Molecular Mechanism of Engineered Zymomonas Mobilis Response to Furfural and Acetic acid Stress

Backgroud: Acetic acid and furfural are two major inhibitors during lignocellulosic ethanol production. In our previous study, we successfully constructed an engineered Zymomonas mobilis ZM532 strain tolerant these double inhibitors by genome shuing, but the molecular mechanisms of tolerance to these inhibitors are still unknown. This study investigated the responses of ZM532 and wild-type ZM4 to acetic acid and furfural using genomics, transcriptomics and label free quantitative proteomics. Results: By Sanger sequencing technology we re-veried of previously identied 19 mutations in ZM532, but we found a total of 23 single nucleotide polymorphisms (SNPs) in the coding sequence (CDS; 4) and intergenic region (19) in ZM532. Six SNPs were novel in this study. We also identied a total of 1865 and 14 novel differentially expressed genes (DEGs) in ZM532 and wild-type ZM4. Among these, 352 DEGs were up-regulated; while and 393 were down-regulated in AF_ZM532 vs RM_532, respectively. However, 442 DEGs were up while 463 were down-regulated in AF_ZM4 vs RM_ZM4. Moreover, 2 up and 8 down-regulated genes were identied in AF_ZM532 vs AF_ZM4; while 7 up and 1 down-regulated genes were found in RM_ZM532 vs RM_ZM. We also identied 1,532 proteins among 107 up and 204 down-regulated proteins detected in ZM4_AF vs ZM4_RM, 123 up and 205 down regulated proteins were identied in ZM532_AF vs ZM532_RM, respectively. In addition, a total of 16 up and 5 down-regulated proteins were identied out of 1462 in ZM4_AF vs ZM532_AF, while 8 up and 5 down-regulated proteins were observed out of 1491 in ZM4_RM vs ZM532_RM. These proteins and genes are involved in amino acid biosynthesis, macromolecules repair, glycolysis, agella assembly,

. But we haven't explored the global transcriptional pro le difference between ZM532 mutant and wild-type ZM4 especially in the conditions of rich media (RM) and media containg acetic acid and furfural (AF) which is potentially important at industrail level. So, in this study, sanger sequencing rstly used to veri ed the mutations in ZM532, and then we applied transcriptomics and proteomics to unravel molecular mechanisms in the wild type (ZM4) and mutant strain (ZM532) under AF and RM conditions. By this we identi ed large number of genes and proteins which are resposible for tolerance against acids and they were also used to hypothesized for media, strain, treatment and growth conditions responsive. Further, we knocked-out and overexpressed two differentially expressed genes (DEGs), ZMO_RS02740 and ZMO_RS06525, to study their modulating roles in acetic acid and furfural (AF) condition. The results provided here deepen our understanding of ethanol production by genetic engineering and/or synthetic biology.
Results And Discussion ZM532 re-sequencing for con rmation of previously identi ed mutations The mutant, ZM532 was constructed by genome shu ing. The ZM532 exhibited superior performance with a shortened fermentation time (30 hours) and higher ethanol production (0.463 g/L/h) compared to the parental strain AQ8-1 in 7.0 g/L acetic acid medium [10]. We identi ed 23 single nucleotide polymorphisms (SNPs) in the coding sequence (4CDS and 19 within the intergenic region) via Sanger sequencing. Six SNPs were novel in this study (Table 1). Wang et al. [10] identi ed 19 identical SNPs in wild type ZM4 located in the CDS (6) and 13 in the intergenic regions, respectively (Table 1).  [10]. c Current study with ZM532 strain. d +/-indicate the presence/absence of variation in the genome, respectively.
The four SNPs in the CDS regions caused variation in the amino acid (AA), resulting in synonymous and non-synonymous mutations (Table 1). In ZMO_RS03765 (arginine-tRNA ligase) one non-synonymous and synonymous AA change was observed at the same time. In contrast, in ZMO_RS00235 and ZMO_RS02620, two non-synonymous AA changes were observed, which were linked to glutaminefructose-6-phosphate aminotransferase and DNA repair protein RadA, respectively (Table 1). While the gene ZMO_RS02620 encodes a DNA repair protein called RadA, which is necessary for cellular survival when cells are exposed to acid stress [29]. Jeong et al. [30] reported that under acid stress in E. coli O157: H7 strand disintegrates and DNA integrity was retained by Dps and RecA-mediated repairs, indicating that DNA repair can play an important role in acid tolerance. Wang et al. [10] identi ed two non-synonymous AA changes in ZMO_RS09165 (IS5/IS1128 family transposase) but these were absent in the present study. Conversely, we found SNPs in the intergenic regions of the pairs of genes: ZMO_RS09160 and ZMO_RS09165; ZMO_RS04290 and ZMO_RS04295; and ZMO_RS07065 and ZMO_RS07070 in ZM532 mutant. Conversely, ZMO_RS04290 and ZMO_RS04295 which encode monofunctional encoding MBPT and cytochrome c to promote glycan chain synthesis in bacterial cell walls and its function is identical to that of DNA polymerases [31]. This could be important to preserve the integrity and tolerance of the cells to the inhibitor. This is consistent with report by Wang et al. [10]. There was also a frameshift mutation in ZMO_RS04405, which codes for ABC transporter substrate-binding protein because of two single nucleotides deletion in the CDS region (Table S1). We also detected − 21 bp deletion in the ZMO_RS05590 (hypothetical protein) while the previous studies identi ed two distinct deletion (-21 bp and − 28 bp) in different locations of the same gene. Additionally, there was a -7 bp deletion in ZMO_RS07255 (carbamoyl phosphate synthase large subunit) and − 1 bp InDel in the intergenic region of ZMO_RS06410-ZMO_RS06415 (Additional le 5: Table S5). ADI genes, as previously reported by Ryan et al. [32], allow Listeria monocytogenes to survive under acidic conditions, with arginine their expression is become higher at low pH. In short, genes involved in the same mutations on the parental and mutant strains have been reported previously and may be play critical role against acids stressors.

