Selection of a Highly Monensin-Resistant Prevotella bryantii Subpopulation with Altered Outer Membrane Characteristics

Abstract

Prevotella bryantii cultures treated with monensin grew more slowly than untreated cultures, but only if the monensin concentration was greater than 1 μM. Cultures that were repeatedly transferred (eight transfers or 25 doublings) with monensin always grew rapidly, even at a 10 μM concentration. The amount of monensin needed to facilitate half-maximal potassium depletion (K_d) from monensin-selected cells was 16-fold greater than “unadapted” wild-type cultures (3,200 versus 200 nM). Cells taken from continuous culture had a K_d of 100 nM, and these inocula could not grow in batch culture when the monensin concentration was greater than 300 nM. Continuous cultures treated with monensin nearly washed out, but the surviving cells had a K_d of 1,300 nM. When wild-type cells were transferred in batch culture with 10 μM monensin, the K_d did not reach its maximum value (3,200 nM) until after eight transfers (25 doublings). K_d declined when monensin was removed, and it took eight transfers to reach the control value (200 nM). The most probable number of wild-type cells was 1,000-fold lower than of the monensin-selected cells, but calculations based on relative growth advantage and K_d indicated that the wild-type culture had 1 to 10% highly monensin-resistant cells. Cell pellets of wild-type cultures were more difficult to disperse than were monensin-selected cells, and water-soluble phenol extracts of monensin-selected cells had 1.8-fold more anthrone-reactive material than did the wild type. Wild-type cultures that were washed in Tris buffer (pH 8.0) released little alkaline phosphatase and were agglutinated by lysozyme. Monensin-selected cultures leaked ninefold more alkaline phosphatase and were not agglutinated by lysozyme. Wild-type colonies taken from high-dilution agar roll tubes retained the lysozyme agglutination phenotype even if transferred with monensin, and monensin-selected colonies were never agglutinated. These observations indicated that wild-type P. bryantii cultures had a subpopulation with different outer membrane characteristics and increased monensin resistance.

The ionophore, monensin, has been used to modify ruminal fermentations since the 1970s, and this antibiotic decreases methane and ammonia losses (24). When energy availability and nitrogen retention are improved, the efficiency of feed utilization increases. Goodrich et al. (10) summarized 228 cattle feeding trials (11,274 animals) and reported that monensin increased the feed efficiency by 6.4%. The efficacy of monensin as a feed additive has persisted, and the average improvement in feed efficiency is still 5.6% (26).

The action of monensin is consistent with its effect on ruminal bacteria in vitro. Monensin is most active against gram-positive bacteria, and these species produce large amounts of hydrogen, a precursor of methane (5), and ammonia, an end product of protein degradation (11, 23). Extrapolation of this mode of action back to the rumen has, however, been stymied by difficulties in enumeration. In vivo, ruminal bacteria are surrounded by capsular material (8), and virtually all of them stain as gram negative or gram variable (12).

Monensin is a metal/proton antiporter (22) that causes potassium depletion (24), and this characteristic has been used as an index of monensin sensitivity (17). Feeding studies indicated that cattle fed monensin had mixed ruminal bacteria that were eightfold more resistant to monensin than untreated controls, but the effect of monensin on gram-negative ruminal bacteria could not be excluded (17). Some gram-negative ruminal bacteria are initially sensitive to monensin and must be adapted before they can grow (5, 20, 21).

Prevotella bryantii is a gram-negative ruminal bacterium that can become highly resistant to monensin, but preliminary experiments indicated that it was initially monensin sensitive. The following experiments sought to define whether this change was a simple adaptation of all cells or a selection of particular cells.

MATERIALS AND METHODS

Culture growth.

