Bacteria and plasmids used in this study are listed in Table 1. Escherichia coli ATCC 25922 was obtained from the China Institute of Veterinary Drug Control. Strain EΔhns is a derivative of E. coli ATCC 25922 via the one-step inactivation of chromosomal gene hns (17). The complementary strain EΔhns/phns was constructed as follows: Firstly, the complete open reading frame of hns was amplified by PCR from the genomic DNA of E. coli ATCC 25922 using primers XhoI-hns–F/HindIII-hns-R (Table 2). Thereafter, the expression plasmid pBAD::hns was generated by inserting the target fragment to the vector pBAD (Karsbad, CA, the United States) and then was introduced to EΔhns by electroporation. The overexpressed hns strain E/phns is a derivative of E. coli ATCC 25922 that was introduced of pBAD::hns by electroporation. All strains were cultured in fresh Luria-Bertani (LB) broth (Beijing Land Bridge Technology Co., Ltd.) at 37°C, where strains EΔhns, EΔhns/phns and E/phns were induced by 0.2% L-arabinose (18).

A single colony of strains, such as E. coli ATCC 25922, EΔhns, EΔhns/phns and E/phns were cultured in LB medium at 37°C. After growth overnight, the cultures were diluted 1:100 in fresh medium and grown to an OD₆₀₀ of 0.5. Following this, the total bacterial RNA was extracted with a TaKaRa MiniBEST universal RNA extraction kit (TaKaRa Bio, Inc., Shiga, Japan). The quantity of extracted RNA was determined by A₂₆₀ measurements and purity was evaluated by the A₂₆₀/A₂₈₀ ratio using a NanoDrop 1000 Spectrophotometer (Thermo Scientific, Hvidovre, Denmark). The cDNA samples were synthesized using the cDNA reverse transcription kit with gDNA Eraser (TaKaRa Bio, Inc.) and then the RT-qPCR was performed with an annealing temperature of 60°C according to our previous study (18). The relative expression levels were calculated by the 2^−ΔΔCt method and compared with the levels of E. coli ATCC 25922. The 16S rRNA gene was chosen as a housekeeping gene, and the primers of RT-qPCR are listed in Table 1. Three independent biological replicates were performed.

Growth curves for E. coli ATCC 25922, EΔhns, EΔhns/phns and E/phns were plotted. Overnight cultures were inoculated in fresh LB medium and grown to an OD₆₀₀ of 0.5. Then the cultures were diluted (1:1,000) in preheated fresh LB broth and inoculated with shaking at 37°C. The OD₆₀₀ was measured periodically at 1-h intervals using an ultraviolet spectrophotometer and shown as means ± standard deviations. Experiments were performed with at least three biological replicates.

Antimicrobial susceptibility testing was tested using the broth microdilution method according to the guidelines of Clinical and Laboratory Standards Institute (CLSI) guidelines (19). The antimicrobial agents are cefotaxime, gentamicin, amikacin, doxycycline, tigecycline, florfenicol, and enrofloxacin. Experiments were performed with at least three biological replicates.

Overnight cultures of E. coli ATCC 25922, EΔhns and EΔhns/phns were diluted (1:1,000) in fresh LB broth and grown to an OD₆₀₀ of 0.5. Bacteria were cultured in the LB broth with different concentrations of gentamicin (0.25, 0.5, 1, and 2 μg/mL) for 24 h. At regular intervals, the culture broths were serially diluted with 0.9% saline and plated onto LB agar plates. Colony forming units (CFU) were counted after 18 h of incubation at 37°C, and the Log₁₀ CFU values were calculated according to previous study (20). Take the Log₁₀ CFU values as the vertical axis, and the incubation time (h) as the horizontal axis to establish the time sterilization curve. Three independent biological replicates were performed.