Overview of Transcriptome under inhibitors (AF)
The RNA-seq yielded a total of 32.79 Gb clean data, averaging 2.73 Gb for each sample with 91 % of bases recording Q30 and above, with a q ≥ 20 (an error probability of 0.03 %) (Additional File 6: Table  S6). The GC-contents in the four distinct groups ranged from 48.26-49.15. A gene was considered differentially expressed (DEG) after comparing the gene expression pro les between RM and AF treatments with fold change (FC) > 1.5 and FDR corrected p < 0.05. A total of 1865 and 14 novel DEGs were identi ed by using the genome of parental strain ZM4 (ATCC 31821) as the Ref. [33].
Differentially expressed genes in ZM532 and ZM4 and their expression pro les In total, we identi ed 745 and 905 DEGs in ZM532 and ZM4, respectively ( Fig. 1A, B, Additional File 7: Table S7, Additional File 8: Table S8). Among these, 352 DEGs were up-regulated; while and 393 were down-regulated in AF_ZM532 vs RM_532, respectively. However, 442 DEGs were up-regulated while 463 were down-regulated in AF_ZM4 vs RM_ZM4 (Fig. 1A-B). In addition, only 2 up and 8 down-regulated genes in AF_ZM532 vs AF_ZM4; while 7 up and 1 down-regulated genes in RM_ZM532 vs RM_ZM ( Fig. 1C-D). The higher number of DEGs detected in the strain ZM4 suggests intense transcriptional alteration in response to the inhibitors due to ZM4 relative sensitivity. We performed hierarchical cluster analysis based on the log 2 FC and FPKM values to validate the DEGs from ZM4 and ZM532 strains ( Figure S1). The analysis clustered the DEGs into two main groups with the two strains clustering together regardless of the treatments. This implies that very few DEGs were distinguishable between these two strains in their response to AF treatments.
Similarly, in AF_ZM4 vs RM_ZM4, the genes were mostly involved in biological process such as translation (GO: 0006412, 67 DEGs), peptide biosynthetic process (GO: 0043043, 67 DEGs) and peptide metabolic process (GO: 0006518, 69 DEGs) ( Fig. 2A). The molecular functions recorded "structural molecule activity (GO: 0005198" in both AF_ZM532 vs RM_532 and AF_ZM4 vs RM_ZM4 as the most enriched terms. In the AF_ZM4 vs RM_ZM4 strain, 14 GO terms were signi cantly enriched in the "biological process" and "molecular function" categories ( Fig. 2B). Interestingly, ribosomes biogenesis (GO: 0042254) and translation (GO: 0006412) which are associated with protein synthesis were the most signi cantly enriched GO terms. These results indicate synthesis of protein in bacteria are markedly altered with modi cation of culture conditions. Interestingly, we found DEGs involved in high oxidoreductase activity were up-regulated in AF_ZM532 vs RM_532 but down-regulated in AF_ZM4 vs RM_ZM4 ( Fig. 2A Subsequently, we searched for candidate genes involved in the tolerance mechanism against the inhibitors by comparing the DEGs between the two strains. We detected 98 DEGs were exclusively involved in the AF resistance in ZM532, including 42 up-regulated DEGs in response to the inhibitors and associated with oxido-reductase activity ( Figure S2A). Additional 647 DEGs were mutually detected in both samples. While in AF_ZM532 vs AF_ZM4 (purple) and RM_ZM532 vs RM_ZM4, only 1 DEGs was codetected in both strain. We identi ed 7 DEGs to RM_ZM532 vs RM_ZM4 and 9 to AF_ZM532 vs AF_ZM4 ( Figure S2B). The most up-and down-regulated DEGs were ZMO_RS02740 (log 2 FC = 6.05) and ZMO_RS06525 (log 2 FC = -2.373) in the ZM532 strain. These candidate DEGs represent important resources for further functional validation in AF resistance in Z. mobilis.
