Modulation of Medium pH by Caulobacter crescentus Facilitates Recovery from Uranium-Induced Growth Arrest

Abstract

The oxidized form of uranium [U(VI)] predominates in oxic environments and poses a major threat to ecosystems. Due to its ability to mineralize U(VI), the oligotroph Caulobacter crescentus is an attractive candidate for U(VI) bioremediation. However, the physiological basis for U(VI) tolerance is unclear. Here we demonstrated that U(VI) caused a temporary growth arrest in C. crescentus and three other bacterial species, although the duration of growth arrest was significantly shorter for C. crescentus. During the majority of the growth arrest period, cell morphology was unaltered and DNA replication initiation was inhibited. However, during the transition from growth arrest to exponential phase, cells with shorter stalks were observed, suggesting a decoupling between stalk development and the cell cycle. Upon recovery from growth arrest, C. crescentus proliferated with a growth rate comparable to that of a control without U(VI), although a fraction of these cells appeared filamentous with multiple replication start sites. Normal cell morphology was restored by the end of exponential phase. Cells did not accumulate U(VI) resistance mutations during the prolonged growth arrest, but rather, a reduction in U(VI) toxicity occurred concomitantly with an increase in medium pH. Together, these data suggest that C. crescentus recovers from U(VI)-induced growth arrest by reducing U(VI) toxicity through pH modulation. Our finding represents a unique U(VI) detoxification strategy and provides insight into how microbes cope with U(VI) under nongrowing conditions, a metabolic state that is prevalent in natural environments.

INTRODUCTION

Depleted uranium (U) is a widespread environmental contaminant, with major sources coming from energy and weapons production (1). The chemical toxicity of U poses a major threat to organisms and the environment. The extent of the environmental risk is dependent on the biogeochemical conditions that affect U speciation and toxicity (1, 2). In the anoxic subsurface, U is usually present in the reduced form U(IV), which is less toxic and less soluble than the oxidized form U(VI). In contrast, in oxic surface waters, the highly toxic and mobile U(VI) predominates (1). At circumneutral pH, U(VI) oxycations are present and form stable, soluble complexes with carbonate, phosphate, and natural organic matter, which greatly influence the mobility and toxicity of U(VI) (1). The formation of these complexes is governed by pH, redox potential, temperature, and ligand concentration (1, 3, 4). Although the relationship between U(VI) speciation and bioavailability is complex since certain ligands can either promote or inhibit U(VI) immobilization depending on the environmental conditions, it is evident that U(VI) in strong complexes or adsorbed to colloidal or particulate matter has reduced bioavailability and toxicity compared to those of the naked U(VI) oxycation (1, 5). Given the high toxicity and bioavailability of U(VI), the development of bioremediation strategies that reduce its mobility and bioavailability in contaminated sites is of significant interest.

Microorganisms play an important role in governing U speciation and transport in the environment, and thus, understanding the microbial factors that influence U transformation is important in predicting U biogeochemical cycling and remediation strategies (6). Extensive studies on U bioremediation in the past decades have predominantly focused on the reduction of U(VI) to U(IV) by dissimilatory metal-reducing bacteria under anaerobic conditions (7). Although less well studied under aerobic conditions, microbes have been reported to detoxify and immobilize U(VI) through surface adsorption, intracellular accumulation, and precipitation (6). In particular, extracellular polysaccharides (EPS) secreted by many microorganisms contain side groups with negatively charged ligands such as hydroxyl, carboxyl, or phosphate, which attract cations like U(VI) and reduce their mobility (8, 9). Additionally, phosphatase enzymes have been reported to mediate U(VI) precipitation through the release of inorganic phosphate from organophosphates (10–14).

The ubiquitous aerobe Caulobacter crescentus holds promise for U(VI) bioremediation because of its ability to mineralize U(VI) and survive in low-nutrient environments (15, 16). The recent isolation of heavy-metal-resistant Caulobacter species from sites contaminated with heavy metals further highlights the environmental relevance of members of this genus (17, 18). C. crescentus undergoes an asymmetric cell division that yields two daughter cells with differing morphological and regulatory features (19). These are a motile, chemotactic swarmer cell and a stalked cell that contains a thin cylindrical protrusion of the cell envelope (stalk). Stalked cells are able to reinitiate cell division, whereas swarmer cells must differentiate into stalked cells in order to divide. The distinct cell cycle-associated morphological features in C. crescentus allow examination of the effect that U(VI) exerts on cell cycle processes. Despite its high tolerance and environmental relevance, little is known about the physiology and detoxification mechanisms of C. crescentus with respect to U(VI). Recently, we showed that extracellular phosphatase activity mediated by PhoY is critical for survival in media containing organophosphates and U(VI) (14). In this study, we characterized the temporary growth arrest response employed by C. crescentus during exposure to U(VI) and identified a unique U(VI) detoxification strategy.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

