Electrophysiological techniques provide a wealth of information regarding the molecular mechanisms that underlie the function and modulation of ion channels. They have revealed that bacterial porins do not behave as static, permanently open pores but display a much more complex and dynamic behavior than anticipated from non-electrophysiological studies. The channels switch between short-lived open and closed conformations (gating activity), and can also remain in an inactivated, non-ion conducting state for prolonged periods of time. Thus the role of porins is not limited to that of a molecular filter, but is extended to the control of outer membrane permeability through the regulation of their activity. Electrophysiological studies have indeed demonstrated that both gating and inactivation are modulated by a variety of physical and chemical parameters and are highly cooperative phenomena, often involving numerous channels working in concert. Cooperativity acts as an amplification mechanism that grants a large population of porins, such as found in the outer membrane, with sensitivity to modulation by external or internal factors. By conferring permeability properties to the outer membrane, porins play a crucial role in the bacterium's antibiotic susceptibility and survival in various environmental conditions. The detailed information that electrophysiology only can provide on porin function and modulation promises to yield a more accurate description of how porin properties can be used by cells to adapt to a changing environment, and to offer mechanisms that might optimize the drug sensitivity of the microorganism.
Porins are trimeric channel proteins located in the outer membrane of Gram-negative bacteria. They have been well described at the biochemical and structural levels. In fact, porins are the only membrane channels whose three-dimensional structure is known at atomic resolution . Numerous reviews on these aspects, as well as on the regulation of porin gene expression, have been published ([2, 3], and references herein). This review will focus on functional properties of the classical non-specific porins (OmpC, OmpF and PhoE), in particular those revealed by electrophysiological techniques. Such techniques are powerful because they provide the high time and molecular resolutions needed to obtain detailed information of the activity and modulation of channels. This type of knowledge is essential for the basic understanding of outer membrane physiology and for biomedical applications. Although very interesting results have been obtained by electrophysiology on other pore-forming proteins of the outer membrane and on structure-function relationships in porins, these topics will not be covered in this review.
2 Studies with planar lipid bilayer
Initial functional work on porins was based on the diffusion of solutes through purified proteins after reconstitution into lipid vesicles. Nakae first coined the term ‘porin’ when demonstrating that solutes up to 600 Da were able to penetrate lipid vesicles only after the ‘matrix protein’ of Escherichia coli had been introduced in the membrane .
Very early on it was recognized that the passage of ions through such large pores should yield sizable currents, and electrophysiological techniques were applied to the study of porins. At the time the most common technique was that of planar lipid bilayer or black lipid membrane . In such a system a lipid bilayer is formed over an aperture located in a septum separating two chambers. Each chamber is filled with a buffered ionic solution and contains an electrode which is used to detect any movement of ions through the bilayer. These electrodes are also used to maintain (‘clamp’) the transmembrane potential to a fixed value, independently of the ionic movement. In the absence of added channel proteins, the conductance, i.e. the ease of flow of current across the bilayer, is very low. When channels are inserted in the membrane, the conductance increases due to the presence of ion-conducting ‘holes’ in the membrane. Typically, membrane bilayers are made by ‘painting’ the aperture with lipids dissolved in an organic solvent (like n-decane) and subsequent evaporation of the solvent. Purified detergent-solubilized channel proteins are added to the so-called cis side, and spontaneously insert in the bilayer over time. Recordings of channel activity is typically made in high salt solutions (1 M).
Porins studied with such a technique are typically first purified by a variety of treatments, including SDS solubilization, trypsinization, or other extractions . Benz and coworkers [6, 7] were able to witness the consecutive incorporation of channel proteins in the membrane, as discrete jumps in the membrane current were detected over time. The size of these individual conductance increments follows a distribution, from which an average value can be calculated and gives an estimate of the channel size. By repeating the measurements in the presence of a salt concentration gradient and a variety of ions, the ionic specificity of the channel can be inferred [8, 9]. These studies were significant in confirming that OmpC and OmpF were large pores with an estimated diameter greater than 1 nm with a slight cationic specificity (permeability of potassium about four and 30 times greater than chloride for OmpF and OmpC, respectively). PhoE, on the other hand, had an anionic specificity, which correlates well with its putative function as a specific permeation pathway for phosphate derivatives. Important findings of these initial studies were that: (1) the current traces never showed any decrease in the conductance which would have represented the closures of porin channels, and (2) the dependence of the membrane conductance with the applied voltage was ohmic, i.e. linearly correlated with potential, over a wide range. This behavior is expected from a population of open ion channels which pass a greater current in response to an increased electrochemical potential, but whose number remains independent of the transmembrane voltage. In other words, the biological implications were that these channels behaved as permanently open pores, whose function was independent of voltage. These observations were, however, challenged by other experiments.