P. bryantii B₁4 was grown anaerobically in a basal medium containing (per liter): 292 mg of K₂HPO₄, 292 mg of KH₂PO₄, 436 mg of NH₄SO₄, 480 mg of NaCl, 100 mg of MgSO₄ · 7H₂O, 64 mg of CaCl₂ · 2H₂O, 4,000 mg of Na₂CO₃, 600 mg of cysteine hydrochloride, 1,000 mg of Trypticase, and 500 mg of yeast extract and volatile fatty acids (23). The medium was prepared under O₂-free CO₂, and sterile, anoxic glucose (10 mM, final concentration) was added to the medium after it was autoclaved. Batch cultures were grown in 18-by-150-mm tubes (10 ml) that were stoppered with butyl rubber or in 500-ml flasks that were continuously bubbled with O₂-free CO₂. The inoculum size was 10% (0.2 optical density units), the transfer interval was 24 h, and the incubation temperature was 39°C.

Continuous cultures were grown anaerobically in chemostats (170-ml culture vessel, 10 mM glucose) at a dilution rate of 0.1 h⁻¹, and at least a 98% turnover of medium was allowed to pass through the culture vessel before samples were taken. Cultures were harvested by centrifugation (10,000 × g, 15 min, 5°C), and cell-free supernatants and concentrated cell suspensions (fourfold) were stored at −15°C until they were analyzed. Optical densities were measured with a Gilford Model 260 spectrophotometer (600 nm, 1-cm cuvette). The salt tolerance of cultures was examined by adding a sterile, anoxic NaCl solution directly to tubes containing basal medium.

Intracellular potassium.

Culture samples (1 ml) were centrifuged (13,000 × g, 5 min, 22°C) through 0.3 ml of silicon oil (50:50 mixture of Dow-Corning 550 and 556; Dow-Corning Corp., Midland, Mich.). The microcentrifuge tubes were frozen (−15°C), and the pellets were removed with a pair of dog nail clippers. The cell pellets were digested at room temperature for 24 h in 3 N HNO₃, and insoluble cell debris was removed by centrifugation (13,000 × g, 5 min, 22°C). Potassium content was determined with a flame photometer (Cole-Parmer 2655-00 digital flame analyzer; Cole-Parmer Instrument Co., Chicago, Ill.).

Intracellular ATP.

Cells from 1 ml of culture were extracted for 20 min with 0.5 ml of ice-cold, 14% perchloric acid, which contained 9 mM EDTA. After centrifugation (13,000 × g, 5 min, 22°C), the supernatant fluid (1 ml) was neutralized with 0.5 ml of KOH-KHCO₃ (1 M each, 0°C). Potassium perchlorate was removed by centrifugation (13,000 × g, 5 min, 22°C), and the supernatant fluid was assayed for ATP by using the firefly luciferin-luciferase method (19). Neutralized extracts were diluted 50-fold with Tris (40 mM, pH 7.75) containing 2 mM EDTA, 10 mM MgCl₂, and 0.1% bovine serum albumin. The luciferase reaction was initiated by adding 100 μl of a purified luciferin-luciferase mix to 100 μl of diluted extract according to the supplier’s recommendations (Sigma Chemical Co., St. Louis, Mo.). Light output was immediately measured with a luminometer (Model 1250; LKB Instruments, Inc., Gaithersburg, Md.) by using ATP as a standard.

Potassium depletion experiments.

Monensin-selected cultures were transferred once in the absence of monensin to reduce carryover. Exponential phase wild-type and monensin-selected cells were washed and resuspended in basal medium lacking ammonia, Trypticase, yeast extract, and volatile fatty acids, and the final cell density was adjusted to one optical density unit (160 μg of protein/ml). Cell suspensions were energized with glucose (4 mg/ml) and incubated anaerobically at 39°C for 30 min. Monensin (0 to 10 μM) was added, and cells were incubated for another 10 min. Potassium depletion was estimated from the decrease in intracellular potassium compared to controls (no monensin).

Monensin binding.

Washed cells (160 μg of protein/ml) were incubated in 50 mM sodium phosphate buffer (50 mM, pH 6.5) containing 10 mM KCl and 5 μM ¹⁴C-labeled monensin (0.62 μCi/mg; Eli Lilly and Co., Indianapolis, Ind.) for 4 h at 39°C. Cells were harvested by centrifugation (7,500 × g, 5 min, 5°C), and the supernatant was removed. Pellets were resuspended in 1 ml of H₂O and transferred to 10 ml of scintillation cocktail (Aqueous Counting Scintillant; Amersham Corp., Arlington Heights, Ill.).

