Limitation of phosphate assimilation maintains cytoplasmic magnesium homeostasis

Significance

Phosphorus (P) is essential for life. As the fifth-most-abundant element in living cells, P is required for the synthesis of an array of biological molecules including (d)NTPs, nucleic acids, and membranes. Organisms typically acquire environmental P as inorganic phosphate (Pi). While essential for growth and viability, excess intracellular Pi is toxic for both bacteria and eukaryotes. Using the bacterium Salmonella enterica serovar Typhimurium as a model, we establish that Pi cytotoxicity is manifested following its assimilation into adenosine triphosphate (ATP), which acts as a chelating agent for Mg²⁺ and other cations. Our findings identify physiological processes disrupted by excessive Pi and how bacteria tune P assimilation to cytoplasmic Mg²⁺ levels.

Abstract

Phosphorus (P) is an essential component of core biological molecules. In bacteria, P is acquired mainly as inorganic orthophosphate (Pi) and assimilated into adenosine triphosphate (ATP) in the cytoplasm. Although P is essential, excess cytosolic Pi hinders growth. We now report that bacteria limit Pi uptake to avoid disruption of Mg²⁺-dependent processes that result, in part, from Mg²⁺ chelation by ATP. We establish that the MgtC protein inhibits uptake of the ATP precursor Pi when Salmonella enterica serovar Typhimurium experiences cytoplasmic Mg²⁺ starvation. This response prevents ATP accumulation and overproduction of ribosomal RNA that together ultimately hinder bacterial growth and result in loss of viability. Even when cytoplasmic Mg²⁺ is not limiting, excessive Pi uptake increases ATP synthesis, depletes free cytoplasmic Mg²⁺, inhibits protein synthesis, and hinders growth. Our results provide a framework to understand the molecular basis for Pi toxicity. Furthermore, they suggest a regulatory logic that governs P assimilation based on its intimate connection to cytoplasmic Mg²⁺ homeostasis.

As a component of a large number of biological molecules, phosphorus (P) is required for key biological functions. These include the formation of cellular boundaries, the storage and transfer of chemical energy, the integration and propagation of information in signal transduction pathways, and the storage, transmission, and expression of genetic information. Bacteria take up P mainly as inorganic phosphate (PO₄⁻³; Pi), which is then assimilated in the cytoplasm via its incorporation into adenosine triphosphate (ATP). ATP functions as the main cellular P-carrier molecule, mediating both the transfer of Pi among biological molecules and the release of chemical energy to power energy-dependent processes (1). Cells must tightly regulate Pi acquisition and utilization because P assimilation is essential but excessive cytoplasmic Pi is toxic (2–9). Here, we elucidate how bacteria coordinate Pi acquisition and consumption to prevent the deleterious effects of unbalanced Pi metabolism.

Following assimilation, the negative charges from Pi groups in biomolecules are neutralized by positively charged ionic species present in the cytoplasm. As such, the majority of cytoplasmic ATP exists as a salt with positively charged magnesium (Mg²⁺), the most abundant divalent cation in living cells. Therefore, the ATP:Mg²⁺ salt, rather than the ATP anion, is the substrate for most ATP-dependent enzymatic reactions (10, 11). In enteric bacteria, ATP stimulates transcription of ribosomal RNA (rRNA) genes (12, 13). The synthesis and activity of ribosomes consume the majority of the ATP in rapidly growing cells (14). During ribosome biogenesis, the negative charges from Pi groups in the rRNA backbone chelate large amounts of Mg²⁺ ions (14, 15). This process reduces electrostatic repulsion among Pi groups in the rRNA backbone, enabling the folding and assembly of functional ribosomes (15, 16). Thus, ATP and rRNA constitute the largest cytoplasmic reservoirs of Pi and Mg²⁺ (12, 14–20). Given this inherent connection between Pi and Mg²⁺, we wondered if the cytotoxic effects of excessive Pi uptake result from Pi assimilation into ATP and subsequent disruption of Mg²⁺ dependent processes.

In the gram-negative bacterium Salmonella enterica serovar Typhimurium (Salmonella), prolonged growth in limiting Mg²⁺ induces a Mg²⁺ starvation response mediated by the MgtA, MgtB and MgtC proteins (21–24). MgtA and MgtB are high-affinity, ATP-dependent Mg²⁺ importers that increase the Mg²⁺ concentration in the cytoplasm (25, 26). MgtC decreases intracellular ATP levels (27), thereby reducing rRNA synthesis, which lowers the steady-state amounts of ribosomes and slows down the translation rate (28). This response reduces the concentration of assimilated P and, consequently, the quantity of Mg²⁺ required as a counter ion, allowing the remaining cytoplasmic Mg²⁺ to stabilize existing ribosomes and to maintain other vital cellular processes that exhibit a strict dependence on Mg²⁺ (21, 28, 29).

MgtC promotes Salmonella survival inside mammalian macrophages and the establishment of systemic infections (30, 31). In macrophages, MgtC inhibits the activity of Salmonella’s own F₁F_o ATP synthase, the enzyme catalyzing ATP production via oxidative phosphorylation (32, 33). MgtC appears to operate by distinct mechanisms inside macrophages versus low-Mg²⁺ media because an MgtC variant has been identified that fails to promote intramacrophage survival while enabling growth and viability under cytoplasmic Mg²⁺ starvation (34). In this paper, we reveal that Pi uptake must be tightly controlled to maintain cytoplasmic Mg²⁺ homeostasis. We establish that during cytoplasmic Mg²⁺ starvation, MgtC lowers ATP levels by inhibiting Pi uptake, thus limiting an ATP precursor instead of interfering with its enzymatic generation (Fig. 1). MgtC hinders the activity of the main Pi uptake system in Salmonella and likely other bacterial species. Counterintuitively, limiting exogenous Pi availability rescues translation, promotes growth, and restores viability to an mgtC null mutant. Excess Pi is toxic even at physiological concentrations of cytoplasmic Mg²⁺ due to its incorporation into ATP and subsequent disruption of Mg²⁺-dependent processes. Our findings provide a conceptual framework to understand the underlying basis of Pi cytotoxicity observed in bacteria and eukaryotes (3–9, 35–38). Moreover, we uncover the regulatory logic by which cells control of P assimilation.

Fig. 1.

