Spontaneous TisB toxin expression modulates Escherichia coli metabolic adaptation during growth transitions

Abstract

Escherichia coli TisB/IstR1 is a type I toxin–antitoxin system. The TisB toxin targets the inner membrane and disrupts the proton motive force (PMF). Under normal growth conditions, tisB transcription is repressed by the LexA SOS repressor, and IstR1 blocks its translation. DNA damage lifts LexA repression, allowing tisB mRNA to overcome IstR1 inhibition and produce TisB. Although spontaneous tisB expression has been reported, its physiological significance remains unclear. We show that tisB is spontaneously induced and TisB is synthesized during the transition to stationary phase, coinciding with decreased LexA repression and a transient decline in istR1 promoter activity. Deletion of tisB does not affect exponential growth but delays entry into and exit from stationary phase, reduces fitness under nutrient limitation and alkaline pH, and impairs stringent response induction, intracellular pH homeostasis, and glycogen accumulation—phenotypes consistent with metabolic dysregulation. These findings indicate that low-level spontaneous TisB production modulates metabolism during the transition to stationary phase. By collapsing the PMF, TisB attenuates nutrient uptake and promotes early stringent-response-driven dormancy before nutrient exhaustion. Our results reveal a physiological role for TisB, positioning it as a metabolic rheostat that fine-tunes adaptation to environmental fluctuations.

This study shows that a bacterial toxin previously linked only to DNA damage also helps Escherichia coli adjust its metabolism during normal growth transitions.

Introduction

Toxin–antitoxin (TA) systems are widespread genetic elements found in bacterial and archaeal genomes, as well as in plasmids and bacteriophages (Jurėnas et al. 2022). TA systems usually consist of a toxin-encoding gene alongside its neutralizing antitoxin counterpart. Toxins are typically proteins that disrupt essential cellular functions, including DNA replication, transcription, translation, membrane integrity, cell wall synthesis, ATP production, and metabolic pathways, while antitoxins can be either proteins or small non-coding RNAs. TA systems are traditionally regarded as selfish genetic elements that spread through horizontal gene transfer (Ramisetty and Santhosh 2017, Fraikin et al. 2020). However, many TA systems have been ‘domesticated’ by co-option into host regulatory networks, such as the SOS and stringent response pathways (Harms et al. 2018). This blurs the boundary between parasitic DNA and functional cellular components, highlighting the potential roles of these elements in the host’s stress responses, adaptation, and survival.

This is also true for Escherichia coli TisB/IstR1, one of the best-studied TA systems. TisB/IstR1 belongs to type I TA systems, where toxins are small proteins and antitoxins are small non-coding RNAs. Type I TA systems are typically restricted to a narrow phylogenetic range and are less frequently associated with mobile genetic elements compared to type II TA systems, for example (Coray et al. 2017). The tisB/istR1 locus has been identified in the recently emerged genera of the Enterobacterales order, likely due to a unique ancestral acquisition event (Su et al. 2022).

TisB is a 29-amino-acid hydrophobic protein that targets the inner membrane, disrupts the proton motive force (PMF), leading to ATP depletion, membrane depolarization, and metabolic arrest (Berghoff et al. 2017, Cayron et al. 2024, Lobie et al. 2025). IstR1 is a non-coding RNA antitoxin that binds to a ribosome standby site on the TisB mRNA, preventing ribosome loading and enabling RNase III-mediated degradation (Wagner and Unoson 2012). Under normal growth conditions, the SOS response repressor LexA binds to an operator site in the tisB promoter, thus strongly repressing toxin gene transcription. The IstR1 coding gene is thought to be constitutively expressed. Under SOS-inducing conditions, when LexA no longer binds to the tisB promoter, the massive synthesis of tisB mRNA overwhelms IstR1 inhibition, leading to TisB production.

The SOS regulon in E. coli comprises about 50 genes, including 6 TA systems. In unstressed cells, these genes are repressed by the LexA protein (Simmons et al. 2008). The SOS regulon can be induced by a wide variety of external or internal stressors that cause the accumulation of persistent single-strand DNA (ssDNA). The RecA protein polymerizes on the ssDNA, activating its coprotease activity, which promotes LexA autocleavage and derepresses SOS regulon genes. Once DNA damage is repaired, ssDNA disappears, and the repression of SOS regulon genes is restored. Because the SOS regulon was initially linked mainly to DNA repair, genes like tisB, which are not directly connected to this process, received little attention for a long time.

The first hypothesis regarding the physiological role of TisB was based on the idea that, upon SOS induction, TisB transiently reduces cellular metabolic activity, thereby rendering cells tolerant to antibiotics (Dörr et al. 2010). However, evidence for a direct role of TisB in antibiotic-induced tolerance remains controversial. For instance, ciprofloxacin (CIP), a fluoroquinolone that targets DNA gyrase, was reported to induce the SOS response and increase TisB-dependent tolerance to CIP (Dörr et al. 2010). In contrast, another study concluded that this apparent TisB-dependent tolerance was not a consequence of SOS induction by CIP but rather reflected the presence of spontaneously formed persister cells in stationary-phase cultures that preexisted before treatment (Edelmann et al. 2021). Moreover, deletion of tisB did not affect the tolerance of either growing or stationary-phase E. coli cells exposed to another SOS-inducing fluoroquinolone, ofloxacin (Völzing and Brynildsen 2015, Goormaghtigh and Van Melderen 2019).

TisB has even been shown to increase the cytotoxicity of trimethoprim (TMP), which is a competitive inhibitor of bacterial dihydrofolate reductase (Miovic and Pizer 1971). In rich medium, the bactericidal effect of TMP results primarily from intracellular depletion of dTTP, which impairs DNA synthesis, causes DNA damage, triggers SOS induction, and ultimately leads to cell death (Giroux et al. 2017). Inactivation of tisB prevents the TisB-induced drop in the PMF, thereby sustaining thymine uptake from the growth medium and enabling the pyrimidine salvage pathway to restore dTTP pools and mitigate the toxic effects of TMP. Therefore, although all of the above antibiotics induce the SOS response, TisB exerts distinct, and in some cases opposite, effects on bacterial tolerance, highlighting that its physiological role cannot be generalized across different antibiotics.

A possible explanation for why the tisB/istR toxin–antitoxin system is integrated into the SOS genotoxic stress response network may be found in its impact on the ecological interactions of E. coli in its natural environment. For instance, it was reported that TisB prevents prophage induction, and therefore cell lysis, through ATP depletion in ciprofloxacin-treated Salmonella (Braetz et al. 2025). Furthermore, TisB-induced collapse of the PMF prevents the import of certain colicins, which are protein toxins produced by enterobacteria to outcompete other bacteria in the intestine (Su et al. 2022). Notably, colicin-encoding genes are also regulated by the LexA repressor (Cascales et al. 2007). Consequently, SOS-inducing stressors can trigger colicin production in some bacterial strains, while in others, they induce TisB expression, thereby providing protection against colicins. Both SOS induction and TisB have been found to play significant roles in E. coli colonization of the mouse intestine (Samuels et al. 2019, Su et al. 2022). This suggests that TA systems such as TisB/IstR1 may play an important ecological role, particularly in host colonization, host–pathogen interactions, and microbial competition during inflammation-driven dysbiosis, a condition known to activate the SOS response (Nedialkova et al. 2014, Samuels et al. 2019).

The different genes controlled by the LexA repressor in E. coli exhibit varying levels of spontaneous expression due to factors such as promoter sequence, structure, and LexA binding affinity (Culyba et al. 2018). While some genes are tightly regulated and expressed only under genotoxic stress conditions, others maintain basal levels of expression even in the absence of DNA damage. This latter category of genes may have evolved low-level expression to maintain cellular homeostasis. Northern blot analysis showed that TisB mRNA can be detected in cells growing in rich LB medium without exogeneous stressors (Berghoff et al. 2017), indicating that tisB may also belong to this category. The observed tisB gene expression may also result from SOS regulon induction in cells suffering spontaneous DNA damage, such as double-strand breaks (Pennington and Rosenberg 2007), or from a reduction of the DNA-binding affinity of LexA at alkaline pH, a condition cells experience as they approach stationary phase (Schuldiner et al. 1986, Dri and Moreau 1994). Regardless of the underlying cause of low-level tisB gene expression in cells growing in rich medium without exogenous stressors, the key questions are whether TisB protein is synthesized and, if so, whether it affects cellular functioning. To address these questions, we investigated tisB gene expression, TisB protein synthesis, and the impact of tisB deletion on E. coli growth and cellular functions in the absence of the exogenous SOS-inducing treatment. Our results demonstrate that the tisB gene is induced and that TisB protein is synthesized even without external DNA-damaging stress, revealing that TisB can bypass dual negative regulation by LexA and IstR1 during the transition to stationary phase. Moreover, we show that TisB helps shape metabolic adaptation during growth-phase transitions.