Label free data and functional annotation of the proteins To elucidate the molecular response, tolerance to acetic acid and furfural inhibitors, the proteomes of the two Z. mobilis strains (ZM4 and its mutant ZM532) were generated under control and treatment conditions. We successfully identi ed a total of 1532 proteins in both samples ( Table 2). The mutant strain ZM532 was more resistant to the inhibitors as compared to ZM4 ( Table 2). The number of proteins detected in the AF samples were lower than in the RM samples ( Figure S3A), suggesting that proteins synthesis in Z. mobilis was inhibited by the treatment. Mass distribution and protein length of most of the identi ed proteins were between 10-60 kDa and 6-20 amino acids, respectively ( Figure S3B, C). There was less variability in the majority of the detected proteins (CV < 0.2) ( Figure S3D). The proteins were functionally annotated by BLASTP (E value ≤ 1e-4) using the COG, GO, KEGG and IPR databases (Table  S10). We successfully annotated 99 % of the total proteins in at least one database (Addition File10: Table S10: Fig. 3A) and 976 proteins (76 %) were annotated in all the four databases. The function of protein is usually associated with its subcellular localization, the capability to predict subcellular localization directly from protein sequences is bene cial to inferred protein functions. The statistical analysis of the proportion of subcellular location (Cell-mPLOC 2.0 website) of the differential protein is shown in Figure S4. The identi ed proteins demonstrated that 52.17% proteins were found in resemblance with cytoplasmic proteins; 32.61% proteins signi cant hits for cell inner membrane proteins, followed by periplasm proteins (5.34%), extracellular proteins (2.57%), agellum proteins (0.59%) and 0.20% proteins with nucleoid protein. Furthermore, PCA was performed based on the protein expression pro le (Addition File10: Table S10). More than half of the total variations were accounted for by the rst two PCs, indicating that AF treatment signi cantly altered the proteome in both strains (Fig. 3B). Differentially expressed proteins in response to the inhibitors (AF) The protein expression data was compared between control and treatment groups to detect the differentially expressed proteins (DEP) based on the fold change > 1.5 and p < 0.05. A total of 107 up and /204 down-regulated proteins out of 1477 were detected in ZM4_AF vs ZM4_RM; while and 123 upregulated and 205 down-regulated proteins of 1474 were identi ed in ZM532_AF vs ZM532_RM, respectively (Addition File11: Table S11; Addition File12: Table S12: Fig. 3C-D). In addition, a total of 16 up and 5 down-regulated proteins were identi ed out of 1462 in ZM4_AF vs ZM532_AF, while 8 up and 5 down-regulated proteins were observed out of 1491 in ZM4_RM vs ZM532_RM ( Fig. 3E-F). Comparative analysis of the DEPs between ZM4 and ZM532 revealed 186 DEPs shared the same pattern of regulation (Addition File13: Table S13: Figure S5). This suggests they represent the core proteome response to the inhibitors, regardless of the tolerance level of the Z. mobilis strains.
Pathway enrichment analysis of the differentially expressed proteins GO and KEGG enrichment analyses were performed to understand the biological pathways activated in response to the inhibitors ( Fig. 4A-B). For GO enrichment analysis, we mainly focused on molecular function and cellular component classes. In the two strains, DEPs related to 'non-membrane-bounded organelle', 'large ribosomal subunit', and 'ribosome' were the enriched terms in the GO cellular component class, indicating major alterations in the ribosome were induced by the inhibitors. These proteins mainly contributed to 'structural molecule activity', 'electron carrier activity' and 'structural constituent of ribosome' (molecular function terms) which denotes that the inhibitors affect the structural integrity and normal ribosome activity. While in ZM4_AF vs ZM532_AF, DEPs maily related to hydrolase activity, edonuclease activity and damage DNA binding were more enrished molecular functions terms. Moreover, hydrolases activity are important against inhibitors, which involved in several critical functions such as maturation, turnover, recycling, and autolysis. In addition, DNA damage binding activity has a direct relationship to decreased nucleotide excision repair and Endonucleases play a role in DNA repair in resistance strain ZM532 (Fig. 4C). But in case of ZM4_RM vs ZM532_RM, most of DEPs linked with plasma membrane at cellular level, while at molecular level DEPs were invloved in N-acetyletransferase activity (Fig. 4D).
A similar result obtained concerning the KEGG enrichment analysis in the two strains, highlighting 'ribosome' as the most affected pathway (Fig. 5A-B). In addition, agellar assembly, peptidoglycan biosynthesis and ribosome were enriched KEEG pathways in ZM4_AF vs ZM532_AF while folate biosynthesis, amino sugar and nucleotide sugar metabolism and ABC transporter were most enriched term in ZM4_RM vs ZM532_RM (Fig. 5C-D).
The identi ed DE proteins were mapped to the reference pathways in the KEGG database, and 21 different biological pathways were obtained in 4 major categories (Fig. 5E). The KEGG pathways (which include 914 proteins) were members of a major group, metabolism, 153 linked to genetic information processing, 59 related to cellular processes, and 87 related to environmental information ( Figure S6). KEGG enrichment analysis in the two strains also a rms 'ribosome' as the most affected pathway in both strains. Analysis of expression fold changes of the proteins involved in selected pathways revealed that the resistant strain (ZM532) strongly delayed the activity in the ribosome by reducing the synthesis of all ribosome-assembly proteins under AF treatment (Table 3), a mechanism known as hypometabolism [34]. In contrast, several ribosome-assembly proteins were either up-regulated or unaffected following AF treatment in ZM4 (Table 3). We speculate that the ability to limit ribosome activity is an effective adaptation mechanism against AF.

Genome, Gene and Protein Correlations
Integrative molecular approaches such as genome, transcriptome and proteome may help us to gain a deep understanding of toxicants' effects at multiple levels of biological organization, while also facilitating in risk assessment. We identi ed two key mutations ZMO_RS00235 and ZMO_RS03765 in our transcriptome data, which encode glutamine-fructose-6-phosphate aminotransferase and arginine-tRNA ligase ( Fig. 6A) as described earlier and our current sanger re-sequencing data [10]. The gene, ZMO_RS00235 was up-regulated in the mutant strain which may contribute to furfural and acetic acid stress tolerance. ZMO_RS00235 was previously reported to be critical for organic acid stress in cells [10,29]. Moreover, ZMO_RS03765 is associated with arginine biosynthesis which is crucial for acid stress. However, no concrete evidence has been adduced for the role of arginine in acid resistance, the cell wall/membrane itself may be important to maintain cell integrity.