Model depicting how Salmonella enterica prevents Pi cytotoxicity. (A) Excessive Pi is imported into the cytoplasm via an inner membrane transport system (denoted by an X). Pi is assimilated by the cells through reactions that synthesize ATP (R₁, R₂, R₃, R_n). Excessive ATP chelates free cytoplasmic Mg²⁺, thereby inducing cytoplasmic Mg²⁺ starvation. (B) Cells express the MgtC membrane protein to inhibit Pi uptake via X. The reduction in the availability of cytoplasmic Pi simultaneously slows all ATP-generating reactions in the cell, freeing Mg²⁺ ions and restoring homeostasis.

Results

The MgtC Protein Limits ATP Accumulation during Low Cytoplasmic Mg²⁺ Stress by an F₁F_o Synthase-Independent Mechanism.

ATP exists as a Mg²⁺ salt in living cells (11, 14). Therefore, when cells face limiting cytoplasmic Mg²⁺ concentrations, they reduce ATP amounts to free Mg²⁺ ions so they can be used for other cellular processes, such as ribosome assembly (28, 29). In macrophages, MgtC inhibits Salmonella’s own F₁F_o ATP synthase, the enzyme responsible for ATP synthesis via oxidative phosphorylation (32). Notably, an mgtC atpB double mutant—lacking both MgtC and the AtpB subunit of the ATP synthase—was reported to harbor similar ATP amounts than an atpB single mutant under cytoplasmic Mg²⁺ starvation conditions (32). This result led to the notion that MgtC prevents a nonphysiological rise in ATP levels by inhibiting the F₁F_o ATP synthase also under cytoplasmic Mg²⁺ starvation (32, 39, 40). Below, we present the result of experiments that reexamine the interpretation of the aforementioned results.

The F₁F_o ATP synthase uses the proton motive force generated by the respiratory electron transport chain to synthesize ATP from adenosine diphosphate and Pi (41, 42). Consequently, an mgtC atpB double mutant (or any other strain lacking a functional ATP synthase) relies exclusively on fermentative pathways to produce ATP via substrate-level phosphorylation (43). However, the low ATP amounts observed in the mgtC atpB double mutant were obtained during growth on medium containing glycerol as the carbon source (32). Given that the glycerol fermentation is extremely inefficient (44, 45), we reasoned that the low ATP amounts in the mgtC atpB double mutant could simply reflect an inability to efficiently ferment glycerol.

To test this notion, we compared ATP amounts in wild-type, mgtC, atpB, and mgtC atpB isogenic strains grown in minimal medium containing readily fermentable glucose as the carbon source and low (10 µM) Mg²⁺ to induce cytoplasmic Mg²⁺ starvation (22, 23). As hypothesized, mgtC and mgtC atpB strains had 42- and 29-fold higher ATP amounts relative to their mgtC⁺ isogenic counterparts, respectively (Fig. 2A), after 5 h of growth, when cytoplasmic Mg²⁺ becomes limiting (28). Hence, atpB inactivation does not abrogate the intracellular ATP accumulation resulting from mgtC inactivation provided that bacteria are fed glucose as carbon source.

Fig. 2.

F₁F_o synthase-independent ATP accumulation in an mgtC mutant during cytoplasmic Mg²⁺ starvation. (A) Intracellular ATP levels in wild-type (14028s), mgtC (EL4), atpB (MP24), or mgtC atpB (MP25) Salmonella following 5 h of growth. Cells were grown in MOPS medium containing 10 μM MgCl₂ and 2 mM K₂HPO₄. (B) Schematic representation of carbon flow through the central metabolic pathways in bacteria with key extracellular (blue) and intracellular (red) metabolites. G6P, glucose-6-phosphate; 6P-Gluc, 6-Phosphogluconate; Gly, glycerol; Pyr, pyruvate; PPP, pentose phosphate pathway; ED, Entner–Doudoroff pathway. (C) Intracellular ATP levels in wild-type (14028s), mgtC (EL4), atpB (MP24), or mgtC atpB (MP25) Salmonella. Cells were grown in MOPS medium containing 10 mM MgCl₂, 2 mM K₂HPO₄, and 25 mM glucose until OD₆₀₀ ∼0.4, washed thrice, and grown for additional 90 min in MOPS medium supplemented with the indicated carbon source (25 mM glucose, 25 mM sodium gluconate, 30 mM l-arabinose, 50 mM sodium pyruvate, or 50 mM glycerol) and containing 2 mM K₂HPO₄ and either 0 (−) or 10 (+) mM MgCl₂. Means ± SDs of three independent experiments are shown. *P < 0.05, **P < 0.01, ****P < 0.0001; ns, no significant difference. (A) Two-tailed t test; (C) two-way ANOVA with Tukey’s correction.

To further test our hypothesis, we compared ATP amounts among the four strains at 90 min following a nutritional downshift from medium containing high (10 mM) Mg²⁺ and glucose, to media lacking Mg²⁺ and containing one of various carbon sources that are metabolized via distinct pathways (Fig. 2B). During cytoplasmic Mg²⁺ starvation, the mgtC strain had higher ATP amounts than the wild-type strain regardless of the carbon source (Fig. 2C). By contrast, the mgtC atpB strain had higher ATP amounts than the atpB single mutant during growth on readily fermentable carbon sources (glucose, gluconate, arabinose, or pyruvate) but was unable to do so during growth on inefficiently fermentable glycerol (Fig. 2 B and C). As a control, all the strains tested displayed similar ATP amounts when resuspended in high Mg²⁺ glucose-containing medium (Fig. 2C). Taken together, these results indicate that the low ATP levels previously observed in an mgtC atpB strain experiencing cytoplasmic Mg²⁺ starvation (32) are caused by an inability to efficiently ferment glycerol as opposed to an inability to inhibit the F₁F_o ATP synthase. Furthermore, these results indicate that MgtC inhibits multiple ATP-generating reactions in the cell, not just the one that is carried out by the F₁F_o complex.

MgtC Inhibits Pi Acquisition Thereby Limiting ATP Synthesis during Cytoplasmic Mg²⁺ Starvation.

