Editor: Ralf Conrad
Analysis of aerotactic band formation by Desulfovibrio desulfuricans in a stopped-flow diffusion chamber
Article first published online: 6 OCT 2005
FEMS Microbiology Ecology
Volume 55, Issue 2, pages 186–194, February 2006
How to Cite
Fischer, J. P. and Cypionka, H. (2006), Analysis of aerotactic band formation by Desulfovibrio desulfuricans in a stopped-flow diffusion chamber. FEMS Microbiology Ecology, 55: 186–194. doi: 10.1111/j.1574-695X.2005.00024.x
Present address: Jan P. Fischer, Max-Planck-Institut für Marine Mikrobiologie, Bremen, Germany
- Issue published online: 6 OCT 2005
- Article first published online: 6 OCT 2005
- Received 20 December 2004; revised 21 June 2005; accepted 27 June 2005
- Sulfate-reducing bacteria;
- energy taxis;
- oxic-anoxic interface;
Aerotactic band formation by Desulfovibrio desulfuricans (DSM 9104) was studied in a stopped-flow diffusion chamber. This chamber allowed us to create reproducible, steep oxygen gradients in a flat capillary, time-lapse video recordings and spatio-temporal analysis of band formation. The cells formed two types of bands. Bands of the first type evolved quickly after starting the experiment and were located near the oxic–anoxic interface. Bands of the second type typically appeared several minutes later and a few millimeters inside the initially anoxic volume of the capillary. Band formation depended on metabolism and could be stimulated by lactate addition, and thus appears to be energy taxis. Mathematical modeling of oxygen diffusion and respiration within the chamber revealed that bands formed preferentially at oxygen concentrations close to 4% air saturation. The swimming speed of the cells was determined by digital single-cell tracking and found to be highest (up to 58 μm s−1) close to the oxic–anoxic interfaces. Motility patterns were influenced by surfaces, at which cells accumulated. Bioconvection sometimes occurred if very dense bands had formed. The ecological implications of these two phenomena are unknown.
Chemotaxis is a well-known phenomenon which has been observed in a wide range of bacterial species. It enables cells to move to favorable environments. Many different chemical species were found to act either as attractant or as repellent or both, depending on their concentration. Beijerinck described bacteria forming sharp bands in oxygen gradients (‘Atmungsfiguren’) and interpreted this as a ‘seeking behavior’ to a certain optimum concentration of oxygen (Beijerinck, 1893). The reaction to oxygen is called aerotaxis. Aerotactic responses were reported for all major taxonomic groups of bacteria and archea (Engelmann, 1881; Taylor et al., 1999). In many cases, chemotactic signaling starts with a sensor protein that transduces the signal through the membrane into the cell, where a regulatory network controls the flagellar motor. Alternatively, it is not the chemical stimulus itself that is detected by the cell but changes in the internal energy level, which are dependent on external factors. Such mechanisms are called ‘energy taxis’ (Taylor & Zhulin, 1998), and have been found in various bacterial species (Rebbapragada et al., 1997; Taylor & Zhulin, 1998). For example, electron acceptors can function as attractants. However, in the case of energy taxis there will be no response if no electron donor is available or if the cells are physiologically not adapted to this acceptor.
Chemotaxis has been most intensely studied with Escherichia coli (Berg & Brown, 1972; Stock & Surette, 1996). This bacterium performs a sequence of straight runs, interrupted by short tumbles during which the swimming direction is changed randomly. Perception of increasing concentrations of an attractant causes the bacterium to swim forward longer than during a period with decreasing or constant concentrations. Other bacteria have different mechanisms to move to favorable environments. Polarly flagellated bacteria such as Desulfovibrio desulfuricans swim backwards for a while instead of tumbling (Eschemann et al., 1999). Thiovulum majus was found to perform U-turns after leaving the optimum environment (Fenchel, 1994). Recently, it was even reported that a bacterium senses chemical gradients spatially rather than temporally (Thar & Kühl, 2003).