Material and methods

Strains and growth conditions

The strains and plasmids used in this study are listed in Table 1. All strains are derivatives of E. coli K-12 MG1655 strain (Touchon et al. 2009). Mutant derivatives were generated by P1vir transduction from the Keio collection (Baba et al. 2006) or our laboratory collection. tisB-gfp strain was constructed using lambda red cloning as previously described (Datsenko and Wanner 2000). Briefly, region Δ1-41 tisB-sfgfp::FRT::Kan^R::FRT was amplified from plasmid “pR6K Δ1-41 tisB-sfgfp” and introduced into cloning strain to replace tisB with Δ1-41 tisB-sfgfp::FRT::Kan^R::FRT. The plasmids containing the PistR1 and PistR2 promoters were constructed using the Circular Polymerase Extension method described in (Quan and Tian 2011), with synthetic DNA fragments carrying the PistR1 and PistR2 promoters including the RBS BBa_B0039 (Table 1), using plasmid pU66 as the backbone. Experiments were conducted using Luria–Bertani (LB), medium (Difco) and conditioned LB (cLB). To prepare cLB, 0.5 ml of an overnight wild-type (WT) culture was inoculated into 500 ml of LB medium and incubated at 37°C with agitation (150 rpm) for 16 h in 2 l Erlenmeyer flasks. The culture was then filtered sequentially through 0.45 µm and 0.22 µm filters (Corning) for sterilization. Buffered LB and cLB were prepared by adding MOPS to a final concentration of 100 mM and adjusting the pH with 2 M HCl or 1 M NaOH. For competition experiments, cLB was supplemented with 1% casamino acids (cLB-CAA) and 0.1% tryptophan. When required, the medium was supplemented with antibiotics (Sigma–Aldrich) at the following concentrations: ampicillin (Amp), 100 μg ml⁻¹; chloramphenicol (Cm), 30 μg ml⁻¹; kanamycin (Kan), 100 μg ml⁻¹; and tetracycline (Tet), 12.5 μg ml⁻¹. Unless otherwise stated, overnight (ON) cultures refer to 10 ml LB cultures grown in 50 ml Falcon tubes at 37°C with shaking at 150 rpm for at least 12 h.

Table 1.

Bacterial strains, plasmids, and DNA primers used in this study.

Strain name	Relevant genotype	Source/construction
Bacterial strains
MG1655 (WT)	Parental strain	Matic lab collection
lexA (Ind−)	lexA1 ins::Kan	Matic lab collection
△tisB	△tisB::FRT	(Su et al. 2022)
tisB::gfp	tisB (∆1-41)-gfp-Kan	Van Melderen lab collection
lexA (Ind−) tisB::gfp	lexA1 ins::FRT tisB (∆1-42)-gfp-Kan	lexA1 x P1 (tisB (∆1-41)-gfp-Kan)
∆rpoS tisB::gfp	∆rpoS::FRT tisB (∆1-42)-gfp-Kan	∆rpoS::FRT × (tisB (∆1-42)-gfp-Kan)
WT [Pless-gfp]	Pless-gfp-mut2	(Zaslaver et al. 2006)
lexA (Ind−) [Pless-gfp]	lexA1 ins::FRT [Pless-gfp-mut2]	(Su et al. 2022)
WT [PtisB-gfp]	PtisB-gfp-mut2	(Su et al. 2022)
lexA (Ind−) [PtisB-gfp]	lexA1 ins::FRT [PtisB-gfp-mut2]	(Su et al. 2022)
WT mkate	intC::pR mkate-cat	MG1655 × P1 (intC::pR-mkate-cat)
WT YFP	intC::pR-YFP-cat	MG1655 × P1 (intC::pR-yfp-cat)
△tisB mkate	△tisB intC::pR-mkate-cat	∆tisB × P1 (intC::pR-mkate-cat)
△tisB yfp	△tisB intC::pR-yfp-cat	∆tisB × P1 (intC::pR-yfp-cat)
△glgA mkate	∆glgA::Kan intC::pR mkate-cat	WT mkate × P1 (Keio ∆glgA::Kan (Baba et al. 2006))
△tisB △glgA mkate	△tisB glgA::Kan intC::pR mkate-cat	△tisB mkate × P1 (Keio ∆glgA::Kan (Baba et al. 2006))
WT PhisG-mkate	intC::PhisG-mkate-Kan	(Micaletto et al. 2024)
△tisB PhisG-mkate	△tisB::FRT intC::PhisG- mkate -Kan	∆tisB × P1 (intC::PhisG- mkate)
△relA △spoT	△relA::FRT △spoT207::cat	Matic lab collection
△relA △spoT PhisG-mkate	△relA::FRT △spoT207::cat intC::PhisG-mkate	(Micaletto et al. 2024)
△tisB lexA (Ind−) PhisG-mkate	△tisB::FRT intC::PhisG-mkate lexA1 ins::kan	△tisB intC::PhisG-mkate × P1 (lexA1 ins::Kan)
△tisB △relA △spoT PhisG-mkate	△tisB::FRT △relA::FRT △spoT207::cat intC::PhisG-mkate-Kan	△tisB △relA △spoT207::cat × P1 (intC::PhisG- mkate)
△rpoS	△rpoS::Kan	MG1655 × P1 (Keio ∆rpoS::kan (Baba et al. 2006))
△relA	△relA::Kan	MG1655 × P1 (Keio △relA::kan (Baba et al. 2006))
△dksA	△dksA::kan	MG1655 × P1 (Keio △dksA::kan (Baba et al. 2006))
△crP	△crP::Kan	MG1655 × P1 (Keio △crP::kan (Baba et al. 2006))
△csrA	△csrA::Kan	MG1655 × P1 (Keio △csrA::Kan (Baba et al. 2006))
△csrB	△csrB::Kan	MG1655 × P1 (Keio ∆csrB::Kan (Baba et al. 2006))
△csrD	△csrD::Kan	MG1655 × P1 (Keio ∆csrD::Kan (Baba et al. 2006))
△glgA	△glgA::Kan	MG1655 × P1 (Keio ∆glgA::Kan (Baba et al. 2006))
∆uvrY	△uvrY ::Kan	MG1655 x P1 (Keio ∆uvrY ::Kan (Baba et al. 2006))
△tisB △crsA	△tisB::FRT △crsA::Kan	△tisB::FRT × P1 (Keio ∆csrA ::Kan (Baba et al. 2006))
△tisB △csrB	△tisB::FRT △crsB::Kan	△tisB::FRT × P1 (Keio ∆csrB ::Kan (Baba et al. 2006))
△tisB △csrD	△tisB::FRT △crsD::Kan	△tisB::FRT × P1 (Keio ∆csrD ::Kan (Baba et al. 2006))
△tisB △glgA	△tisB::FRT △glgA::Kan	△tisB::FRT × P1(Keio ∆glgA::Kan (Baba et al. 2006))
△tisB ∆uvrY	△tisB::FRT △uvrY::Kan	△tisB::FRT × P1 (Keio ∆uvrY::Kan (Baba et al. 2006))
Plasmids
pUA66–Pless::gfp-mut2	pSC101 ori, KanR, Pless-gfpmut2	(Zaslaver et al. 2006)
pUA66–PtisB:: gfp-mut2	pSC101 ori, KanR ptisB- gfp-mut2, gfp under tisB promoter	(Su et al. 2022)
pUA66–PistR1:: gfp-mut2	pSC101 ori KanR PistR1- gfp-mut2, istR1 promoter region cloned upstream of the synthetic BBa_B0039 RBS and gfp-mut2 gene	This study
pUA66–PistR1::gfp-asv	pSC101 ori KanR PistR1- gfp-asv coding for the GFP protein ASV-tagged for degradation	This study
pUA66–PistR2:: gfp-mut2	pSC101 ori KanR PistR2- gfp-mut2, istR2 promoter region cloned upstream of the synthetic BBa_B0039 RBS and gfp-mut2 gene	This study
pUA66–PistR2::gfp-asv	pSC101 ori KanR PistR2- gfp-asv coding for the GFP protein ASV-tagged for degradation	This study
pHluorin2	pBeloBAC11 derivative encoding the E. coli optimized pH sensor PHluorin2 under the control of the proDp-0034b region	(Cayron et al. 2024)
pUA66–PlexA::gfp-mut2	pSC101 ori KanR PlexA-gfp-mut2	(Zaslaver et al. 2006)
pUA66–PdinG::gfp-mut2	pSC101 ori KanR PdinG-gfp-mut2	(Zaslaver et al. 2006)
pUA66–PuvrD::gfp-mut2	pSC101 ori KanR PuvrD-gfp-mut2	(Zaslaver et al. 2006)
pUA66–PsulA::gfp-mut2	pSC101 ori KanR PsulA-gfp-mut2	(Zaslaver et al. 2006)
pUA66–PdinB::gfp-mut2	pSC101 ori KanR PdinB-gfp-mut2	(Zaslaver et al. 2006)
pR6K-sfgfp	pR6K γ ori KanR	Kind gift from E. Gueguen
pR6K Δ1-41 tisB-sfgfp	pR6K-sfgfp derivative containing Δ1-4 tisB-sfgfp::FRT::KanR::FRT used for the construction of strain tisB-gfp	This study
Name	Sequence (5’-3’)	Construction
DNA primers
PisrtR1_int_Fw	TGCGCGGAACACGTATACTT	Control of PistR1 integration
PistR2_int_FW	CAAAACGCAAACTGTGATCG	Control of PistR2 integration
ASV_int_Rev	TTCGTTTAAGCTTATTAAACTGATGC	Control of ASV integration
Lin_pUA66_gfp-ATG_Fw	CCATCCAGTTTACTTTGCAGGGC	Linearization of pUA66 plasmid
Lin_pUA66_Fw	GAAAGGCTCAGTCGAAAGAC	Linearization of pUA66 plasmid
Lin_pUA66_Rev	CTCGAGGTGAAGACGAAAGG	Linearization of pUA66 plasmid
rrnB_Ter_Rev	GGATTTGTCCTACTCAGGAGAGC	Amplification gfp-asv gene and Sequence full PistR1 and PistR2
Seq_promoter_Fw	CCATCCAGTTTACTTTGCAGGGC	Sequence full PistR1 and PistR2 construction
BBa_gfp_Rev	TTGGGACAACTCCAGTGAAAAGTTCTTCTCCTTTACGCATCATCGAGGCCTTTCCTGTGT	Amplification gene GB1 and GB2
GFP-ASV_rev	AACTGATGCAGCGTAGTTTTCG	Amplification gfp-asv
42mRNA TisB GFP for	GATTACGAATTCGCACTAAGTCCTGGCTGAAACG	Used with primer “42mRNA TisB GFP rev“ on K12 strain to amplify tisB coding sequence with the 41 UTR region of tisB
42mRNA TisB GFP rev	CAGCCGAGCTCTTCAGGTATTTCAGAAC	Used with primer “42mRNA TisB GFP for“ on K12 strain to amplify tisB coding sequence with the 41 UTR region of tisB
Pr6K_sfgfp_tisB_rev	CTTAGTGCGAATTCGTAATCATGTCATAGC	Used with primer “ Pr6K_sfgfp_tisB_for” on plasmid pR6K-sfgfp to insert PCR product from primer “42mRNA TisB GFP for” and “42mRNA TisB GFP rev“
Pr6K_sfgfp_tisB_for	ATACCTGAAGAGCTCGGCTGGCTCC	Used with primer “ Pr6K_sfgfp_tisB_rev” on plasmid pR6K-sfgfp to insert PCR product from primer “42mRNA TisB GFP for” and “42mRNA TisB GFP rev“
Verif_pr6K_for	CATTAATGCAGCTGGCAC	Used with primer “ Verif_pr6K_gfp_rev “ on plasmid “pR6K Δ1-41 tisB-sfgfp” to verify the construction
Verif_pr6K_gfp_rev	GATAGCCCAGTAGCTGAC	Used with primer “ Verif_pr6K_gfp_for “ on plasmid “pR6K Δ1-41 tisB-sfgfp” to verify the construction
Δ1-41_tisB_gfp_chrom_for	ATTACTGTTTATTTATACAGTAAACTTCTATAATATCACTCACTAAGTCCTGGCTGAAAC	Used with primer “Δ1-41_tisB_gfp_chrom_rev” on plasmid “pR6K Δ1-41 tisB-sfgfp” to amplify Δ1-41 tisB-sfgfp:FRT:KanR:FRT for the construction of strain tisB::gfp via lambda red.
Δ1-41_tisB_gfp_chrom_rev	CTTTGGTGCGACTTGAATCTGAACATATGAATATCCTCCTTAG	Used with primer “Δ1-41_tisB_gfp_chrom_for” on plasmid “pR6K Δ1-41 tisB-sfgfp” to amplify Δ1-41 tisB-sfgfp:FRT:KanR:FRT for the construction of strain tisB::gfp via lambda red.
Verif_tisB_for	CCTTGAGATACCGCGTGCACAACAG	Used with primer “verif_tisB_rev” on strain tisB::gfp to verifiy construction
Verif_tisB_rev	CAGGAATGGGGAATTGTTTAGCGGCAG	Used with primer “verif_tisB_for” on strain tisB::gfp to verifiy construction
Name	Sequence (5’-3’)	Chemically Synthesized DNA fragment
Construction of the PistR1-gfp and PistR2-gfp reporters
Gene Block 1	TTCCCAACCTTACCAGAGGGCGCCCCAGCTGGCAATTCCGACGTCTAAGAAACCATTATTATCATGACATTAACCTATAAAAATAGGCGTATCACGAGGCCCTTTCGTCTTCACCTCGAGAAATACGTCAAAATTCGTGCCGAAATTGCGCGTTCTGCGCGGAACACGTATACTTTCAGTGTTCACACAGGAAAGGCCTCGATGATGCGTAAAGGAGAAGAACTTTTCACTGGAGTTGTCCCAA	Gene Block to introduce PistR1-BBa_B0039-gfp in pUA66 backbone
Gene Block 2	TTCCCAACCTTACCAGAGGGCGCCCCAGCTGGCAATTCCGACGTCTAAGAAACCATTATTATCATGACATTAACCTATAAAAATAGGCGTATCACGAGGCCCTTTCGTCTTCACCTCGAGTTACTGTATAAATAAACAGTAATATTTGGACAAAACGCAAACTGTGATCGATACTCAACTATTCACACAGGAAAGGCCTCGATGATGCGTAAAGGAGAAGAACTTTTCACTGGAGTTGTCCC	Gene Block to introduce PistR2-BBa_B0039-gfp in pUA66 backbone

