Bacterial strains and plasmids used in this study are listed in Supplementary Table S1. Z. mobilis ZM4 (ATCC 31821) was used as the parental strain in this study. Generally, ZM4 and its derivative strains are cultured at 30°C with shaking at 100 rpm in RM medium (50 g/L glucose, 10 g/L yeast extract, 2 g/L KH₂PO₄, and 1.5% agar for solid). To avoid the effect of other amino acid nutrients in RM medium, MM medium (50 g/L glucose, 1 g/L KH₂PO₄, 1 g/L K₂HPO₄, 1 g/L (NH₄)₂SO₄, 0.5 g/L NaCl, 0.42 g/L MgCl₂ 6H₂O, 0.001 g/L calcium pantothenate, and 1.5% agar for solid) was used in this study for Na₂SO₄ supplementation experiments, and 4 g/L Na₂SO₄ was added. The final concentration of exogenous ethanol added to the RM and MM medium is 8% and 3% (v/v), respectively.

E. coli DH5α was used for plasmid construction in this study. All E. coli strains were cultured in LB medium (10 g/L tryptone, 5 g/L yeast extract, 10 g/L NaCl, and 1.5% agar for solid) at 37°C, 250 rpm. When required, 100 and 200 μg/ml spectinomycin was added for E. coli or Z. mobilis, respectively.

The shuttle plasmid pEZ15A was used for hfq overexpression in Z. mobilis ZM4 with its native promoter and the recombinant strain named ZM4-hfq. The plasmid pL2R was used for hfq deletion using the native Type I-F CRISPR-Cas system of Z. mobilis (Zheng et al., 2019). Plasmid pEZ15A was transformed into the wild type and the hfq deletion strain, which named ZM4 and ZM4-Δhfq, respectively.

For the gene-deficient mutant construction, interference plasmids were initially constructed with spacer. The spacer was designed to bear the entire 32-bp sequences containing a 5′-CCC-3′ PAM. The editing plasmid pL2R was digested with Bsa I at 37°C overnight. Oligonucleotides were annealed by first heating the reaction mixture to 95°C for 5 min and subsequently cooling down gradually to room temperature. The annealed spacer and the digested linear DNA vector were enzymatically linked with T4 ligase at 18°C overnight, and then transformed into DH5α competent cells to generate genome engineering plasmids.

For all plasmid construction, primers were designed to contain a region of 15–25 nucleotides that overlap with adjacent DNA fragments and synthesized by TsingKe (Beijing, China) (Supplementary Table S2). All plasmids were assembled by Gibson assembly method (Gibson et al., 2009). Gene and vector fragments amplified by primer pairs were purified and then ligated through the T5 exonuclease (NEB, WA, United States). The gene fragment was mixed with the vector at a 3:1 M ratio, 0.5 U T5 exonuclease (NEB, MA, United States) and 0.5 μl buffer 4 (NEB, MA, United States) were then added, with ddH₂O added to the final volume of 5 μl. All reagents were mixed and incubated on ice for 5 min before being added to chemically competent E. coli cells. After ice incubation for 30 min and heat-shock for 45 s at 42°C, held it on ice for 2 min before 100 μl NZY medium (5.0 g/L yeast extract, 5 g/L NaCl, 1.2 g/L MgCl₂, 1.5 g/L MgSO₄, 3.6 g/L glucose, 10 g/L casein enzymatic hydrolysate NZ amine®) were added to the mixture above, and recovered at 37°C, 250 rpm. The cells were plated on LB agar plates containing 100 μg/ml spectinomycin, the recombinant was screened by colony PCR and confirmed by Sanger Sequencing (TsingKe Biotechnology, Beijing, China).

After identification, the recombinant plasmid was transformed into Z. mobilis competent cells via electroporation (0.1-cm electrode gap, 1600 V, 200 Ω, 25 μF) using a Gene Pulser® (Bio-Rad, CA, United States) following the method developed for Z. mobilis (Okamoto and Nakamura, 1992). The correct colonies were selected by colony PCR. After three to five generations purification on RM agar plates with 200 μg/ml spectinomycin supplementation, the correct recombinant was stored at −80°C.

