Correspondence: Lana Shabala, Australian Food Safety Centre of Excellence, University of Tasmania, Private Bag 54, Hobart, Tasmania 7001, Australia. Tel.: +613 6226 6278; fax: +613 6226 2642; e-mail: firstname.lastname@example.org
The regulation of membrane-transport activity is crucial for intracellular pH homeostasis, maintenance of cell osmotic potential, nutrient acquisition, signalling, and adaptation of bacterial cells. The non-invasive microelectrode ion flux estimation (MIFE) technique is a powerful tool for kinetic studies of membrane-transport processes across cellular membranes. Since 2001, when this technique was first applied to the study of membrane-transport processes in bacterial cells (J Microbiol Methods46, 119–129), a large amount of information has been accumulated. This review summarizes some of these findings and discusses the advantages and applicability of this technique in studying bacterial adaptive responses to adverse environmental conditions. First, various methodological aspects of the application of this novel technique in microbiology are discussed. Then, several practical examples (‘case studies’) are described. The latter include changes in membrane-transport activity in response to various stresses (acidic, osmotic, and temperature stresses) as well as flux changes as a function of bacterial growth stage and nutrient availability. It is shown that non-invasive ion flux measurements may provide a significant conceptual advance in our understanding of adaptive responses in bacteria, fungi and biofilms to a variety of environmental conditions. The technique can also be used for the rapid assessment of food-processing treatments aimed at reducing bacterial contamination of food and for the development of strategies to assess the resistance of organisms to antimicrobial agents.
Cell membranes are traditionally considered to act as semi-permeable barriers allowing the preferential uptake of some nutrients, and preventing or restricting the accumulation of undesirable chemicals (Maloney & Wilson, 1996), with loss of membrane integrity leading to cell death (Yeagle, 1989). Similar to in other organisms, the efficient regulation of membrane-transport activity in bacterial cells is crucial for intracellular pH homeostasis, maintaining cell osmotic potential, nutrient acquisition, and the disposal of toxic products of cell metabolism. Recent progress in electrophysiology (Orlov et al., 2002; Martinac, 2004; Ohmizo et al., 2004) and molecular genetics (Wood, 1999; Zimmermann et al., 1999; Epstein, 2003; Kuo et al., 2003) has revealed the crucial role of plasma membrane transporters in perception and signalling in response to virtually every known environmental factor. Changes in plasma membrane potential and/or ion flux modulations are amongst the earliest cellular events in response to temperature, hormonal stimuli, elicitors, osmotic stress and mechanical stimulation in many organisms (Wood, 1999; Zimmermann et al., 1999), and show a strong link with bacterial cell viability (Silverman et al., 2003). However, little is known of early events in the adaptive response of bacterial cells occurring at the cell membrane that enable their survival in hostile environments or their growth in foods. Therefore, a better understanding of the mechanisms of adaptation of microorganisms may offer insights into methods of controlling their growth.
Causal links between membrane-transport processes and other metabolic or physiological processes in bacterial cells are not always clear. For these links to be established, the kinetics of transport processes must be monitored, followed by rigorous comparison of the revealed time constants for transport processes with those known for other cellular metabolic processes. As the bacterial life cycle and adaptive responses are fast, time resolution is a critical issue.
Historically, the transport of ions into and out of the cell has been studied using various methods of chemical analysis and tracer techniques (Onoda & Oshima, 1988; Soupene et al., 1998). More recently, attention has been focussed on the transporters themselves, located in the plasma membrane of cells. A number of methods have been employed, including nuclear magnetic resonance (NMR) spectroscopy (Gilboa et al., 1991; Gillies, 1994), fluorescent dyes (Knight et al., 1991; Jones et al., 2000; Novo et al., 2000), the patch-clamp technique (Cui & Adler, 1996; Martinac, 2004), and various types of ion-selective electrodes (Orlov et al., 2002; Ohmizo et al., 2004). Each of the techniques has particular advantages that make it best suited for a specific application. There are, however, also limitations that make their use problematic in some circumstances. For example, ion imaging by fluorescence microscopy is based on fluorescence probes that accumulate inside cells and change their fluorescence properties when bound to distinct ions (Roos, 2000). Despite having excellent temporal and spatial (especially when confocal microscopy is used) resolution, the practical use of this method is jeopardized by several major pitfalls, such as problems with dye-loading, photobleaching, the interaction of ion probes with cell metabolism, etc. (Roos, 2000). The patch-clamp technique is based on the formation a ‘giga seal’ between the plasma membrane patch and a microelectrode glass pipette, enabling measurements of very low (pA range) currents through a specific ion channel (Hamill & Martinac, 2001). However, being a highly sophisticated method, patch clamping requires a high level of technical skills, and its application to bacterial cells is very limited (Levina et al., 1999; Martinac, 2004). NMR, apart from having a relatively low time resolution, detects atoms with magnetic moments only, and, therefore, not every nutrient can be studied (Gillies, 1994).