In addition, we also identi ed two important mutations such as ZMO_RS06410 and ZMO_RS04295 in the proteome data (Additional le 14: Table S14) as reported previously [10] and current re-sequencing data. ZMO_RS06410 might improve fusidic acid resistance and methicillin-resistance, it may also be useful for Z. mobilis to survive acid stress [35,36]. One of the genes, ZMO RS04295 encoding MBPT and cytochrome c to promote glycan chain synthesis in bacterial cell walls and its function is identical to that of DNA polymerases [31]. This could be important to preserve the integrity and tolerance of the cells to the inhibitors. These mutations have roles in acids tolerance as cytochrome C may provide some protective layer sheet against acetic acid and furfural stresses.
To identify some corresponding relationships, the transcriptome data was combined with the proteome data. A total of 662, 578, 1379 IDs were identi ed in ZM532_AF vs ZM532_RM, AF_532 vs RM_532, ZM4_AF vs ZM4_RM, AF_ZM4 vs RM_ZM4 and ZM532_AF vs ZM4_AF, AF_532 vs AF_ZM4 by both RNAseq and proteomics ( Figure S7A-C). In the three groups, 111, 138 and 1 unique proteins related to transcriptome DEGs, were identi ed, respectively. Correlation analysis was performed between the multiples of genes (proteins) reported by transcriptome and proteome study in the three groups ( Figure   S7A-C). Among mRNA and the corresponding protein, the Pearson correlation coe cient was positive (Pearson = 0.233, 0.217 and 0.014) for all groups, respectively ( Figure S7D-F). As a result, we suggest that it is critical to assess protein expression in order to understand phenotypic changes and not relying solely on transcriptional level.
Candidate genes and proteins involved in motility of cell, membrane components and envelope biogenesis contributed in stability of cell membrane Several DEPs were associated with membrane, cell wall, cell motility and envelope biogenesis in the presence of acetic and furfural acids. Proteins associated with membrane and membrane components were up-regulated with enhanced tolerance to acidic acid and furfural. In bacteria, presence of TonBdependent transporters are necessary for active absorption of complexes such as carbohydrates and iron [37,38] as iron assimilation is crucial in some bacteria. TonB-dependent transport proteins may contribute to resistance to osmotic pressure [39]. Here, our proteomics results revealed that TonB transporter proteins; ZMO_RS07620, ZMO_RS06825, ZMO_RS07005 and ZMO_RS06600 were downregulated in the resistant strain ZM532. Contrary, a down-regulated gene ZMO_RS00805 encoded TonBdependent siderophore receptor in ZM4 (Additional le 14: Table S14). Nonetheless, our transcriptomics results revealed TonB-dependent transporter genes (ZMO_RS00540, ZMO_RS04385, ZMO_RS07620 and ZMO_RS07730) were down-regulated in the mutants, ZM532 and ZM4 (Additional le 7, 8: Table S7, S8). The down-regulation of TonB-dependent transporter proteins are consistent with results by [40]. ZM532 (mutant strain) recorded more down-regulated proteins compared to ZM4. This may be ascribed to the down-regulation of the TonB-dependent transporters resulting in decreased energy for the absorption of substrates and reserve energy for stressors.
Acetic acid substrates are transported by ABC transporters as acetic acid tolerance mechanism [40]. The transcriptomics results demonstrated that genes encoding ABC transporters such as ZMO_RS08315, ZMO_RS04690, ZMO_RS01550, ZMO_RS02545 and ZMO_RS01130 were signi cantly up-regulated in ZM532 (Additional le 7, 8: Table S7, S8). The up-regulation of ABC transporter genes were consistent with results from acidic-pH treatment [41,42]. The up-regulation of ABC transporters may enhance and sustain the cytoplasmic pH homeostasis in mutant strain ZM532 under AF conditions. Additionally, RND e ux pump is essential for supply of different metabolites with the proton antiport (including antibiotics, and basic organic solvents). This is effective for detoxifying external toxic compounds and internal harmful intermediates [43]. RND e ux pump was encoded by ZMO_RS03470 in the proteome while ZMO_RS03475, and ZMO_RS06835 were signi cantly down-regulated under acidic condition in both ZM4 and ZM532 (Additional le 7, 8: Table S7, S8; Additional le 14: Table S14). This is in consonance with report by Wu et al. (2012) in furfural. Might be down-regulated RND e ux pump encoded genes contributed essestial role against acids resistance. Hence, the effect of acids on above genes involved in RND e ux pump system is poorly studied needs to be studied their function in detail. Chemotaxis is a bacterial mechanism in response to chemical stimuli. A chemotaxis protein CheA, ZMO_RS00345 was down-regulated in ZM4 while CheB ZMO_RS03930 was up-regulated in ZM532 under acidic conditions.
Flagellum-related proteins (ZMO_RS02810 ( iF), ZMO_RS02680 (motA), ZMO_RS02770 ( hA), ZMO_RS02705), cell division protein (ZMO_RS01795) protein were down-regulated in both ZM4 and ZM532 in our proteome data. Interestingly, ZMO_RS02795 ( agellin), ZMO_RS02890 ( agellar protein FliS), ZMO_RS02760 ( gM), ZMO_RS02720 ( agellar hook-basal body complex protein), ZMO_RS02685 (Flagellin) were up-regulated in ZM532 (Additional le 14: Table S14). Up-regulation of these proteins may be tolerance mechanism to acetic acid and furfural. It should be noted, however, that the effect of acids on the above mentioned genes and proteins are raely discussed. In our transcriptome we found a number of genes (ZMO_RS02675, ZMO_RS02680, ZMO_RS00870, ZMO_RS02690, ZMO_RS02825, ZMO_RS02700) encoding proteins associated with chemotaxis and agellar structure were signi cantly down-regulated under the acidic condition in both strains (Additional le 7, 8: Table S7, S8). Our results showed that many genes associated with agellar assembly and chemotaxis were reserved. Limiting agellum biosynthesis is an energy usage technique to reserve the comparatively expensive energy reserve in stress conditions. Moreover, proteins associated with chemotaxis and agellum were downregulated in E. coli under low-pH conditions and high-osmolality decreasing the agellum biosynthesis resulting in limited proton penetration [42].
