Aerotactic responses in bacteria to photoreleased oxygen


*Corresponding author. Tel.: +1 (808) 956 6945; Fax: +1 (808) 956 5339, E-mail address:


Bacterial aerotaxis is a rapid response towards or away from oxygen. Here we report on the use of computer-assisted motion analysis coupled to flash photolysis of caged oxygen to quantify aerotactic responses in bacteria. The caged compound (μ-peroxo)(μ-hydroxo)bis[bis(bipyridyl) cobalt(III)] perchlorate liberates molecular oxygen upon irradiation with near-UV light. A mixture of cells and the caged oxygen compound was placed in a capillary tube and challenged by discrete stimuli of molecular oxygen produced by photolysis. We then recorded the rate of change of direction (rcd) as an estimate of tumble frequency in response to liberated oxygen and measured the signal processing (excitation) times in Bacillus subtilis, Bacillus halodurans and Escherichia coli. This computer-assisted caged oxygen assay gives a unique physiological profile of different aerotaxis transducers in bacteria.


Aerotaxis is the migratory response towards or away from oxygen and is used by various microorganisms to swim toward an optimal oxygen concentration for their metabolism. Beginning with Engelmann's report in 1881 that bacteria such as Bacterium termo and Spirillum tenue move to surround oxygen-producing plant cells, physiological methods have played an essential role in characterizing different aerotactic responses in motile bacteria[1]. Baracchini and Sherris[2] developed the aerotactic capillary assay and demonstrated that different microbial species form rings of different radii around air bubbles. The assay relies on oxygen consumption from bacterial respiration, which creates oxygen gradients between the open ends and the interior of the capillary. Later, a temporal assay was developed which involved step changes in the oxygen concentration within a gas perfusion chamber[11]. However, inherent limitations in the architecture of the perfusion system precluded rapid kinetic measurements of the excitation phase.

Aerotaxis sensors in Archaea and Bacteria include the heme-containing myoglobin-like HemAT transducers in Halobacterium salinarium and Bacillus subtilis[5–7], FAD-containing Aer transducers in Escherichia coli[3,4,8] and Pseudomonas putida[9], and the Tsr chemoreceptor in E. coli[4]. The HemAT transducers bind oxygen and are therefore true oxygen sensors. Capillary assays showed aerotactic band formation in wild-type B. subtilis cells and in methyl-accepting chemotaxis proteins (MCP) minus cells overexpressing hemAT-Bs from a plasmid, but not in cells deleted for all 10 putative MCP-like transducers[5]. In HemAT-Bs, a 176-residue N-terminal domain is responsible for binding heme and sensing oxygen[6], with His123 binding to the heme group[7].

In contrast to HemAT, the aerotaxis sensor Aer in E. coli does not bind oxygen directly, but likely infers the oxygen level from changes in the redox state of the electron transport chain[8]. This trait gives Aer the ability to sense other electron acceptors or molecules that can alter internal energy supply. It has therefore been termed an ‘energy sensor’. Tsr lacks a cofactor, so sensing must differ from that by Aer or the HemAT transducers. In E. coli, aerotaxis has been assayed spatially by capillary, bubble, and succinate-swarm assays, as well as by temporal assays in a gas perfusion chamber [3,4]. Aerotaxis was completely abolished in an aer tsr double mutant and restored by the expression of either Aer or Tsr from a plasmid[4]. Aer has also been implicated in aerotaxis in P. putida[9].

Techniques for resolving the excitation phase in bacterial chemotaxis signaling include ionophoretic stimulation and photorelease of caged chemoeffectors [10–14]. The photorelease assay is based on the rapid liberation or uncaging of an effector molecule by UV light directed at a population of swimming cells. Extensive studies using such caged compounds have demonstrated that E. coli responds to step increases in chemoeffector concentration within milliseconds [13,14]. Recently, (μ-peroxo)(μ-hydroxo)bis[bis(bipyridyl)cobalt(III)] perchlorate (i.e. caged oxygen compound) has been synthesized and used to study the reduction of diatomic oxygen to water via cytochrome c oxidase [15–17]. To assess the feasibility of using caged oxygen to study aerotaxis, we developed a temporal assay based on the photorelease of oxygen and tested it in B. subtilis, E. coli and Bacillus halodurans.

2Materials and methods

2.1Bacterial strains and growth conditions

B. subtilis OI1085 (wild-type), OI3545 Δten (all ten transducers deleted), OI3555 Δten+hemAT-Bs (overexpression of HemAT-Bs), and Δten+hemAT-BsH123A (overexpression of HemAT-BsH123A) were grown in nutrient broth for 48 h for sporulation [5,7]. Spore morphology was inspected under the microscope before washing with sterile HPLC-grade water and freezing with liquid nitrogen. Motile cells were obtained by growing frozen spores in Luria–Bertani (LB) media supplemented with the appropriate antibiotics. Experimental cultures were grown in a shaker at 37°C to OD600= 1.5.