Overnight cultures of E. coli ATCC 25922, EΔhns and EΔhns/phns were diluted 1:100 in fresh LB medium and incubated at 37°C. Samples equivalent to an OD₆₀₀ of 0.5 were removed and suspended with 5 mmol·L⁻¹ N-2-hydroxyethylpiperazine-N'-2- ethanesulfonic acid (HEPES, pH 7.0, plus 5 mmol·L⁻¹glucose). Thereafter, the OD₆₀₀ of the above bacteria suspension was standardized to 0.5 using the same buffer. Fluorescent probe NPN (10 μM) was used to evaluate the OM integrity of strains (21). Fluorescence intensity was measured using a Sparpk 10M Microplate reader (Tecan) with the excitation wavelength at 350 nm and the emission wavelength at 420 nm.

The culture suspension of E. coli ATCC 25922 and EΔhns were prepared according to the method described in 2.6. Thereafter, the bacterial suspensions were centrifuged, washed softly with phosphate-buffered saline (PBS) buffer for three times, and mixed with 2.5% (w/v) glutaraldehyde in 0.1 M PBS buffer at 25°C for about 2 h away from light. Subsequently, the samples were dehydrated in sequential with a graded ethanol series (30%, 50%, 70%, 90%, and 100%), transferred to a culture dish, dried overnight, and observed using SEM (Hitachi, Japan).

Bacteria were cultured and standardized to an OD₆₀₀ of 0.5 using similar protocols described in 2.6. DiSC₃ (5) (0.5 μM) (Aladdin, Shanghai, China) was added to achieve a final concentration of 2 μM in the mixture to determine the transmembrane electric potential (Δψ). The dissipation for Δψ in E. coli ATCC 25922, EΔhns and EΔhns/phns was measured with the excitation wavelength of 622 nm and emission wavelength of 670 nm according to the method described in the previous study (22).

The accumulation of ethidium bromide (Beyotime, Shanghai, China) in the cells was monitored as previously described with some modifications (21). Strains of E. coli ATCC 25922, EΔhns and EΔhns/phns were grown to an OD₆₀₀ of 0.5, then centrifuged and suspended with 5 mmol·L⁻¹ HEPES (pH 7.0, plus 5 mmol·L⁻¹glucose). Subsequently, the OD₆₀₀ of the bacterial suspension was standardized to 0.3 using the same buffer. 5 μM of ethidium bromide was added to determine the efflux pumps activity. Fluorescence intensity was determined with a Spark multifunctional microplate reader with excitation wavelength 530 nm and emission wavelength 600 nm. Each experiment was conducted in triplicate at least three times.

Total RNA of strains E. coli ATCC 25922 and EΔhns were extracted by the RNAprep Pure Cell/Bacteria Kit (TIANGEN Biotech (Beijing) Co., Ltd) and sequenced using the Illumina NovaSeq 2000 system by Novogene Bioinformatics Technology Co., Ltd. Differentially expressed genes (DEGs) were identified by using the fragments per kilobase of transcript per million mapped reads method with P_adj < 0.05 and fold change (FC) of | log₂FC | > 1. RNA-sequencing reads were aligned to E. coli ATCC 25922 (Ref-Seq accession no. CP025268).

Statistical analysis and the generation of plots were performed using GraphPad Prism 8. Unpaired t-test (normally distributed data) between two groups was used to calculate P-values. Differences with P < 0.05 were considered as significant difference. Significance levels are indicated by numbers of asterisks (ns, No significant difference, ^* P < 0.05, ^** P < 0.01, ^*** P < 0.001, ^**** P < 0.0001).

The mRNA expression levels of hns gene in E. coli ATCC 25922, hns deletion strain EΔhns, complemented strain EΔhns/phns and overexpressed strain E/phns were measured by RT-qPCR (Figure 1). Compared with strain EΔhns, the hns expressions of strain EΔhns/phns increased by 44-fold, from 0.006 ± 0.001 increasing to 0.264 ± 0.015, although they were still sharply lower than that of E. coli ATCC 25922 (P < 0.0001), demonstrating that the complement of hns could only partially restore the function of H-NS in the deletion strain EΔhns, which may be closely linked to the location of H-NS, such as on chromosome or plasmids (1, 2, 23). Meanwhile, we observed that the hns expressions of strain E/phns (52.232 ± 4.182) were 52.23-fold increments in comparison with the reference strain E. coli ATCC 25922, proving that strain E/phns overexpressed hns.