MPN estimates.

Wild-type and monensin-selected cultures were serially diluted (10-fold increments) in the basal medium containing or lacking monensin (10 μM). The tubes were incubated at 39°C for 96 h, and growth was scored (as “+” or “−”) by the increase in optical density (600 nm). The most probable number (MPN) was estimated from replicate (n = 5) dilutions (14).

Agar roll tubes.

Wild-type and monensin-selected cultures were serially diluted (10-fold increments) into molten basal agar medium (47°C, 2% agar) containing or lacking monensin (10 μM). The tubes were rolled in an ice bath to solidify the agar, and the roll tubes were incubated at 39°C for 96 h. Isolated colonies (n = 11) from the highest-dilution tubes of wild-type and monensin-selected cultures were picked and inoculated into basal medium broth containing glucose and incubated at 39°C.

Lysozyme agglutination.

Cultures were harvested by centrifugation (10,000 × g, 10 min, 5°C), and the cell-free supernatant was discarded. The cell pellet was resuspended in Tris buffer (30 mM, pH 8.0) or Tris buffer with lysozyme (1 mg or 70,000 U/ml; Sigma). Agglutination was monitored microscopically or via sedimentation after low-speed centrifugation (120 × g, 2 min, room temperature).

Lipopolysaccharide extraction.

Wild-type and monensin-selected cultures (1 liter, 5.5 optical density units) were harvested by centrifugation, and lipopolysaccharides were extracted by the hot phenol (85%) method described by Westphal and Jann (27). The aqueous phase above the phenol was removed, dialyzed for 24 h (3,000-kDa cutoff) against 10,000 volumes of Tris buffer (10 mM and 1 mM MgCl₂, [pH 7.5], 4°C), lyophilized to dryness, and resuspended in Tris buffer (30 mM, pH 7.0). Anthrone-reactive material (carbohydrate) was determined by the method of Bailey (3).

Lysozyme inhibition.

Micrococcus luteus cells (Sigma) were resuspended in Tris buffer (30 mM, pH 7.0, 0.1 to 0.6 mg of cells/ml), lysozyme was added (0.1 μg or 7 U per ml), and lysis was estimated from the decrease in optical density (600 nm, 1-cm cuvettes). Lipopolysaccharide from water-soluble, phenol extracts of P. bryantii cultures and lysozyme were mixed in cuvettes and incubated (37°C, 10 min) prior to addition of M. luteus cells to estimate lysozyme inhibition by lipopolysaccharide (2.8 and 5.6 mg of anthrone-reactive material per mg of lysozyme).

TLC.

Equal amounts of anthrone-reactive material from lipopolysaccharide extracts (typically, 1 to 3 μg of hexose equivalent) were applied to spots 1 cm apart on thin-layer chromatography (TLC) plates (Merck Art 5737 silica gel 60-kieselguhr F₂₅₄ precoated; layer thickness, 0.25 mm) and separated by immersing the bottom 1 cm of the plate in n-propanol–ethyl acetate–water (7:1:4). Dried plates were sprayed with an anisaldehyde-based reagent (27 ml of ethanol, 0.3 ml of acetic acid, 1.5 ml of sulfuric acid, 1.5 ml of p-anisaldehyde) and placed in a 110°C oven for 5 min. Under these conditions, many sugars stain as unique colors.

Other analyses.

Protein from NaOH-hydrolyzed cells (0.2 N NaOH, 100°C, 15 min) was assayed by the method of Lowry et al. (18). The ratio of protein to optical density for both wild-type and monensin-selected cultures was 160 μg of protein/ml per optical density unit (1-cm cuvettes, 600 nm). Alkaline phosphatase was determined with a colorimetric assay by using p-nitrophenyl phosphate (PNPP) as described by Garen and Levinthal (9).