Caulobacter crescentus NA1000, Escherichia coli K-12, Pseudomonas aeruginosa PAO1, and Pseudomonas putida KT2440 were used in this study. Growth experiments were performed in (i) PYE medium, containing 0.2% (wt/vol) Bacto peptone (Difco), 0.1% yeast extract (Difco), 1 mM MgSO₄, and 0.5 mM CaCl₂, (ii) phosphate-buffered minimal medium (M2G) (20), or (iii) modified M5G medium (21) supplemented with 5 mM glycerol-2-phosphate in place of inorganic phosphate. Cultures were grown at 30°C with shaking to mid-exponential phase prior to use in experiments. Unless otherwise specified, an inoculum size of 9 × 10⁷ cells/ml was used in all C. crescentus experiments.

Uranium toxicity experiments.

Uranyl nitrate hexahydrate [(UO₂)(NO₃)₂ · 6H₂O) [here designated U(VI)] was obtained from SPI Supplies (West Chester, PA), and 100 or 200 mM stock solutions were prepared in 0.1 N nitric acid and then filtered through a 0.22-μm membrane. The U(VI) concentration of the stock solution was confirmed by inductively coupled plasma mass spectrometry (ICP-MS). Spontaneous U(VI) precipitation was observed upon addition of uranyl nitrate to PYE, with a maximum solubility of ∼75 μM. The U(VI) concentrations described in this article correspond to the initial added U(VI); solubility measurements at each added U(VI) concentration are shown in Fig. S1 in the supplemental material. The pH of each growth medium, including those lacking added uranyl nitrate, was standardized through the addition of an equivalent amount of HNO₃. Where indicated, the supernatant from cell cultures was collected by centrifugation at 20,000 × g for 8 min and then filtered through a 0.22-μm membrane. Heat-killed cells were prepared by incubation at 70°C for 15 min, which was followed by CFU measurements and propidium iodide staining to confirm treatment efficacy and phase-contrast microscopy to confirm cell integrity. The optical density at 600 nm (OD₆₀₀) was determined using a BioTek Synergy H1 microplate reader. CFU were determined by diluting and plating on PYE agar.

Cells were synchronized as described previously (22). Briefly, exponential-phase cells were resuspended in 10 mM phosphate buffer (pH 7.0), added to an equal volume of Percoll (Pharmacia), and centrifuged at 10,000 × g for 20 min at room temperature. The lower swarmer band was removed, washed twice with the phosphate buffer, and resuspended in PYE.

To compare the effect of U(VI) on growth of Escherichia coli K-12, Pseudomonas aeruginosa PAO1, and Pseudomonas putida KT2440 to that on growth of C. crescentus, cells grown to mid-exponential phase (OD₆₀₀ of ∼0.25) in PYE were washed once in PYE to remove the spent medium and inoculated into PYE containing 350 μM U(VI). Since growth recovery was dependent on metabolic activity, the inoculum was normalized by biomass, as determined by cell dry weight (see Table S1 in the supplemental material). The cell dry weight was determined by growing all strains to an OD₆₀₀ of 0.5, centrifuging at 20,000 × g for 8 min, washing twice in 1× phosphate-buffered saline (PBS) (pH 7), and drying overnight at 65°C. There was less than a 3-fold difference in cell number among the four strains after normalization by biomass (data not shown).

Assays of U(VI) concentration.

The U(VI) concentration was determined spectrophotometrically using the Arsenazo III reagent (23). A 1% (wt/vol) Arsenazo III stock solution was prepared in 6.25% (wt/vol) trichloroacetic acid (TCA) and filtered with a 0.22-μm membrane. To measure the soluble U(VI) concentration, samples were centrifuged at 20,000 × g for 5 min at room temperature, 40 μl of supernatant was acidified with an equal volume of 12.5% (wt/vol) TCA before the addition of 60 μl Arsenazo III stock solution, and the absorbance at 652 nm was measured. Total U(VI) concentration was determined as described above, except an equal volume of 12.5% (wt/vol) TCA was added before centrifugation. The U(VI) concentration was calculated using a standard curve.