Seed culture of Z. mobilis was prepared by reviving the frozen glycerol stocks in 50 ml flasks containing 40 ml RM medium. After culturing overnight without shaking to the mid-exponential phase, the seed culture was harvested and inoculated in 50 ml shake flasks containing 40 ml RM or MM medium with an initial OD_600 nm value of 0.1. During the fermentation, cell growth in terms of the absorbance value (OD₆₀₀) was measured at 600 nm by a spectrophotometer (UV-1800, AOE, China) at different time points. Samples were centrifuged at 12,000 rpm for 2 min, collected supernatants were passed through 0.22-μm filters and stored at −80°C for subsequent HPLC analysis if needed. Glucose and ethanol in the culture supernatant were determined using a HPLC system LC-20AD (Shimadzu, Japan) with an Aminex HPX-87H column (300 × 7.8 mm, Bio-Rad, CA, United States) at 65°C. Elution was performed with 5 mM H₂SO₄ at 0.5 ml/min.

Cell culture samples in different conditions were collected at the mid-exponential phase. According to standard Illumina protocols, RNA-Seq was carried out by GENEWIZ (Suzhou, China) with a library construction kit (NEBNext® Ultra™ Directional RNA Library Prep Kit for Illumina®) and an Illumina HiSeq 2000 instrument. After RNA-seq fastq data quality was evaluated by FastQC software (Babraham Bioinformatics, United Kingdom). Data were filtered and those passed the quality control were imported to the CLC Genomics Workbench (version 14.0) for reads trimming and RNA-Seq analysis to get the RPKM value (reads mapping to the genome per kilobase of transcript per million reads sequenced) of each gene. Genome sequence of Z. mobilis was used as the reference for RPKM calculation (Yang S. et al., 2018).

Gene expression normalization, analysis of variance (ANOVA), and hierarchical clustering analysis were conducted using JMP Genomics (Ver. 9.0, SAS Inc., NC, United States) to identify differentially expressed genes at different conditions. Significantly differentially expressed genes were identified with p-value ≤0.05 and log₂-fold change ≥1 (significant induction) or ≤−1 (significant inhibition) as selection thresholds. The one-way ANOVA analysis was performed by JMP Genomics and only those with p-value ≤0.05 were considered for further analysis.

Intracellular ROS levels were measured by 2′,7′-dichlorodihydro-fluorescein diacetate (H₂DCF-DA) (Beyotime Biotechnology, Hangzhou, China) (Yan et al., 2022). Cells were collected at the mid-log phase (total OD₆₀₀ was 1.0) at 12,000 rpm for 1 min, washed with phosphate buffer saline (PBS) and then resuspended in 500 μl PBS with the final concentration of H₂DCF-DA at 100 μM. After incubating in darkness for 1 h at 30°C, 100 rpm, cells were collected, washed three times with PBS and resuspended in 300 μl PBS. The CytoFLEX FCM flow cytometry (Beckman Coulter, CA, United States) was used to detect the DCF fluorescence with excitation and emission wavelength of 488 nm and 525 nm, respectively. Cells with fluorescence intensities ranging 10³–10⁵ were selected and counted, and 20,000 cells were analyzed for each sample.

The plasmid pESE which is modified from pESD (Yi et al., 2013) was used and the TEV protease mediated Yeast Endoplasmic Reticulum Sequestration Screening (YESS) system was construction in Z. mobilis for the Protein-Protein interaction detection (Supplementary Figure S1). For the recombinant plasmid construction, two target genes and the vector fragment were PCR amplified and purified, and then assembled by the Golden gate assembly strategy. The reaction included 0.05 mol gene fragments, 20 ng pESE, 1 μl T4 ligase buffer (Thermo), 0.2 μl T4 ligase (Thermo), 0.3 μl Bsa I (NEB), 0.1 μl BSA (Takara), and with ddH₂O added to a final volume of 10 μl. All reagents were mixed and incubated on 37°C for 3 h before being added to chemically competent E. coli cells. After being incubated on ice for 30 min, heat-shocked for 45 s at 42°C, and held on ice for 2 min, 100 μl of NZY medium was added into above mixture and recovered for at least 1 h at 37°C with shaking at 250 rpm. The cells were plated on LB agar plates containing 50 μg/ml ampicillin, the recombinants were then screened by colony PCR and confirmed by Sanger Sequencing (TsingKe Biotechnology, Beijing, China).