Although cultures of D. desulfuricans do not show sustainable growth in the presence of oxygen (Marschall et al., 1993) they were found to translocate protons and to form ATP by aerobic respiration (Dilling & Cypionka, 1990). Cell suspensions were found to form bands in oxygen gradients (Eschemann et al., 1999; Sass et al., 2002). Band formation at a certain position in a gradient must include both positive as well as negative responses to the stimulus. As D. desulfuricans does not grow aerobically, its aerotactic behavior might be directed by a defense strategy, enabling them to quickly re-establish anoxic conditions (Cypionka, 2000).
Chemotaxis can be studied on the population level or by tracking individual cells. First, test tubes were used to study band formation macroscopically (Beijerinck, 1893). For quantitative assessments, Adler (1966) introduced the capillary technique. Thin capillaries filled with a chemical stimulant were placed into a chamber containing a bacterial suspension. The accumulation of bacteria in the capillary after a certain time was used as a measure for the chemotactic response. Flattened capillaries were a further improvement of the method enabling observations on the microscopic level. A sophisticated tracking microscope was built by Berg & Brown (1972). They were able to follow single cells and to record their swimming paths in three dimensions.
To create reproducible, steep gradients, Ford et al. (1991) introduced the stopped-flow diffusion chamber (SFDC) in chemotaxis studies with E. coli. The SFDC allows two different fluids to be brought into contact via an impinging flow. However, this method has not been used widely. Problems with the SFDC might arise from disturbances of the fluids after stopping the flow.
In the present study, we have used an advanced version of the stopped-flow diffusion chamber to study chemotaxis on the microscopic and the macroscopic level. Digital time-lapse photography and image analysis, accompanied by mathematical modeling, were used to observe and later describe the complex behavior of D. desulfuricans in oxygen gradients.
Experiments were performed with Desulfovibrio desulfuricans strain CSN (DSM 9104). The cells are gram-negative, slightly curved rods with a length of 3–5 μm and a single polar flagellum (Cypionka, 1989). The cells were cultivated at room temperature (20–24°C) in a mineral medium according to (Cypionka & Pfennig, 1986) with 10 mM sulfate and 20 mM lactate. Resazurin was used as a redox indicator.
Stopped-flow diffusion chamber
Motility was studied with a modified stopped flow-diffusion chamber (SFDC, Ford et al., 1991). In this chamber, two fluids were pumped at the same rate from both ends into a flat cuvette, forming an impinging flow. Two outlets in the middle of the cuvette allow for efflux during pumping (Fig. 1). One fluid was an anoxic bacterial suspension, the other one was an aerated filtrate of this suspension. The resulting flow-pattern showed a sharp front in the middle of the cuvette, separating the two fluids from each other. A steep gradient of oxygen and cell densities is generated. After stopping the flow, diffusion of oxygen takes place and the cells react to the gradient.
The SFDC was constructed using standard laboratory material and epoxy resin. The base was formed by a glass slide (76 × 26 mm), the top was a long coverslip (24 × 60 mm). For the side walls, two cut hypodermic needles (0.80 × 120 mm) were mounted on the object slide at a distance of 8 mm. In the middle of the needles, small slits functioned as outlets. These could be closed simultaneously after pumping by two stainless steel wires inside the needles which were used as valves (Fig. 1, bottom). Cut autosampler tubes glued on the ends of the cuvette formed the inlets. Thus, a laminar flow was generated and the injection of gas bubbles into the chamber was avoided. To inject the suspensions, a syringe pump for two parallel gas-tight syringes (Fortuna Optima® 10 mL, Walther Graf and Co. GmbH, Wertheim, Germany) was used. This ensured equal fluxes on both sides. The syringes were connected with the SFDC by teflon tubing (inner diameter 0.8 mm).
To prepare cell suspensions for the experiments, H2S was removed from the cultures by gassing with CO2 for 15 min. Subsequently, the pH was titrated to 7.0 with NaOH. Experiments were performed without centrifuging and washing of the cells, as this strongly affected motility and chemotactic responses. Half of the suspension was filtered through a syringe filter (NalgeNunc® SFCA, Nalge Europe Ltd, Neerijse, Belgium, 0.2 μm) to remove the cells and aerated to ensure air saturation before transferring to the glass syringe.
