Influence of Amino Acids on Low-Density Escherichia coli Responses to Nutrient Downshifts

A vast bibliography on nutrient effects on high-density culturesexists, while it has been overlooked that low densities of starvingcells are often the rule in natural environments. By means ofa novel sensitive ß-galactosidase assay, we examinedEscherichia coli transitions to minimal media when the cellconcentration was 100 to 10,000 cells per ml. As in high-densitycultures, the enzyme activity depended on amino acid availabilityand was subject to catabolite repression and stringent control.In all conditions tested, despite the presence of other nutrientsources, the relationship between ß-galactosidaseactivity and the L-amino acid pool was hyperbolic. The affinityconstant when the amino acid pool was the only nutrient sourceaveraged 14 µM after 90 min and increased up to 222 µMafter 4.5 h. While investigating the transition from lag phaseto exponential phase, we observed that the cells did not enterinto starvation mode in the presence of amino acids, even whenthe nutrient amount was insufficient to support full survival.Based on these premises, the switch from starvation to hungerwas investigated in relation to the amino acid pools. A criticalrange of concentrations at which E. coli linearly synthesizedß-galactosidase despite, at the same time, sufferinga large decrease in cell viability was then recognized. Sinceboth ß-galactosidase production and the dilution ratewere reduced by more than half in the absence of leucine, weexamined the contribution of leucine to cell recovery capabilities.

INTRODUCTION

Top

Introduction

The natural state in which microorganisms conduct their livesoscillates between feast and famine (20). The growth rate ineither soil or water is often not measurable, with duplicationtimes in the order of days or weeks (16), because most bacteriaare under starving conditions and only bacteria inhabiting "hotspots," such as the gut, rhizosphere soil, or sites around vegetabledebris, have wide access to nutrients (7). Simulating environmentalfluctuations of nutrient conditions via controlled nutritionalshifts for Escherichia coli showed the activation/deactivationof many different molecular systems coordinating the regulationof the cell metabolism by carbon, nitrogen, and amino acids,like the cyclic AMP (cAMP)-cAMP receptor protein modulon (34),the nitrogen-regulated response (31), and the stringent modulon(4, 9, 24). However, upon the complete exhaustion of an essentialnutrient, cells were shown to enter into stationary phase througha general stress response controlled by the rpoS-encoded sigma-Ssubunit of the RNA polymerase (17). When nutrients were limitedyet not completely absent, the cells exhibited the hunger response,which partly overlapped with the stress response (13). It hasbeen thoroughly shown how nutrient downshifts or upshifts alsoinfluence the induction of ß-galactosidase due tocatabolite repression (22), stringent control (30), and stationary-phasecontrol (14). On the other hand, the lack of suitable analyticalapproaches and the need to uniform experimental conditions directedresearch mainly towards high bacterial cell densities (around10⁸ to 10⁹ cells per ml), in continuous or batch cultures (21),despite the fact that numbers of cells of different entericspecies in fluid bodies are low (commonly <100 cells perml) and reach 10⁵ cells per ml only in proximity to a sourceof contamination.

The aim of this research was to analyze the effect of nutrientdownshifts on the bacterial metabolic response at cell densitiesof 100 to 10,000 cells per ml so as to find the minimum nutrientconditions at which cells resumed growth and/or expressed anadaptive behavior by synthesizing catabolic enzymes. For a model,we used E. coli that was grown in rich medium and diluted inminimal medium and ß-galactosidase as an indicatorof bacterial metabolism. The amino acid pool proved necessaryand sufficient in supporting the cell recovery from the nutritionalgap and allowing a faster ß-galactosidase expression,in substantial agreement with results at high bacterial densities.We examined in depth the relationship between amino acid concentrationand ß-galactosidase activity and found it hyperbolic,regardless of other nutrient additions. Then, we compared ß-galactosidaseand growth kinetics during the cell transition to the exponentialphase and observed that the presence of the amino acid poolwas sufficient to induce E. coli to synthesize ß-galactosidaseinstead of entering a safer survival-starvation mode, even atthe cost of large numbers of individual losses. Finally, weinvestigated the consequences deriving from single amino aciddeficiencies and, particularly, the significant reductions inß-galactosidase synthesis and cell growth caused,in all strains tested, by the removal of leucine from the aminoacid pool.