The correct recombinant plasmids were transformed to S. cerevisiae EBY100 competent cells and then plated on SD-UT agar plates (2% galactose, 0.67% yeast nitrogen base w/o amino acids, 1% casamino acids, 1.5% agar) at 30°C. Single colony was selected and cultivated in 25 ml tube containing 5 ml SD-UT medium for 14 h with shaking at 200 rpm. The seed culture was then harvested and transferred into 25 ml tube containing 5 ml SD-DT medium (2% glucose, 0.67% yeast nitrogen base w/o amino acids, 1% casamino acids, 1.5% agar) with an initial OD_600 nm value of 0.8. After induction at 30°C for 8 h, 10⁶ cells were collected by centrifugation, washing two times with 150 μl PBS buffer at 3,000 rpm, 4°C for 1 min, and finally resuspended in 20 μl PBS. For Fab display analysis, cells were incubated with 0.1 μl anti-HA-FITC and/or 0.1 μl anti-FLAG-iFlor647 (GenScript, Nanjing, China) at 4°C for 15 min in darkness. After washed three times with PBS (pH 7.4) supplemented with 0.5% BSA at 3,000 rpm, 4°C for 1 min, the sample was resuspended in 300 μl PBS (pH 7.4) with 0.5% BSA. Flow cytometry analysis was performed by the Beckman Coulter CytoFLEX (Brea, CA) equipped with 488 and 633 nm lasers and 525/40 and 660/20 nm band-pass filters.

To investigate the effect of hfq on ethanol tolerance in Z. mobilis, recombinant strains with hfq overexpression (ZM4-hfq) or hfq deletion (ZM4-Δhfq) were constructed and evaluated in terms of cell growth, sugar utilization, and ethanol production. As shown in Figure 1A, the hfq overexpression strain ZM4-hfq grew normally without ethanol treatment and achieved a final OD₆₀₀ value of 4.92 that was similar to the wild-type with OD₆₀₀ value of 4.84. While, the hfq deletion strain ZM4-Δhfq grew slightly slower than wild type with a final OD₆₀₀ value of 4.20. Corresponding to the slower cell growth, ZM4-Δhfq exhibited slower glucose consumption and ethanol production than the control and ZM4-hfq (Figure 1B). These results indicated that Hfq plays roles for normal cell growth even without ethanol stress inhibition.

Fermentation performance was further evaluated with 8% exogenous ethanol added into the media. Cell growth of ZM4 was effectively improved with hfq overexpression. ZM4-hfq grew into the stationary phase around 24 h and achieved a final OD₆₀₀ value of 3.10, while wild-type strain had an OD₆₀₀ value of 2.33 only after 56 h (Figure 1C). Consistently, hfq deletion in Z. mobilis significantly decreased the ethanol tolerance. The maximum biomass of strain ZM4-Δhfq exposure to 8% ethanol was quite lower than that of the control with OD₆₀₀ of 1.78 vs. 2.33. Corresponding to the cell growth, ZM4-hfq strain accelerated the glucose consumption and ethanol production, while strain ZM4-Δhfq was in reverse. As shown in Figure 1D, all glucose was consumed by ZM4-hfq to achieve a maximum ethanol titer of 25.82 g/L within 24 h, almost 6 h faster than that of control, while 6.83 g/L residual glucose was still detected at 36 h by strain ZM4-Δhfq. All these results indicated that hfq in Z. mobilis contributes to the ethanol tolerance, just as its positive roles responding to other stresses such as furfural and acetic acid (Yang et al., 2010; Nouri et al., 2020).

Previous reports have shown that Hfq is associated with the regulation of numerous cellular pathways (Vogel and Luisi, 2011; Sharma et al., 2018; Han et al., 2019). To gain insights into the genes regulated by Hfq, and to illustrate the molecular mechanism of the positive role of Hfq overexpression on ethanol tolerance in Z. mobilis, RNA-Seq was employed to explore the global transcriptional differences of ZM4 and its hfq derivatives ZM4-hfq and ZM4-Δhfq under different conditions of E0 (control, without exogenous ethanol treatment) and E8 (8% exogenous ethanol treatment) (Figure 1). Differentially expressed genes (DEGs) between different strains across different growth conditions were analyzed through analysis of variance (ANOVA) with a total of 628 DEGs identified (Supplementary Tables S3, S4).