Cellular ATP can be synthesized by several catabolic reactions (46–48). How, then, can MgtC control the activity of multiple ATP-producing enzymatic reactions? We reasoned that regardless of the identity of the enzymes catalyzing ATP formation, the overall rate of ATP synthesis in the cell could be restricted by the availability of substrates. MgtC could, therefore, function by inhibiting the synthesis or acquisition of an ATP precursor. Interestingly, at the onset of cytoplasmic Mg²⁺ starvation, a temporary shortage in the concentration of free cytoplasmic Mg²⁺ destabilizes the bacterial ribosomal subunits (28). The resulting decrease in translation efficiency reduces ATP consumption and, consequently, the recycling of Pi from ATP. This lowers the concentration of cytoplasmic Pi, transiently activating the cytoplasmic Pi-starvation sensing two-component system composed of the PhoB and PhoR proteins (PhoB/PhoR) (17). Three lines of evidence led us to hypothesize that MgtC prevents ATP synthesis by limiting Pi influx into the cell. First, PhoB/PhoR activation is hampered in an mgtC mutant (17), indicating that this strain experiences excess cytoplasmic Pi. Second, mgtC and mgtC atpB strains have higher intracellular steady-state Pi amounts than wild-type and atpB Salmonella strains (Fig. 3A), respectively. Thus, MgtC prevents Pi accumulation in an atpB-independent fashion. And third, when Mg²⁺ and Pi are abundant, heterologous expression of the mgtC gene activates PhoB/PhoR, further suggesting that MgtC causes a shortage in cytoplasmic Pi (SI Appendix, Fig. S1) (17).

Fig. 3.

MgtC-dependent inhibition of Pi transport and assimilation into ATP during cytoplasmic Mg²⁺ starvation. (A) Total intracellular Pi in wild-type (14028s), mgtC (EL4), atpB (MP24), or mgtC atpB (MP25) Salmonella following 5 h of growth in MOPS medium containing 10 μM MgCl₂ and 2 mM K₂HPO₄. (B) Relative radioactive orthophosphate (³²Pi) uptake in wild-type (14028s) or mgtC (EL4) cells. Bacteria were grown in MOPS medium containing 10 μM MgCl₂ and 500 μM K₂HPO₄ during 3 h before the addition of ³²Pi to the cultures. Levels of ³²Pi accumulated in cells were determined after 30 min of labeling by liquid scintillation counting, as described in Materials and Methods. ³²Pi uptake values were normalized against the wild-type strain. (C) Intracellular ATP levels in wild-type (14028s) or mgtC (EL4) Salmonella following 5 h of growth in MOPS medium containing 10 μM MgCl₂ and the indicated concentration of K₂HPO₄ (μM Pi). Means ± SDs of at least three independent experiments are shown. **P < 0.01, ****P < 0.0001; ns, no significant difference. (A–C) Two-tailed t tests.

To test our hypothesis, we measured transport of radiolabeled Pi (³²Pi) following MgtC expression from its native chromosomal location in response to cytoplasmic Mg²⁺ starvation. We established that mgtC cells accumulated four times more radioactivity than the isogenic wild-type strain after 30 min of growth in the presence of ³²Pi (Fig. 3B). This result suggests that the increased steady-state intracellular Pi levels observed for mgtC and mgtC atpB mutants (Fig. 3A) arises from increased uptake of extracellular Pi rather than to increased Pi release from intracellular sources. Note that all ³²Pi transport assays were performed in the presence of cold Pi, at a molar ratio of 1:25 (³²Pi:Pi), to prevent expression of the PstSCAB Pi transporter resulting from a lack of Pi in the growth medium (1). Therefore, this assay underestimates Pi influx into cells (see Discussion and Materials and Methods).

If MgtC restricts Pi import to the cytoplasm and Pi is required for ATP synthesis, a decrease in Pi availability in the growth medium should correct the high ATP amounts of the mgtC mutant. To test this prediction, we measured ATP amounts in wild-type and mgtC cells grown in minimal media with low Mg²⁺ and decreasing concentrations of exogenous Pi. Note that bacteria are able to grow in medium lacking exogenous Pi due to the residual Pi present in the mixture of casamino acids supplemented to the culture medium (see Materials and Methods). Strikingly, a decrease in exogenous Pi from 500 to 0 µM lowered ATP amounts in the mgtC mutant to wild-type levels (Fig. 3C). By contrast, ATP amounts remained invariably low in the wild-type strain (Fig. 3C). Importantly, in the absence of exogenous Pi, wild-type and mgtC strains had similar ATP amounts (Fig. 3C). Taken together, these results indicate that MgtC controls ATP synthesis by limiting Pi uptake.

MgtC Inhibits a Noncanonical Pi Transport System.

Salmonella encodes two bona fide Pi import systems: pitA and pstSCAB (Fig. 4A). While PitA functions as a metal:phosphate (M:Pi)/proton symporter (1, 49–51), PstSCAB is a high affinity, ATP-dependent Pi transporter (Fig. 4A) (1, 8, 51, 52). In addition to pitA and pstSCAB, Salmonella also harbors an yjbB homolog, which encodes a sodium/phosphate symporter. Whereas YjbB promotes Pi export (53), we made the cautious assumption that YjbB may also import Pi (Fig. 4A).

Fig. 4.

A noncanonical Pi transport system is inhibited by MgtC. (A) Schematic representation of inorganic Pi transporters harbored by Salmonella enterica and Escherichia coli. Note that PitB (gray, dashed outline) is absent from Salmonella. M:Pi, metal-phosphate complex; X, inferred, uncharacterized Pi transporter inhibited by MgtC. (B) Relative ³²Pi uptake in wild-type (14028s) or ∆3Pi (RB39) carrying either pVector (pUHE-21) or pMgtC (pUHE-MgtC). Bacteria were grown in MOPS medium containing 250 μM MgCl₂ and 500 μM K₂HPO₄ until OD₆₀₀ ∼0.2. Cultures were then propagated for 15 min in the presence of 750 μM IPTG prior to the addition of ³²Pi. Transport of ³²Pi was allowed to take place for 20 min. Intracellular ³²Pi accumulation was determined by liquid scintillation counting, as described in Materials and Methods. ³²Pi uptake values were normalized against the wild-type pVector strain. **P < 0.01, ***P < 0.001, unpaired two-tailed t test. (C) Growth curve of wild-type (14028s) or ∆3Pi (RB39) Salmonella. Cells were grown in MOPS medium containing 10 mM MgCl₂ and either 0 (−Pi) or 500 (+Pi) μM of K₂HPO₄. (D) Fluorescence from wild-type (14028s), pitA (MP1251), yjbB (MP1252), and pitA yjbB (MP1479p) Salmonella carrying pPpstS-GFPc and either pVector or pMgtC. Cells were grown in MOPS medium containing 250 μM MgCl₂ and 500 μM K₂HPO₄. A total of 250 μM of IPTG were added after 2 h of growth. (E) Fluorescence from wild-type (14028s), pitA (MP1251), yjbB (MP1252), pitA yjbB (MP1479p), and mgtC (EL4) Salmonella carrying pPpstS-GFPc or pVector (the promoterless GFP plasmid pGFPc). Cells were grown in MOPS medium containing 10 μM MgCl₂ and 500 μM K₂HPO₄. (F) Intracellular ATP levels in wild-type (14028s), mgtC (EL4), mgtC pitA (MP1254), mgtC pstSCAB (MP1720), mgtC yjbB (MP1255), or mgtC ∆3Pi (RB43) Salmonella. Cells were grown in MOPS medium containing 10 mM Mg and 2 mM 2-AP OD₆₀₀ ∼0.4. Cells were subsequently washed thrice with MOPS medium lacking MgCl₂ and any P source and grown for additional 90 min in MOPS medium lacking Mg²⁺ and containing 2 mM K₂HPO₄. ***P < 0.001, unpaired two-tailed t tests against the wild-type strain. For all graphs (B–F), means ± SDs of at least three independent experiments are shown.