Ionophores and chemicals.

Monensin and lasalocid (Sigma) stock solutions were prepared separately as sterile, anoxic solutions in ethanol, and the final concentration of ethanol in cultures was always less than 2%. Ethanol (2%) had no effect on the growth of P. bryantii. All other chemicals used were reagent grade.

Replication.

All experiments were performed two or more times, and the measurements were highly reproducible. The coefficient of variation (standard deviation/mean) was always less than 10%.

RESULTS

Batch cultures.

When P. bryantii B₁4 was grown anaerobically in batch culture, monensin had little effect on the initial growth rate until the concentration was greater than 1 μM (Fig. 1a). When 10 μM monensin was provided, the initial growth rate was 50% slower. If cultures were transferred repeatedly (eight transfers or 25 doublings) with 10 μM monensin, the depression in the initial growth rate was not observed, and the monensin-selected cultures always grew as rapidly as the untreated controls (Fig. 1a). Wild-type and monensin-selected cultures had an intracellular potassium concentration of 1,200 nmol/mg of protein, but wild-type cultures that were treated with monensin had an intracellular potassium concentration of only 500 nmol/mg of protein (Fig. 2a). When the growth rate of monensin-treated cultures increased, there was an increase in intracellular potassium. Wild-type and monensin-selected cultures had an intracellular ATP concentration of 4 nmol/mg of protein, but wild-type cultures that were treated with monensin had an intracellular ATP concentration of 8 nmol/mg of protein (Fig. 2b). When the growth rate of monensin-treated cultures increased, there was a decrease in intracellular ATP.

FIG. 1.

(a) Effect of monensin on the initial growth rate of wild-type (○) or monensin-selected (●) P. bryantii B₁4 batch cultures. (b) Effect of monensin on the initial growth rate of inocula that were taken from continuous culture (D = 0.1 h⁻¹). Open symbols (□) show inocula from a continuous culture that had not been treated with monensin, and closed symbols (■) indicate inocula that were obtained from a continuous culture that had been treated with monensin (10 μM). In each case, the initial cell density was 0.2 optical density units.

FIG. 2.

(a) Relationship between intracellular potassium and the growth rate of wild-type cultures without monensin (○), wild-type cultures treated with 10 μM monensin (●), and monensin-selected cultures that were treated with 10 μM monensin (▴). (b) Relationship between intracellular ATP and the growth rate.

Wild-type cultures that were washed and incubated in a medium lacking nitrogen and containing glucose did not grow, but the intracellular potassium concentration was nearly as high as that of the growing cells (950 versus 1,200 nmol/mg of protein). When monensin was added, intracellular potassium declined and reached a new steady-state concentration within 10 min. Potassium losses were dependent on the monensin concentration, and a double-reciprocal plot of monensin concentration versus potassium depletion was linear (Fig. 3a). The amounts of monensin needed to cause half-maximal potassium depletion (K_d) from wild-type and monensin-selected cells were 200 and 3,200 nM, respectively. Subsequent experiments demonstrated that the changes in K_d occurred gradually rather than abruptly. When wild-type cultures were transferred repeatedly (eight transfers or 25 doublings) with monensin, K_d reached its maximum value, and it took eight transfers (25 doublings) without monensin for the K_d to return to its control value (Fig. 3b).

FIG. 3.

(a) Double reciprocal plot of 1/monensin concentration versus 1/potassium depletion from wild-type (○) or monensin-selected (●) P. bryantii B₁4 cultures. The amount of ionophore needed to catalyze half-maximal potassium depletion is determined from −1/K_d, the intercept of the abscissa. (b) K_d of wild-type cultures (▴) transferred with 10 μM monensin and monensin-selected cultures (▵) transferred without monensin. Dotted lines indicate wild-type (bottom) and monesin-selected (top) cultures.

Continuous cultures.