The expression of hfq gene was examined first. As expected, the transcript of hfq was not detected in ZM4-Δhfq, while the expression of hfq in ZM4-hfq had a 3.2-fold upregulation compared with the wild-type ZM4 (p-value ≤0.05) (Supplementary Tables S3, S4). The gene expression of hfq confirmed that recombinant strains ZM4-Δhfq and ZM4-hfq were both constructed successfully, and the global gene expression was then further analyzed.

Consistent with previous studies (He et al., 2012; Yang et al., 2013), the presence of ethanol made many genes differentially expressed (E8 vs. E0) including 92 upregulated and 61 downregulated genes, which were mainly associated with cell wall/membrane biogenesis, metabolism, transcription, and general stress response. The effect of ethanol stress on three strains ZM4, ZM4-hfq, and ZM4-Δhfq were further compared and analyzed. The results showed that exogenous ethanol stress regulated more than 200 genes in all three strains (Supplementary Table S3). This result suggested that despite of hfq overexpression, Z. mobilis remained challenged by the ethanol stress, which was consistent with the decreased cell growth and reduced biomass accumulation of ZM4-hfq in the presence of 8% ethanol compared with that without ethanol treatment (Figures 1A, C).

The detailed gene expression information of different strain comparisons was analyzed as well (Supplementary Table S4). Unlike the dramatic transcriptomic changes in response to ethanol stress, less differentially expressed genes were observed among strains. Under E0 condition without exogenous ethanol treatment, 58 genes were differentially expressed in ZM4-Δhfq compared with ZM4, and 143 genes were differentially expressed when compared with ZM4-hfq. Meanwhile, only 30 genes were detected to be differentially regulated in ZM4-hfq compared with ZM4. When 8% exogenous ethanol was supplemented into RM (E8 condition), more differentially expressed genes were identified. Eighty genes were differentially expressed in ZM4-hfq compared with ZM4, while 77 genes and 181 genes were differentially expressed in ZM4-Δhfq compared with ZM4 and ZM4-hfq, respectively. The result of more genes differentially expressed under E8 condition was consistent with the fermentation performance analyses that hfq recombinant strains were obviously different from the parental strain ZM4 under ethanol stress conditions (Figure 1). All these genes identified between strain comparisons could be closely associated with the positive role of Hfq on ethanol tolerance in Z. mobilis.

Bacterial flagellum is a complex and dynamic nanomachine appended on the cell body that provides motility (Soutourina and Bertin, 2003). Recent studies reported that these complex organelles also play an important role in bacterial survival, reproduction, and pathogenicity, such as adhesion to a variety of substrates, secretion of virulence factors, and formation of biofilms (Nedeljkovic et al., 2021). Gene expression without ethanol supplementation (E0) showed that 20 flagellar related genes (total 34 genes in Z. mobilis), including flgBCEGHIJKL, flhA, fliCEFGHIKL, and motAB encoding flagellar structure proteins, motor proteins and biosynthesis proteins were upregulated in ZM4-Δhfq compared with the wild-type ZM4 or ZM4-hfq, and most of these genes were upregulated as well in the presence of 8% ethanol (E8) (Table 1, Supplementary Table S4).

Moreover, the expression of fliA, which encodes an alternative sigma factor specific for the flagellar regulons including flgK and fliC (Ohnishi et al., 1990), was two folds upregulated in ZM4-Δhfq compared with the wild-type ZM4 with or without ethanol presence. These results collectively suggested that Hfq in Z. mobilis might constitutively mediate through the sigma factor FliA to negatively regulate the expression of flagellar proteins, which has been reported in other Gram-negative microbes, such as Cronobacter sakazakii (Kim et al., 2015). Previous transcriptomic analyses in Z. mobilis reported that flagellar related genes were downregulated to help conserve energy from cell motility for survival in the stressful conditions (He et al., 2012; Yang et al., 2013). Therefore, the less ethanol tolerance in ZM4-Δhfq might be attributed to the limited energy resulting from the upregulation of the energy-costly flagellar assembly process without the negative regulation of Hfq.