MATERIALS AND METHODS

Top

Materials and Methods

Bacterial strains and reagents. Escherichia coli ATCC 25922 and Klebsiella pneumoniae ATCC 13883were purchased from the American Type Culture Collection (ATCC,Manassas, VA). Citrobacter freundii ATCC 8090 and Escherichiacoli K-12 ATCC 23716 were purchased from the Deutsche Sammlungvon Mikroorganismen und Zellkulturen (Braunschweig, Germany).All reagents, unless otherwise specified, were purchased fromSigma-Aldrich, St. Louis, MO.

Cell preparation and counts. Fresh cultures were prepared by overnight growth in Luria-Bertanimedium (LB) (Difco, Sparks, MD) at 37°C. Tests were performedon diluted cell samples after centrifugation (1,800 x g, 4 min)and pellet resuspension in 0.85% NaCl physiological solution.Due to the limited number of cells used in the experiments (500to 20,000 cells per ml), optical density (turbidity) measurementsdid not provide significant values. Thus, we performed cellcounts by MacConkey agar plating, with 40 h of incubation at37°C (Oxoid, Basingstoke, United Kingdom).

Induction of ß-galactosidase synthesis and the ß-galactosidase assay. The synthesis of ß-galactosidase was induced with0.1 mM of the synthetic inducer IPTG (isopropyl-ß-D-thiogalactopyranoside)(6) in phosphate buffer (PB) at pH 7.2, which was made of 47.7mM Na₂HPO₄, 22.0 mM KH₂PO₄, and 0.5 mM MgSO₄·7H₂O unlessotherwise indicated. All tests were performed in 100- by 15-mmglass tubes containing 2.25 ml of PB; they were incubated ina water bath at 37°C. The 20 mM stock solutions of eachL-amino acid were prepared in PB and stored frozen. Glasswareand media were autoclaved (10⁵ Pa; 121°C) and buffers andsolutions were filter sterilized (0.45 µm) before use.

ß-Galactosidase activity was determined by the hydrolysisof MUG (methylumbelliferyl-ß-D-galactoside) (1). Atthe end of the induction time, 130 µl of MUG in dimethylsulfoxide (5 mg/ml), 25 µl of 1% (wt/vol) sodium dodecylsulfate, and 20 µl of chloroform were added to the cultures;after vortexing, the hydrolysis mixture was incubated for 60min at 37°C. The hydrolysis was stopped by adding 100 µlof 2 N NaOH. After the solution was gently mixed, 2.21 ml ofthe solution, containing the fluorescent product of ß-galactosidasehydrolysis 4-methylumbelliferone, was transferred to a 10- by10-mm quartz cuvette by avoiding the sampling of chloroformphase. The heating block of the spectrofluorimeter, a PerkinElmerLS50B (Wellesley, MA), was set at 37°C, the excitation wavelengthat 362 nm, and the emission wavelength at 445 nm, with bothslit widths at 2.5 nm. ß-Galactosidase was expressedas standard enzymatic units. One unit will hydrolyze 1.0 µmolof o-nitrophenyl ß-D-galactoside to o-nitrophenoland D-galactose per min at pH 7.3 and 37°C.

Preparation of MOPS medium. MOPS (potassium morpholinepropane sulfonate) medium was preparedaccording to the method of Neidhardt et al. (27) and comprised40 mM MOPS, 4 mM Tricine, 10 µM FeSO₄·7H₂O, 1.32mM K₂HPO₄, 9.52 mM NH₄Cl, 276 µM K₂SO₄, 500 nM CaCl₂·2H₂O,523 µM MgCl₂, 50 mM NaCl, 3 nM (NH₄)₆Mo₇O₂₄·4H₂O,400 nM H₃BO₃, 30 nM CoCl₂, 10 nM CuSO₄, 80 nM MnCl₂, and 10nM ZnSO₄.