When batch cultures were inoculated with cells taken from glucose-limited continuous culture (dilution rate of 0.1 h⁻¹), the initial growth rates were similar to normal batch cultures, but even small amounts of monensin caused a decrease in the initial growth rate (Fig. 1b). Cells taken from glucose-limited continuous culture had a K_d value of only 100 nM. When monensin was added directly to the continuous culture vessel, intracellular potassium decreased, and the optical density declined at a rate similar to the dilution rate (Fig. 4). Because monensin was added to the continuous-culture vessel rather than to the medium reservoir, the monensin concentration decreased logarithmically with time and the culture eventually recovered. The next morning (24 h later), the optical density and the intracellular potassium concentrations approximated pretreatment values. When a second dose of monensin was added to the culture vessel, intracellular potassium decreased 50%, but optical density did not decline. Continuous cultures that had received a second dose of monensin were more resistant to monensin, and they grew rapidly in batch culture even if the monensin concentration was high (Fig. 1b). Monensin-treated continuous cultures had a K_d value of 1,300 nM.

FIG. 4.

Optical density (○) and intracellular potassium (▴) of wild-type P. bryantii continuous cultures (D = 0.1 h⁻¹, pH 6.7). Monensin (10 μM) was added directly to the culture vessel at time zero and at 25 h.

Cross-resistance.

Wild-type P. bryantii was more sensitive to lasalocid than monensin, and growth was completely inhibited at lasalocid concentrations of ≥10 μM. The K_d of lasalocid was 115 nM. Wild-type cultures that were transferred repeatedly (eight transfers or 25 doublings) with increasing concentrations of lasalocid eventually grew in the presence of 5 μM lasalocid, but the growth rate was always 30% lower than that of controls. Lasalocid-selected cells had a lasalocid K_d of 1,400 nM and a monensin K_d of 1,600 nM. Monensin-selected cells were nearly as resistant to lasalocid as lasalocid-selected cells, and lasalocid K_d values were comparable (1,350 and 1,400 nM, respectively).

Survival characteristics and salt tolerance.

Monensin-selected cultures were more sensitive to glucose starvation than were the wild-type cultures, and the MPNs after 8 days of starvation at 39°C were 10¹ and 10⁴ viable cells per ml, respectively. The growth rates of wild-type and monensin-selected cultures declined when NaCl was added to the growth medium, but monensin-selected cells were more sensitive. When 300 mM NaCl was added to basal medium, the growth rates of monensin-selected and wild-type cultures were 0.2 and 0.34 h⁻¹, respectively.

Outer membrane characteristics.

When wild-type and monensin-selected cultures (1.0 optical density unit or 160 μg of protein/ml) were treated with 5 μM ¹⁴C-labeled monensin, the monensin binding levels were similar (5.2 ± 0.4 and 4.9 ± 0.3 nmol of monensin/mg of cell protein, respectively). However, cell pellets of wild-type cultures were more difficult to disperse than cells from monensin-selected cultures. Wild-type cells that were washed and resuspended in Tris buffer agglutinated when lysozyme was added, and this agglutination could be monitored visually, as sedimentation, or microscopically (Fig. 5). The monensin-selected cells did not agglutinate or sediment when lysozyme was added. Lysozyme agglutination of wild-type cells could be counteracted by salt additions (either 100 mM KCl or NaCl) or by decreasing the pH to values less than 7.0 with HCl. When washed cell suspensions were extracted with phenol, the water-soluble extract above the phenol layer contained anthrone-reactive material. When the phenol extracts were subjected to TLC, the wild-type and monensin-selected cells were similar, but the monensin-selected cells had nearly twice as much anthrone-reactive material as did the wild-type cells (0.18 ± 0.007 versus 0.1 ± 0.008 mg of hexose equivalents/mg of cell protein). The phenol extracts inhibited the ability of lysozyme to lyse M. luteus cells, but extracts from monensin-selected cells had a K_i that was 3.4-fold less than that for wild-type cells (Fig. 6).

FIG. 5.