Growth kinetics of strains E. coli ATCC 25922, EΔhns, EΔhns/phns and E/phns were established respectively based on optical density at 600 nm values (Figure 2). No difference in growth was observed between E. coli ATCC 25922 and E/phns, which inferred that hns overexpressions had no effect on the growth of E. coli. It's worth noting that the OD₆₀₀ values of strains EΔhns and EΔhns/phns were significantly lower than those of E. coli ATCC 25922 at 3–6 h incubation time, although the OD₆₀₀ values rose to the control level at the later incubation, in line with previous studies (24). The aforementioned results reflect that hns deletion will lead to reduced adaptability of E. coli, although it does not affect the bacterial survival.

To determine whether hns deletion alters the susceptibility of E. coli to a variety of drugs, such as gentamicin, amikacin, cefotaxime, doxycycline, tigecycline, florfenicol, and enrofloxacin, the MIC values were determined for E. coli ATCC 25922, EΔhns, EΔhns/phns and E/phns (Table 1). In contrast to E. coli ATCC 25922 (MICs = 0.008 ~ 2 μg/mL), the MICs for strain EΔhns were significantly decreased, with gentamicin decreased by 16 folds, followed by amikacin (8-fold) and enrofloxacin (4-fold), while cefotaxime, doxycycline, tigecycline and florfenicol displaying 2-fold decreases. As expected, the MICs for strain EΔhns/phns partially recovered, while for strain E/phns was the same as that of the control. Collectively, our data illustrate that protein H-NS negatively regulates the multidrug resistance of E. coli.

To further investigate the impact of hns deletion on the susceptibility of strains, we conducted in vitro time-dependent killing assay of gentamicin against E. coli ATCC 25922, EΔhns and EΔhns/phns. As shown in Figure 3, strains E. coli ATCC 25922, EΔhns and EΔhns/phns were inhibited in a concentration-dependent manner after incubation with different concentrations of gentamicin from 0 h to 24 h. In addition, the bactericidal effect of gentamicin against EΔhns displayed markedly stronger than that against E. coli ATCC 25922. For strain EΔhns, the number of viable bacteria began to decrease obviously after incubation with 0.25 μg/mL of gentamicin for 3 h (Figure 3A), and all died after incubation with 2 μg/mL for 12 h (Figures 3C, D), whereas for E. coli ATCC 25922, the bactericidal effect was observed only after incubation with 2 μg/mL of gentamicin for 4 h (Figure 3D). In conclusion, hns deletion leads to increased susceptibility of E. coli ATCC 25922 to gentamicin.

Notably, there was no significant difference in the bactericidal effect of gentamicin against strains EΔhns and EΔhns/phns, suggesting that (1) the expressions of hns located on pBAD could not completely complement the deletion of hns on chromosome, which had been also confirmed by the expression of hns in recombinant bacteria (Figure 1) and the results of in vitro antimicrobial susceptibility testing (Table 1); (2) the deletion of hns gene might lead to bacterial damages that were difficult to reverse even after the gene hns was complemented, such as membrane damages, although further studies were needed.

To understand the molecular mechanism by which H-NS regulates the susceptibility of E. coli, we determined the OM permeability changes in E. coli ATCC 25922, EΔhns and EΔhns/phns using 1-N-phenylnaphthylamine (NPN) as a hydrophobic fluorescent probe (Figure 4A). The results demonstrated that strain EΔhns exhibited dramatically stronger NPN uptakes than that of E. coli ATCC 25922, implying that the absence of hns led to the elevated OM permeability of E. coli. In parallel, compared with strain EΔhns, the fluorescence values of complementary strain EΔhns/phns decreased significantly, although it still showed higher than E. coli ATCC 25922, which was in agreement with the expression levels of hns gene in the complementary strain. Soon afterwards, we further determined the permeability of inner membrane (IM) of the aforementioned bacteria using propidium iodide (PI) as a fluorescent probe (21) and no evident changes were observed (Figure 4B). In summary, hns deletion increases OM permeability of E. coli, however has little effect on IM.