Induction of ß-galactosidase synthesis and ß-galactosidase assay in the absence of a single amino acid. Overnight cultures in LB (0.5 ml) were centrifuged, dilutedto 1,000 to 5,000 (E. coli) or 20,000 (C. freundii) cells perml in 0.1 mM IPTG in PB, and aliquots of 2.25 ml were incubatedat 37°C in the presence of different amino acid pools. After90 min, ß-galactosidase was measured as describedabove. The amino acid pool, designated AA, contained all 20natural L-amino acids in equimolar quantities, each at 13.3µM (1/20 each; total concentration, 266 µM). Certainamino acid mixture AA-x contained 19 natural L-amino acids inequimolar quantities, each at 14.0 µM (total concentration,266 µM), without the indicated amino acid (x).

RESULTS

Top

Results

Dependence on the presence of amino acids of ß-galactosidase kinetics in the lag phase. We subjected Escherichia coli ATCC 25922 to a remarkable nutritionaldownshift, from the highly nutritional LB to poor minimal mediumPB, and we assessed the metabolic performances of E. coli bymonitoring the synthesis of the inducible enzyme ß-galactosidasein environments with low cell densities (1,000 to 5,000 cellsper ml) under different nutrient conditions. A short incubationtime (90 min) was selected so as to evaluate the cell behaviorduring the lag phase. The presence of amino acids supportedthe synthesis of ß-galactosidase (Fig. 1) alreadybefore bacterial growth. In the absence of external amino acids,ß-galactosidase activity was very low (<5%), whileit increased to 75% of maximum activity at the given conditions,when the amino acid concentration was increased from 10 to 50µM. The induced ß-galactosidase activity wasplotted against the total amino acid concentration ranging between0 and 200 µM (0 to 10 µM each). The Hill sigmoidalcurve was selected at first to interpolate experimental data.Since the best-fitting calculated n value (measure of the degreeof sigmoidicity) was close to 1 (0.98), the study was redirectedto the hyperbolic function, which proved more suitable for theapplication. Both K_m (Table 1) and V_max constants were calculatedby setting the ß-galactosidase activity measured inthe absence of amino acids (ß-gal_AA=0) to 0 unitsto be used as a blank reference. Second, we took into accountthe real values of ß-gal_AA=0, in order to calculateK_AA, which represents the amount of external amino acids requiredto induce 50% of the V_ß_-galmax (V_max plus ß-gal_AA=0), the hypothetical maximum ß-galactosidaseactivity. The difference between K_m and K_AA was termed Int_AA,a value for the intracellular supply of amino acids and representingthe concentration of external amino acids that the cell wouldrequire in order to induce an enzymatic activity equivalentto a ß-gal_AA=0.

View larger version (10K):
[in this window]
[in a new window]

FIG. 1. ß-Galactosidase activity affected by amino acid concentration. The final curve was the result of 20 different experiments conducted in an 8-month period to reduce effects due to climatic influence. The experimental points ( ${blacksquare}$ ) are expressed as mean percentages of ß-galactosidase activity versus amino acid concentrations ± standard errors (SE) (error bars) of 3 to 20 tests. At first, the absolute value of ß-gal_AA=0 was calculated in every test and subtracted from the experimental points, in order to evaluate ß-galactosidase activity only due to added amino acids. The upper limit of the ß-galactosidase/cell production V_max was then determined in every test by nonlinear least-squares analysis (Microcal Origin version 5.0 software) in the following hyperbole equation: y = (V_max x x)/(K_m + x). After that, the following equation was calculated: V_ß_-galmax = V_max + ß-gal_AA=0. In order to recalculate, in every test, ß-gal_AA=0 and all experimental points as percentages of V_ß_-galmax, we assumed the limiting value V_ß_-galmax to be the highest hypothetical ß-galactosidase activity per cell (100%) inducible at the test conditions. Finally, by introducing the mean ß-gal_AA=0 as a constant value in the hyperbole equation y = ß-gal_AA=0 + [(V_max x x)/(K_m + x)], we generated a final curve which was calculated from the means of normalized experimental points.