Effect of lysozyme on wild-type (left test tube and upper micrograph) and monensin-selected (right test tube and lower micrograph) P. bryantii cultures. Cultures (test tubes) were centrifuged (120 × g, 2 min, room temperature). The microscopic magnification (micrographs) is ×1,800.

FIG. 6.

Inhibition of lysozyme (0.1 μg/ml) activity by lipopolysaccharide extracts from wild-type (○) and monensin-selected (●) P. bryantii. The initial velocity (V_o) is expressed as the decrease in optical density of M. luteus cells per minute, and the initial S_o is the concentration of M. luteus cells in milligrams per milliliter. The closed triangles (▴) show the effect of lysozyme on M. luteus when no lipopolysaccharide extract was added.

Sonicated extracts of wild-type and monensin-selected cells had similar amounts of alkaline phosphatase (893 versus 959 nmol of PNPP/mg of cell protein/min, respectively), but monensin-selected cells that were washed and incubated in Tris buffer (pH 8.0) released 9 times more alkaline phosphatase into the cell-free supernatant (3.7 versus 35.5 nmol of PNPP/mg of cell protein/min). Wild-type cells that were washed and incubated in Tris buffer (pH 8.0) were relatively resistant to lysozyme, and large numbers of spheroplasts were not observed for at least 90 min (39°C, 1.0 mg of lysozyme/mg of cell protein). The monensin-selected cells were more susceptible to lysozyme, and the rate of spheroplast formation was at least threefold faster.

Adaptation versus selection.

Monensin-selected cells grew faster than wild-type cells when the monensin concentration was 10 μM, and monensin-selected cells achieved an optical density of 2.0 in 4.0 h (Fig. 7a). Over this same time interval (4.0 h), the wild-type cells had only achieved an optical density of 0.49. Based on this comparison (optical densities of 2.0 versus 0.49), the relative enrichment of monensin-selected cells by monensin would have been approximately 4 to 1 after each transfer. By using the 4-to-1 relative enrichment and the K_d values of each cell type (Fig. 3), it was possible to calculate the population K_d (percent wild-type × 213 nM + percent monensin-selected × 3,200 nM = the K_d of the total population) after each transfer (Fig. 7b). If the initial population had 0.1% monensin-resistant cells and 99.9% monensin-sensitive cells, the calculated K_d would not increase significantly until the fourth transfer. If the initial population had 1 and 99% monensin-resistant and monensin-sensitive cells, respectively, the calculated K_d increased after only two transfers and was in close agreement with the measured values. If the initial population had 10 and 90% monensin-resistant and monensin-sensitive cells, respectively, the calculated K_d increased very rapidly.

FIG. 7.

(a) Optical density of wild-type (○) or monensin-selected (●) P. bryantii cultures that were grown with 10 μM monensin. The arrow shows the fourfold relative growth advantage. (b) Calculated changes in K_d based on initial populations of 0.1, 1, or 10% monensin-resistant cells. Measured K_d values are indicated (▴).

When wild-type P. bryantii cultures were serially diluted in basal medium (broth) containing 10 μM monensin, the MPN was 6 logs lower than for controls without monensin (10³ versus 10⁹ viable cells/ml). Monensin-selected cultures had an MPN that was 1,000-fold greater (10⁶ viable cells/ml) when monensin (10 μM) was present. This comparison indicated that the wild-type culture only had 0.1% monensin-resistant cells, but it should be realized that highly diluted cultures have a much greater ratio of monensin to cells than do cultures after routine transfer (typically, 0.2 optical density units at time zero).

Differences in lysozyme agglutination allowed us to monitor monensin resistance visually as either sedimentation after slow-speed centrifugation (Fig. 5, test tubes) or microscopically (Fig. 5, micrographs). When wild-type cultures were transferred in basal medium lacking monensin, the phenotype was always “agglutinated,” but the phenotype shifted to “nonagglutinated” after four transfers with monensin. If “wild-type” cultures lacking monensin were serially diluted in agar roll tubes lacking monensin and colonies (n = 11) were inoculated into broth containing and lacking monensin, the phenotype remained “agglutinated.” Conversely, if “monensin-selected” cultures containing monensin were serially diluted in agar roll tubes containing monensin and colonies (n = 11) were inoculated into broth containing and lacking monensin, the phenotype remained “nonagglutinated.”

