Cyanide Production by Chromobacterium piscinae Shields It from Bdellovibrio bacteriovorus HD100 Predation

ABSTRACT Predation of Chromobacterium piscinae by Bdellovibrio bacteriovorus HD100 was inhibited in dilute nutrient broth (DNB) but not in HEPES. Experiments showed that the effector responsible was present in the medium, as cell-free supernatants retained the ability to inhibit predation, and that the effector was not toxic to B. bacteriovorus. Violacein, a bisindole secondary metabolite produced by C. piscinae, was not responsible. Further characterization of C. piscinae found that this species produces sufficient concentrations of cyanide (202 µM) when grown in DNB to inhibit the predatory activity of B. bacteriovorus, but that in HEPES, the cyanide concentrations were negligible (19 µM). The antagonistic role of cyanide was further confirmed, as the addition of hydroxocobalamin, which chelates cyanide, allowed predation to proceed. The activity of cyanide against B. bacteriovorus was found to be twofold, depending on the life cycle stage of this predator. For the attack-phase predatory cells, cyanide caused the cells to lose motility and tumble, while for intraperiplasmic predators, development and lysis of the prey cell were halted. These findings suggest that cyanogenesis in nature may be employed by the bacterial strains that produce this compound to prevent and reduce their predation by B. bacteriovorus.

One well-known secondary metabolite with defensive capabilities is indole. With respect to predation, indole is both toxic to the bacterivorous nematode Caenorhabditis elegans (5) and capable of blocking predation of Escherichia coli and Salmonella by the bacterial predator Bdellovibrio bacteriovorus HD100 (6). Indole is produced by both Gram-positive and Gram-negative bacteria (7,8) and can reach concentrations of up to 1.1 mM within mammalian guts (9,10). With C. elegans, a significant loss in viability was seen when indole was present at 0.5 mM or higher (5), while concentrations between 1 and 2 mM inhibited B. bacteriovorus mobility and hindered its development within the bdelloplast (6).
In addition to violacein, Chromobacterium violaceum also produces cyanide (25,26), a potent inhibitor of cytochrome c oxidase (27). As a secondary metabolite, cyanide is generated by C. violaceum using glycine through an oxidative decarboxylation step (25,26,28), although other amino acids may also be used (29). Studies with Pseudomonas aeruginosa PAO1, which is also cyanogenic, found that the cyanide produced by this bacterium was sufficient to completely kill C. elegans cultures (30), illustrating its potential as a deterrent against predation.
Similarly, Serratia marcescens is known to produce several detergents, or serrawettins, that repel C. elegans (31). Pradel et al. (31) found that purified serrawettin W2 elicited a lawn avoidance response from C. elegans with E. coli OP50, i.e., the microbe typically used to cultivate this nematode, and that this response was mediated by its sole Toll-like receptor, encoded by tol-1. Prior to their study (31), the role of this receptor specifically in C. elegans avoidance of S. marcescens was known (32), but how C. elegans recognized this pathogen was not known. Burlinson et al. (33) also describes the characterization of the Pseudomonas fluorescens NZ17 EDB gene cluster, whose product repels C. elegans but does not kill it. Although the exact compound has yet to be identified, they were able to show that it is either poorly diffusible or associated with the bacterial cell, as C. elegans needed to encounter P. fluorescens NZ17 before the aversion response was initiated (33). Furthermore, using two C. elegans mutants, tax-2 and tax-4 mutants, they also demonstrated that this response involves both chemotactic and nonchemotactic elements.
In this study, we show that C. piscinae is preyed upon by B. bacteriovorus HD100 when provided in HEPES buffer but that it was resistant when the tests were performed in dilute nutrient broth (DNB). Through a series of experiments, we identified that the effector was present within the cell-free supernatants and, similar to the EDB cluster product, it was inhibitory but not overtly toxic toward B. bacteriovorus HD100. the concentration of the inhibitor from this time point was sufficient to block B. bacteriovorus HD100. Using the supernatants after 24 h, we also found that the activity is dose dependent (Fig. 1D).