Verification of the ΔtisB genetic construct

We previously confirmed that the phenotypes associated with the ΔtisB allele used in the present study are TisB-dependent and not artifacts of the genetic construct (Su et al. 2022). This was assessed using specific controls under SOS induction with lethal trimethoprim treatment. Because TisB reduces PMF-dependent import of thymine/thymidine, it enhances trimethoprim lethality. Consequently, deletion of tisB increased survival, whereas TisB expression decreased it. Using survival after trimethoprim treatment as a readout, we tested the following controls: (i) deletion of tisB with and without the antibiotic resistance cassette; (ii) combined deletion of tisB and istR to test for TisB-independent effects of the sRNA IstR; (iii) combined deletion of tisB and the downstream gene emrD to evaluate potential polar effects; and (iv) plasmid-based complementation (pBAD-tisB⁺). These experiments revealed no effect of the resistance marker, no polar effects on emrD, no IstR-dependent phenotype in the absence of tisB, and restoration of the expected survival phenotype upon plasmid-borne tisB expression. Together, these results demonstrate that the observed phenotypes are specifically TisB-dependent and not due to the genetic construct.

Measurement of growth curves and expression levels of fluorescent protein-based transcriptional reporters

Overnight cultures were diluted 1:10 in 10⁻² M MgSO₄, and a 1.5 μl aliquot of this dilution was inoculated into 150 μl of culture medium in black, transparent-bottom 96-well plates. To prevent evaporation, each well was overlaid with 50 μl of mineral oil. Plates were placed in a TECAN Spark plate reader, and growth​ and fluorescence were measured every 5 min under double-orbital agitation at 37°C. Growth was monitored by measuring OD_600nm. For GFP-based reporters, excitation and emission wavelengths of 480 nm and 510 nm, respectively, were used. For mKate-based reporters, excitation and emission wavelengths of 588 nm and 635 nm, respectively, were used.

We also used CytoFlex cytometer (Beckman Coulter) to measure expression of different fluorescent proteins-based transcription reporters in individual cells. For this experiment, 1 µl of bacterial culture was transferred in 250 µl PBS in a microplate, and fluorescence was measured in at least 50 000 cells.

Gentamicin-Texas Red (GTx) conjugate transport

To observe the import of gentamicin (GM), we used a gentamicin-Texas Red (GTx) conjugate which was obtained by mixing 4.4 ml of GM 50 mg/ml with 0.6 ml of Texas Red (Molecular Probes) 2 mg/ml and agitated at 4°C during 48 h (Dai et al. 2006). GTx was added to the cultures at a final concentration of 2.5 μg/ml when OD₆₀₀_nm = 0.6 for 15 min. Cells were then collected and washed several times with 10⁻² M MgSO₄ to remove unbound Tx. For microscopy imaging, 1 µl of sample is spotted on 1.5% agarose pads prepared with 10⁻² M MgSO₄. Fluorescence detection was performed using a 561 nm excitation laser and a 630/660 nm emission filter. For each observation field, 16 images were acquired by varying the Z parameter with a 0.1 µm step in order to obtain images at different focal planes, achieving a depth of field suitable for the width of the bacteria. Using the Z-projection tool in ImageJ, the images were summed into a single image for analysis. The mean fluorescence of each cell was obtained after analysis using the MicrobeJ plugin (Ducret et al. 2016) on Fiji (ImageJ).

Protein extraction

Escherichia coli cultures were grown to the desired optical density (OD₆₀₀​_nm) and harvested by centrifugation at 4000 × g for 10 min at 4°C. The resulting cell pellet was resuspended in lysis buffer (50 mM Tris-HCl, pH 8.0; 150 mM NaCl; 1% Triton X-100; 0.5% sodium deoxycholate; 0.1% SDS; 1 mM PMSF) at a ratio of 1 ml per 10 ml of the original culture volume.

Lysozyme was added to a final concentration of 1 mg/ml to facilitate cell wall disruption, and the suspension was incubated on ice for 30 min. Cell lysis was further enhanced by sonication using a Diagenode Bioruptor at 30 s intervals for five cycles, keeping samples at 4°C using in a temperature-controlled water bath. The lysate was centrifuged at 10 000 × g for 20 min at 4°C to remove cell debris, and the supernatant containing soluble proteins was collected and stored at −80°C. Protein concentration was determined using a DC Protein Assay Kit (Bio-Rad).

Western blotting

Before electrophoresis, 4 × Laemmli Sample Buffer (Bio-Rad, 161-0747) was added to the samples, which were then heated at 95°C for 5 min to denature the proteins. Proteins were separated using precast TGX 12% acrylamide denaturing gels (Bio-Rad, 456-1033) in 1X TGS running buffer (Tris, Glycine, SDS; Bio-Rad, 161-0772), prepared by diluting 100 ml of 10X TGS buffer in 900 ml of Milli-Q water.

A Kaleidoscope protein ladder (Bio-Rad, 161-0375) was used as a molecular weight marker. Electrophoresis was performed at 25 mA constant current for 10 min, followed by 50 mA constant current for 1 h. After separation, proteins were transferred onto a PVDF membrane (Bio-Rad, 162-0177) using transfer buffer (100 ml 10× TG buffer (Bio-Rad, 161-0771), 200 ml methanol, 700 ml Milli-Q water).

The PVDF membrane was activated in methanol for 5 min before use. Protein transfer was carried out at 200V constant voltage for 30 min, followed by extensive rinsing in Milli-Q water and incubation in TBST buffer (20 mM Tris, 500 mM NaCl, pH 7.5, 0.1% Tween-20), prepared from 10× TBS buffer (Bio-Rad, 170-6435). Membranes were then blocked with 5% non-fat milk or BSA in TBST for 1.5 h at room temperature to prevent non-specific antibody binding.

