Escherichia coli cell envelope is crucial for stress sensing and signal transduction, mediated by numerous protein–protein interactions to enable adaptation and survival. Interfering with these interactions might affect envelope integrity leading to bacterial death. The outer membrane lipoprotein (RcsF) is the stress sensor of the regulator of capsule synthesis (Rcs) phosphorelay that senses envelope threats. RcsF interacts with two essential proteins, IgaA (repressing the Rcs system) and BamA (inserting β-barrel proteins in the outer membrane). Disturbing RcsF interactions may alter Rcs signaling and/or membrane integrity thus affecting bacterial survival. Here, we derived the sequence of a peptide mimicking RcsF (RcsFmim), based on the in silico docking of RcsF with IgaA. expression of rcsFmim caused?3-to-4-fold activation of the Rcs system and perturbation of the outer membrane. Both effects result in decreased E. coli growth rate. We anticipate that RcsFmim present a candidate for future antibacterial peptide development.

　　The Gram-negative cell envelope is a unique and complex macromolecular structure. Formed of an inner membrane and an outer membrane separated by the periplasmic space where the peptidoglycan resides (Silhavy et al. 2010), it represents a formidable barrier to physical and chemical stressors. Cell envelope integrity is a keystone in bacterial survival, adaptation and homeostasis. Accordingly, cell envelope proteins which play key roles in response to environmental stressors and participate in key complexes and signaling pathways are promising targets for antimicrobial development. Disturbing or interfering with their function will likely lead to loss of membrane integrity and an overall negative effect on bacterial survival and adaptation.

　　RcsF, a 12 kDa outer membrane lipoprotein is a typical example of such proteins. It is the stress sensor of the Regulator of capsule synthesis (Rcs) system, a complex phosphorelay that sense and respond to cell envelope threats (Stout and Gottesman 1990; Majdalani and Gottesman 2005). The Rcs system functions in parallel to two other two component systems?(TCSs), the Conjugative pilus expression (Cpx) and the Bacterial adaptive response (Bae) systems (Leblanc et al. 2011; Delhaye et al. 2016; Grabowicz and Silhavy 2017; Erin Wall et al.2018).

　　Within the Rcs system, RcsF senses specific cues, and interact with IgaA, an essential inner membrane protein functioning as the negative regulator of the Rcs system. The interaction between RcsF and IgaA is considered to cause derepression of the Rcs phosphorelay, leading to autophosphorylation of the histidine kinase (RcsC), the transfer of the phosphate group to the phosphotransfer auxiliary protein (RcsD) and finally the phosphorylation of the cytosolic response regulator (RcsB) (Majdalani and Gottesman 2005; Wall et al. 2018 and references therein).(Fig.?1).

　　Fig. 1

　　Schematic diagram of RcsF interactions and role in the activation of the Rcs system. In the outer membrane, RcsF interacts with BamA (the essential β- barrel of the BAM machinery, which also mediates RcsF exposure to the cell surface) and OmpA (RcsF-OmpA interaction is not shown) (see text for references). Also, RcsF interacts with the periplasmic domain of the inner membrane protein, IgaA leading to derepression of the Rcs system. The Rcs system is composed of the inner membrane histidine kinase (RcsC) and the cytosolic response regulator (RcsB) in addition the phosphotransfer protein (RcsD) and the negative regulator essential protein (IgaA). Black arrows indicate the direction of the phosphorylation cascade

　　In addition, RcsF interacts with BamA (Cho et al. 2014; Konovalova et al. 2014; Rodríguez-Alonso et al. 2020), an essential component of the Bam machinery mediating the insertion of β-barrel proteins in the outer membrane of E. coli (Konovalova et al. 2016; Malinverni and Silhavy 2011) and the outer membrane porin OmpA (Konovalova et al. 2014; Cho et al. 2014; Dekoninck et al. 2020).

　　based on its important role and key interactions, we reasoned that a peptide interfering with the RcsF function within the Rcs system and/ or interaction might also affect the outer membrane thus affecting bacterial survival. In addition, the crystal structure of RcsF was determined previously as well as its complex with BamA (Leverrier et al. 2011; Rogov et al. 2011; Rodríguez-Alonso et al. 2020). This makes RcsF primary structure a good candidate for derivation of a sequence of antibacterial peptide. An additional advantage is RcsF relatively small molecular mass (12?kDa for the mature protein), so a peptide simulating its structure could potentially overcome the outer membrane barrier.