To test MgtC’s role on inhibition of PitA, PstSCAB, or YjbB, we measured ³²Pi uptake following ectopic MgtC expression in wild-type and isogenic pitA pstSCAB yjbB triple mutant (3∆Pi) Salmonella. ³²Pi uptake was slightly (17%) lower in wild-type Salmonella expressing MgtC than in those with the vector control (Fig. 4B). By contrast, MgtC reduced Pi uptake by 70% in the 3∆Pi background relative to the vector control (Fig. 4B). Because MgtC decreases Pi influx in a pitA-, pstSCAB-, and yjbB- independent manner, we reasoned that there must be an additional Pi influx system. In support of this notion, the 3∆Pi mutant grew using Pi as the sole P source (Fig. 4C), albeit with a longer lag and to a lower final optical density unit (OD₆₀₀) than the wild-type strain (Fig. 4C).

Two additional lines of evidence indicate that MgtC inhibits a noncanonical Pi transporter. First, by decreasing cytoplasmic Pi amounts (Figs. 3 and 4B), heterologous MgtC expression elicits a dose-dependent activation of PhoB/PhoR that results in transcription of the PhoB-activated pstSCAB operon (SI Appendix, Fig. S1) (17). Hence, removal of the Pi transporter targeted by MgtC should result in MgtC-irresponsive constitutively high pstS expression because MgtC would no longer impact PhoB/PhoR activation. We determined that MgtC expression (either heterologously in medium containing 250 µM Mg²⁺ or natively in medium containing 10 µM Mg²⁺) induced pstS transcription in pitA, yjbB, and pitA yjbB mutants (Fig. 4 D and E). We did not examine the pstSCAB mutant because the PstSCAB transporter participates in the regulation of the PhoB/PhoR two-component system through a physical interaction (54). Consequently, inactivation of the transporter through mutations in pst genes leads to PhoB/PhoR hyperactivation, effectively preventing signal transduction (1, 3, 54, 55). In addition, we reasoned that it would be highly unlikely for cells to harbor a regulatory circuit whereby MgtC inhibition of PstSCAB transporter activity would promote PstSCAB expression (SI Appendix, Fig. S1) (17), requiring more MgtC protein, ad infinitum.

And second, deleting the MgtC-inhibited Pi transporter should correct the exacerbated ATP accumulation of the mgtC mutant (Fig. 3C). However, mutations in either pitA, pstSCAB, or yjbB or a combined deletion of all these genes did not lower ATP amounts in the mgtC mutant (Fig. 4F). Note that due to the longer lag time of 3ΔPi mutant strains growing with Pi as the sole P source (Fig. 4C), we cultured all tested strains in minimal medium containing 2-aminoethylphosphonic acid [2-AP] as the P source and subsequently subjected them to a nutritional downshift to medium supplemented with 2 mM Pi and lacking Mg²⁺ for 90 min prior to ATP quantification. Because 2-AP is imported into the cytoplasm through an organic P system (PhnSTUV) prior to removal of Pi, this allows cells to bypass the need for inorganic P source and grow at the same rate. In sum, these results indicate that MgtC inhibits Pi uptake by targeting an unidentified Pi transporter.

Phosphate Limitation Rescues the Translation and Growth Defects of an mgtC Mutant Experiencing Cytoplasmic Mg²⁺ Starvation.

An mgtC mutant experiencing cytoplasmic Mg²⁺ starvation exhibits ribosomal assembly defects, inefficient translation, growth arrest, and a loss of viability (21, 28). This is presumably due to Mg²⁺ chelation by excess ATP molecules because the mgtC mutant phenotypes are corrected by promoting ATP hydrolysis (28, 29). Given that MgtC lowers ATP synthesis by inhibiting Pi acquisition (Figs. 3 A and C and 4B), these phenotypes should also be rescued by limiting Pi access.

As predicted, wild-type and mgtC strains showed similar translation rates in the absence of exogenous Pi (Fig. 5 A and B). By contrast, during growth in 500 µM Pi, the translation rate of the mgtC mutant was 4.7-fold lower than that of the wild-type strain (Fig. 5 A and B). Moreover, decreasing exogenous Pi in the growth medium from 500 to 0 µM progressively increased growth of the mgtC mutant (Fig. 5C). Remarkably, in the absence of exogenous Pi, the mgtC mutant displayed growth yield and kinetics indistinguishable from that of the wild-type strain (Fig. 5C). As expected, growth of wild-type and mgtC strains was indistinguishable in medium containing 500 µM Pi when the Mg²⁺ concentration was raised from 10 to 500 µM (Fig. 5C), thereby preventing Mg²⁺ starvation. Furthermore, the loss of viability of the mgtC mutant (21) in medium containing 500 µM Pi was suppressed by not adding Pi (Fig. 5D). The mutant maintained the same number of colony-forming units (CFU) per OD₆₀₀ as the wild-type strain. Altogether, these results establish that MgtC prevents the toxic effects of excessive Pi uptake.

Fig. 5.

