Methods of using non-growing bacteria to synthesize target products

Abstract

This disclosure generally relates to a non-growing bacteria that continues to function in a stressed environment, e.g., starvation. Also provided in the present disclosure are improved methods of target product synthesis using the bacterial cells in the cultivation methods described herein.

Description

[0001] Attorney Docket No.15670-0435WO1 METHODS OF USING NON-GROWING BACTERIA TO SYNTHESIZE TARGET PRODUCTS CLAIM OF PRIORITY This application claims the benefit of U.S. Provisional Application Serial No. 63 / 700,288, filed on September 27, 2024. The entire contents of the foregoing are incorporated herein by reference. SEQUENCE LISTING The instant application contains a Sequence Listing which has been submitted electronically in ASCII format and is hereby incorporated by reference in its entirety. Said ASCII copy, created on September 25, 2025, is named 15670- 0435WO1_SL_ST26.xml and is 2,804 bytes in size. FEDERALLY SPONSORED RESEARCH OR DEVELOPMENT This invention was made with Government support under Grant No. GM152133 awarded by the National Institutes of Health. The Government has certain rights in the invention. TECHNICAL FIELD This disclosure generally relates to a non-growing bacteria that continues to function in a stressed environment, e.g., starvation. Also provided in the present disclosure are improved methods of target product synthesis using the bacterial cells in the cultivation methods described herein. BACKGROUND Engineered bacteria are widely used as bio-factories to produce useful products, including biofuel, enzymes, drugs, and cosmetics. When used as microbial catalysts, these bacteria must separate the incoming substrate between the synthesis of bacterial cell mass and the synthesis of a desired product. Although two-stage bioprocesses can accommodate this tradeoff through temporal separation of the growth and production phases, the intricate couplings between protein synthesis, protein turnover, and cell viability during starvation, reflect a hitherto little-studied subject of bacterial physiology that is critical to bacterial fitness. Further Attorney Docket No.15670-0435WO1 understanding this class of phenomena exposes unknown bacterial vulnerabilities – inspiring new strategies targeting non-growing cells as provided by the present disclosure. Better understanding of proteome dynamics in non-growing bacteria enables the rational design of genetic circuits in stationary cells, leading to new strategies for synthetic biology, e.g., in extending the lifetime and functionalities of non-growing cells, thereby turning them into highly efficient and evolutionarily stable bio-factories. SUMMARY Survival through periods of starvation is crucial for bacterial fitness. A lot is known about what genes are expressed during starving bacteria and whether they contribute to viability, exemplified by studies of the “stress response” proteins driven by the general stress sigma factor RpoS in E. coli. However, much less is known about the metabolic processes sustaining gene expression during starvation, how they are controlled and paid for, and in what ways they benefit survival. The present disclosure provides a comprehensive physiological study of E. coli cells under starvation conditions (e.g., glucose starvation, nitrogen starvation), quantifying the temporal dynamics of resource generation and utilization, and connecting them to cell viability. We focused on a 24-h preparatory period where the proteome was remodeled after entering starvation, well before the onset of the death phase. We found two proteins degraded for every protein synthesized, and 5-6 proteins degraded for every protein accumulated, resulting in the turnover of over 50% of the proteome for a moderate gain in stress proteins. The present disclosure demonstrates that such extensive global proteolysis required a little-known member of the RpoS regulon, YeaG, which worked in concert with members of know proteolytic machinery. YeaG was predicted to exert its activity in complex with YeaH and YcgB. The total amount of protein synthesis was dictated by the amount of protein degradation via YeaG in complex with YeaH and YcgB, reflecting a dominant, supply-driven strategy of resource allocation for starving cells. This YeaG-mediated proteome remodeling accounted for a major share of the viability boost provided by RpoS. However, in a mutant where energy can be supplemented without cell growth, YeaG-mediated global proteolysis still occurred but was largely unnecessary, suggesting that a major physiological function of Attorney Docket No.15670-0435WO1 proteome remodeling was to increase the energy efficiency of starving cells. Such knowledge on the physiology and molecular biology of starving bacteria offers new insights on bacterial vulnerabilities and inspires new strategies for engineering efficient bacterial bio-factories. Provided herein are bacterial cells comprising one or more loss-of-function mutations in genes yeaG, yeaH, ycgB, and / or aceA, or bacterium-specific gene equivalent thereof, wherein the bacterial cells maintain synthesis of at least one target product during stress. In some embodiments, a bacterial cell disclosed herein can be from a bacterium, wherein the bacterium belongs to a genus selected from the group consisting of Acidovorax, Acinetobacter, Actinomyces, Alcanivorax, Arthrobacter, Brevibacterium, Bacillus, Clostridium, Corynebacterium, Deinococcus, Dietzia, Escherichia, Gordonia, Marinobacter, Mycobacterium, Micrococcus, Micromonospora, Moraxella, Nocardia, Pseudomonas, Psychrobacter, Rhodobacter, Rhodococcus, Salmonella, Streptomyces, Thalassolituus, Thermomonospora, Vibrio, and Zymomonas. In some embodiments, a bacterial cell disclosed herein can be from a bacterium selected from the group consisting of Escherichia coli, Bacillus subtilis, Streptomyces spp., Corynebacterium glutamicum, Lactococcus lactis, Pseudomonas putida, Myxococcus xanthus, Thermus thermophilus, Clostridium spp., Agrobacterium tumefaciens, and Vibrio natriegens. In some embodiments, a bacterial cell disclosed herein can be from a bacterium selected from the group consisting of Escherichia coli (E. coli), Pseudomonas putida (P. putida), Bacillus subtilis (B. subtilis), and Vibrio natriegens (V. natriegens). In some embodiments, a bacterial cell disclosed herein can be (i) an E. coli cell comprising one or more loss-of-function mutations in genes yeaG, yeaH, ycgB, and / or aceA; (ii) an P. putida cell comprising one or more loss-of-function mutations in genes equivalent to yeaG, yeaH, ycgB, and / or aceA; (iii) a B. subtilis cell one comprising or more loss-of-function mutations in genes prkA, yhbH, and / or spoVR; or (iv) an V. natriegens cell comprising one or more loss-of-function mutations in genes PN96_RS08585, PN96_RS08590, PN96_RS08595, and / or aceA. In some embodiments, a bacterial cell disclosed herein can be an E. coli cell comprising a deletion of the yeaG, yeaH, and / or ycgB gene. In some embodiments, the one or more loss-of-function mutations in genes yeaG, yeaH, ycgB, and / or aceA, or bacterium-specific gene equivalent thereof, can Attorney Docket No.15670-0435WO1 comprise a deletion of the yeaG, yeaH, ycgB, and / or aceA gene, or bacterium-specific gene equivalent thereof. In some embodiments, the one or more mutations occurs in (i) a gene that comprises the complex of yeaG, yeaH, and ycgB; or (ii) a gene whose activity is affected by the complex of yeaG, yeaH, and ycgB. In some embodiments, a bacterial cell disclosed herein can be an E. coli cell comprising a deletion of the yeaG, yeaH, and / or ycgB gene. In some embodiments, a bacterial cell disclosed herein can be a bacterial cell from an E. coli K-12 strain. In some embodiments, a bacterial cell disclosed herein can be derived from an E. coli K-12 parental bacterial strain. In some embodiments, a bacterial cell disclosed herein can be a bacterial cell from an engineered strain of E. coli. In some embodiments, a bacterial cell disclosed herein can be derived from an engineered E. coli. parental bacterial strain. In some embodiments, the stress occurs during a stationary phase of product synthesis. In some embodiments, the bacterial cell does not grow during the stationary phase of product synthesis. In some embodiments, the stress occurs during starvation. In some embodiments, the starvation comprises carbon starvation, nitrogen starvation, phosphorus starvation, sulfur starvation, iron starvation, oxygen starvation, amino acid starvation, or combinations thereof. In some embodiments, the starvation is glucose starvation. In some embodiments, a bacterial cell disclosed herein can maintain synthesis of the at least one target product for 1-14 days during stress. In some embodiments, a bacterial cell disclosed herein can have viability at about 40% to about 100% of initial colony form units (CFUs). Also provided herein are compositions comprising any of the bacterial cells disclosed herein. Provided herein are also kits for use in making any of the bacterial cells disclosed herein, preparing any of the compositions disclosed herein, and / or for use in any of the methods disclosed herein. In some embodiments, a kit can comprise at least one parental bacterial strain (e.g., an E. coli strain) and, optionally, at least one reagent suitable for preparing any of the bacterial cells disclosed herein. In some embodiments, a kit can comprise any of the bacterial cells disclosed herein and a suitable culture medium for use in producing any of the target products according to the methods disclosed herein. Attorney Docket No.15670-0435WO1 Provided herein are also methods of producing at least one target product using any of the bacterial cells disclosed herein. In some embodiments, a method of producing at least one target product using a bacterial cell disclosed herein can comprise: (a) cultivating the bacterial cells in a suitable culture medium under conditions that permit the bacterial cells to produce the at least one target product, wherein the at least one target product is released into the culture medium; and (b) isolating at least one target product from the culture medium. In some embodiments, the suitable culture medium lacks at least one essential nutrient for growth. In some embodiments, the at least one essential nutrient for growth is selected from the group consisting of nitrogen, phosphorus, potassium, sulfur, carbon, an amino acid, a nucleotide, and a vitamin. In some embodiments, the at least one essential nutrient for growth is glucose. In some embodiments, step (a) comprises cultivating the bacterial cells for about 1 to about 15 days. In some embodiments, step (a) comprises cultivating the bacterial cells for at least 6 days. Provided herein are methods of increasing at least one target product using any of the bacterial cells disclosed herein. In some embodiments, a method of increasing at least one target product comprises (a) growing a plurality of the bacterial cells disclosed herein in a glucose-limited culture medium; (b) adding acetate into the glucose-limited culture medium once glucose is depleted from the glucose-limited culture medium; and (c) culturing the plurality of the bacterial cells for about 1 to about 15 days in the acetate culture medium. In some embodiments, methods of increasing at least one target product using any of the bacterial cells disclosed herein can increase the amount of target product by at least 1-fold, 1.5-fold, 2-fold or more than 2-fold as compared to the amount of target product prepared using a method with a parental bacterial strain. In some embodiments, methods of producing at least one target product and / or increasing at least one target product using any of the bacterial cells disclosed herein can comprise a bacterial fermentation process. In some embodiments, the bacterial fermentation process is selected from the group consisting of batch fermentation, fed- batch fermentation, and continuous fermentation. In some embodiments, the bacterial cell remains in a stationary phase of the bacterial fermentation process for about 1 to about 15 days. In some embodiments, the bacterial cell remains in a stationary phase of the bacterial fermentation process for at least 6 days. Attorney Docket No.15670-0435WO1 In some embodiments, the at least one target product is selected from the group consisting of a carbohydrate, an amino acid, a protein, and a nucleic acid. In some embodiments, the at least one target product is a protein, an antimicrobial, an antibiotic, an enzyme, a probiotic, a toxin, a biofuel, an organic acid, an amino acid, a vitamin, a biopolymer, a polysaccharide, a secondary metabolite, or combinations thereof. In some embodiments, the at least one target product is butanediol. In some embodiments, the at least one target product is 2,3-butanediol. In some embodiments, the at least one target product is 1,4-butanediol. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. The term “about” as used herein can allow for a degree of variability in a value or range, for example, within 10%, within 5%, or within 1% of a stated value or of a stated limit of a range. Methods and materials are described herein for use in the present invention; other, suitable methods and materials known in the art can also be used. The materials, methods, and examples are illustrative only and not intended to be limiting. All publications, patent applications, patents, sequences, database entries, and other references mentioned herein are incorporated by reference in their entirety. In case of conflict, the present specification, including definitions, will control. Other features and advantages of the invention will be apparent from the following detailed description and figures, and from the claims. DESCRIPTION OF DRAWINGS FIGS.1A-1F: Resource allocation and protein synthesis in starving cells. FIG.1A: Culture of E.coli strain K-12 NCM3722 (WT)59,60growing exponentially in glucose minimal medium at mid-log was diluted to a fresh medium with limiting glucose (2 mM); exponential growth continued for several doublings until optical density (OD600, top panel) reached ~0.4 when growth was abruptly arrested (vertical dashed line, t=0) due to glucose depletion (bottom panel). Unless otherwise indicated, all starvation experiments in this study were performed this way, which was highly repeatable (see FIG.7A). FIG.1B: Viability of WT and ∆rpoS cultures at various times after starvation was characterized by counting colony forming units (CFU) on Luria Bertani (LB) plates (see FIG.7B). FIG.1C: Illustration of amino acids (AA) Attorney Docket No.15670-0435WO1 and ATP (energy) generation through protein turnover during starvation. Protein degradation released free AAs. A portion of these AAs were reincorporated into newly synthesized proteins, others were catabolized to generate ATP (needed to fuel protein synthesis and maintain vital cellular functions). Non-catabolizable AAs were excreted (see FIG.8A). FIG.1D: Total protein abundance of WT and ∆rpoS cultures was measured by radiolabeling (see FIG.9). The results were expressed as percentages of the Initial Total Protein Mass (%ITPM), which was the total protein content during exponential growth1.1% ITPM is ~40,000 copies / cell. For CFU measurements, the value of each biological replicate was obtained as the average over three technical replicates. The vertical bars represent standard errors. FIG.1E: The estimated upper bound in protein synthesis, assuming that the protein loss measured in FIG.1D was converted completely to ATP, and this ATP obtained was used exclusively to fuel protein synthesis at 4 ATP per peptide bond. Plotted are estimates based on the AA-ATP conversion efficiency given in the legend for WT and ∆rpoS cells. FIG.1F: The expected amount of protein degradation was obtained as the sum of the expected synthesis in FIG.1E and the measured amount of protein loss in FIG.1D. FIGS.2A-2G: Proteome remodeling by WT and ΔrpoS cells during glucose starvation. FIG.2A: Illustration of absolute abundance estimation for individual proteins. From protein samples collected at time t=0 (initial), 1.5 h, 6 h and 24 h during glucose starvation, mass spectrometry based proteomics32provided the fractional abundance of each detected protein, ^^i(^^) = ^^i(^^) / ^^total(^^), where ^^i(^^) is the mass of an individual protein i, and ^^total(^^) = Σi^^i(^^) is the total protein mass detected. Because proteins captured by proteomics accounted for over 90% of total protein mass in the culture, we took ^^total(^^) to measured protein mass ^^(^^) as given in FIG.1D, and estimated ^^i(^^) as ^^i(^^) ⋅ ^^(^^). The result was plotted as %ITPM (corresponding to ~40,000 copies / cell) for two exemplary proteins:(1) HNS, a nucleoid-associated protein, which exhibited a rising proteome fraction,^^HNS, but its protein amount ^^HNS was nearly constant after taking into account of thetotal protein loss; and (2) CysK, a metabolic enzyme, which showed a definitive drop in protein amount after including the total protein loss. FIGS.2B and 2C: The time course of protein mass ^^i(^^) in WT culture for five proteins with the largest increase (FIG.2B) or decrease (FIG.2C) in amount at 6 h into starvation, among all proteins Attorney Docket No.15670-0435WO1 that changed by at least 1.5-fold. For proteins of typical lengths, 1% ITPM corresponded to ~40,000 copies / cell. The right axis in FIG.2C indicates the abundance of MetE. FIG.2D: The time course of summed mass of proteins belonging to the four functional groups that increased in WT culture during the first 6 h of glucose starvation. FIG.2E: The time course of summed mass of proteins belonging to the two functional groups that showed by far the largest decreases in WT culture during the first 6 h of glucose starvation. FIG.2F: The time course of the sum of all detected proteins whose mass increased from that at the time of starvation. Circles: WT culture; squares: ΔrpoS culture. FIG.2G: The time course of the sum of all detected proteins whose mass decreased from that at the time of starvation. Circles: WT culture; squares: ΔrpoS culture. FIGS.3A-3J: Characteristics of YeaG-dependent proteolysis. FIG.3A: Total protein amount remaining in the culture for ΔyeaG (diamonds), Δdps (triangles), and ΔraiA Δrmf Δhpf (Δ3R, asterisks) after glucose depletion. The data for WT and ΔrpoS cells were the same as in FIG.1D. FIG.3B: Viability of the 5 cultures described in panel. FIG.3C: Direct measurement of protein degradation after glucose depletion. Proteins were labeled with14C-leucine during exponential growth and then chased with excess12C-leucine upon glucose depletion (t=0). The amount of14C-leucine in the supernatant reflected degradation of proteins synthesized before t=0; see FIGS.15A-15E. Plotted are the amount of pre-starvation proteins degraded for the 5 strains described in FIG.3A during the first 24 hours of glucose starvation. FIG.3D: Direct measurement of protein degradation after glucose depletion for strains deleted of one or more proteases, or with the deletion of yeaH, which was co-transcribed with yeaG from the same operon. FIG.3E:14C-leucine was added for specific periods after glucose depletion (see legend) to label proteins synthesized during those periods (see FIG.15D). The culture was subsequently chased with excess12C-leucine, and the amount of14C-leucine released in the medium was recorded and plotted. The slope of the rise of each set of symbols from zero (various lines) indicated the degradation rate for the labeled proteins, ~5% / h for those synthesized during growth, and 15-20% higher for those synthesized during the first 4 h of starvation. FIG.3F: Degradation of protein during exponential growth was measured with minor modification (FIG.15E), for WT and ΔyeaG cells each grown in glucose and in mannose as the sole carbon source. The degradation rates, Attorney Docket No.15670-0435WO1 taken as the initial slope of the data points, were ~1.3% / h for WT cells growing on glucose (black circles), and similar for ΔyeaG cells in glucose or mannose medium (grey symbols), but over 2x higher for WT cells grown in mannose (black triangles). This result was rationalized by the ~8-fold increase of YeaG protein levels in cells growing in poor carbon sources such as mannose compared to glucose31,32. Proteomic data is shown for some exemplary functional groups (AA biosynthesis (FIG.3G), TCA cycle (FIG.3H), RpoS regulation (FIG.3I), and ribosome hibernation (FIG. 3J)) expressed in WT and ΔyeaG cells, obtained as described in FIGS.12, 13. FIGS.4A-4F: Flux of protein synthesis and degradation. FIG.4A: The protein degradation flux ^^D(^^) at various time ^^ after glucose depletion, obtained by measuring the amount Δ^^(^^) of14C-leucine released into the medium over the course of a period Δ^^ (about an hour) after cells grown in14C-leucine were chased by the addition of excess12C-leucine at time ^^; see FIGS.15A-15E. The degradation flux at time ^^ was taken to be Δ^^(^^) / Δ^^. Circle: WT cells; diamonds: ∆yeaG cells. Dashed line: degradation flux during exponential growth (from FIG.3F). FIG.4B: The protein synthesis flux ^^S(^^) at various time ^^ after glucose depletion, obtained by adding14C-leucine to the stationary culture and measuring the depletion of radioactivity in the medium; see FIGS.19A-19C. Dashed lines in FIGS.4A and 4B represent the best linear fits to the data for WT cells (excluding the data at ^^ =0): ^^S (^^) =2.56 ⋅ (1 – ^^ / 25.9ℎ) and ^^S (^^) =5.27 ⋅ (1 – ^^ / 26.2ℎ). Vertical bars in FIGS.4A and 4B represent technical errors. FIG.4C: The data for WT cells in panels FIGS.4A and 4B were partitioned into 6 bins, and the average degradation flux of each bin was plotted against the average synthesis flux of that bin. Except for the first bin (0-2 h, filled circle), the data from the other 5 bins exhibited a linear relation (dashed line) with slope 1.83 ± 0.1 and y-intercept 0.16 ± 0.17% ITPM / h. Error bars: SEM within bin. FIG.4D: Total protein loss, computed as Δ^^loss(^^) = ∫ ^^^^′[^^D(^^′) – ^^S(^^′)] (line), compared to the measured data taken from FIG.1D (circles). FIG.4E: Amount of protein synthesized, computed as Δ^^synth(^^) = ∫ ^^^^′^^S(^^′) (top line), compared to the data from proteomics (circles, FIG.2F). FIG.4F: Amount of protein degraded, computed as Δ^^degr(^^) = ∫ ^^^^′^^D(^^′) (top line), compared to the data from proteomics (circles, FIG.2G). FIGS.5A-5F: Coordination of protein synthesis and degradation. FIGS.5A and 5B depict illustrations of two strategies of flux coordination. In the demand- Attorney Docket No.15670-0435WO1 driven strategy (FIG.5A), depletion of energy (the box filled in gray) due to the demand of synthesis removed the inhibition of proteolysis. In the supply-driven strategy (FIG.5B), increase in energy due to increased proteolysis enabled increased protein synthesis. The two strategies could be distinguished by perturbing the energy pool (the open arrow) and observing whether the response was dominated by changes in the synthesis or degradation flux. Perturbation of the energy pool could be implemented by supplementing acetate to cells deleted of aceA, as acetate could be used to generate energy but not for biomass growth; see FIGS.21A-21H. FIG.5C: ∆aceA cells were grown in glucose MOPS medium. When the exponentially growing culture reached OD = 0.4, cells were washed and resuspended in fresh MOPS medium without any carbon source (^^ =0).6 h later, the culture was either supplemented with 5 mM acetate (filled symbols) or no acetate (open symbols). Total protein content in each tube was monitored through 24 h. FIG.5D: In the experiment described in panel FIG.5C, protein synthesis flux was measured as described (see FIG.4B and FIGS.19A-19C) for two cultures of ∆aceA cells, with and without acetate supplement at 6 h after glucose depletion, both with14C-leucine added at ^^ =6ℎ. The amount of synthesis, taken as the reduction of radioactivity in the medium, was measured and plotted for the subsequent 45 min. FIG.5E: WT cells were grown exponentially in14C-leucine and entered glucose starvation as described in FIG.1A.6 h after starvation, the culture was chased with excess12C-leucine, with and without the addition of 35 μM of Chloramphenicol (filled and open symbols, respectively). The amount of radioactivity in the medium, taken as the amount of protein degraded (FIGS.15A-15E) was monitored and plotted for the subsequent 45 min. FIG.5F: Total protein abundances (FIG.3A, FIGS.9A-9D) were measured for WT and ∆yeaG cultures for up to 6 days after glucose starvation. The initial rapid total protein loss of the WT culture (25% in 1 day (d), circles) stopped abruptly after the first day, followed by a much slower loss of ~1.3% / d (dashed black line). The low degradation flux after day 1 is of the magnitude of the offset ^^0 in Eq. (1) which may reflect the energy flux demanded by cell maintenance. The ∆yeaG culture did not have a big initial protein loss; however, it exhibited a 2-fold larger long-term loss flux (~2.5% / d, diamonds). FIGS.6A-6K: Effects of energy supplement and protein synthesis on long- term viability. FIG.6A depicts an overview of the physiological functions of “stress Attorney Docket No.15670-0435WO1 proteins” synthesized during starvation. Some play a protective role (upper arrow), others serve to support energy maintenance (lower arrows), either by improving the efficiency of ATP conversion or by reducing the cost of maintenance (e.g., by reducing futile cycles). Only the conversion of AA to energy is shown but fatty acids and nucleotide also contribute. FIG.6B: Viability of glucose-starved ∆aceA cells (filled circles), with excess acetate (5 mM acetate added at the time of glucose depletion (t=0) (FIG.21C) and further addition as needed to maintain above 1 mM throughout the experiment). Viability data for WT cells (open circles) was the same as that in FIG.7A. FIG.6C: Total protein amount for ∆aceA cells (filled symbols) were taken as described