　　In this work, we derive the sequence of a novel peptide RcsFmimic (RcsFmim) from RcsF based on the in?silico prediction of RcsF-?IgaA interaction map. We show that expression of RcsFmim-coding gene causes a constitutive activation of the Rcs system and affects outer membrane permeability resulting in decreased E. coli growth. We anticipate that RcsFmim presents a likely candidate for future antibacterial peptide development.

　　The bacterial strains used in this study are all derivatives of E. coli K12 MG1655 carrying a chromosomal rprA::lacZ fusion at the lambda attachment site (DH300) (Majdalani et al. 2002). This strain was used previously as a reporter strain to monitor the Rcs system activation in the absence and presence of Rcs- specific inducing cues (Cho et al. 2014; Asmar et al. 2017; Hussein et al. 2018; Lach et al. 2023). The rcsB and rcsF knockout strains were generated by P1 vir transduction from the corresponding Keio collection strain (Baba et al. 2006) (obtained from the National Bioresource Project, Japan) to E. coli MG1655 DH300 and were checked by corresponding PCR primers complementary to the upstream and downstream genomic locus of each gene.

　　Unless otherwise stated, LB-Lennox (MP biomedical) was used to culture E.?coli?MG1655?DH300 at 37?°C containing (whenever necessary) the following concentration of antibiotics: chloramphenicol (25?μg/μl), spectinomycin (50 μg/ml) and kanamycin (20?μg/μl). Whenever required, isopropyl β-?thiogalactoside (IPTG) was added at a final concentration of 100 μM, L- arabinose and D-fucose at a final concentration of 0.2% weight/ volume.

　　Plasmids used in this study are all derived from either pSC232 or pNH401 previously described in (Hussein et al. 2018). pSC232 is derived from pSC101 (Dominique and Bouché 1991), modified by inserting lacIq and trc promoter from pTrc99a plasmid and is a kind gift from Dr Seun-Hyun Cho, Collet lab. pNH401 is derived from the L- arabinose inducible pBAD33 (Guzman et al. 1995) as described in (Hussein et al. 2018). rcsFmim nucleotide sequence was synthesized as gene block by IDT (Belgium) and was cloned between NcoI and XbaI restriction sites in pSC232 or SacI and XbaI restriction sites in pNH401. In both cases, the XbaI site was located upstream of an in- frame tag (3X flag tag in pSC232 and 5X His tag in pNH401) followed in both cases by the KpnI restriction site. Standard molecular biology techniques were followed, using Phusion polymerase (Thermo), restriction enzymes (Thermo) and E. coli DH5α (Novagen) as cloning strain.

　　We constructed a homology model of IgaA periplasmic domain (IgaAperip) (comprised between amino Asp361 to Tyr655) using Phyre2 server (Kelley et al. 2015). The interaction of RcsF (PDB2Y1B) with IgaAperip was predicted using the protein–protein docking protocol of the Molecular Operating Environment (MOE 2019.0102 Chemical Computing Group, Montreal-Canada) (Chemical Computing Group ULC 2023).

　　For prediction of the three-dimensional structure of RcsFmim, we used wild type RcsF crystal structure (PDB2Y1B) and the amino acids residues 16 to 45 were deleted. In parallel, we built a homology model for RcsFmim using its predicted amino acid sequence and wild type RcsF as a template using Phyre2. The deduced amino acid sequence of RcsFmim was deposited in a patent request to the Egyptian patent office (request number EG/P/2023/1561). RcsFmim- IgaAperip and RcsFmim- BamA interactions were predicted as described above.

　　Another homology model for IgaAperip was built using AlphaFold2 via ColabFold server and its interaction with RcsFmim was predicted using AlphaFold2-Multimer chimera visual interface (Mirdita et al. 2022).

　　Predictions using AlphaFold2 were performed by Ann Analysis for Computational Chemistry Academic services (Egypt).