Only 75% of the cells from wild-type cultures were initially agglutinated and precipitated by lysozyme, but many of the remaining cells could be agglutinated if they were harvested by centrifugation (10,000 × g, 10 min, 5°C) and subjected to lysozyme treatment a second time. If wild-type cultures were treated with lysozyme four times, there was little further agglutination, and this cell population accounted for approximately 2.5% of the original optical density.

DISCUSSION

Many ruminal bacteria are currently being reclassified (15), and Prevotella ruminicola B₁4 was recently reclassified as P. bryantii (2). Attwood et al. (1) indicated that B₁4 did not persist in the rumen after inoculation, and they hypothesized that ruminal fluid contained a “toxin” that was specific to this bacterium. Recent 16S rRNA sequence analyses of mixed ruminal DNA, however, indicated that DNA clones often contained “the signature sequence for P. bryantii strain B₁4” (29). P. bryantii B₁4 grows rapidly with high concentrations of monensin (5); produces succinate, a precursor of propionate (12); and has been used a model of monensin resistance in the rumen (7).

Previous work with Streptococcus bovis, a monensin-sensitive bacterium, indicated that monensin catalyzed a decrease in ATP, but this bacterium generates its proton motive force via membrane-bound ATPase (25). When wild-type cultures of P. bryantii B₁4 were initially treated with monensin, ATP was high (not low) and ATP decreased as the culture’s specific growth rate increased. The effect of monensin on ATP of P. bryantii is consistent with its catabolic scheme. P. bryantii B₁4 has a cytochrome-linked, fumarate-reductase (16, 28), and ATP per se is not needed to regenerate a proton motive force.

P. bryantii cultures initially grew more slowly with monensin (10 μM) than untreated controls. Monensin catalyzes potassium depletion from sensitive bacteria (17, 24), and cultures initially exposed to monensin could not maintain normal concentrations of intracellular potassium. The monensin-selected cultures had normal intracellular potassium concentrations, and experiments with nongrowing cells indicated that that the concentration of monensin needed to catalyze half-maximal potassium depletion (K_d) was 16-fold greater than wild-type cells (3,200 versus 200 nM). Cells grown in continuous culture were even more sensitive to monensin than those grown in batch culture (K_d values of 100 and 200 nM, respectively), but the continuous cultures that were treated with monensin recovered and had a K_d of 1,300 nM.

Chen and Wolin (5) studied the effect of monensin and lasalocid, another ionophore that is used in the feed industry, on P. brevis, a closely related ruminal species, and indicated that they were able to generate monensin- and lasalocid-resistant strains with no cross-resistance. The strains generated by Chen and Wolin are no longer available, but our work indicated that monensin-selected P. bryantii cells were more resistant to lasalocid than were wild-type cells and vice versa. Lasalocid is a more hydrophobic molecule than monensin, and previous work indicated that S. bovis was more sensitive to lasalocid than to monensin (6). P. bryantii was never able to grow with 10 μM lasalocid, and the K_d for lasalocid was lower than the K_d value of monensin (115 versus 213 nM).

Newbold et al. (20, 21) indicated that P. ruminicola (now P. albensis) cultures that were selected with the ionophore, tetronasin, bound less ¹⁴C-labeled tetronasin than did wild-type cultures, and these authors hypothesized that the tetronasin-selected cells had smaller outer membrane porins. Our results indicated that monensin-selected P. bryantii cultures bound as much ¹⁴C-monensin as did wild-type cultures, but monensin-selected cell pellets were easier to disperse than were wild-type cells. Because phenol extracts from monensin-selected cells had twice as much anthrone-reactive material as the wild-type cells, it appeared that monensin resistance was mediated via an accumulation of hydrophilic carbohydrate rather than a decrease in porin size.