Violacein is not the inhibitor. As one of the secondary metabolites produced by C. piscinae is the bisindole violacein ( Fig. 2A), we initially thought that it was the inhibitor due to its structural similarity with indole (6). Cultures of C. piscinae grown in DNB produced, on average, between 0.3 and 0.5 mg/liter of violacein, or around 1 to 1.5 M, although its concentration in the cell-free supernatants was close to zero, since it is hydrophobic and primarily associates with the C. piscinae cells. When C. piscinae is grown in NB medium, the violacein concentration can reach roughly 5 mg/liter (15 M) (17). To evaluate whether violacein inhibited predation, we used the higher concentration (Fig. 2B). The same set of experiments was also performed using a commercial preparation of violacein that was produced by Janthinobacterium lividum to determine whether there are any differences based upon the host strain (Fig. 2B). In both cases, predation of E. coli MG1655 by B. bacteriovorus HD100 was not blocked.
C. piscinae produces cyanide, which inhibits predation. In addition to violacein, C. piscinae was also found to produce a significant amount of cyanide when grown in DNB medium, with an average of 202 M after 24 h (Fig. 3A, inset). Parallel experiments FIG 1 (A) Impact of the media on predation of C. piscinae by B. bacteriovorus HD100. Control cultures (unpredated) and predated cultures are indicated by the letters C and P, respectively, after the medium. A predator-to-prey ratio (PPR) of approximately 0.03 was used. Values are means plus standard deviations (error bars). This experiment was repeated eight times. (B) Filter-sterilized supernatants from C. piscinae grown in DNB protect E. coli MG1655 from predation by B. bacteriovorus HD100. Supernatants from C. piscinae cultures incubated for 24 h in HEPES or DNB were tested. The initial E. coli viability and unpredated and predated E. coli viability are shown. A PPR of approximately 0.03 was used. Values that are not statistically significantly different (P Ͻ 0.05) are indicated by the same letter (letter a or b). This experiment was repeated three times. (C) Correlation between the C. piscinae growth in DNB and the inhibitory activity with E. coli as the prey. At each time point, a sample of the C. piscinae culture was taken to measure the OD 600 . The cells within another aliquot were removed by filtration, and the inhibitory activity of the supernatant was assessed in predation tests with E. coli MG1655 as the prey. The results show a good correlation between the growth of C. piscinae and the inhibitory activity of its supernatant. A PPR of approximately 0.02 was used. This experiment was repeated three times. (D) The inhibitory activity of the DNB supernatants is dose dependent. Cultures of C. piscinae grown in DNB for 24 h were filter sterilized. The supernatants were then diluted into HEPES, and their inhibitory activity was studied using E. coli MG1655 as the prey. The E. coli viabilities were measured after 24 h. The values for pairs of samples (predated and unpredated samples) were compared. Values that are significantly different are indicated by asterisks as follows: *, P Ͻ 0.05; **, P Ͻ 0.01; ***, P Ͻ 0.001. This experiment was repeated four times.
Cyanogenesis Adversely Impacts Bacterial Predation ® within NB medium found that after 24 h, the cyanide concentration was between 600 and 800 M, while cultures incubated in HEPES produced very little or no cyanide (Fig. 3A, inset). A strong correlation is evident when the inhibitory activities of the supernatants (Fig. 1C) are compared with their respective cyanide concentrations (Fig. 3A). In P. aeruginosa, the cioAB genes encode a cyanide-insensitive oxidase, making this organism resistant to its own cyanide production (34). A BLASTP analysis using the P. aeruginosa CioA and CioB amino acid sequences (NCBI GenBank accession no. CAA71555.1 and CAA71556.1, respectively) found homologues encoded by C. piscinae genes (C. piscinae CioA [NCBI accession no. KIA81027] with 48% identity with P. aeruginosa CioA and C. piscinae CioB [NCBI accession no. KIA81026] with 43% identity with P. aeruginosa CioB). However, neither of these genes were found within the B. bacteriovorus HD100 genome, helping to explain why C. piscinae is resistant to cyanide while this predatory bacterium is not. This compound is produced by various bacterial strains. (B) Violacein does not negatively impact predation. Both a preparation purified from cultures of C. piscinae and a commercially available violacein were tested. E. coli MG1655 was used as the prey for B. bacteriovorus HD100. The initial E. coli viability (Initial), viability of E. coli cultures exposed to DMSO for 24 h with or without predation (DMSO), and viability of E. coli cultures exposed to 20 mg/liter violacein (in DMSO) for 24 h with or without predation (Violacein) are shown. A PPR of approximately 0.03 was used. Values are means plus standard deviations (error bars). Values that are not statistically significantly different (P Ͻ 0.05) are indicated by the same letter (letter a, b, or c). This experiment was repeated three times.