For protein detection, membranes were incubated overnight at 4°C with primary antibodies (anti-GFP (D5.1)) Rabbit mAb Ref. 2956, Cell Signaling Technology and anti-Sigma70 antibody (NeoClone, #W0004), diluted in blocking buffer, followed by three washes with TBST for 5–10 min each. The membrane was then incubated with an HRP-conjugated secondary antibody (Abcam AB 6721) for 1.5 h at room temperature, followed by an additional three washes with TBST to remove unbound secondary antibodies. Finally, the membrane was incubated with chemiluminescent Western blot reagents and visualized using the Bio-Rad imaging system (Chemidoc MP). For dot blot quantification, the « blot analysis » function in the ImageJ software (version 1.52a) was used. This method consists of subtracting the background and measuring the integrated density of each band.

Competition experiments

To assess the relative fitness of different strains in various media, overnight cultures were mixed at a 1:1 ratio and subsequently diluted in the corresponding media to a final 1:1000 dilution in round-bottom 96-well microtiter plates. The final volume per well was adjusted to 150 µl, and 50 µl of mineral oil was added to prevent evaporation. An aliquot of the initial 1:1 mixture was preserved to determine the initial strain ratio, allowing for the calculation of the competitive index (CI) and relative competitive fitness (RCF). Plates were incubated at 37°C with agitation for 24 hl. In 3-day competition experiments, cultures were diluted 1:1000 in fresh medium every 24 hl. Prior to sample collection, the contents of each well were briefly mixed by pipetting.

Competing strains were distinguished using a Beckman Coulter Gallios flow cytometer, which allowed fluorescence-based differentiation. Data acquisition was performed at a rate of up to 2000 events per second, analyzing a total of 50 000 cells per sample. Strains carried chromosomally integrated constitutive fluorescent transcriptional reporters, either P_RLambda-YFP or P_RLambda-mkate, inserted at the intC phage integration site. To account for potential fitness effects of each fluorescent reporter, competitions were performed with both strains carrying each reporter variant, resulting in two independent competitions per experiment: one with WT cells expressing mKate versus mutant cells expressing YFP, and another with the fluorescent labels swapped to control for tag-related bias.

Relative fitness was estimated using the competitive index calculated as: CI_n = [Day_n (% WT/mutant)]/[Day₀ (% WT/mutant)], and the relative competitive fitness calculated as the slope of the linear regression over three days following the initiation of competition of ln(freq WT/freq mutant).

DNA content analysis

DNA content quantification was conducted following a previously established protocol (Boye and Løbner-Olesen 1991). Overnight cultures were diluted 1:1000 in fresh LB medium and incubated at 37°C with agitation. Samples were taken at 15, 30, and 60 min.

To inhibit cell division and prevent initiation of new replication cycles, cephalexin (10 μg ml⁻¹) and rifampicin (150 μg ml⁻¹) were added simultaneously. While rifampicin blocks replication initiation, it permits ongoing replication rounds to complete. Cells were incubated with both antibiotics at 37°C for 90 min.

After incubation, cells were washed with 10⁻² M MgSO₄​ and fixed in 100% cold methanol for 1 h. Fixed cells were then washed with phosphate-buffered saline, and DNA was stained with DAPI (4′,6-diamidino-2-phenylindole; 0.3 μg ml⁻¹). For each time point, at least 50 000 cells were analyzed using a Beckman Coulter Gallios flow cytometer, with data acquisition at a maximum rate of 2000 events per second. DAPI fluorescence was excited at 355 nm using an ultraviolet laser and detected at 450 nm.

The number of peaks in the DNA content histograms corresponded to fully replicated chromosomes, indicating the number of replication origins present at the time of drug treatment. The number of genome equivalents (N) was determined using stationary-phase cells (1N) as a reference.

Microfluidics experiments

Cells from overnight cultures were introduced into a microfluidic mother machine device via centrifugation. This device, previously described (Mathieu et al. 2016), consists of separate growth channels, each with a closed end and an open end that connects to a feeding channel. Fresh LB medium was continuously supplied at 500 mbar using a perfusion pump (Elveflow OB1MK3). For each experiment, images were acquired over 5 h at 5-min intervals using MetaMorph software and a Nikon Eclipse Ti fluorescence microscope, equipped with a motorized stage and a Photometrics CoolSnap HQ2 charge-coupled device (CCD) camera. Image analysis was performed using ImageJ with the BACMMAN plugin (Fleurier et al. 2022). The first division time was defined as the interval between the first acquired image and the completion of the cell’s first division. The division time of a single cell was defined as the interval between its birth and subsequent division (Ollion et al. 2019).

Internal pH

To assess whether the tisB mutant struggles to maintain internal pH (pH_i) during growth in an alkaline medium, WT cells and various derivatives were transformed with a plasmid carrying the gene encoding the pH-sensitive pHluorin2 indicator protein (Mahon 2011). pHluorin2 is a modified variant of GFP containing multiple amino acid substitutions, including the F64L mutation, which enhances protein folding and maturation (Heim et al. 1995, Mahon 2011). This GFP derivative exhibits a bimodal excitation spectrum with peaks at ∼395 nm and 475 nm and an emission maximum at 509 nm. Its fluorescence properties depend on the protonation state of the chromophore and are therefore sensitive to pH. Upon acidification, excitation at 395 nm decreases with a corresponding increase in the excitation at 475 nm. Conversely, at alkaline pH, excitation at 395 nm increases while the 475 nm excitation peak intensity decreases (Arce‐Rodríguez et al. 2019). The ratio of fluorescence excited at these two wavelengths can thus be used to quantify intracellular pH variations in E. coli cells (Miesenböck et al. 1998).

Growth curves and fluorescence measurements were recorded over 12 h at 37°C with double-orbital agitation in flat-bottom black 96-well plates using a Tecan microplate reader. Optical density (OD) was measured at 600 nm, while fluorescence was recorded at two excitation wavelengths 400 nm (Fluo₄₀₀) and 485 nm (Fluo₄₈₅)—with emission at 520 nm. Differences in pH_i between genotypes were calculated based on the ratio of Fluo₄₀₀/Fluo₄₈₅.

Glycogen determination

Overnight cultures (20 ml) were harvested by centrifugation at 4000 rpm for 10 min at 4°C. Cell pellets were resuspended in 20 ml Tris–EDTA–acetate buffer (50 mM, pH 7.8) and centrifuged again at 4000 rpm for 10 min at 4°C. The pellet was then resuspended in 200 mM acetate buffer containing a protease inhibitor tablet at a ratio of 3 ml buffer per g of wet cell pellet. Aliquots (300 µl) of the suspension were transferred to sonication tubes and lysed by sonication (seven cycles of 30 s on/30 s off). Lysates were clarified by centrifugation at 12 000 rpm for 5 min at 4°C, and the resulting supernatant was used for glycogen determination with the Enzychrom Glycogen Assay Kit (fluorimetric), according to the manufacturer’s instructions.

For Lugol staining, 10 μl of overnight cultures of different E. coli strains were spotted onto LB agar plates and incubated overnight at 37°C. Subsequently, the plates with colonies were kept at 4°C for 24 h, after which 10 μl of Lugol solution (iodine–potassium iodide solution) was deposited onto each colony. Colonies containing glycogen turn dark brown due to the formation of a glycogen-iodine complex, whereas those lacking glycogen remain yellow to pale brown.

Statistical analysis

Statistical analyses were performed using GraphPad Prism (version 10.5.0). The choice of statistical test was guided by the nature of the data. Parametric tests (t-test) or non-parametric tests (Mann–Whitney test) were applied after assessing whether the data followed a normal distribution. For one-day competition experiments, statistical significance was determined as deviations of the Competitive Index (CI) from a hypothetical median of 1. For 3-day competition experiments, significance was assessed as deviations of the mean slope from zero using a one-sample t-test. Statistical differences between curves were evaluated by comparing the areas under the curve (AUC).

Results

Expression of the tisB gene and synthesis of TisB protein in E. coli cells in the absence of exogenous SOS-inducing stress

In this study, we used the E. coli K-12 MG1655 strain, referred to as the wild type (WT), along with its mutant derivatives (Table 1). To assess whether the tisB gene is expressed in E. coli cells growing in LB medium without any exogenous stress, we used a low-copy plasmid carrying the istR-tisB promoter region upstream of a gene encoding the fast-folding green fluorescent protein GFPmut2 (Su et al. 2022). We also used plasmids carrying transcriptional reporters for six LexA-controlled genes: dinB, dinG, lexA, ruvA, sulA, and uvrD, along with a promoter-less control plasmid. Stationary-phase cultures were diluted into fresh LB medium, and cell growth and fluorescence were monitored using a microplate reader. All LexA-controlled genes showed induction during the late exponential growth phase (Fig. 1A). Among them, P_tisB-gfp was the last to be induced during the transition to stationary phase. Its induction was lower in lexA (Ind⁻) cells, which cannot activate the SOS response, compared to WT cells (Fig. S1A), confirming that tisB is a bona fide component of the SOS regulon. However, P_tisB-gfp activity in the lexA (Ind⁻) strain remained higher than that of the promoter-less control (Fig. S1A), indicating that LexA repression does not completely abolish tisB expression during this growth phase.

Figure 1.