Effect of phosphate limitation on the translation rate, growth, and viability of an mgtC mutant during cytosolic Mg²⁺ starvation. (A) Quantification and (B) Sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) analysis of the rate of protein synthesis (l-azidohomoalanine [AHA] labeling) of wild-type or mgtC (EL4) Salmonella. Cells were grown in MOPS medium containing 10 mM MgCl₂ and 2 mM K₂HPO₄ until OD₆₀₀ ∼0.4. Cells were subsequently washed thrice with MOPS medium lacking MgCl₂, K₂HPO₄, and amino acids and grown for additional 90 min in MOPS medium lacking methionine and containing 10 μM MgCl₂ plus the indicated concentration of K₂HPO₄ (μM Pi). Means ± SDs of four independent experiments are shown. **P < 0.01, unpaired two-tailed t test. ns, no significant difference. The gel is representative of four independent experiments. Samples from 0 and 500 μM Pi were resolved and imaged from different gels (indicated by dashed lines). (C) Growth curve of wild-type (14028s) or mgtC (EL4) Salmonella. Cells were grown in MOPS medium containing the indicated concentrations of MgCl₂ and K₂HPO₄. Means ± SDs of three independent experiments are shown. (D) Viable cell count of wild-type (14028s) or mgtC (EL4) Salmonella following 16 h of growth in MOPS medium containing 10 μM MgCl₂ and 0 or 500 μM K₂HPO₄. Cell suspensions were normalized to the same OD₆₀₀ and diluted, and 5 μL was spotted on plates. Images were taken after incubation of plates at 37 °C for 18 h and are representative of three independent experiments.

Excess Pi Is Toxic Even under High Mg²⁺ Conditions.

Excessive Pi uptake hinders bacterial growth even when Mg²⁺ is not limiting and cells are not anticipated to experience cytoplasmic Mg²⁺ starvation (3, 9). For instance, mutations that increase PstSCAB activity or expression cause heightened Pi uptake and growth inhibition in Escherichia coli (1–3, 5). The data presented in the previous sections suggest that the growth inhibition resulting from PstSCAB overexpression results from excessive Pi incorporation into ATP and subsequent chelation of Mg²⁺, that, in turn, give rise to elevated ATP levels, inhibition of translation and growth, and induction of MgtC expression (21, 27, 28). Because these phenotypes are observed during cytoplasmic Mg²⁺ starvation, they may be corrected by increasing the availability of free cytoplasmic Mg²⁺ (e.g., through the provision of excess Mg²⁺ in the growth medium) or the reversal of Pi assimilation (e.g., through enzymatic hydrolysis of ATP) (21, 28, 29).

First, we determined the phenotypic consequences of ectopic PstSCAB expression in medium containing high (10 mM) Pi and intermediate (0.1 mM) Mg²⁺ concentrations, two conditions that do not typically result in cytoplasmic Mg²⁺ starvation (22, 23) and, thus, MgtC expression (SI Appendix, Fig. S2). We established that ATP amounts were four times higher in PstSCAB-expressing bacteria than in control strains harboring either an empty vector or a plasmid expressing the inner membrane protein PmrB (Fig. 6A). Notably, the PstSCAB-dependent increase in ATP amounts is accompanied by reductions in growth rate and growth yield, phenotypes not observed in the control strains (Fig. 6B).

Fig. 6.

Effect of PstSCAB on growth, P assimilation, and Mg²⁺ homeostasis. (A) Intracellular ATP levels of cultures depicted in B and C. Measurements were conducted at 5 h of growth. (B, C) Growth curves of wild-type (14028s) Salmonella carrying pVector (pUHE-21), pPstSCAB (pUHE-PstSCAB), or pPmrB (pUHE-PmrB) in MOPS medium containing 0.1 (B) and 10 mM MgCl₂ (C). (D) Growth curve of wild-type (14028s) Salmonella carrying either pVector or pPstSCAB and either pVector2 (pBbB2K-GFP) or pATPase (pBbB2K-AtpAGD). In all experiments, cells were grown in MOPS medium containing 10 mM K₂HPO₄ and the indicated MgCl₂ concentration. Ectopic protein expressions were induced by adding 250 μM IPTG to the cultures after 2 h of growth. Means ± SDs of three independent experiments are shown. (A, B) ***P < 0.001, ****P < 0.0001, unpaired two-tailed t test against pVector.

And second, the growth defect resulting from PstSCAB expression is caused by an increase in ATP amounts resulting in chelation of free Mg²⁺. This is because PstSCAB expression caused only a minor reduction in growth rate, enabling cells to reach the same growth yields as the control strains when the Mg²⁺ concentration was increased 100-fold (to 10 mM) (Fig. 6C). Importantly, Mg²⁺ rescued growth of the PstSCAB-expressing strain even though ATP amounts remained fourfold higher than those of the control strains harboring the empty vector or the PmrB-expressing plasmid (Fig. 6A). In addition, a plasmid-encoded soluble subunit of the ATPase (17) corrected the growth defect caused by PstSCAB expression (Fig. 6D) whereas the vector control did not (Fig. 6D). Taken together, these results indicate that excess Pi imported into the cytoplasm is rapidly assimilated into ATP, causing a decrease in the levels of free cytoplasmic Mg²⁺ and inhibiting growth.

Excessive Pi Uptake Impairs Translation and Promotes MgtC Expression during Growth in High Mg²⁺.

If excessive Pi uptake leads to physiological conditions resembling cytoplasmic Mg²⁺ starvation (Fig. 6 A–D) (21, 28, 29), then increased Pi uptake resulting from PstSCAB expression should also decrease translation rates. Furthermore, because a reduction in translation rates caused by cytoplasmic Mg²⁺ starvation promotes mgtC transcription (22, 23, 27, 28, 56), excessive Pi uptake should also promote MgtC expression. We tested these predictions by measuring temporal fluorescence changes in wild-type strains harboring both a plasmid-borne gfp transcriptional fusion to the promoter and leader regions of mgtC, and either the empty vector, the pPmrB plasmid, or the pPstSCAB plasmid.

Fluorescence from mgtC-gfp increased 1 h following PstSCAB induction (Fig. 7A). This increase was absent or delayed in the control strains harboring the empty vector or expressing PmrB (Fig. 7A). Note that during growth in 25 µM Mg²⁺, the control strains display a minor increase in fluorescence after 4 h of growth. This small increase in fluorescence results from late onset of cytoplasmic Mg²⁺, which is delayed when compared to cells grown at 10 µM Mg²⁺ (SI Appendix, Fig. S2) (17, 22, 23, 57). The effect of PstSCAB expression on the time and fluorescence levels of the mgtC-gfp strain was inversely related to the availability of Mg²⁺ in the growth medium. That is, fluorescence resulting from PstSCAB expression in cultures grown in 25 µM Mg²⁺ were higher than those grown in 50 µM Mg²⁺, which, in turn, were higher than those grown in 0.1 mM Mg²⁺ (Fig. 7A). Bacteria grown in 0.1 mM Mg²⁺ displayed a relatively mild increase in mgtC-gfp activity 1 h after PstSCAB expression, and their fluorescence levels rose rapidly at 7.5 h post induction (Fig. 7A). The delayed mgtC induction likely reflects the time at which bacteria no longer neutralize the excess of intracellular Pi incorporated into ATP because they exhausted the Mg²⁺ available in the growth medium and, consequently, are unable to efficiently stabilize their ribosomes (28). We conclude that increased ATP production resulting from excessive Pi assimilation disrupts translation by sequestering free Mg²⁺ ions.