The aim of this review is to introduce microbiologists to the microelectode ion flux estimation technique (referred to here as MIFE) and to address various methodological issues relevant to non-invasive ion flux measurements from bacterial cells (e.g. cell immobilization procedure, potential confounding effects of changes in temperature, ionic strength of solution etc.). Then, several ‘case studies’ are described, illustrating the possibility of using MIFE to study bacterial adaptive responses to various stresses (temperature, osmolality, acid stress) as well as to analyse the effect of nutrient availability (glucose) and bacterial growth stage on membrane-transport activity. We also show that the power of this technique is enhanced when it is used in combination with other advanced tools, such as fluorescence ratio imaging microscopy. Overall, we conclude that in situ measurements of net ion fluxes from bacterial cells can provide a valuable tool with which to study mechanisms mediating bacterial adaptive responses to environmental stresses.
The MIFE technique
Theory of non-invasive ion flux measurements
The theory of non-invasive MIFE ion flux measurements was the subject of a recent review (Newman, 2001). Briefly, if an ion is taken up by a living cell, its concentration in the proximity of the cell surface will be lower than that further away. Vice versa, if the ion is extruded across the plasma membrane, there will be a pronounced electrochemical potential gradient directed away from the cell surface. Ions in solution move down a concentration gradient and also down an electrical potential. Consequently, if the combined electrochemical potential gradient is measured, the net ion flux (mol m−2 s−1) can be calculated from that gradient using the Nernst equation (Newman, 2001):
where c is the ion concentration (mol m−3); u is the ion mobility (m s−1 per Newton mol−1); z is the ion's valence; F is the Faraday number (96 500 C mol−1); g is a factor found from the measured Nernst slope for the electrode during calibration; dV is the voltage gradient measured by the electrometer between two positions (V); and dx is the distance between two positions (m) (see Fig. 1a for details).
Ions crossing the cell surface in solution are carried to or from that surface by diffusion. In static or slowly changing conditions, when the convection or water uptake is negligibly small, measurements of the net diffusive flux of the ion in solution close to the sample indicate the net flux of the ion across the sample surface. Thus, the principle of the MIFE technique is the use of slow square-wave movement of ion-selective electrode probes between two positions, close to (position 1), and distant from (position 2) the sample surface (Fig. 1a).
Recorded at the two positions, voltage characteristics are converted into concentration parameters using the calibrated Nernst slopes of the electrodes. Net fluxes of specific ions can then be calculated from the measured voltage gradient at the surface. Different equations are used for objects of different basic geometry (e.g. having cylindrical, spherical or planar diffusion profiles) (Shabala et al., 1997; Newman, 2001). For the simplest case of a planar geometry (applicable to bacterial monolayers and biofilms), the net flux J (mol m−2 s−1) of a specific ion is calculated as (Newman, 2001)
MIFEFLUX software (University of Tasmania, Hobart, Australia) automatically performs the required calculations based on the geometry of the measured cell (cell diameter and probe distance above the cell surface – in the case of spherical or cylindrical geometry) and provides tabulated results of measurements as net ion fluxes (nmol m−2 s−1) for import into an ASCII-format spreadsheet.
The MIFE experimental setup is built around an inverted microscope system with long-distance objectives providing total magnification up to × 500. Microelectrodes are held in electrode holders (WI 64-0918, SDR Clinical Technology, Middle Cove, Australia) mounted on a three-dimensional hydraulic micromanipulator (Fig. 1b). An open-type experimental chamber placed on the microscope stand enables easy access to both sides of the measured object. The standard non-polarizing Ag/AgCl reference electrode is positioned in the chamber.
The MIFE system uses a stepper motor-driven micromanipulator to move 3 ion-selective electrodes. For studies with immobilized bacteria, electrode holders are positioned at an angle of 30° to the surface of the cover slip. The electrodes are oscillated, usually at either 0.1 or 0.05 Hz, between two positions, close (usually 15–20 μm) and more distant (70–100 μm) from the bacterial monolayer. The voltage output from the electrodes is amplified and digitized using an analogue-to-digital interface card (DAS 08, Measurement Computing Co., Middleboro, MA) on an IBM-compatible PC. The card also controls the stepper motor of the manipulator and is used for offset adjustment of the four-channel electrometer. Custom-designed software, the CHART program (University of Tasmania, Australia), is used to control the interface, collection and storage of the digitized output.
Microelectrode preparation for measurements
The MIFE uses liquid-type ion-selective microelectrodes (LIXs) with the tip diameter in the micrometre range. The LIX microelectrodes are faster than glass microelectrodes, with response times usually in the range of seconds (Kim et al., 1996).