View this table:
[in this window]
[in a new window]

TABLE 1. Kinetic parameters with amino acid as the only nutrient source^a

The study was extended to E. coli K-12 with results similarto those of studies with E. coli ATCC 25922 (Fig. 2). Also,Citrobacter freundii (Fig. 3) was taken into consideration toverify similarities/differences of enteric non-E. coli strains.A close hyperbole-fashioned relationship was recorded at lowamino acid concentrations, despite the fact that C. freundiihad a 10-fold-lower ability to produce the ß-galactosidaseenzyme, and consequently, a 10-fold-higher number of cells wasrequired to equalize E. coli test sensitivity.

View larger version (11K):
[in this window]
[in a new window]

FIG. 2. ß-Galactosidase activity of E. coli K-12 affected by amino acid concentration. See Materials and Methods for the procedure. The AA curve (for all 20 amino acids; solid line) was determined as described in the legend to Fig. 1. Microcal Origin version 5.0 software was used to plot the effects of AA-leucine (dashed line) and AA-isoleucine (dotted line) by interpolation. The experimental points ( ${blacklozenge}$ , ${blacksquare}$ , and •) are expressed as mean percentages ± SE (error bars) of three to six tests.

View larger version (12K):
[in this window]
[in a new window]

FIG. 3. ß-Galactosidase activity affected by amino acid concentration. The procedure used is as described in the legend to Fig. 1 except for the use of Citrobacter freundii ATCC 8090 at 17,000 to 19,000 cells per ml. The experimental points ( ${blacksquare}$ ) are expressed as mean percentages ± standard deviations (SD) (error bars) of two or three tests.

Retaining the hyperbolic relationship between the amino acids and the synthesis of ß-galactosidase in the presence of other carbon or nitrogen sources. We improved the nutritional quality of the minimal medium PBby adding several carbon and nitrogen sources singularly andin combination. However, when amino acids were lacking, thesynthesis of ß-galactosidase in E. coli was severelyrestricted (ß-gal_AA=0 < 5%), except when the completeMOPS minimal medium (27) was used (ß-gal_AA=0 <20%). The presence of amino acids stimulated, at different degrees,the enzyme production, depending on the presence of other nutrientsfor absolute yields (Table 2) . Despite the variability of mediumcomposition, the correlation with the amino acid concentrationwas unaffected. As expected, the addition of glucose inducedsevere catabolite repression, impairing 60% of the ß-galactosidaseproduction, partly reactivated by cyclic AMP, which also relievedcatabolite repression caused by amino acids as the only carbonsources (10 to 20%). The value of K_AA_-_n, i.e., K_AA in the presenceof other nutrients (n), was reduced by 80% by the simultaneouspresence of succinate and ammonium chloride. Also, maltose andlactate reduced K_AA_-_n values (between 40 and 60%), while othernutrients were noneffective. The use of the widely recognizedMOPS allowed us to investigate a less severe nutritional downshiftdue to the presence of micronutrients (see Materials and Methods);for this reason, the cell production of ß-galactosidasewas nearly 20% even in the absence of amino acids, but the presenceof external amino acids was still required to stimulate 80%of the ß-galactosidase production both in the presenceand absence of glycerol.