Schindler and Rosenbusch showed that deviation from the ohmic dependence could be observed when the bilayers were maintained at voltages greater than 150 mV [10, 11]. They attributed this behavior to the voltage-dependent closing of porins. These results were later confirmed by other studies on a variety of porins [12–15]. It is likely that this behavior had not been apparent in the concurrent studies by Benz and coworkers because many functional properties of porins, including voltage-dependence and conductance, are highly dependent on purification and reconstitution procedures [16–18]. In general, the milder the treatments, the lower the voltage at which the porins start to close (threshold voltage). In their work, Schindler and Rosenbusch  were using a ‘raised-bilayer’ technique, which avoids the use of organic solvents, and did not purify porins but added outer membrane vesicles.
The hallmark features of porin behavior monitored by the planar lipid bilayer technique are: (1) the pores remain open for prolonged periods of time (seconds to minutes), (2) the conductance of presumed single channels is large (>150 pS in 100 mM KCl), (3) the pores close only at high voltages (typically greater than 100 mV, but the threshold voltage was lower than 100 mV in some studies ), and (4) closing transitions are often grouped in three steps, possibly representing the successive closures of each monomer of a trimer .
3 Complex behavior revealed with patch-clamp
Finer details of the functional properties of porins have recently been obtained by using patch-clamp . This technique permits the study of single channels in a small patch of membrane at a lower noise and higher time resolution than in black lipid membranes. Thus, channels can be investigated in their native environment even if they belong to small cells. A glass pipette, of ∼1–2 μm tip diameter, is placed at the surface of a cell or lipid vesicle. A membrane patch, harboring one or a few channels, is drawn inside the tip, and the currents passing through the ion channels are monitored over time under a clamped transmembrane voltage. The technique was applied successfully to live bacterial cells and spheroplasts [20, 21], as well as reconstituted liposomes containing inner or outer membrane fragments [22, 23].
In patch-clamp experiments, the detection of currents passing through single channels requires that a tight mechanical and electrical seal be formed between the internal wall of the glass pipette and the lipid bilayer. The total electrical resistance of such a seal and the patch of membrane under study must be in the order of 1×109 ohm (1 gigaohm). It is surprising that such a high resistance can be obtained when the outer membrane of a bacterium is drawn into the pipette [20, 21]. Indeed, the area of such a membrane patch is roughly equal to the surface area of an E. coli cell, and should contain several thousand open porins. Such a leaky membrane does not have a gigaohm resistance. There are two possible explanations for such an observation: either the actual patch isolated at the tip of the pipette is from the inner membrane (possibly due to rupture of the outer membrane), or most porins are maintained in closed states and the permeability of the outer membrane, in these conditions, is extremely low. Electron micrographs of giant spheroplasts showed an intact outer membrane and electrophysiological evidence that patches were made on the outer membrane was presented . For example, the activity of a few (10–20) porin channels could be recorded in such experiments. Such a low ratio of open to closed channels in the cell has also been suggested by others .
The biological implications of this observation are profound: it argues that in some conditions cells can maintain a large population of their porins in the closed state, and hence that porin activity is regulated. One might then wonder why the cell would synthesize so many porins. This situation is in fact akin to that of nerve cells which harbor thousands of voltage-dependent channels – most of them kept closed at resting membrane potentials. The opening of the channels, however, can be triggered by depolarizing voltages and lead to the major membrane permeability changes that underlie neuronal excitability. By having a pool of synthesized, but regulated ion channels, the neurons are primed to respond quickly to excitation. Bacteria are not excitable cells, but they can face rapid changes in their environment. The presence of a large number of porins whose activity state can be regulated prepares bacteria to respond quickly to these challenges. It has been suggested that a large number of open porins is only needed when cells are grown in conditions of low food supply . In rich laboratory media, less than one percent of the expressed porins need to be in the open state to sustain growth. Cells must then be able to regulate the activity of porins in a reversible way that depends on external conditions. As discussed below, factors that keep porins in closed states have started to be identified. A major focus of future research will aim to illustrate that closed porins can be opened.
Porins behave differently in patch-clamp experiments on spheroplasts than in planar lipid bilayers : they remain open for shorter periods of time (20–30 s) and display frequent transient closures of 1 msec or less on average, they show a much reduced single channel conductance (<100 pS in 100–150 mM KCl), they close cooperatively in numbers that are not multiples of three, and they become inactivated at negative voltages only (periplasmic side) of lesser magnitude than reported in planar lipid bilayers. These characteristic electrophysiological features were also observed when porins are studied in reconstituted outer membrane fractions [22, 25–27]. In our initial reconstitution studies [22, 25], porins were not purified and one may question the identity of the channels. Further experiments with well-characterized porin mutants later confirmed that these activities indeed represented those of porin channels .