for WT cells (open symbols) in FIGS.9A-9D, 10A, 10B but with excess acetate maintained throughout. FIG.6D: Degradation and synthesis fluxes of WT (filled and open black bars, respectively) are taken from FIGS.4A and 4B. Degradation flux of ∆aceA cells (filled grey bars) was measured relative to that of WT cells in parallel for each time point. Synthesis flux was estimated as the difference of the degradation flux at each time point and the protein loss flux at the corresponding time point (total protein content two hours apart). Shown is the abundance of the 4 protein groups (ribosome hibernation (FIG.6E), RpoS regulation (FIG.6F), recycling (FIG.6G), and carbon catabolism (FIG.6H)) that were increased in WT cells (FIG.2D) as open circles; also shown are those for ∆aceA cells (filled circles) and ∆aceA ∆yeaG cells (filled diamonds), each supplemented with excess acetate throughout. FIG.6I: Same as FIG.6B but including ∆aceA ∆yeaG cells supplemented with excess acetate throughout (filled diamonds). Viability of ∆yeaG cells (open diamonds) was taken from FIG.3B. FIG.6J: Same as FIG.6C but including ∆aceA ∆yeaG cells supplemented with acetate from t=0 (filled diamonds). Total protein of ∆yeaG cells (open diamonds) is taken from FIG.3A. FIG.6K: Same as FIG.6I but including ∆aceA ∆dps (filled triangles) and ∆aceA ∆rpoS cells supplemented with excess acetate throughout. FIGS.7A-7I: Growth curves, viability, and reductive cell division. FIG.7A: The behaviors of stationary cells depend on the dynamics of how nutrient depletion sets in. In this study, we applied the “natural runout” protocol where glucose was rapidly depleted by a culture of exponentially growing cells (FIG.1A). The course of growth arrest was highly reproducible, in terms of the timing and the OD at runout (at OD ≈ 0.4 for 2 mM glucose), as demonstrated by the plot of OD for four biological Attorney Docket No.15670-0435WO1 replicates. Unless otherwise indicated, all measurements reported herein pertain to the amount per culture at various times after glucose runout (t=0). FIG.7B: The number of colony forming units (CFU) per mL of the starving culture, normalized by the OD at the time of glucose depletion (OD≈0.4), referred to here and elsewhere as “iOD”. The viability curve exhibited a sharp rise during thefirst several hours of glucose starvation, then declined gradually over the course of the next 10 days. The gradual decline in viability was divided naturally into two phases, the “stationary phase” where viability barely dropped, followed by a “death phase” where the viability dropped exponentially. The exponential nature of the viability drop in the death phase resulted from nutrient recycling24. The data disclosed herein focused on what happened during thefirst 24 h of the stationary phase and its impact on the duration of the rest of the stationary phase. The initial sharp rise of the viability curve was due to the “reductive cell division”, a well-known process by which a sub-population of growth-arrested cells completed the current round of DNA replication and carried out another round of cell division to separate the newly generated chromosomes, thus increasing cell number in the absence of biomass accumulation29,30. As provided herein, the CFU data was always plotted relative to 6 h after glucose depletion to avoid referring to the reductive division. In FIGS.7C-7I, we provided some details of the reductive division process. FIG.7C shows the CFU of WT and ∆rpoS cultures during the initial hours after glucose depletion, exhibiting an approximate doubling of cell number, attributed to reductive cell division. The two cultures grew similarly and had similar number of cells / OD during exponential growth, reflecting the fact that RpoS was hardly expressed during exponential growth on glucose25. The data herein showed that these two strains were also similar in the early hours of glucose starvation. FIGS.7D and 7E depict representative microscopy images of wild-type cells at the moment before growth arrest i.e., 0 hours (FIG.7D) and at 6 hours (FIG. 7E). Scale bars indicate 5 μm. Shown are histograms of cell lengths (FIG.7F) and widths (FIG.7G) at 0 h (black line) and 6 h (grey line) after glucose depletion for wild-type cells. Shown are average cell length (FIG.7H) and average cell width (FIG.7I) of wild type cells (black circles) and ∆rpoS (open squares) at various times after glucose depletion. For FIGS.7F-7I, approximately 200 individual cells for each timepoint and each strain were measured. Attorney Docket No.15670-0435WO1 FIGS.8A-8C: Catabolism of amino acids for energy regeneration. FIG. 8A: As discussed herein, there are 12 catabolizable AAs; the other 8 AAs are non- catabolizable: leu, ile, val, met, lys, his, tyr, phe. The levels of these AAs rise over the course of 24-h glucose starvation15, except for lys, whose level turned around after ~5 h, suggesting its conversion to other substrates, possibly GABA31. The pool of trp, which is catabolizable, nevertheless accumulated over thefirst 12 h of starvation, suggesting the lack of trp catabolic enzyme, TnaA. The pool of arg rose in thefirst 1- 2 h15before dropping gradually downwards, suggesting catabolism after thefirst hour. FIG.8B: The proteomic data collected herein showed that TnaA and enzymes of the arginine degradation pathway (AstABCDE) accumulated over thefirst 6 h of starvation, corroborating the temporal excretion patterns observed for these two AAs, although differences in strains and conditions allowed only a qualitative comparison. FIG.8C: The grey bars show the number of ATP molecules derivable from each catabolizable AA, obtained from using Flux Balance Analysis with results tabulated in Table 4. Trp and Arg, the two catabolizable AAs which accumulated for some time after starvation, are shown as open bars. The average ATP derivable per AA, weighted by the frequency of that AA in the E. coli expressed proteome (black bars, are shown as the horizontal dotted and dashed lines, with Trp and Arg included (7.8 AA / ATP) and excluded (6.6 ATP / AA), respectively; see Table 4 for details. FIGS.9A-9D: Total protein abundance measurement. For sensitivity and reliability, we used a radio-labeling approach to quantify the amount of proteins in a culture. FIG.9A depicts an illustration of the experimental procedure. To label intracellular protein, cells were grown in the presence of14C-leucine until glucose ran out (t=0) at OD ≈ 0.4 (FIG.1A). (Leucine supplement did not affect exponential growth nor viability during glucose starvation, see FIGS.10A-10B). Unincorporated leucine was immediately removed from the medium by washing with media lacking glucose. Washed cells were resuspended in fresh medium without glucose (and no leucine) but additionally supplemented with 0.86 mM sodium acetate. The acetate supplement was included to reconstitute the medium at the time of glucose depletion, which contained 0.86 mM acetate due to overflow metabolism during exponential growth on glucose (about 2mM acetate / OD)12; no other metabolites were detected in the medium during growth15,32. Aliquots of the resuspended culture were sampled at regular time intervals and were precipitated with trichloroacetic acid (TCA). FIG. Attorney Docket No.15670-0435WO1 9B: The direct method to measure protein abundance would be to measure the radioactivity in the precipitate (black bars). However, this approach gave rise to variable results due to, e.g., loss of the precipitated materials during subsequent acetone-washes. Instead, we measured the radioactivity in the supernatant (gray bars). As the total radioactivity of the culture (horizontal arrow) cannot change, a decrease in the radioactivity from the TCA precipitate was coupled to an increase in the radioactivity of the supernatant. Thus, by monitoring the (much more accessible) radioactivity in the supernatant, radioactivity in the precipitate (and hence protein abundance) could be quantified. This problem with TCA precipitate was not limited to radio-labeling, but affected common approaches to measuring protein abundances, e.g., the Bradford assay33. It is for this reason that we used the radiolabeling assay instead of the common assays to quantify protein abundances in the data provided herein. FIG.9C: WT culture was grown in glucose +14C-leucine and washed at t=0 as described in FIG.9A. Samples taken during glucose starvation were analyzed. Squares represented total radioactivity measured in the sampled culture, taken as 100%, and triangles represented percent of total radioactivity in the supernatant. Circles were the differences between squares and triangles, which we took to be radioactivity in the TCA precipitate, reflecting the total protein amount remaining in the culture. The left vertical axis is the fraction of radioactivity in each component of the culture. The right axis represents the same data, but on the scale of absolute protein abundance, taking the measured protein abundance in exponential growth (333 µg / OD / ml12) as the protein abundance at t=0. FIG.9D: The results obtained in this manner were highly repeatable, as shown by independent biological replicates for WT and ∆rpoS strains (black and grey symbols, respectively) during thefirst 24 h of glucose starvation. FIGS.10A and B: Validation of leucine-labelling to monitor protein abundance. FIG.10A: The presence of external leucine had no detectable effect on growth rate and yield of E. coli. This was expected as E. coli cannot be grown on leucine as the sole carbon source and lacks leucine degradation pathways (see FIGS. 8A-8C), making it an ideal choice for radiolabeling experiments. FIG.10B: Leucine incorporation into proteins (TCA precipitable fraction, FIGS.9A-9D) for WT cells (circles) and leucine auxotroph, ∆leuA cells (triangles), plotted at different culture OD600during exponential growth on glucose supplemented with 400 M leucine. Attorney Docket No.15670-0435WO1 Comparable leucine incorporation by WT and ∆leuA cells suggested that when supplemented with leucine, de novo leucine biosynthesis was negligible for WT cells. Thus, the relationship between radioactivity and leucine content (specific radioactivity of leucine) in proteins was the same as that for externally provided leucine, as assumed in the analysis of FIG.9C. The straight line is a two-parameter linear fit of the data. The x-intercept of the fit (~0.15 OD) corresponded to the loss in biomass during centrifugation procedures. The slope, 265 nmol / OD / ml was equivalent to 8.8% of protein content (3 mol / OD / mL based on 330 g protein / OD / mL12). This matched closely the leucine frequency of 8.7% listed in Table 4. This agreement further reinforced the lack of leucine catabolism. FIGS.11A-11C: Turnover of RNA and lipids. The degradation of RNA in starving bacteria is known for over 60 years34and has remained an active topic of study in bacteriology ever since18,35–39. Most of the RNA degraded is ribosomal RNA (rRNA)18,37–39while tRNAs are relatively stable under various starvation conditions40. The extent to which rRNA is degraded has been estimated by many studies to be between 30-50% within 24 h of starvation18,37–39,41,42. It is known that once ribosomal degradation is initiated, its disintegration is rapid35,43. There is evidence that several ribosomal proteins require the presence of rRNA for their stability6,7,18. Molecularly, it is thought that the ClpX-degradation tag that is carried on the C-termini of several ribosomal proteins44become exposed in the absence of rRNA, leading to their degradation. However, Kaplan & Apiron35showed that ribosomal proteins are fairly stable during starvation, suggesting that either not all r-proteins maybe subject to degradation, or some r-proteins might still be protected by smaller fragments of rRNA. A recent proteomic study corroborated the existence of two r-proteins species, one that degrades at the same rate as rRNA suggesting that they require rRNA for their protection, and the second species that are stable even in the absence of rRNA18. The degradation and utilization of lipids by starving bacteria is also an intriguing topic as fatty acid has very high energy content per mass despite its low amount in biomass (Table 5). Vast portions of phospholipids were found degraded during starvation by V. cholerae45; also interference of fatty acid metabolism is bactericidal and a direction of drug development46. Previous studies in E. coli found expression of fatty acid degradation during adaptation to starvation, and mutants unable to make this adaptation exhibited long-term viability loss16,47. Interaction between Acyl carrier Attorney Docket No.15670-0435WO1 protein and SpoT48, the only enzyme that synthesizes and degrades ppGpp in E. coli, adds another layer of connection. For these reasons, the degradation of RNA and lipids were monitored during the course of starvation in addition to proteins. The data in FIGS.11A-11C show that they did not play major roles in the conditions we focused on in herein, i.e., during the first 24 h of glucose depletion where major proteome remodeling takes place. FIG.11A: Total RNA abundance in WT (black bars), ∆rpoS (medium gray bars), and ∆yeaG (light gray bars) cultures 6 and 24 hours into glucose starvation, as a percentage of RNA abundance at entry into starvation. Total RNA was measured by the perchlorate method49. RNA abundance in glucose steady state (equivalent to the t=0 h, 100% value timepoint here) has previously been established to be ~95 µg / OD / ml12. FIG.11B: Total RNA is a good indicator of ribosome content during exponential growth as most ribosomes are expected to be intact. However, during starvation conditions as RNA is degraded, total RNA abundance does not necessarily reflect the intact ribosome content as a few nicks along the rRNA might render the ribosomes without function without causing measurable reduction in total RNA abundance. To gain an understanding on intact ribosome levels, we additionally measured the abundance of full-length 16S and 23S rRNA (striped bars) by TapeStation, an automated electrophoresis method. The amounts of total RNA (open bar, from FIG.11A) are shown for comparison. The data indicated that the ribosomes were largely intact for WT cells through the first 24 h. FIG.11C: Total fatty acid abundance in WT cultures at the time of glucose depletion. The molar abundance of each fatty acid moiety was quantified by gas chromatograph mass spectrometry. The results were weighted by the molecular weight of each species and summed over all species to yield the total fatty acid mass. Vertical bars represent standard errors. FIGS.12A-12I: Proteome remodeling by WT cells during starvation. FIG. 12A: Change in protein levels, expressed as percent of the total initial protein mass (% ITPM), for wild type cells between the onset of starvation (at 0 h) and 6 hours after glucose runout. Proteins which increased (N=157) or decreased (N=298) by more than 0.01% ITPM are highlighted and protein names are shown for proteins with changes above 0.1% ITPM. Note: 0.01% ITPM corresponds to about 400 copies / cell. FIG.12B: The total abundance of proteins belonging to the five coarse- grained categories over time after entry into starvation. FIGS.12C-12F show a Attorney Docket No.15670-0435WO1 breakdown of each of the five categories shown in FIG.12B into the corresponding subgroups (central dogma (FIG.12C); biosynthesis (FIG.12D); central carbon + energy (FIG.12E); and envelope, motility, transport (FIG.12F)). The “housekeeping, recycling, stress” category in FIG.12B is broken down in more detail, into its subgroups of housekeeping (FIG.12G); stress (FIG.12H); and recycling (FIG.12I). FIG.13: Changes of exemplary proteins in each functional group. For each of the 16 selected protein groups among those shown in FIG.12C-12I, the time courses of the top 4 proteins with the largest change in amount between 1.5 and 6 hours of starvation were plotted. In the translation machinery subgroup, tufAB represented the overall abundance of the elongation factor EF-Tu. FIGS.14A-14O: Proteome remodeling by ΔrpoS cells. Shown are the remodeling of the five main functional categories subgroups (central dogma (FIG. 14A); biosynthesis (FIG.14B); central carbon + energy (FIG.14C); envelope, motility, transport (FIG.14D); and housekeeping, recycling, stress (FIG.14E)) for wild type and ΔrpoS cells during glucose starvation. Proteomics data was collected and processed in the same batch. FIG.14F: The net protein gain at 6 h of starvation, obtained by summing over the abundances of all proteins which gained in abundance compared to exponential growth (onset of starvation) was similar for WT and ΔrpoS cells. Black bars indicate the total gain originating from proteins with net gain in both strains; striped bars indicate contributions from proteins with net gain in one strain but not in the other. FIG.14G: Breakdown of net protein gains unique to either strain (the striped bars in the previous panel) into individual protein groups. Shown are changes in the total abundances of the groups (RpoS regulation (FIG.14H), ribosome hibernation (FIG.14I), recycling (FIG.14J), and carbon catabolism (FIG.14K)) which gained in WT cells (the groups shown in FIG.2D). Also shown are abundances of the 3 known cytoplasmic protease systems: HsIUV (FIG.14L), Lon (FIG.14M), and ClpXAP (FIG.14N). FIG.14O shows the five members of the RpoS-regulon that increased the most during glucose starvation: RpoS-regulon members were defined as the proteins which gained abundance in WT cells by more than 3-fold in WT cells by 6 h after glucose runout, and whose abundance in WT at 6 h was at least 3-fold the abundance in ΔrpoS cells by the same time after glucose runout. Attorney Docket No.15670-0435WO1 FIGS.15A-15E: Measurement of the extent and flux of protein degradation. FIG.15A depicts an illustration of the experimental procedure. To label intracellular protein, exponentially growing cells were inoculated into the same glucose medium as described in FIG.1A, except for the supplement of 100 μM14C- leucine. The cells grew for ~4 doublings until glucose ran out at OD600 ~ 0.4 (t=0). At the desired time for which the extent of protein degradation was to be measured (for example, at 6 h into starvation as illustrated), unincorporated leucine was removed from the culture by washing with medium lacking glucose and leucine. Cells were resuspended in fresh medium without glucose but containing excess (1 mM)12C- leucine to initiate the chase of the radiolabel. Aliquots of the culture were sampled at regular time intervals and precipitated with trichloroacetic acid (TCA). Radioactivity in the TCA supernatant was measured to monitor the release of14C-leucine from protein degradation. The data of FIGS.3C, 3D, and 16 were taken this way, with the wash of14C-leucine and the chase of 12C-leucine occurring at the time of glucose runout (t=0). FIG.15B: For the measurement of “instantaneous” protein degradation flux, the amount of radioactivity in the TCA supernatant was measured at short (15 min) intervals after chase. The results, plotted as triangles, were expressed as percentages of the total radioactivity in the culture at the moment of chase. This represented a lower bound on the fraction of proteins degraded since the time of chase. The line fitted to the triangles indicated a degradation rate of ~3.3% proteins / h. The circles represented the same measurements repeated by adding chloramphenicol (50 g / ml) at the moment of chase (see below). The protein degradation flux values shown in FIG.4A were obtained by further dividing the instantaneous rate as measured in FIGS.15A and 15B by the instantaneous protein abundances (FIG.3A). FIG.15C: The expected dynamics of leucine exchange after the chase. Immediately after chasing the label with excess of non-labelled leucine, all label (circles) was present within proteins. Over time, the label was released into the cytoplasmic pool at a rate that depended on the protein degradation (JD) and synthesis (JS) fluxes. Furthermore, the cytoplasmic leucine pool rapidly equilibrated with the non-labelled excess leucine in the medium (gray circles). Under the assumption that the fluxes Jin and Jout(occurring at timescales of seconds) were much larger than JDand Js(occurring at timescales of hours), reincorporation of released label would be negligible. This assumption was tested by adding chloramphenicol at the moment of Attorney Docket No.15670-0435WO1 chase to inhibit Js. If reincorporation of label into synthesized proteins was appreciable, then chloramphenicol treatment would have resulted in an increased release of label. This was ruled out by data shown in FIG.15B (compare circles and triangles), validating that reincorporation of label was negligible. Thus, the rate of label accumulation measured in FIG.15B could be taken as the protein degradation rate at the time of measurement. FIG.15D: Unlabeled cultures were grown in glucose and14C leucine was added at the moment of glucose depletion. Thus, only proteins synthesized after t=0 are labelled. After 2 hours in starvation, the label was chased with12C leucine allowing the measurement of the degradation rate of the proteins synthesized in starvation. This procedure was also repeated for adding14C leucine 2 h after glucose depletion and chasing with12C leucine 2 h later to obtain the degradation rate of proteins synthesized during t=2-4 h after starvation. FIG.15E: Exponentially growing cultures in the presence of 50 μM14C leucine were shifted to an otherwise identical medium but containing 1 mM12C leucine instead of the label, and degradation was monitored while growth continued. FIGS.16A-16D: Genetic characterization of YeaG. FIG.16A depicts a graphical representation of YeaG’s two putative protein domains: the N-terminal AAA+ ATPase domain associated with diverse cellular and the Eukaryotic-like Serine / Threonine Kinase (eSTK) domain52–55. The Walker B (AAA+ domain) and eSTK functional motifs are annotated at their respective locations (gray line) in each domain. Alanine mutations were made to known functional amino acids in each motif and are indicated by arrows56,57. Mutations E263A (gag to gcg) (SEQ ID NO:1 : GIMEFVEMFKAPI) and K426A (aag to gct) (SEQ ID NO:2 : TRFAFKILSRVF) were introduced into the yeaG gene driven by its own promoter at its native locus on the chromosome. Shown are the extent of protein degradation (FIG.16B) and the relative viability (FIG.16C) during the first 24 hours of starvation for the yeaG.E263A (square, dashed line) and yeaG.K426A (triangle, dotted line) cultures. The data for WT and ∆yeaG are the same as those shown in FIGS.3B, 3C. FIG. 16D: The amount of YeaG protein measured at 0 h and 6 h into starvation (open and filled bars) for the wild type strain, and the two point mutants: E263A, K426A. FIGS.17A-17L: Comparison of proteome remodeling dynamics in WT and ΔyeaG cells. Shown are time courses for the total abundances of the five main protein functional categories (central dogma (FIG.17A); biosynthesis (FIG.17B); central Attorney Docket No.15670-0435WO1 carbon + energy (FIG.17C); envelope, motility, transport (FIG.17D); and housekeeping, recycling, stress (FIG.17E)) for wild type and ΔyeaG cells in glucose starvation. Proteomics data was collected and processed in the same batch. Shown are the dynamics of 7 proteins (raiA (FIG.17F); rmf (FIG.17G); rpoS (FIG.17H); dps (FIG.17I); yeaG (FIG.17J); rbsB (FIG.17K); and aldA (FIG.17L)) with the largest increase in protein abundance within 6 hours of starvation for wild type and ΔyeaG cells. FIGS.18A-18I: Proteome comparison between WT and ΔyeaG cells. FIGS. 18A-18C show proteome adaptation to glucose depletion, measured as change in protein mass (%ITPM) for WT cells (x- axis) and ΔyeaG cells (y-axis) in different time intervals during starvation. To capture changes of different magnitudes and present both the positive and negative changes in a single plot, the plott4d data were transformed with the arc-tangent function, i.e., instead of the change in abundance Δ^^iof a protein ^^, we plotted tan-l(Δ^^i). The arc-tangent scale had the advantage that it was linear with Δ^^ifor small values of Δ^^ibut saturated to a constant for large values of Δ^^i, so that Δ^^i of vastly different orders of magnitude could be shown in the same plot. FIG.18A: The initial adaptation (0 to 1.5 h) was similar in the two strains (data close to the diagonal), consistent with the low level of YeaG expressed in WT cells (FIG.2C). FIG.18B: Between 1.5 h and 6 h, differences were seen in the adaptation pattern as proteins tend to change less in ΔyeaG cells compared to WT cells. FIG.18C: After 6 hours, most of the proteins that decreased in abundance in WT showed a lack of change in ΔyeaG strain (data spread out around negative x- axis). FIGS.18D-18F: Same data as FIGS.18A-18C but highlighting proteins that were quickly turned over in exponential phase according to a recent study which followed the fate of amino acids that were synthesized de novo28; high-lighted proteins have half-life <2 hours for cells growing in carbon-limited chemostat with 3 h doubling times. Changes in the levels of these proteins were well correlated in the two strains, indicating that YeaG was not necessary for their degradation. Instead, YeaG enabled degradation of proteins that were stable during exponential growth. FIG.18G: Protein loss (reduction in protein abundance) for proteins with different cellular localization (according to annotation provided by Ecocyc8) between 1.5 h and 6 h from starvation. Cytoplasmic proteins were strongly reduced, particularly in the WT strain. The abundance of periplasmic and cytoplasmic proteins was much less Attorney Docket No.15670-0435WO1 affected during the same time interval. FIG.18H: According to a recent transcriptomics data of glucose-starved E. coli 58, the mRNA levels of amino acid biosynthetic (AAB) enzymes dropped sharply in the first 1.5 h and remained low through the course of glucose starvation. The plot here was obtained by summing the data of Ref.