Fig. 7.

Effect of PstSCAB expression on mgtC transcription and cellular translation rates during growth under conditions of moderate and high Mg²⁺. (A) Fluorescence from wild-type (14028s) Salmonella carrying transcriptional reporter pPmgtC-GFPc and either pVector (pUHE-21), pPstSCAB (pUHE-PstSCAB), or pPmrB (pUHE-PmrB). Full time course (Left) and inset between 2 and 7 h (Right) of the experiments are shown. (B) Quantification and (C) SDS-PAGE analysis of the rate of protein synthesis (l-azidohomoalanine [AHA] labeling) of wild-type (14028s) Salmonella carrying either pVector, pPmrB, or pPstSCAB. In all experiments, cells were grown in MOPS medium containing 10 mM K₂HPO₄ and the indicated concentration of MgCl₂. For AHA labeling, bacteria were cultured in MOPS medium lacking methionine (see Materials and Methods). A total of 250 μM IPTG was added to the cultures following 2 h of growth. AHA was incorporated to the cultures at 4.5 h. Means ± SDs of three independent experiments are shown. Gels are representative of three independent experiments. ***P < 0.001, unpaired two-tailed t test against pVector.

To determine whether excessive Pi uptake compromises ribosome activity, we measured translation rates in the same strain set. During growth in medium containing 25 µM Mg²⁺, PstSCAB expression caused a twofold reduction in translation rates relative to control strains (Fig. 7 B and C). Notably, PstSCAB expression also caused a relative reduction in translation rates during growth in medium containing a 400-fold excess (10 mM) of Mg²⁺ (Fig. 7 B and C). However, at 10 mM Mg²⁺, translation rates were approximately twofold higher across all strains (Fig. 7 B and C). Taken together, these results indicate that the rapid assimilation of too much imported Pi reduces the availability of free cytoplasmic Mg²⁺, thereby lowering translation efficiency and promoting MgtC expression.

Discussion

We have identified a physiological connection between Pi assimilation and Mg²⁺ homeostasis whereby cells exert tight control over Pi uptake to avoid uncontrollable ATP synthesis and depletion of free cytoplasmic Mg²⁺ (Fig. 1). We establish that the Salmonella MgtC protein lowers ATP amounts by inhibiting Pi uptake (Figs. 3 A–C and 4B), thereby limiting the availability of an ATP precursor rather than interfering with the enzymatic catalysis of ATP-generating reactions per se (Figs. 2 A–C and 3 A–C). MgtC hinders the activity of a yet unidentified transporter that functions as the main Pi uptake system in Salmonella (Fig. 4 A–F). This inhibitory activity could take place through a number of different mechanisms. These include 1) a direct physical interaction with the transporter and 2) indirectly through interaction(s) with cellular component(s) that inhibit the Pi transporter activity, synthesis, or degradation. Furthermore, even when cytoplasmic Mg²⁺ is not limiting, excessive Pi uptake leads to increased ATP synthesis, which reduces free cytoplasmic Mg²⁺, impairing translation and inhibiting growth (Figs. 6 A–D, 7 A–C and 8).

Fig. 8.

Model illustrating how limitation of phosphate assimilation maintains cytoplasmic Mg²⁺ homeostasis in Salmonella enterica. (Top) Overview of the temporal adaptation of S. enterica to Mg²⁺ starvation. Mg²⁺ (green) and Pi (blue) concentrations in the environment, lipopolysaccharide (LPS), and cytoplasm are depicted as gradients with dark colors denoting high concentration and light colors representing low concentrations. (Bottom) Schematic depicting molecular events and responses underlying the adaptation. (A) During homeostasis, Pi is imported into the cytoplasm through dedicated inner membrane (IM) transport systems (X [unknown transporter] and PitA). Cells assimilate imported Pi through the synthesis of ATP, which exists as a salt with positively charged Mg²⁺ (ATP:Mg²⁺). ATP:Mg²⁺ 1) mediates the transfer of Pi among biological molecules, 2) powers energy-dependent enzymatic reactions, and 3) promotes ribosome biogenesis (1). ATP hydrolysis for the release of energy recycles Pi and Mg²⁺, whereas the biosynthetic transfer of Pi typically recycles cytoplasmic Mg²⁺. (B) After consuming the Mg²⁺ present in the environment, cells eventually experience a shortage in cytoplasmic Mg²⁺ levels (22, 23, 57). Insufficient cytoplasmic Mg²⁺ impairs ribosomal subunit assembly, lowering translation efficiency (28). This reduces the consumption of ATP by translation reactions and, consequently, decreases the recycling of Mg²⁺ and Pi from ATP:Mg²⁺ (17). (C) Salmonella enterica expresses the MgtC membrane protein as a homeostatic response to cytoplasmic Mg²⁺ starvation. MgtC inhibits Pi uptake through an unknown transporter (X), thereby preventing assimilation Pi into ATP. As the levels of ATP and ribosomes decrease, free Mg²⁺ ions necessary for core processes such as translation are recovered, increasing the efficiency of protein synthesis and the recycling of Mg²⁺ and Pi from ATP:Mg²⁺. The density and size of cartoons represent the concentrations of Mg²⁺, Pi, and ribosomes.

The Physiological Connection between Pi and Mg²⁺ Homeostasis.

The capacity of MgtC to maintain physiological ATP levels during cytoplasmic Mg²⁺ starvation had been so far ascribed to its inhibitory effect on Salmonella’s F₁F_o ATP synthase (32). Here, we demonstrate that this inaccurate conclusion resulted from an experimental setting arising from the propagation of atpB strains in a poorly fermentable carbon source (Fig. 2 A–C). Whereas the function of MgtC in macrophages is associated with the activity of the F₁F_o ATP synthase (32), the specific role of this protein within this biological context is not well understood. Cytoplasmic Mg²⁺ starvation can trigger MgtC expression by impairing the activity of bacterial ribosomes (Fig. 7) (27, 28, 40). However, MgtC promotes Salmonella survival irrespective of whether macrophages can restrict the access of bacteria to Mg²⁺ (30, 58, 59). This suggests that a macrophage-induced stress, other than Mg²⁺ starvation, impairs Salmonella translation, mimicking a physiological signal experienced during cytoplasmic Mg²⁺ starvation. That is, when inside macrophages, Salmonella may “sense” cytoplasmic Mg²⁺ starvation even though Mg²⁺ may not be limiting in the subcellular compartment harboring Salmonella (60).