Pentacyclic triterpenoid lipids are class of hopanoids responsible for regulating and maintaining membrane stability, uidity, pH, homeostasis and membrane integrity in Gram-positive and Gramnegative bacteria [44]. In the current study, proteins associated with hopanoid and terpenoid biosynthesis pathways (hpnJ and dxs) were down-regulated in both ZM532 (mutant) and ZM4 (wild-type) (Additional le 14: Table S14). Similarly in transcriptome, hopanoid genes (ZMO_RS03910, ZMO_RS04350, ZMO_RS07175) were down-regulated in ZM532 and ZM4 while ZMO_RS03920 was up-regulated in ZM4 (Additional le 7, 8: Table S7, S8). Previously, IspG, Dxs1, and IspA of ZM4 under ethanol stress were down-regulated [24,26,27,45], and similar pattern was recorded in our study under stress conditions. However, acetic acid and furfural stressors may cause detrimental effect on terpenoid biosynthesis and stability of cell membrane.
TatD is a DNA repair exonuclease contributed acid resistance In additition, ZMO_RS04890, encoded TatD family hydrolase were up-regulated gene in mutant strain ZM532 and was absent on wild-type strain ZM4 under acidic condition may crucial against acids resistance. Follow-up studies further showed that TatD bears 3′-5′ exonuclease activity that processes single-stranded DNA in DNA repair ( Fig. 6G; Additional le 7, 8: Table S7, S8) and participates in DNA fragmentation during apoptosis in S. cerevisiae [46] and Trypanosoma brucei [47]. Previous study showed that TatD-knockout cells are less resistant to the DNA damaging agent hydrogen peroxide [48]. Hydrogen peroxide may induce various DNA lesions, not only double-strand breaks, but also oxidation and deamination of bases and sugar modi cations [49,50]. TatD has ability to remove deaminated nucleotide from DNA chain, inferring that it may be involved in H 2 O 2 -induced-DNA repair [48]. Since TatD is an evolutionarily conserved protein, it should have an important cellular role. However, our understanding of this protein is largely hampered due to lack of knowledge of its biological functions and structure-to-function relationship. So, in the future, we can provide evidence that TatD is a 3′-5′ exonuclease that may process single-stranded DNA in DNA repair.
Contribution of OstA in furfural and acetic acid tolerance ZMO_RS01205, ostA encoded organic solvent tolerance protein was up-regulated gene in mutant strain ZM532 and absent in wild-type strain ZM4 under stress condition (Additional le 7, 8: Table S7, S8; Table  S14). An earlier studies reported gene ostA is one of the genes contributing to the level of organic solvent tolerance [51,52]. In the future, it will be critical to investigate the single and combined effects of genes on the increase in Ost activity in response to salt and acid stress. The overexpression of previously reported transcriptional regulator proteins may be one of most effective method for increasing the Ost of Z. mobilis and other microorganism. Moreover, It will also assist to identify the transcriptional regulator proteins which are important in the Ost mechanisms in ZM4.

Candidate genes and proteins involved in energy generation and conservation
Transmembrane ATPases decompose ATP into ADP to release energy for importing cell metabolism compounds and exporting the contaminants to inhibit cellular processes. In our transcriptomics data, ATP synthase family proteins such as ZMO_RS01070, ZMO_RS04090, ZMO_RS01900 were up-regulated in ZM532 and ZM4 but their expressions were higher in ZM532 than ZM4. Also, ZMO_RS04930 and ZMO_RS05235 were up-regulated exclusively in ZM4 under the acid inhibitions (Additional le 7, 8: Table   S7, S8). Bacteria can increase the activity of H + -ATPase in the presence of acids to reduce proton loss. In bacteria, protons are observed to be transported from cells via H + -ATPase, an ATP-consuming process [43]. As a result, increased H + -ATPase activity and energy accumulation improve cells' ability to regulate pHi homeostasis under stress environment. Hence, the up-regulation of genes may helful to H + pump out from the cytoplasm by using ATP as previously reported by Rutkis et al. [53] and Yang et al. [40], respectively. Our ATP synthase family up-regulated proteins results are inline with these results. In the proteomics data, ZMO_RS01435 which encodes ATPases was up-regulated exclusively in ZM532 while ZMO_RS02975 (encoding F0F1 ATP synthase subunit C) was up-regulated only in ZM4 (Additional le 14: Table S14). The F1Fo ATP synthase can be used to activate and hydrolyze ATP to pump H + for intracellular pH homeostasis [54,55]. Additionally, a previous study revealed F1F0 ATPase hydrolyzes ATP to pump protons when respiration is disrupted culminating in intracellular neutral phase retention of the mitochondrial membrane [56]. Moreover, proteins involved in the respiratory chain for energy production like dA, ZMO_RS07890, gloB and wrbA were up-regulated with higher expression levels in ZM532 than ZM4 (Additional le 14: Table S14). The transcriptome pro le indicates ZMO_RS04970, ZMO_RS03175, ZMO_RS04970 were up-regulated in both strains and are involved in respiratory chain for energy production while ZMO_RS06760, ZMO_RS07025 and ZMO_RS02530 were down-regulated (Additional le 7, 8: Table S7, S8). Previous studies have con rmed that these proteins are essential of higher ethanol production and growth [42] however, up-and down-regulation of genes involved in various biological processes in response to various stressors.