The kinetic behavior of porins is complex. This is mostly due to two factors: the high degree of cooperative behavior and the existence of multiple conformational (open and closed) states. These observations were already made with planar lipid bilayers studies [11, 18], but patch-clamp experiments have provided additional and more detailed information on these interesting phenomena. A typical patch-clamp recording of porin activity shows a trace representing the total current flowing through many, mostly open, channel monomers (Fig. 1). This current dwells at a preferred level, which we call ‘baseline’ (BL) in Fig. 1. Two types of deflections arise from this baseline: those that correspond to a reduction of total current and thus represent channel closures (upward deflections in Fig. 1), and those that correspond to an increased total current and thus represent openings of additional channels (downward deflections in Fig. 1). Both phenomena are reversible, such that the trace appears as the succession of short-lived current jumps away from the baseline. Typically the openings of additional channels are extremely transient and appear as downward spikes. The closures, on the other hand, are often more prolonged and appear as square-top deflections. Closures can occur with two distinct kinetics. Fig. 1 shows the reversible transitions of channels into a short-lived closed state of an average duration of the order of a few milliseconds. Channels can also occupy a long-lived ‘inactivated’ state of a duration that can reach several seconds to minutes [25, 26]. In some cases, the entry into the inactive state is apparently irreversible . The fact that the kinetics of the openings and closing transitions from a common baseline level are so different indicates that porins can display kinetically distinct states (at least two closed states and two open states ). Presumably they represent different stable closed and open conformations of the proteins that can be visited independently.
Patch-clamp recording of current through (A) a porin-free or (B) a porin-containing patch. In (B), the dominant current level is denoted ‘BL’. It represents the total amount of current flowing through ∼15 porin monomers. Dashed horizontal lines mark the current levels corresponding to the closures of 1–4 channel monomers (C1–C4). Short-lived openings of additional channels are seen as downward spikes originating from the baseline. The preparation of outer membrane is from a strain that expresses both OmpC and OmpF; thus the channel activity is likely to represent that of heterotrimers. The single channel current is 1.8 pA. The pipette voltage is −60 mV.
The closing transitions of fast kinetics can involve one or more channels (Fig. 1). In the latter case, stepwise closures are typically not seen, indicating that the channels behave as highly cooperative units. Interestingly, the cooperativity extends beyond trimers, since the simultaneous closures of any number of channels can occur [25, 26]. For example, Fig. 1 shows a current record where cooperative closures of 2, 3 and 4 channels can be observed (C2, C3, C4). The maximum number of monomers closing cooperatively becomes larger as the transmembrane potential increases. This may be due to a combination of effects on the intrinsic closing rate constant of individual channels (see below) and on the cooperativity itself [25, 26]. The transitions to the inactivated state also rarely involve single channels. Berrier and colleagues  report that during the inactivation process most patches showed conductance steps that may represent single trimers, although some variability does exist.
On the basis of the size of the conductance jumps and the kinetics of closures, it seems probable that the closing transitions reported in planar lipid bilayers correspond to the slow inactivation of trimeric units seen in patch-clamp with large potentials (>80–100 mV). The reason for the lack of fast kinetics in planar lipid bilayers is not apparent at this time, and can only be addressed once the same porin preparation is used in both types of experiments carried out in parallel with the same bilayer composition and solutions. However, it is clear that porins behave as non-independent channels. This dynamic behavior, combined with the potential modulation of the channels by voltage and endogenous compounds (see below), can have profound effects on the activity state of a whole population of channels, and hence on the overall permeability properties of the outer membrane.
The entry into the short-lived closed state or into the inactivated state is voltage-dependent, with faster rates at higher potentials. In all patch-clamp studies, the voltage dependence is clearly asymmetric; however, there is some discrepancy regarding the polarity of the potentials involved. This is likely ascribed to a different orientation preference during reconstitution, which might be due to the fact that in one study, porins are first purified and then reconstituted , while in the other, the porins are studied within reconstituted outer membrane fractions that are likely to still contain some lipopolysaccharides . From the work on live cells, it appears that the voltages favoring the closed states are negative on the periplasmic side . The frequency of rapid transitions that characterize the fast closing kinetics show variability, and appears to depend on the nature of the porins investigated. When membrane fractions are made from wild-type strains expressing both OmpF and OmpC, and thus containing a predominance of heterotrimers , the majority of the patches display a high frequency of closing transitions [25, 27]. When patches containing similar numbers of homotrimers (either OmpF or OmpC) are studied, the closing frequency can be reduced by a factor of three or more [28, 29]. This interesting phenomenon, and its significance, are still poorly understood, and are under further investigation.