β-Galactosidase assays.

NJH199 (24), an NA1000 derivative lacking CC_1634 (lacA) and containing a plasmid-borne translational P_urcA-lacZ fusion, was grown in PYE containing U(VI) in the presence of 2 μg/ml chloramphenicol. This reporter fusion included a 1-kb region upstream of the urcA start codon through the first 24 nucleotides of urcA, fused to lacZ (24). To terminate cell growth and any further protein synthesis, tetracycline was added to a final concentration of 1 μg/ml and cells were placed on ice. β-Galactosidase activity was assayed as previously described (25).

Fluorescence microscopy.

The membrane potential probe bis-(1,3-dibutylbarbituric acid) trimethine oxonol [DIBAC₄(3)] (Molecular Probes, Inc.), which enters cells with depolarized membranes, and the dead cell stain propidium iodide ReadyProbes (PI) (Molecular Probes, Inc.) were used to assess cell viability during growth arrest induced by 350 μM U(VI). DIBAC₄(3) was added at 1 μg/ml, and PI was used according to the manufacturer's instructions. After dye addition, samples were incubated for 30 min at 30°C, concentrated by 10-fold, and then immobilized using a thin layer of PYE-agarose. Phase-contrast and fluorescence microscopy images were obtained using a Leica DMI 6000B microscope with an HCX PL APO 63×, 1.40 numerical aperture, oil PH3 CS objective and a Leica DFC360 FX camera. Fluorescein isothiocyanate (FITC) and rhodamine filters were used for DIBAC₄(3) and PI, respectively.

MT190 (26), an NA1000 derivative with the enhanced cyan fluorescent protein (eCFP) gene (ecfp) fused to the endogenous parB gene, was synchronized as described above and inoculated into PYE containing 350 μM U(VI). At different time points, cells were fixed in 4% formaldehyde in 1× PBS (pH 7) for 30 min at room temperature, and the U(VI) precipitates were solubilized with 1 M NaHCO₃ (pH 8.5) (27). Cells were washed with 1× PBS, spotted onto a 1% agarose pad, and imaged using an Axiovert 200M microscope (Zeiss, Minneapolis, MN) equipped with a Photometric CoolSNAP HQ charge-coupled device (CCD) camera. Images were acquired with a 100× differential interference contrast (DIC) objective and an eCFP filter set (Chroma, Bellows Falls, VT). Images were processed using ImageJ (28), and fluorescent foci were counted manually.

SEM.

Samples were fixed and U(VI) precipitates were solubilized with bicarbonate as described above. Fixed cells were washed four times with double-distilled water (ddH₂O) and allowed to dry overnight onto a silicon wafer. Scanning electron microscopy (SEM) images were taken with a JEOL JSM 7401-F field emission SEM (FESEM) instrument.

RESULTS

U(VI) induces a reversible growth arrest in C. crescentus.

The effect of U(VI) on the growth of Caulobacter crescentus NA1000 was determined in PYE with exponential-phase cells as the inoculum. U(VI) concentrations are reported as total added U(VI). U(VI) caused a temporary growth arrest that delayed the onset of exponential growth in a U(VI) concentration-dependent manner (Fig. 1A). No growth recovery with U(VI) concentrations at or above 500 μM was observed after 2 weeks (data not shown). At a fixed U(VI) concentration (300 μM), the length of growth arrest was also dependent on the initial cell density, with lower inoculum sizes extending the growth arrest duration (Fig. 1B). At below ∼7 × 10⁵ cells/ml, no growth recovery was observed, suggesting that there is a minimal cell density required for growth recovery (data not shown). Growth arrest was observed regardless of the growth phase of the inoculum (see Fig. S2A in the supplemental material), suggesting that U(VI) exposure arrested growth rather than extending a normal lag phase. Furthermore, direct addition of U(VI) to mid-exponential-phase cultures induced growth arrest but required higher U(VI) concentrations than when U(VI) was introduced during inoculation (see Fig. S2B in the supplemental material). After recovery from growth arrest, C. crescentus achieved the same growth rate and approximate final cell density as the control without U(VI) (Fig. 1). Thus, it appears that C. crescentus has accommodated the U(VI) stress and/or the U(VI) toxicity was greatly reduced during growth arrest.