Upregulation of ferredoxin promote high proton transport capacity against acids stressors Ferredoxin (Fdx) refers to a class of small proteins that bind inorganic clusters containing two to four iron atoms and an equal number of sulfur atoms [57]. The whole-genome sequencing of bacteria and archaea has uncovered number of genes encoding these proteins [58]. Fdxs are typically ascribed to electron transport activities, and some of them are necessary for metabolism to work properly [59] Table  S14). Up-regulation of these proteins is important for cell recovery from DNA damage caused by these inhibitors.

Candidate genes and proteins involved in chaperones participate in resistance
Proteins associated with posttranslational alteration, protein turnover, and molecular chaperone complex responsive to stress shock could react to various stress conditions, including extreme temperature, depletion of cellular resources, concentrations of ions and toxic substances [73]. Transcriptional level of ZMO_RS02930, Novel00004 (dnaK) and ZMO_RS08760 encode GroEL protein for adaptation to acidic stress [41]. This protein (GroEL protein) was however, highly expressed in ZM532 than in ZM4. An earlier study revealed that dnaK is crusical for microbe survival in evironmental stress conditions [74] becides, dnak play signi cant role in refolding of damaged proteins. Two novel genes (Novel00013 and Novel00014), which encode GroEL were up-regulated exclusively in the mutant strain ZM532, which may account for the robustness of our mutant strain against acetic acid and furfural stresses and could lead to high ethanol production ( Fig. 6B; Additional le 7, 8: Table S7, S8). Previous studies have con rmed that these proteins are necessary for normal growth of E. coli under toxic antibiotic [11,26] and temperature stress conditions [75]. Our transcriptomics result showed that the expression level of Clp protease complex, like ZMO_RS01740 (clpA), ZMO_RS04250 (ATP-dependent Clp protease proteolytic subunit), ZMO_RS04255 (clpX), ZMO_RS06375 (clpB), and ZMO_RS07775 (clpS) were up-regulated in both ZM4 and mutant ZM532 but the expression level of Clp protease were higher in mutant ZM532 compared with ZM4 ( Fig. 6B; Additional le 7, 8: Table S7, S8). These may be involved in protein remodeling and reactivation [41,[76][77][78]] to enhance the expression of these proteins to protect DNA and protein from damage in acidic cytoplasm. However, our transcriptomics results demonstrated that sulfur encoding genes (ZMO_RS00040, ZMO_RS06540, ZMO_RS00045 ZMO_RS03345, and ZMO_RS00025) were up-regulated in ZM532 and ZM4 but their expressions were higher in ZM532 compared to ZM4 ( Fig. 6E; Additional le 7, 8: Table S7, S8). As furfural and acetic acid could inhibit sulfhur amino acid biosynthesis either by restricting the availability of reduced sulfur (H 2 S) from sulfate or by inhibiting the incorporation of reduced sulfur into cysteine. The inhibition of sulfate reduction is unlikely to represent the initial action of furfural that inhibits growth [79]. Up-regulation of these genes may have contributed to the improved tolerance of Z. mobilis to acetic acid and furfural stressors.
Our proteomics data showed that molecular chaperone complexes (ZMO_RS06375 (ClpB), ZMO_RS01740 (ClpA)) were up-regulated in both strains (Additional le 14: Table S14).This contradicts results of previous studies [27]. Such proteins belong to the multi-chaperone system induced by stress, essential for the folding of newly synthesized polypeptides. Another chaperone protein, ZMO_RS04435 (Hsp20 family protein) which regulates bacteria growth and survival under different stresses was upregulated in the mutant ZM532. Hsp20 was found to stabilize both archaeal and bacterial membrane lipids and small HSPs in microbial pathogenesis [80-83]. However, chaperone ZMO_RS03810 (peptidylprolyl isomerase) which can maintain overall reduction in the level and folding of OMPs and to the induction of the periplasmic and ZMO_RS07675 (tetratricopeptide repeat protein) involves sensing and treatment of defective or incomplete protein structures under stress responses as previously discussed [82, 83] both proteins exclusively found only in mutant strain ZM532. For inhibitor tolerance of Z. mobilis cells, control of these stress response molecular chaperones may indeed be helpful. ZMO_RS01850 (iron-sulfur cluster assembly accessory protein) was up-regulated which may assemble or x oxygen-labile FeS clusters from extracellular iron chelators under oxidative stress and iron uptake (Additional le 14: Table S14). ZMO_RS08485 (grxD), ZMO_RS03370 (grxC), and ZMO_RS04910 (thioredoxin) bind FeS clusters and transport the clusters to speci c enzymes. The enzymes were upregulated in both strains (ZM4 and mutant ZM532). ZMO_RS04910 (thioredoxin) was exclusively up- Up-regulated proteins, Pgk, gpmA and ZMO_RS07905 (glucokinase) were found only in mutant strain ZM532 in our proteomics data while pgl was up-regulated in both strains (Additional le 14: Table S14). Moreover, alcohol dehydrogenase encoded ZMO_RS07165 and ZMO_RS05560 which were up-regulated in both strains but these genes were doubly expressed in ZM532 compared to ZM4. Lactoylglutathione lyase (ZMO_RS03400), hydroxyacylglutathione hydrolase (ZMO_RS03395), 2-hydroxyacid dehydrogenase (ZMO_RS05565), glucose-6-phosphate isomerase (ZMO_RS05445), galactose-1epimerase (ZMO_RS03970), and glucose-6-phosphate isomerase (ZMO_RS05445) were up-regulated in ZM532. These genes may partly account for the robustness of our mutant strain ZM532 against acetic acid and furfural stresses ( Fig. 6C; Additional le 7, 8: Table S7, S8). The up-regulation of these genes stimulate more ATPs for acidic tolerance as established by previous reports [41,42].