The alkaline phosphatase of gram-negative bacteria is a periplasmic protein, and most bacteria do not release alkaline phosphatase from the periplasm unless they are treated with EDTA (9). Wild-type cells that were washed in Tris buffer (without EDTA) did not release much alkaline phosphatase, but larger amounts were detected when monensin-selected cells were treated in a similar fashion. Monensin-selected cells were more susceptible to lysozyme than were wild-type cells, and water-soluble phenol extracts from monensin-selected cells did not inhibit lysozyme as strongly as did extracts from wild-type cells.

Lysozyme is a basic protein that can agglutinate gram-negative bacteria (4, 13) via its interaction with the lipid A moiety of the outer membrane (4). Wild-type P. bryantii cultures were agglutinated by lysozyme, but monensin-selected cells were not. Wild-type cultures shifted from “agglutinated” to “nonagglutinated” phenotype and vice versa when monensin was added or removed from the medium. Colonies from agar roll tubes had similar morphologies, but each agglutination phenotype was retained regardless of monensin treatment. These results supported the idea that the wild-type culture had two populations of cells, and only a certain percentage of the total population could become highly monensin resistant. MPN values indicated that only 1 of 1,000 cells had the capacity to become highly monensin resistant, but it should be realized that monensin is a highly lipophilic substance that concentrates in cell membranes. When the cell number is low (in the case of high serial dilutions), the ratio of monensin to cells is higher and monensin is more toxic.

The K_d of wild-type cultures did not reach its maximum value until after eight transfers, but the K_d increased significantly after only two transfers with monensin. Monensin-selected cells grew faster than wild-type cells when monensin was present (10 μM), but this growth advantage was only fourfold after each transfer (Fig. 7a). If only 1 of 1,000 cells had the capacity to become highly monensin resistant, there would have been little increase in K_d until the third or fourth transfer (Fig. 7b). Subsequent calculations of K_d and relative growth indicated that the wild-type culture probably had from 1 to 10% monensin-resistant cells. Because monensin-sensitive cells could be agglutinated and precipitated by lysozyme, whereas highly monensin-resistant cells could not, it was possible to estimate the relative population size after repeated lysozyme treatments. Approximately 2.5% of the cells were never agglutinated by lysozyme, and this value was in agreement with calculations based on relative growth advantage and changes in K_d (Fig. 7b). The wild-type culture had few highly monensin-resistant cells, but their K_d was 16-fold higher. The impact of highly monensin-resistant cells on the overall K_d estimate was significant, and simple algebraic calculations ([100% × 200 nM] = [2.5% × 3,200 nM] + [97.5% × K_d of monensin-sensitive cells]) indicated that the K_d of monensin-sensitive cells was actually only 123 nM.

When monensin-selected cultures were transferred in medium lacking monensin, the K_d value declined and approached the wild-type value in eight transfers (25 doublings). Highly monensin-resistant cells seemed to grow as fast as monensin-sensitive cells when monensin was not present, but these cultures were more sensitive to energy starvation and salt stress. Based on these results, it is conceivable that the highly monensin-resistant cells have a survival disadvantage.

Previous work with mixed ruminal bacteria taken directly from cattle indicated that monensin supplementation caused an almost immediate increase in K_d, and the K_d reached its maximum value in only 2 days (17). Because the change in K_d was very rapid, it appeared that the rumen already had a large population of monensin-resistant bacteria. Wild-type P. bryantii and mixed ruminal bacteria from cattle not consuming monensin had similar K_d values (200 and 180 nM, respectively), and the K_d values after monensin selection were also comparable (3,200 and 1,800 nM, respectively). Batch cultures of P. bryantii were routinely selected with 10 μM monensin, a value that exceeds the in vivo concentration, but it should be noted that continuous cultures were much more sensitive and could be inhibited by as little as 0.04 μM monensin (Fig. 1b). Based on these results, it is conceivable that monensin-dependent changes in the K_d of mixed ruminal bacteria could be at least partially due to a selection of gram-negative bacteria.