　　Abstract

　　Introduction

　　Materials and methods

　　Prediction of IgaA periplasmic domain three-dimensional structure and RcsF- IgaA interaction:

　　β-Galactosidase assay

　　Assessment of E. coli growth and calculation of logarithmic growth rate

　　Outer membrane integrity/ permeability testing

　　Figure preparation and statistical analysis

　　Results

　　Structural design of RcsF mimic peptide

　　RcsFmim activates the Rcs system

　　RcsFmim slows E.?coli growth rate predominantly in an Rcs- independent manner

　　RcsFmim permeabilizes E.?coli cell envelope

　　Discussion

　　References

　　Acknowledgements

　　Author information

　　Ethics declarations

　　#####

　　Β- galactosidase activity was measured in exponentially growing E.?coli MG1655 DH300 transformed with various plasmids as described previously (Miller 1972). In case of measuring Rcs activation in response to stress, mecillinam (amdinocillin) (Sigma-Aldrich, Germany) was added at a final concentration of 0.3 μg/ml when the bacterial OD600 reached 0.2 and incubated at 37°C with agitation for one hour before assessing β- galactosidase activity. Briefly, 20?μl of the culture were permeabilized by mixing with 80?μl of permeabilization solution (100?mM Na2HPO4, 20?mM KCl, 2?mM MgSO4, 0.8?mg/ml CTAB (hexadecyltrimethylammonium bromide), 0.4?mg/ml sodium deoxycholate and 5.4 μL/ml β-mercaptoethanol). After one hour incubation at room temperature, 600?μl of substrate solution (60?mM Na2HPO4, 40?mM NaH2PO4, 1?mg/ml o-nitrophenyl-β-D-galactoside (ONPG) (Sigma-Aldrich, Germany) and 2.7?μl/ml β-mercaptoethanol were added and the tubes were incubated at 30?oC until development of the yellow color. The reaction was then stopped by addition of 700?μl of 1?M Na2CO3 and the absorbance at 420?nm was assessed spectrophotometrically. The Miller units were calculated according to the following equation.

　　All β- galactosidase assays were repeated at least in three different biological events.

　　To monitor E.?coli growth, the corresponding strains were cultured overnight in LB-Lennox containing 25 μg/ml chloramphenicol and were grown at 37?°C overnight. On the following day, the cultures were diluted 1/100 in fresh LB-Lennox containing 25 μg/ml chloramphenicol supplemented with either 0.2% L-arabinose or 0.2%?D-?fucose. The OD600 was measured in BioTek plate reader every hour for six hours. The growth rate (k) of the logarithmic phase in liquid cultures was calculated using the exponential equation Nf?=?Ni ekΔt. Alternatively, aliquots from the growing cultures were serially diluted and then spotted on LB- agar- chloramphenicol supplemented with either 0.2% L-arabinose or 0.2%?L-?glucose. Number of Colonies Forming units (CFU/ ml) were calculated and normalized to the CFU/ ml of empty pBAD33/ E. coli DH300 or empty pBAD33/ rcsB::kan tested simultaneously. Each test was repeated at least in three biological replicates.

　　To test the effect of rcsFmim expression of E. coli outer membrane integrity, mid- log phase cultures of E.?coli?DH300 or rcsF or rcsB null mutants transformed with either empty pBAD33, rcsFmim-pBAD33 or prcsFwt-pBAD33 were serially diluted and spotted on LB-?agar- arabinose plates containing 2% sodium dodecyl sulphate (SDS) or 1?mM?ethylene diamine tetra acetic acid disodium salt (EDTA). The plates were incubated overnight at?37?°C. Number of Colonies Forming units (CFU/ ml) were calculated and normalized to the CFU/ ml of empty pBAD33/ E. coli DH300 or empty pBAD33/ rcsB::kan tested simultaneously. Each test was repeated at least in three biological replicates.

　　To assess the outer membrane permeability of the same strains to the fluorescent probe 8-anilino-1-naphthalenesulfonic acid (ANS) (Loh et al. 1984; Lamers et al. 2013; Sch?fer & Wenzel 2020; Tag ElDein et al. 2021), pellets of mid- log phase cultures of E.?coli?DH300 or rcsF or rcsB null mutants transformed with either empty pBAD33, rcsFmim-pBAD33 or prcsFwt-pBAD33 grown in Mueller Hinton broth were washed and resuspended in 5 mM HEPES, and the fluorescent probe ANS (ANS, Sigma-Aldrich, Germany) was added (at a final concentration of 3 mM). Relative fluorescence was calculated as the ratio of ANS fluorescence in the tested strains (normalized to their OD600) to the isogenic strain transformed with the empty plasmid. Excitation and emission wavelengths for ANS were 375 and 510?nm, respectively. The assay was performed in independent triplicates with each biological replicate consisting of three technical replicates.

　　based on the predictions described in the previous section, we thought that RcsFmim could positively or negatively affect signaling through the Rcs phosphorelay (either through the possible interaction with IgaAperip or BamA).