[0058] for all genes belonging to the AAB group at each time point. FIG. 18I: Due to their very low mRNA levels and hence the lack of synthesis (after the first 1.5 h), the decrease in the abundance of the AAB proteins could be attributed primarily to protein degradation. Here, we estimated the degradation rate of each AAB protein ^^ as ln[^^i(^^2) / ^^i(^^l)] / (^^2 − ^^l), where ^^i(^^) is the mass of the protein measured by proteomics, and ^^l, ^^2 are two consecutive time points where data were taken. The degradation rates of WT were always larger than the ΔyeaG strain, and both were seen to decrease as the cultures enter deeper into stationarity. Horizontal white lines are medians. Filled boxes indicate 25-75% intervals. Whiskers indicate 10-90% intervals. For comparison, the bold horizontal lines indicate the whole- proteome degradation rate for WT cells, obtained by dividing the measured protein degradation fluxes (dashed line in FIG.4A) by the protein abundances measured at the beginning of each time window (FIG.1D). FIGS.19A-19C: Measurement of protein synthesis flux. FIG.19A: Illustration of the measurement procedure. Cells were grown exponentially until glucose ran out at OD ≈ 0.4 (t=0) as shown in FIG.1A. At the desired time of measurement, 6 h in this example, the culture was pulsed with a small amount of14C- leucine. The cultures were then sampled at regular time intervals and precipitated with TCA. FIG.19B: All of the leucine pulsed in at 6 h (horizontal arrow) was contained in the TCA supernatant initially. As proteins were synthesized, the14C level in the supernatant (gray bars) decreased while that in the precipitate (black bars) increased. The amount of leucine incorporated into proteins was obtained by monitoring the radioactivity in the supernatant; the absolute abundance obtained from the known relation between radioactivity and leucine concentration, and the leucine composition of the proteome. FIG.19C: Wild type cells were grown as described in FIG.19A and14C leucine was pulsed at 6 h of starvation. Samples were collected at short time intervals and precipitated with TCA. The amount of leucine incorporated into proteins, shown on the right axis, was taken to be the total radioactivity pulsed in minus the radioactivity measured in the supernatant, at ~1500 CPM / nmol. From the Attorney Docket No.15670-0435WO1 known leucine content of cells (265 nmol leucine / OD / mL, or 106 nmol for 1 ml of culture at OD ≈ 0.4, FIG.10B) the leucine incorporation rate was converted to a percentage of the initial total protein mass (left axis): Amount of protein synthesized (as % ITPM) at time t = nmol of leucine incorporated at time t per OD⋅ml of the initial culture ÷ 106 nmol / iOD / ml *100. The flux of protein synthesis was the slope of the linear-fit of the data. The data in FIG.4B were obtained in this way for cultures with14C-leucine pulsed at various time after starvation, with the linear fit over data taken for a period of 2 hours at the interval of 0.5 hours for measurements within 12 hours of starvation, and over a period of 4 hours at the interval of 1 hour for measurements made after 12 h of starvation. FIGS.20A-20E: The effect of overflow acetate on starvation response and viability. FIG.20A: Acetate was added in culture media after removing cells via filtration at various times after 2 mM glucose ran out at OD ≈ 0.4 (t=0, FIG.1A) for WT and ∆yeaG cultures.0.86 mM of acetate in the medium at the onset of starvation (due to acetate overflow, ~2 mM / OD12) was taken up in the subsequent 2 hours. This acetate taken up presumably relieved the need for an increased degradation flux compared to synthesis flux as seen in the first 1-2 h of starvation (FIG.4C). This amount of overflow acetate was important for the physiology of the starving cells, as we showed in the next panels by comparing the “runout” treatment studied for most of this work with the “washout” treatment. Here, “washout” refers to pelleting growing cultures at OD600 ≈ 0.4 and resuspending the pellet in fresh medium with no glucose or acetate. FIG.20B: The viability of the cells clearly suffered in the washout treatment compared to runout (solid and dashed line, respectively), both in the first several hours where the rise due to reductive division was reduced after washout, and after 3 days if one compared to viable cell count to the peak count at 6 h. Most of this viability loss by the washout treatment was rescued by adding back the 0.86 mM acetate when washed cells are resuspended with fresh medium (grey line), suggesting that the difference in viability resulted from some long-term effect exerted by the acetate taken up in the first 2 h. Possible effects included the use of acetate to fuel the synthesis of stress proteins, and the storage of acetate as fatty acid (used latter to supplement cell maintenance). FIG.20C: Direct measurement of protein synthesis (FIGS.19A-19C) showed that protein synthesis was drastically reduced for cells subjected to the washout treatment. Measurements of the full-length 16S (FIG.20D) Attorney Docket No.15670-0435WO1 and 23S (FIG.20E) rRNA by TapesStation (see FIGS.11A-11C) showed that the ribosomes were substantially disrupted for cells subjected to the washout treatment, with only 20% of its 16S rRNA intact 24 h after washout, indicating greatly diminished translational machinery under this treatment. Altogether, the data in FIGS. 20B-20E indicated that the uptake of overflow acetate in the first 2 h drastically affected the starvation response. Therefore, we focused our efforts to characterize starvation resulting from runout but not washout. FIGS.21A-21H: ∆aceA strain and acetate supplement. FIG.21A depicts a diagram indicating key acetate-utilizing metabolic pathways, including energy generation via the TCA cycle and oxidative phosphorylation, fatty acid (FA) biosynthesis, and gluconeogenesis via the glyoxylate shunt (grey text, and arrows). Anabolism of AA via the glyoxylate shunt encoded by aceA and aceB. Expression of both genes was repressed during growth on glucose and turned on during the first hours of glucose starvation59. We deleted aceA to remove the cell’s capacity to synthesize AAs from acetate, such that acetate could be used as an energy source (and possibly be stored as fatty acids), but not for cell growth. FIG.21B: OD600of WT and ∆aceA cultures, which grew exponentially on glucose and entered starvation via glucose depletion shown in FIG.1A. The ∆aceA culture did not grow even when supplemented with 5 mM acetate after glucose depletion, validating the inability to grow on acetate. FIG.21C: Varying concentrations of acetate were added to the ∆aceA culture after glucose depletion. Acetate was consumed by the culture despite its lack of growth. The rate of consumption was similar for acetate concentration at 1- 5 mM, indicating that the uptake was not concentration limited. In all subsequent experiments involving acetate-supplemented ∆aceA strains (including those in combination with other mutations studied, such as ∆yeaG, ∆dps, or ∆rpoS), the concentration of acetate was routinely monitored and additional supplementations made as needed such that acetate concentrations always remained above 1 mM. In the absence of acetate supplement, WT and ∆aceA cells behaved similarly: they grew similarly on glucose (FIG.21E), and after washed into fresh medium without glucose or acetate, the viability dropped similarly (FIG.21F) and also suffered similar protein loss (FIG.21D). FIG.21G: Viability of ∆aceA ∆dps cells supplemented with acetate (ac) (filled triangles) was substantially higher than that of ∆dps cells (open triangles, same data as that shown in FIG.3B), but lower than ∆aceA cells Attorney Docket No.15670-0435WO1 supplemented with acetate (filled circles, same data as that shown in FIG.6B). The data indicated that energy supplement could partially rescue the deleterious effect of dps deletion. FIG.21H: Viability of ∆aceA ∆rpoS cells supplemented with acetate (ac) (filled squares) exhibited higher viability that of ∆rpoS cells (open squares, same data as that shown in FIG.3B), but lower than ∆aceA cells supplemented with acetate (filled circles, same data as that shown in FIG.6B). Thus, again energy supplement partially rescued the deleterious effect of rpoS deletion. FIGS.22A-22D: Coordination of AA-ATP conversion with protein synthesis – four scenarios. FIG.22A depicts the optimal scenario where the two fluxes were just balanced. Setting the AA recycling flux too low could reduce protein synthesis due to energy limitation (FIG.22B) while setting the recycling flux too high could deprive ribosomes of substrates and push energy into fueling futile cycles (FIG.22C). A natural coordination mechanism exists based on mass action: If the recycling flux is too low, then the resulting energy limitation would first limit tRNA charging, which comprises half of the 4-ATP-per-bond cost of protein synthesis. The slowdown in tRNA charging would in turn result in the rise of the AA pools. This rise would increase the flux of AA recycling towards energy generation, at least by mass action through increase of the substrate pools. The only requirement of this strategy relying on mass action is that the ^^Ms of the recycling enzymes be larger than those of tRNA synthetases. [Additionally, there could be regulatory mechanisms (transcriptional, translational, or allosteric) to upregulate the expression or activity of the AA recycling enzymes in response to the build-up of AA pools.] In the presence of energy supplement, such as acetate supplement to ∆aceA cells (FIG.21A), mass action was sufficient to direct the flux of AA from protein degradation largely to protein synthesis (FIG.22D) as seen in FIGS.5C, 5D. In this scenario, the residual AA flux going to energy instead of protein synthesis would be given by the ratio of the ^^M for protein synthesis over that for energy catabolism. FIGS.23A-23Q: Proteome remodeling by energy-supplemented ΔaceA cells.5 mM of acetate was added to the ΔaceA cells at the time glucose was depleted, enabling additional energy production but not AA biosynthesis throughout the first 24 h of starvation (see FIGS.21A-21H). FIG.23A: Protein net gain and loss for ΔaceA cells was compared to WT cells. An increase in net protein gain (open gray circles) and a decrease in protein loss (filled gray circles) was seen for the ΔaceA strain, Attorney Docket No.15670-0435WO1 indicating that amino acids from degraded proteins were almost entirely reincorporated into newly synthesized proteins, indicating that AA catabolism was mainly used to generate energy to support protein synthesis, and this catabolism was not needed in the presence of acetate supplement. The adaptation of individual proteins in ΔaceA cells was compared to WT from 0 h to 6 h (FIG.23B) and 6 h to 24 h (FIG.23C), shown in the arc-tangent scale introduced in FIGS.18A-18I. Early adaptation (0 to 6 h) was similar between the two strains; the ΔaceA strain showed additional protein synthesis (upper left quadrant) after 6 hours into starvation. Shown are the dynamics of the five main protein functional categories (central dogma (FIG. 23D); envelope, motility, transport (FIG.23E); biosynthesis (FIG.23F); central carbon + energy (FIG.23G); housekeeping, recycling, stress (FIG.23H)) for wild type (open circles; same data as in FIG.12B), ΔaceA cells (filled circles) and ΔaceAΔyeaG cells (diamonds). The latter strain (see FIGS.24A-24E) was grown similarly as ΔaceA cells with acetate supplementation at the onset of starvation. Shown are the dynamics of ribosome hibernation proteins, Rmf (FIG.23I) and RaiA (FIG.23J), or selected proteins belonging to the RpoS regulon (rpoS (FIG.23K); yeaG (FIG.23L); dps (FIG.23M); katE (FIG.23N); sodC (FIG.23O); osmC (FIG. 23P); and otsA (FIG.23Q). These proteins displayed a mixed behavior in ΔaceA and ΔaceA ΔyeaG cells with acetate supplement: Rmf, SodC, OsmC and OtsA reached much higher levels in ΔaceA cells compared to WT or ΔaceAΔyeaG cells. However, levels of YeaG, Dps and KatE were the highest in wild-type. Finally, the levels of RaiA and RpoS reached similar maximum values across strains, with a decrease at 24 h in the presence of YeaG. FIGS.24A-24K: Proteome comparison of ΔyeaG cells with ΔaceAΔyeaG cells under energy supplement. Shown are the protein abundance of the five main coarse-grained protein groups (central dogma (FIG.24A); envelope, motility, transport (FIG.24B); biosynthesis (FIG.24C); central carbon + energy (FIG.24D); and housekeeping, recycling, stress (FIG.24E). Despite the energy supplementation (via acetate), the proteome dynamics in the ΔaceAΔyeaG strain were similar to that in the ΔyeaG strain, and both strains showed smaller changes compared to WT. Shown are comparisons for four protein groups upregulated at the entry of carbon starvation (ribosome hibernation (FIG.24F); RpoS regulon (FIG.24G); recycling (FIG.24H); and carbon catabolism (FIG.24I). The ΔaceAΔyeaG strain behaved similarly to the Attorney Docket No.15670-0435WO1 ΔyeaG strain, except for a slight decrease in catabolic proteins. Proteome remodeling in ΔaceAΔyeaG and in the ΔyeaG strain, shown here at the individual protein level (arc-tangent scale), were highly similar in both early (0 to 6 h; FIG.24J) and late (6 h to 24 h; FIG.24K) time intervals. FIGS.25A and 25B: Sources of amino acids for energy generation during starvation. FIG.25A depicts a schematic of amino acids degradation pathways. FIG. 25B depicts a sematic of three composite pathways, where the pathways are indicated by the circled numbers. FIG.26: Energy content of biomass components. The ATP derivable from each biomass component, listed in Table 5, was calculated based on the ATP yield of each monomer using Flux Balance Analysis (FBA) and the biomass composition for E. coli cells growing on glucose. The left bar depicts the biomass composition, and the right bar depicts the amount of ATP per gCDW. FIGS.27A-27D: Autophagy-like processes were activated in bacteria under stress. FIG.27A shows rate of degradation of pre-synth proteins from 0 to 6 h of starvation. The amount of protein mass fraction during starvation was measured over time in E. coli (FIG.27B), V. natriegens (FIG.27C) and P. putida (FIG.27D) strains. FIGS.28A-28C: The protein complex YeaG / YcgB / YeaH was required for global proteolysis. E. coli strains HE1090 (which harbored wild-type (WT) yeaG+; “WT”) and HE1091 (which harbored KEIO-knockout yeaG:kan alleles, “ΔyeaG”) were subjected to stress (starvations) for 0 to 6 h. FIG.28A shows the percent of degradation of pre-synth proteins over time of starvation. FIG.28B shows a schematic of the protein complex YeaG / YcgB / YeaH as predicted by Deepmind Alphafold 3. FIG.28C shows the percent viability of the bacterial cells over time of starvation. FIGS.29A-29D: Production of meso-2,3-Butanediol (BDO). FIG.29A depicts a schematic of BDO production by a microbial catalyst. FIG.29B shows the amount of BDO produced over time by WT and ΔyeaG strains. FIG.29C shows the rate of BDO produced over days by WT and ΔyeaG strains. FIG.29D shows that the loss of WT productivity was attributed to the loss of BudC activity / total protein. FIG.30 shows the amount of BDO produced by WT and ΔyeaG in glucose depleted media. Attorney Docket No.15670-0435WO1 FIG.31 shows the rate of BDO production by ΔyeaG bacterial cells where the OD was 0.36 (left panel, see also FIG.29C) compared to BDO production by ΔyeaG bacterial cells where the OD was 2.7 (right panel). FIG.32 shows the amount of BDO produced over time by WT and ΔyeaG strains (as depicted in FIG.29B) compared to the known amount of BDO produced by an E. Coli producer strain with extensive metabolic engineering (see Boecker et al., Microb Cell Fact 20, 63 (2021)). DETAILED DESCRIPTION Generally, bacteria are typically most productive in late exponential growth phase and productivity drops rapidly in stationary phase, even before death kicks in. When used for microbial product production, the bacterial biocatalyst first grows to optimal density before it is then switched to a growth-arrested state during which the product is synthesized. However, a substantial reduction in metabolic activity is often observed after cellular growth arrest, even in the presence of sufficient substrate. A particular problem with production during growth is that nutrients largely go into making bacterial cell mass (and thrown away) rather than the desired microbial products. As such, there is a need to create bacteria that can function in growth- arrested states while retaining high metabolic activity. The present disclosure provides bacterial cells and methods of using thereof to prolong the stationary phase and increase its productivity of target products (e.g., microbial products). As shown herein, the current challenges of microbial product production were resolved, in part, by imposing growth arrest by removing nutrients essential for growth and extending the productivity of non-growing bacteria by deleting bacterial autophagy and supplementing energy. For example, the ΔyeaG bacterial cells used herein for 2,3-BDO production reached 3 / 4 of max yield without strain engineering and, using the methods disclosed herein, even an engineered strain can be predicted to reach max production titer. The key advantages to the compositions and methods of the present disclosure are that the methods work easily with existing producer strains and method of growth arrest (e.g., does not interfere with production during growth); anoxia is alleviated due to growth arrest, enabling scaling to high-density cultures; and physiology-based engineering is emphasized which, optionally, can complement genetic engineering of bacterial strains. Attorney Docket No.15670-0435WO1 Bacteria The present disclosure provides microorganisms for use in synthesizing target products. A “microorganism” refers to eukaryotes and in particular prokaryotes, and more particular bacteria (e.g., a bacterial cell). In some embodiments, the microorganism (e.g., bacterial cell) is a bacterium. In some embodiments, the bacterium can belong to a genus selected from Acidovorax, Acinetobacter, Actinomyces, Alcanivorax, Arthrobacter, Brevibacterium, Bacillus, Clostridium, Corynebacterium, Deinococcus, Dietzia, Escherichia, Gordonia, Marinobacter, Mycobacterium, Micrococcus, Micromonospora, Moraxella, Nocardia, Pseudomonas, Psychrobacter, Rhodobacter, Rhodococcus, Salmonella, Streptomyces, Thalassolituus, Thermomonospora, Vibrio, and Zymomonas. In some embodiments, the bacterium is a species selected from Escherichia coli (E. coli), Bacillus subtilis, Streptomyces spp., Corynebacterium glutamicum, Lactococcus lactis, Pseudomonas putida, Myxococcus xanthus, Thermus thermophilus, Clostridium spp., and Agrobacterium tumefaciens. In some embodiments, the bacterium is a species of Escherichia, Pseudomonas, and Vibrio. In some embodiments, the bacterium can be Escherichia coli, Pseudomonas putida, B. subtilis, or Vibrio natriegens. A bacterium (e.g., a bacterial cell, a bacterial strain) as disclosed herein can be derived from a parent bacterium. A bacterium derived from a parent bacterium refers to a bacterium that has been modified in some manner to deviate from its parental strain. For example, a bacterium disclosed herein can be derived from a parent E. coli K-12 strain. Bacterium discussed herein can be modified by any type of manipulation, or combination of such manipulations, selected from chemical, biochemical or microbial, in particular genetic engineering techniques. In the latter, the bacterium is referred to as a “genetically engineered” or a “genetically modified” bacterium. A bacterium (e.g., a bacterial cell, a bacterial strain) as disclosed herein can be physically or environmentally “altered” or “modified” to express a gene product at an increased or lower level relative to level of expression of the gene product by the starting bacterium (e.g., the parental bacterium). For example, a bacterium can be treated with or cultured in the presence of an agent (chemical or genetic) known or suspected to increase or decrease the transcription and / or translation of a particular Attorney Docket No.15670-0435WO1 gene and / or translation of a particular gene product such that transcription and / or translation are increased or decreased. Alternatively, a bacterium can be cultured at a temperature selected to increase or decrease transcription and / or translation of a particular gene and / or translation of a particular gene product such that transcription and / or translation are increased or decreased. “Genetically modified” refers to a bacterium altered in the above sense by means of genetic engineering techniques available in the art, as for example transformation, mutation, and / or homologous recombination. Methods of genetically altering bacterium are generally known in the art and are acceptable for use herein. See, e.g., in Sambrook et al., MOLECULAR CLONING, A LABORATORY MANUAL (3d ed.2001); Kriegler, GENE TRANSFER AND EXPRESSION: A LABORATORY MANUAL (1990); and CURRENT PROTOCOLS IN MOLECULAR BIOLOGY (Ausubel et al., eds., 2010). In some embodiments, a genetically modified bacterial cell disclosed herein can comprise one or more loss-of-function mutations in and / or a deletion of at least one gene associated with a regulon of global proteolysis, e.g. the RpoS regulon (see, e.g., Schellhorn, Front Microbiol.2020 Sep 17;11:560099 for a review of the RpoS regulon). As one of skill in the art can appreciate, gene nomenclature often depends on the specific bacterium, and a gene with a similar function or a homologous sequence may be named differently in other species of bacterium. The genes of interest herein are identified by the gene nomenclature used for E. Coli (e.g., yeaG, ycgB, yeaH, aceA). As used herein, where the present disclosure refers to a gene using its standard nomenclature in E. coli, it is understood to also encompass the bacterium- specific gene equivalent thereof. A bacterium-specific gene equivalent to an E. coli gene of interest can comprise a similar genomic sequence, encode for a protein having a similar amino acid sequence, and / or encode a protein having an equivalent function to that of the E. coli gene of interest. Methods of identifying bacterium- specific gene equivalents to E. coli genes are generally known in the art (see, e.g., Carhuaricra-Huaman & Setubal (2024). Step-by-Step Bacterial Genome Comparison. In: Setubal, J.C., Stadler, P.F., Stoye, J. (eds) Comparative Genomics. Methods in Molecular Biology, vol 2802. Humana, New York, NY) and are suitable for use herein. For reference, a bacterium-specific gene can be equivalent to: yeaG (Gene Attorney Docket No.15670-0435WO1 ID: 946297; NCBI Reference Sequence: NP_416297.1); ycgB (Gene ID: 946365; NCBI Reference Sequence: NP_415706.1); yeaH (Gene ID: 946296; NCBI Reference Sequence: NP_416298.1); and aceA (Gene ID: 948517; NCBI Reference Sequence: NP_418439.1). In some embodiments, a gene of interest herein can comprise yeaG, ycgB, yeaH, and / or aceA and the gene equivalent thereof specific to any of the bacterium belonging to a genus selected from Acidovorax, Acinetobacter, Actinomyces, Alcanivorax, Arthrobacter, Brevibacterium, Bacillus, Clostridium, Corynebacterium, Deinococcus, Dietzia, Escherichia, Gordonia, Marinobacter, Mycobacterium, Micrococcus, Micromonospora, Moraxella, Nocardia, Pseudomonas, Psychrobacter, Rhodobacter, Rhodococcus, Salmonella, Streptomyces, Thalassolituus, Thermomonospora, Vibrio, and Zymomonas. In some embodiments, a gene of interest herein can comprise yeaG, ycgB, yeaH, and / or aceA and the gene equivalent thereof specific to Bacillus subtilis, Streptomyces spp., Corynebacterium glutamicum, Lactococcus lactis, Pseudomonas putida, Myxococcus xanthus, Thermus thermophilus, Clostridium spp., and Agrobacterium tumefaciens. In some embodiments, a gene of interest herein can comprise the bacterium- specific gene equivalent to yeaG, ycgB, yeaH, and / or aceA in Pseudomonas putida. In some embodiments, a gene of interest herein can comprise the bacterium-specific gene equivalent to yeaG, ycgB, and / or yeaH in Bacillus subtilis, wherein the B. subtilis gene prkA is equivalent to the yeaG gene; the B. subtilis gene yhbH is equivalent to the yeaH gene, and the B. subtilis gene spoVR is equivalent to the ycgB gene. In some embodiments, a gene of interest herein can comprise the bacterium- specific gene equivalent to yeaG, ycgB, yeaH, and / or aceA in Vibrio natriegens, wherein the V. natriegens gene PN96_RS08585 is equivalent to the yeaG gene; the V. natriegens gene PN96_RS08590 is equivalent to the yeaH gene, the V. natriegens gene PN96_RS08595 is equivalent to the ycgB gene, and the V. natriegens gene aceA is equivalent to the aceA gene. The present disclosure provides, in part, that a protein complex comprised of yeaG, ycgB, and yeaH (referred to herein as “the GBH complex”) was required for global proteolysis. In some embodiments, a genetically modified bacterial cell disclosed herein can comprise one or more loss-of-function mutations in at least one gene that comprises the GBH complex. In some embodiments, a genetically modified Attorney Docket No.15670-0435WO1 bacterial cell disclosed herein can comprise one or more loss-of-function mutations in genes yeaG, ycgB, and yeaH. In some embodiments, a genetically modified bacterial cell disclosed herein can comprise a deletion of at least one gene that comprises the GBH complex. In some embodiments, a genetically modified bacterial cell disclosed herein can comprise a deletion of at least one of yeaG, ycgB, and yeaH. In some embodiments, a genetically modified bacterial cell disclosed herein can comprise one or more loss-of-function mutations in the ycgB gene. In some embodiments, a genetically modified bacterial cell disclosed herein can comprise a deletion of the ycgB gene from its genome. In some embodiments, a genetically modified bacterial cell disclosed herein can comprise one or more loss-of-function mutations in the yeaH gene. In some embodiments, a genetically modified bacterial cell disclosed herein can comprise a deletion of the yeaH gene from its genome. In some embodiments, a genetically modified bacterial cell disclosed herein can comprise one or more loss-of- function mutations in the yeaG gene. In some embodiments, a genetically modified bacterial cell disclosed herein can comprise a deletion of the yeaG gene from its genome. In some embodiments, a genetically modified bacterial cell disclosed herein can comprise one or more loss-of-function mutations in the aceA gene. In some embodiments, a genetically modified bacterial cell disclosed herein can comprise a deletion of the aceA gene from its genome. In some embodiments, a genetically modified bacterial cell disclosed herein can comprise one or more loss-of-function mutations in the yeaG gene and the aceA gene. In some embodiments, a genetically modified bacterial cell disclosed herein can comprise a deletion of the yeaG gene and the aceA gene from its genome. In some preferred embodiments, a genetically modified bacterial cell disclosed herein can comprise a E. coli bacterial cell that has a deletion of the yeaG gene from its genome (i.e., E. coli ΔyeaG). In some embodiments, a parental bacterium subject to any of the genetic modification disclosed herein can be an engineered bacteria strain. Any of the engineered bacteria strains known in the art are suitable for use herein as the parental bacterium. Non-limiting examples of engineered bacteria strains and methods of synthesizing said engineered bacteria strains are described in, for example, Chavez- Gonzalez et al., Advances in food bioproducts and bioprocessing technologies. CRC Attorney Docket No.15670-0435WO1 Press, 2019; Ledesma-Amaro et al., Frontiers in Bioengineering and Biotechnology 8 (2020): 221; Singh, ed. Microbial cell factories engineering for production of biomolecules. Academic Press, 2021; and Shah, Microbial Biotechnology: Integrated Microbial Engineering for B3-Bioenergy, Bioremediation, and Bioproducts: Elsevier Science & Technology, 2025. A manipulation of the parental bacterium can result in the manipulated bacterium (e.g., a genetically modified bacterium) having at least one change of a biological feature as compared to the parent bacterium from which it was derived. As an example, the coding sequence of a heterologous enzyme may be introduced into the parent bacterium or a coding sequence of the parent bacterium may be deleted. Such modifications can result in at least one feature added to, replaced in or deleted from said parent bacterium. In some embodiments, a bacterial cell of the present disclosure can maintain the synthesis of a desired target product during stress (e.g., starvation, e.g., glucose starvation) whereas a bacterial cell from the parent bacterium does not exhibit this feature. In some embodiments, a bacterial cell of the present disclosure can synthesize a higher yield of a target product during stress (e.g., starvation, e.g., glucose starvation) compared to a parental bacterial cell. In some embodiments, a bacterial cell of the present disclosure can synthesize at least a 1-fold, 1.5-fold, 2- fold, 2.5-fold, or 3-fold higher yield of a target product during stress (e.g., starvation, e.g., glucose starvation) compared to a parental bacterial cell. In some embodiments, a bacterial cell of the present disclosure can maintain the synthesis of a desired target product during its stationary phase (e.g., while the bacterial cell is not growing). In some embodiments, a bacterial cell of the present disclosure can synthesize a target product for a longer period under during stress (e.g., starvation, e.g., glucose starvation) compared to a parental bacterial cell. In some embodiments, a bacterial cell of the present disclosure can maintain synthesis of a target product for at least about or about 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 20, 25, or 30 days. In some embodiments, a bacterial cell of the present disclosure can maintain synthesis of a target product for at least 6 days. In some embodiments, a bacterial cell of the present disclosure can maintain synthesis of a target product for at least 8 days. In some embodiments, a bacterial cell of the present disclosure can maintain synthesis of a target product for about 1-30 days, about 1-25 days, about 1-20 days, about 1-15 Attorney Docket No.15670-0435WO1 days, about 1-14 days, about 1-13 days, about 1-12 days, about 1-11 days, about 1-10 days, about 1-9 days, about 1-8 days, about 1-7 days, about 1-6 days, or about 1-5 days during stress (e.g., starvation, e.g., glucose starvation). In some embodiments, a bacterial cell of the present disclosure can have a higher viability under during stress (e.g., starvation) compared to a parental bacterial cell. In some embodiments, a bacterial cell of the present disclosure can have a viability of at least about or about 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, or 100% initial colony form units (CFUs). In some embodiments, a bacterial cell of the present disclosure can have a viability ranging from about 40% to about 100% CFUs. In some embodiments, a bacterial cell of the present disclosure can maintain viability during its stationary phase (e.g., while the bacterial cell is not growing). In some embodiments, a bacterial cell of the present disclosure can maintain protein synthesis under during stress (e.g., starvation). In some embodiments, a bacterial cell of the present disclosure can maintain a higher rate of protein synthesis under during stress (e.g., starvation) compared to a parental bacterial cell. In some embodiments, a bacterial cell of the present disclosure can maintain protein synthesis during its stationary phase (e.g., while the bacterial cell is not growing). Provided herein are compositions comprising a bacterial cell of the present disclosure. Such compositions can be aqueous, in particular a composition can comprise at least one bacterial cell and a suitable culture medium. Suitable culture mediums for bacterium are generally known in the field (see, e.g., Vinci & Parekh, eds. Handbook of industrial cell culture: mammalian, microbial, and plant cells. Springer Science & Business Media, 2002; Behera & Varma, Microbial biomass process technologies and management. Basel, Switzerland: Springer International Publishing, 2017) and are suitable for use herein. In some embodiments, a composition can comprise at least one bacterial cell disclosed herein and a culture medium that is essentially free of contaminating microorganisms. Compositions disclosed herein can be used in any of the methods contemplated herein, e.g., microbial catalysis, fermentation. Compositions disclosed herein can be stored frozen (e.g., less than 0°C, e.g., about -20°C to -80°C) and thawed prior to use. Compositions contemplated herein for frozen storage can comprise one or more cryoprotectants known in the art (see, e.g., Alonso, Saúl. "Novel preservation techniques for microbial cultures." Novel food fermentation technologies. Cham: Attorney Docket No.15670-0435WO1 Springer International Publishing, 2016.7-33, Moldenhauer, James R. "Preservation of Cultures for Fermentation Processes." Practical Fermentation Technology (2008): 125-166). Target Products Bacteria (e.g., a bacterial cell disclosed herein, e.g., a bacterial cell comprising one or more loss-of-function mutations in genes yeaG, yeaH, ycgB, and / or aceA as disclosed herein) can be used according to the methods herein as microbial catalysts. Microbial catalysts are used to convert biomass into a desired product, e.g., a target product. Through genetic modification of the bacterial cells disclosed herein, metabolic engineering, evolutionary adaption, and / or optimization of the conditions for the process (such as temperature, nutrition, oxygenation, etc.) can control the bacterial cells herein to produce the precise target product from the available resources in an efficient manner. In some embodiments, a bacterial cell disclosed herein can synthesize a target product during its stationary phase (e.g., when the bacterial cell is not growing). Target products can be synthesized using the bacterial cells disclosed herein according to the methods disclosed herein, e.g., microbial catalysis, fermentation. Non-liming examples of target products that can be produced by a bacterial cell disclosed herein can include amino acids (lysine, glutamic acid), organic acids (succinic acid), biofuels (butanol, fatty acid esters), acetone, industrial enzymes (proteases, amylases, lipases), polymers, bioplastics, and specialty chemicals (organic acids such as citric acid, lactic acid, glutamic acid). In some embodiments, a target product that can be produced by a bacterial cell disclosed herein can be a