Phase-Variable Control of Multiple Phenotypes in Acinetobacter baumannii Strain AB5075

Strains and growth conditions.

A. baumannii strain AB5075 (26) and an isogenic abaI::Tc mutant were obtained from Colin Manoil at the University of Washington. Strains ATCC 17978, AB1712, and AB0057 (27, 28) were obtained from Robert Bonomo at Case Western Reserve University. Strain AB307 was obtained from Anthony Campagnari at the University of Buffalo. Fresh clinical isolates of A. baumannii were obtained from Sarah Satola at the Emory University School of Medicine. These isolates included AB732, AB248, AB45, AB991, AB997, AB1009, AB1011, AB1013, AB1020, AB499, AB552, AB1040, AB1051, AB1056, and AB1062. Growth conditions consisted of modified LB containing 10 g of tryptone, 5 g of yeast extract, and 5 g of NaCl per liter. Agar was added to 1.5% or 0.8% as indicated below. Motility assays were conducted on LB plates containing 0.3% Eiken agar. For all experiments involving opaque and translucent variants, a freezer stock of each was used that contained less than 0.5% of the other colony type.

For the growth of strains in this study, the optical densities represent the non-dilution-adjusted values when readings were above 1.0. The dilution-adjusted densities for cells at an A₆₀₀ of 1.1 corresponded to 1.8 when diluted. For an optical density of 1.3, cells corresponded to an optical density of 2.5 when diluted and for cells at an optical density of 1.7, cells corresponded to an optical density of 4.0 when diluted.

Conditioned medium was prepared from A. baumannii or Escherichia coli by growing cells in 25 ml of LB to an A₆₀₀ of 1.7. Cells were then pelleted by centrifugation, and the resulting supernatant was adjusted to pH 6.5 with HCl and filter sterilized using a 0.22-μm filter unit attached to a syringe. The cell-free spent medium was immediately frozen at −80°C and used within 2 weeks. To add back nutrients, a 10× concentrate of tryptone and yeast extract was added back to a final concentration of 0.25× relative to LB.

For experiments examining the effects of pH, LB was adjusted to pH 8.5 with sodium hydroxide. To prepare 0.25× LB, standard LB was diluted with sterile water. For the above-mentioned conditions and the effects of various stress conditions, opaque and translucent freezer stocks were inoculated into LB and grown stationary overnight at room temperature to begin each experiment. LB cultures (2 ml) with various stressors were inoculated from the above-mentioned starter cultures and incubated at 37°C with vigorous shaking. Cultures were grown to the desired optical density, serially diluted in LB, and plated on 0.5× LB–0.8% agar. Plates were incubated at 37°C overnight, and CFU per milliliter and number of variant colonies were determined.

Microscopy.

Cells were grown in LB and visualized by phase-contrast microscopy (1,000×) at log phase (optical density at 600 nm [OD₆₀₀] = 0.2) and early stationary phase (OD₆₀₀ = 1.1). For samples at log phase, cells were concentrated 10-fold by centrifugation. Images were taken using a Lumenera camera and photographed using the Infinity Analyze software program.

Biofilm analysis.

For biofilm studies, strains to be tested were grown to an A₆₀₀ of 0.3 in LB and 150 μl was used to inoculate 8 wells of a polystyrene 96-well microtiter plate. Plates were incubated at 37°C for 24 h, and the A₆₀₀ for each well was determined to measure growth. The medium was removed and replaced with 200 μl of 10% crystal violet solution, and cells were stained for 20 min. Each well was then rinsed 4 times with water and allowed to dry. The crystal violet was solubilized with 300 μl of 33% acetic acid and then diluted with 600 μl of 33% acetic acid. The absorbance of each sample was read at 585 nm. Experiments were repeated in duplicate, with similar results. Biofilm assays in glass tubes were conducted by growing the cells in 2 ml of LB at 37°C for 24 h. At this time medium was removed and replaced with a solution of 10% crystal violet, and staining was carried out for 20 min. Tubes were washed 4 times with 4 ml of sterile water and dried.

Surface motility assays.

Cultures of opaque and translucent variants were grown to stationary phase at 37°C and adjusted to an A₆₀₀ of 0.7. A 1-μl drop of each colony type was spotted in duplicate on the same 0.3% Eiken agar plate (150-mm plates) and grown for 8 h at 37°C. Motility diameters were measured, and the values of 4 independent measurements for each strain were determined. The differences in motility between opaque and translucent variants had a P value of 0.01 as determined using the Student t test.

Antibiotic susceptibility assays.