The impact of cyanide on predation was further confirmed in tests using potassium cyanide (KCN) ( Fig. 3B and C). When the KCN concentration was 200 M or greater, predation was completely blocked. As shown in Fig. 3B and C, under these conditions, the prey and predator populations after 24 h were similar with their initial values, i.e., Cyanogenesis Adversely Impacts Bacterial Predation ® 3.1 ϫ 10 8 CFU/ml for E. coli and 1.3 ϫ 10 7 PFU/ml for B. bacteriovorus HD100. Treatment of B. bacteriovorus HD100 with 200 M KCN significantly reduced the motility of this predatory bacterium. Movie S1 in the supplemental material shows the predatory cells when present in HEPES buffer alone with an average speed of 18.2 Ϯ 4.4 m/s (n ϭ 210), while in Movie S2, they were exposed to 200 M KCN and had an average swimming speed of only 4.5 Ϯ 2.8 m/s (n ϭ 210). The reduced motility (6,35) helps to explain the weaker predatory activities seen with cyanide.
To test whether cyanide alone was responsible or whether some other compound also acted as an inhibitor, two additional experiments were performed (Fig. 3D): purging and treatment of the supernatant with vitamin B 12a . In a neutral (pH 7) environment, cyanide is present primarily as hydrogen cyanide (HCN), which is less dense than air and volatile. Taking advantage of this characteristic, we purged the media with nitrogen gas and reduced the cyanide concentration in the media from 230 M to only 20 M. Likewise, vitamin B 12a , or hydroxocobalamin, acts as a scavenger of cyanide, forming cyanocobalamin (36,37), which prevents cyanide from binding to cytochrome c oxidase. After either treatment, B. bacteriovorus HD100 was able to prey upon E. coli MG1655 (Fig. 3D), confirming that cyanide is the main, if not sole, inhibitor responsible.
Cyanide halts B. bacteriovorus HD100 development in the bdelloplast. Although cyanide inhibited predation, it was not very toxic to attack-phase (AP) B. bacteriovorus HD100, as shown in Fig. 4A. The viability decreased slightly from 100 M cyanide, with the greatest loss in viability seen at 400 M cyanide. Even at this concentration, however, the loss in viability was only 51%. When similar experiments were performed using bdelloplasts, no toxic effect was seen for any of the concentrations tested (Fig. 4B). In the control samples, the number of predators increased after 4 h and was on average 5.6-fold higher than the initial populations. A similar increase was seen in the 50 M samples, although there appears to have been a delay since the number of B. bacteriovorus did not increase until 6 h. For the higher concentrations, there was no observable increase in the predator numbers after either 4 or 6 h. When bdelloplasts within the control cultures (0 M cyanide) and cultures treated with 200 M cyanide were imaged using confocal microscopy, we found that the development of the intraperiplasmic predator was halted after the addition of cyanide (Fig. 4C). In the control cultures, the predators within the bdelloplasts were elongated at 3 h, while only AP B. bacteriovorus HD100 cells were seen at 6 h, since lysis of the prey had already occurred. In contrast, the addition of 200 M cyanide after 1 h halted the growth and development of the intraperiplasmic predator, as shown in Fig. 4C. Even after 6 h, the bdelloplasts are essentially indistinguishable from those imaged just before cyanide was added.

DISCUSSION
In this study, we show that predation of C. piscinae by B. bacteriovorus HD100 occurs in HEPES buffer, although it took 3 days to see a 60-fold drop in prey viability. Compared to other prey strains, this is rather protracted and implied that C. piscinae is inherently resistant to predation by B. bacteriovorus HD100. When the same tests were performed in DNB, there was no loss in the viability of C. piscinae, a result that implied that this bacterium produces an inhibitor when provided with amino acids. The only other report found describing the predation of a Chromobacterium strain was a study by Jurkevitch et al. (38). They found that four out of the five predatory strains they tested were capable of predating on C. violaceum but used HEPES buffer as the medium. On the basis of our findings here, this would be permissive for predation to occur.