MgtC inhibits Pi uptake independently of all known Pi importers (PitA and PstSCAB) as well as the Pi exporter YjbB. When heterologously expressed in E. coli, MgtC activates the PhoB/PhoR system (SI Appendix, Fig. S3). In addition to PstSCAB and PitA, E. coli harbors PitB, a Pi importer absent from Salmonella (Fig. 4A) (1, 51, 61). Notably, E. coli strains containing mutations in all of these three Pi transport systems are still able to grow on minimal medium with Pi as the only P source (51, 62). This indicates that E. coli encodes an additional Pi transporter(s) that may be targeted by MgtC in Salmonella. Given that MgtC sequelogs promote growth in low Mg²⁺ in a large number of distantly related species (34, 63–69), the Pi transporter targeted by MgtC is likely to be widespread in bacteria.

If MgtC inhibits the activity of a single Pi importer, what then prevents Pi uptake by other transport systems? PhoB promotes transcription of the high affinity PstSCAB Pi importer in response to a decrease in cytoplasmic Pi. During the initial stages of cytoplasmic Mg²⁺ starvation, the ribosomal subunits are unable to assemble efficiently (28). This leads to a decrease in translation efficiency and a concomitant reduction in ATP hydrolysis and free cytoplasmic Pi, which triggers pstSCAB transcription (17). Whereas MgtC inhibits the main Pi transporter, expression of MgtA and MgtB results in an influx of Mg²⁺ into the cytoplasm (21–26). Both responses restore ribosomal subunit assembly thereby increasing translation efficiency (Fig. 8). ATP consumption during translation (most notably the charging of transfer RNAs and the synthesis of guanosine triphosphate, which is subsequently used by elongation factors) (28) replenishes intracellular Pi, which represses pstSCAB transcription by inactivating PhoB/PhoR (17).

The availability of the Pi:Zn²⁺ salt regulates pitA transcription (70). In addition, PitA is posttranscriptionally repressed during Mg²⁺ starvation in E. coli. This repression is orchestrated by the Mg²⁺-sensing PhoP/PhoQ two-component system (71), which, in Salmonella, activates transcription of mgtCB and mgtA (21). Because growth in low Mg²⁺ decreases pitA mRNA levels in both Salmonella and Yersinia pestis (72, 73), inhibition of PitA expression appears to be a common feature of the Mg²⁺ starvation response in enteric bacteria. Given that PitA transports M:Pi salts, a reduction in PitA abundance may hinder uptake of other metal cations, such as zinc, that can readily replace scarce Mg²⁺ in enzymatic reactions (74–76). In this context, it is interesting to note that the Pho84 Pi transporter of Saccharomyces cerevisiae can promote metal toxicity by importing M:Pi salts (77–80).

If excessive ATP is toxic during cytoplasmic Mg²⁺ starvation, why, then, does MgtC function by hindering Pi uptake and not by directly inhibiting ATP-generating enzyme(s)? Depending on the growth condition, the production of ATP via the oxidation of carbon can occur via several distinct pathways, each involving dozens of enzymes (48). While we can conceive that a single protein may have the capacity to directly inhibit a myriad of distinct enzymes, evolution has likely provided cells with a more parsimonious solution for this problem. Because Pi is an ATP precursor, limitation of Pi uptake enables cells to hinder all ATP generating reactions independently from the metabolic pathway(s) used to oxidize the available carbon source(s) (Figs. 2B and 7). Hence, the inhibition of Pi uptake by MgtC allows Salmonella to lower ATP synthesis, eliciting a physiological response to a lethal depletion of cytoplasmic Mg²⁺, whether carbon is being metabolized via classic glycolysis, the Entner–Doudoroff, the Pentose Phosphate, the Tricarboxylic Acid Cycle, or any other energy-generating pathway (Figs. 2B and 8).

On the Activation of PhoR by MgtC.

Choi et al. reported that MgtC promotes PstSCAB expression and Pi uptake due to PhoB/PhoR activation resulting from a direct physical interaction between the MgtC and PhoR proteins (81). Our data provide independent pieces of evidence demonstrating that MgtC actually inhibits Pi uptake, which helps maintain cellular viability during cytoplasmic Mg²⁺ starvation.

First, the proposal that MgtC furthers Pi uptake (81) is incompatible with the fact that Pi is acutely toxic to cells experiencing cytoplasmic Mg²⁺ starvation (Fig. 5 C and D) and that MgtC is produced in response to cytoplasmic Mg²⁺ starvation (23, 24). Second, during cytoplasmic Mg²⁺ starvation, MgtC expression follows PhoB/PhoR activation and pstSCAB transcription, and MgtC production results in inactivation of PhoB/PhoR (SI Appendix, Fig. S4 A and B) (17). Third, this temporal chain of events is supported by a disruption in protein synthesis due to excessive Pi exacerbating cytoplasmic Mg²⁺ starvation and by influx of extracellular Pi via PstSCAB inducing mgtC expression (Figs. 6 A–D and 7 A–C). And fourth, substitution of the PhoR leucine 421 by an alanine was reported to disrupt PhoR interaction with MgtC, preventing PhoB/PhoR activation in cytoplasmic Mg²⁺ starvation (81). However, wild-type and phoR^L421A strains display similar pstS transcriptional kinetics in response to ectopic or chromosomal MgtC expression (SI Appendix, Fig. S5). Notably, cytoplasmic Mg²⁺ starvation also disrupts translation homeostasis in E. coli (28), thereby inducing a shortage in free cytoplasmic Pi and triggering PhoB/PhoR activation (SI Appendix, Fig. S6) (17), even though E. coli lacks MgtC.

Pi Toxicity and the Control of P Assimilation.