KCl (additional 100 mM) was added to the cell suspension in order to increase the specific gravity and thus avoid convection. Then the SFDC was mounted perpendicularly with the cell suspension at the bottom. The chamber was flushed with 3 mL of the solutions before stopping the flow and closing the outlet valves.
The chamber was illuminated from the sides by means of two fluorescent tubes and a blind enabling darkfield observations. The intensity of the scattered light gave a measure for cell density of the suspension. Image acquisition was done using a digital camera (Philips® PCVC740K, Philips, Eindhoven, the Netherlands) with a resolution of 1280 × 960 pixels. Pictures were taken every 20 s, experiments lasted between 20 and 60 min. For calibration, cell suspensions with known densities were measured. Suspensions of up to 5 × 108 cells mL−8 showed good linearity between cell density and brightness of the image. In very dense suspensions, cell concentrations were underestimated.
Image analysis and visualization
The images of the time-lapse recordings were analyzed in several steps. First, a region of interest was selected. Averaging over the width of the cuvette reduced the effects of noise in the images. The resulting course of brightness along the longitudinal axes of the chamber served as a measure for cell density (Fig. 2). All pictures of a series were treated in the same way, providing a three-dimensional plot.
The SFDC was placed horizontally under a microscope (Leitz DM RBE, phase contrast, × 40 objective, Leitz, Wetzlar, Germany). No KCl was added to the cell suspension in these experiments because it would not have helped to avoid convection here, as it does in the vertical setup. Recordings of the swimming patterns were made with a digital camera (Kappa CF 15/2 Kappa, Gleichen, Germany), stored in MiniDV format and subsequently transferred to a computer. Tracking of single cells by analysis of the video sequences was done with home-made MATLAB-based software.
Model-based reconstruction of oxygen concentration
Direct measurement of oxygen concentration was not possible with electrodes due to the closed construction of the chamber. Therefore, simple mathematical models were used for estimating the oxygen distribution in the chamber. The analytical solution of the diffusion equation gave the change in oxygen profile with time in the absence of respiring cells. We assumed infinite length of the chamber (which was reasonable due to the reservoirs at the ends of the chamber). Because the gradient was symmetric to the longitudinal axis, the problem could be treated as one-dimensional. At the beginning, a step-function with an oxygen concentration of 0% at one side of the chamber and air-saturation at the other was assumed. With these initial conditions, the general diffusion equation for one dimension:
can be solved as described by Crank (1975).
To take the activity of cells into account, the diffusion equation was amended by a respiration term f(z, c),
with z being the cell density and c the oxygen concentration. D is the diffusion coefficient of oxygen in water, c0 denotes the initial concentration of oxygen and t is the time after starting the experiment. The shape of f(z, c) was obtained by fitting respiration rates of D. desulfuricans CSN measured at different oxygen concentrations. A constant rate of 80 nmol O2 (mg protein min)−1 was assumed at oxygen concentrations below 140 μM. At higher concentrations, the rate decreased by 0.38 per μM oxygen (J. Kranczoch and H. Cypionka, unpublished data). For simulations, this equation was solved numerically. Time steps of sufficiently small size were chosen to ensure accuracy and stability of the numerical scheme.
Band formation in cell suspensions
Two different types of bands developed within the oxygen gradients generated by means of the stopped-flow diffusion chamber (SFDC). The width of both bands were 0.2–1 mm. Bands of type 1 evolved quickly after starting the experiment (20 s to 5 min), and were located near the oxic–anoxic interface. They remained at the same position until disintegration, which took place after 10–30 min.
Bands of the second type appeared typically several minutes after starting the experiment. Their position was a few millimeters inside the initially anoxic volume of the SFDC. Once started, the development of these bands was completed after a short time (Figs 2 and 3). Bands of type 2 often moved towards the oxygenated volume with traveling speeds of millimeters per hour.