Violacein is formed by a variety of bacterial strains, including C. piscinae (17), through a condensation reaction involving two tryptophan molecules (39,40). Given violacein's bisindole chemical structure and the established activity of indole against B. bacteriovorus HD100 (6), we initially thought that this compound was the C. piscinae inhibitor being produced. According to Choi et al. (17), C. piscinae produces slightly less than 4.5 mg/liter of violacein when grown for 24 h in NB medium. Although this is equal to a concentration of 15.4 M, which is far lower than the 1 mM needed for indole to inhibit predation (6), it was much higher than what was actually seen in the DNB cultures (1.2 M). Tests with both crude and commercial violacein preparations at the higher concentration found that this compound does not inhibit predation, consequently, violacein is not responsible for the loss in activity seen.
In addition to violacein, C. piscinae also produced cyanide when grown in DNB medium. Although the production of cyanide by C. violaceum was established long ago (25,41,42), this is the first report demonstrating that C. piscinae also produces cyanide. Cyanide is formed during the oxidative decarboxylation of glycine (25,26,28) and of glutamate and methionine (29). As the DNB medium used in this study contains both FIG 4 (A) Cyanide is mildly, yet significantly, toxic to attack-phase B. bacteriovorus HD100. B. bacteriovorus cultures were exposed to different micromolar concentrations of cyanide for 24 h, after which the viable populations were determined. A total of 12 independent cultures were tested. Each symbol represents the value for one culture, and the average result for a particular cyanide concentration is indicated by a light gray circle. Values were compared to the value for the no-cyanide control samples. Values that are significantly different (P Ͻ 0.001) are indicated (***). This experiment was repeated 12 times. (B) Cyanide prevents release of the predatory cells from the bdelloplast. This graph shows the numbers of viable predators, which include both free-swimming predators and those within the bdelloplast, seen during and after a single predation event. The relative number illustrates the titer burst seen when lysis of the bdelloplast occurs. This experiment was repeated six times. (C) Confocal microscopic images showing the impact of cyanide on the development of the predatory cells within the bdelloplasts. E. coli MG1655 prey express the red fluorescent protein (red), while the predator is expressing the Venus yellow fluorescent protein (green). In the presence of 200 M cyanide, elongation and development of the intraperiplasmic predatory cells, and the resulting lysis of the bdelloplast, are halted. This agrees with the results shown in panel B. The 0-h image was taken just prior to introducing the predatory strain. Bars ϭ 10 m.

Cyanogenesis Adversely Impacts Bacterial Predation
® beef extract and an enzymatic digest of gelatin, all three of these amino acids are provided in this medium. When grown in DNB, C. piscinae produced an average of 202 M cyanide after 24 h, although most (143 M) was produced within the first 6 h (Fig. 3A). The production trend in Fig. 3A is quite similar with that published by Michaels et al. (25) during their characterization of C. violaceum. In both cases, there was a lag period that lasted for the first 4 h, after which the cyanide concentrations rapidly increased. The maximum cyanide concentration obtained with C. violaceum was approximately 16 mg/liter, or 600 M, which is threefold higher than that produced by C. piscinae, i.e., 202 M. Aside from likely variations in the two strains, this difference can be attributed to the composition of the medium used in each study, i.e., DNB (0.8 g/liter) here and 10 g/liter peptone in their study, as Michaels et al. (25) also reported that peptone greatly influenced the final cyanide yields with C. violaceum. As a demonstration of this, when C. piscinae was grown in NB (8 g/liter), the final cyanide concentrations were comparable at 600 to 800 M.
Aside from C. violaceum and C. piscinae, bacterial cyanogenesis is best characterized in pseudomonads (28,43), where it has been linked to both reduced nematode viability and egg hatching rates (30,33,44,45). For instance, exposure to P. aeruginosa PAO1 completely (100%) killed C. elegans cultures, while only 15% of the nematode population was killed by an isogenic noncyanogenic mutant strain (30). Likewise, in Siddiqui et al. (44), the egg hatching rates for Meloidogyne javanica in control cultures were 90%. This rate dropped significantly (to 40%) when the cultures were exposed to the cyanogenic Pseudomonas protegens CHA0, but it dropped only slightly (85%) with P. protegens CHA77, an isogenic mutant lacking the ability to produce cyanide. Both studies illustrate the activity of cyanogenic bacteria against predatory nematodes and the potential benefit they gain from producing cyanide.