Is there any physiological relevance to the voltage-dependence? It is likely that this question will remain answered poorly until voltage-independent mutants can be isolated or constructed. It has been suggested that the voltage dependence may provide a defense mechanism in case of the wrongful insertion of porins in the inner membrane . One can also argue that there might be conditions in which even the outer membrane would experience, possibly transiently, potentials that would be high enough to promote porin closures. For example, cells synthesize periplasmic membrane-derived oligosaccharides (MDOs) when they are grown under conditions of low osmolarity . These compounds are too large to leak out of the cell through the porins, and constitute a pool of fixed, highly negatively charged, molecules. They can reach concentrations of several millimolar and hence create Donnan potentials of up to −100 mV on the periplasmic side. In a very elegant study, Sen and coworkers  showed that such high potentials were unable to diminish the flux of antibiotics through porins and concluded that porins did not display voltage-dependent behavior in vivo. In fact, this result is not surprising since patch-clamp experiments on live cells  demonstrated that only positive potentials on the periplasmic sides were able to inactivate porins (see above). Such positive potentials could be established when cells, adapted to low osmolarity and harboring periplasmic MDOs, are suddenly facing an external high salt concentration, and experience a rush of positively charged ions in the periplasm. The resulting depolarization could, through the closure of porins, allow the membrane to be ‘sealed off’ until the appropriate regulatory mechanisms are in place to deal with the osmotic upshock. In such a situation, the decreased membrane permeability would improve survival of the cell. This appealing hypothesis, however, still needs to be tested rigorously.
4 Modulation of porins
Porins appear as much more plastic and dynamic entities than previously deduced from non-electrophysiological experiments. Recently, systematic studies of porin modulation have been initiated. There is ample evidence that ion channels are influenced by membrane composition, and porins appear to require the presence of lipopolysaccharides (LPS) for function . The addition of LPS during reconstitution affects the conductance of OmpC porin specifically, in a way that depends on the glycolipid structure, and reveals an additional conductance, of larger magnitude, that is similar to that of OmpF pores .
Effects of outer membrane composition on porins can also be mediated by the protein environment, in particular by the size and nature of the porin clusters. Already mentioned is the possible effect of trimeric composition on gating kinetics. In addition, the size of the clusters of activated channels affects the voltage sensitivity, kinetics and reversibility of channel closure, and the channel conductance . Interestingly, larger clusters have lower conductance values and reluctance to re-opening after voltage-induced closure. These features suggest that in the native environment the permeability properties of the outer membrane as a whole and its regulation by voltage might be influenced in subtle ways by the quantity and the nature of porins.
Alterations in the ionic composition of the surrounding milieu can also impact the behavior of porin channels. There is agreement among various groups that an acidic pH promotes stabilization of closed states and shifts threshold voltages to lower values [12, 18, 33, 34], but the same reports are conflicting with regards to the effect on conductance. Decreasing the ionic strength promotes the appearance of channel populations of higher conductances , but it is not clear that the effect is on the size properties of the channel per se or on cooperativity.
An important, and potentially medically relevant aspect of porin modulation is the inhibition of porin activity. The ionic flow through specific porins, such as LamB or Tsx, is blocked by the presence of the larger, specific, substrates for these porins . The non-specific porin PhoE also passes reduced currents in the presence of large anionic compounds . In all these cases, the effects appear to be due only to the physical block of the ion conduction pathway. Recently, kinetic effects have been documented for the modulation of OmpC and OmpF by MDOs and polyamines. When presented to the periplasmic side of the channels, a crude fraction of MDOs can induce an increase in the frequency of closures . The effects are observed at millimolar concentrations, which is a physiological level when cells are growing in low osmolarity . These effects are not mimicked by a mixture of succinate and phosphoglycerol, and it was inferred that they may be mediated by the ethanolamine moiety of these complex molecules.