In some experiments, double bands occurred (Fig. 3c), with a second band mostly arising at the time of decomposition of the first band (Fig. 3b and d). Cells building the second band originated from the first one. In some cases, only bands of one type developed (Fig. 3a). Several times, cultures grown under the same conditions and treated in the same way differed in banding patterns; however, repeated experiments with the same cell suspension gave reproducible results. Dilution of a suspension sometimes changed the behavior of the cells (Fig. 3c and d). Furthermore, the response depended on the age of the culture. The fastest reaction and sharpest bands were observed in cultures from the late exponential growth phase. Younger and older cultures typically showed slower or no development of bands, although the cells in these suspensions were motile. Band formation in old cultures could be stimulated by the addition of lactate (5 mM).
In control experiments with immobile cells the initial front remained sharp all the time, indicating that effects of Brownian motion could be ignored.
Response to small oxygen pulses
To examine the reaction of the cells to small oxygen pulses, the same anoxic cell suspension was filled into both ends of the SFDC, then the oxygen concentration of one side was increased to 50 μM by injecting air-saturated suspension. The cells started to form a band immediately, which was located in the oxygenated part of the chamber, and the cells forming it came from the anoxic part (Fig. 4). Obviously, the cells swam towards higher levels of oxygen. The band was not as narrow as in the experiments with air-saturated filtrate and the relative cell accumulation was lower. The decomposition of this band started after 15 min, and after 45 min there was still a slightly increased cell density in the center of the SFDC.
Modeling of the oxygen concentrations during band formation
The oxygen concentration and its variations with time and space were inaccessible to a direct measurement with electrodes due to the construction of the chamber. To obtain information about the concentration field, simple mathematical models were used for estimation.
First, the oxygen gradient was calculated assuming oxygen diffusion without respiring cells, and thus overestimating the concentration. Several observed bands of type 2 were located at a calculated oxygen concentration of about 4% (Fig. 5). This was independent of the dilution of the suspension. Surprisingly, bands in diluted suspensions sometimes developed earlier than those in concentrated suspensions.
A numerical model of oxygen distribution was used to include respiration of the cells in dependence of their density. If a respiration rate of 80 nmol O2 (mg protein min)−1 was assumed (J. Kranczoch and H. Cypionka, unpublished data), the oxic–anoxic interface was close to the observed bacterial bands (Fig. 6, middle). The bands (type 2) formed at the position and time of the deepest penetration of oxygen into the anoxic volume. Higher respiration rates (160 nmol O2 mg−1 mL−1) lead to a situation where the band was totally in the anoxic volume (Fig. 6, left). Assuming lower rates (40 nmol O2 mg−1 mL−1) made the bands lie deeply in the oxic part (Fig. 6, right). Therefore, it is concluded that type 2 bands form at the oxic–anoxic interface and stop further oxygen penetration by respiration. If both band types occur, type 1 is not able to stop the intrusion of oxygen.
Bioconvection and surface effects
The SFDC was mounted vertically with the oxygenated part on top. In some experiments with strong bands of type 1, bioconvection was observed. This phenomenon occurs if the accumulation of cells, which have a higher specific gravity than the medium, forms packets that are big and heavy enough to overcome the viscosity of the medium and drop down. In this case, convective fingers or plumes of cells fell out of the band (Fig. 7). The cell density in adjacent fingers changed with time, indicating that cells swam in and out from one finger into another.
Within strong bands of type 1, the cell density was often observed to be highest near the glass walls of the capillary. This could be seen as parallax when watching the capillary from an inclined perspective. In cases of bioconvection, this effect could also be observed: the convective structures were located in two different planes. This accumulation of cells could be explained by decreased swimming velocity close to walls. The effect was most pronounced near bands as the speed differences should be highest in these areas.
Observations on the individual cell level
Using the SFDC under a microscope allowed us to examine the behavior of individual cells. Video recordings were made over a transect through the chamber, from the oxic–anoxic interface into the anoxic suspension. The recordings commenced quickly after starting the experiment, before band formation made tracking of single cells impossible. It turned out that the swimming speed of the cells decreased with decreasing oxygen concentrations. In the oxic part, the average speed was 25 μm s−1, whereas in the anoxic part it was only 5 μm s−1 (Fig. 8). The average speed decreased under anoxic conditions and also the maximum speed reached by single cells.