Correct ion-selective microelectrode fabrication is a crucial step in ion flux measurements. Several approaches for microelectrode preparation are described in the literature (Shabala et al., 1997; Shipley & Feijo, 1999; Smith et al., 1999; Newman, 2001). The procedure developed in our laboratory enables storage of the prepared electrode blanks at room conditions for up to six weeks without loss of microelectrode performance. This is in striking contrast to other methods (Smith et al., 1999). Microelectrode fabrication includes several distinct steps: (i) pulling out electrode blanks; (ii) baking and silanizing the blanks (making the electrode surface hydrophobic); (iii) back-filling silanized blanks; and (iv) front-filling electrode tips with appropriate LIX. The electrode blanks are made using 1.5-mm (OD) non-filamentous borosilicate glass capillaries (GC 150-10, CDR Clinical Technology, Middle Cove, Australia). The blanks are pulled to <1-μm-diameter tips using a vertical pipette puller (PP 830, Narishige, Tokyo, Japan). Then electrode blanks are placed upright, base down, in a stainless-steel rack and oven-dried at 250°C overnight. Ten to 15 min before silanization, electrodes are covered by a steel lid that creates a closed container around the blanks, and 40–50 μL of tributylchlorosilane (90796, Fluka Chemicals, Busch, Switzerland) is injected under the lid. The lid is removed after 10 min and electrode blanks are baked at 250°C for a further 30 min. By this procedure, the surface of the electrode blanks is made hydrophobic, enabling entry of hydrophobic LIXs (organic cocktails; see Table 1 for details) into the tip of the prepared microelectrode.
Table 1. Specific details of the major types of commercially available LIXs used in the laboratory. All resins are from Fluka (Busch, Switzerland)
Dried and cooled electrode blanks are stored in a closed container for up to six weeks. A LIX-containing tube is constructed using a glass microcapillary with ∼30–50-μm tip diameter dipped into the stock LIX and thus containing a column of cocktail c. 1 mm long. The microelectrode blank is mounted horizontally on a three-dimensional micromanipulator and the electrode tips are flattened to achieve a tip diameter of 2–3 μm by gently placing the blank against a flat glass surface while viewing under a stereo microscope. Blanks with proper tip size are back-filled with appropriate back-filling solutions using a syringe with a thin metal needle (MF34G-5, WPI, Sarasota, FL). Immediately after back-filling, the electrode tip is front-filled with the corresponding LIX. Specific details about the major types of commercially available LIXs (ready-to-use cocktails), and the composition of back-filling solutions, are given in Table 1. Most of the prepared electrodes can be used immediately after preparation, while others (such as H+ and Cl−) require some conditioning time (∼1 h) to ensure a stable response. Once filled, the electrodes may last up to 24 h, if not longer, without significant changes in their characteristics.
Calibration of the microelectrodes
Electrodes are fitted to the electrode holder and then calibrated against a set of three standards with a series of concentrations covering the expected range of the ion in question before and after use. Electrodes with responses of less than 50 mV per decade for monovalent ions and 25 mV per decade for divalent ions, and with correlation coefficients less than 0.999 are discarded. Both the slope and intercept of the calibration line are used to calculate the concentrations of ions during the experiment. The resistance of the electrodes is typically 1−4 g Ohm. The electrode resistance and the length of the LIX in the electrode tip are critical for electrode signal-to-noise resolution, as discussed below. High values are usually associated with increased ‘noise’ of electrode response. The Ag/AgCl reference electrodes are fabricated in a similar way from a pulled and broken glass microcapillary, and filled with 1 M KCl in 2% agar. A chlorided silver wire (galvanized in 0.25 N HCl for 2–3 min) is inserted in a glass microcapillary and sealed with parafilm. A small tip diameter (∼50 μm) ensures little K+ leakage from the reference electrode.
Owing to the small size of bacteria, microelectrode ion flux measurement from a single bacterial cell is not practically feasible. To overcome this problem, a novel approach has been developed, in which ion fluxes are measured from the surface of a bacterial monolayer (a population of cells immobilized on the supporting surface and forming a dense single-cell layer) (Shabala et al., 2001a).
Cells are immobilized using poly-L-lysine for cell adhesion to solid surfaces. Polycationic poly-L-lysine molecules are known to absorb strongly to various solid surfaces, leaving cationic sites. The interaction between polyanionic cell surfaces and a charged cationic surface results in cell adhesion. The immobilization procedure for MIFE measurements includes several basic steps: (i) a glass cover slip cleaned using ethanol followed by rinsing with running distilled water and drying; (ii) application of a few drops of poly-L-lysine (0.1% weight in volume aqueous solution, P 8920, Sigma diagnostics, St Louis, MI) for ∼3 min followed by another rinse with running distilled water and drying; (iii) application of ∼30 μL of concentrated bacterial culture to the prepared cover slip followed by elimination of unattached cells by dipping the cover slip into a beaker with ‘experimental’ solution (Shabala et al., 2001a, 2002a, b). The density of the cell monolayer can be assessed under the microscope using stained cells (Shabala et al., 2002a). On average, 70–80% of the glass slide is normally covered evenly with bacterial cells. These values can be used later to express the magnitude of net ion fluxes per bacteria rather than per surface area. Measurements from several sites are usually averaged to minimize further the variability of flux values.