Role of porin in acid resistance
We identi ed up-regulated gene ZMO_RS08390, encoding carbohydrate porin when compared to resistance strain with wild-type (AF_ZM532 vs AF_ZM4) (Fig. 6H) facilitating the passive transport of various chemicals. For example, non-speci c porins, such as OmpA, found in outer membrane proteins, promote the passive transport of many small molecules [90,91].
Additionally, this protein is related to peptidoglycan via a exible periplasmic motif that interacts noncovalently with peptidoglycans [92]. Because porins are linked to antibiotic resistance in Gram-negative bacteria because they enable the passive diffusion of drugs throughout the outer membrane. Although prior research suggested that porins regulate the antibiotic resistance, but contribution of porin in resistance to acids (furfural and acetic acid) largely unknown and not studied yet.
In addition, our transcriptomics data showed amino acids such as histidine, cysteine, arginine, and serine were differentially expressed under acetic acid and furfural stresses. The genes encoding histidine include ZMO_RS05290, ZMO_RS05275, ZMO_RS02495, ZMO_RS05250 (succinyl arginine dihydrolase), and ZMO_RS03345 (cysteine synthase A) were up-regulated in both strains and in line with [79]. However, ZMO_RS05290 and ZMO_RS02495 were only found in the mutant ZM532 ( Fig. 6F; Additional le 7, 8: Table S7, S8). The up-regulation of these genes in the mutant ZM532 may enhance tolerance to acid stresses. Previous studies also con rmed that tolerance of E. coli to furfural can be improved signi cantly by serine, arginine, histidine and aromatic amino acids [79]. Arginine and lysine can also enhance Salmonella typhimurium's resistance to acetic acid stress [94].

Validation of differentially expressed genes under inhibitory (AF) conditions by qPCR
The results of qPCR showed three DEGs (ZMO-RS02740, ZMO-RS00080, and ZMO-RS08110) were upregulated in ZM532 while ZMO-RS03395, ZMO-RS08600 and ZMO-RS06525 were down-regulated in the same strain ZM532. Conversely, among the selected DEGs in ZM4, ZMO-RS00065 and ZMO-RS02800 were up-regulated, while ZMO-RS01385 and ZMO-RS03775 were down-regulated which are in consonance with the transcriptome results ( Figure S8; Additional le 7, 8: Table S7, S8). These genes had high expression either as up-or down-regulated in RNA-seq results, hence, give clue for their potential for functional validation in our subsequent experiments.
Veri cation of RNA-seq candidate genes involved in bacteria tolerance mechanism DEGs detected in ZM4 or ZM532, ZMO_RS02740 and ZMO_RS06525 in RNA-seq were selected for veri cation through Type I-F CRISPR-Cas system technology following recommended procedure ( Figure  S9) [95]. CRISPR-Cas Type I-F edited Z. mobilis revealed that the protospacer-bearing plasmids had signi cant interference activity. We transferred the DNA cleavage of interest to a PAM-anking sequence on the chromosome for self-targeting and genome engineering. The ZMO_RS02740 (204 bp) and ZMO_RS06525 (1275 bp) were selected as engineering targets. Plasmids were primarily constructed to import a leader-repeat-spacer-repeat cassette of an arti cial CRISPR expression individually (Fig. 7A). A donor DNA comprising of two homology arms for supporting homologous recombination engineered to carry expected mutations to improve the reliability of selected genotypes by self-targeting (Fig. 7B). By using genome engineering plasmids pKO-ZMO_RS02740 and pKO-ZMO_RS06525 (FigureB), both target genes were successfully deleted in ZM4 and ZM532 (Fig. 7C-D). The genotypes of randomly selected transformants in ZM532 and ZM4 were analyzed by colony PCR and Sanger sequencing and con rmed deletion of both genes (Fig. 7C-D).