In addition, gene expression in response to ethanol stress showed that above flagellar related genes were downregulated in all three strains, especially in ZM4-Δhfq with more than 2-fold repression (Supplementary Table S3). The downregulation of flagellar proteins in ZM4-Δhfq indicated that other mechanisms of flagellar regulation may exist in Z. mobilis especially under the ethanol stress condition, and further study is needed to unravel the underlying mechanisms. Altogether, the gene expression data suggested that the negative regulation of Hfq on these flagellar related genes is beneficial but may not be sufficient for the ethanol tolerance in Z. mobilis.

A gene cluster involved in sulfate assimilation and cysteine biosynthesis pathway was identified by comparing gene expression among strains under ethanol stress. Gene expression under ethanol treatment (E8) showed that all six genes associate with sulfate assimilation, cysCND (ZMO0003-0005) and cysHIJ (ZMO0007-0009), were upregulated in ZM4-hfq compared with ZM4-Δhfq or ZM4. Similar upregulation of gene expression in ZM4-hfq was observed in other genes involved in cysteine biosynthesis, including serA1 (ZMO1685) and serC (ZMO1684) (Table 1; Figure 2, Supplementary Table S4). However, such upregulated gene cluster was not induced under the condition without ethanol stress (E0). These results indicated that the sulfur metabolism including the cysteine biosynthesis was induced under the ethanol stress condition by hfq overexpression in Z. mobilis. Previous study confirmed that cysteine supplementation enhanced the tolerance in Z. mobilis to different inhibitors including ethanol (Yan et al., 2022). Therefore, hfq overexpression in ZM4-hfq might regulate the sulfate assimilation and cysteine biosynthesis for enhanced ethanol tolerance.

To evaluate the hypothesis that hfq overexpression enhanced sulfur metabolism and therefore played a protective role against the ethanol stress, 4 g/L Na₂SO₄ as the substrate for sulfate assimilation was supplemented into the medium and the cell growths of ZM4 and its hfq recombinant strains were detected (Figure 3). To avoid the effect of other sulfur-containing nutrients in RM medium, MM medium was used (Yan et al., 2022). Consistent with the result in rich medium (Figure 1), hfq overexpression strain ZM4-hfq exhibited an enhanced tolerance to 3% ethanol in MM media and achieved a final OD₆₀₀ of 1.51, clearly higher than the wild-type with OD₆₀₀ of 1.33. While, the hfq deletion strain ZM4-Δhfq had the lowest cell growth with the final OD₆₀₀ of 0.93 among three strains (Figure 3A). When Na₂SO₄ was added in the media, the cell growth of ZM4-hfq under ethanol treatment was further improved to a final OD₆₀₀ of 1.65, while no influence was observed for hfq deletion strain ZM4-Δhfq.

Moreover, an improvement responding to ethanol stress was detected as well in the wild-type ZM4 with the Na₂SO₄ supplementation, which suggested that the native gene expression of hfq in ZM4 would effectively regulate the sulfate assimilation process with Na₂SO₄ as the substrate. Since that MM medium is regarded as a stressful environment for Z. mobilis (Yang et al., 2014), Na₂SO₄ supplementation also slightly improved the cell growth of ZM4-hfq and ZM4 without ethanol treatment (Figure 3B). However, there is still no improvement observed for ZM4-Δhfq. Taken together, Na₂SO₄ supplementation results suggested that the hfq overexpression in ZM4-hfq positively regulated the sulfate assimilation process, and Na₂SO₄ could be assimilated to deal with the stressful condition.

Previous studies in E. coli, S. cerevisiae, and Z. mobilis demonstrated that the sulfate assimilation was induced in response to stresses (Miller et al., 2009; Kanna and Matsumura, 2015; Zhang Y. et al., 2019). Recent study in Z. mobilis 8b further confirmed that cysteine supplementation in the growth media boosted glutathione synthesis or H₂S release effectively leading to the reduced accumulation of reactive oxygen species (ROS) induced by inhibitor stress (Yan et al., 2022). To evaluate the impact of upregulated sulfur metabolism on ROS for ethanol tolerance improvement in ZM4-hfq, the cellular ROS levels in hfq recombinant strains under ethanol treatment were further examined.

