The wild-type strain Z. mobilis CP4 was purchased from the China Center for Type Culture Collection (CCTTCC; Accession NO.CICC10232). F211 is a strain of Z. mobilis with high furfural tolerance obtained in the early stage of our laboratory (Huang et al., 2018). Zymomonas mobilis was cultured in Rich Medium (10g/L yeast extract, 2g/L KH₂PO₄, 20g/L glucose, and different concentration of furfural as required) at 30°C without shaking.

Nine deletion strains (Supplementary Table S1) were constructed by homologous recombination as described in our previous study with some modification (Wang et al., 2013; Huang et al., 2018). The left and right homologous arm DNA fragments of the target gene were ligated to flank both sides of a chloramphenicol resistance gene, generating the homologous recombinant fragments (Section 1.1 of Supplementary Materials and Methods). The homologous recombinant fragments were ligated into the SphI and SacII sites of pUC19 to yield the plasmid pT-gene (Supplementary Table S1) by using the Gibson assembly (GA; Gibson et al., 2009). The target plasmids were then transformed into CP4 competent cells by electroporation (Ma et al., 2012). The target gene was replaced with the chloramphenicol resistance gene by homologous recombination using its native RecA recombinase (Supplementary Figure S1; Savvides et al., 2000; Huang et al., 2018). Transformants were screened on RM agar plates with 50μg/ml chloramphenicol. Positive single colonies were confirmed by PCR using the appropriate geneGF/geneGR, geneGF/T-CM-R, and T-CM-F/geneGR (Supplementary Table S2) primer pairs (Supplementary Materials and Methods).

Overexpression vectors were constructed by ligating the target wild-type gene or its mutated gene into multiple clone sites of pEZ15Asp (pE)/pHW20a (pH) vector. The target genes were amplified from the genomes of CP4 and F211 using the primers gene OF/OR (Supplementary Table S3), and the P _gap promoter (Yang et al., 2019) was amplified from the genome of CP4 using the primers pgap-F/pgap-R (Supplementary Table S3). Then, the two fragments were fused by overlap-extension PCR using the primers pgap-F/gene-OR. The fusion fragment was inserted into the PstI and EcoRI sites of pE/pH to yield the plasmid pE-gene/pH-gene by using the aforementioned GA (Gibson et al., 2009). Positive clones were further verified by DNA sequencing. The correctly sequenced recombinant plasmids were transformed into CP4 competent cells by electroporation. The target plasmids were confirmed by PCR using the appropriate pEZ-F/pEZ-R (Supplementary Table S3) primer pairs.

Strain was precultured in RM at 30°C to the late exponential stage, and the cultures were concentrated to an OD₆₀₀ of 20 by centrifugation. The concentrated cells were inoculated into 100ml RM medium containing 3.5g/L furfural with an initial OD₆₀₀ of 0.2, and the OD₆₀₀ value was measured every 6h to detect the growth of the cells for the evaluation of furfural tolerance. The degradation rate of furfural was calculated by the following formula:

where “T” indicated the fermentation time, and “C_initial” and “C_final” indicated “initial medium furfural concentration” and “final medium furfural concentration,” respectively. All experiments were performed in triplicates, and statistical differences were analyzed using a one-way analysis of variance (ANOVA). *p<0.05 was considered statistically significant, and **p<0.01 was considered extremely significant.

Strains were cultured in RM medium containing 3g/L furfural for 36h, and 10μl of the culture was taken to observe the flocculation under a light microscope (Olympus BX51, Japan). Five milliliter culture of each strain was collected and was shaken vigorously for 15s and then rested for 5min. Suspension of 1ml was sampled from the upper section to measure OD₆₀₀ and quantify the non-flocculating cells (Cf), and 1.2units of cellulase (Novozymes, Denmark) was added to the remaining culture, which was incubated at 50°C for complete de-flocculation of the bacterial flocs to measure the concentration of total cells (Ct). The flocculation efficiency of Z. mobilis was calculated based on the formula:

R%=(1−CfCt)×100 (Xia et al., 2018).

Strain F211 and CP4 were cultured statically in RM medium with 3g/L furfural. The cultures were collected by centrifugation when the OD₆₀₀ reached 0.8–1.4, and the cells were washed with PBS buffer (pH=7.4, 137mM NaCl, 2.7mM KCl, 10mM Na₂HPO₄·12H₂O, and 1.4mM KH₂PO₄). The cells were collected by centrifugation and freeze in liquid nitrogen for 1h. The samples were submitted to Novogene (China, Beijing) for RNA-Seq.