　　Accordingly, we cloned rcsFmim sequence in the medium copy number plasmid pSC232 with a C-terminal 3X flag tag. To this end, we mutated the lipobox Cys to Ser, to direct RcsFmim to the periplasm rather than the outer membrane, thus out ruling the rerouting of RcsFmim to the inner membrane (el Rayes et al. 2021). The plasmids were expressed into E. coli?K-12?MG1655 DH300 carrying a genomic rprA-lacZ fusion on the chromosome to assess Rcs system activation. rprA encodes a small non- coding RNA and its synthesis is controlled by the Rcs system (Majdalani et al. 2002). From now on, E.?coli?K-12?MG1655 DH300 will be designated as wild type strain.

　　Induction of expression of rcsFmim by adding 100?μM IPTG to the wildtype strain culture (where chromosomal rcsF is present) resulted in three-to-four-fold activation of the Rcs system (Fig.?4a). This activation was comparable to the isogenic strain transformed with prscFwt. When we treated the cultures expressing rcsFmim with to 0.3 μg/ml mecillinam (amdinocillin), a β-?lactam antibiotic known to induce the Rcs system, we observed a variation of Rcs activation over a wide range (Fig.?4b). To further investigate the effect of rscFmim independently of chromosomal rscF, we transformed prcsFmim-pSC232 in E.?coli?DH300 rcsF::kan thus testing if rcsFmim could relieve IgaA- mediated repression of the Rcs system. Rcs activation in rscF null mutant expressing rcsFmim in the absence and presence of 0.3 μg/ml of mecillinam were inconsistent, suggesting a possible perturbation of the cell envelope (Fig.?4b).

　　Fig. 4

　　Effect of RcsFmim on Rcs system activity and response to stress. a Activation of the Rcs system by expression of rcsFmim. Wild type E. coli DH300 containing either empty pSC232 or rcsFwt-pAM238 or prcsFmim-pSC232 were grown until OD600?=?0.6–0.8 and the β-galactosidase activity was measured as previously described. Error bars denote standard error of the mean and double asterix denote statistically significant difference at P?≤?0.05. b Effect of RcsFmim on Rcs system response to mecillinam. Wild type or rcsF null E. coli DH300 containing either rcsFwt-pAM238 or prcsFmim-pSC232 were grown until OD600?=?0.2 and then treated with mecillinam at a final concentration 0.3?μg/ml. The β-galactosidase activity was measured after one hour as previously described and the activation was calculated relative to the E.?coli DH300 containing empty pSC232

　　Taken together, our results suggest that RcsFmim could affect basal Rcs system signaling but might not be sufficient to replace wildtype RcsF.

　　As previously mentioned, RcsF interacts with two essential proteins in the outer and inner membrane of E. coli, BamA and IgaA, respectively (Konovalova et al. 2014; Cho et al. 2014; Hussein?et al.?2018; Rodríguez-Alonso et al. 2020). based on the predicted IgaAperip-?RcsFmim and BamA-?RcsFmim interactions, which seem overlapping with wild type RcsF interactions sites, we reasoned that RcsFmim peptide could potentially affect E.?coli growth. To enable a tighter control of expression of rcsFmim, we cloned its sequence in the medium copy number pNH401 (Hussein et al. 2018), derived from pBAD33 (Guzman et al. 1995) under the control the L-arabinose inducible promoter PBAD. First, we confirmed that rcsFmim-pBAD33 can induce the Rcs system in the presence of 0.2% L-arabinose (Supplementary Fig.?4). Then, we monitored the growth of E.?coli DH300 harboring rcsFmim-pBAD33 in comparison to the isogenic strains transformed with rcsFwt-pBAD33 upon induction with 0.2%?L-?arabinose. We observed that only the strain expressing rcsFmim showed severe growth delay (Fig.?5, Supplementary Fig.?5 and 6). To test if this growth delay resulted from the activation of the Rcs system, we transformed rcsFmim-pBAD33 and rcsFwt-pBAD33 in E. coli DH300 rcsB::kan (so turning OFF the Rcs system). To our surprise, RcsFmim resulted also in a significant delay of bacterial growth rate that was only abolished by the addition of D-fucose (a non-metabolizable analogue of L-arabinose that represses expression from the PBAD promoter) (Fig.?5, Supplementary Fig.?7).