As shown in Figure 4, the intracellular ROS accumulation was detected in the wild-type ZM4 with ethanol treatment (10.19%–15.53%, p-value <0.05). Corresponding to the cell growth performance responding to ethanol stress (Figure 3), the intracellular ROS level was decreased by hfq overexpression in ZM4-hfq compared with the wild-type control ZM4 (15.53%–6.42%, p-value <0.05), while increased by hfq deletion in ZM4-Δhfq (15.53%–26.07%, p-value <0.05). Similar alterations were detected in strain comparisons without ethanol presence (Figure 4). Collectively, the results suggested that hfq overexpression in Z. mobilis was involved in the positive regulation of sulfate assimilation and cysteine biosynthesis, and thus contributed to the effective elimination of the ROS induced by ethanol inhibition and finally enhanced the ethanol tolerance of Z. mobilis.

The major stress sigma factor RpoS of E. coli and Salmonella is the master regulator of oxidative stress response, and depends on Hfq for translation (Frohlich and Gottesman, 2018). Z. mobilis does not possess a rpoS gene, but its genome contains other five sigma factors genes. RpoE (sigma E) in Z. mobilis potentially is a RpoS-like regulatory factor, and plays a critical role in various stress conditions including ethanol stress responses (Seo et al., 2005). Recent study in the facultative phototrophic bacterium Rhodobacter sphaeroides demonstrated that sigma factor RpoE is involved in a specific photo-oxidative stress response, and sRNA RSs0019 controlled by the sigma factor RpoE contributes to the balance of sulfate uptake and biosynthesis of sulfur-containing amino acids to guarantee appropriate cysteine and glutathione levels under the given environmental conditions (Hess et al., 2014). These evidences suggested a potential role of rpoE in the regulation of sulfur metabolism against the oxidative stress in Z. mobilis.

In this study, the gene expression of rpoE (ZMO1404) was not influenced in both hfq recombinant strains. However, Hfq may only influence the RpoE activity without affecting its protein levels just as observed in R. sphaeroides (Berghoff et al., 2011). In addition, recent study in Z. mobilis identified two sRNAs, Zms4 and Zms6, that might affect a wide range of sRNAs through modulating hfq expression indirectly and specifically co-regulate some pathways important to ethanol stress, including sulfate assimilation and cysteine biosynthetic processes (Han et al., 2020). All these results support the role of Hfq on sulfur metabolism in response to ethanol stress. However, considering the versatile function of Hfq (Dos Santos et al., 2019), further studies are necessary to define the regulation of Hfq in Z. mobilis.

Z. mobilis can regulate general stress-response genes especially those in heat shock response to protect proteins from the damage caused by the stressful environments (Yang et al., 2014; Yi et al., 2015; Yang et al., 2020). Consistent with previous studies (He et al., 2012; Yang et al., 2013), many genes involved in the heat shock response were significantly upregulated in response to ethanol stress in ZM4, such as grpE (ZMO0016), groES (ZM O 1928), and groEL (ZM O 1929) (Supplementary Table S3). Further analysis discovered that more general stress response genes were induced by ethanol in ZM4-Δhfq, including 11 genes involved in protein remodeling and reactivation, six genes involved in DNA replication and repair, and three genes responsive for oxidative stress. This was in accordance with the cell growth performance that ZM4-Δhfq was more sensitive to ethanol stress (Figure 1). However, most of these stress response genes were significantly repressed by ethanol in strain ZM4-hfq.

Moreover, the gene expression in the strain comparisons showed that these stress response genes, especially genes encoding the heat shock protein (HSP) including grpE (ZMO0016), dnaK (ZMO0660), dnaJ (ZMO0661), groES (ZMO1928), and groEL (ZMO1929) were significantly downregulated in ZM4-hfq compared with the wild-type ZM4 or ZM4-Δhfq under 8% ethanol (Table 1; Supplementary Table S4). These results collectively suggested that the stress responses induced by ethanol might be effectively alleviated by hfq overexpression in ZM4-hfq. Therefore, the strain did not have to upregulate its energy-costly protein repair system to conserve energy for ethanol stress responses and cell growth.