Methodological aspects of non-invasive ion flux measurements in microbiology
Before applying the MIFE technique to study bacterial stress responses, it was critical to address many important methodological issues in order to ensure the absence of unwanted side-effects on measurements. These include potentially confounding effects of environmental ‘hurdles’ such as the ionic strength of solution, temperature fluctuations, and solution stirring on measuring electrodes. It was, therefore, crucial to ensure that such artefacts were eliminated or, at least, taken into account in microbial studies.
An important methodological aspect relates to cell immobilization and the associated confounding effects of ion fluxes from the supporting surface used to attach bacterial cells. Lew (Lew, 2000) reported significant K+ efflux from the surface of the gelan gum used to anchor root hairs in Arabidopsis seedlings. Some type of plastics, or even a freshly cut glass surface, may be a significant source of H+ efflux (S. Shabala, unpublished). Moreover, even diluted solutions show to some extent convection, stratification and other medium inhomogeneities. These confounding effects have to be checked, and the background flux (if any) has to be subtracted from the ‘biological’ signal during analysis.
Effect of ionic strength
Variations in the ionic strength of solutions might significantly affect the characteristics of ion-selective electrodes and result in inaccurate estimates of ionic concentrations (Nobel, 1974; Hille, 1992) and, ultimately, of net ion fluxes. Methodological experiments revealed that the presence of high Na in solution caused a significant (70% for 90 mM NaCl) underestimation of the actual K+ and Ca2+ concentrations (and thus fluxes) if electrodes were calibrated without the addition of Na to standards. Such underestimation occurred as a result of the substantial shift in the electrode intercept (the Nernst slope of electrodes was not affected; data not shown). This has to be taken into account during ion flux measurements in solutions of high ionic strength, especially when dealing with marine bacteria (living normally in high-Na habitats) or while mimicking conditions of food-preservative environments (high NaCl concentrations). In practice, each electrode must be calibrated in two sets of standards (with and without NaCl present), and flux should be calculated separately for data segments, obtained before and after osmotic stress was applied.
Effect of solution change
One of the advantages of MIFE is the possibility of performing real-time measurements of flux changes. However, some kinetic experiments might require replacement of the bath solution (acid and osmotic treatment; application of metabolic inhibitors, etc). As MIFE theory assumes non-stirred layer conditions (Shabala et al., 1997; Newman, 2001), it is important to estimate quantitatively the timing required for this procedure. Depending on the geometry of the measuring chamber and its volume, between 30 s and 2 min is usually required for all turbulent flow to cease (Shabala et al., 1997; Shabala & Hariadi, 2005). The interval should be discarded from the subsequent flux analyses. In cases for which high time resolution is critical, such as in signal transduction experiments, an efficient practical way to minimize this timing is by simply adding an equal volume of double-strength solution to the measuring chamber. In this case, the time required to reach equilibrium and meet the unstirred-layers conditions does not exceed 25–30 s (Shabala & Hariadi, 2005).
The effect of temperature on LIX characteristics is another methodological issue important for the application of this technique to microorganisms, particularly in relation to the role of membrane-transport processes in bacterial adaptation to chilling or high-temperature stress. From general knowledge, it was expected that both the slope and the intercept of the calibration curve for ion-selective electrodes might be slightly affected by the changing temperatures. It remained to be answered, however, how significant these changes were and whether they should be taken into account while undertaking experiments at various ambient temperatures. Methodological experiments showed that, although both electrode slopes and intercepts were slightly affected by temperature variation, in all cases tested (a range between 4 and 40°C) the Nernst slope remained above 50 mV/decade (−54.5 for room temperature, −58.1 for +40°C, and −50.6 for+4°C), with linear correlation R>0.999 (Fig. 2). It was also found that, for practical purposes, no temperature correction is needed in the temperature range between 4 and 22°C (maximal inaccuracy in ion flux calculations ≤4.5%; data not shown). At higher temperatures (above 32°C), however, resins often became very ‘noisy’ (see Fig. 6 for example). Thus, it is highly recommended that the performance of specific LIXs should be tested at high temperatures if these are to be applied in experiment in order to ensure that the signal-to-noise ratio is acceptable, and that net flux responses can be distinguished from the background noise.