To date, bacterial cyanogenesis has been identified only in Gram-negative microbes (46), and this may be evolutionarily important given the selective activity of B. bacteriovorus, which attacks only Gram-negative strains. Cyanide acts by binding to and inhibiting cytochrome c oxidase (27), a key enzyme in the respiratory electron transport chain, making it a potentially strong poison for B. bacteriovorus HD100, since this microbe is regarded as a strict aerobe.
Although the effect cyanide has on bacterial predation has not been reported, it was studied from the perspective of electron transport systems of B. bacteriovorus and oxidation of NADH using attack-phase predatory cells or cell lysates (47). The authors found that 5 mM KCN completely inhibited respiration, while 0.5 mM KCN reduced NADH oxidation by 75% in cell-free extracts of B. bacteriovorus strain 6-5-S. Similarly, another study reported that the amino acid uptake by B. bacteriovorus 109D (H-I), a host-independent variant that can grow axenically, was inhibited by as much as 91% by 500 M cyanide, while lower concentrations had lesser impact (48). Moreover, transport of phosphate into both B. bacteriovorus 109J and 109D was completely inhibited by 1 mM cyanide.
Although both studies used cyanide concentrations that were higher than the inhibitory concentrations found in this study, i.e., 100 to 200 M, their findings help to explain the activity of cyanide as seen here. As cyanide impairs electron transport and energy production, as well as inhibits transport of critical nutrients (i.e., phosphate and amino acids) needed for the growth and development of the predator within the bdelloplast, the implication would be a halting of growth for intraperiplasmic predators when the cyanide concentration is sufficiently high. This is what was observed in Fig. 4C. However, our findings also clearly demonstrate that the effects of cyanide were not permanent, since exposed B. bacteriovorus cells still formed plaques and grew later when the cyanide was diluted. This was true for both the free-swimming B. bacteriovorus (Fig. 4A) as well as those within bdelloplasts (Fig. 4B). This is further exemplified in Movie S2 in the supplemental material, where predatory cells exposed to cyanide were still viable, although their ability to swim was severely impaired. This impairment was anticipated, since the flagellar motor activity is dependent upon the proton motive force, which is generated by the electron transport chain (ETC). Given that cyanide binds to cytochrome c oxidase, a member of the ETC, and inhibits any further electron transport, the proton motive force would quickly be spent and bacterial motility partially halted. This was seen in both P. fluorescens (49) and E. coli (50). For P. fluorescens, 2 min was sufficient to achieve a stable level of inhibition, illustrating the rapid loss of motility during exposure. For E. coli, exposure to 100 M cyanide led to both a 63% reduction in the average speed of this bacterium and a 38% reduction in the number of cells that were motile. Both of these studies illustrate the same response as seen with B. bacteriovorus HD100 when exposed to cyanide in this study.
Conclusions. In addition to the clear impact bacterial cyanogenesis has on macrofaunal predators (i.e., nematodes and protozoa), this study illustrates an additional benefit for cyanogenic bacterial strains, namely, that they may be shielded and protected from bacterial predation. The amount of cyanide produced by C. piscinae when provided DNB was sufficient to inhibit predation by B. bacteriovorus and halt the development of the predator within the bdelloplast. From an ecological perspective, this suggests that bacterial cyanogenesis may be used to protect natural microbial communities from predation by B. bacteriovorus. Given the small amount of nutrients present in this medium and the clear impact they had on the predation results, this study also suggests that caution should be taken in defining bacterial strains that are susceptible to predation and those that are resistant, as secondary metabolites clearly can play a role.

MATERIALS AND METHODS
Strains and culturing methods. The strains used in this study were Bdellovibrio bacteriovorus HD100 (DSMZ 50701), Chromobacterium piscinae LMG 3947, and E. coli MG1655 (ATCC 700926). Culturing of B. bacteriovorus HD100 was performed as described previously (51,52) using E. coli MG1655 as the prey. C. piscinae and E. coli MG1655 were grown from frozen stocks on nutrient broth (NB) agar plates at 30°C overnight. A single colony was inoculated into fresh NB medium and grown for 24 h at 30°C and 250 rpm.