Strains to be tested were grown to an OD₆₀₀ of 0.3 and then diluted to an OD₆₀₀ of 0.05 with sterile LB. The MICs for various antibiotics were determined using Etest strips on LB agar plates according to the manufacturer's instructions (bioMérieux). The MICs for each antibiotic were determined after 12 h of growth at 37°C. All susceptibility tests were done in duplicate.

Galleria mellonella killing assays.

G. mellonella waxworms were purchased from Wild Bird Goodies Inc., Cumming, GA. A 2-ml culture was inoculated with an opaque or translucent colony and grown shaking at 37°C to an A₆₀₀ of 0.6. Serial dilutions were prepared in cold LB, and 3 ×10⁴ cells of each colony type were injected into the hemolymph of G. mellonella (200 to 250 mg each). Worms were incubated in a petri plate at 37°C in a humidified incubator. At 24-h intervals, viability was assessed by checking for movement, and worms not exhibiting any movement after prodding with a pipette tip were considered dead and were typically dark brown to black. For all experiments, the serial dilutions used for the inoculations were also plated on 0.8% agar plates to determine overall colony counts of the opaque and translucent cultures and the percent reversion to the other colony type. Only experiments derived with cultures exhibiting less than a 0.5% reversion rate were reported. The reported values represent the averages of 4 independent experiments with at least 10 worms for each strain for each experiment.

Phase-variable control of colony opacity in A. baumannii AB5075.

The highly virulent and multidrug-resistant A. baumannii strain AB5075 (26) was observed to form two distinct colonies on agar plates, termed translucent or opaque (Fig. 1A). There were three requirements to accurately view these colony variants: (i) colonies had to be viewed under a dissecting microscope with oblique indirect illumination, (ii) colonies needed to be at a density greater than approximately 200 per plate, and (iii) 0.5× LB plates with an agar concentration of 0.8% enhanced the colony differences, although they could still be clearly seen on normal (1×) LB–1.5% agar plates if the first two requirements were met. The importance of the indirect illumination is highlighted by the inability to detect colony differences on 0.8% agar plates when visualized under normal room lighting (Fig. 1B). All subsequent experiments in this study were done using 0.5× LB–0.8% agar plates unless indicated otherwise.

When a single pure opaque colony was restreaked, translucent colonies could occasionally be seen. In a similar manner, translucent colonies occasionally gave rise to opaque colonies. Opaque colonies often exhibited obvious translucent sectors or a mottled appearance due to translucent variants arising at high frequency during colony growth (Fig. 1C). However, in the translucent variants, rare opaque sectors were seen, but translucent colonies that exhibited a mottled appearance like the opaque form were never observed. The inability of translucent colonies to produce mottled colonies was not due to the inability to give rise to opaque variants (see below). A single lineage of opaque and translucent variants could be selected by taking a pure opaque colony (designated O₁), selecting a translucent variant (T₂) from O₁, and then selecting an opaque variant (O₃) from T₂ and then a translucent variant (T₄) from O₃ so strains O₁→T₂→O₃→T₄ were obtained, which were all derived from the same opaque colony.

The above data suggested that the colony opacity phenotype was unstable. To examine this further, the reversion rates of 10 individual opaque or translucent colony variants grown for 16 h were determined by plating dilutions of a resuspended pure colony and examining the resulting colony opacity phenotypes under a dissecting microscope. The rate of reversion from opaque to translucent was highly variable and ranged from 0.09% to 0.85%, with an average reversion rate of 0.29% ± 0.23%, or 1/345 cells. The rate of reversion from translucent to opaque was also highly variable, ranging from 0.05 to 0.45%, with an overall rate of reversion of 0.21% ± 0.16%, or 1/455 cells. However, the differences in reversion rates between opaque and translucent colonies were not statistically significant (P = 0.28).

Phase variation rates increase at high cell density.

The ability of opaque colonies to develop the sectored and/or mottled appearance resulting from translucent variants arising at high frequency during growth required at least 24 h of colony development (Fig. 1C). This suggested that phase variation increased in a density-dependent manner. To examine this possibility, the frequency of phase variation for both opaque and translucent colonies was examined in broth culture at various cell densities (Table 1). The frequency of phase variation in cultures at late log phase (A₆₀₀ = 0.9) was essentially undetectable (<1/10⁵), However, at an optical density of 1.7, the frequency of phase variation from opaque to translucent cells increased 39-fold above that at an optical density of 0.9 and the frequency of translucent to opaque cells increased 215-fold (Table 1).