View this table:
[in this window]
[in a new window]

TABLE 2. V_ß-galmax-n and K_AA-n under different nutrient conditions

Longer-term influence of amino acid concentration on growth rate and ß-galactosidase activity. We investigated the effect of amino acids on ß-galactosidaseproduction and the dilution rate after a 4.5-hour incubation(Fig. 4a to c). In the absence of amino acids, ß-galactosidaseactivity was 300 times lower than the V_ß_-galmax, witha mean value of 1.5 x 10^–9 U/cell, and cell growth approachedzero, with a –0.03 h^–1 mean growth rate {µ_mean= t_x^–¹ x 2.303 x [log₁₀ (cells at t_x) – log₁₀ (cellsat t₀)]} (2) corresponding to a 10% decrease in the number ofculturable cells (Fig. 4c). At a 5 µM amino acid concentration,ß-galactosidase activity increased by 17-fold (2.5x 10^–8 U/cell). In the range of 5 to 70 µM, theenzyme synthesis was linearly correlated with the amino acidconcentration, whereas the cell duplication was not sufficientlysupported when the concentration was lower than 40 µM,such that after 270 min, the number of culturable cells suffereda significant 60 to 80% decrease in all experiments involvingamino acid concentrations higher than 0 µM but lower than40 µM (Fig. 4c). It was only at amino acid concentrationshigher than 40 µM that a positive µ_mean was measuredand the cells reached the exponential stage. When the aminoacid concentration was raised to 1 mM, the behavior of ß-galactosidasewas similar to that obtained after 90 min of incubation (Fig.4a), and the µ_mean was also fitted with a hyperbolic function,while the ß-galactosidase activity per cell showedan exponential decay in correspondence with the beginning ofcell growth (Fig. 4b). Because of longer incubation time andconsequent cell proliferation, the K_AA constant, indicativeof the need for amino acids, increased from 14 to 222 µM.Since the Int_AA values were equivalent in the lag and exponentialphases, we assumed that this parameter was dependent only oninoculum history, confirming its reliability as a parameterfor measuring the internal amino acid supplies (Table 1).

View larger version (12K):
[in this window]
[in a new window]

FIG. 4. (a) ß-Galactosidase activity affected by amino acid concentration. Overnight cultures in LB (0.5 ml) were centrifuged and diluted to 50 to 300 cells per ml in 0.1 mM IPTG in PB, and then aliquots of 2.25 ml were incubated at 37°C in the presence of growing concentrations of amino acids. After 270 min, ß-galactosidase activity was measured as described in Materials and Methods. The experimental points ( ${blacksquare}$ ) are expressed as mean percentages ± SD (error bars) of five to seven tests. The curve was determined as described in the legend to Fig. 1. (b) ß-Galactosidase activity and mean growth rate affected by amino acid concentration. See the description for Fig. 4a for the procedure. ß-Galactosidase activity ( ${blacksquare}$ ) was expressed as enzymatic units per the final number of cells. Cell counts were performed as described in Materials and Methods. ß-Galactosidase activity versus amino acid concentration was fitted as an exponential decay of the third order via Microcal Origin version 5.0 software. Mean growth rate (µ_mean; ${blacklozenge}$ ) versus amino acid concentration was determined by nonlinear least-squares analysis via Microcal Origin version 5.0 software with the following hyperbole equation: y = µ_AA=0 + [(µ_max x x)/(K_m + x)]. Since the hyperbolic relationship was valid only for µ values of >0, µ_AA=0 is a fictitious value which was calculated only for regression purposes. For µ_mean values of $≤$ 0, see Fig. 4c. The experimental points ( ${blacksquare}$ and ${blacklozenge}$ ) are expressed as means ± SE (error bars) of five to seven tests. (c) ß-Galactosidase activity and mean growth rate affected by amino acid concentration. See the description for Fig. 4a for the procedure. The curve representing ß-galactosidase activity ( ${blacksquare}$ ) was determined by linear least-squares analysis. µ_mean ( ${blacklozenge}$ ) was calculated as described for Fig. 4b. The experimental points ( ${blacksquare}$ and ${blacklozenge}$ ) are expressed as means ± SE (error bars) of five to seven tests.