For the predation studies, the prey cells were pelleted by centrifugation (15 min, 2,000 ϫ g) and resuspended to an optical density at 600 nm (OD 600 ) of 1.0 in either dilute nutrient broth (DNB; 1/10 NB) or HEPES buffer (pH 7.2), both supplemented with 3 mM MgCl 2 and 2 mM CaCl 2 . The cultures were split in half, and one half had B. bacteriovorus HD100 added to a predator-to-prey ratio (PPR) of approximately 0.03. The other half was used as an unpredated control sample. The initial prey and predator viabilities and viabilities after 24 h were determined as described previously using top agar plates (51).
C. piscinae supernatant experiments. C. piscinae was cultured as described above for 24 h in NB (30°C and 250 rpm), after which the culture was diluted 1:100 into either HEPES buffer or DNB medium. These cultures were incubated for an additional 24 h at 30°C and 250 rpm. At this time, the supernatants were collected by removing the C. piscinae cells through centrifugation (15 min, 2,000 ϫ g) and sterilized using a 0.22-m filter (Millipore, USA). To these supernatants (DNB and HEPES), E. coli MG1655 was added as described above to an OD 600 of 1.0. The cultures were split in half, and one half had B. bacteriovorus HD100 added to a predator-to-prey ratio (PPR) of approximately 0.03. The other half was used as an unpredated control sample. The initial prey and predator viabilities and the viabilities after 24 h were determined as described previously (51). A similar protocol was used for the supernatant dosedependent experiments. For these tests, the cell-free C. piscinae DNB supernatants were diluted using HEPES buffer to achieve the desired percent supernatant. E. coli MG1655 was then added to an OD 600 of 1.0, and the predation tests were performed as described above.
To study the impact of the C. piscinae growth stage on predation, DNB cultures were prepared as described above, and samples were taken at set times (0, 3, 6, 9, 12, and 24 h) for OD 600 measurement and predation tests. The C. piscinae cells were removed from these samples as described above using centrifugation and filtration. E. coli MG1655 was resuspended in the supernatant samples to an OD 600 of 1.0, and B. bacteriovorus HD100 was added to a PPR of approximately 0.03. After 24 h at 30°C and 250 rpm, the OD of the different cultures were measured at 600 nm.
Violacein preparation and tests. To test the effects of violacein, we purchased violacein from Sigma-Aldrich (USA), prepared using Janthinobacterium lividum, as well as purified violacein from the C. piscinae cultures grown in our lab. To purify violacein, we used the ethanol extraction method as described previously (17) followed by crystallization using acetone (11). The concentration of violacein within the samples was determined using high-performance liquid chromatography (HPLC) as described previously (17). This method was used for both the preparation and the supernatant samples.
Stock solutions (100ϫ) containing 0.05 g/liter and 0.5 g/liter of either violacein preparation were prepared in dimethyl sulfoxide (DMSO) (Sigma-Aldrich, USA). To E. coli MG1655 cultures in HEPES, prepared as described above to an OD 600 of 1.0, the violacein stock was added at a 1:100 dilution. The cultures were split in half, and immediately afterwards, B. bacteriovorus HD100 was added to one of the tubes to give a PPR of approximately 0.03. All the tubes were incubated at 30°C with shaking at 250 rpm for 24 h. The prey and predator viabilities were determined initially (0 h) and after 24 h.
Cyanide concentration determination. To measure the cyanide concentrations in the culture media, we used a modified method based upon previous studies (53,54). Briefly, 0.1 M o-dinitrobenzene (Fluka, Germany) and 0.2 M p-nitrobenzaldehyde (Sigma-Aldrich, USA) solutions were prepared in 2-methoxyethanol (Sigma-Aldrich, USA). For each measurement, a fresh 1:1 mixture of these two solutions was made, and this was mixed 77:23 with the supernatant samples (100 l total). Afterwards, 1.8 l of 5 M NaOH was added to each sample, giving a final concentration of 0.09 M NaOH, which was required to adjust the pH prior to measurement. The supernatants were diluted as needed to ensure that the cyanide concentration was within the measurable range. After mixing, the samples were incubated for 30 min at room temperature, and then 900 l of 2-methoxyethanol was added to each tube. From each sample, 100 l was aliquoted into the wells of transparent 96-well plates (SPL, South Korea), and the OD 578 was determined. The cyanide concentrations were calculated using a calibration curve prepared with known KCN solutions.