Spontaneous expression of the tisB gene and synthesis of TisB protein. (A) Kinetics of the spontaneous expression of six LexA-controlled genes, measured by monitoring GFP fluorescence intensity. A promoter-less (P_less) plasmid was used as a negative control. Each data point represents the mean value (± SEM) from three independent experiments. The shaded time window indicates the period during which the P_tisB-gfp reporter begins to be expressed above the P_less control, while bacterial growth slows down. Growth was assessed by measuring optical density at 600 nm. (B) Flow cytometry analysis of fluorescence at the growth endpoints of WT and lexA (Ind⁻) strains carrying the P_tisB-gfp reporter or P_less control, grown in LB medium. For each strain and experiment, a minimum of 50 000 cells was analyzed. Bars represent the mean value (± SEM) from at least eight independent cultures from two independent experiments. Statistically significant differences are indicated with asterisks: **** P < 0.0001 (t-test). “ns” indicates no significant difference. (C) Spontaneous synthesis of the TisB protein was assessed by Western blot. Total protein was extracted from overnight cultures of WT, lexA (Ind⁻), and ΔrpoS strains carrying a chromosomal tisB-gfp reporter fusion, grown in LB medium, and used for TisB-GFP detection. WT without the tisB-gfp reporter was used as a negative control. Antibodies against GFP were employed to detect the TisB-GFP fusion, while antibodies specific to the RpoD protein served as a loading control. The band corresponding to the TisB-GFP fusion is indicated with arrows.

To further investigate the difference in P_tisB-gfp expression between WT and lexA (Ind⁻) cells growing in LB medium, we performed flow cytometry analysis of fluorescence at the growth endpoints. We found that the P_tisB-gfp reporter was induced significantly more strongly in WT than in lexA (Ind⁻) cells (Fig. 1B). The two-dimensional representation of the Forward Scatter (FSC) and Fluorescence (FL1) values revealed the presence of highly fluorescent subpopulations (Fig. S1B). These subpopulations, which represent about 2% of cells in a LexA-proficient background, correspond to canonical SOS induction, as they disappear in the population of lexA (Ind⁻) cells. However, the P_tisB-gfp reporter was still significantly more induced in both WT and lexA (Ind⁻) cells compared to their respective P_less controls, indicating again that LexA repression does not fully block tisB gene expression (Fig. 1B).

To determine whether the TisB protein is synthesized during stationary phase, we used a strain carrying a chromosomal tisB-gfp gene fusion and detected TisB-GFP fusion protein using an anti-GFP antibody (Fig. 1C and Fig. S2ABC). As a positive control, we induced the SOS response with the DNA-damaging agent mitomycin C (MMC), which increased TisB–GFP levels by more than 66-fold compared with untreated WT cells (Fig. S2C). Notably, untreated WT cells already displayed detectable TisB–GFP production, indicating spontaneous expression, although at substantially lower levels than in MMC-treated cultures (Fig. 1C and Fig. S2BC). In contrast, TisB–GFP levels in the lexA(Ind⁻) strain remained at background levels. Quantification showed that TisB–GFP levels in WT cells were ∼44-fold higher than those observed in the lexA(Ind⁻) background (Fig. S2A). Given that the RpoS sigma factor (σ³⁸) of RNA polymerase (RNAP) regulates gene expression during the transition from growth to stationary phase (Battesti et al. 2011), a period when LexA-controlled genes are spontaneously induced (Fig. 1A), we investigated its potential role in TisB production. However, TisB-GFP synthesis occurred independently of RpoS (Fig. 1C and Fig. S2AB).

Therefore, the tisB gene is induced during the transition to stationary phase, and TisB protein is synthesized during stationary phase, even in the absence of exogenous SOS-inducing stressors.

The ΔtisB strain exhibits reduced fitness compared to WT in nutrient-depleted and alkalinized growth media

While high-levels of TisB production, whether artificially induced or triggered by SOS-inducing treatments, are known to disrupt cellular processes (Unoson and Wagner 2008, Su et al. 2022, Cayron et al. 2024), the impact of low-level spontaneous TisB production in WT cells remains unexplored. To address this knowledge gap, we first compared the growth of WT and ΔtisB strains in LB medium and observed no significant difference (Fig. 2A). We next assessed growth in LB medium conditioned by 16 h growth of the WT strain (conditioned LB: cLB), which mimics the nutritional and metabolic state of late stationary phase, when tisB is induced, TisB protein is synthesized, nutrients are depleted, and extracellular metabolites accumulate. Because TisB dissipates the PMF and can impair PMF-dependent import, as shown for thymine and lactose (Su et al. 2022), WT and ΔtisB strains may differ in their ability to take up residual nutrients from cLB, a difference that could translate into altered growth under these conditions. Both strains grew slowly in cLB, but again, no significant differences in their growth curves were observed (Fig. S3A).

Figure 2.

Impact of tisB gene deletion on E. coli growth and fitness. (A) Twenty-four-hour batch culture competitions were performed between WT and ΔtisB strains in different media: LB and cLB; LB and cLB buffered to pH 6.0 or pH 8.5; and cLB supplemented with 1% casamino acids plus 0.1% tryptophan (cLB + CAA). Competitive fitness was measured by calculating the Competitive Index (CI). Asterisks indicate statistically significant deviations of the CI from a hypothetical median of 1 (one-sample t-test: P < 0.05). “ns” indicates no significant difference. Each bar represents the mean value (± SEM) of six independent experiments. (B) Three-day batch culture competitions were performed between WT and ΔtisB strains in different media: LB and LB buffered to pH 6.0 or pH 8.5. Overnight cultures were diluted 1:1000 into fresh medium every 24 h. Relative competitive fitness was estimated as the slope of the linear regression of ln (frequency WT/frequency ΔtisB) over time. Asterisks indicate significant deviations of the mean slope from zero (one-sample t-test: *P < 0.05; **P < 0.005). “ns” indicates no significant difference. Each bar represents the mean value (± SEM) of at least six independent experiments. (C) and (D) show the growth of individual WT and ΔtisB cells in LB medium, monitored using the “mother machine” microfluidic device. In both panels, box plots represent the minimum, maximum, and median values. (C) Lag phase: Time required for cells to resume division after stationary phase, measured in at least 70 individual cells per experiment. The box plots show data from one representative experiment out of four independent experiments. Asterisks indicate statistically significant differences (Mann–Whitney test; P < 0.01). “ns” indicates no significant difference (D) Generation time: Time between two consecutive cell divisions, measured in at least 100 tracked cells in four independent experiments. The box plots show data from one representative experiment out of four independent experiments. “ns” indicates no significant differences (Mann–Whitney test). (E) and (F) DNA content in individual (E) WT and (F) ΔtisB cells measured 15 min after refreshing overnight cultures in LB, using the cephalexin-rifampicin method. DNA was stained with DAPI, and the amount of DNA per cell was analyzed by flow cytometry. For each strain, 50 000 cells were analyzed. The number of genome equivalents (N) per cell was calculated using stationary-phase 1N cells as a reference. One representative result from five independent experiments is shown. For statistical analysis see Table 2.

To detect small fitness differences that may not be apparent from individual growth curves, we performed 24 h batch culture competition experiments in both LB and cLB (Fig. 2A). The competitive index showed that the WT and ΔtisB strains had similar fitness in LB medium, but ΔtisB exhibited significantly reduced fitness compared to WT in cLB. While fresh LB has a pH of 7, cLB has a pH of 8.5 and is largely depleted of nutrients, which in LB consist mainly of peptides and amino acids (Sezonov et al. 2007). Therefore, we investigated which of these two factors underlies the reduced fitness of ΔtisB strain in cLB. When cLB was buffered to pH 6, the ΔtisB strain not only overcame its fitness defect but even outcompeted the WT, whereas buffering cLB to pH 8.5 had no effect. Supplementation with casamino acids + tryptophane alleviated the fitness defect of the ΔtisB strain relative to WT, though to a lesser extent than buffering cLB to pH 6.

Next, to separate the effects of pH and nutrient depletion, both of which impacted ΔtisB fitness in cLB, we examined how varying LB medium pH levels affected the growth of WT and ΔtisB strains. Buffering LB to pH 8.5 reduced growth of both strains relative to unbuffered LB, but their growth curves remained similar (Fig. S3B). In contrast, buffering LB to pH 6 had no effect on the growth of either strain compared to unbuffered LB. Finally, to uncover small but meaningful differences that might appear at any stage of the growth cycle, we performed a three-day competition in batch culture (Fig. 2B). We assessed the fitness of the ΔtisB strain relative to WT and found that ΔtisB exhibited significantly lower fitness than WT in both unmodified LB and LB buffered to pH 8.5, but not in LB buffered to pH 6.

Therefore, ΔtisB strain has reduced fitness compared to WT under conditions of nutrient-depleted and elevated pH.

TisB impacts both entry into and exit from stationary phase, but not exponential growth

Bacterial growth under batch culture typically proceeds through three phases: lag (a preparatory period before cell division), exponential (characterized by rapid cell division), and stationary (growth cessation). The reduced fitness of the ΔtisB strain compared to WT in batch culture competitions could stem from defects in any of these growth phases. To identify which growth phase is affected by the ΔtisB deletion, we examined the growth of single cells in LB medium using a microfluidic “mother machine” device coupled with microscopy (Wang et al. 2010). Cells emerging from overnight stationary phase exhibited a significantly longer lag phase in the ΔtisB strain than in WT (Fig. 2C). However, the generation time, i.e. the interval between successive divisions, was not significantly different between strains (Fig. 2D).