carbohydrate, an amino acid, a protein, a nucleic acid, or any combination thereof. In some embodiments, a target product that can be produced by a bacterial cell disclosed herein can be a protein, an antimicrobial, an antibiotic, an enzyme, a probiotic, a toxin, a biofuel, an organic acid, an amino acid, a vitamin, a biopolymer, a polysaccharide, a secondary metabolite, or combinations thereof. In some embodiments, a target product that can be produced by a bacterial cell disclosed herein can be a carbohydrate. In some embodiments, a target product that can be produced by a bacterial cell disclosed herein is 2,3-butanediol. In some Attorney Docket No.15670-0435WO1 embodiments, a target product that can be produced by a bacterial cell disclosed herein is 1,4-buytanediol. Methods of Use Bacterial cells of the present disclosure (e.g., a bacterial cell comprising one or more loss-of-function mutations in genes yeaG, yeaH, ycgB, and / or aceA) can be used in a method of producing at least one target product. In general, methods of producing at least one target product can comprise cultivating the bacterial cells in a suitable culture medium. Culturing a population of the bacterial cells disclosed herein can comprise use of a base culture medium that supports the initial cell growth. A feed culture medium can be added to prevent nutrient depletion. The controlled addition of the nutrient directly affects the growth rate of the culture and helps to avoid overflow metabolism and formation of side metabolites. However, methods of synthesizing certain target products using the bacterial cells disclosed herein as microbial catalysts, can require that the culture medium lack at least one essential nutrient for growth. For example, microbial catalysts under nitrogen limitation (and thus starvation) can produce a range of target products, primarily by reallocating cellular resources away from protein synthesis and toward the accumulation of carbon-based storage compounds. This strategy is most commonly applied to produce lipids and carbohydrates (e.g., butanediol; e.g., 2,3-butanediol and / or 1,4-buytanediol). As such, the bacterial cells disclosed herein are uniquely equipped for synthesizing target products when cultivated in culture mediums with minimal to no essential nutrients as the bacterial cells provided herein maintain their activity in stressed environments (e.g., starvation). Non-limiting examples of specific limitation conditions under which bacterial cells disclosed herein can be cultured in include iron limitations, sulphate limitations, nitrogen limitations, potassium limitations, oxygen limitations, phosphorus limitations, carbon limitations, and gradients and combinations thereof. For example, specific iron and / or sulphate limitation can impact the synthesis of iron- sulfur proteins and cytochromes and can manipulate the electron transport chains of the organism. This specific limitation condition can be used alone or in combination with nitrogen and / or phosphorus limitation to increase the production of organic acids including, but not limited to, lactic acid, acetic acid, formic acid, and pyruvic acid. Attorney Docket No.15670-0435WO1 The specific limitation condition of a potassium gradient can be used to generate products of oxidative metabolism. This specific limitation condition can be used alone or in combination with nitrogen and / or phosphorus limitation to increase the synthesis of organic acids including, but not limited to, lactic acid, acetic acid, formic acid and pyruvic acid. The specific limitation condition of oxygen limitation can be utilized to disrupt the redox balance of the organism. Oxygen limitation can be used alone or in combination with nitrogen and / or phosphorus limitation, iron and / or sulfur limitation, and / or potassium limitation to increase the synthesis of organic acids including, but not limited to, lactic acid, acetic acid, formic acid and pyruvic acid. In some embodiments, bacterial cells disclosed herein can be cultured in a suitable culture medium that lacks at least one essential nutrient for growth. In some embodiments, bacterial cells disclosed herein can be cultured in a suitable culture medium that lacks nitrogen, phosphorus, potassium, sulfur, and or carbon. In some embodiments, bacterial cells disclosed herein can be cultured in a suitable culture medium that lacks an amino acid, a nucleotide, and / or a vitamin that is an essential nutrient for growth. In some embodiments, bacterial cells disclosed herein can be cultured in a suitable culture medium that lacks nitrogen. In some embodiments, bacterial cells disclosed herein can be cultured in a suitable culture medium that lacks glucose. In some embodiments, methods disclosed herein can result in the bacterial cells to produce target products wherein the target products are released into the culture medium. Methods of isolating a target product from the culture medium are generally known in the art and are suitable for use herein. Non-limiting examples of methods for isolating the target product from the culture medium can include centrifugation, filtration (e.g., microfiltration, ultrafiltration), flocculation and coagulation, and sedimentation. Optionally, once the target product is recovered from the culture medium the target product can be subjected to product purification. Methods of recovery and purification of products resulting from microbial catalysis and / or fermentation are both generally known in the art (see, e.g., Gupta, K. et al. (2022). Microbial Fermentation: Basic Fundamentals and Its Dynamic Prospect in Various Industrial Applications. In: Verma, P. (eds) INDUSTRIAL MICROBIOLOGY AND BIOTECHNOLOGY. Springer, Singapore) and are suitable for use herein. Attorney Docket No.15670-0435WO1 In some embodiments, the methods provided herein can further include culturing a population of the bacterial cells disclosed herein (e.g., bacterial cells comprising one or more loss-of-function mutations in genes yeaG, yeaH, ycgB, and / or aceA) in a fermentation system. Non-limiting examples of fermentation systems suitable for use with the methods disclosed herein include a single fermenter, multiple fermenters in series, a membrane fermenter, a fixed-bed fermenter, a fluidized-bed fermenter, a single autoclave, multiple autoclaves in series, a plug flow fermenter, a pneumatically agitated fermenter, a gas-lift fermenter with an external loop having forced circulation, a bubble column fermenter, a fixed (packed)-bed column fermenter, a single horizontal fermenter having multiple compartments, and a multistage column fermenter. Each individual fermenter or autoclave of the fermentation system can also be referred to herein as a reactor or bioreactor of the fermentation system. Fermentation is a biochemical process in which microorganisms, such as bacteria (e.g., a bacterial cell disclosed herein, e.g., a bacterial cell comprising one or more loss-of-function mutations in genes yeaG, yeaH, ycgB, and / or aceA as disclosed herein) convert sugars into microbial products (e.g., target products). Fermentation typically encompasses several stages during the incubation or fermentation period, including, but not limited to, inoculation (seeding), growth phase, stationary phase, and death phase. At the inoculation stage, the desired microorganisms are inoculated into the fermentation vessel, often from a pure culture or a seed culture. A seed culture is a small volume of microbial culture used to initiate a fermentation process. It typically consists of a specific strain or combination of strains of bacteria that are selected for their ability to produce desired products or perform specific metabolic activities. In some embodiments, a seed culture comprises at least one bacterial cell capable of maintaining protein synthesis during starvation as disclosed herein. In some embodiments, a seed culture comprises at least one bacterial cell which has one or more loss-of-function mutations in genes yeaG, yeaH, ycgB, and / or aceA. In some embodiments, a seed culture comprises at least one bacterial cell which does not comprise a functional GBH complex. In some embodiments, a seed culture comprises at least one bacterial cell which does not comprise yeaG (e.g., a ΔyeaG bacterial cell). Once the seed culture reaches a certain cell density and / or activity Attorney Docket No.15670-0435WO1 level, it can be transferred to a larger fermentation vessel to inoculate the fermentation medium and start the production process. Microbial growth and fermentation activity can then enter the exponential or growth phase, characterized by rapid cell division and metabolite production. Furthermore, during the stationary phase, as nutrient availability decreases and waste products accumulate, microbial growth slows down, and the population reaches a plateau. Metabolite production continues, but at a reduced rate. Eventually, nutrient depletion, waste accumulation, and other unfavorable conditions can lead to a decline in microbial activity. This may result in cell death, resulting in a reduction in biomass and microbial product production. The present disclosure provides bacterial cells with superior activity during the stationary phase which are capable of generating target products even when the cell population reaches a plateau (e.g., the bacterial disclosed herein cells stop growing and / or their growth rate is minimal (reduced) compared to the rate observed during their growth phase) and / or nutrient availability is limited (e.g., minimal glucose and / or other essential nutrients are present in the station phase). In some embodiments, the bacterial disclosed herein can remain in a stationary phase of the bacterial fermentation process for at least or about 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 days or for more than about 15 days. In some embodiments, the bacterial disclosed herein can remain in a stationary phase of the bacterial fermentation process for about 1-15, 1-14, 1-13, 1-12, 1-11, 1-10, 1-9, 1-8, 1-7, 1-6, 1-5, 1-4, 1-3, or 1-2 days. Industrial fermentation (also called commercial-scale fermentation) is any large-scale microbial process, whether aerobic or anaerobic, in which microbes (e.g., a bacterial cell disclosed herein, e.g., a bacterial cell comprising one or more loss-of- function mutations in genes yeaG, yeaH, ycgB, and / or aceA as disclosed herein) are employed for the mass production of target products. Various scales of fermentations can be conducted depending on the production requirements, ranging from laboratory-scale research (e.g., from milliliters to a few liters) to large-scale commercial production. Large-scale usually includes volumes ranging from hundreds to thousands of liters (L) including but not limited to about 100 L, about 500 L, about 1,000 L, about 2,000 L, about 3,000 L, about 4,000 L, about 5,000 L, about 6,000 L, about 7,000 L, about 8,000 L, about 9,000 L, about 10,000 L, about 20,000 L, about 30,000 L (about 10,000 gallons), about 40,000 L, Attorney Docket No.15670-0435WO1 about 50,000 L, about 60,000 L, about 70,000 L, about 80,000 L, about 90,000 L, about 100,000 L, about 110,000 L, about 120,000 L, about 130,000 L, about 140,000 L, about 150,000 L, about 160,000 L, about 170,000 L, about 180,000 L, about 190,000 L (about 50,000 gallons), about 200,000 L, about 210,000 L, about 220,000 L, about 230,000 L, about 240,000 L, about 250,000 L, about 260,000 L, about 270,000 L, about 280,000 L, about 290,000 L, and about 300,000 L. Several fermentation methods used in various industries for producing a wide range of products, including batch fermentation, fed-batch fermentation, and continuous fermentation. In standard batch fermentation, microorganisms are cultured in a closed vessel (fermenter) with a specific substrate and optimal conditions for growth and metabolism. The fermentation progresses until the substrate is depleted or the desired product concentration is achieved, without any subsequent additions or removals. Fed-batch fermentation is a modification of batch fermentation where additional nutrients or substrates are added to the fermentation vessel during the fermentation process to maintain optimal growth conditions for the microorganisms, leading to higher product concentrations and yields. Continuous fermentation involves the continuous addition of fresh substrate and removal of fermentation products from the fermentation vessel while maintaining steady-state conditions. This method allows for continuous production without the need for stopping and starting the process. The economic viability of industrial fermentations is highly dependent on the final yield and titer of the desired target product and productivity (e.g., the amount of product produced with respect to time) of its biosynthesis. The final yield in industrial fermentation is the amount of desired product obtained after the fermentation process is complete. It indicates the efficiency of the fermentation process and the effectiveness of the microorganism in converting the substrate into the desired product. The final yield is typically expressed in terms of the quantity of the product obtained per unit volume of fermentation broth or per unit weight of substrate used. Achieving a high final yield is important for maximizing production efficiency and reducing production costs in industrial fermentation processes. Yield is often expressed as yield percentage (yield %), calculated as the ratio of the weight of the product (in grams) to the weight of the input carbon source (in grams), multiplied by 100 ((g of product) / (g of input carbon source) × 100). This value represents the Attorney Docket No.15670-0435WO1 percentage of input carbon that is converted into the desired product. The titer of the desired metabolite in industrial fermentation is the concentration or amount of the target compound produced by the microorganisms during the fermentation process. It indicates the productivity of the fermentation process and is usually measured at the end of fermentation. The titer is typically expressed in terms of mass concentration (e.g., grams per liter) or volumetric concentration (e.g., moles per liter) of the desired metabolite in the fermentation broth. Achieving a high titer is essential for maximizing the yield of the desired product and optimizing the overall efficiency of the fermentation process. In some embodiments, the titer is calculated as the amount of product (in grams) produced per liter of fermentation volume ((g of product) / (1 L of fermentation volume)). Productivity in the context of industrial fermentation is the efficiency with which microorganisms produce the desired metabolites or products during the fermentation process. It is a measure of the rate at which the target compound is synthesized or generated per unit of time and resources, typically expressed as grams per liter per hour (g / L / h) or moles per liter per hour (mol / L / h). Productivity is influenced by various factors such as the choice of microorganism, fermentation conditions (e.g., temperature, pH, agitation, aeration), substrate availability, and genetic engineering techniques. High productivity is desirable in industrial fermentation as it directly impacts the efficiency and economics of the production process. In some embodiments, productivity is calculated as the amount of product (in grams) produced per liter of fermentation volume per hour post-inoculation ((g / L of product) / (hours post-inoculation)). The economic viability of industrial fermentation is highly dependent on the final yield and titer of the desired metabolite and productivity (e.g., the amount of target product produced with respect to time) of its biosynthesis. In some embodiments, productivity is not affected by depleting a carbon source (e.g., glucose). In some embodiments, productivity is not affected by depleting a nitrogen source. In some embodiments, productivity is not affected by high cell density of the bacterial cells disclosed herein. The methods described herein can maintain fermentation titer and / or yield even when the microbial catalyst (e.g., a bacterial cell which does not comprise yeaG, yeaH, ycgB, and / or aceA; e.g., a ΔyeaG bacterial cell) is under starvation-induced stress, thereby increasing the productivity compared to Attorney Docket No.15670-0435WO1 other methods. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. Methods and materials are described herein for use in the present invention; other, suitable methods and materials known in the art can also be used. The materials, methods, and examples are illustrative only and not intended to be limiting. All publications, patent applications, patents, sequences, database entries, and other references mentioned herein are incorporated by reference in their entirety. In case of conflict, the present specification, including definitions, will control. Other features and advantages of the invention will be apparent from the following detailed description and figures, and from the claims. EXAMPLES The invention is further described in the following examples, which do not limit the scope of the invention described in the claims. Materials and Methods The following materials and methods were used in the Examples below. Growth and Starvation Media All growth and starvation experiments herein, including the collection of all proteomics samples, were performed in defined MOPS-buffered minimal medium120. The medium includes 40 mM MOPS, 4 mM Tricine (adjusted to pH 7.4 by NaOH), 1.32 mM KH2PO4, 0.1 mM FeSO4, 0.276 mM Na2SO4, 0.5 μM CaCl2, 0.523 mM MgCl2, as well as the following concentrations of micronutrients: 0.03 µM (NH4)6Mo7O24, 4 µM H3BO3, 0.3 µM CoCl2, 0.1 µM CuSO4, 0.8 µM MnCl2, 0.1 µM ZnSO4*7H2O. Osmolarity was adjusted using 50 mM NaCl121.10 mM NH4Cl was provided as the nitrogen source and glucose was added in limiting or excess amounts depending on the experiment as described in the Cell Growth and Starvation section. Cell Growth and Starvation All cell culturing was performed at 37°C in water baths shaking at ~240 r.p.m. To maintain sufficient aeration, 3 ml cultures were grown in 16 mm diameter test Attorney Docket No.15670-0435WO1 tube. For experiments that required larger volumes such as the total RNA assay, proteomics samples, and long-term CFU assays, 10 ml cultures were grown in 20 mm diameter test tubes. Growth rate and viability differences between 3ml and 10ml culture were not distinguishable from noise. To achieve reproducible, steady-state exponential growth prior to starvation, we followed a standard three-stage culturing approach95: seed culture, pre-culture and experimental culture. LB medium was inoculated (seed culture) from a colony on an LB plate and grown for about 6 hours until there was visible cell growth. Cells were diluted between 103- and 104- fold into MOPS medium (pre-culture) with 10 mM glucose such that cultures were in steady-state exponential growth and below an optical density (OD600) of ~0.6 after 12-16 hours. The exponentially growing pre- culture was washed in pre-warmed MOPS medium with no glucose and diluted to 0.02 OD600 in fresh pre-warmed MOPS medium (experimental culture) with 2 mM or 10 mM glucose depending on whether a run-out or wash-out experiment was being performed. Starvation via glucose run-out In glucose run-out experiments, 2 mM glucose was included in the experimental culture such that growth arrested in mid-exponential phase (OD600 ~0.4). Time 0 (t=0) of starvation for all run-out experiments was defined as the minute when cultures were measured to exit exponential growth. Due to the rapid exit from exponential growth in 2 mM glucose cultures at OD600 ~ 0.4, there is an uncertainty of ~5 minutes in the Time 0 value. This uncertainty in time itself is insignificant as our focus is hours after Time 0. However, this uncertainty can introduce up to 10% uncertainty in the OD at Time 0 from one run to another. To avoid this uncertainty affecting the runout OD which all measurements are normalized to, we define the “runout OD” operationally as the OD600reading 15 minutes after the measured Time 0, since by then the increase of OD has surely been arrested. This “runout OD” is also referred to as the “initial OD” or “iOD”. Starvation via Wash For various assays, washing away the growth media was required to remove components of the growth media no longer wanted during starvation. These assays included: RNA abundance, protein abundance, and protein degradation extent and Attorney Docket No.15670-0435WO1 flux. The goal of a wash-out experiment was typically to replicate a run-out condition. For glucose wash-out experiments, 10 mM glucose was included in the experimental culture such that glucose was not limiting during exponential growth. At OD600~0.4, cells were pelleted via centrifugation (9,000 r.p.m., 3 minutes), washed with 1ml of glucose-free fresh MOPS media, pelleted again, and resuspended in the desired volume of glucose-free fresh MOPS media to achieve an OD600~0.4 at entry into glucose starvation, similar to run-outs. If washing was needed but a run-out condition desired, for example in certain experiments involving14C-leucine, 0.86 mM / 0.4 OD600 sodium acetate was added in the post-wash medium to mimic the concentration of acetate due to overflow metabolism. Strain Construction Bacterial strains used in this study are described in Error! Reference source not found.. The wildtype strain is NCM372285,122; all other strains were derived from this parental strain. Single structural genes dps, yeaG, aceA, yeaH, clpP and leuA were individually deleted from NCM3722 by P1 transduction using appropriate kanamycin (Km) resistant mutants (JW0797-1, JW1772-1, JW3975-3, JW1773-1, JW0427-1 and JW0073-1) from Keio Collection as donors. In some cases, the Km resistance gene (kmr) was flipped out from P1 transductants by pCP20 [PMID: 10829079]. This yielded six deletion mutant strains HE961, HE845, NQ1026, HE988, HE1017 and HE1057, in which dps, yeaG, aceA, yeaH, clpP and leuA were deleted, respectively. TABLE 1 Strain name Strain Description Reference NCM3722 Wild type Refs.22, 23 Attorney Docket No.15670-0435WO1 Each double mutant was made by P1 transducing one mutation to a recipient strain harboring the other mutation. Using this strategy, the following double mutants were yielded: (1) HE963 (deleted for yeaG and aceA by transferring the aceA mutation into the yeaG single mutant); (2) HE969 (deleted for dps and aceA by transferring the aceA mutation into the dps single mutant); (3) HE968 (deleted for rpoS and aceA by transferring the aceA mutation into the rpoS single mutant); and (4) HE962 (deleted for dps and yeaG by transferring the yeaG mutation into the dps single mutant). For the lon / hslV mutant, A strain annotated as lon / clpP / hslV triple mutant was obtained from the Markus Basan Lab. PCR and sequencing analyses revealed that this mutant carries the complete clpP gene. Therefore, this so-called triple mutant was actually a lon / hslV double mutant, which is now named HE996 (deleted for lon and hslV). For the clpP / clpX / lon mutant, the contiguous chromosomal region carrying clpP, clpX and lon was deleted using the Lambada-Red method [PMID: 10829079]. The kmr gene was amplified from pKD4 using chimeric oligos ClpP-P1 and Lon-P2. The purified PCR products were electroporated into NCM3722 cells expressing Lambada-Red proteins encoded by pKD46. The Km resistant colonies were confirmed by PCR for the replacement of genes clpP, clpX and lon by kmr. This yielded strain HE1014, in which clpP, clpX and lon are deleted. The triple ribosome hibernation mutant strain HE587 was made by two rounds of P1 transduction, first by combining rmf and hpf mutations in NCM3722 followed by transferring the raiA mutation into the rmf / hpf double mutant. To create the mutations of yeaG_E263A (the 81stcodon GAG changed to GCG) or yeaG_K426A (the 132ndcodon AAG changed to GCT) on chromosome, Attorney Docket No.15670-0435WO1 two major steps were employed. First, the yeaG gene plus its promoter region were deleted. Second, the same DNA fragment harboring the E263A or K426A mutation was moved back to the chromosome at the same yeaG locus. Using the same Lambda-red approach, a 1656-bp chromosomal region, located at -351 to +1305 with respect to the yeaG translation start site, was first replaced by a kmrgene that was then flipped out by pCP20. This yielded strain ∆pyeaG-yeaG. Fusion PCR was used to introduce the E263A alteration to the yeaG gene carried in the pyeaG-yeaG fragment. First, two fragments sharing a 43-bp overlapped region in between, pyeaG-yeaG’_E263A (harboring the E263A mutation at the 3’ end) and E263A_’yeaG (harboring the E263A mutation at the 5’ end), were individually amplified from the NCM3722 genomic DNA. These two fragments were then fused together by PCR. The fused product, pyeaG-yeaG_E263A (that is, the entire pyeaG-yeaG fragment carrying the E263A mutation in yeaG) was ligated with a DNA fragment (carrying a neo gene and the sc101 ori) amplified from the low-copy plasmid pZS24 [PMID 9092630], yielding pZS24_pyeaG-yeaG_E263A. Using this plasmid as the template, the pyeaG-yeaG_E263A fragment plus the upstream neo gene (that is, neo::pyeaG-yeaG_E263A) was amplified. The purified products were introduced into strain ∆pyeaG-yeaG that over-expressed the Lambda-red proteins. Several Km resistant colonies were verified first by PCR and then by sequencing for the integration of pyeaG-yeaG_E263A at the yeaG locus. This yielded strain HE1006, in which the yeaG carries the desired E263A mutation at its native locus. Similarly, the “neo::pyeaG-yeaG_K426A” cassette was constructed and then integrated onto the chromosome, yielding strain HE1008, in which the yeaG carries the K426A mutation. In both strains HE1006 (harboring yeaG_E263A point mutation) and HE1008 (harboring yeaG_K426A point mutation), the modified yeaG gene (with only one codon alteration) is located at its native locus and it has no other mutations except a neo gene (encoding Km resistance) situated immediately upstream of -351. HE1090 and HE1091, the strains used for BDO production as detailed herein, harbored wild-type yeaG+ and KEIO-knockout60yeaG::kan alleles, respectively. Both HE1090 and HE1091 carried three copies of the tetR gene driven from the PLtet promoter61, as well as the plasmid pZA-PLtet:budABC harboring the chloramphenicol Attorney Docket No.15670-0435WO1 resistance gene. To construct the pZA-PLtet:budABC plasmid, the entire operon budABC from the genome of Enterobacter cloacae dissolvens SDM63was inserted downstream of the PtetM2 promoter in the pZA vector62. Viability Assay Viability was measured by counting colony forming units (CFU) on Luria Broth (LB) agar plates. Details of plate preparation were important for reproducible results. Preparation of LB agar plates for viability curves: 80g LB agar powder was mixed with ddH20 to 2 liters. The solution, along with a magnetic stir bar, was autoclaved for 30 minutes at 120°C. Upon the autoclave cooling to ~80°, the LB agar solution was moved onto a stir-plate at room temperature and stirred until the temperature of the media cooled to ~60°C, before agar began to solidify. Near a flame for sterility, 30ml of the LB agar solution was dispensed into sterile 100x15mm petri dishes using a manual 25ml pipette such that no bubbles were present on the plate surface. Plates were left face up, partially covered by a lid until the solution solidified. Lids were placed on the plates, and they were flipped upside down (agar side up, to avoid excess water loss) to dry for ~20 hours at room temperature in stacks of 5 plates each. Any condensation on plate lids was wiped off with a Kimwipe and plates were stored upside down (agar side up) in stacks of 20-25 at 4° in plastic plate sleeves for up to 1 week. Preparation of plates for cell plating: Plates were removed from storage at 4°C ~40 minutes-1hr before cell plating and warmed in a 37°C incubator upside down (agar side up) for 25 minutes. They were then dried face up with lids removed for 5 minutes in a PCR hood. Plates were then transferred with lids to a 37° incubator and stored upside down (agar side up) for ~a few minutes-1hour. Preparation of cells for plating: Cultures at various time into starvation were diluted in glucose-free MOPS media to ~100 CFU / 100µl. Cells were diluted via 10X serial dilution, sequentially mixing 100µl of sample in 900µl of MOPS media with 10mM NH4Cl and no glucose. To yield a final concentration of 100 CFU / 100µl, the final dilution step varied depending on runout OD600and the viability of the cell population at that timepoint based on previous measurements. For each sample, 100 µl of diluted culture was pipetted on to three LB plates at each timepoint to obtain technical replicates. Diluted culture was spread on the plate with 5 glass beads. Attorney Docket No.15670-0435WO1 After spreading, diluted culture absorbed into the plate within minutes. Plates were incubated upside down (agar side up) for 20-24 hours in a 37° incubator until visible colonies formed. Analysis of culture viability: Colonies (CFU) were counted manually. Each biological timepoint plotted throughout the manuscript represents the average of three technical replicates. Biological replicates were performed in separate tubes on different days. Biological replicates are binned into ~2-hour windows. Unless otherwise noted, all data is normalized to CFU at 6 hours into glucose starvation to account for reductive division during the first few hours of runout. Relative Protein Abundance 3 ml cultures were grown in the presence of limiting glucose (2 mM) and 50μM14C-leucine [at 6.7μCi / μmol]. The addition of leucine was confirmed to not affect growth, yield, and survival of E. coli. Furthermore, leucine uptake was comparable between wild type and a leucine auxotroph (HE1057) suggesting that the externally supplied leucine is the only source of leucine incorporated into proteins. Upon glucose runout, cells were washed to remove all external leucine and resuspended in MOPS buffer without glucose and leucine but supplemented with 0.86 mM acetate to make up for the acetate released during growth, unless specified otherwise. At various time intervals 100 μl of the culture was withdrawn and precipitated with 200 μl of 20% Trichloroacetic acid (TCA) for at least 30 minutes on ice. One half of the precipitation reaction was then centrifuged at maximum speed to separate the precipitate and the supernatant. The supernatant and the other half of the reaction were mixed with 2 ml scintillation cocktail (LiquiscintTM, National Diagnostics), and their radioactivity were measured using a Beckman LS6500 counter. The difference between the radioactivity in the total reaction and the supernatant was taken to be the radioactivity of the protein precipitate. This method of subtractive estimation allowed us to bypass the typical need to repeatedly wash the precipitate with acetone126which in our hands lead to 30-60% loss of material. The precipitate radioactivity at various times relative to that at the time of glucose runout was taken as the relative protein abundance. Attorney Docket No.15670-0435WO1 Total RNA and 16S and 23S rRNA Abundance Total RNA and 16S and 23S rRNA abundance in cultures were measured during growth and at various time points during starvation. Total RNA mass in cultures was measured using the perchlorate method118, with minor modifications as described in [Ref.124], yielding total RNA µg / ml culture. Briefly, 10 ml samples were collected by centrifugation (21,000 r.p.m. for 3 minutes), cell pellets washed twice with cold 0.7M HClO4 and digested with 0.3 M KOH for 1 hour at 37°. Cell digestion was halted by addition of cold 3M HClO4, samples centrifuged (21,000 r.p.m. for 3 minutes) and supernatant collected, including supernatant from two pellet washes with cold 0.5M HClO4. Supernatant OD260 was measured using a spectrophotometer. For 16S and 23S rRNA, total RNA was extracted at the desired timepoints using the TRIZOL reagent (ThermoFisher, catalog number 15596018) according to the vendor’s protocol and the samples were evaluated using Agilent TapeStation at the UCSD Institute of Genomic Medicine (IGM) core facility. In both RNA assays, t=0 samples were collected during exponential growth between 0.3 and 0.4 OD600 in order to capture RNA abundance before entering starvation. RNA abundance was normalized to the t=0 OD600. For time points during starvation, RNA abundance was normalized to the OD at growth arrest (iOD defined above), yielding RNA abundance / iOD / ml. RNA abundances / (i)OD / ml are presented as a percentage of RNA abundance / OD / ml at t=0. Fatty Acid Abundance 10 ml cultures were collected during growth and at various time points during starvation via centrifugation (21,000 r.p.m., 3 minutes). For the t=0h time point, samples were collected during exponential growth between 0.3 and 0.4 OD600before entering starvation. Fatty acid analysis was performed at the UCSD Lipidomics Core. Total fatty acid is obtained by adding up the abundances of all fatty acid µg / OD•ml obtained from GC-MS analysis following the method described in119. Preparation of Proteins Samples for Mass Spectrometry For proteomic analysis, cultures were grown using the standard MOPS media and cell growth protocol described above and samples were prepared using a modified version of the method described in138.10 ml of culture were harvested via centrifugation (21,000 r.p.m., 3 minutes) at various time-points during glucose Attorney Docket No.15670-0435WO1 starvation such that visible pellet was observed. T=0, steady-state, samples were collected at OD600~0.4 and cultures entered starvation at OD600~0.4. Cell pellets were washed once with 150 mM NaCl, rapidly frozen on dry ice and stored at -80° C. Cell lysis was performed by adding 75µl of 100% TFA to each cell pellet in a laminar flow fume hood, pipette mixing, and heating pellets on a heat block at 55°C for precisely 5 minutes. After 5 minutes, 2M pH-unadjusted Tris-base was added with brief vortex mixing to neutralize the dissolved cell pellet in a laminar flow fume hood. Lysed samples were stored at -80C. Protein Quantification and ITPM We quantified the absolute protein abundance in starved cells as a fraction of the initial total protein mass (ITPM), which specified the absolute amount of protein referenced to the total protein mass at the onset of starvation, 333 mg / OD / mL or 616 mg proteins / gCDW for NCM3722 cells grown in glucose minimal medium12. Additionally, a fraction of ITPM is also used as a unit for the amount of individual proteins (in FIG.2A and all plots of proteomic data). Conversion of protein mass fractions to cellular concentrations or copy numbers have been described extensively in Ref.