Overcoming the problem of the signal-to-noise ratio
With the electrode tip diameter being only a few micrometres, electrode resistance is usually quite high (in the g Ohm range), and often leads to rather high electrical noise (Shipley & Feijo, 1999). When combined with the relatively small signal (flux in question) from bacterial cells, this usually causes a problem of low signal-to-noise ratio and, thus, jeopardizes flux resolution, especially with high ionic background. Even if the electrodes are ‘perfect’ and all the circuits are properly grounded, a certain thermal noise remains in both LIXs and amplifiers (Ryan et al., 1990). Another source of noise is the limited resolution of the data acquisition card (Newman, 2001). There are, however, several ways to overcome this problem. First, an increase in electrode tip diameter and a reduction in length of the LIX column in the electrode tip will result in a reduction of the electrode resistance and an improved signal-to-noise ratio (Fig. 3a). It can be further improved by using a ‘proper’ microelectrode tip shape that is constructed by adjustment of the setting parameters of a microelectrode puller. Second, the problem of low signal level might be resolved by increasing the electrode travel range between positions 1 and 2. The effect is illustrated in Fig. 3b. The larger the distance, the ‘steeper’ the electrochemical gradient, and, thus, the stronger the signal per se (while the noise level remains unchanged). It is important to remember that, as ion diffusion in solution has a limited rate, the latter method might jeopardize studies on fast (signalling) events from bacterial cells. Therefore, a reasonable compromise is required, and our studies on more than 10 bacterial species suggested that the most rational is a travel range of 70–100 μm, with position 1 being 10–15 μm above the cell layer.
The problem of the signal-to-noise ratio becomes crucial when experimental conditions of interest have high levels of the ion in question (Shipley & Feijo, 1999). For example, at very low values (e.g. at high concentration of H+ in solution, such as pH 3.5 or lower), small changes in the electrochemical gradient potential for H+ arising from H+ flux might become ‘invisible’ against that background. This makes reliable measurements of H+ fluxes an almost impossible task. The problem can be solved by performing ‘recovery’ experiments, in which microorganisms (an immobilzsed bacterial monolayer or a biofilm) are treated at the desired conditions for a required time, but flux measurements are only recorded on ‘return’ to measurable conditions, with low concentrations of the ion in question. For example, bacteria can be treated at pH 3, when H+ concentration is too high (10−3 M of H+ concentration) for reliable H+ flux measurements. The bacteria are then returned to pH 6 (10−6 M of H+ concentration), and cell recovery from pH stress is measured by the MIFE technique. This approach has been validated in our work on Listeria monocytogenes (Shabala et al., 2002b). Similar protocols can be designed to study fluxes of other ions, for example Na+ fluxes in salinity experiments, etc.
Practical applications of the MIFE technique to study bacterial adaptive responses to environment (‘case studies’)
The MIFE technique has already been applied to study specific features of membrane-transport processes associated with bacterial growth and adaptive responses to stresses commonly applied in the food industry for food preservation. Similar experiments are applicable to other microbial systems, including fungi, yeasts and biofilms.
Effect of the stage of growth on net ion flux
It is well established that bacteria in the stationary phase of growth are more resistant to many environmental stresses than in the exponential phase of growth (Kolter et al., 1993). Transport of sugars and amino acids and extrusion of wastes are fundamental requirements for all organisms that often occur in co-transport with H+, Na+ or other ions (Maloney & Wilson, 1996). However, information regarding net ion flux at these distinguishing growth phases is lacking. Earlier we reported consistent changes in net ion fluxes (H+, Ca2+, K+, NH4+) in Escherichia coli as cells passed from the exponential to stationary phase (Shabala et al., 2001a). In the present study, we extended this work to compare the net H+ flux from immobilized bacterial cells of different species (and from different genera). Flux measurements were performed under static conditions for two distinct phases of growth: exponential and stationary. Results are shown in Table 2.
Table 2. Net H+ flux (nmol m−2 s−1) in the exponential and stationary phases of growth defined for various bacteria (n=14–21). SEM is the standard error of 3–4 experiments, with measurements from 4 to 7 sites on the monolayer in each individual experiment
Escherichia coli M23
Listeria monocytogenes Scott A
In these experiments, cells were grown in appropriate (for each bacterial strain) media (Shabala et al., 2001a, 2002a, b) and immobilized on a cover slip as described above. Bacteria used were of OD540=0.3A (for exponential-phase cells) and 16 h of growth (for stationary-phase cells). One mL of cells at an appropriate stage of growth was harvested by centrifugation (2500 g for 5 min), and washed twice in experimental medium, followed by cell immobilization on a glass cover slip as described above. The experimental medium contained 10 mM of glucose.
For all bacteria studied, a significant (P=0.01) difference in the magnitude of net H+ fluxes was found between the exponential – and stationary-phase cells (Table 2), with a substantial decrease in net H+ efflux through the plasma membrane as bacterial cells progressed from the exponential to stationary stage of their growth. A similar trend was earlier demonstrated for K+ and NH4+ in E. coli (Shabala et al., 2001a). This may be caused by the decreasing metabolic activity of stationary-phase cells, resulting from the decrease in ATP pool at the stationary phase of growth (Kahru & Vilu, 1983; Nesmeyanova, 2000). Increased acid tolerance in stationary-phase populations was shown to correlate with a decrease in permeability of the cell envelope to protons (Jordan et al., 1999). This implies that both the efflux and influx of protons are reduced, resulting in a decrease in the net H+ flux value (Table 2).