E. coli strains RP437 (wild-type) and RP5882 *(tsr) were provided by Dr. J.S. Parkinson, University of Utah. E. coli strain RP437 and isogenic strains RP5882 Δ(tsr), BT3391 pTrc Δ(aer), BT3312 Δ(aer tsr), BT3388 Δ(aer tar tsr trg tap), and BT3388 pGH1 Δ(aer tar tsr trg tap)aer+ (aer overexpression) were all grown in LB medium supplemented with 1 μM thiamine and selected for motility on tryptone 0.28% agar plates also supplemented with 1 μM thiamine. Motile cells were grown at 30°C in LB media with 1 μM thiamine to OD600= 0.40–0.50.

The gene replacement vector pKO3 (G.M. Church, Harvard University, Cambridge, MA) was used to create an in-frame deletion of the E. coli aer gene. Upstream and downstream regions of the aer gene were amplified by PCR and cloned into the Sma I site of pKO3. A 0.9-kb erythromycin cassette was then amplified by PCR and cloned into engineered Bgl II restriction sites between the upstream and downstream components. The vector was transformed into E. coli strain HCB339 (Δtsr, Δtar-tap, trg::Tn10) and RP437, and gene deletion was carried out by the methods of Link et al.[18] to create the aer strains BT3388 and BT3391, respectively. The only variation in procedure included growing colonies from the 43°C LB agar plate in 5 ml LB medium with 7.5% sucrose overnight before plating onto LB agar containing 7.5% sucrose. This alteration increased the efficiency of strain creation to almost 100%. Gene replacements were confirmed by PCR using primers flanking the aer gene.

The alkaliphilic B. halodurans strain C-125 provided by Dr. H. Takami, Japan Marine Science and Technology Center was grown on Horikoshi II medium[19] for 48 h or until sporulation. Spore morphology was inspected under the microscope, after which spores were washed with HPLC-grade water and frozen with liquid nitrogen. Motile cells were obtained by growing frozen spores in Horikoshi II medium to OD600= 1.5 in a shaker at 37°C.

2.2Preparation of caged oxygen

Two mg of (μ-peroxo)(μ-hydroxo)bis[bis(bipyridyl) cobalt(III)] perchlorate, ɛ= 4900 M−1 cm−1 (Molecular Probes, Eugene, OR, USA), was dissolved in 500 μl of 50 mM potassium phosphate buffer (pH 7.4), diluted to 13.1 mM, frozen with liquid nitrogen and stored at −70°C until further use (during initial studies we used caged oxygen compound synthesized according to MacArthur et al.[15]).

2.3Motion analysis

Caged oxygen compound (4.37 mM final concentration) was added to B. subtilis and E. coli cells in LB media and to B. halodurans in Horikoshi II media (OD600= 0.35). Flat capillary tubes (0.1×1.0 mm, #5010, VitroCom, NJ, USA) were then half-filled with the bacterial suspension/caged oxygen mixture. The capillaries were sealed at both ends with Critoseal (Monoject. Sci. Division, MO, USA), placed on the microscope stage, and observed with a 40× glass phase-contrast objective. A green interference filter was interposed between the sample and light source to enhance video contrast and facilitate tracking of motile cells. Data collection occurred 30 to 45 min after preparation of the capillary tubes. Duration of the near-UV light flash was 900 ms for B. subtilis and 100 ms for E. coli and B. halodurans. Images of swimming bacteria were captured with a CCD camera and digitized with a VP110 video digitizer (VP120 digitizer, Expert Vision 2D/AT Release 3.1 version software; Motion Analysis, Santa Rosa, CA, USA). The digitizer was equipped with a −5 V event tone marker that signals the opening of the shutter (UniBlitz, Vincent Associates, Rochester, NJ, USA) with a 5 s delay. A custom-made NOT gate circuit connected to a Grass stimulator was interposed between the digitizer and shutter driver/timer (UniBlitz, model T132) to convert the event tone to a +35 V pulse needed to activate the shutter system. Illumination from a 100-W mercury short arc lamp (HBO 100 W/2, Osram, Munich, Germany) was passed through the electronic shutter and filtered through a Nikon fluorescence cube (360±20 nm excitation, >400 nm barrier filter, 400 nm dichroic mirror). Intensity of the UV light, measured with a silicon photodiode (XRL340B #5348, International Light, Newburyport, MA, USA), was 1.60×10−4 W cm−2. The amount of oxygen released from 4.37 mM caged oxygen was estimated to be approximately 3.27× 10−13 mol (0.046 pmol) for a 900-ms flash and 3.63×10−15 mol (0.0051 pmol) for a 100-ms flash (calculations not shown) using a quantum yield of 0.1[16].