[0013] . Unlike exponential phase, the transition to starvation conditions on which we focus in this work is characterized by an initial increase in copy number (without biomass accumulation) due to reductive division, followed by an approximately constant cell copy number for several days before the onset of the death phase (FIGS.7A, 7B). Thus, changes in protein abundance as measured by fractional ITPM corresponded to changes in the protein copy number per cell, as long as the number of viable cells remains constant. Using the measured total protein abundance at the onset of starvation, 333 mg / OD / mL mentioned above and the measured CFU per initial OD·ml (FIG.7B), we converted the protein mass of each protein from fractional ITPM to copy number per cell based on its molecular weight. For a representative protein 250 residues in size, and using 1.76·109CFU / iOD / mL (the value 6 h into starvation, after reductive division), we obtained the following copy numbers (Table 2): TABLE 2 Fraction of ITPM #proteins / cell 0.01% 414 0.1% 4,140 Attorney Docket No.15670-0435WO1 For proteins of different sizes, the conversion to copy numbers is changed in inverse proportion to the mass of the protein, see Table 3 below for a protein at 0.1% ITPM: TABLE 3 Protein size #proteins / cell 2,3-BDO production Media and Growth The minimal medium used in this study was a phosphate-buffered medium with pH≈6.2 containing 80 mM KH2PO4, 20 mM K2HPO4, 0.5 mM Na2SO4, 10 mM NaCl, 0.1 μM MnCl2, 0.1 μM CoCl2, 20 μM FeSO4, 0.1 μM ZnSO4, 0.014 μM (NH₄)₆Mo₇O₂₄, 0.1 μM NiSO4, 0.1 μM CuSO4, 0.1 μM SeO2, 0.1 μM H3BO4, 50 μM CaCl2, 0.5 mM MgCl2 supplemented with glucose and NH4Cl as carbon and nitrogen sources at the concentrations described for each experiment. Growth and starvation experiments were typically performed using test-tube orflask cultures incubated in a water bath shaker maintained at 37°C and shaking at 250 rpm such that good aeration was kept. 2,3-BDO production at a low cell density HE1090 and HE1091 strains werefirst grown in LB broth containing 25 μg / ml chloramphenicol for 4-6 hours, and then overnight in the minimal media containing 20 μg / ml chloramphenicol, 40 mM glucose, and 10 mM NH4Cl. The cells were harvested, washed in the minimal medium containing 20 μg / mL chloramphenicol, 40 mM glucose, and 1.5 mM NH4Cl, and inoculated in the same medium to OD600 = 0.03 and grown until OD600 reached around 0.24. The culture was then diluted 20-fold in the same medium containing chlortetracycline (cTc) to the final concentration of 38 ng / ml and grown until ammonium ran out. Upon growth cessation, sodium acetate was added to thefinal concentration of 30 mM. The time of ammonium depletion, manifested by an abrupt stoppage of OD increase, was taken as “t=0”. Time of all subsequent measurements taken were recorded with respect to this growth arrest time. Attorney Docket No.15670-0435WO1 2,3-BDO production at a high cell density HE1091 strain wasfirst grown in LB broth containing 25 μg / ml chloramphenicol for 4-6 hours, then diluted 40,000-fold in the minimal media containing 20 μg / ml chloramphenicol, 40 mM glucose, 3 mM NH4Cl, and 25 ng / ml cTc, and grown and let starved for ammonium overnight. The cells were harvested, washed in the ammonium-free minimal medium containing 40 mM glucose, and inoculated to OD600 = 2.6-2.7 in the same medium containing 30 mM sodium acetate and 30 mM sodium butyrate. The time of inoculation was taken as “t=0”. Quantitation of 2,3-BDO: The concentration of 2,3-BDO in the extracellular medium was quantified using HPLC (Shimadzu LC-20AB solvent delivery unit connected to a Shimadzu RID-20A refractive index detector). The extracellular medium was collected by transfer of a portion of the culture to a spin X centrifuge filter (Costar; 8169) followed by centrifugation at 10,000 rpm for 1 minute with Eppendorf Centrifuge 5424 and stored at -80°C. The solvent was 0.01 M H2SO4 with aflow rate of 0.4 mL / min. HPLC peaks were analyzed using the python code created with Google Gemini. The peak area was converted to concentration based on a 2,3- BDO standard of known concentrations and normalized to the height of the phosphate peak in the same sample to calibrate for evaporation of the culture. The concentration of glucose in the extracellular medium was quantified enzymatically following manufacturer’s instruction (Sigma; GAHK20) and normalized for evaporation of the culture as described above. Quantitation of 2,3-BDO Dehydrogenase Activity: Cells containing 2,3-BDO dehydrogenase (encoded by budC of Enterobacter cloacae dissolvens SDM) were collected at various timepoints into ammonium starvation by centrifugation at 16,000 x g for 3 minutes followed by discarding the supernatant. Collected cell pellets were flash frozen in dry ice and stored at -80°C until use. To start the assay, cell pellets were suspended in the extraction buffer containing 100 mM potassium phosphate, pH 7.0, 1 mM EDTA, and 0.5 mM PMSF to an OD600 = 2. The cell suspension was sonicated on ice-water for 5 seconds at an amplitude of four of the low mode in an MSE sonicator. The sonication was repeated four times with an interval of 15 seconds on ice- water at each cycle. The extract was clarified by centrifugation at 21,130 x g for 15 minutes at 4°C and the supernatant was transferred and diluted in the extraction buffer. Attorney Docket No.15670-0435WO1 The cell extract at different dilution was added to the reaction mixture containing 80 mM potassium phosphate, pH 7, 0.25 mM NADH, and 25 mM acetoin in a 96-well plate and NADH oxidation was monitored at 28-30°C as decrease in absorbance through a band-passfilter of 365+ / -10 nm with Tecan GENios Pro. The proteins in the extract were quantified with Biuret assay. The 2,3-BDO dehydrogenase activity was calculated as the slope of the rate of acetoin-dependent decrease in A365 against the protein amounts added to the reaction mixture and represented as μmol NADH oxidized per minute per mg soluble proteins. Example 1: Constraints on protein synthesis during starvation. To study the temporal dynamics of starving bacteria in a defined, robust manner, we set up exponentially growing cultures of E. coli K-12 cells to abruptly run out of glucose, the sole carbon source, at a fixed, intermediate optical density (OD, a proxy for biomass density1), such that the increase of biomass is halted within a few minutes and the culture remains fully aerobic and replete with all other nutrient components2(FIGS.1A, 7A). To avoid extensive death and the associated nutrient recycling due to cannibalism3, we used MOPS medium in which the death phase is delayed4(FIG.7B). In these conditions, viability decreased minimally for wildtype cells over the next several days (circles, FIG.1B), after an initial rise due to reductive cell division5–7(FIGS.7C-7I). Viability dropped rapidly in a strain lacking rpoS (squares, FIG.1B), which encoded the general stress sigma factor RpoS8–10. These data reflected the importance of the protective roles offered by the expression of the RpoS regulon in starving bacteria. Protein synthesis requires amino acids (AA) as substrates and ATP as fuel. To understand how these resources are derived and utilized during starvation, we focused on the first 24 hours (h) after glucose depletion where most of the gene expression takes place12,14,15. Since the viability of WT cells was largely maintained during this period (FIG.1B), nutrient recycling from the dead cells in the culture was negligible and resources had to be recycled internally within individual starving cells. Recycling of AAs from protein turnover provides, in principle, an endless source of AAs16–18. De novo AA biosynthesis from recycling other biomass components can also occur; Attorney Docket No.15670-0435WO1 however, quantitative estimates indicated that the capacities of these processes are limited. Deriving energy from biomass components was more challenging. Some AAs derived from protein turnover could be catabolized to generate energy. Non- catabolizable AAs were excreted into the medium (FIG.8A) as illustrated in FIG. 1C, reflecting protein loss which we measured to reach ~25% of the initial total protein mass (ITPM) by 24 h of glucose depletion (circles, FIGS.1D, 9A-9D, 10A, 10B). Using flux-balance analysis (FBA20), we obtained a maximal conversion efficiency of ~8 ATP per AA on average (FIG.8C). To generate the required energy needed for protein synthesis, at 4 ATPs per peptide bond, on average an extra one- half of a protein needed to be degraded for every protein synthesized, a 50% energy surcharge of energy was derived completely from AA catabolism. In reality, the conversion efficiency could easily be 2-fold lower, since many AA recycling enzymes were not expressed during growth on glucose, and the more energy-efficient oxidoreductase and cytochrome were also not fully employed during rapid growth (Table 4). TABLE 4: Calculation of Energy Yield for Individual Amino Acids Residue Residue Residue frequency mass WT Δnuo ΔcyoΔnuoΔc o Attorney Docket No.15670-0435WO1 In brief, maximum yields per amino acid were calculated using Flux Balance Analysis (FBA). The aerobic energy yield depended critically on the efficiency of the electron transport chain, which can vary between 1 to 4 protons per electron19; this efficiency was likely not maximal for E. coli cells grown on glucose14,20. Thus, we compared yields obtained for the default metabolic model (iML151521), able to use the most efficient conversion factor, to those obtained with in-silico mutants lacking either the efficient NADH dehydrogenase NDH-I (encoded by the nuo operon), the efficient cytochrome oxidase BO3 (encoded by the cyo operon), or both. The average ATP yield per amino acid, weighted by the residue frequency in E. coli proteome, varied from about 8 ATP / AA in the most efficient case, to about 4 in the least efficient case. Furthermore, arginine and tryptophan were not used during the early hours of glucose starvation15, consistent with the low levels of enzymes of Arg degradation encoded by the ast operon, and low level of Tryptophanase encoded by tnaA; see FIG.8B. The bottom rows of Table 4 show the numbers recalculated excluding the contribution of these two amino acids. Additional energy could be obtained from the catabolism of other biomass components, particularly from RNA and lipids6,17,21,22. Their smaller biomass contents along with the requirements of RNA and lipids for cellular function and integrity limited the amount they contributed towards energy. For example, to contribute the amount of energy equivalent to the 25% of the protein loss observed, it would take 85% of the total RNA or 130% of the total lipid content (Table 5), while the measured fractional loss of RNA and lipids in the first 24 h were no more than that of the protein loss (black bars, FIG.11A). Thus, we focused on the turnover of proteins as the major supplier of energy in this disclosure. TABLE 5: Calculation of Energy Yield for Biomass Components ATP yield (mmol / g) Energy content(mmol ATP / gCDW)Monomer Fraction f drouonyΔcΔ9 Attorney Docket No.15670-0435WO1 3.0 9 6701810.1 Table 5 summarizes the maximum energy content from each biomass component herein. The biomass composition was taken from Ref.