To clarify the impact of hns deletion on the morphology of E. coli, morphological changes of strains E. coli ATCC 25922 and EΔhns were examined by SEM. In comparison with E. coli ATCC 25922, the cell surface of strain EΔhns exhibited different degree of deformations and devastation (Figure 5). As shown in Figure 5, long rod-shaped bacteria increased significantly in strain EΔhns, speculating that it might be related to the slow growth caused by the deletion of the hns gene (Figures 5C, D). In addition, some cells of strain EΔhns displayed observable membrane damage, which may be one of the reasons for the increased uptake of aminoglycosides by strain EΔhns. Recently, Lv et al. proved that hypoionic shock-induced cell membrane damage could dramatically increase the bacterial uptake of aminoglycosides, which enhanced bactericidal action of the antibiotics (25).

Earlier studies have documented that the SOS response is an extremely important molecular instrument of bacteria which allows it to deal with diverse DNA damages (26), but at the same time, activation of the SOS response accelerates the synthesis of the SulA protein, which can arrest cell division (27). In the present study, strain EΔhns not only grew slowly, but also showed cell membrane damage. Therefore, we speculate that the deletion hns gene in E. coli ATCC 25922 leads to bacterial membrane damage, thus activating the SOS response, which in turn hinders bacterial growth.

It is well-known that aminoglycosides do not require energy to cross the OM of bacteria, where OM damage and increased expression of porin-related genes can accelerate absorption, while PMF is required to provide energy when they pass through the IM (28–30). Bacterial PMF, an energetic pathway located on the bacterial membrane, consists of Δψ and the transmembrane proton gradient (ΔpH).

To verify whether PMF altered among strains E. coli ATCC 25922, EΔhns and EΔhns/phns, we first assessed the dissipation of Δψ by the fluorescent probe 3,3′-Dipropylthiadicarbocyanine Iodide [DiSC₃ (5)]. When Δψ dissipates, the fluorescent probe DiSC₃ (5) is released into the buffer solution, resulting in a significant increase in fluorescence intensity. Our findings showed that compared with E. coli ATCC 25922, the fluorescence intensity of strain EΔhns was decreased dramatically (P < 0.0001) (Figure 4C), indicating that the dissipation of Δψ in strain EΔhns was sharply repressed. Further, the fluorescence intensity of complemented strain EΔhns/phns reverted to the levels of E. coli ATCC 25922. Similarly, ΔpH was also determined by the fluorescent probe BCECF-AM (20 μM, Beyotime, Shanghai, China) and the results demonstrated that no observable changes were found in the tested bacteria (Figure 4D). These results imply that the elevated PMF of strain EΔhns is mainly due to the inhibition of Δψ dissipations, which enhances the uptake of aminoglycosides.

Given that the intracellular accumulation of antibiotics is determined by both influx and efflux (31), we also determined efflux pumps activity changes in E. coli ATCC 25922, EΔhns and EΔhns/phns using ethidium bromide as a hydrophobic fluorescent probe (Figure 4E). Compared with E. coli ATCC 25922 and complementary strain EΔhns/phns, the fluorescence intensity of strain EΔhns exhibited higher (P < 0.0001), suggesting that its efflux activity was significantly inhibited.

To gain a deeper understanding of H-NS regulation mechanism on the susceptibility of E. coli, we further performed the transcriptomic analysis. In comparison with the control E. coli ATCC 25922, Gene Ontology (GO) annotation analysis (Figure 6A) showed that DGEs of strain EΔhns were mainly correlated with adhesion, stress response and chemotaxis of biological processes, lyase activity of molecular function, and pili of cellular component. Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis (Figure 6B) revealed that DEGs in strain EΔhns were enriched significantly in energy metabolisms.