Mutations that increase Pi uptake via the PstSCAB transport system inhibit growth in a wide range of bacterial species (1–9). However, the underlying molecular basis for this inhibition has remained elusive. That MgtC promotes growth and viability during cytoplasmic Mg²⁺ starvation (21, 23) by inhibiting Pi uptake (Figs. 3 A and B and 4B) and, consequently, ATP synthesis (Fig. 3C) (29) prompted us to examine the physiological basis of Pi toxicity observed in the context of the aforementioned mutations. We established that increased Pi transport via PstSCAB causes a rise in ATP concentrations (Fig. 6A). Increased ATP disrupts the pools of free cytoplasmic Mg²⁺ (Fig. 7A), thereby inhibiting translation (Fig. 7 B and C), growth (Fig. 6B), and triggering MgtC expression when the Mg²⁺ concentration in the growth medium is sufficiently high to silence its expression (Fig. 7A) (22, 23, 27). Our results both establish that the toxic effects of excessive Pi are manifested following its assimilation into ATP and shed light into the underlying causes of Pi toxicity by revealing a logic for cellular control of P assimilation. The rapid synthesis of ATP (Fig. 6 A–C), and ATP-derived highly charged Pi anions-containing molecules, such as rRNA and poly-Pi (28, 82, 83), depletes the pools of free cytoplasmic Mg²⁺. Therefore, when biosynthetic precursors are abundant, but Mg²⁺ is limiting, cells adjust Pi uptake as an effective way to inhibit ATP-generating reactions (Fig. 1).

Materials and Methods

Bacterial Strains, Plasmid Constructs, Primers, and Growth Conditions.

The bacterial strains and plasmids used in this study are listed in SI Appendix, Table S1, and oligonucleotide sequences are presented in SI Appendix, Table S2. Single gene knockouts and deletions were carried out as described (84). Mutations generated via this method were subsequently moved into clean genetic backgrounds via phage P22-mediated transduction as described (85). For chromosomal point mutations, detailed strain construction is described below. Bacterial strains used in recombination and transduction experiments were grown in Luria–Bertani (LB) medium at 30 °C or 37 °C (84–86). When required, the LB medium was supplemented with ampicillin (100 μg/mL), chloramphenicol (20 μg/mL), kanamycin (50 μg/mL), and/or l-arabinose (0.2% wt/vol).

Unless stated otherwise, physiological experiments with bacteria were carried out at 37 °C with shaking at 250 rpm in MOPS medium (87) lacking CaCl₂ (to avoid repression of the PhoP/PhoQ system) (88) and supplemented with 0.1% (wt/vol) bacto casamino acids (BD Difco), 25 mM glucose, and the indicated amounts of MgCl₂ and K₂HPO₄. Experiments were conducted as follows: After overnight (∼16 to 20 h) growth in MOPS medium containing 10 mM MgCl₂ and 2 mM K₂HPO₄, cells were washed three times in medium lacking Mg²⁺ and Pi, inoculated (1:100) in fresh medium containing the indicated concentrations of MgCl₂ and K₂HPO₄, and propagated for the corresponding amount of time. It should be noted that at a concentration of 0.1% (wt/vol) bacto casamino acids (BD Difco), the medium already contains ∼163 μM Pi. During physiological experiments, selection of plasmids was accomplished by the addition of ampicillin at 100 μg/mL (overnight growth) or 30 μg/mL (experimental condition), chloramphenicol at 20 μg/mL (overnight growth) or 10 μg/mL (experimental condition), and/or kanamycin at 50 μg/mL (overnight growth) or 20 μg/mL (experimental condition). Unless specified otherwise, heterologous expression of proteins was achieved by treatment of cultures with 250 µM (pMgtC, pPstSCAB) isopropyl β-D-1-thiogalactopyranoside (IPTG). ATPase expression from pBbB2k-AtpAGD was attained without the addition of the inductor.

Construction of Plasmids Used in This Study.

Phusion High-Fidelity DNA Polymerase (New England Biolabs) was used in PCR to amplify DNA inserts. These DNA inserts were cloned into linearized plasmids (17, 27, 89–91) using NEBuilder HiFi DNA Assembly Cloning Kit (New England BioLabs). A detailed construction of plasmids is described in the SI Appendix.

Construction of phoR^L421A Strain.

Chromosomal point mutation was generated as described (84, 86). A detailed procedure is described in the SI Appendix.

Estimation of Intracellular ATP.

Intracellular ATP was estimated as described (17). A detailed description of this method is found in the SI Appendix.

Estimation of Intracellular Pi.

Total Pi in the samples was estimated from crude cell extracts using the molybdenum blue method (17, 92). A detailed description of this method is found in the SI Appendix.

Phosphate Transport Assay.

Wild-type (14028s) or mgtC (EL4) Salmonella were grown in MOPS medium containing 10 μM MgCl₂ and 500 μM K₂HPO₄ during 3 h. Wild-type (14028s) or ∆3Pi (RB39) cells harboring either pVector or pMgtC were grown in MOPS containing 250 μM MgCl₂ and 500 μM K₂HPO₄ until OD₆₀₀ ∼0.2, at which point, MgtC expression was induced for 15 min with the addition of 750 μM IPTG. To assay the transport of Pi, 20 μCi of radioactive Pi solution (10 μL from a 2 mCi K₂H³²PO₄ at a concentration of 2 mM K₂H³²PO₄, PerkinElmer catalog number NEX055) was added to 1 mL cell suspension. At the indicated time points, 50 μL of each sample was submitted to rapid filtration through 0.45 μm mixed cellulose ester membrane filters (Whatman) with an applied vacuum. The filters were washed three times with 1 mL PBS buffer and subsequently soaked in 5 mL scintillation fluid (Research Products International). The amount of radioactivity taken up by the cells was determined with a scintillation counter (Triathler multilabel tester, HIDEX) using the ³²P-window and by counting each vial for 20 s. Radioactive counts per minute were normalized by protein content using a Rapid Gold BCA Protein Assay Kit (Pierce). ³²Pi uptake of each sample was normalized against the corresponding control in each independent experiment.

Monitoring Bacterial Growth and Gene Expression via Fluorescence.

Bacterial growth and fluorescence measurements were performed using SpectraMax i3x plate reader (Molecular Devices). The detailed protocols are described in the SI Appendix.

l-azidohomoalanine Protein Labeling and Quantification.

In vivo labeling of proteins and estimation of translation rates were performed as described (26), with minor modifications highlighted in the SI Appendix.

Image Acquisition, Analysis, and Manipulation.

Plates, gel, and membrane images were acquired using an Amersham Imager 600 (GE Healthcare Life Sciences). ImageJ software (93) was used to crop the edges and adjust the brightness and contrast of the images. These modifications were simultaneously performed across the entire set images to be shown.

Data Availability

All study data are included in the article and/or SI Appendix.

Change History

March 16, 2021: The author line has been updated.