　　Fig. 5

　　RcsFmim affects E. coli growth rate. Overnight cultures of either wild type or rcsB null E. coli DH300 containing rscFwt-pBAD33 or rcsFmim-pBAD33 were diluted in fresh LB media containing either 0.2% l-arabinose or 0.2% d-fucose and the OD600 was measured each hour for six hours. Triple asterix indicate statistically significant difference P?≤?0.01

　　Taken together, our results indicate that RcsFmim affects E.?coli growth predominantly in an Rcs independent manner despite causing Rcs system activation.

　　The delayed growth of E.?coli strains expressing rcsFmim independently of Rcs activation and the inconsistency of the complementation experiments, suggested that expression of rcsFmim might perturb E.?coli cell envelope.

　　Indeed, wild type and rcsB null mutants of E. coli DH300 transformed with rcsFmim-pBAD33 showed increased sensitivity towards 2%?SDS and 1?mM EDTA (Supplementary Fig.?8). To confirm further that RcsFmim affects E.?coli envelope integrity, we assessed the permeability of the outer membrane by measuring the uptake of the fluorescent probe ANS. ANS is a neutral probe that fluoresces upon interaction with hydrophobic ligands or in non-polar environment. Enhanced fluorescence of ANS is an indication of outer membrane perturbation in Gram-negative bacteria (Loh et al. 1984; Lamers et al. 2013; Sch?fer & Wenzel 2020; Tag ElDein et al. 2021) Accordingly, we assessed ANS fluorescence of mid- log phase E.?coli DH300 transformed with either rcsFmim-pBAD33 or prcsFwt-pBAD33 compared to the isogenic strain transformed with empty pBAD33.

　　only strains transformed with rcsFmim-pBAD33 showed enhanced fluorescence, thus indicating outer membrane defect (Fig.?6). We noticed, that expressing rcsFmim in the absence of chromosomal rcsF (rcsF::kan strain) or when the Rcs system is turned OFF (rcsB::kan strain), resulted in a higher fluorescence than in the wild type strain (plausible explanations are mentioned in the discussion).

　　Fig. 6

　　RcsFmim impairs E. coli outer membrane. The ANS fluorescence of E. coli DH300, rcsF::kan and rcsB::kan transformed with either rcsFwt-pBAD33 or rcsFmim-pBAD33 were calculated relative to that of E. coli DH300 containing empty pBAD33. Error bars represent the standard error of the mean, double asterix and triple denote statistical significance P?

　　Altogether, our results show that RcsFmim moderately alter Rcs signaling but severely affect bacterial growth through perturbation of outer membrane integrity (Fig.?6).

　　The problem of antibiotic resistance is a global health challenges threatening humanity, especially in the post-COVID-19 era, as the mortality rates of antimicrobial resistance patients are on the rise (Elmahi et al. 2022; Sulayyim et al. 2022; WHO 2017). Combined efforts are needed to develop non-classically acting antimicrobials. Since Gram-negative cell envelope integrity is a cornerstone in bacterial survival and homeostasis, cell envelope and/ or signaling pathways monitoring its integrity might provide important druggable targets (Hartmann et al. 2010; Bem et al. 2015).

　　Despite the presence of in-depth studies of bacterial signaling pathways including TCSs, the research area of interfering with signaling pathways sensing and responding to cell envelope challenges via altering protein–protein interaction is largely unexplored although promising (Kahan et al. 2021).

　　In this work, we suggest a potential antibacterial peptide based on the confirmed but unmapped interaction of RcsF- IgaA. The RcsF-IgaA interaction is reported to increase under Rcs inducing stresses (Cho et al. 2014; Hussein et al. 2018) and to be the first step of activation of the Rcs system. We hypothesized that altering this interaction can potentially compromise this pathway and/ or membrane integrity. Antimicrobial peptides are generally formed of 50–100 amino acids and have a general mechanism of action of disrupting the cell envelope (Barreto-Santamaría et al. 2021; Bechinger & Gorr 2017; Sch?fer & Wenzel 2020). Hence, we predicted that interfering with RcsF-IgaA interaction will lead to a sequence for a probable new antimicrobial peptide affecting signaling through the Rcs system and/ or affecting cell envelope integrity.