The integrity of the extracted RNA was tested using the Agilent Bioanalyzer 2,100 system (Agilent Technologies, CA, United States). cDNA library preparation and sequencing were conducted by the Novogene Technology Company in Beijing, China. The clean reads were obtained by sequencing and mapped to the Z. mobilis CP4 genome sequence (GenBank: CP006818.1) using Bowtie2-2.2.3. Differentially expressed genes (DEGs) between the two samples were identified using the DESeq package (1.20.0), and |log2(FoldChange)|>1 and q-value<0.005 were used as the threshold to identify significant differences in gene expression analysis of Gene Ontology (GO), which assigns genes into functional categories, was performed using the GO R packages. KOBAS software could perform KEGG enrichment analysis on differentially expressed genes to determine the metabolic pathways of genes.

The expression levels of 22 DEGs representing different functional categories were assessed by real-time quantitative reverse transcription PCR (qRT-PCR) to validate the reliability of the RNA-Seq data. The culture conditions of CP4 and F211 were the same as the RNA-Seq experiment. When the strains concentration reached OD₆₀₀=0.8–1.4, about 10⁸ cells were collected from each strain, and the cells are treated in liquid nitrogen for 5min. Subsequently, the bacterial cells were ground in a mortar pre-cooled with liquid nitrogen, and the powder was transferred to 100μl TE buffer (containing 400μg/ml lysozyme), mixed, and placed at room temperature for 4min to dissolve the cell wall. RNA was extracted with TRIzol reagent (Invitrogen, Carlsbad, CA, United States) according to the manufacturer’s instructions.

DNase (RQ1 RNase-Free DNase, Promega, United States) was added to the RNA samples to remove genomic DNA. The PCR reaction with primer pair 16SrRNA-F/16SrRNA-R was performed on the total RNA in order to check for genomic DNA contamination. No product was observed, demonstrating that the sample had no genomic DNA contaminants and was suitable for qRT-PCR assay. Subsequently, cDNA was synthesized using iScript cDNA Synthesis Kit (Bio-Rad) with the purified total RNA of CP4 or F211 as a template following the manufacturer’s instructions. Real-time qRT-PCR experiments were performed using CFX96 Real-Time System (Bio-Rad). Each reaction contained 2μl of diluted (1/10) cDNA, Taq SYBR Green qPCR Premix (Yugong Biolabs Inc., Jiangsu, China) and the corresponding primer pairs (Supplementary Table S4). The 16S rRNA gene was used as a reference to normalize the qRT-PCR data. The expression level of each gene was calculated according to the 2(-Delta Delta C(T)) method (Livak and Schmittgen, 2001).

Eleven SNPs (single nucleotide polymorphisms) were found in high furfural-tolerant F211 compared with the sequence (GenBank No. CP006818) of CP4 from Agricultural Research Service Culture Collection (NRRL No. B-14023; Huang et al., 2018). The initial strain CP4 (CICC 10132) of F211 and the reference CP4 may have different SNPs. Therefore, in this study, except for the three synonymous mutated genes of ZCP4_0096, ZCP4_0593, and ZCP4_1887, seven missense mutated genes and one nonsense mutated gene of F211, along with the starting strain, CP4, were PCR-amplified and sequenced. F211 and CP4 shared consistent sequences of ZCP4_1431, ZCP4_1,616, ZCP4_0244, and ZCP4_0525. The ZCP4_1919, ZCP4_1419, and ZCP4_0270 genes in F211 carried missense mutations in G335A, C85T, and C493T, resulting in a substitution of V119I, G29S, and G165S, respectively. The mutation of ZCP4_0970 (1075G>A) converted a glutamine codon to a stop codon (Q359X), producing a truncated protein lacking 153 amino acids at the C-terminal end (Table 1). Therefore, these four genes were selected for functional studies related to furfural tolerance.

The deletion and overexpression strains of these genes were constructed to study furfural tolerance. In RM medium containing 3.5g/L furfural, the deletion and overexpression strains of ZCP4_1419 and ZCP4_1919 genes did not show obvious growth differences (Supplementary Figures S3A,B), indicating that these two genes have no effect on furfural tolerance. The ZCP4_0970 gene encodes the TolC family protein of the bacterial type I secretion system, transporting proteins to the extracellular space coupled with ATP consumption (Delepelaire, 2004). The deletion strain Δ0970 (pE) showed a 13 and 7.6% increase in the maximum cell density and ethanol production, respectively, compared to the control CP4 (pE) (Figures 1A,B). In contrast, the overexpression strain CP4 (pE-W0970) had a 47 and 13.9% decrease in the maximum cell density and ethanol production, respectively, compared to the control CP4 (pE) (Figures 1A,B). The results showed that the downregulation of ZCP4_0970 can increase the furfural tolerance of strains, which indicates that the G1075A mutation in the ZCP4_0970 gene resulted in functional inactivation or attenuation, improving furfural tolerance.