The dependence on high cell density for colony opacity switching could be due to nutrient depletion or increased pH from ammonia generated by the deamination of amino acids present in LB. Therefore, to determine the effects of pH and nutrient depletion, the phase variation frequency was examined at an optical density of 1.1 to determine if these treatments prematurely stimulated phase variation. This optical density was chosen because impact of cell density on stimulating phase variation was found to begin at an optical density of 1.3; therefore, the ability of pH and nutrient depletion to prematurely stimulate phase variation could be determined. When cells were grown in LB at 0.25× (nutrient limited) concentrations, the phase variation frequency was lower than in 1× LB (Table 1). In a similar manner, when cells were grown in broth at pH 8.5, there were no significant changes in the phase variation frequencies relative to pH 7.0 (Table 1).

Phase variation is regulated by the accumulation of an extracellular signal.

The above-described data indicated that the density-dependent increase in phase variation was not due to nutrient depletion or increased pH. In bacteria, cell-to-cell communication is mediated by the secretion of extracellular signals in a process termed quorum sensing (29, 30). This process can regulate activities at high cell density and could be responsible for the increase in phase variation. To assess the role of an extracellular signal, spent culture supernatants were prepared from a culture of an opaque variant at an A₆₀₀ of 1.6 and the phase variation frequencies of opaque and translucent variants grown to an optical density of 1.1 in this conditioned medium or in LB only were compared. Before use, the pH of conditioned medium was adjusted to 6.5 and nutrients were added back via the addition of a 10× tryptone-yeast extract concentrate to 0.25× the original concentration. The rate of phase variation for both opaque and translucent cells was increased 30- to 35-fold in cells grown in conditioned medium relative to that seen in LB alone (Table 1). This stimulatory activity in conditioned medium was stable to heating at 100°C for 20 min (data not shown). When conditioned medium was prepared from E. coli MG1655 at the same optical density as used for A. baumannii, the rate of phase variation was not significantly increased over the levels seen during growth in LB alone (Table 1). This indicated not that the increase in phase variation with A. baumannii conditioned medium was due to a nonspecific conditioning effect of bacterial growth but that an A. baumannii signal was required. Consistent with this, conditioned medium prepared from a second A. baumannii strain, ATCC 17978, was also capable of activating phase variation in both directions (data not shown).

To determine if the secreted signal was dependent on the abaI autoinducer synthase (31), the phase variation frequency of an abaI mutant was examined. However, there were no significant differences in the phase variation frequency in either direction between opaque and translucent variants in the abaI mutant compared to the wild type (data not shown). In addition, a quorum sensing signal produced in an AbaI-dependent manner is 3-OH C₁₂-homoserine lactone (3-OH C₁₂-HSL) (31, 32). The ability of synthetic 10 μM 3-OH C₁₂-HSL to stimulate phase variation was examined, but no significant differences were observed compared to growth in LB without 3-OH C₁₂-HSL (Table 1).

Effects of stress conditions on phase variation.

To determine if stress influenced the frequency of phase variation, cells were tested in LB under stress conditions similar to those reported by McQueary et al. (33). These conditions include high salt (300 mM NaCl), no salt, H₂O₂ (50 μM), low iron (100 μM 2,2′-dipyridyl), high iron (100 μM FeCl₃), EDTA (100 μM), and ethanol (1%). The frequency of phase variation was decreased in the absence of salt, with the conversion of opaque to translucent decreased 1.9-fold and the conversion of translucent to opaque decreased 3.6-fold compared to values with LB (Table 1). The remaining stress conditions did not significantly alter the switching frequency from opaque to translucent, or from translucent to opaque, relative to that of cells grown in LB only (data not shown). In addition, the opaque and translucent colony morphologies were not altered in the presence of the above-listed stress treatments (data not shown).

Phenotypic differences between opaque and translucent variants. (i) Cell shape.

Cells taken directly from opaque or translucent colonies had essentially indistinguishable morphology when examined under a phase-contrast microscope (data not shown). However, when grown in liquid medium, cells from the opaque and translucent colony variants exhibited significant morphological differences, which became more prominent as cells increased in density and approached stationary phase. In cells at log phase (OD₆₀₀ = 0.2), both colony variants exhibited the coccobacillus morphology typical of A. baumannii, although cells from the translucent variants were slightly larger (Fig. 2). However, at early stationary phase (OD₆₀₀ = 1.1), cells from the translucent variant were highly elongated compared to cells from the opaque form, with numerous long rods that were up to 5-fold longer than the opaque cells and occasionally had a bloated appearance (Fig. 2).