An inhibition by positively charged, amine-bearing, compounds has been clearly demonstrated in the study of the interactions of polyamines with porins. Cadaverine was first shown to induce closing of E. coli porins in a concentration- and voltage-dependent manner . Although this compound is small enough to penetrate through porin, its effect is distinct from a simple block, because no significant reduction in single channel conductance is detected, and because the polyamine appears to inhibit porin by stabilizing closed states. In the presence of the inhibitor the porin channels visit closed states more frequently and for more prolonged periods of time. In addition, there is an increase in the number of channels that close cooperatively. These effects are observed with all four natural polyamines (putrescine, cadaverine, spermidine and spermine), on both OmpF and OmpC [27, 29, 36, 37]. The voltage dependence and the lack of effect at alkaline pH (where the polyamines are less positively charged) indicate that ionic interactions between the channel and the drug are likely. Indeed, the mutation of aspartate 105 of OmpC into glutamine abolishes the increase in closing frequency seen in wild-type . This residue is located at the tip of the L3 loop, which intersects the pore at half the height of the barrel, and is likely to be one of the sites of contact of the polyamine with the channel. It also appears that multiple binding sites for polyamines exist on porins because mutations on the L3 loop have abolished only certain aspects of polyamine modulation but not others . Other molecular determinants involved in the binding and inhibition of polyamines are presently under investigation.
The impact of this inhibition on the overall permeability properties of the outer membrane is demonstrated with non-electrophysiological experiments performed on live cells. All four natural polyamines inhibit chemotaxis and β-lactam antibiotic flux through porins, in concentration ranges that agree with those found by electrophysiology . Thus, the permeability of the outer membrane can be drastically reduced by these compounds. Koski and Vaara  showed that cadaverine, putrescine and spermidine are associated with the outer membrane, probably through interactions with LPS. Thus, polyamines can be found in the vicinity of porins and may play modulatory roles in vivo. A complete assessment of the physiological relevance of this type of inhibition is likely to emerge from the study of porin activity in conditions that alter polyamine synthesis and secretion, and in polyamine-deficient mutants.
An unexpected aspect of porin modulation, that of increased closing probability of OmpC by hydrostatic pressure, has recently been described . Using patch-clamp, the authors show that OmpC, but not PhoE, is sensitive to the application of positive pressure concurrently with negative pipette potentials. This other highly asymmetric form of modulation also appears to involve the L3 loop, as in the case of polyamine inhibition. Indeed, insertions and site-directed mutations in the L3 loop conferred to the insensitive PhoE porin a pressure-dependent modulation that was similar to that of OmpC. It is likely, however, that the increased closing activity is not due to a direct effect of the pressure forcing a movement of the L3 loop across the channel, but rather to a redistribution of ions within the pore which might alter the electrostatic forces that exist between the L3 loop and the opposite wall of the β-barrel. This hypothesis is supported by recent modeling studies of the involvement of L3 in voltage-dependent closure of porins .
5 Concluding remarks
Electrophysiological studies of bacterial porins provide a valuable tool to follow the activity of single proteins in real time under a variety of conditions, and thus are particularly useful for the study of modulation and relationships between structure and function. They have revealed the complexity of porin function and demonstrated that the channel's functional state can be influenced by many parameters. This dynamic behavior must be highly significant for the survival of bacteria, which can encounter changing and severe environmental conditions. The implications from this complex behavior of porins need to be taken into account in the design of therapeutic strategies, since the modulation of porins can drastically affect the overall outer membrane permeability and possibly even the antibiotic susceptibility of the microorganism. Continued progress in electrophysiological investigations of porins will refine current models on the relationships between porin function and control of outer membrane permeability.
Michael Benedik is gratefully acknowledged for his critical reading of the manuscript.
CowanS.W., Schirmer, T., Rummel, G., Steiert, M., Ghosh, R., Pauptit, R.A., Jansonius, J.N. and Rosenbusch, J.P. (1992) Crystal structures explain functional properties of two E. coli porins. Nature 358, 727–733.
HamillO.P., Marty, A., Neher, E., Sakmann, B. and Sigworth, F.J. (1981) Improved patch-clamp techniques for high-resolution current recordings from cells and cell-free membrane patches. Pflügers Arch. 391, 85–100.
BerrierC., Coulombe, A., Houssin, C. and Ghazi, A. (1989) A patch-clamp study of ion channels of inner and outer membranes and of contact zones of E. coli, fused into giant liposomes. FEBS Lett. 259, 27–32.
BerrierC., Coulombe, A., Houssin, C. and Ghazi, A. (1992) Fast and slow kinetics of porin channels from Escherichia coli reconstituted into giant liposomes and studied by path-clamp. FEBS Lett. 306, 251–256.
SaintN., Prilipov, A., Hardmeyer, A., Lou, K.-L., Schirmer, T. and Rosenbusch, J. P. (1996) Replacement of the sole histidinyl residue in OmpF porin from E. coli by threonine (H21T) does not affect channel structure and function. Biochem. Biophys. Res. Commun. 223, 118–122.