[0014] , and the stoichiometric ratios of each component (amino acids, nucleotides, lipids) in each macromolecule were used to calculate the ATP yield per macromolecule mass (columns under “ATP yield”; unit is mmol ATP per g of the corresponding macromolecule). The product of ATP yield and biomass abundance yielded the total energy content per gram of cell dry weight (CDW). The energy yields and content depended on the efficiency of the electron transport chain. We report here the yields obtained in four different cases, as explained in Table 4. The DNA energy content was not available in practice as it was not possible to degrade DNA without losing viability. The maximal amounts of proteins synthesizable from the ATP derivable from measured protein loss were plotted in FIG.1E (grey lines), for the maximal conversion efficiency of 8 ATP / AA and one-half of that value. The corresponding amounts of protein degradation needed to sustain such synthesis, obtained as the sum of the synthesis amount and the loss amount, were plotted in FIG.1F (grey lines). These plots illustrated the substantial cost of protein synthesis for carbon-starved cells, involving the degradation of over half of the pre-starved proteome in the first 24 h. We note that this amount of protein turnover, while large in absolute magnitude, could in principle be accommodated by the known rate of proteolysis during exponential growth (2~3% / h23,24) if it was maintained over the first 24 h of starvation. However, the macroscopic extent of protein loss (FIG.1D) appeared to go much beyond the limited targets of the well-studied proteolytic machineries25, Attorney Docket No.15670-0435WO1 involving largely proteins with quality problems and those that were specific subjects of regulated proteolysis26–28. The above energy budget analysis was repeated for ∆rpoS cells, which showed a similar RNA loss (FIG.11A), but a much-reduced protein loss compared to WT cells (squares, FIG.1D). The data suggested much-reduced ATP regeneration and protein synthesis fluxes in ∆rpoS cells (FIG.1E), which reflected a reduced need for energy generation by ∆rpoS cells due to their reduction in protein synthesis (silencing of the RpoS-regulon), or due to their reduced proteolytic capacity. Example 2: Sources of amino acids for protein synthesis during starvation. Upon the abrupt stoppage of carbon influx (FIG.1A), E. coli continued to synthesize proteins for many hours1. A large portion of the AAs needed for protein synthesis was derived from the turnover of existing proteins as established by the data herein. Here we discuss other sources to derive AAs. Glycogen is E. coli’s main carbon storage mechanism. However, glycogen comprises of only 2.5% of the E. coli cell dry weight (CDW) and has been previously found to be consumed within 10 min of starvation2. RNA is the 2nd most abundant macromolecule in E. coli after proteins, comprising ~18% CDW for cells growing exponentially on glucose (Ref. [3], Table 5). However, the conversion of RNA to AA is carbon-inefficient, since only the ribose component of a nucleotide is recycled by E. coli as a carbon substrate4. There were 0.52 mmol nucleotide residues / gCDW (Table 5), and thus 0.52 mmol ribose / gCDW. Assuming that ribose is converted to AA 1-to-1, then the amount of cellular RNA could potentially make 0.52 mmol AA / gCDW. This is equivalent to only 9% of proteins since the amount of AA fixed in proteins is 5.7 mmol / gCDW. Additionally, given that the majority of RNA for exponentially growing cells is ribosomal RNA5and a number of ribosomal proteins rely on binding to rRNA for their stability6,7, the usage of RNA has an extra cost of dissembling (and hence destabilizing) the ribosomes. Lipids and osmolytes (mostly glutamate) are the other catabolizable biomass components. However, they cannot make the full set of 20 AAs after glucose depletion. This arises due to the structure of the AA biosynthesis pathways. Summarizing the pathways enlisted in Ecocyc8, we note that they can be classified into two groups based on their starting points in central carbon metabolism: the 13 Attorney Docket No.15670-0435WO1 “lower” amino acids derived from TCA intermediates or pyruvate (asp and 5 AAs derived from asp (asn, lys, met, thr, ile), glu and 3 AAs derived from glu (gln, prol, arg), and ala, leu, val derived from pyr), and the 7 “upper” amino acids derived from glycolytic intermediates (ser, gly, cys which can be converted from pep, and trp, tyr, phe, his which requires ribose or f6p). Glutamate and Acetyl-coA, derived from the turnover of lipids, are readily converted to TCA intermediates and hence can be used to synthesize the “lower” amino acids. But gluconeogenesis is required to make the upper amino acids. It was previously shown3 that even with a large amount of external acetate, it took E. coli many hours to synthesize the enzymes of gluconeogenesis in order to enable the conversion of the TCA intermediates to the upper AAs. For a starving culture with only limited pools of Ac-coA and glutamate, this becomes very difficult. In principle, it is possible to use the turnover of RNA to synthesize the upper AAs, while using lipids and glutamate to synthesize the lower AAs. For example, the amount of glutamate as an osmolyte is 0.1 mmol / gCDW9. If this amount is used to synthesize the 13 lower AAs, and 0.05 mmol / gCDW of ribose (from 10% RNA turnover) is used to synthesize the 7 upper AAs, then (assuming an approximate 1-to- 1 conversion) this can produce 0.15 mmol / gCDW of a complete set of AAs, thus enabling the synthesis of 2.7% of proteins (which requires 5.6 mmol AA / gCDW as mentioned above). As demonstrated by the data provided herein, this is a very low AA yield compared to the AA flux that E. coli can generate from global proteolysis. Example 3: Sources of amino acids for energy generation during starvation. As shown in FIG.25A, each pathway starts with a catabolizable AA and ends on a TCA intermediate, alpha ketoglutarate (akg), succinate (succ), fumarate (fum), or on pyruvate (pyr) or acetyl coA (ac-CoA). Energy products are shown, with QH2being a shorthand for ubiquinol. In some cases, cycles with transaminase reactions (marked parenthetically by GDH) have been accounted for by an additional NAD(P)H, as will be illustrated in FIG.25B. Enzymes catalyzing each reaction are indicated next to each reaction arrow: those listed in on the same line are all needed for the reaction; those stacked vertically form parallel pathways and are replaceable. Attorney Docket No.15670-0435WO1 Reaction arrows solid where dashed arrows represent alternate pathways not the most efficient from the energetic perspective. We made use of previously published data for E. coli10to classify the enzyme based on their levels during growth on glucose minimal media or on casamino acids (cAA) in the absence of glucose, since the latter requires AA catabolism for energy generation and hence serves as a good benchmark for recycling reactions used in carbon-starved conditions. Enzymes in green are expressed in glucose minimal media to levels comparable (50% or higher) to those on cAA and are therefore immediately available for AA catabolism at the onset of carbon starvation. TCA enzymes, glutamate dehydrogenase (GDH), and the pyruvate dehydrogenase complex (PDH), connecting the TCA intermediates (magenta), are all highly expressed, as well as the important enzyme MaeA, connecting malate to pyruvate. The rest of the enzymes are either expressed at low levels in both glucose and cAA media (protein mass fractions below 0.01% ITPM, grey boxes), or significantly upregulated in cAA media compared to glucose (grey). In the latter case, such strong upregulation suggests that these enzymes play a catabolic role during growth on cAA, but they must be synthesized when glucose is no longer available as the sole carbon source. We further detail in FIG.25B several composite pathways, indicated by the circled numbers. Pathway (1) shows the degradation of glutamine to glutamate using the GS / GOGAT pathway. Although the conversion of glutamine (gln) to glutamate (glu) can in principle be cataylzed by either one of two glutaminases encoded in the E. coli genome, glsA or glsB, they are not expressed. Instead, pathway (1) involves only enzymes that are abundantly expressed during growth11. Pathway (2) in FIG.25B shows the degradation of arg via the AST pathway. The pathway requires a succ-CoA, and yields succinate and glutamate as the products of AstE. While one succ molecule is needed to regenerate succ-CoA at the expense of one ATP, the glu is catabolized to akg via GDH, and then to succ via akg dehydrogenase (the SUC enzymes). This balances the ATP across the pathway (as indicated by the strikethrough), while producing additional energy in terms of NAD(P)H; an additional NADPH is obtained by GDH balancing the transaminase reaction encoded by AstC. The net effect of the pathway is to convert arg to succ and 2 CO2, with the generation of 4 NAD(P)H, as indicated by the abbreviated pathway in FIG.25A. Attorney Docket No.15670-0435WO1 Finally, pathway (3) in FIG.25B shows the possible interconversion of prp- coA to pyruvate using the prp operon, which enables propionate catabolism. However, the low levels of the PrpCDB enzymes in rich media suggest that threonine is degraded via glycine and serine as indicated in FIG.25A. Example 4: Energy content of biomass components. The ATP derivable from each biomass component, listed in Table 5, was calculated based on the ATP yield of each monomer using Flux Balance Analysis (FBA), and the biomass composition for E. coli cells growing on glucose, taken from Ref.