Effect of acid stress and glucose availability
The MIFE technique was used for kinetic studies of a number of food-related bacteria in response to acidic treatment (e.g. L. monocytogenes, L. innocua, E. coli, L. lactis, L. bulgaricus). From a pragmatic point of view, acid stress responses and adaptation of bacterial organisms to low pH are of importance for food microbiologists. Organic acids and their derivatives are widely used in the food industry as food preservatives. The acquired acid tolerance of food-related bacteria can have important implications for their survival in environments and food products and for bacterial colonization in the human gut. Bacteria have a variety of genetic mechanisms that respond to changes in environmental pH (Hall et al., 1995), but the physiology of pH adaptation is still poorly understood.
The applicability of the MIFE technique for these purposes is illustrated by the kinetics of the L. monocytogenes ATCC 19115 response to acidification of the medium (Fig. 4). A shift in external pH (pH0) from 6.5 to 4.5 (in the absence of glucose) resulted in increased H+ influx (net uptake), which may be interpreted as the inability of L. monocytogenes to pump H+ out at low external pH in the absence of an energy source. The addition of glucose, as an energy source (at 18 min in Fig. 4), to the medium enabled cells to expel protons to restore cytoplasmic pH (pHi) (open symbols in Fig. 4).
The observed data are consistent with the hypothesis that resistance to acid stress is an energetically expensive process (Miyagi et al., 1994). It has been demonstrated that H+-ATPase is involved in the acid tolerance response of L. monocytogenes (Davis et al., 1996; O'Driscoll et al., 1996). The activity of H+-ATPase is strongly dependent on cell energetic status. H+-ATPase activity was shown to correlate with the cytoplasmic ATP levels (Kobayashi et al., 1986; O'Sullivan & Condon, 1999), which depend on glucose concentration in the medium. Therefore, glucose availability might significantly affect the responses of bacteria to acid stress. Glucose is used as an energy source in L. monocytogenes. Reduced H+ extrusion demonstrated in glucose-deprived L. monocytogenes cells is in accord with reports on other bacteria (Kobayashi et al., 1986; O'Sullivan & Condon, 1999) and suggests that H+ flux kinetics measured from L. monocytogenes in response to acid stress (Fig. 4) may originate from changes in the H+-ATPase activity.
Unequivocal interpretation of H+ flux data, however, can be done only in combination with pharmacological experiments. Being essentially a pH sensor, an H+ microelectrode will detect any pH gradients, including those generated by respiration or other carbonate excretion. A simple way to address this issue is to measure H+ flux kinetics in response to the stress factor in the presence of some pharmacological agent affecting H+ transport systems. In the above case, glucose-induced H+ efflux was significantly (by ∼80%; P<0.01) suppressed when 1 mM DCCD (dicyclohexylcarbodiimide, a specific inhibitor of H+-ATPase Sturr & Marquis, 1992) was added to the bath solution (Fig. 4, closed symbols), suggesting that H+-ATPase activity was indeed contributing to the measured H+ flux from L. monocytogenes cells.
High salt concentrations are often used for food preservation. It is well accepted that bacterial growth inhibition by high NaCl levels is a result of the osmotic component of the salt stress (Wood, 1999). The latter arises from a dramatic reduction in the cell turgor pressure in response to lowered water activity (aw) in an external solution under saline conditions. Maintenance of cell turgor is essential for bacterial growth (Csonka & Epstein, 1996; Wood, 1999), and is achieved via a complex network of interacting plasma membrane transporters. In E. coli, for example, a major contributor to cytoplasmic osmolarity is K+ (Cayley et al., 1991; Epstein, 2003). In addition to K+, several so-called ‘compatible solutes’ (such as proline, betaine, trehalose, ectoine) are involved in cell osmotic adjustment (Giaever et al., 1988; Jebbar et al., 1992). Some of these are transported across the plasma membrane from the external medium through various membrane transporters; others are synthesized de novo. The latter process is believed to be regulated by cytosolic pH and K+ content (Csonka & Epstein, 1996).
In this study, L. monocytogenes ion flux responses to hyperosmotic stress were measured (Fig. 5). Net fluxes of K+, H+ and Mg2+ were measured in control (basic nutrient solution; see (Shabala et al., 2002a, b) for details), and after adding 850 mM mannitol (a resulting change in solution osmotic potential of 2 MPa). A significant (P=0.01) increase in net K+ uptake (influx) and decrease in Mg2+ flux were measured immediately after stress onset (Fig. 5). Net H+ flux was also shifted to more negative values (larger efflux), and gradually recovered over the next 30–40 min.