Specifically, the genes with significantly increased expression were involved in glycolysis and electron transport chain of respiratory process (Figure 6) in strain EΔhns. As shown in Figure 7A, the expression levels of glycolytic-related genes presented distinctly elevated. In parallel, the genes involved in respiratory chain, such as NADH-Q oxidoreductase-related gene ndh and cytochrome oxidase-related genes appA, appB, appC, cydA, and cydB, were all obviously up-regulated (Figure 7B). Previous studies have verified that an increase in glycolytic metabolites can facilitate the conversion of redox energy to PMF, thereby promoting aminoglycosides uptake (28, 32, 33). Accordingly, to identify whether the respiratory chain is activated, we determined the intracellular NAD⁺/NADH ratio in strains EΔhns, EΔhns/phns and E. coli ATCC 25922 using an NAD⁺/NADH Assay Kit (WST-8, Beyotime, Shanghai, China). As expected, the NAD⁺/NADH ratio of strain EΔhns was remarkably higher than that of E. coli ATCC 25922 (P < 0.0001) (Figure 4F), suggesting that the deletion hns gene in E. coli ATCC 25922 can stimulate glycolysis and promote the conversion of redox energy into PMF, thus facilitating the uptake of aminoglycosides.

Further analysis found that many efflux-related genes (Figure 7C) presented down-regulated, such as clbM (34), emrE (35, 36), acrD (37), and acrAB (38), which were helpful to reduce drugs efflux. To confirm whether the expression levels of above genes were altered in strain EΔhns, we examined their mRNA relative expression levels using RT-qPCR (Figure 8A). The results demonstrated that the expression levels in strain EΔhns were evidently lower than those in E. coli ATCC 25922, approximately decreased by 99.0% (clbM), 68.3% (emrE), 49.7% (acrB), and 39.2% (acrA), respectively. Thus, reduced expression levels of genes associated with the efflux system can inhibit efflux activity, thereby helping to slow drugs efflux.

Due to porin-related genes ompN, ompF, ompC, and ompG (Figure 7D) showed obvious upregulations, we also examined their mRNA expression levels in strains E. coli ATCC 25922 and EΔhns using RT-qPCR. We found that the expression levels of these genes were markedly higher in strain EΔhns than that in E. coli ATCC 25922 (Figure 8B), with the highest levels of ompN (i.e., 22.21-fold higher), followed by ompF (5.06-fold higher) and ompG (3.06-fold higher), whereas the levels of ompC exhibited only a marginal increase. Porins OmpC and OmpF passively diffuse small hydrophilic molecules (≤ 500 Da) into the periplasm (39), and their expressions are mutually exclusive, i.e., when ompF is on, ompC is off, and vice versa (40). OmpG, which belongs to the subclass of porins, harbors a larger channel than OmpF and OmpC, and also allows for translocation (41). OmpN is one of the minor porins, although its translocation function has not been reported (42). Therefore, it is conceivable that the elevated expression levels of porin-related genes will further destabilize the OM barrier and contribute to the increase of OM permeability of strain EΔhns.

Intriguingly, there were many down-regulated DEGs in the flagellar system in strain EΔhns (Figure 7E), compared to the reference strain E. coli ATCC 25922. Further, it can also be seen from Figure 5 that the flagella of E. coli ATCC 25922 are clearly visible, while that of strain EΔhns is almost invisible. Many studies have revealed that flagellum is a locomotive organelle and can affect bacterial adhesion, invasion and biofilm formation (37, 43). To further confirm whether the motility of strain EΔhns has changed, we fulfilled the swimming motility assay of strains E. coli ATCC 25922, EΔhns, EΔhns/phns, and E/phns using LB plates with 0.3% agar according to the method of previous studies (44, 45). The results demonstrated that strain EΔhns had the shortest swimming distance (≈ 9.3 mm), followed by EΔhns/phns (≈ 12.8 mm) and E. coli ATCC 25922 (≈ 13.8 mm), while E/phns had the longest distance (≈ 14.7 mm), suggesting that H-NS modulates the motility of E. coli by positively governing the expression of flagellate-related genes.