Purging and hydroxocobalamin experiments. The C. piscinae cell-free DNB supernatants were prepared as described above. Each supernatant sample (20 ml) was then purged with nitrogen gas (flow rate, 3 liters/min) for 1 h. After purging, the medium was reconstituted to 20 ml using sterile water, and the predation experiments were performed as described above using E. coli MG1655 as the prey.
Hydroxocobalamin (vitamin B 12a ) was purchased from Sigma-Aldrich (USA). Stock solutions of this vitamin were prepared using sterile water and subsequently filter sterilized before use. For the experiments, the C. piscinae cell-free DNB supernatants were prepared as described above. To each, 200 M hydroxocobalamin was added just prior to introducing the prey or predator.
Potassium cyanide inhibition of predation. Potassium cyanide (KCN) was purchased from Sigma-Aldrich (USA). A stock solution (1 M) was prepared using sterile deionized water and filter sterilized (0.22-m syringe filter; Millipore [USA]). The cyanide stock was diluted to a 2ϫ concentration using sterile HEPES buffer and subsequently mixed 1:1 (vol/vol) with an E. coli culture, prepared in HEPES to an OD of 2.0. Immediately afterwards, B. bacteriovorus HD100 was added to the samples to obtain a PPR of approximately 0.03. After 24 h at 30°C and 250 rpm, the surviving E. coli and resulting B. bacteriovorus HD100 populations were determined.
Potassium cyanide impact on attack-phase and intraperiplasmic predator viabilities. The KCN stock solution was diluted into cultures of B. bacteriovorus HD100 to achieve the desired concentration. These cultures were then incubated at 30°C and 250 rpm for 24 h, after which the number of surviving B. bacteriovorus HD100 was determined. For the intraperiplasmic studies, predation of E. coli was performed for 1 h at 30°C and 250 rpm using a PPR of approximately 0.03. At this time, KCN was added to achieve the desired concentration. The number of predators were determined after an additional 3 and 5 h using top agar plates as described previously (52).
Confocal microscopy. For the microscopic images, we used B. bacteriovorus HD100 bearing plasmid pMQ572, which expresses the Venus yellow fluorescent protein (55), and E. coli expressing the Dsred fluorescent protein via plasmid pHTK3 (56). The pMQ572 plasmid for B. bacteriovorus HD100 was constructed by Robert Shanks' group and is similar to another construct that expresses the fluorescent tdTomato protein (31,57). pMQ572 was introduced into B. bacteriovorus HD100 by conjugation as described previously (58) with slight modifications. For these experiments, approximately 2 ϫ 10 10 B. bacteriovorus HD100 bacteria were pelleted by centrifugation (16,000 ϫ g, 15 min) and resuspended in 2 ml HEPES. The donor strain, E. coli S17pir/pMQ572, was grown to stationary phase (16 h) in LB medium with 10 g/ml gentamicin added at 30°C and 250 rpm. The cells were then pelleted by centrifugation (20 ml, 16,000 ϫ g, 15 min) and resuspended in 2 ml HEPES buffer. Both the donor and recipient (50 l) were mixed and spotted onto nitrocellulose filters (0.22-m pore size) (Pall Life Sciences, USA), which were placed on top of DNB agar plates. The plates were then incubated overnight at 30°C. After 24 h, the bacteria were resuspended in 2 ml HEPES buffer and serially diluted. These samples were used to prepare DNB top agar plates with E. coli MG1655 harboring pMQ572 as the prey. These plates had 10 g/ml gentamicin added to them to select for the transconjugants. The plates were incubated at 30°C until plaques were visible. The plaques were then excised, and the recombinant predatory cells were grown in HEPES buffer with gentamicin added using E. coli MG1655/pMQ572 as the prey. It was noticed that when these predatory cell cultures were grown in DNB, they were more fluorescent than cultures in HEPES, and thus, DNB was used to prepare the recombinant cultures for the experiments. Consequently, the predation tests were conducted as described above in DNB except the PPR was approximately 1.0. After 1 h, the culture was divided, and KCN was added to a final concentration of 200 M. Samples were taken at set times and imaged using a Carl Zeiss LSM 780 multiphoton confocal microscope controlled by ZEN 2012 software.
Statistical analyses. All experiments were performed at least three times. Statistical analysis was done using Student's t test to compare two sets of data). For comparing three or more data sets, analysis of variance (ANOVA) was performed followed by the Tukey posthoc test.
Data availability.