In our study, a sophisticated setup was used to analyze aerotactic reactions on both the macroscopic and individual cell level. The original design of the SFDC (Ford et al., 1991) was improved. By implementation of gate valves inside the outlets, hydrodynamic disturbances were avoided. The use of coverslips for the construction of capillaries enhanced the optical performance, and convection was prevented by vertical mounting and use of solutions with different buoyancies. Thus, reproducible steep gradients were generated, which allowed us to analyze cell behavior and to model oxygen distribution in time and space.
Band formation in oxygen gradients strongly depended on the growth phase of the cultures. Motility of the cells was not a sufficient precondition for this. Obviously, the behavior is not directed simply by oxygen sensing. Instead, band formation was stimulated by the addition of lactate, providing electron donors in excess. Thus, band formation in Desulfovibrio desulfuricans appears to be energy taxis, coupling electron transport and behavior (Taylor & Zhulin, 1998; Alexandre et al., 2004). However, whereas normal energy taxis supports growth of individual cells, in this case it might rather be a defense strategy that helps to establish favorable conditions for the population (Cypionka, 2000). The activity of the cells prevents oxygen intrusion into the anoxic volume and finally allows anaerobic growth of the culture.
Dilling & Cypionka measured maximum respiration rates at oxygen concentrations below 4% air saturation (about 10 μM O2) (Cypionka, 2000). In accordance with this, our model calculations indicated that the bands formed at these concentrations (Figs 5 and 6). When oxygen concentrations below 4% were applied (Fig. 4) the cells swam towards oxygen, thus increasing their energy level.
Of the two types of bands observed, the second type can be explained by the mechanisms discussed above. The rapidly developing bands of the first type are more difficult to understand. We assume the following explanation: In the beginning the cells are overwhelmed by oxygen diffusing at high speed due to the initially steep gradient. Regardless of their individual swimming direction, oxygen concentration increases with time, and increases above their optimum energetization. Cells entering the oxic zone from the anoxic side are trapped for the same reason. Consequently, zones with increased cell density at the oxic side, and decreased density at the anoxic side of the interface develop. The band is sharp in the beginning and broadens with time.
Consequently, the oxygen gradient flattens, enabling the cells to escape the first band and to form a new band (type 2) at the preferred oxygen level (Fig. 3b). At this time, a new band already may have been established by other cells, leading both types coexisting for a while (Fig. 3c). These double bands may therefore be different from those found with E. coli by Adler (1966), who proposed different metabolic reactions within the different bands.
Surfaces influenced bacterial accumulation in our experiments. Although surface effects are ubiquitous, they have often been neglected in chemotaxis studies. The swimming speed of bacteria decreases when they approach solid surfaces (Frymier & Ford, 1997). This can be explained by hydrodynamic models (Ramia et al., 1993). The same models predict clockwise or anticlockwise tracks for cells near surfaces, and have been confirmed with E. coli (Vigeant & Ford, 1997). Earlier studies describe swimming in circles in bands of D. desulfuricans (Eschemann et al., 1999; Cypionka, 2000). This observation was interpreted as typical for band-forming cells. However, as the best microscopic view on thick cell suspensions can be obtained by focusing the layer close underneath the coverslip, it might be that the circular moving was caused by surface effects. The same moving pattern could be demonstrated in the absence of an oxygen gradient by time-resolving analysis of video recordings from cells swimming close to the coverslip (Fig. 9).
Although we found higher cell densities near the walls only within the bands, it is possible that this is a general feature that could only be observed as parallax if sharp bands occur. So far it is not clear whether this is a hydrodynamic effect or a real behavioral response. Perhaps hydrodynamic forces start to establish an uneven cell density which results in an uneven oxygen distribution by respiration. Then this would give way to self-stabilizing patterns due to aerotaxis. In very dense suspensions, cells may interact with their neighbors as they do with surfaces (Ramia et al., 1993).