　　We based our reasoning that Rcs is a complex phosphorelay, with three auxiliary proteins (RcsF, IgaA and RcsD). One main advantage of altering signaling through the Rcs system is its highly conserved nature in various Enterobacterial genera (Wall, et al. 2018; Majdalani & Gottesman 2005; Meng et al. 2021). Accordingly, an Rcs-acting peptide may be efficient against E. coli, Salmonella spp, Klebsiella sp and Enterobacter spp.

　　based on previous work on RcsF structure (Leverrier et al. 2011; Rogov et al. 2011) and our docking analysis, we predicted 16 amino acids residues to be important for RcsF-IgaA interaction and/ or RcsF signaling role. We included those 16 amino acid residues in a hypothetical peptide, RcsFmim, and predicted its interaction with IgaA.

　　Prior to submission of this work, Lach and co-workers published the results of three independent genetic screens which identified, among others, RcsF mutants defective in signaling to IgaA. Some of the RcsF residues identified in their study were included between Gln93 to Lys134, identified through our in silico approach and included in RcsFmim (Lach et al. 2023).

　　Previous work by Rodriguez- Alonso showed that interaction of wild type RcsF with BamA β-barrel could be divided into three zones, Zone 1 (Z1) including Arg488, Asn643, Tyr 465 and Gln466, Zone 2 (Z2) including Arg592 and Arg634 and Zone 3 (Z3) which is one of the components of the proposed lateral gate of the β barrel (Rodríguez-Alonso et al. 2020). Computational docking of RcsFmim with BamA revealed that RcsFmim occupied the interaction region in BamA and established some hydrogen bonds similar to wild type RcsF in addition to extra hydrogen bonding with other amino acids in the interaction zones (Arg391, Asn459, Arg 526, Glu470, Gly486 and Pro782). Hence, we hypothesized that RcsFmim might interact with BamA or compete with the wildtype RcsF causing an envelope-disrupting phenotype. Both this assumption, structure of RcsFmim and its conformational changes upon binding to partners are yet to be confirmed experimentally.

　　In addition, we demonstrate that expression of rscFmim resulted in increased membrane permeability and a significant delay in E. coli cell growth. This growth delay seems to be related to a general stress on the cell envelope rather than the specific activation of the Rcs system for three reasons. First, the growth defect of E. coli containing a medium copy number plasmid with rcsFmim is not abolished when the Rcs system is turned OFF (rcsB null mutants). Second, normal growth is restored only in presence of D-fucose whether the Rcs system is functional or not. Third, E. coli expressing rcsFmim exhibited increased permeability to ANS indicating outer membrane defect (Fig.?6). The permeability increased in rcsF null mutant, probably due to the absence of adequate interaction with wild type RcsF protein partners, leaving room for only aberrant interactions with RcsFmim. Similarly, we observed an increased fluorescence in rcsB::kan strain compared to?wild type strain when both contained rcsFmim-pBAD33, which may outline a limited role of a functional Rcs system in balancing the perturbation caused by RcsFmim.

　　Interestingly, rcsFmim does not hinder chromosomal RcsF interaction with IgaA, as demonstrated by the ability of the wildtype strain transformed with prscFmim-pBAD33 to respond to mecillinam. However, we could not conclude whether rscFmim is sufficient to complement the rscF null mutant.

　　The specific in situ target(s) of RcsFmim, is/are yet to be explored. It is noteworthy to mention that RcsF interacts with the outer membrane porin OmpA, which also plays an important role in preserving the outer membrane integrity (Konovalova et al. 2014; Cho et al. 2014; Dekoninck et al. 2020). Whether the effect of RcsFmim is mediated through an aberrant interaction or a stalled complex formation with one or more of the studied proteins (IgaA or BamA, or OmpA) is still under investigation.

　　Overall, we suggest the sequence of the new peptide RcsFmim, that may serve as an interesting research tool to study cell envelope perturbation and the Rcs system signaling. Future studies are also needed to test whether the purified RscFmim can be uptake by intact E. coli, causing a bacteriostatic or bactericidal effect.

　　Below is the link to the electronic supplementary material.

　　ccDownload:/content/pdf/10.1007/s00203-023-03733-3.pdf