In many bacteria, RpoH (Sigma 32) is associated with the stress responses especially the heat shock proteins (HSPs) (Krajewski et al., 2014; Roncarati and Scarlato, 2017). Corresponding to the upregulation of HSPs, the expression of rpoH (ZMO0749) was induced two folds in ZM4-Δhfq compared with ZM4 under ethanol treatment (Supplementary Table S4). Interestingly, recent study in Z. mobilis reported that rpoH overexpression led to a decrease in ethanol tolerance (Benoliel et al., 2020). Such evidence suggested that RpoH as well as HSPs at specific concentrations is required for Z. mobilis to respond to ethanol stress, and rpoH upregulation could make Z. mobilis sensitive to ethanol stress as observed in ZM4-Δhfq (Figure 1).

Different from the upregulation of rpoH in ZM4-Δhfq, the expression of rpoH was not affected in ZM4-hfq compared with ZM4 under ethanol treatment, even though the RpoH-transcribed heat shock regulons were downregulated (Supplementary Table S4). As a result, Hfq was speculated to be involved in the regulation of heat shock response via the alternative sigma factor RpoH. Previous studies demonstrated that RpoH activity is modulated via the DnaK/J-mediated negative feedback loop (Roncarati and Scarlato, 2017; Yura, 2019). In addition, Hfq was recently discovered to be capable to establish many protein-protein interactions in bacterial species (Yonekura et al., 2013; Dos Santos et al., 2019). Collectively, we speculated that hfq was associated with the indirect influence to RpoH activity via interaction with DnaK and/or DnaJ to help alleviate the cytoplasmic stress responses in Z. mobilis under ethanol stress.

Different methods have been developed to characterize the protein-protein interactions, such as co-immunoprecipitation (Co-IP), bacteria/yeast two-hybrid (B2H/Y2H), affinity-based techniques, and split protein complementation assays (Miura, 2018; Pichlerova and Hanes, 2021). Recently, Yi et al. (2013) developed a TEV protease mediated Yeast Endoplasmic Reticulum Sequestration Screening (YESS) system which can be easily applied in eukaryotic and prokaryotic species to explore the protein-protein interactions. In this system, the intracellular protein-protein interaction is converted to quantitative fluorescence signals through a split TEV protease-mediated proteolytic reaction, which possesses many advantages such as signal amplification, clean background and high efficiency. To illustrate the hypothesis above, the YESS system was established in Z. mobilis (Supplementary Figure S1) and then applied to detect the interaction between Hfq and DnaK/J.

As shown in Figure 5 and Supplementary Figure S2, the protein-protein interaction in YESS system was quantized by the D-value with flow cytometry. If the D-value >0, the protein pair has interaction, otherwise without interaction. Firstly, several known protein pairs of negative (PurF-RpiB, PurF-HisZ) and positive (HisG-HisZ, Prs-HisZ, Prs-RpiB, Prs-PyrE) ones were selected based on the UniProt and String database, and further evaluated. As expected, all the negative ones (PurF-RpiB, PurF-HisZ) displayed D-value <0, while the positive ones (HisG-HisZ, Prs-HisZ, Prs-RpiB, Prs-PyrE) showed D-value >0 (Figure 5 and Supplementary Figure S2). The results confirmed the YESS system was successfully established in Z. mobilis.

The protein pairs of Hfq-DnaK and Hfq-DnaJ were then characterized, which were linked with N-TEV or C-TEV respectively to avoid the structure interference in the protein interaction. Our results showed that the Hfq protein had direct interactions with DnaK and DnaJ in Z. mobilis, even though the interaction was not very strong with D-value from 8.45 to 22.98. The data indicated that Hfq was associated with the modulation of RpoH by direct interaction with DnaK/J. Taken together, the regulation of the RpoH-mediated cytoplasmic stress responses by Hfq contributed to the ethanol tolerance in Z. mobilis.