The aerotactic behavior of D. desulfuricans is not adequately described by a simple model of reaction to a single stimulus which alters the length of straight runs. The reactions are more complex, swimming speed is changing, and surface effects have to be considered. Under certain conditions, even bioconvection may occur. Our observations were made with laboratory cultures; however, comparable cell densities and steep oxygen gradients can be reached in densely populated natural habitats. More detailed information on oxygen distribution within the gradients could be obtained by applying planar optodes (Glud et al., 1996), which allow fine-scale two-dimensional mapping of oxygen concentrations.
The authors thank Kai Wirtz (Geesthacht), Hans-Peter Grossart (Neuglobsow) and Andrew Kvassnes Sweetman (Bremen) for helpful comments on the manuscript.
- 1966) Chemotaxis in bacteria. Science 153: 708 – 716. (
- 2004) Ecological role of energytaxis in microorganisms. FEMS Microbiol Rev 28: 113 – 126. , & (
- 1893) Über Atmungsfiguren beweglicher Bakterien. Zentrabl Bakteriol Parasitenkd 14: 827 – 845. (
- 1972) Chemotaxis in Escherichia coli analysed by three-dimensional tracking. Nature 239: 500 – 504. & (
- 1975) The Mathematics of Diffusion. 2nd edn. Oxford University Press, London. (
- 1989) Characterization of sulfate transport in Desulfovibrio desulfuricans. Arch Microbiol 152: 237 – 243. (
- 2000) Oxygen respiration by Desulfovibrio species. Annu Rev Microbiol 54: 827 – 848. (
- 1986) Growth yields of Desulfotomaculum orientis with hydrogen in chemostat culture. Arch Microbiol 143: 366 – 369. & (
- 1990) Aerobic respiration in sulfate-reducing bacteria. FEMS Microbiol Lett 71: 123 – 128. & (
- 1881) Neue Methode zur Untersuchung der Sauerstoffausscheidung pflanzlicher und tierischer Organismen. Pflügers Arch Gesammte Physiol 25: 285 – 292. (
- 1999) Aerotaxis in Desulfovibrio. Env Microbiol 1: 489 – 494. , & (
- 1994) Motility and chemosensory behaviour of the sulphur bacterium Thiovulum majus. Microbiology 140: 3109 – 3116. (
- 1991) Measurement of bacterial random motility and chemotaxis coefficients: I. Stopped-flow diffusion chamber assay. Biotech Bioeng 37: 647 – 660. , , & (
- 1997) Analysis of bacterial swimming speed approaching a solid–liquid interface. AIChE J 43: 1341 – 1347. & (
- 1996) Planar optrodes, a new tool for fine scale measurements of two dimensional O2 distribution in benthic microbial communities. Mar Ecol Prog Ser 140: 217 – 226. , , & (
- 1993) Influence of oxygen on sulfate reduction and growth of sulfate-reducing bacteria. Arch Microbiol 159: 168 – 173. , & (
- 1993) The role of hydrodynamic interaction in the locomotion of microorganisms. Biophys J 65: 755 – 778. , & (
- 1997) The Aer protein and the serine chemoreceptor Tsr independently sense intracellular energy levels and transduce oxygen, redox, and energy signals for Escherichia coli behavior. Proc Natl Acad Sci USA 94: 10541 – 10546. , , , , , & (
- 2002) Growth and chemosensory behavior of sulfate-reducing bacteria in oxygen-sulfide gradients. Microbiol Ecol 40: 47 – 54. , , , , & (
- 1996) Chemotaxis. Escherichia coli and Salmonella. 2nd edn (NeidhardtFC, eds), pp. 1103 – 1129. ASM Press, Washington, DC. & (
- 1998) In search of higher energy: metabolism dependent behaviour in bacteria. Molecular Microbiology 28: 683 – 690. & (
- 1999) Aerotaxis and other energy-sensing behavior in bacteria. Annu Rev Microbiol 53: 103 – 128. , & (
- 2003) Bacteria are not too small for spatial sensing of chemical gradients: an experimental evidence. Proc Natl Acad Sci USA 100: 5748 – 5753. & (
- 1997) Interactions between motile Escherichia coli and glass in media with various ionic strengths, as observed with a three-dimensional-tracking microscope. Appl Environ Microbiol 63: 3474 – 3479. & (