In E. coli, the first response to increased medium osmolarity is stimulation of K+ uptake. At least three K+ uptake systems are present in the E. coli plasma membrane (Trk, Kdp, and Kup) and all are upregulated by hyperosmotic stress within minutes, or even seconds (Csonka & Epstein, 1996; Wood, 1999). It appears that a similar scenario is applicable also to L. monocytogenes cells, although it remains to be seen what specific K+ transport mechanisms are involved. Pharmacological experiments using specific K+ channel blockers (such as tetraethylammonium, TEA+) or non-selective cation channel blockers (such as La3+ or Gd3+) may provide further insights into specific ionic mechanisms mediating bacterial responses to osmotic stress. The use of specific K+-transport mutant strains may also contribute to a resolution of this issue.
Temperature fluctuations are a common feature of many microbial ecosystems, and low temperatures are routinely used for food preservation. However, many bacteria, including the food-borne pathogen L. monocytogenes, are able to grow at chill temperatures (Ross et al., 2000), making storage by refrigeration alone inadequate to control microbial growth in many foods. The maintenance of bacterial growth at low temperatures is achieved by changes in lipid composition of the plasma membrane (Russell et al., 1995). The lipid bilayer provides a suitable environment for the proper functioning of the membrane protein complexes involved in bioenergetic and biosynthetic functions. To enable optimal metabolism over a range of temperatures, bacteria maintain optimal fluidity of the membrane through alterations in the degree of acyl chain saturation and branching, or acyl chain length, in bacterial membrane (Russell et al., 1995). The lipid composition, therefore, has a direct effect on the activity of plasma membrane transporters (Arav et al., 1996). Thus, study of the temperature-induced ion flux kinetics may provide a rapid and convenient (although indirect) way to determine the temperature at which changes in membrane lipid composition occur and, thus, to predict bacterial behaviour and evaluate temperature treatments used for food safety. This is further illustrated in Fig. 6, where the kinetics of L. monocytogenes (LO28) membrane-transport activity after low-temperature stress was monitored.
Immobilized stationary-phase cells were kept at 4°C for 2 h, followed by slow return to 37°C with a temperature-rate change of 1°C/min (Fig. 6a). A temperature-controlled chamber (HCC-100A, Dagan, MN) was used to control temperature changes linearly with the desired rate change. A progressive increase in net H+ efflux was observed (Fig. 6a), consistent with our previous observations in plant tissues (Shabala & Shabala, 2002). Pharmacological experiments showed that this temperature-induced H+ efflux was strongly (∼80%) suppressed by CCCP (carbonyl cyanide m-chlorophenylhydrazone; protonophore) and vanadate (sodium orthovanadate; a specific inhibitor of P-type ATPase), implicating the involvement of H+-ATPase (Shabala & Shabala, 2002). This is in agreement with the recent reports of (Soini et al., 2005) showing an increase of the cellular ATP concentration in response to temperature up-shift in bacteria. At the same time, incomplete inhibition of H+ flux kinetics by vanadate and CCCP suggests that other metabolic process (such as elevation of the cellular respiration and glucose uptake Soini et al., 2005) might also contribute to measured net H+ fluxes, although to a lesser extent. An apparent critical temperature (Tc), at which a sudden change in the activity of H+ transporters responsible for H+ efflux occurred, was about 7°C (Fig. 6b). It was shown earlier that this apparent temperature showed a strong dependence on the rate of temperature change (Shabala & Shabala, 2002). Experiments with two different rates of temperature change might allow quantification of the absolute critical temperature at which membrane-transport activity undergoes dramatic changes (Shabala & Shabala, 2002), thus enabling comparison of the relative chilling resistances of various bacterial strains and mutants.
It has been suggested that, for plants, in response to a change from the liquid-crystalline to the gel state, sensor proteins undergo a conformational change acting as a primary event in the transduction of the temperature signal (Murata & Los, 1997). There is strong evidence that this putative sensor protein may be a Ca2+ channel (Monroy & Dhindsa, 1995). Via a complex signal transduction pathway (involving numerous second messengers), activation of des genes occurs, leading to enhanced biosynthesis of fatty acid desaturases and sequential adaptive changes in membrane fluidity (Murata & Los, 1997). While the molecular genetics of the later stages of this process are well known, the mechanism of putative temperature sensors remains to be investigated. Application of the MIFE technique could be a powerful tool with which to gain such knowledge.
Combination of non-invasive ion flux measurements with fluorescence microscopy
The power of the MIFE technique increases many-fold when it is used in combination with other physiological or biophysical tools. This is illustrated by the results of our studies on L. monocytogenes acid stress responses from a combined application of MIFE and fluorescent ratio imaging microscopy (FRIM) techniques (Shabala et al., 2002a). The principle of the FRIM technique is that a bacterial cell is loaded with a non-fluorescent dye (such as 5(6)-carboxyfluorescein diacetate-N-succinimidyl ester). The dye is cleaved inside bacterial cells by non-specific cellular esterases and becomes fluorescent (Breeuwer et al., 1996). Fluorescent emission is recorded at two wavelengths, only one of which is pH-dependent. The established ratio is pH-dependent but concentration-independent, and therefore changes in the ratio reflect alterations of the pH inside bacterial cells. Intracellular pH is then quantified using a calibration curve.