[0014] . The results are summarized as stacked bars in FIG.26 with left bar being the biomass composition, and the right bar being the amount of ATP per gCDW. The ATP derivable from proteins, 44 ATP mmol / gCDW, was about 50% of the total derivable from biomass. ATP derivable from RNA (13 ATP mmol / gCDW) and lipids (14 ATP mmol / gCDW) each comprised about 1 / 3 of the remaining 50%. The ATP yield per nucleotide residue (~25 ATP per nucleotide) was actually 3x higher than that of an average AA (~8 ATP / AA); however, the biomass composition was such that the number of nucleotides was only ~1 / 10 of the number of AAs. Lipids had even higher energy content, but constitute an even smaller fraction of the biomass (Table 5). The energy content of the remaining biomass components was negligible, and the latter were often rapidly depleted upon entering starvation. For example, glycogen was consumed within 10 min of starvation, while the free glutamate pool, comprising a few percent of the total AA in proteins, was depleted within ~20 min. Lipid degradation could follow naturally from “cell dwarfing”16,17but cannot get far ahead of the overall decrease of cellular mass, given the necessity to preserve membrane integrity. Thus, the only expandable biomass component other than proteins was RNA. In fact, RNA could be quickly synthesized de novo by metabolic enzymes once nutrients return, as long as biosynthetic enzymes are present. However, a major drawback of RNA degradation was that the loss of RNA (mostly rRNA) led to the loss of ribosome proteins, since several r-proteins were prone to proteolysis without being in the ribosome complex6,7,18. Thus, the catabolism of RNA for energy has a serious long-term cost, i.e., the loss of r-proteins, which threatens the cells’ long- term viability and recovery. Attorney Docket No.15670-0435WO1 Example 5: Proteome remodeling by WT and ∆rpoS cells. To understand the substantial difference in protein turnover between WT and ∆rpoS cells and its implication on proteome remodeling, we performed proteomic analysis for these and other strains after glucose depletion using data-independent acquisition (DIA) mass spectrometry29,30. The proteomic data collected were expressed in terms of the mass fraction of individual proteins, with abundance calibrated using ribosome profiling data31,32Combining the mass fraction data with the measurements of total protein abundances (FIG.2A), we obtained the time courses of the absolute abundances of over 2000 proteins during the first 24 h of starvation. First, we provided an analysis of the WT cells: Despite the decrease of total protein amount in the WT culture (FIG.1D), ~150 proteins exhibited increases of > 0.01% ITPM (~400 copies per cell) by 6 h after starvation (FIG.12A). Among the proteins with the largest increase at that time were several members of the RpoS- regulon (Dps33,34and YeaG35,36), the ribosome hibernation factor RaiA37,38and several enzymes involved in carbon catabolism. The time courses of the five proteins that were increased the most by amount are shown in FIG.2B. We also found ~300 proteins which decreased by > 0.01 %ITPM 6 h after starvation (FIG.12A). These proteins were dominated by metabolic enzymes and components of the translational apparatus, with the time- courses of the five proteins that decreased the most amount shown in FIG.2C, including MetE, the most abundantly expressed protein during exponential growth on glucose39,40, and glutamate dehydrogenase (GDH), whose degradation during starvation was investigated in great detail41. A simplified view of protein dynamics was gained by following the total abundances of various protein functional groups. We classified proteins into 5 categories, each further divided into 5-6 groups; altogether this classification accounts for ~1200 proteins comprising ~85% of the proteome in steady state growth. The time course of the total protein abundance of each category is shown in FIG. 12B, and the abundance of each group in FIGS.12C-12I, with exemplary group members shown in FIG.13. FIG.2D shows the abundances of the 4 groups that gained abundances in the first 6 h: the RpoS-regulon, ribosome hibernation factors, and carbon catabolism each gained 1-1.5% ITPM, while recycling enzymes gained 0.5% ITPM. Groups that decreased were dominated by amino acid biosynthesis Attorney Docket No.15670-0435WO1 (AAB) enzymes and ribosomal proteins (FIG.2E). The net gain and loss by all detected proteins are shown as circles in FIGS.2F, 2G, respectively, revealing that a lot more proteins have reduced abundances than increased abundances, consistent with the general expectation based on resource allocation (FIGS.1E, 1F). However, it should be noted that net gain / loss provided only lower bounds for synthesis / degradation, since a protein could be synthesized and degraded as elaborated further herein. We repeated the proteome analysis for ∆rpoS cells. The net loss by ∆rpoS cells was much smaller than that of WT cells (squares, FIG.2G). An analysis of the proteome revealed that the loss of ribosomal proteins and biosynthetic enzymes were suppressed in ∆rpoS cells after the first several hours (FIGS.14A, 14B), and this difference accounted for most of the differences in net protein loss. On the other hand, the net protein gain by ∆rpoS cells was surprisingly similar to that of WT (FIG.2F). Analyzing the composition of the synthesized proteome revealed that the two strains differ in which proteins were synthesized (FIGS.14F-14G), with ∆rpoS cells losing the gain in recycling enzymes in addition to not expressing the RpoS regulon, but gaining in biosynthetic and TCA cycle enzymes, possibly reflecting the lack of sigma factor competition42,43in ∆rpoS cells. The similar amounts of proteins gained by ∆rpoS and WT cells demonstrated that there was active protein synthesis in ∆rpoS cells. Thus, the reduced total protein loss observed in ∆rpoS cells (FIG.1D) could not be attributed to reduced protein synthesis. Rather, ∆rpoS cells seemed to have lost the ability to degrade proteins after several hours of starvation. Since the levels of the well-known proteolysis machineries were hardly affected in ∆rpoS cells (FIGS.14L-14N), that data suggested that a member of the RpoS regulon mediated the extensive protein loss found in WT cells during starvation (FIGS.12C-12I). Example 6: YeaG-mediated global proteolysis. We individually deleted dps and yeaG, the most abundantly expressed proteins of the RpoS regulon after glucose depletion44(FIG.2B) and which were not expressed in ∆rpoS cells (FIG.14O). ∆yeaG cells exhibited a marked reduction in protein loss, with total protein abundance similar to ∆rpoS cells and much smaller than WT cells (diamonds, FIG.3A). In contrast, ∆dps cells exhibited protein loss at Attorney Docket No.15670-0435WO1 similar level as WT cells (triangles, FIG.3A), consistent with the known functional role of Dps in protecting the chromosome34,45. As another control, we constructed a strain (∆3R) and deleted of all 3 ribosome hibernation factors (RaiA, Rmf, Hpf), two of which were very highly expressed in early hours of starvation (FIG.2C). Protein loss exhibited by ∆3R cells was similar to WT (FIG.3A). ∆yeaG cells exhibited substantial viability loss compared to WT cells (FIG.3B), similar to the effect of deleting dps, reaching ~2 / 3 of the viability loss of ∆rpoS cells after 3 days. This underscored the importance of YeaG to the survival of starving cells, possibly through its effect on protein loss. To characterize the degradation process more directly, we adapted a radio- labelling method to quantify the extent of total protein degradation46(FIGS.13A- 13C). By pre-labeling proteins with14C-leucine during exponential growth and measuring its release during starvation, we found ~30% of labeled proteins were degraded in WT cells within the first 6 hours of starvation (circles, FIG.3C). ∆yeaG cells showed a similar amount of degradation as WT cells during the first 2 h, reaching 15% degradation by 6 h (diamonds, FIG.3C). A dependence on YeaG was not expected the first 2 h, since proteomic data showed low YeaG levels in the first 1.5 h (FIG.2B). From 2 h to 24 h however, the amount of protein degraded in ∆yeaG cells was 4-5x smaller than that of WT cells, with the latter reaching ~50% of the initial protein amount by 24 h. Protein degradation in the other strains behaved similarly in agreement to their respective total protein loss (FIG.3A), with ∆dps cells and ∆3R cells degrading similarly as WT cells and ∆rpoS cells similarly as ∆yeaG cells. YeaG, previously identified as a serine / threonine kinase35, with an additional AAA+ ATPase domain36,47, was studied mostly for its role in nitrogen starvation36,48. We deleted mutated key residues in each domain36,49,50, and found these point mutants exhibited similar degradation activity as ∆yeaG cells, along with their drops in viability (FIGS.16A-16D), suggesting that both domains were required for YeaG- mediated degradation. To identify additional requirements for YeaG activity, we found deletion of yeaH, the other gene in the operon with yeaG, reduced protein degradation to the same extent as ∆yeaG cells (FIG.3D). Additionally, we employed Alphafold to predict protein interactions. While the co-folding of the YeaG and YeaH proteins yielded a low score, a proteome-wide search highlighted the YeaH-YcgB as Attorney Docket No.15670-0435WO1 a high-confidence pair of interacting proteins. A further Alphafold query showed the three proteins YeaG, YeaH and YcgB forming a trimer, shown in FIG.28B, with YcgB bridging between the YeaG and YeaH subunits. Finally, deletion of these three proteins were found to have the same phenotype in a large-scale survey across bacteria63. Thus, we predicted that YeaG, YcgB and YeaH formed a protein complex, henceforth termed GBH, mediating protein degradation. Of the known cytoplasmic proteases25, ClpP was required for proteolysis while the deletion of Lon and HslV showed little effect (FIG.3D). Efforts to delete all 3 proteases led to an unstable strain. Together, our data suggested the GBH complex was an unfoldase which worked together with ClpP and perhaps some of its partners to carry out the extensive degree of proteolysis observed. Aside from the degradation of protein synthesized during exponential growth (FIGS.3C, 3D), we also characterized the degradation of proteins synthesized during starvation, by adding14C-leucine for a duration after glucose depletion and quantifying the subsequent release of leucine label (FIG.15D). We found continued release of labeled leucine (FIG.3E), indicating that the newly synthesized proteins were unstable themselves, and were degraded at faster rates. The presence of YeaG also enhanced proteolysis during exponential growth. Although hardly detected during growth on glucose, YeaG was elevated during growth on poor carbon sources such as mannose31. Accordingly, we found the rate of protein degradation was reduced for ∆yeaG cells grown in mannose but not in glucose (FIG.3F). We repeated the proteomic analysis for ∆yeaG cells and compared its proteome dynamics with WT cells. The protein functional categories (FIGS.3G, 3H, 17A-17E) showed the lack of remodeling after several hours of starvation, consistent with the overall decrease in proteolysis measured in FIGS.3A, 3C. Plotting changes in protein abundances over different time intervals (FIGS.18A-18C) showed that the diminished protein loss was pervasive across many proteins in ∆yeaG cells after 1.5 h of starvation. Using available rates of protein degradation recently reported during exponential growth24, we found those proteins unstable during exponential growth remained unstable during starvation in ∆yeaG cells (FIGS.18D-18F); thus it was largely those proteins deemed stable during exponential growth that were affected by yeaG deletion during starvation. Sorted according to cellular localization, the proteins affected by yeaG were overwhelmingly cytoplasmic (FIG.18G). Among these, we Attorney Docket No.15670-0435WO1 used the set of AA biosynthesis enzymes to estimate a range of YeaG-dependent degradation rates in WT cells (since they are not synthesized during starvation based on their low mRNA levels (FIG.18H), so that their losses could be equated with degradation). We found a 2-fold spread in degradation rates (4-8% / h) for WT cells during the first 6 h of starvation, declining to 2-5% / h in the next 6 h (FIG.18I); the degradation rates were strongly reduced in ∆yeaG cells. These data indicated that YeaG enabled widespread and largely non-specific degradation of cytoplasmic proteins during starvation. Turning to the proteins groups that increased in starving WT cells (FIG.2D), we note that the gains by the RpoS regulon and ribosome hibernation proteins were diminished in ∆yeaG cells (FIGS.3I, 3J). Comparisons for exemplary members of these groups are shown in FIGS.17F-17L, with Dps and Rmf, two proteins that gained the most in WT cell, losing 50% or more of their gains in ∆yeaG cells. The reduced accumulation of these and other stress proteins, whose deletions strongly affected viability (FIG.3B), were possibly responsible for the viability loss of ∆yeaG cells. Example 7: Relation between protein synthesis and degradation. Why would the deletion of YeaG, a mediator of proteolysis and part of the GBH complex, result in reduction in the accumulation of stress proteins such as Dps, particularly when the RpoS protein itself is increased in ∆yeaG cells (FIG.17)? This question was raised long ago51when a similar effect was found upon the deletion of clpP. Reduced expression of RpoS-regulated genes was also reported as a puzzle in previous studies of ∆yeaG cells under N-starvation36,52. Within the resource allocation framework proposed here (FIG.1C), the answer lies generally in the reliance of (YeaG-dependent) proteolysis to generate resources for protein synthesis. To probe the validity of this framework and answer these questions, we next measured directly the protein degradation and synthesis fluxes and characterized their mutual relationship over the first 24 h of starvation. The radio-labeling method as used herein was modified to measure the “instantaneous” flux of degradation at different time after starvation (FIGS.9A-9C). As shown in FIG.4A, the degradation flux sharply increased from the value during exponential growth (dashed line) upon entering starvation for both WT and ∆yeaG Attorney Docket No.15670-0435WO1 cells; this initial increase was YeaG-independent, possibly resulting from a general problem with protein quality as the cytoplasm condenses shortly after glucose depletion4,53. While the degradation flux of ∆yeaG cells (diamonds) dropped sharply after 1-2 hours towards the basal level, it was maintained in WT cells (circles) for the first 5-6 hours before declining gradually over the course of the next 20 hours. This data was quantitatively consistent with the loss of pre-shift proteins measured in FIG. 3C and with the degradation rates estimated based on AA biosynthesis enzymes (FIG.18I) and additionally provided a refined temporal measure of the boost in proteolysis due to YeaG. We also reversed the radio-labeling protocol to obtain the “instantaneous” synthesis flux (FIGS.19A-19C). As seen in FIG.4B, the synthesis flux of ∆yeaG cells dropped rapidly compared to that of WT cells, reaching the resolution of our measurements soon after 6 h. However, the synthesis flux of WT cells was maintained at sufficiently high level, allowing for a quantitative comparison with the degradation flux. The synthesis and degradation fluxes, denoted as ^^S(^^) and ^^D(^^), respectively, both decreased over time (circles, FIGS.4A, 4B). They were quite similar the first 1-2h, but a gap developed subsequently. A scatter plot of the two fluxes (FIG.4C) revealed a simple linear relation between them except for an aberrant point associated with the first hours of starvation (arising from a small amount of overflow acetate in the medium as explained in FIGS.20A-20E). The linear relation, written as^^D(^^) = (1 + ^^) ⋅ ^^S(^^) + ^^0 (Eq. 1),had a slope of ~1.8, indicating that on average 1.8 proteins were degraded for every protein synthesized, a surcharge of ^^ ≈ 80% for protein synthesis through the first 24 h of starvation. This surcharge was consistent with the minimum surcharge of 50% needed to generate ATP from AA catabolism as described above. If we interpret this 80% surcharge as going completely into the energy need of protein synthesis, then at 4 ATP / peptide bond, it implied that 4 / ^^ ≈ 5 ATP could be generated for every AA catabolized, which was ~60% of the maximum conversion efficiency of ~8 ATP / AA predicted by FBA (FIG.8C). This level of reduction from the maximum ATP conversion efficiency was not surprising given that the NADH-ATP conversion was not expected to be maximal54(Table 4) and not all AA recycling pathways were expressed (FIG.8B). Another feature of the data in FIGS.4A, 4B was the Attorney Docket No.15670-0435WO1 approximately linear time-dependence of the decrease in synthesis and degradation fluxes (dashed lines). These linear forms allowed us to compute the expected total protein loss (line, FIG.4D), which compared very well with the independently measured protein loss (circles), thereby validating both measurements. Additionally, we compared the total proteins synthesized by the culture after a period of time ^^ in starvation, obtained from the time-integral of ^^S(^^), to the net protein gain at time ^^ obtained from proteomic data (FIG.2F). FIG.4E showed that the protein gained at 24 h of starvation is only ~1 / 3 of the proteins synthesized. This indicated a high degree of futile synthesis, i.e., ~2 / 3 of the protein synthesized (~30% ITPM) were degraded, as reflected by a similar gap in the amount of proteins degraded and the net protein loss (FIG.4F). The high degree of futile synthesis was expected mechanistically based on the high degradation rate of proteins synthesized during starvation (FIG.3E) and the non-specific nature of YeaG-dependent degradation (FIG.17I). Thus, the combined effects of the energy surcharge for protein synthesis and the futility of synthesis due to persistent degradation made it very costly to accumulate desired proteins in starving cells, requiring the turnover of 5-6 proteins to gain one protein by 24 h of starvation. Another important feature of the linear relation in FIG.4C was the y- intercept, or the term ^^0in Eq. (1), which is independent of the extent of protein synthesis and could be interpreted as the flux of degraded proteins used to supply the maintenance energy, i.e., an ATP maintenance flux, ^^ = (4 / ^^) ⋅ ^^0. The value of the intercept gave ^^ ≤ 0.1 mmol ATP / gCDW / h, which was ~30x smaller than the maintenance flux extrapolated from exponential growth54,55. The smallness of this maintenance flux compared to the ATP flux for protein synthesis (4 ⋅ ^^S) during the first 24 h showed that protein degradation in this period was almost exclusively geared towards protein synthesis. Example 8: Strategy of flux coordination. The tight linear relation between synthesis and degradation fluxes (FIG.4C) caused us to ask how the two are coordinated. There exists in principle two opposing types flux coordination strategies. A demand-driven strategy in which protein degradation is set by the demand for energy, in this case, the need for protein synthesis (FIG.5A). An opposing strategy is supply-driven, in which synthesis is set Attorney Docket No.15670-0435WO1 by the amount of proteolysis (FIG.5B). These two strategies could be distinguished by perturbing the ATP pool: An injection of energy inhibits protein synthesis in the demand-driven strategy, but increases synthesis in the supply-driven strategy. We devised an approach to implement this energy perturbation: As described in FIG.21A, the glyoxylate shunt is required for biomass generation when E. coli grows on acetate as the sole carbon source56. By deleting aceA, one of the genes encoding the glyoxylate shunt57, E. coli, can use acetate for energy generation via the TCA cycle or convert it to fatty acid, but cannot generate most of the biomass components necessary for growth. Accordingly, ∆aceA cells could not grow on acetate as the sole carbon source (FIG.21B), but acetate was still consumed (FIG. 21C), presumably to generate energy. ∆aceA cells, when depleted of glucose, exhibited protein loss similar to that of WT cells (FIG.21D). When 5 mM of acetate was added to the culture 6 h into glucose starvation, protein loss was abruptly arrested (filled triangles, FIG.5C), reflecting the immediate balance of the degradation and synthesis fluxes as would be expected if the derivation of energy from AA catabolism was relieved by the external energy supplement. This adjustment could involve a combination of reduction in degradation (demand-driven strategy) and increase in synthesis (supply-driven strategy). Direct measurement in FIG.5D showed an increase of the synthesis flux, from an unmeasurable amount to 2~2.5%ITPM / h within 15 min, comparable to the change in the protein loss flux (from ~2.1%ITPM / h to nearly zero) seen in FIG.5C. The result thus supported the supply-driven strategy of flux coordination (FIG.5B). The rapid recovery of protein synthesis flux upon energy supplementation indicated the existence of a dynamic mechanism to quickly balance the allocation of AA derived from protein turnover, toward ATP vs protein synthesis, without requiring new proteins; one possible mechanism based on mass action alone is depicted in FIG. 22. As a separate test of the coordination strategy between synthesis and degradation, we added chloramphenicol to the culture of WT cells 6 h into glucose starvation. The abrupt stoppage of protein synthesis, hence the abrupt relief of the energy demand, led to an arrest of protein degradation in the demand-driven strategy (FIG.5A). Instead, protein degradation only reduced by 10~20% in the next 45 min (FIG.5E), again supporting the supply-driven strategy. Together, these results Attorney Docket No.15670-0435WO1 indicated that not only was YeaG-mediated proteolysis used to fuel protein synthesis, but the amount of this degradation dictated the amount of protein synthesis possible. Interestingly, YeaG-mediated protein loss appeared to stop abruptly after the first 24 h (FIG.5F), which matched with the time dependences of the measured synthesis and degradation fluxes (dashed lines, FIGS.4A, 4B). As the effect of YeaG on viability extended for days after (FIG.3B), these data suggested that YeaG- mediated proteome remodeling was a costly but critical investment that E. coli cells make in this 24-h preparatory period after starvation to improve their long-term viability. Example 9: Effect of global proteolysis on long-term viability. We now turn to the physiological roles of YeaG-fueled protein synthesis on cell viability. The protein synthesized, including the RpoS regulon, ribosome hibernation factors, and recycling enzymes catalyzing the catabolism of AA, nucleotides, and fatty acids (FIGS.12A-12I and17) could contribute to extending cell survival by a variety of means (FIG.6A). Qualitatively, the effects of these factors belong to two classes, those that protect essential cellular components from damage and those that reduce the cost of cell maintenance. For example, Dps compacts the chromosome45and protects it from damage, while ribosome hibernation factors protect the ribosome from degradation58. The recycling enzymes instead reduce the demand on the turnover of biomass components for energy generation by increasing their efficiencies (FIGS.8A-8C). Additionally, global proteolysis could reduce the enzymes involved in futile cycling and thus further reduce the cost of maintenance. It is possible to distinguish these two classes of effects by comparing the effect of energy supplement to long-term viability after glucose depletion. Energy supplement was provided by adding acetate to ∆aceA cells (FIG.21A): In the absence of acetate supplement, ∆aceA cells grew similarly as WT and their viability dropped similarly as WT (FIGS.21E, 21F). With supplement that maintained acetate at several mM in the medium, the viability of ∆aceA cells was clearly improved over that of WT cells after several days (FIG.6B). Further, protein loss was reduced in ∆aceA cells (FIG.6C), consistent with the link between protein loss and energy need (FIG.1C), and the transient rescue of protein loss after acetate addition (FIG.5C). The lack of protein loss in energy-supplemented cells resulted not from reduction in Attorney Docket No.15670-0435WO1 degradation but from an increase in synthesis (FIG.6D), again consistent with the supply-driven strategy of flux coordination (FIG.5B). To see whether the increased viability in energy-supplemented cells was attributable to increased synthesis of stress protein, particularly the protective proteins such as Dps and ribosome hibernation factors, we repeated the proteomic analysis for ∆aceA cells with acetate supplement after glucose depletion (FIGS.23A-23Q). Indeed, many proteins gained abundance compared to WT cells at the individual level (FIGS.23A-23C) and for coarse categories (FIGS.23D-23H, compare open and filled circles). However, a closer look at what proteins increased in gain raised the question about the link to viability: Of the 4 groups that increased for WT cells shown in FIG.2D, while ribosome hibernation proteins clearly increased compared to WT (filled vs open circles, FIG.6E), the RpoS regulon level was mixed (FIG.6F), with some members increasing, and other member, including Dps, decreasing (FIGS.23L- 23Q). Also, while substantial increases went to biomass recycling and carbon catabolism (FIGS.6G, 6H), it was not clear that these proteins were needed in the presence of energy supplement. To further clarify the role of YeaG-mediated proteolysis, we constructed a ∆aceA ∆yeaG double mutant and characterized this strain in glucose starvation with acetate supplement. This strain exhibited a substantial gain in viability over ∆yeaG (and even WT) cells lacking acetate supplement (filled diamonds, FIG.6I), but little loss in total protein as did ∆yeaG cells (FIG.6J). In fact, the proteome composition of ∆aceA ∆yeaG cells was very similar to that of ∆yeaG cells (FIGS.24A-24K). Thus, YeaG and hence YeaG-fueled proteome remodeling appeared unnecessary in extending cell viability in the presence of energy supplement, implicating their functional roles being primarily in helping WT cells to lowering energy demand for cell maintenance in the absence of external energy supplement, e.g., by increasing the ATP conversion efficiency and reducing the futile cycles (FIG.6A). This did not mean that protective enzymes play no role, as ∆aceA ∆dps and ∆aceA ∆rpoS cells exhibited reduced viability compared to ∆aceA cells even in the presence of energy supplement (FIG.6K). Nevertheless, energy supplement reduced the rate of viability drop even for those deletion mutants (FIGS.21G, 21H), indicating that lowering the energy demand for maintenance could be an important function even for these protective enzymes. Attorney Docket No.15670-0435WO1 In growing cells, protein synthesis is fueled by nutrient uptake while proteolysis provides auxiliary services such as quality control and alternative means of regulation25. During carbon-starvation however, proteolysis is the primary means a starving cell has to dispense its reserve resource, which is its own biomass, for protein synthesis and other energy expenditures. Thus, fundamentally proteolysis occupies a position “upstream” of protein synthesis in starving cells. Due to the huge cost of protein synthesis, ~2 proteins degraded for each protein synthesized (FIG.4C), or 5- 6 proteins degraded for each protein gained (FIG.4E), starving cells must resort to global proteolysis to synthesize even a moderate amount (~10% ITPM) of stress proteins to sustain survival (FIG.3B). E. coli managed this proteome remodeling dynamics very carefully, via a supply-driven strategy of resource allocation (FIG. 5B) implemented by the GBH protein complex. The large cost of protein synthesis provided a rationale for why the alternative demand-driven strategy was not selected: if there was a demand from transcription to make 20% ITPM of certain proteins, then a cell would need to turnover over more than 100% of its proteome to meet this demand (and die along the way). Instead, the supply-driven strategy provided cells with a set amount of resources to prepare for long-term survival. Example 10: Methods of using non-growing bacteria to synthesize target products (e.g., microbial products). When cells face stress conditions, such as nutrient deprivation, oxygen deficiency, and endoplasmic reticulum (ER) damage, autophagy is highly induced to maintain metabolic and energy balance. Herein, the present disclosure demonstrates an autophagy-like process that bacteria turn on while under stress (FIGS.27A-27D). Indeed, ~30% cellular proteins were degraded within 6 h of growth arrest (FIG. 27A). Target of proteolysis was very broad, including most biosynthesis enzymes in E. coli (FIG.27B) and was observed to be even faster for V. natriegens (FIG.27C) and P. putida (FIG.27D), which are typically used as chassis organisms for synthetic biology applications. Targets of proteolysis are very broad, including most enzymes in E. coli and even GFP. FIGS.28A-28C show the extent of global protein degradation in E. coli strains NCM3722 (which harbors wild-type (WT) yeaG+; “WT”) and HE845 (which harbors KEIO-knockout yeaG:kan alleles, “ΔyeaG”) that were subjected to stress Attorney Docket No.15670-0435WO1 (starvations) for 0 to 6 h. These data demonstrated that the GBH protein complex was required for global proteolysis. Removing global proteolysis accelerates death (energy provided by recycling); however, viability can be restored by supplying an energy source. As such, the present disclosure showed that a bacterial strain lacking the functional GBH complex (ΔyeaG), when used in addition to an energy supply, promoted bioproduction by non-growing cells. In general, we deleted stress-induced autophagy in a bioproduction strain, which did not affect exponential growth since it is turned on during stress. We induced growth arrest by removing essential nutrient(s) for growth. For example, (1) removing nitrogen (N) and / or phosphorus (P) for the production of carbohydrates; (2) removing phosphorus (P) and / or potassium (K) for the production of amino acids and proteins; (3) removing sulfur (S) for the production of nucleic acids; and (4) removing an essential amino acid, nucleotide, or vitamin for bacteria with natural or synthetic auxotrophy for the corresponding metabolite. Next, we adjust growth-arrested medium to provide for cell maintenance and promote productivity. Finally, we collect the target product for as long as possible. As an example, we applied the method to production of Meso-2,3-Butanediol (a carbohydrate, referred to here below as BDO for brevity). BDO is an industrially relevant chemical used as a precursor of many products. It is synthesized from glucose via pyruvate with stoichiometry of one BDO molecule per glucose molecule and its synthesis is catabolized by the three heterologous enzymes, BudB, BudA, and BudC (see FIG.29A). BDO diffuses out of the cell and is easily detected in the medium. Growth arrest was induced in the bacteria by removing the nitrogen source in the glucose minimal medium. Subsequently, we added acetate (at pH 6.2) to supplement energy and increase the internal pyruvate pool. We monitored BDO accumulation in the medium after growth arrest and compared the effect of yeaG deletion to the ancestor strain. FIG.29B shows that the ΔyeaG strain produced ~2x more BDO than the WT strain. FIG.29C shows that productivity was maintained for ~6 days after growth arrest. FIG.29D shows that the loss of WT productivity was attributed to the loss of BudC and other enzymes. Existing literature (see Boecker et al., Microb Cell Fact 20, 63 (2021)) reports that the maximum BDO yield on glucose achieved was ~80% (after extensive Attorney Docket No.15670-0435WO1 metabolic engineering). FIG.30 shows that the BDO yield was ~60% for the ΔyeaG strain, which was almost twice as that for the WT strain, demonstrating that only the single knock-out allele, ΔyeaG, could achieve ~3 / 4 of the yield attained through extensive strain engineering. Dependence of productivity on culture density can play a role in the scalability of target product production. Particularly as maintaining productivity at high density can be difficult due to anoxia. Non-growing cells avoid anoxia since oxygen consumption can be greatly reduced. In FIG.29C, the OD of the culture of the ΔyeaG strain at t=0 was 0.36. As shown in FIG.31, BDO productivity was maintained by ΔyeaG cells even at 7.5x higher cell density (OD=2.7). With extensive metabolic engineering on microbes, 80-130 g / L of BDO titer, which corresponds to 1.0 to 1.5 M, can be achieved. The data herein demonstrated that one can get a titer of 1.25 M BDO if productivity of ΔyeaG is maintained till OD=50. This amount of BDO produced by ΔyeaG could be further increased by strain engineering or, as indicated by the data herein, increasing glucose uptake, improving flux matching, and the like. In FIG.32, the amount of BDO known to be produced by the producer strain in Boecker et al., was plotted against the BDO production of the WT and ΔyeaG strains as determined herein (see FIG.29B). This comparative analysis demonstrated that if the high activity observed in the bacteria of Boecker et al., (20 mM BDO / iOD / day) was extended to ~8 days for ΔyeaG cells, the titer of 80g / L could be reached at OD=5, merely 2x larger than the density demonstrated herein. References 1. Basan, M. et al. Inflating bacterial cells by increased protein synthesis. Mol. Syst. Biol. 11, 836 (2015). 2. Phaiboun, A., Zhang, Y., Park, B. & Kim, M. Survival kinetics of starving bacteria is biphasic and density-dependent. PLoS Comput. Biol.11, e1004198 (2015). Schink, S. J., Biselli, E., Ammar, C. & Gerland, U. Death Rate of E. coli during Starvation Is Set by Maintenance Cost and Biomass Recycling. Cell Syst 9, 64-73.e3 (2019). 3. Schink, S. et al. Survival dynamics of starving bacteria are determined by ion homeostasis that maintains plasmolysis. Nat. Phys.1–7 (2024). 4. Rockabrand, D., Arthur, T., Korinek, G., Livers, K. & Blum, P. An essential role for the Escherichia coli DnaK protein in starvation-induced thermotolerance, H2O2 resistance, and reductive division. J. Bacteriol.177, 3695–3703 (1995). 5. Nyström, T. Stationary-phase physiology. Annu. Rev. Microbiol.58, 161–181 (2004). 6. Pletnev, P., Osterman, I., Sergiev, P., Bogdanov, A. & Dontsova, O. Survival guide: Escherichia coli in the stationary phase. Acta Naturae 7, 22–33 (2015). Attorney Docket No.15670-0435WO1 7. Mulvey, M. R. & Loewen, P. C. Nucleotide sequence of katF of Escherichia coli suggests KatF protein is a novel sigma transcription factor. Nucleic Acids Res.17, 9979–9991 (1989). 8. Lange, R. & Hengge-Aronis, R. Identification of a central regulator of stationary-phase gene expression in Escherichia coli. Mol. Microbiol.5, 49–59 (1991). 9. Battesti, A., Majdalani, N. & Gottesman, S. The RpoS-mediated general stress response in Escherichia coli. Annu. Rev. Microbiol.65, 189–213 (2011). 10. Groat, R. G., Schultz, J. E., Zychlinsky, E., Bockman, A. & Matin, A. Starvation proteins in Escherichia coli: kinetics of synthesis and role in starvation survival. J. Bacteriol.168, 486–493 (1986). 11. Matin, A. The molecular basis of carbon-starvation-induced general resistance in Escherichia coli. Mol. Microbiol.5, 3–10 (1991). 12. Hengge-Aronis, R. Survival of hunger and stress: the role of rpoS in early stationary phase gene regulation in E. coli. Cell 72, 165–168 (1993). 13. Urchueguía, A. et al. Genome-wide gene expression noise in Escherichia coli is condition-dependent and determined by propagation of noise through the regulatory network. PLoS Biol.19, e3001491 (2021). 14. Gervais, T. et al. E. coli leverages growth arrest to remodel its proteome upon entry into starvation. bioRxiv 2024.02.29.582700 (2024) doi:10.1101 / 2024.02.29.582700. 15. Mandelstam, J. Turnover of protein in growing and non-growing populations of Escherichia coli. Biochem. J 69, 110–119 (1958). 16. Mandelstam, J. The intracellular turnover of protein and nucleic acids and its role in biochemical differentiation. Bacteriol. Rev.24, 289–308 (1960). 17. Reeve, C. A., Bockman, A. T. & Matin, A. Role of protein degradation in the survival of carbon-starved Escherichia coli and Salmonella typhimurium. J. Bacteriol.157, 758– 763 (1984). 18. Link, H., Fuhrer, T., Gerosa, L., Zamboni, N. & Sauer, U. Real-time metabolome profiling of the metabolic switch between starvation and growth. Nat. Methods 12, 1091–1097 (2015). 19. Orth, J. D., Thiele, I. & Palsson, B. Ø. What is flux balance analysis? Nat. Biotechnol. 28, 245–248 (2010). 20. DiRusso, C. C. & Nyström, T. The fats of Escherichia coli during infancy and old age: regulation by global regulators, alarmones and lipid intermediates. Mol. Microbiol.27, 1–8 (1998). 21. Sekar, K. et al. Bacterial Glycogen Provides Short-Term Benefits in Changing Environments. Appl. Environ. Microbiol.86, (2020). 22. Pine, M. J. Turnover of intracellular proteins. Annu. Rev. Microbiol.26, 103–126 (1972). 23. Gupta, M. et al. Global protein turnover quantification in Escherichia coli reveals cytoplasmic recycling under nitrogen limitation. Nat. Commun.15, 5890 (2024). 24. Mahmoud, S. A. & Chien, P. Regulated Proteolysis in Bacteria. Annu. Rev. Biochem. 87, 677–696 (2018). 25. Chung, C. H. & Goldberg, A. L. The product of the lon (capR) gene in Escherichia coli is the ATP-dependent protease, protease La. Proc. Natl. Acad. Sci. U. S. A.78, 4931– 4935 (1981). 26. Flynn, J. M. et al. Overlapping recognition determinants within the ssrA degradation tag allow modulation of proteolysis. Proc. Natl. Acad. Sci. U.S.A.98, 10584–10589 (2001). 27. Flynn, J. M., Neher, S. B., Kim, Y. I., Sauer, R. T. & Baker, T. A. Proteomic discovery Attorney Docket No.15670-0435WO1 of cellular substrates of the ClpXP protease reveals five classes of ClpX-recognition signals. Mol. Cell 11, 671–683 (2003). 28. Gillet, L. C. et al. Targeted data extraction of the MS / MS spectra generated by data- independent acquisition: a new concept for consistent and accurate proteome analysis. Mol. Cell. Proteomics 11, O111.016717 (2012). 29. Ludwig, C. et al. Data-independent acquisition-based SWATH-MS for quantitative proteomics: a tutorial. Mol. Syst. Biol.14, e8126 (2018). 30. Mori, M. et al. From coarse to fine: the absolute Escherichia coli proteome under diverse growth conditions. Mol. Syst. Biol.17, e9536 (2021). 31. Wu, C. et al. Enzyme expression kinetics by Escherichia coli during transition from rich to minimal media depends on proteome reserves. Nat Microbiol 8, 347–359 (2023). 32. Almirón, M., Link, A. J., Furlong, D. & Kolter, R. A novel DNA-binding protein with regulatory and protective roles in starved Escherichia coli. Genes Dev.6, 2646–2654 (1992). 33. Nair, S. & Finkel, S. E. Dps protects cells against multiple stresses during stationary phase. J. Bacteriol.186, 4192–4198 (2004). 34. Tagourti, J., Landoulsi, A. & Richarme, G. Cloning, expression, purification and characterization of the stress kinase YeaG from Escherichia coli. Protein Expr. Purif. 59, 79–85 (2008). 35. Figueira, R. et al. Adaptation to sustained nitrogen starvation by Escherichia coli requires the eukaryote-like serine / threonine kinase YeaG. Sci. Rep.5, 17524 (2015). 36. Wada, A., Yamazaki, Y., Fujita, N. & Ishihama, A. Structure and probable genetic location of a “ribosome modulation factor” associated with 100S ribosomes in stationary-phase Escherichia coli cells. Proc. Natl. Acad. Sci. U. S. A.87, 2657–2661 (1990). 37. Prossliner, T., Skovbo Winther, K., Sørensen, M. A. & Gerdes, K. Ribosome Hibernation. Annu. Rev. Genet.52, 321–348 (2018). 38. Pedersen, S., Bloch, P. L., Reeh, S. & Neidhardt, F. C. Patterns of protein synthesis in E. coli: a catalog of the amount of 140 individual proteins at different growth rates. Cell 14, 179–190 (1978). 39. González, J. C., Peariso, K., Penner-Hahn, J. E. & Matthews, R. G. Cobalamin- independent methionine synthase from Escherichia coli: a zinc metalloenzyme. Biochemistry 35, 12228–12234 (1996). 40. Maurizi, M. R. & Rasulova, F. Degradation of L-glutamate dehydrogenase from Escherichia coli: allosteric regulation of enzyme stability. Arch. Biochem. Biophys. 397, 206–216 (2002). 41. Farewell, A., Kvint, K. & Nyström, T. Negative regulation by RpoS: a case of sigma factor competition. Mol. Microbiol.29, 1039–1051 (1998). 42. Gruber, T. M. & Gross, C. A. Multiple sigma subunits and the partitioning of bacterial transcription space. Annu. Rev. Microbiol.57, 441–466 (2003). 43. Weber Harald, Polen Tino, Heuveling Johanna, Wendisch Volker F. & Hengge Regine. Genome-Wide Analysis of the General Stress Response Network in Escherichia coli: σS- Dependent Genes, Promoters, and Sigma Factor Selectivity. J. Bacteriol.187, 1591– 1603 (2005). 44. Martinez, A. & Kolter, R. Protection of DNA during oxidative stress by the nonspecific DNA-binding protein Dps. J. Bacteriol.179, 5188–5194 (1997). 45. Pine, M. J. Steady-state measurement of the turnover of amino acid in the cellular proteins of growing Escherichia coli: existence of two kinetically distinct reactions. J. Attorney Docket No.15670-0435WO1 Bacteriol.103, 207–215 (1970). 46. Sultan, A. et al. Phosphoproteome Study of Escherichia coli Devoid of Ser / Thr Kinase YeaG During the Metabolic Shift From Glucose to Malate. Front. Microbiol.12, 657562 (2021). 47. Brown, D. R., Barton, G., Pan, Z., Buck, M. & Wigneshweraraj, S. Nitrogen stress response and stringent response are coupled in Escherichia coli. Nat. Commun.5, 4115 (2014). 48. Ogura, T. & Wilkinson, A. J. AAA+ superfamily ATPases: common structure--diverse function. Genes Cells 6, 575–597 (2001). 49. Hänzelmann, P. & Schindelin, H. Structural Basis of ATP Hydrolysis and Intersubunit Signaling in the AAA+ ATPase p97. Structure 24, 127–139 (2016). 50. Stephani, K., Weichart, D. & Hengge, R. Dynamic control of Dps protein levels by ClpXP and ClpAP proteases in Escherichia coli. Mol. Microbiol.49, 1605–1614 (2003). 51. Switzer, A. et al. A novel regulatory factor affecting the transcription of methionine biosynthesis genes in Escherichia coli experiencing sustained nitrogen starvation. Microbiology 164, 1457–1470 (2018). 52. Shi, H. et al. Starvation induces shrinkage of the bacterial cytoplasm. Proc. Natl. Acad. Sci. U. S. A.118, e2104686118 (2021). 53. Mori, M., Cheng, C., Taylor, B. R., Okano, H. & Hwa, T. Functional decomposition of metabolism allows a system-level quantification of fluxes and protein allocation towards specific metabolic functions. Nat. Commun.14, 4161 (2023). 54. Mori, M., Hwa, T., Martin, O. C., De Martino, A. & Marinari, E. Constrained Allocation Flux Balance Analysis. PLoS Comput. Biol.12, e1004913 (2016). 55. Cronan, J. E., Jr & Laporte, D. Tricarboxylic Acid Cycle and Glyoxylate Bypass. EcoSal Plus 1, (2005). 56. Maloy, S. R. & Nunn, W. D. Genetic regulation of the glyoxylate shunt in Escherichia coli K-12. J. Bacteriol.149, 173–180 (1982). 57. Prossliner, T., Gerdes, K., Sørensen, M. A. & Winther, K. S. Hibernation factors directly block ribonucleases from entering the ribosome in response to starvation. Nucleic Acids Res.49, 2226–2239 (2021). 58. Soupene, E. et al. Physiological studies of Escherichia coli strain MG1655: growth defects and apparent cross-regulation of gene expression. J. Bacteriol.185, 5611–5626 (2003). 59. Brown, S. D. & Jun, S. Complete Genome Sequence of Escherichia coli NCM3722. Genome Announc.3, (2015). 60. Baba, T., Ara, T., Hasegawa, M., Takai, Y., Okumura, Y., Baba, M., Datsenko, K.A., Tomita, M., Wanner, B.L. & Mori H. (2006). Construction of Escherichia coli K-12 in- frame, single-gene knockout mutants: the Keio collection. Mol Syst Biol.2:2006.0008. 61. Lutz, R. & Bujard, H. (1997). Independent and Tight Regulation of Transcriptional Units in Escherichia coli via the LacR / O, the TetR / O and AraC / I1-I2 regulatory elements. Nucl Acids Res.25: 1203–10. 62. Xu, Y., Wang, A., Tao, F., Su, F., Tang, H., Ma, C. & Xu, P. (2012). Genome sequence of Enterobacter cloacae subsp. dissolvens SDM, an efficient biomass- utilizing producer of platform chemical 2,3- butanediol. J Bacteriol.194: 897-8. 63. Price MN, Wetmore KM, Waters RJ, Callaghan M, Ray J, Liu H, Kuehl JV, Melnyk RA, Lamson JS, Suh Y, Carlson HK, Esquivel Z, Sadeeshkumar H, Chakraborty R, Zane GM, Rubin BE, Wall JD, Visel A, Bristow J, Blow MJ, Arkin AP, Deutschbauer Attorney Docket No.15670-0435WO1 AM. Mutant phenotypes for thousands of bacterial genes of unknown function. Nature. 2018 May;557(7706):503-509. OTHER EMBODIMENTS It is to be understood that while the invention has been described in conjunction with the detailed description thereof, the foregoing description is intended to illustrate and not limit the scope of the invention, which is defined by the scope of the appended claims. Other aspects, advantages, and modifications are within the scope of the following claims.