Next, we used a rifampicin/cephalexin run-out assay (Boye and Løbner-Olesen 1991) to monitor chromosome replication dynamics in individual cells exiting from overnight stationary phase. After inoculation into fresh LB medium, samples were collected at 15-, 30-, and 60-min intervals to track replication progression during early growth resumption. DNA content distributions in both strains showed distinct peaks corresponding to one, two, four, or more chromosomal equivalents per cell (Fig. 2E and F and Table 2). Stationary phase cells had one chromosomal equivalent, while cells that completed/initiated chromosome replication cycles had two, four, or more chromosomal equivalents per cell. After 15 min, the ΔtisB cell population exhibited a higher proportion of cells with one chromosomal equivalent and a smaller proportion of cells with four chromosomal equivalents compared to WT. This difference persisted at 30 min, with the proportion of ΔtisB cells containing four chromosomal equivalents remaining significantly reduced. By 60 min, these differences disappeared, consistent with the “mother machine” results, which showed that ΔtisB cells exhibit a longer lag phase but eventually achieve similar growth.

Table 2.

DNA content in individual WT and ΔtisB cells during growth restart.

Strain		Time (min)1
		15				30				60
		Number of chromosomal equivalents per cell2
		1	2	4	8	1	2	4	8	1	2	4	8
		Percentage of cells with n chromosomal equivalents at each time point for each strain
WT	Number of experiments	4	4	4	4	6	6	6	6	6	6	6	6
	Mean value	18.87	64.54	17.31	0	12.39	56.23	25.37	3.99	21.84	42.77	28.44	4.86
	Standard deviation	1.04	7.6	3.89	-	7.97	10.32	13.59	5.35	23.7	15.38	11.68	6.56
△ tisB	Number of experiments	4	4	4	4	6	6	6	6	6	6	6	6
	Mean value	26.64	63.3	9.51	0	20.48	53.27	18.44	4.24	15.54	49.43	26.29	9.35
	Standard deviation	2.92	4.12	2.89	-	14.54	7.76	12.33	5.91	6.07	16.19	6.96	14.77
WT vs △ tisB t-test 3		***	ns	**		ns	ns	*	ns	ns	ns	ns	ns

¹Time after transferring cells from overnight stationary phase to fresh LB medium. DNA content was measured using the cephalexin-rifampicin method. At each time point, DNA was stained with DAPI, and the amount of DNA per cell was analyzed by flow cytometry. For each strain and time point, 50 000 cells were analyzed.

²The number of genome equivalents (N) per cell was calculated using stationary-phase 1N cells as a reference

³Paired t-test P values : *** P < 0.001; ** P < 0.01; * P < 0.05; ns: non-signifiant.

See also Fig. 2E and F.

Because exit from stationary phase can depend on the mode of entry (Jõers and Tenson 2016), we next asked whether TisB also affects the transition into stationary phase. When E. coli cells experience nutrient depletion, particularly a shortage of amino acids, uncharged tRNAs trigger the ribosome-associated RelA protein to synthesize ppGpp (Dalebroux and Swanson 2012). The accumulation of ppGpp triggers the stringent response, which facilitates the transition to stationary phase by downregulating rRNA and tRNA synthesis to conserve cellular resources and upregulating amino acid biosynthesis genes to mitigate starvation. To investigate whether TisB influences induction of the stringent response, we used a reporter fusion of the hisG promoter with mkate, which encodes a red fluorescent protein. Because hisG encodes the enzyme catalyzing the first step of histidine biosynthesis and is positively regulated by ppGpp (Barker et al. 2001), the P_hisG-mkate reporter serves as a bona fide stringent response, as it is not induced in the ΔrelA ΔspoT (ppGpp⁰) mutant (Micaletto et al. 2024) (Fig. 3AB).

Figure 3.

Effect of tisB gene deletion on stringent response induction and proton motive force function. (A) and (B) Stringent response induction in WT, lexA(Ind⁻), and ΔrelA ΔspoT strains and their ΔtisB derivatives growing in (A) cLB and (B) LB media was monitored using the P_hisG-mKate reporter integrated at the intC chromosomal site. Fluorescence normalized to OD is plotted over time. Data are from one representative experiment out of at least three independent replicates. For statistical analysis, see Fig. S4. (C) The uptake of the gentamicin-Texas Red (GTx) conjugate in WT and ΔtisB cells was monitored using a fluorescence microscope. For each cell, fluorescence values corrected for cell area were calculated. The results are from three independent experiments, with a total of 144, 130, 640, and 662 cells analyzed for WT non-treated control (NT), ΔtisB NT, WT GTx, and ΔtisB GTx conditions, respectively. GTx was used at the MIC for both strains. Violin plots showing the median (middle line) and the 25th and 75th percentiles are presented. Asterisk indicates statistically significant differences (Mann–Whitney test; P < 0.05).

During growth in cLB, the P_hisG-mkate reporter was induced earlier and significantly more strongly in WT than in ΔtisB cells during the transition to stationary phase, and this effect of TisB on stringent response induction occurred independently of LexA regulation (Fig. 3A and Fig. S4AB). A similar, albeit smaller and statistically nonsignificant, trend was observed in LB (Fig. 3B and Fig. S4AB). Statistical analysis of these effects was performed using the area under the curve (AUC) of fluorescence data, corrected for cell density by normalization to OD_600nm, based on seven independent experiments (Fig. S4AB). To further investigate a small difference between WT and ΔtisB cells growing in LB medium, we performed flow cytometry analysis of fluorescence at the growth endpoints for WT and ΔtisB strains, which increased statistical power. Using this approach, we found that the P_hisG-mkate reporter was induced significantly more strongly in WT than in ΔtisB cells (Fig. S4C). Together, these results indicate that TisB promotes activation of the stringent-response reporter during the transition to stationary phase and under nutrient-depletion conditions.

TisB may accelerate induction of the stringent response by collapsing the PMF, thus limiting nutrient import. Therefore, we tested whether deletion of tisB gene affects PMF-dependent import under our experimental conditions, i.e. in cells growing without any external stress. To do this, we exploited the fact that the PMF is required for the import of aminoglycoside antibiotics such as gentamicin (GT) (Webster et al. 2022). We used a gentamicin-Texas Red (GTx) conjugate to visualize GT import in WT and ∆tisB cells by fluorescence microscopy, and found that ∆tisB cells imported significantly more GTx than WT cells (Fig. 3C; Mann–Whitney test, P < 0.0001). We have previously shown that GT kills more efficiently SOS-induced ∆tisB than WT cells (Su et al. 2022), indicating that this antibiotic penetrates ∆tisB cells more readily. This observation is fully consistent with the increased GTx uptake we report here.

Therefore, TisB facilitates both entry into and exit from stationary phase, without affecting exponential growth. Its impact on stationary phase entry is likely linked to its ability to collapse the PMF, thereby promoting the stringent response through reduced nutrient import.

Differential regulation of the istR1 promoter across the E. coli growth cycle

We demonstrated that the tisB gene is induced and that TisB protein is produced (Fig. 1), revealing that TisB can bypass dual negative regulation: transcriptional repression by LexA and translational inhibition by IstR1. We found that repression of the tisB promoter by LexA is leaky in the bulk of the population, while in a subpopulation of cells, the tisB gene is strongly induced due to spontaneous SOS activation during the transition to stationary phase (Fig. 1AB and Fig. S1AB). This, however, raises the question of how tisB mRNA escapes IstR1-mediated translational inhibition.

Following transcription, the sRNA IstR binds to tisB mRNA, preventing its translation and mediating its degradation by RNase III (Wagner and Unoson 2012). The istR locus encodes two sRNAs with shared terminators but distinct promoters. The shorter sRNA, IstR1, is the constitutively expressed antitoxin, while the longer sRNA, IstR2, is induced by SOS-inducing treatments but has an unknown physiological role (Vogel et al. 2004, Wagner and Unoson 2012). To investigate whether the two istR promoters are active under our experimental conditions in WT strain, we used transcriptional reporter fusions of the istR1 and istR2 promoters with the gene encoding an unstable GFP (GFP-ASV). We found that the P_istR1-gfp-asv reporter was expressed during both exponential and stationary phases in LB and cLB, whereas the P_istR2-gfp-asv reporter showed only weak expression (Fig. S5AB). The fact that unstable GFP fluorescence did not diminish over time suggests that both promoters are continuously active, albeit at different levels.

To better characterize the temporal activity of the istR1 promoter, we next compared reporter constructs with stable (GFP-mut2) and unstable (GFP-ASV) GFPs in the WT strain growing in LB in a microplate reader (Fig. 4A). The fluorescence from the unstable GFP increased during the exponential phase, plateaued upon entry into stationary phase, and then began a slow secondary increase during stationary phase. Fluorescence from the stable GFP also increased during the growth phase, levelled off during early stationary phase, and then began increasing again, although at a higher rate than that of the unstable GFP. In cLB medium, both reporters exhibited initial strong increase in fluorescence, although the increase was delayed for the unstable GFP (Fig. 4B). Unlike in LB, unstable GFP fluorescence plateaued and did not show a secondary increase in stationary phase, while stable GFP the fluorescence continued to rise, albeit at a lower rate than during growth. These results indicate that the istR1 promoter is active during stationary phase both in LB and cLB.

Figure 4.

Activity of istR-1 promoter in LB and cLB media and in different genetic backgrounds. (A) and (B) The activity of the pistR-1 promoter during E. coli growth was monitored using plasmids carrying the pistR-1 promoter fused to fast-folding GFP variants: mut2, and mut2 with an ASV degradation tag to enhance proteolytic turnover. Growth was measured by measuring optical density at 600 nm. Promoter activity was measured by monitoring fluorescence under different growth conditions: in LB (A) and cLB (B). A promoter-less plasmid was used to subtract background fluorescence. Shown are the mean (± SD) of fluorescence normalized to OD₆₀₀ versus time, along with OD versus time, from three independent experiments. (C) Activity of the pistR-1 promoter in different genetic backgrounds during growth in LB medium. A plasmid carrying the P_pistR1gfp-asv fusion was introduced into WT and various mutant strains defective in global regulators: ΔdksA, ΔrelA ΔspoT, lexA(Ind⁻), Δcrp, and ΔrpoS. The experiment was repeated for each background in at least 3 independent experiments. One representative experiment is shown. (D) Activity of the pistR-1 promoter in WT and ΔrpoS strains carrying a plasmid with the P_pistR1gfp-asv reporter fusion during growth in LB medium. Fluorescence was measured by flow cytometry at different stages of the cell growth cycle. 50 000 cells per time point and strain were analyzed. The experiment was repeated three times, and one representative experiment is shown.