FIG 1.

U(VI)-induced growth arrest in C. crescentus. (A) Cell growth with U(VI) added at the beginning of incubation. Error bars represent the standard deviation for three biological replicates. (B) Effect of inoculum size on growth arrest duration with 300 μM U(VI). Data represent the average for two biological replicates that varied by less than 20%. (C) CFU measurements during growth in the presence of U(VI). Error bars represent the standard deviation for three biological replicates.

CFU measurements during the course of growth arrest indicated that cell viability is dependent on U(VI) concentration. At 350 μM or lower, there was no change in cell viability throughout growth arrest (Fig. 1C). Dead cell staining with either propidium iodide or bis-(1,3-dibutylbarbituric acid) trimethine oxonol [DIBAC4(3)] indicated that greater than 99% of cells remained alive, suggesting that the lack of a change in CFU is a result of growth arrest rather than equivalent rates of cell growth and death (data not shown). Furthermore, the lack of DIBAC4(3) staining suggests that the membrane remained polarized (29), indicative of a metabolically active state. However, a decrease in viability is evident with U(VI) at or above 400 μM (Fig. 1C). The CFU count dropped approximately 25-fold at 400 μM U(VI) before growth recovery, while no detectable viability (<10 CFU/ml) was observed at 500 μM U(VI) after 24 h (data not shown). This reduction in cell viability at higher U(VI) concentrations provides a likely explanation for why a small increase in U(VI) concentration disproportionally lengthens the duration of growth arrest, as shown in Fig. 1A. Since 350 μM U(VI) elicits growth arrest without affecting viability, it was used to induce growth arrest in the experiments described below, unless specified otherwise.

Growth arrest is not specific to C. crescentus.

To determine whether the temporary growth arrest response is specific to C. crescentus, the effect of U(VI) on the growth of Escherichia coli K-12, Pseudomonas aeruginosa PAO1, and Pseudomonas putida KT2440 was tested. Exposure to U(VI) (350 μM) caused growth arrest in all species tested, although the duration of growth arrest was different among species (Fig. 2). P. aeruginosa and P. putida entered exponential phase after ∼35 h and ∼75 h of growth arrest, respectively, which was significantly longer than for C. crescentus (∼8 h). E. coli was the most sensitive to U(VI), as no growth recovery was evident after 2 weeks; growth recovery was observed with a U(VI) concentration of 150 μM or lower (data not shown). These data suggest that C. crescentus is more resistant to U(VI) than the other three species tested, consistent with previous observations (16). The temporary growth arrest observed for all four species tested suggests that stress-induced growth arrest may be a common mode of U(VI) toxicity among aerobic bacteria.

FIG 2.

Comparison of growth arrest duration for Caulobacter crescentus, Escherichia coli K-12, Pseudomonas aeruginosa PAO1, and Pseudomonas putida KT2440. Growth curves are with no U(VI) (A) and with 350 μM U(VI) (B). Data represent the average for two biological replicates that varied by less than 30%.

U(VI) inhibits progression of swarmer cells to stalked cells.

To determine whether the morphology of C. crescentus was altered during growth arrest, we visualized cells using scanning electron microscopy (SEM). Compared to exponential-phase cells prior to U(VI) addition, no gross morphological changes or accumulation of any particular cell type was observed during the first 4 h of growth arrest (Fig. 3A). Cells with shorter stalks were observed toward the end of growth arrest (7 h), suggesting that U(VI) may affect stalk development. To further test this notion, isolated swarmer cells were exposed to U(VI) and visualized by SEM (Fig. 3B). No detectable morphological changes were observed within the first 5 h of growth arrest, indicating that stalk development was inhibited. During the transition to exponential phase, cells that had reached the predivisional stage either appeared to completely lack stalks or possessed abnormally short stalks, consistent with the results from unsynchronized cells. It should be noted that swarmer cells exhibited a slightly longer growth arrest (by ∼2 h) than a mixed population (data not shown), which may reflect the additional time required for the initial transition from swarmer to stalked cell and recovery from the added stress of the synchronization procedure.

FIG 3.