Claims

Attorney Docket No.15670-0435WO1 WHAT IS CLAIMED IS:

1. A bacterial cell comprising one or more loss-of-function mutations in genes yeaG, yeaH, ycgB, and / or aceA, or bacterium-specific gene equivalent thereof, wherein the bacterial cell maintains synthesis of at least one target product during stress.

2. The bacterial cell of claim 1, wherein the bacterial cell is from a bacterium, wherein the bacterium belongs to a genus selected from the group consisting of Acidovorax, Acinetobacter, Actinomyces, Alcanivorax, Arthrobacter, Brevibacterium, Bacillus, Clostridium, Corynebacterium, Deinococcus, Dietzia, Escherichia, Gordonia, Marinobacter, Mycobacterium, Micrococcus, Micromonospora, Moraxella, Nocardia, Pseudomonas, Psychrobacter, Rhodobacter, Rhodococcus, Salmonella, Streptomyces, Thalassolituus, Thermomonospora, Vibrio, and Zymomonas.

3. The bacterial cell of claim 2, wherein the bacterial cell is from a bacterium selected from the group consisting of Escherichia coli, Bacillus subtilis, Streptomyces spp., Corynebacterium glutamicum, Lactococcus lactis, Pseudomonas putida, Myxococcus xanthus, Thermus thermophilus, Clostridium spp., Agrobacterium tumefaciens, and Vibrio natriegens.

4. The bacterial cell of claim 3, wherein the bacterial cell is from a bacterium selected from the group consisting of Escherichia coli (E. coli), Pseudomonas putida (P. putida), Bacillus subtilis (B. subtilis), and Vibrio natriegens (V. natriegens).

5. The bacterial cell of any one of claims 1-4, wherein the bacterial cell is: (i) an E. coli cell comprising one or more loss-of-function mutations in genes yeaG, yeaH, ycgB, and / or aceA; (ii) an P. putida cell comprising one or more loss-of-function mutations in genes equivalent to yeaG, yeaH, ycgB, and / or aceA; (iii) a B. subtilis cell one comprising or more loss-of-function mutations in genes prkA, yhbH, and / or spoVR; orAttorney Docket No.15670-0435WO1 (iv) an V. natriegens cell comprising one or more loss-of-function mutations in genes PN96_RS08585, PN96_RS08590, PN96_RS08595, and / or aceA.

6. The bacterial cell of any one of claims 1-5, wherein the one or more loss-of- function mutations in genes yeaG, yeaH, ycgB, and / or aceA, or bacterium- specific gene equivalent thereof, comprises a deletion of the yeaG, yeaH, ycgB, and / or aceA gene, or bacterium-specific gene equivalent thereof.

7. The bacterial cell of any one of claims 1-6, wherein the bacterial cell is an E. coli cell comprising a deletion of the yeaG, yeaH, and / or ycgB gene.

8. The bacterial cell of any one of claims 1-6, wherein the one or more mutations occurs in (i) a gene that comprises the complex of yeaG, yeaH, and ycgB; or (ii) a gene whose activity is affected by the complex of yeaG, yeaH, and ycgB.

9. The bacterial cell of any one of claims 1-8, wherein the bacterial is a cell from an E. coli K-12 strain.

10. The bacterial cell of any one of claims 1-8, wherein the bacterial is a cell from an engineered strain of E. coli.

11. The bacterial cell of any one of claims 1-10, wherein the stress occurs during a stationary phase of product synthesis.

12. The bacterial cell of claim 11, wherein the bacterial cell does not grow during the stationary phase of product synthesis.

13. The bacterial cell of any one of claims 1-12, wherein the stress occurs during starvation.

14. The bacterial cell of claim 13, wherein the starvation comprises carbon starvation, nitrogen starvation, phosphorus starvation, sulfur starvation, iron starvation, oxygen starvation, amino acid starvation, or combinations thereof.

15. The bacterial cell of claim 13, wherein the starvation is glucose starvation.Attorney Docket No.15670-0435WO1 16. The bacterial cell of any one of claims 1-15, wherein the bacterial cell maintains synthesis of the at least one target product for 1-14 days during stress.

17. The bacterial cell of any one of claims 1-16, wherein the bacterial cell has viability at about 40% to about 100% of initial colony form units (CFUs).

18. A composition comprising the bacterial cell of any one of claims 1-17.

19. A method of producing at least one target product using the bacterial cell of any one of claims 1-17, or the composition of claim 18, the method comprising: (a) cultivating the bacterial cells in a suitable culture medium under conditions that permit the bacterial cells to produce the at least one target product, wherein the at least one target product is released into the culture medium; and (b) isolating at least one target product from the culture medium.

20. The method of claim 19, wherein the suitable culture medium lacks at least one essential nutrient for growth.

21. The method of claim 20, wherein the at least one essential nutrient for growth is selected from the group consisting of nitrogen, phosphorus, potassium, sulfur, carbon, an amino acid, a nucleotide, and a vitamin.

22. The method of claim 20, wherein the at least one essential nutrient for growth is glucose.

23. The method of any one of claims 19-22, wherein step (a) comprises cultivating the bacterial cells for about 1 to about 15 days.

24. The method of claim 23, wherein step (a) comprises cultivating the bacterial cells for at least 6 days.

25. A method of increasing at least one target product, the method comprising: (a) growing a plurality of the bacterial cells of any one of claims 1-17 in a glucose-limited culture medium;Attorney Docket No.15670-0435WO1 (b) adding acetate into the glucose-limited culture medium once glucose is depleted from the glucose-limited culture medium; and (c) culturing the plurality of the bacterial cells for about 1 to about 15 days in the acetate culture medium.

26. A method any one of claims 19-25, wherein the method is a bacterial fermentation process.

27. The method of claim 26, wherein the bacterial fermentation process is selected from the group consisting of batch fermentation, fed-batch fermentation, and continuous fermentation.

28. The method of claim 26 or 27, wherein the bacterial cell remains in a stationary phase of the bacterial fermentation process for about 1 to about 15 days.

29. The method of claim 28, wherein the bacterial cell remains in a stationary phase of the bacterial fermentation process for at least 6 days.

30. The bacterial cell of any one of claims 1-17, the composition of claim 18, or the method of any one of claims 19-29, wherein the at least one target product is selected from the group consisting of a carbohydrate, an amino acid, a protein, and a nucleic acid.

31. The bacterial cell of any one of claims 1-17, the composition of claim 18, or the method of any one of claims 19-29, wherein the at least one target product is a protein, an antimicrobial, an antibiotic, an enzyme, a probiotic, a toxin, a biofuel, an organic acid, an amino acid, a vitamin, a biopolymer, a polysaccharide, a secondary metabolite, or combinations thereof.

32. The bacterial cell of any one of claims 1-17, the composition of claim 18, or the method of any one of claims 19-29, wherein the at least one target product is 2,3-butanediol or 1,4-butanediol.

Images