The temporal dynamics of the expression of GFP fluorescence reporters suggests that the istR1 promoter is continuously active, but undergoes differential regulation across growth phases. The transition from growth to stationary phase in E. coli is governed by global regulators, including the stringent response and the RNAP sigma factor RpoS (Dalebroux and Swanson 2012, Bouillet et al. 2024). Although previous work reported that the stringent response does not affect istR or tisB expression (Maciąg-Dorszyńska et al. 2025), we sought to verify this under our conditions. We therefore tested istR1 promoter activity using an unstable GFP-ASV reporter in Δcrp [lacking the cAMP receptor protein CRP (Sonnleitner 2025)], a relA spoT double mutant (ppGpp⁰) than cannot synthetize ppGpp, and ΔdksA (lacking the protein that modulates RNAP in conjunction with ppGpp) strains, all of which are tisB-proficient (Fig. 4C). We observed no difference in promoter activity between these strains and WT, confirming that istR1 activity is independent both of the stringent response and cAMP regulation. A similar result was obtained for the lexA (Ind⁻) strain. In contrast, deletion of the rpoS gene abolished istR1 promoter activity in stationary phase cells, consistent with a previous report showing that RpoS directly binds to the istR promoter during the transition from exponential to stationary phase (Peano et al. 2015). We also compared istR1 promoter activity between WT and ΔrpoS strains using an unstable GFP-ASV reporter and FACS analysis. We found that istR1 expression was lower in the ΔrpoS strain in overnight cultures, and even more so during the transition to stationary phase, corroborating the data obtained using a microtiter plate reader (Fig. 4D). This difference was very small at the early exponential phase (OD_600nm = 0.1) and undetectable during mid- and late-exponential growth (OD_600nm = 0.45 and 0.8).

Taken together, these data indicate that the activity of the istR1 promoter varies dynamically across the bacterial growth cycle. During exponential growth, it is likely driven by RpoD (σ⁷⁰), while in stationary phase it is activated by RpoS. The switching from RpoD to RpoS RNAP sigma factors may explain the plateau in the istR1 promoter activity observed during the transition to stationary phase. Concomitant with this transition, tisB promoter activity increases, which shifts the stoichiometry in favor of tisB mRNA, enabling it to out-titrate istR1, avoid istR1-mediated degradation, and permit TisB synthesis.

The loss of tisB gene causes significant metabolic perturbations

Our data show that deletion of the tisB gene reduces cell fitness in alkalinized growth media (Fig. 2A and B). To further investigate this phenotype, we examined whether tisB deletion affects the cell’s ability to maintain pH homeostasis. WT and ΔtisB cells were cultured in unbuffered LB and LB buffered at pH 8.5, and intracellular pH was measured using the pHluorin2 biosensor during growth (Fig. 5A). Internal pH did not differ significantly between the two strains in unbuffered LB. However, under alkaline conditions, ΔtisB cells exhibited a significantly different internal pH compared to WT cells at the end of the growth curve (Wilcoxon matched-pairs signed-rank test, P = 0.007). These results suggest that TisB contributes to maintaining pH homeostasis under alkaline conditions.

Figure 5.

Impact of ΔtisB on the maintenance of intracellular pH and glycogen production. (A) Maintenance of intracellular pH in WT and ΔtisB cells containing the pHluorin2 reporter at the final point of growth in LB medium and in LB adjusted to pH 8.5. Intracellular pH was calculated using the pHluorin2 fluorescence ratio. Each bar shows the median values with 95% confidence interval (CI) from eight independent experiments. Asterisk indicate statistically significant differences (Mann–Whitney test; P < 0.05). “ns” indicates no significant differences. (B) Quantity of intracellular glycogen in WT and ΔtisB stationary-phase cells. Each bar represents the mean value (± SEM) of six independent experiments. Asterisks indicate statistically significant differences (Mann–Whitney test; P < 0.05). (C) Impact of tisB gene deletion on glycogen biosynthesis in the overnight colonies of WT strain, strains mutated in major glycogen biosynthesis regulators (△csrA, △csrB, △csrD, and △uvrY), lexA(Ind⁻), △rpoS, and ppGpp⁰ strains, and the △glgA strain as negative control. Glycogen accumulation in bacterial colonies was visualized using Lugol staining. Independent experiments were performed at least three times, and one representative experiment is shown. (D) Twenty-four-hour batch culture competitions were performed between WT and ΔtisB strains, and their ΔglgA derivatives, in LB and cLB media. Competitive fitness was measured by calculating the Competitive Index (CI). Asterisks indicate statistically significant deviations of the CI from a hypothetical median of 1 (one-sample t-test: *P < 0.05; ** P < 0.01). “ns” indicates no significant difference. Each bar represents the mean value (± SEM) of three independent experiments.

Among the key metabolic adaptations during the growth-to-stationary phase transition is glycogen production, which serves as an energy reserve under nutrient limitation (Wilson et al. 2010). Glycogen accumulates when glucose is available but other essential growth nutrients become limiting. In E. coli, glycogen metabolism is linked to many cellular processes, including the stringent response, carbon source transport, envelope integrity, and energy production. Because this study shows that TisB is induced during the transition to stationary phase and influences induction of the stringent response, and because previous reports indicated that TisB affects carbon source transport, envelope integrity, and energy production (Berghoff et al. 2017, Su et al. 2022, Leinberger et al. 2024), we investigated whether it also influences glycogen accumulation.

First, we quantified glycogen in WT and ΔtisB stationary phase cells, and found that ΔtisB produced significantly more glycogen compared to the WT strain (Fig. 5B) (Wilcoxon matched-pairs signed rank test, P = 0.03). Next, we performed a comprehensive phenotypic analysis of the impact of tisB gene deletion on glycogen production in colonies of different mutants of genes coding for direct or indirect regulators of glycogen metabolism using the Lugol staining method (Fig. 5C). We found that deletion of tisB enhanced Lugol staining in a GlgA-dependent manner, consistent with the role of GlgA as glycogen synthase (Wilson et al. 2010).

Then, we tested the impact of tisB gene deletion on glycogen biosynthesis in strains mutated for major glycogen biosynthesis regulators (Wilson et al. 2010, Pourciau et al. 2020). CsrA is a regulator of carbohydrate metabolism that prevents glycogen biosynthesis and impacts glycogen degradation. CsrB and CsrC are small regulatory RNAs that negatively regulate the activity of CsrA, whereas CsrD promotes their degradation, thereby indirectly enhancing CsrA activity when glucose is abundant. UvrY is part of the BarA/UvrY two-component system, which plays a positive regulatory role in glycogen synthesis by activating the expression of CsrB and CsrC. BarA/UvrY is induced by metabolic signals associated with nutrient availability and carbon source status, especially short-chain carboxylic acids like acetate and formate, which accumulate during growth on glycolytic carbon sources.

We found that tisB-proficient WT, △csrA, △csrB, △csrD, and △uvrY strains produced less glycogen than their ΔtisB derivatives (Fig. 5C). This suggests that TisB does not impact the functioning of these regulators. However, we did not observe a difference in Lugol staining between tisB-deficient and proficient variants of △rpoS and ppGpp⁰ strains (Fig. 5C). These results are expected as ppGpp regulates RpoS levels (Bouillet et al. 2024), while RpoS induces glgS expression in stationary phase (Hengge‐Aronis and Fischer 1992). GlgS negatively regulates flagellar motility and production of biofilm polysaccharides, which compete with glycogen biosynthesis for the same metabolite pools (Rahimpour et al. 2013). Therefore, deletion of rpoS gene indirectly prevents glycogen synthesis by causing depletion of precursors for glycogen biosynthesis. Analogously, TisB may reduce glycogen synthesis by preventing nutrient import and thereby the availability of precursors for glycogen biosynthesis. Finally, we found that the impact of TisB on glycogen production is independent of LexA repressor inducibility (Fig. 5C).

Next, we investigated whether higher glycogen levels in ΔtisB strain compared to the WT represents merely a marker of metabolic perturbation or whether they contribute to the fitness cost associated with tisB deletion during 24 h competitions (Fig. 5D). Deletion of the glgA gene did not significantly affect fitness of the WT strain in LB or cLB. Similarly, glgA deletion did not significantly alter fitness of the ΔtisB strain in LB, but it significantly reduced the fitness cost of tisB deletion in cLB (CI (WT/△tisB) vs CI (WT/△tisB △glgA), t-test: P = 0.005). Nevertheless, the CI of WT/△tisB △glgA remained significantly higher than that of WT/WT (t-test: P = 0.02), indicating that additional factors contribute to the fitness cost of tisB deletion in cLB medium.

Overall, our results demonstrate that tisB deletion induces metabolic imbalances in E. coli cells. This is illustrated by impaired ability to regulate internal pH under alkaline conditions and through dysregulated glycogen accumulation, which is expected to disrupt carbon resource allocation and cellular energetics, thereby directly compromising bacterial fitness.