Adaptive changes of C. crescentus following exposure to 350 μM U(VI). (A) Representative SEM images of cells collected at different time points during growth in the presence of U(VI). The time following U(VI) addition and the growth stage of cells are indicated above each image. Scale bars, 1 μm. (B) Representative SEM images after exposure of synchronized swarmer cells to U(VI). The time following U(VI) addition and the growth stage of cells are indicated above each image. Scale bars, 1 μm. (C) Replication initiation following U(VI)-induced growth arrest was visualized in a synchronized population containing a chromosomal ecfp-parB fusion. The percentage of the population with two CFP-ParB foci at each time point is indicated. Top panels, DIC; bottom panels, CFP-ParB fluorescence. (D) Control depicting ParB-CFP foci after 1 h of growth in medium lacking U(VI). (E) ParB-CFP foci within representative filamentous cells shortly after growth recovery (OD₆₀₀ of 0.15).

To determine whether chromosome replication initiation, which occurs concomitantly with the transition from swarmer to stalked cell, was also inhibited by U(VI), synchronized cells containing a ParB-CFP fusion (26) were monitored throughout growth arrest (Fig. 3C). ParB binds to parS sites in close proximity to the origin of replication and is involved in chromosome partitioning. After replication initiation, the new origin is moved to the opposite pole, so oriC can be quantified by examining ParB-CFP foci within a cell (30). Within the first 8 h of growth arrest, there was no change in the number of cells with two ParB-CFP foci, suggesting that U(VI) inhibited either chromosome replication initiation or segregation (Fig. 3C). In contrast, 67% of cells inoculated in medium lacking U(VI) had two fluorescent foci after 1 h (Fig. 3D). Taken together, these data suggest that U(VI) blocks the G₁-to-S transition of the C. crescentus cell cycle.

Interestingly, shortly after cells had transitioned into exponential phase (OD₆₀₀ of 0.2), approximately 15% of the population appeared filamentous (Fig. 3A and B). The filamentous cells possessed multiple ParB-CFP foci, indicative of DNA replication without subsequent cell division (Fig. 3E). Although bacterial filamentation can arise from several different stresses (31), it is commonly associated with DNA damage and likely indicates cell division inhibition. Normal cell morphology was restored by the end of exponential phase (Fig. 3B).

Preexposure to U(VI) does not increase the ability to cope with subsequent U(VI) exposure.

As a first step to deduce the mechanism that leads to growth recovery, we tested whether growth recovery of C. crescentus is a result of a beneficial mutation that confers U(VI) resistance. Cells grown on U(VI)-containing agar plates did not show increased viability compared to naive cells when regrown on U(VI)-containing plates (Fig. 4A). Additionally, exponential-phase cells that had recovered from U(VI)-induced growth arrest were not more resistant to U(VI) than cells without previous U(VI) exposure; both cultures experienced the same length of growth arrest when reinoculated into U(VI)-containing medium (Fig. 4B). Thus, preexposure to U(VI) did not shorten the duration of growth arrest. These data suggest that growth recovery is not caused by acquired mutations that confer U(VI) resistance. Furthermore, the retained U(VI) sensitivity in recovered cells suggests that if recovery from growth arrest is a result of altered gene/protein expression that confers U(VI) resistance, such changes are not preserved after growth recovery. This failure of preexposure to elicit enhanced U(VI) resistance is in contrast to the case with other metals (e.g., Cd) and toxic chemicals (e.g., H₂O₂), where inducible detoxification pathways are responsible for resistance (32, 33). Recovery from growth arrest may, therefore, be a result of U(VI) detoxification rather than accommodation.

FIG 4.

Sensitivity of C. crescentus to repeated U(VI) exposure. (A) Cells grown on PYE agar containing either 0, 500, or 700 μM U(VI) (labeled at top of image) were respotted on PYE agar containing 0, 500, or 700 μM U(VI) (labeled at bottom of image). The spotting represents 10-fold serial dilutions (10⁰ to 10⁶), ordered from top to bottom. Results are representative of three replicates. (B) Cells grown to mid-exponential phase in PYE containing 0 (open symbols) or 350 μM (filled symbols) U(VI) were reinoculated into fresh PYE containing 350 μM U(VI), and growth was monitored.