(ii) Surface motility.

The surface motility of opaque and translucent colony variants was examined on 0.3% Eiken agar plates. The opaque form exhibited a significantly (P = 0.01) increased rate of surface motility (44 ± 3.1 mm) relative to that of the translucent variant (30 ± 4 mm) (Fig. 3A).

(iii) Biofilm formation.

The ability of opaque and translucent variants to form biofilms was examined on two different abiotic surfaces, polystyrene and glass. After 24 h of growth, cells from the translucent variant were better able to form biofilms on both polystyrene and glass (Fig. 3B). In addition, prolonged incubation of opaque and translucent variants under static conditions in LB resulted in the formation of a prominent cell pellicle in translucent variants that formed across the air-liquid interface (data not shown). This pellicle was absent in the opaque variant.

(iv) Virulence in a Galleria mellonella waxworm model.

To assess if virulence differences were present between the opaque and translucent colony variants, a Galleria mellonella waxworm model was used (34–36). Overall, the opaque variant exhibited a higher rate of killing (85% ± 5%) than the translucent variant (47% ± 10%) (Fig. 3C). Since the translucent form was less virulent, the colony opacity phenotype of cells was determined in worms injected with translucent cells that died between 24 and 48 h. In all cases, bacteria recovered from dead worms exhibited a strong shift from translucent cell populations to the opaque form, with values ranging from 35 to 90% opaque, compared to the less than 0.5% opaque cells in the starting translucent population that was injected into the G. mellonella waxworms.

(v) Antibiotic resistance.

Opaque and translucent variants were tested for resistance against a panel of antibiotics representing different classes (Table 2). There were no significant differences in the levels of resistance to the following antibiotics, with the following MICs (in parentheses) observed for both opaque and translucent variants: tetracycline (2 μg/ml), cefoxitin (96 μg/ml), colistin (3 μg/ml), and rifampin (1.5 μg/ml). For the aminoglycoside class of antibiotics, there were significant differences between opaque and translucent variants for gentamicin (>256 μg/ml for opaque and 32 μg/ml for translucent), amikacin (256 μg/ml for opaque and 128 μg/ml for translucent), and tobramycin (48 μg/ml for opaque and 16 μg/ml for translucent). The levels of resistance for trimethoprim, ciprofloxacin, and ceftriaxone could not be determined, as the levels of resistance for AB5075 were above the highest concentration of antibiotic present in the Etest strips.

Additional strains of A. baumannii exhibit the opaque/translucent colony phenotype.

To determine if the variable colony phenotype was limited to A. baumannii strain AB5075, the ability of additional strains of A. baumannii to form colony variants was investigated. Four additional isolates, ATCC 17978, AB1712, AB307-0294, and AB0057, were tested for differences in colony morphology; AB1712, AB307-0294, and AB0057 produced opaque and translucent colony variants. The rate of phase variation for AB0057 and AB307-0294 ranged from 1/10² to 1/10³ for both opaque and translucent variants. However, the rate of phase variation for AB1712 was greatly favored in the translucent to opaque direction (1/10²) compared to the opaque to translucent direction (ranging from 1/10⁴ to 1/10⁵). We were unable to detect variants of ATCC 17978, and the colony morphology of this strain was distinct from that of most other A. baumannii isolates and exhibited a translucent phenotype with irregular and spreading (motile) colonies on 0.8% agar plates. The inability of ATCC 17978 to undergo phase variation was not due to decreased production of the extracellular signal that stimulates phase variation, as spent culture supernatants of ATCC 17978 promoted switching of both opaque and translucent variants of AB5075 (data not shown).

Since the opaque form was more virulent in Galleria mellonella, it was of interest to determine if the opaque form also predominated in vivo during human infections. Fresh clinical isolates that were passaged only once after the primary isolation from body sites were tested for the ability to produce opaque or translucent colonies. Among the isolates, 7 were predominately opaque (AB991, AB1009, AB1011, AB1020, AB499, AB552, and AB1051) and 5 were translucent (AB997, AB732, AB248, AB45, and AB1040). The opaque isolates AB991, AB1009, AB1011, and AB1020 were capable of phase variation at frequencies of 1/10² to 1/10³ for both the opaque and translucent forms. Isolates AB1040, AB1051, and AB499 were capable of phase variation at significantly lower frequencies (approximately1/10⁵). Lastly, the translucent clinical isolates AB997, AB732, AB248, and AB45 did not switch to the opaque form at detectable frequencies (<1/10⁵).