Furfural destroys the tertiary structure of proteins and inhibits protein synthesis (He et al., 2012; Todhanakasem et al., 2014). Tolc inactivation leads to defects in the secretion system and the increase in ATP levels (Delepelaire, 2004), thereby providing more energy for cell growth and increasing furfural tolerance. Interesting, the deletion of the ZCP4_0970 gene destroys the efflux system, also yielding a reducing tetracycline resistance (Delepelaire, 2004; De Cristóbal et al., 2006), which causes the inability of transformants to grow on the tetracycline selection plates. And the original initial pHW20a plasmid containing tetracycline resistance gene (Dong et al., 2011) could not be transformed successfully into Δ0970 strain. So, the plasmid pEZ15Asp with an ampicillin resistance gene (Yang et al., 2016) was used as a starting vector for construction of wild-type and mutated ZCP4_0970 gene overexpression vectors.

ZCP4_0270 encodes a bifunctional protein with diguanylate cyclase (DGC)/phosphodiesterase (PDE), which synthesizes/degrades bis-(39–59)-cyclic dimeric guanosine monophosphate (c-di-GMP). The deletion strain Δ0270 (pH) and overexpression strain Δ0270 (pH-W0270) did not show significant growth differences compared to those of the control strains CP4 (pH) and Δ0270 (pE-W0270) (Figures 1C,D), indicating that deletion or overexpression of the native ZCP4_0270 gene did not affect furfural tolerance. The specific growth rate, degradation rate of furfural in the medium (0–48h), and ethanol production of the mutated gene overexpression strain Δ0270 (pH-M0270) were 62, 8.3, and 38.5% higher than those of the control strain Δ0270 (pE-M0270) (Figures 1C,D), respectively, indicating that the C493T mutation of the ZCP4_0270 gene can promote the degradation of furfural, thereby improving furfural tolerance.

Flocculated yeast showed a higher tolerance to furfural than free yeast (Westman, 2014). The increased c-di-GMP, which is regulated by PDEs, can promote the flocculation of Z. mobilis (Tan et al., 2014). Therefore, the mutated ZCP4_0270 may improve furfural tolerance by promoting flocculation, and thus, the flocculation of F211 and Δ0270 (pH-M0270) was investigated. Flocculation was observed in both strains (Figures 2A,C), and the flocculation rates of F211 and Δ0270 (pH-M0270) were 600 and 470% higher than that of CP4, respectively (Figure 2B). The flocculation rate of the wild-type ZCP4_0270 gene overexpressing strain Δ0270 (pH-W0270) and the engineered strains of ZCP4_0970 and ZCP4_1419 were similar to those of the control CP4 (Figure 2B). These results show that the mutated ZCP4_0270 gene in F211 causes flocculation, resulting in an improved furfural tolerance.

The bifunctional DGCs/PDEs encoded by the ZCP4_0270 gene can regulate the levels of the second messenger c-di-GMP, which coordinates diverse aspects of bacterial growth and behavior, including motility, virulence, biofilm formation, and cell cycle progression (Jenal et al., 2017). Therefore, enhanced tolerance by the mutated ZCP4_0270 also resulted in a change in the physiological, biochemical, and expression levels of some genes. Consequently, RNA-Seq was performed to analyze the transcriptome differences between F211 and CP4 under furfural stress and to explore the potential furfural tolerance mechanism.

A total of 139 DEGs were identified in F211 compared with the control, with 77 and 62 genes being upregulated and downregulated, respectively. The expression levels of the three missense genes in F211 did not change significantly, indicating that its enhanced tolerance was due to the changed protein function caused by the mutated ZCP4_0270 and ZCP4_0970 genes, rather than the changed transcript abundance. DEGs were subjected to GO enrichment analysis. The five GO clusters with the highest enrichment were as follows: “localization,” “transport,” “establishment of localization,” “cofactor binding,” and “response to stimulus” (Supplementary Figure S4A). The DEGs were enriched in 39 pathways, found using a Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis, 27 of which were different from those previously reported (He et al., 2012; Yang et al., 2020). The significant KEGG terms were primarily associated with microbial metabolism in diverse environments, biosynthesis of secondary metabolites and antibiotics, bacterial chemotaxis, and sulfur metabolism (Supplementary Figure S4B).

We screened 22 DEGs representing different GO clusters from 139 DEGs (Table 2) for qRT-PCR verification. Based on the qRT-PCR results, the expression trends of these candidate genes were consistent with the RNA-Seq data (Figure 3). A comparison of the two methods indicated a high level of concordance (R²=0.7833), indicating that the RNA-Seq data were reliable.