On a daily basis, both 90-min and 4.5-hour incubation testswere significantly repeatable, whereas we observed large day-to-dayvariations for V_ß_-galmax (±40%) and K_AA (±50%),determinations with both constants not being statistically correlatedwith each other, with the number of cells, or with externalclimate conditions. We believed that these discrepancies couldbe ascribed to the fact that, even under identical circumstances,every cell culture expressed a different and exclusive behavior.However, the high number of experiments performed (20 90-minexperiments and 7 4.5-hour experiments) allowed the calculationof mean values.

Variation in ß-galactosidase production caused by the absence of a single amino acid. We evaluated the levels of ß-galactosidase synthesisof different strains affected by a single amino acid deficiencyin a low-cell-density environment (Table 3). In E. coli ATCC25922, most amino acid deficiencies were not influential; however,in a few circumstances, the absence of a single amino acid accountedfor a major decrease, depending on the final amino acid poolconcentration. Up to an 80% loss in the ability of this strainto synthesize ß-galactosidase was caused by leucineor tryptophan deficiency. For instance, E. coli K-12 was extremelysensitive to single deficiencies of amino acids of the ilv pathway,while C. freundii was scarcely receptive to the need for a particularamino acid, except for leucine. Eventually, the absence of leucineyielded a more-than-50% reduction of ß-galactosidaseactivity and growth rate in all strains tested (E. coli K-12and ATCC 25922, C. freundii ATCC 8090, and K. pneumoniae ATCC13883).

View this table:
[in this window]
[in a new window]

TABLE 3. Effect of a missing amino acid on the synthesis of ß-galactosidase

Tests of cells in the lag phase evidenced that when the aminoacid concentration was further increased (>1 mM), ß-galactosidaseactivity approached closely the asymptotical maximum V_ß_-galmaxat first but then started to decrease (data not shown). Thistendency was amplified in the absence of leucine (Fig. 2), whenthe enzyme activity suffered a remarkable reduction after thepeak corresponding with an amino acid concentration of 10 µM.Finally, mixtures with an overall amino acid concentration of200 µM, including 1 to 20 amino acids (from 200 to 10µM each), were tested sequentially. As expected, ß-galactosidaseactivity increased progressively with the increase in the numberof different amino acids present in the solution (data not shown).

DISCUSSION

Top

Discussion

Three main findings emerged from the present study involvingnutritional downshifts in environments in which E. coli haslow cell density: (i) the recovery of ß-galactosidaseactivity and the concentration of the 20 natural amino acidswere hyperbolically related, and the shape of their relationwas not affected by the simultaneous presence of other carbon/nitrogensources; (ii) a 4.5-hour absence of nutrients induced a starvationbehavior characterized by high rates of cell survival but nomeasurable ß-galactosidase activity, whereas low concentrationsof amino acids (5 µM) were sufficient to promote a transitionto an adaptive mode, in which E. coli resumed enzyme synthesiswhile simultaneously enduring a population drop by between 60and 80 percent; and (iii) ß-galactosidase activityand growth rate were negatively affected by leucine deficiencyin all conditions tested.

ß-Galactosidase activity in environments populated at low cell densities after nutritional downshift. Earlier experiments carried out with high-density cultures provedthat when E. coli is transferred to a nutrient-limited medium,the overall synthesis of protein (38), including ß-galactosidase(39), is reduced. Kuroda et al. (23) substantiated the dependenceof ß-galactosidase activity on the presence of aminoacids after nutrient down-transition via Northern blot analysisat high cell densities and indicated translation as the criticalstep; the resulting ß-galactosidase activity was foundto be closely related to amino acid concentration despite thefact that the measured amount of IPTG-induced lacZ mRNA wasnot influenced whatsoever. We achieved similar results withlow cell densities by using an experimental procedure for quantitativelymonitoring ß-galactosidase activity, and additionally,we assessed the correlation between the enzyme activity andthe amino acid concentration. We observed a hyperbolic relationshipand defined the cell affinity for the amino acid pool underconditions for the maximum expression of scavenging capabilities,due to both glucose and ammonium shortage, via a Michaelianconstant (K_AA). In order to calculate ß-galactosidaseproduction, we replaced the Monod parameter µ_max, normallyexpressing maximum growth rate, with V_ß_-galmax, whichrepresented the hypothetical maximum achievable rate of ß-galactosidasesynthesis.

