need to move from the neighborhood of its surface receptor in order to exert its effects (36). Concentrations of colicin El up to 3.6 pg/ ml also did not increase the conductance of a diphytanolylphosphatidylcholine planar bilayer separating aqueous phases containing 0.1 M potassium chloride.' In any case, the consequence of free potassium movement and entry of other monovalent cations is that the cell should no longer be able to maintain a state of charge separation across the inner membrane. In other words, colicin El in the presence of potassium or other monovalent cations should cause a cellular membrane potential ta be dissipated. Feingold (19) observed that addition of the uncoupler CCCP after an acid pulse caused alkalinization of the medium if the cells had first been treated with colicin El. This was attributed to compensating potassium movement potentiated by the presence of the colicin. The rate of alkalinization was low in the absence of colicin and no pH change was observed upon addition of colicin El alone. The main inference from these experiments, and from measurements of the colicin-induced K' leakage and decrease in the intracellular ATP level, was that colicin El causes an increase in membrane permeability to potassium ions, but not to protons.
In previous work from this laboratory dealing with the mechanism of action of colicin El, we have considered the nature of structural changes in the cell envelope (inner and outer membrane) associated with the primary process of mem-' J. M. Gould   brane de-energization (e.g., 21,37). This work has shown that de-energization of the cell by colicin El, or by the uncoupler FCCP, causes a change in the rotational motion of the amphiphilic fluorescence probe ANS and the hydrophobic probe PhNap. An effective permeability barrier in the outer membrane to the hydrophobic probe is decreased upon de-energization of the envelope by colicin El and FCCP, resulting in increased binding of the probe to the cells (21). The increase in the binding of PhNap after colicin treatment was first reported by Nivea-Gomez et al. (22) for cells treated with colicin Ia. The time course of these structural changes for colicin El (6) and colicin Ia (22) is very similar to that of the earliest biochemical changes, and raised the question as to whether such structural changes caused by colicin El could in fact be a primary event in the transmission of the lethal effect of this colicin (37). This hypothesis has the conceptual problem of explaining how a single protein added to the cell envelope can cause such an extensive structural change in the cell envelope. Degradative enzyme activity with rapid kinetics associated with early biochemical events following colicin El addition has not yet been detected (38). Since changes in the H+/O ratio and dissipation of the membrane potential occur as rapidly as the first detectable biochemical changes, it would seem that the structural changes and the change in the effective permeability barrier of the outer membrane could be an immediate consequence of the decrease in cytoplasmic membrane potential. One cannot specify at this time how a collapse in electrical potential across the inner membrane could cause immediate structural changes in the envelope other than to say that the local electric field density across the membrane is, of course, very high. There is precedent for electrostrictive effects on membranes (391, and it is clear from structural studies on the Escherichio coli envelope that there are specific connections between the peptidoglycan layer and outer membrane (40).
The concentration of potassium required in the external medium for an increase in H+/O ratio and dissipation of the membrane potential is not large (Fig. 5) and can be partly supplied by the intracellular K+ lost to the medium. The data of Figs. 5 and 6 imply that the dissipation of the membrane potential should be even more complete in the presence of high external potassium levels. On the other hand, it has been found that the survival of sensitive cells treated with colicin K or El is actually enhanced on agar plates containing high (-100 mM) potassium, relative to those made with low potassium concentrations (41). The latter experiments suggest that the main, if not the only, cause of cell death is the decrease of the intracellular potassium concentration below the physiological levels needed to sustain protein synthesis and glycolysis. The experiments reported here, in conjunction with the general ideas discussed above on membrane de-energization caused by colicins El and K, imply that restoration of high intracellular potassium levels should only be sufficient to restore cell viability in the presence of active, bound colicin if the membrane is not required to do work (i.e. ATP synthesis or active transport) or if the effect of colicin on the membrane is reversed during the incubation.
There are a variety of other colicins which seem to have a mode of action similar to that of El. These include colicin K, mentioned above, colicin Ia (42), colicin A (43, 441, and possibly S8 (45). Bacteriocins JF246 from Serrutia marcescens and bacteriocin 1580 acting on gram-positive bacteria (44) also seem to resemble El. The conclusions reached in this paper about the mode of action of El have been shown to apply to colicin K (281, and possibly apply as well to the above listed colicins and bacteriocins. Finally, it should be noted that the conclusions reached in this paper on the mechanism of colicin El action are based largely on the model for proton et&x and membrane energization in bacteria proposed by Scholes and Mitchell (2'7) based on the chemiosmotic hypothesis of Mitchell (10,11). However, some aspects of the proton emux from bacterial membranes induced by an oxygen pulse are not easily explained by the chemiosmotic hypothesis (14). This of course does not change the empirical similarity between the effects of colicin El (plus M+) and permeant ions such as SCN-or valinomycin plus K+, or the conclusion that these conditions all lead to a loss of the ability of the membrane to become electrically polarized by vectorially oriented proton transfer reactions, and therefore to the de-energisation of the bacterial cell membrane.