C. crescentus reduces U(VI) toxicity during growth arrest.

To evaluate the possibility that C. crescentus recovers from growth arrest by reducing U(VI) toxicity, the supernatant from a C. crescentus culture at the end of U(VI)-induced growth arrest was collected and used as growth medium. Upon reinoculation, the supernatant did not elicit growth arrest but instead yielded growth kinetics identical to those of the control without U(VI) (Fig. 5A). In contrast, growth arrest was evident with supernatant collected from heat-killed cells or an abiotic control incubated with U(VI) for the same period of time, indicating that U(VI) detoxification is a physiological process. The heat-killed cells served to control for any process that may reduce U(VI) toxicity that is metabolism independent (e.g., cell surface sorption). To further characterize this reduction in U(VI) toxicity, an aliquot of supernatant was extracted every ∼75 min during growth arrest and reinoculated with fresh C. crescentus cells. The time required for these cultures to reach mid-exponential phase (OD₆₀₀ of 0.2), termed the growth time, was measured. The growth time was inversely proportional to the amount of time that the supernatant had been exposed to C. crescentus during growth arrest (Fig. 5B). In other words, supernatant extracted at later time points of growth arrest was less toxic to cells. In contrast, no reduction in growth time was observed for the supernatant collected from an abiotic control. Thus, C. crescentus reduced the toxicity of U(VI) throughout growth arrest.

FIG 5.

Cell-mediated reduction in U(VI) toxicity by C. crescentus during growth arrest. (A) Supernatant from live cells (5.4 × 10⁷ cells/ml), heat-killed cells (5.4 × 10⁷ cells/ml), or an abiotic control was collected prior to the end of growth arrest and used as growth medium for reinoculation. Growth in fresh medium with or without 350 μM U(VI) (open symbols) is shown for comparison. Data represent the average for two biological replicates that varied by less than 30%. (B) Supernatant from growth-arrested cultures (bottom panel) was collected every ∼75 min (roman numerals) and used as growth medium for reinoculation. The growth time (top panel) represents the length of time for the reinoculated culture to reach an OD₆₀₀ of 0.2. The dashed line represents the growth time for cells inoculated in a medium lacking U(VI) (i.e., no growth inhibition). The growth time in supernatant collected from an abiotic control is shown for comparison. Error bars represent the standard deviation for three biological replicates. (C) Time course of P_urcA-lacZ expression (top panel) at different U(VI) concentrations. The growth curve at each U(VI) concentration is displayed in the bottom panel for reference.

To further examine the reduction in U(VI) toxicity during growth arrest, the expression of P_urcA, a promoter that was previously shown to specifically sense U(VI) (24), was monitored using a plasmid-borne translational fusion of the first 8 amino acids of UrcA to LacZ under the control of the urcA promoter. Strikingly, the dynamics of P_urcA-lacZ expression correlated precisely with the duration of growth arrest (Fig. 5C). The expression of the P_urcA-lacZ reporter increased sharply during the initial period of growth arrest and reached a maximum toward the end of growth arrest. The reporter activity declined immediately upon growth recovery and returned to basal levels during exponential phase, likely due to cessation of further P_urcA-lacZ expression and protein turnover and dilution during cell division. This result suggests that the U(VI)-induced stress sensed by P_urcA has been eliminated during growth arrest, consistent with the observed reduction in U(VI) toxicity.

U(VI) toxicity is reduced by modulation of medium pH through metabolic activity.

Since pH is known to control U(VI) speciation and hence U(VI) toxicity (1), we monitored the medium pH. With 350 μM U(VI), C. crescentus increased the medium pH by approximately 0.4 unit during growth arrest (from 5.7 to 6.1) (Fig. 6A). The medium pH continued to increase throughout growth, reaching ∼8.1 by early stationary phase. This increase was dependent on metabolic activity, since no pH change was observed over the same time period when sodium azide, an inhibitor of cytochrome oxidase, was present (data not shown). It should be noted that in the absence of U(VI), cell growth was not severely altered within this pH range (see Fig. S3A in the supplemental material) and medium alkalinization occurred throughout C. crescentus growth (data not shown). Medium alkalinization has frequently been noted during E. coli growth in LB and is attributed to ammonia excretion during amino acid catabolism (34). To test whether the observed pH increase was responsible for alleviating U(VI) toxicity, the pH of the U(VI)-containing growth medium (pH 5.7) was artificially increased to 6.1, the measured value at the end of growth arrest, prior to inoculation. Growth arrest was significantly shortened and P_urcA expression was greatly reduced compared to a control with no pH adjustment (Fig. 6B and C). The metabolism-dependent modulation of medium pH and hence U(VI) toxicity provides a likely explanation for the observed effect of inoculum size on growth arrest duration; one would expect a larger inoculum to increase pH at a higher rate.

FIG 6.