When flash photolysis was performed immediately after bacteria were introduced into the capillary tube, wild-type B. subtilis failed to respond because receptors were saturated with oxygen. A 30–60 min incubation period was used to assure a semi-anaerobic environment. A field of view (approximately 150×150 μm) containing 20–40 cells with a rate of change of direction (rcd) = 700–800° s−1 for B. subtilis and 700–1200° s−1 for E. coli was selected for observation. Data collection began 5 s before the flash, and an average of 1000 paths (30 repetitive assays) were recorded for 30 s. A stage micrometer was used to calibrate image measurements (scale factor = 0.795 μm/pixel). Although UV light acts as a repellent, it does not prevent a positive response to oxygen to occur.

2.4Agar plug assay and effect of 360±20 nm UV light for B. subtilis

Cobalt bipyridine is released as a byproduct of photolysed caged oxygen. To ensure that oxygen and not the byproduct is the attractant, we performed an agarose-in-plug bridge time-lapse dark-field microscopy assay[20]. LB growth medium and 10 mM serine were used as positive controls, with buffer as a negative control. UV light is the third possible stimulus, which might cause a change in tumble frequency. To rule out this possibility, we performed a flash experiment with buffer replacing caged oxygen.

2.5Error calculations

The mean rcd value (rcdmean) and rcd standard deviation value (rcdS.D.) was calculated using rcd values between the 2–5 s baselines. A truncated rcd plot between 5–7.5 s range with a cubic polynomial algorithm to generate a best-fit curve yielded the peak rcd value (rcdpeak). The error in Δrcd = rcdpeak−rcdmean±rcdS.D.. The standard deviations produced with the best-fit curve were small (<0.005 s), and were therefore ignored in these calculations. We found a 0.02-s delay in shutter signal speed and corrected the results accordingly. Excitation response rates, kex, were determined according to Khan et al. and MacArthur et al. [14,15] from single exponential fits. The response half time, t1/2(= 1n 2/kex) was determined by a logistic fit according to Renate et al.[21].

3Results and discussion

Baracchini and Sherris were among the first to describe the aerophilic nature of Bacteria[2]. In this report we demonstrate the measurement of excitation responses to discrete stimuli produced by photolysis of caged oxygen in Bacteria using computerized video analysis.

3.1Oxygen sensing in B. subtilis

HemAT-Bs in B. subtilis binds oxygen reversibly and triggers aerotactic responses[5]. Thus, by using the caged oxygen compound to release diatomic oxygen at a given moment, it is possible to test behavioral responses directly related to oxygen-binding characteristics of HemAT-Bs. We have tested this method with wild-type B. subtilis and observed a rapid smooth swimming response towards photoreleased oxygen. Wild-type B. subtilis cells increase tumble frequency when flashed with 900 ms UV (360±20 nm) in the absence of caged oxygen (data not shown). This result is consistent with previously observed swimming behavior of UV-irradiated bacteria [22–24]. When flashed with UV in the presence of caged oxygen, the cells exhibited a strong smooth response with an excitation time of 0.28±0.04 s (Fig. 1A). A strain lacking in all ten transducers, Δten, did not respond to released oxygen (Fig. 1B), indicating that the presence of oxygen-sensing transducers is essential for aerotactic responses. Indeed, in a strain with overexpression of hemAT-Bsten+hemAT-Bs), photorelease of oxygen triggered a strong smooth response with an excitation time of 0.31±0.04 s similar to that of the wild-type (Fig. 1C).

Figure 1.

Responses of B. subtilis strains to photoreleased oxygen. Wild-type (A) and Δten+hemAT-Bs (C) showed a decrease in rcd (increased smooth swimming response) following a 900-ms flash and subsequent oxygen release, but Δten (B) and Δten+hemAT-BsH123A (D) showed no significant response.

To further confirm that these responses are triggered by HemAT-Bs, we tested B. subtilis strain Δten+hemAT-BsH123A. We have previously demonstrated that His123 is a proximal residue in HemAT-Bs and plays an important role in oxygen binding[7]. Replacing this proximal histidine with an alanine residue destroys the heme-binding capability of HemAT-Bs[7]. Fig. 1D shows that the mutant strain Δten+hemAT-BsH123A does not undergo an aerotactic response.