Microelectrode ion flux estimation measurements of net H+ from bacterial cells in response to acid treatment provide valuable information about the early stages of cell adaptive responses as well as about the kinetics of the response. Owing to the large number of contributing factors, however, an unambiguous interpretation of the mechanisms responsible for the measured pH changes (interpreted by the MIFE technique as net H+ flux) is sometimes problematic. The combination of the MIFE and FRIM techniques enabled the kinetics of protons at either side of the bacterial membrane to be monitored simultaneously (Shabala et al., 2002a).
As shown in Fig. 7, a clear relationship between H+ flux patterns and cell pHi existed. In glucose-sufficient cells (solid line in Fig. 7), net H+ efflux (extrusion) was observed for the range of external pH0=4 and above. In this range, pHi was maintained at the optimal (for Listeria) pHi=7.5 level (Fig. 7, solid line). Once the ability of the cell to extrude H+ failed (at pH0<4), pHi fell dramatically, indicating the inability of cells to maintain normal metabolism and resulting in cell death (as indicated by growth experiments Shabala et al., 2001a).
In glucose-deficient cells (dashed line in Fig. 7), no H+ efflux was observed, even at optimal (relatively high) pH0 values. Accordingly, even a slight acidic stress (pH0 changes from 6 to 5) caused an immediate and significant (P=0.01) pHi drop. An increase in the stress severity resulted in a progressive increase in net H+ influx into the cell (Fig. 7, dashed line), further contributing to pHi decline.
The activity of H+-ATPase increases with decreasing pH0 for a number of bacteria (Kobayashi et al., 1986; O'Sullivan & Condon, 1999). Keeping in mind the important role of H+-ATPase in the acid tolerance response (ATR) of L. monocytogenes (Cotter et al., 2000), the data here suggest that a similar acid stress-induced activation of H+-ATPase takes place in L. monocytogenes. It was shown that the magnitude of response of both pHi and H+ flux depended on glucose concentration in the medium (Shabala et al., 2002a) and that the lowest pH0 at which ΔpH did not collapse occurred at the highest used (10 mM) glucose concentration, reflecting the ability of L. monocytogenes cells to expel protons. The absence of H+ extrusion in the trial without glucose (Fig. 7, dashed line) is in accord with this and suggests the involvement of H+-ATPase. The above findings might be of importance in understanding the mechanism of ATR induction. The kinetics of pHi and net H+ fluxes are shown to be complementary in nature and provide evidence that plasma membrane H+ transporters play a central role in L. monocytogenes pH homeostasis and ATR.
Combining fluorescent microscopy to measure pHi and MIFE techniques proved to be a powerful tool in studying bacterial response to acidic stress. Further possible applications of the synergism of the two techniques include studies of bacterial response to other stresses such as organic acids, salinity, temperature, etc.
Prospects and conclusions
In this work, an introduction to the MIFE technique and to the feasibility of its applications in microbiology was undertaken. Similar to studies on mammalian cells and plants, the application of non-invasive ion flux measuring techniques in microbiology might provide unique opportunities to answer specific questions concerning how bacteria cope with environmental stresses. A better understanding of physiological processes might provide insights into early events associated with environmental perception and signalling in microorganisms, including those in natural biofilms. This might offer efficient measures to control bacterial growth with minimal doses of treatments used, further contributing to the ‘Hurdle Technology’ (Leistner, 1987) approach used in the food industry. Applications also include the analysis of stress resistance of cells to antimicrobial compounds and investigation of complex microbial communities, such as biofilms. Our preliminary experiments suggest the feasibility of ion flux measurements from the biofilm surface, essentially following basic protocols for bacterial ‘monolayers’. Ion flux measurements might also lead to greater progress in balancing the nutritional requirements and optimization of culture growth conditions (pH, osmolality, temperature, etc.) for beneficial food microorganisms in biotechnology and food fermentations.
From the data presented, it is apparent that the MIFE technique by itself is a powerful tool, capable of providing valuable information about the ‘blueprints’ of cell adaptive responses to the environment. Its power, however, might be many-fold higher when combined with other electrophysiological or cellular techniques such as fluorescence microscopy and patch clamp. Recent progress in molecular genetics has made available many mutant bacterial isolates and strains with altered adaptive response characteristics. For many of them, the genome sequence is known. These ‘structural genomics’ data might, and should, be complemented by comprehensive studies into the ‘functional genomics’ of these organisms. Non-invasive ion flux measurements seem to be ideally suited for this purpose.
This work was supported by an ARC Discovery (DP0559874) grant to S.S. and T.R., and a University of Tasmania IRGS grant (S0014384) to L.S.