Reduction of U(VI) toxicity through modulation of growth medium pH. (A) Measurement of medium pH during growth of C. crescentus with 350 μM U(VI). Data represent the average for two biological replicates that varied by less than 10%. (B) Effect of pH adjustment of the medium on growth arrest duration. The pH of PYE containing 350 μM U(VI) (pH 5.7) was adjusted to 6.1 (open symbols) prior to inoculation. Data represent the average for two biological replicates that varied by less than 20%. (C) Effect of pH adjustment on P_urcA-lacZ expression during growth arrest induced by 350 μM U(VI). Symbols are the same as for panel B. (D) Measurement of medium pH during growth of P. aeruginosa, P. putida, and E. coli with U(VI). The concentrations of U(VI) used are 350 μM, 175 μM, and 90 μM for P. aeruginosa, P. putida, and E. coli, respectively. pH is plotted in the top panel, while OD₆₀₀ is plotted in the bottom panel. Data represent the average for two biological replicates that varied by less than 10%. (E) Soluble (closed symbols) and total (open symbols) concentrations of U(VI) measured during growth of C. crescentus with 350 μM U(VI). The dashed line marks the time when growth recovery occurred.

To test whether pH modulation was also responsible for growth recovery in other bacterial species, we monitored the medium pH during growth in U(VI) for E. coli, P. aeruginosa, and P. putida (Fig. 6D). We note that lower U(VI) concentrations were chosen for E. coli and P. putida, because of their high sensitivity toward U(VI). All three strains increased the medium pH during growth arrest. Compared to that for C. crescentus, a higher pH (∼6.5 compared to 6.1) was reached for these species before growth recovery was observed, consistent with the higher U(VI) tolerance observed for C. crescentus. Medium alkalinization was not observed for E. coli at 350 μM U(VI), consistent with the lack of growth recovery (data not shown). Importantly, pH adjustment also shortened the growth arrest duration of these three species (data not shown), suggesting that pH modulation may be a general U(VI) detoxification mechanism.

To determine whether the pH increase affected U(VI) solubility, we monitored the soluble U(VI) concentration during the course of growth arrest for C. crescentus (Fig. 6E). It should be noted that spontaneous U(VI) precipitation was evident upon addition of U(VI) to the PYE growth medium, independent of the presence of cells. These U(VI) precipitates were not responsible for the observed U(VI) toxicity, since growth arrest was still observed upon their removal by centrifugation (Fig. 5A). The solubility of U(VI) increased slightly during growth arrest at a rate that was indistinguishable from that for an abiotic control (Fig. 6E). Artificially increasing the growth medium pH from 5.7 (pH at the beginning of growth arrest) to 6.1 (pH toward the end of growth arrest) did not significantly affect U(VI) solubility (see Fig. S3B in the supplemental material). Together, our data point toward a pH-mediated detoxification mechanism that does not involve a change in U(VI) solubility.

U induces a similar reversible growth arrest in minimal media.

Attempts to characterize the effect of U(VI) on C. crescentus growth in minimal media were made, taking into account U(VI) solubility and phosphate limitation issues. Although M2G (20) is commonly employed as the defined minimal medium for C. crescentus, the limited solubility of U(VI) (∼10 μM) in this phosphate-buffered medium precluded phenotypic determination of U(VI) toxicity. Chelation with a 10:1 molar ratio of citrate or bicarbonate to U(VI) did not increase U(VI) solubility in M2G (data not shown). Furthermore, phosphate limitation resulting from abiotic U-P_i precipitation hindered the use of low-phosphate minimal media (e.g., M5G [21]). Nevertheless, we found that the increased range of U(VI) solubility in modified M5G (M5G-G2P), where inorganic phosphate was replaced by glycerol-2-phosphate, allowed growth to be characterized. Similar to the case with PYE, U(VI) caused a reversible growth arrest in M5G-G2P in a concentration-dependent manner (see Fig. S4A in the supplemental material). In contrast to the case with PYE, the soluble U(VI) concentration decreased during growth arrest (see Fig. S4B in the supplemental material), due to precipitation with P_i released from G2P through cellular phosphatase activity (14). Since the soluble P_i concentration remained low throughout growth arrest (see Fig. S4C in the supplemental material), a contribution of P_i limitation to the growth arrest cannot be excluded. Further work is under way to explicate the toxic effect of U(VI) in minimal media and to further characterize the mechanism of U(VI)-induced cell cycle arrest.