Other factors affected the kinetics of enzymatic production,with different outcomes for V_ß_-galmax-n and K_AA_-_n,as evidenced by multiple-nutrient experiments (Table 2). Asin high-density environments, catabolite repression stronglyreduced V_ß_-galmax-n, in view of the fact that ß-galactosidasewas repressed in cAMP-deficient glucose medium and the V_ß_-galmax-nof cells in MOPS was higher than that of cells in glycerol-supplementedMOPS. Since the addition of cAMP to the glucose medium leftK_AA_-_n unchanged while K_AA_-_n was unexpectedly increased in cAMP-supplementedmedia having amino acids as the only carbon source, the effectof cAMP on K_AA_-_n could not be significantly defined. On theother hand, K_AA_-_n was distinctly subject to cell dilution effects.The remarkable K_AA_-_n reductions in nutrient media with nitrogenand carbon sources in a balanced ratio and in glycerol-supplementedMOPS were ascribable to a shorter lag phase (data not shown),with an earlier resumption of cell growth and consequent faster,but not greater, ß-galactosidase production.

Downshift responses triggered by the presence of amino acids: starvation or hunger? After 4.5 h of incubation, ß-galactosidase activityand cell growth showed conflicting behaviors at amino acid concentrationsbetween 5 and 35 µM, since the enzyme activity grew linearlywhile the chances of bacterial survival were significantly morereduced during the whole interval than under conditions in whichnutrient was absent (Fig. 4c). To find an interpretation ofthese results, we investigated the cell responses before andafter this interval. In the absence of amino acids, ß-galactosidaseactivity was extremely low, cell growth ceased, and the numberof culturable cells endured only a 10% decrease. We deducedthat the cells switched to a maintenance metabolism characterizedby improved resistance to stress conditions and consistent withthe rpoS-mediated survival response (15, 17). On the other hand,cells growing at suboptimal dilution rates, like at amino acidconcentrations higher than 35 µM, express the "in betweenfeast and famine" hunger response (13), which is elicited bycarbon or nitrogen limitations (18) and characterized by improvedscavenging skills, due to the overexpression of transport andassimilation systems (12). In view of the fact that the activationof the rpoS-mediated survival response causes inhibition ofthese genes, hunger and starvation responses are in conflict(29). Since the expression of RpoS is inversely correlated withspecific growth rate (19) and, above all, attenuated by theinfluence of amino acids on the leucine responsive regulatoryprotein (Lrp) regulon (5), a central system of regulation offeast-to-famine transitions (35), we have associated this aminoacid-mediated cell growth to the exponential phase with therising expression of a hunger response.

In the transition from 0 to 5 µM, the lack of the starvationresponse for an adaptive response caused a fall in the survivalrate from 90% to between 20 and 40%. Since an effective hungerresponse involves specific growth rates from 0.1 to 0.9 h^–1(13), we deduced that the hunger response was not expressedwith homogeneous efficacy. Only cells fitted with advanced capabilitiesfor the transport of amino acids and their use as both carbon(26) and nitrogen sources (32) were able to survive in the absenceof ammonium and sugars without expressing a starvation stressresponse. Cell viability stabilized during the 5 µM-to-35µM interval, and the cells that survived synthesized ß-galactosidaselinearly with respect to the amino acid pool concentrations.These observations agree with the reported slow shut-off ofRpoS transcription in concurrence with the reactivation of lacZafter diauxie (14). We have argued whether, instead of havingsimply died, the nonculturable cells would have entered a viablebut not culturable state; however, we discarded this possibilitywhen, by using tests in the absence of nutrients as an experimentalmodel for the maximum degree of starvation achievable in suchtime at those conditions, no more than 10% reduction due tostarvation was observed in culturability. The minimum aminoacid nutritional limit for allowing full survival at those celldensities was thus identified to be between 40 and 50 µM.