Discussion

Full-scale TisB production typically requires strong SOS induction triggered by exogenous DNA-damaging agents (Wagner and Unoson 2012, Su et al. 2022). Its established role under such genotoxic stress is to protect bacterial cells from environmental toxins, such as colicins and antibiotics and from bacteriophage induction (Su et al. 2022, Braetz et al. 2025). Although spontaneous low-level tisB expression in E. coli has been occasionally alluded to in the literature (Berghoff et al. 2017), its physiological significance during natural growth transitions, has never been thoroughly investigated. Thus, whether TisB can be produced and play a role outside the context of genotoxic stress remains an open question.

We observed that TisB escapes dual negative regulation, transcriptional repression by LexA and translational interference by IstR1, during the transition from exponential to stationary phase in the absence of exogenous DNA-damaging stress (Fig. 6). This observation suggests that TisB production may occur under physiological conditions distinct from the canonical SOS response. LexA repression can be relieved due to spontaneous DNA damage and/or a reduction in LexA’s DNA-binding affinity in cells exposed to alkaline pH during the transition to stationary phase, as previously described (Schuldiner et al. 1986, Dri and Moreau 1994, Pennington and Rosenberg 2007). However, spontaneous LexA-dependent SOS induction, previously measured using the P_sulA-gfp reporter, has been detected in only about 1% of cells (Pennington and Rosenberg 2007, Simmons et al. 2009). Using the P_tisB-gfp reporter, we similarly found that around 2% of cells show strong LexA-dependent induction, consistent with observations made using the P_sulA-gfp reporter. If TisB protein is produced solely in such a small fraction of cells, deletion of the tisB gene would not be expected to have a major impact on the analyzed phenotypes at the population level. However, we also detected significant low-level, LexA-independent P_tisB-gfp expression in the bulk of the population, which could be explained by increased leakiness of LexA repression. This is consistent with our observation that many of the effects of tisB deletion are LexA-independent.

Figure 6.

Growth-phase–dependent regulation of the istR1/tisB toxin–antitoxin system and TisB production. (A) Bacterial growth in LB medium proceeds through lag, exponential, and stationary phases, accompanied by pH fluctuations and progressively nutrient depletion. The transition to stationary phase is characterized by the stringent response induction and a switch in RNA polymerase sigma factors from RpoD to RpoS, redirecting transcription from growth promoting-genes toward stress resistance and cellular maintenance. (B) During the transition to stationary phase, we observed induction of tisB promoter activity, likely reflecting both SOS induction in a subpopulation of cells and increased leakiness of LexA repression in most cells. Concurrently, istR1 promoter activity plateaus, possibly due to sigma factor switching. The simultaneous increase in tisB expression and reduction in istR1 transcription may shift the stoichiometry in favor of tisB mRNA, allowing it to out-titrate IstR1 and enable TisB synthesis. While RpoS later induces istR1 transcription to limit further tisB mRNA translation, existing TisB protein likely persists because low ATP levels reduce proteolysis in stationary-phase cells. By collapsing the proton motive force and lowering ATP production, TisB limits nutrient uptake and promotes stringent response induction in an impoverishing medium before complete resource depletion.

In addition to partial relief of transcriptional repression, TisB expression may also be facilitated by reduced translational interference from the IstR antitoxin. Translational interference by the IstR antitoxin may be reduced when its production rate transiently declines during the transition to the stationary phase. We propose that the observed plateau in istR1 transcription during this growth transition reflects a kinetic gap in the sigma factor switch from RpoD to RpoS. This switch, which affects transcription of genes with promoters recognized by both sigma factors, is driven by the anti-sigma factor Rsd, whose levels rise during this growth transition. Rsd rapidly sequesters free RpoD thus preventing RpoD-RNAP holoenzyme formation (Jishage and Ishihama 1998). This transition is further accelerated by ppGpp, which actively destabilizes RpoD-complexes, activates transcription of anti-adaptor proteins to stabilize RpoS and redirects RNAP toward RpoS-dependent promoters (Dalebroux and Swanson 2012). Modeling suggests that the plateau in the expression of genes with promoters recognized by both sigma factors arises because rapid shutdown of RpoD activity is not immediately compensated for by RpoS (Kusano et al. 1996), which binds core RNAP with less than half the affinity of RpoD and initiates transcription less efficiently (Mauri and Klumpp 2014). Consequently, during this growth transition, the departure of RpoD outpaces the recruitment of RpoS, creating a transient plateau in istR1 accumulation. Experimental demonstration of the proposed model requires further investigation.

The simultaneous increase in tisB expression and reduction in istR1 transcription rate may shift the stoichiometry in favor of tisB mRNA, allowing it to out-titrate IstR1 and enable TisB synthesis during transition to stationary phase. Later in stationary phase, RpoS-driven istR1 transcription likely arrests new TisB synthesis, whose excessive accumulation could otherwise interfere with growth resumption following a nutritional upshift. Thus, TisB production appears to be confined to a narrow temporal window during early stationary phase. However, while new TisB synthesis is inhibited, already synthesized TisB likely persists, protected by its insertion into the inner cell membrane and by the declining ATP levels, which reduces protease activity in stationary-phase cells (Pu et al. 2019). This may explain why we were able to detect TisB in stationary-phase cells despite the increased expression of istR1.

The level of spontaneous TisB production is lower than that induced by the DNA-damaging agent MMC. The key question, therefore, is whether TisB produced through spontaneous regulatory relaxation can affect metabolic adaptation during growth transitions. We observed that deletion of tisB does not affect exponential growth in LB medium, but it leads to delayed and reduced induction of the stringent response during entry into stationary phase, slower exit from stationary phase, and reduced fitness under nutrient-depleted and alkaline conditions. Taken together, these phenotypes point to a broad role for TisB in coordinating metabolic responses during growth transitions. In addition, we found that tisB deletion impaired the ability to maintain intracellular pH under alkaline conditions and increased glycogen accumulation, indicating substantial metabolic perturbations.

One condition under which tisB deletion strongly reduced fitness was alkaline pH. Alkaline stress has been shown to induce the SOS and heat shock regulons in E. coli, reflecting damage to cellular macromolecules such as DNA and proteins (Padan et al. 2005). High pH can also compromise membrane integrity by disrupting hydrogen bonding within lipids (Lowe et al. 2022). Consequently, the reduced capacity of tisB mutant to maintain cellular homeostasis under alkaline conditions, together with their broadly deleterious pleiotropic effects, likely contributes to the decreased fitness observed for tisB mutant at high pH.

Another prominent metabolic alteration associated with tisB deletion was increased glycogen accumulation. Although glycogen normally helps cells withstand starvation and supports growth recovery (Wilson et al. 2010, Yamamotoya et al. 2012), it was shown that its excessive accumulation can be detrimental to E. coli because it reduces the replicative lifespan (Boehm et al. 2016) and causes metabolic imbalance (Timmermans and Van Melderen 2009). By acting as a carbon and energy sink (Dedhia et al. 1994, Wei et al. 2000), excess glycogen sequesters resources that would otherwise support essential processes. We showed that elevated glycogen levels in the ΔtisB strain compared to WT are not merely a marker of metabolic perturbation but also contribute to the fitness cost associated with tisB deletion in conditioned medium. Taken together, the deleterious effects of tisB deletion suggest that TisB plays a role in optimizing metabolic adaptation during resource limitation.

Based on these observations, we propose a mechanistic model in which TisB actively shapes the transition from growth to dormancy. We propose that TisB accelerates this transition by slowing cellular metabolism through reduced ATP production and collapsing the PMF, which limits nutrient import and promotes stringent response induction in an impoverishing medium before complete resource depletion. In this framework, TisB functions as a metabolic checkpoint, ensuring that cells enter dormancy with sufficient residual resources to support eventual growth resumption, thereby avoiding starvation traps, i.e. states of resource exhaustion that compromise regrowth.

Beyond facilitating entry into dormancy, TisB may also enhance the capacity for growth restart by reducing ATP production in stationary phase cells. This hypothesis is supported by observations in Salmonella enterica serovar Typhimurium and Saccharomyces cerevisiae, where nutrient depletion lowers ATP production, reduces activity of ATP-dependent proteases, and ultimately limits the degradation of functional proteins (Yeom and Groisman 2021). This preservation of cellular components allows cells to more efficiently resume rapid growth once nutrients become available.

In conclusion, by distinguishing between resource limitation (hunger) and complete resource depletion (starvation), we refine our understanding of TisB’s function. Rather than responding to starvation, TisB appears to help cells adapt to incipient nutrient limitation. Through this early response, TisB optimizes resource allocation and prevents energy waste under suboptimal conditions. In this way, TisB functions as a metabolic rheostat, modulating the transition between growth and dormancy. We further show that TisB can facilitate the reverse transition, from dormancy to growth. This perspective not only highlights a distinct regulatory layer but also provides new insights into the ecological advantages conferred by TA systems in fluctuating environments.

Finally, our findings on TisB’s role in modulating bacterial metabolism point to potential applications in antimicrobial strategies. Small molecules that inhibit TisB activity could hinder bacterial entry into protective dormant states associated with antibiotic tolerance, while also preventing the TisB-induced collapse of the PMF that restricts the uptake of certain antibiotics, such as the aminoglycoside GT. Such approaches could ultimately enhance antibiotic efficacy.

Conflicts of interest

None declared.

Funding

This work was supported by the French Agence Nationale de la Recherche (ANR) through grants awarded to I.M.: LST (ANR-21-CE12-006-01), DREAM (ANR-20-PAMR-0002), and IMfdmut (ANR-24-CE12-7386-01), as well as by the Initiative d’Excellence (IdEx ANR-18-IDEX-0001, project MICROBEX), which is part of the “France 2030” investment program launched by the French government and implemented by Université Paris Cité. T.O. was supported by a FNRS-FRIA fellowship. L.B. was supported by the Erasmus+ Programme of the European Union (Student Mobility for Traineeships).