DISCUSSION

Here, we demonstrated that in the presence of U(VI), C. crescentus entered a reversible, nonreplicative state that was maintained until U(VI) was sufficiently detoxified. Metabolic activity during growth arrest increased the growth medium pH, detoxified U(VI), and facilitated growth recovery. A loss of viability was observed during prolonged growth arrest periods with high U(VI) concentrations (>350 μM), highlighting the importance of U(VI) detoxification for cell survival. Growth recovery was not observed below a certain threshold inoculum size, indicating that there is a minimal quorum of C. crescentus required to overcome U(VI) toxicity. Together these data point toward a community-driven U(VI) detoxification process.

Growth recovery was marked by proliferation with normal growth kinetics but sustained susceptibility to U(VI), arguing against accumulation of U(VI)-resistant mutants. Antimicrobial compounds such as heavy metals and antibiotics are common causes of growth arrest in microorganisms (32, 35–38). In many cases, recovery from growth arrest is attributed to the induction of a genetically encoded resistance mechanism or the selection of antimicrobial-tolerant mutants during antimicrobial exposure (32, 36). However, recent studies of microbial antibiotic resistance suggest that resistance can stem from a reduced growth rate, which significantly diminishes the potency of antibiotics (39). Our results indicate that C. crescentus is inherently susceptible to and phenotypically tolerant of U(VI) in a similar manner. In accordance with our data, substantial degradation of cellular RNA was observed when Metallosphaera sedula was exposed to U(VI), suggesting that critical cellular processes were aborted as a dynamic mechanism for resisting U(VI) toxicity (40). Thus, the temporary growth arrest observed for C. crescentus may represent a survival strategy against U(VI) toxicity.

The cellular and morphological changes observed during U(VI) exposure indicate a disruption of the cell cycle. C. crescentus exposed to high levels of U(VI) arrested cell cycle progression; the stalked-cell transition and the initiation of DNA replication were blocked in swarmer cells, while cell division was inhibited in stalked and predivisional cells. Recent proteomic analysis indicated that the levels of two master cell cycle regulators, CtrA and DnaA, were significantly reduced during U(VI) exposure (41). During the C. crescentus cell cycle, the abundance of these regulators is tightly controlled in order to precisely regulate the timing of DNA replication. DnaA is required for replication initiation (42), and CtrA inhibits replication in swarmer cells (43) but also acts as a transcription factor for ∼100 genes that are critical for cell cycle progression (44). Downregulation of DnaA and CtrA has been observed under conditions that cause cell cycle arrest (e.g., carbon starvation) (42). Thus, it is conceivable that U(VI)-induced cell cycle arrest is at least in part a consequence of CtrA and DnaA depletion, although the precise nature of the U(VI)-induced cellular stress and how it leads to cell cycle disruption remain to be determined.

The finding that growth arrest is alleviated through alkalinization of the growth medium highlights the importance of pH in U(VI) toxicity. Given that U(VI) speciation is strongly influenced by pH in aquatic systems (1) and the absence of a pH-mediated change in U(VI) solubility during growth arrest, we speculate that U(VI) detoxification during growth arrest was a result of altered U(VI) speciation. It is generally accepted that the uncomplexed uranyl ion is the most toxic and bioavailable form of U(VI) while chelation with ligands such as phosphate, carbonate, and organic acids reduces U(VI) toxicity (1). Previous studies suggest that an increase in pH within a range of 5 to 6 decreases U(VI) bioavailability by lowering the uranyl concentration in aquatic environments (45, 46). Consistently, U(VI) complexation by organic or inorganic phosphate was shown to reduce U(VI) toxicity in C. crescentus (14), while lower-pH conditions that favor the presence of the uncomplexed uranyl ion appear to be more toxic to bacteria (47). Further analytical techniques will be required to substantiate a speciation dependency of U(VI) toxicity in our system.

In conclusion, we have demonstrated that U(VI) causes a reversible growth arrest in C. crescentus, with growth recovery mediated through a metabolism-dependent increase in medium pH. This detoxification mechanism is not specific to C. crescentus, suggesting that it may be a general mechanism of U(VI) defense by aerobic bacteria. Together, our results provide insight into how microbes cope with (metal) stress under nongrowing conditions that are prevalent in the natural environment and highlight the need to consider the nongrowing state, not simply active growing cells, when predicting microbial behaviors or planning a bioremediation system.