Amino acid diversity as crucial as amino acid quantity: the importance of leucine. Another aspect that was investigated concerned the importanceof the single amino acids. Despite the fact that the same levelof nutrient concentration was maintained, the ß-galactosidaserate decreased dramatically when a few components were not present,confirming that amino acid diversity played a critical role.Although most of the reductions in ß-galactosidaseactivity, due to single or multiple deficiencies, were ascribedto the logical effort of the cell to biosynthesize the requiredbuilding blocks (4), some amino acids produced distinctive strain-dependentinhibition of E. coli, like at higher cell densities (33). Weverified that a single isoleucine deficiency inhibited the synthesisof ß-galactosidase in K-12 (Fig. 2), since the presenceof valine impairs the expression of genes responsible for thesynthesis of enzymes of the common biosynthetic pathway, inducinga stringent response due to amino acid starvation (8).

Leucine was the only amino acid significantly affecting enzymesynthesis and growth in all strains tested. Interestingly, thesereductions of ß-galactosidase activity and dilutionrate in leucine-deficient environments agree favorably withobservations reported in literature covering leucine-dependenteffects on the Lrp (3), which is activated in minimal medium,due to carbon or ammonia limitation (28). Above all, leucinewas shown to counteract the negative regulation of Lrp on alanine,serine, and threonine catabolism (25, 37) and neutralize the70% reduction in ß-galactosidase synthesis measuredin lrp mutants (36). Our results support the assumption thatE. coli interprets leucine as an indication of nutrient availability,and thus, in the presence of leucine, the starvation responseis slowed down (10).

In summary, we developed a novel sensitive tool for calculatingß-galactosidase activity, which could also be extensivelyused for regular analyses of gene expression based on geneticfusions with the lac operon. By means of this method, we investigatedlow-cell-density environments and recognized effects due tocatabolite repression and stringent control, in agreement withwhat is common knowledge regarding environments with high bacterialdensity. While doing so, we observed that the presence of lowconcentrations of amino acids was actually sufficient for E.coli to switch from survival to adaptation when moving to adverseenvironments.

Cells subject to the above-mentioned feast-to-famine transitionsfind themselves in a nutrient-deprived regimen, similar to thatcharacterizing cells in stationary phase after nutrient exhaustion.For this reason, our results, which pointed out the key roleplayed by amino acids, particularly of leucine, in the cellresponses to nutrient disadvantage, are also consistent withthe findings that cells in stationary phase tend to escape starvationby optimizing tools for amino acid scavenging and catabolism(41) via a genetic development of lrp mutants (40), which exhibita growth advantage in stationary phase phenotype (11).

Future research will address the investigation of longer-termlow-cell-density nutritional shortages in the presence or absenceof amino acids with the purpose of exploring the borders betweenstarvation and hunger.

ACKNOWLEDGMENTS

This work was supported by the Italian Ministry of Universityand Research (MIUR), via L.488/92, grant N. S634-P, and alsoby the Italian private foundation Fondazione Cassa di Risparmiodi Terni e Narni.

We thank Paolo Nannipieri for the unrivalled support in therevision process, Mark Falvo for the English revision and CentroServizi Bibliotecari, Universitá di Perugia, SezioneTerni for providing references.

FOOTNOTES

* Corresponding author. Mailing address: ISRIM Scarl, Località Pentima Bassa, 21, 05100 Terni, Italy. Phone: 0039-0744547882. Fax: 0039-0744470174. E-mail: s.bodini{at}isrim.it

Published ahead of print on 9 February 2007.

Present address: Division of Cell and Molecular Biology, Immunologyand Infection, Imperial College, SAF Building, SW7 2AZ ImperialCollege Road, London, United Kingdom.

REFERENCES

Top