3.2Oxygen sensing in E. coli and B. halodurans

To test our method in other bacteria, we analyzed aerotactic responses in E. coli and B. halodurans cells towards flash photoreleased oxygen. Wild-type E. coli RP437 showed a positive response to photoreleased oxygen with an excitation time of 0.21±0.04 s (Fig. 2A). A single deletion of tsr or aer genes did not eliminate the aerotactic response (Fig. 2C, D). A strain lacking Tsr, RP5882 Δ(tsr), shows an excitation time of 0.22±0.04 s (Fig. 2C), and BT3391 pTrc Δ(aer) shows an excitation time of 0.20±0.04 s (Fig. 2D). Both the double mutant BT3312 Δ(aer tsr) (Fig. 2E) and the strain missing all five transducers BT3388 Δ(aer tar tsr trg tap, Fig. 2B) did not show significant response to photoreleased oxygen, indicating that oxygen-sensing transducers are required for aerotactic responses. Indeed, overexpression of aer in BT3388 strain yielded a large decrease in rcd with an excitation time of 0.59±0.04 s, confirming the oxygen-sensing capability of this receptor (Fig. 2F). Flash experiments without caged oxygen showed the expected negative response, indicating that the UV acted as a repellent (data not shown), as has been previously reported [22–24].

Figure 2.

Responses of E. coli strains to photoreleased oxygen. Wild-type (A), Δtsr (C), Δaer (D), and Δ(aer tar tsr trg tap) aer+ (F) showed a decrease in rcd following a 100-ms flash and subsequent oxygen release, but Δ(aer tar tsr trg tap) (B) and Δ(aer tsr) (E) showed no significant response.

Our results showed that strains with Tsr adapted more slowly to changing oxygen concentrations than strains with Aer. Tsr-containing strains not only took longer to reach the pre-stimulus stage but also showed shorter excitation response times than did Aer-containing strains. When the transducer genes hemAT-Bs and aer were overexpressed, excitation times became longer (Δten+hemAT-Bs∼0.31±0.04 s and aer+ 0.59±0.04 s). Spudich and Koshland[25] showed that the duration of the smooth swimming response is proportional to receptor occupancy, i.e. the number of receptors titrated. The prolonged duration of the responses in these two strains (Fig. 2C, F) might be due to a larger number of Aer receptors that require a longer time for the adjustment to occur.

We have previously demonstrated that the globin-coupled sensors (GCS) motif-type transducer from B. halodurans C-125 (GCSBh) binds diatomic oxygen [7,26]. We postulate that GCSBh should display an aerotactic response towards photoreleased oxygen. Indeed, Fig. 3A shows that when cell suspensions containing caged oxygen were flashed with UV light, B. halodurans strain C-125 exhibited increased smooth swimming behavior with an excitation time of 0.20±0.04 s (Fig. 3A). In the absence of caged oxygen, the UV flash caused no decrease in rcd or increase in smooth swimming (Fig. 3B). HemAT-Bs and GCSBh are homologs[26]. Our results suggest that GCSBh is responsible for oxygen-sensing behavior in B. halodurans, as HemAT-Bs is in B. subtilis.

Figure 3.

Response of B. halodurans C-125 to photoreleased oxygen. B. halodurans C-125 showed a decrease in rcd following a 100-ms flash in the presence of caged oxygen (A), but did not significantly respond when caged oxygen was absent (B).

In addition to oxygen, our assay introduces two other potential stimuli, UV light and cobalt bipyridine. To ensure that cobalt bipyridine does not act as an attractant, we performed an agarose plug chemotaxis assay[20]. Wild-type cells did not accumulate around an agarose plug filled with either water or cobalt bipyridine (data not shown). However, these cells formed a diffuse band around an agarose–serine plug and a thick band around an agarose–LB–serine plug (data not shown), indicating that they were capable of chemotaxis. Thus, cobalt bipyridine is not an attractant and did not contribute to the smooth oxygen response. UV light does (data not shown), however, act as a repellent in control experiments, suggesting that B. subtilis, E. coli and B. halodurans possess a sensing mechanism for this stimulus.

In conclusion, our results show that, despite the low quantum yield of caged oxygen (0.1), the obligate aerobe B. subtilis as well as the facultative aerobes E. coli and B. halodurans were able to respond to diatomic oxygen produced by photolysis of a caged oxygen compound. Therefore, the technique is applicable to a wide range of microorganisms.


We thank Sandy Parkinson for critical comments and JoAnn Radway for editorial help and comments on the manuscript. This investigation was supported by the National Science Foundation Grant No. MCB 0080125 to M.A., MCB 9904713 to R.W.L., and by University of Hawaii intramural funds awarded to M.A.