Cell growth, glucose consumption, and ethanol production of mutant strains ∆ZMO_RS02740 and ∆ZMO_RS06525 under acetic acid and furfural conditions Four mutant strains (ZM532∆ZMO_RS02740, ZM4∆ZMO_RS02740, ZM532∆ZMO_RS06525, and ZM4∆ZMO_RS06525) were investigated under rich media and furfural (3 g/L) with acetic acid (5 g/L) conditions, respectively. Furfural and acetic acid affect glucose consumption, cell growth, and ethanol production (Fig. 8A, B,C and D). With same initial OD600, when strains were cultivated for 36 hours, the ZM4∆ZMO_RS06525 OD600 values was increased by 5.6% compared with the wild-type strain ZM4 under the same initial OD600. This OD600 value decrease when ZMO-RS02740 was knocked out in ZM532 and ZM4. The growth activity and glucose consumption of mutant strains ZM532∆ZMO_RS02740 and ZM4∆ZMO_RS02740 were decreased and thus, increasing fermentation time from 42 h in ZM532 to 55 h. Ethanol production was 58% higher in ZM532 than that of ZM532∆ZMO_RS02740. However, in ZMO_RS06525 knockout in ZM4, time of fermentation was signi cantly decreased from 60 h for ZM4 to 42 h for ZM4∆ZMO_RS06525, which contributed to 45.54% increase in ethanol production (Table 4). These results demonstrate the mutant ZM532 has more ability to convert sugar to ethanol and to withstand toxic conditions. Moreover, these ndings are consistent with our transcriptome data. Values are the means and standard deviations of representative experiment with three technical replicates, and error bars indicated standard deviation, **p < 0.01, ***p < 0.001 represents the difference between the control group and mutants strains Evaluation of candidate resistance genes under furfural and acetic acid tolerance by complementary study Lignocellulose inhibitors are composed of aldehydes such as HMF, furfural, weak acids, particularly acetic acid [11]. Ethanol and the toxicity of these inhibitors are in uenced by bacterial cell, lipid structure and uidity, membrane permeability, and physiological processes, including intake of nutrients, electron transport chain, and absorption and energy transduction [96]. Resistance to these inhibitors is a complex phenotype, which is controlled by mysterious regulatory mechanisms. Synthesis of resistant strains by functional and evolutionary engineering is a valuable way to distinguish genetic elements that are important to the resistance of inhibitors [39]. Four plasmids bearing candidate operons were constructed on the basis of a shuttle vector pEZ15Asp with Ptet as the promoter to investigate the impact of these genetic variants on combined acetic acid and furfural resistance. These plasmid constructs were then separately transferred into competent cells of ZM532 and ZM4, including the empty vector pEZ15Asp as the control. Besides, recombinant strains expression pro le were examined in without stress and with stress (acetic acid and furfural) conditions to analyze their effect on cell growth. Since the production of ethanol in Z. mobilis is closely linked to cell growth and substantially reduced by the inhibitory effects of toxic compounds [97]. Hence, these results suggested that the of ZM406525 encoding a major facilitator superfamily (MFS) containing recombinant strain failed to contribute to the resistance of acids in Z. mobilis and ZM532, which is consistent with our RNA-Seq outcome (Fig. 8F). Many major facilitator superfamily transporters are essential for microorganisms to grow under stress conditions. Several superfamily transporters of major facilitators are important for microorganisms to develop under conditions of stress [98]. Gram-negative bacteria can reduce their entry by establishing a low permeability barrier to restrict the intracellular concentration of toxic inhibitors [99]. This non-speci c phenomenon, such as the down-regulation of ZMO06525, which encodes a major facilitator superfamily (MFS) transporter protein was present in ZM4;while all strains had approximately similar growth rates under normal conditions (Fig. 8E). In addition, the up-regulated expression of ZMORS02740 (Chemotaxis protein Mot A) was similar to our RNA-seq results (Fig. 8F). For the Ptet promoter, the fermentation time of ZMORS02740 was reduced as compared to mutant strain ZM532, which may be ZMORS02740 coordination with some other genes and linker genes for acids resistance. When we combined this gene with Ptet promoter, their balance was disturbed resulting in reduced fermentation time. Our mutant strain ZM532 had a higher hydrolysate metabolic performance in comparison to the parental one and other recombinant strains. This suggests that one gene is may not be adequate to explain the tolerance of acids, and the synergetic effects of several mutations in uencing protein structural modi cations.

Conclusion
One of the main challenges of cost-competitive production of bioethanol from lignocellulosic biomass is development of resistant strains toward stresses. Exploitation of the global regulatory landscape may show different impacts on bacterial metabolism leading to the overlap of cell stress responses. In our previous study we constructed a mutant ZM532 by genome shu ing, which is superior to the parental strain and Z. mobilis. However, the molecular mechanisms underlying the enhanced tolerance and shortened fermentation time was largely unknown. Therefore, genetic changes, proteins and gene expression pro le under AF (acetic acid and furfural) stress or without stress conditions were investigated using transcriptomics and proteomics to unravel the molecular mechanisms in the wild type (ZM4) and mutant strain (ZM532). We also studied the functions of differentially expressed genes and proteins, and their relevant role in tolerance mechanism against inhibitors using the GO and KEGG databases. Our results revealed that the strain, ZM532 is more capable of converting biomass to ethanol; and enhanced tness in the toxicant-containing environment will bene t from this. Thus, ZM532 can be used to enhance bioethanol production under acetic acid and furfural conditions with ZM4 as a biocatalyst within a shorter fermentation period and greater productivity than ZM4. Overall, the Z. mobilis furfural and acetic acid tolerance molecular mechanism presented in this study may useful to synthetic biology focused on enhancing biological processes involved in ethanol production.

Bacterial strains and fermentation conditions
In this study, Z. mobilis, ZM4 and its mutant ZM532 were used. Both strains were cultivated on Rich Medium (RM) containing agar plates at 30°C [100]. All cell culture plates were incubated at 30°C until colonies were grown and stored at 4°C. Then, both strains were cultured in RM at 30°C without shaking for 16 h [100]. The strains were sub-cultured to fresh inoculum RM media or on RM plates added to agar power for 16 h at 30°C. Inoculation into fermentation medium was conducted when the initial cell density of optical density 600 (OD 600 ) for ZM4 and ZM532 were between 0.1 and 0.2. Cell pellets were extracted by centrifugation at 3,000×g for 4 min at 4°C and then inoculated in groups without inhibitors for 8 hours and with inhibitors (acetic acid and furfural) for 36 hours fermentation period. Both fermentations and cultring were performed in triplicates. The groups of ZM4 and ZM532 susceptible to acetic and furfural acids combination were named as AFZM4, and AF532, respectively, while other groups without inhibitors, the cells grown in RM were considered as control groups designated as RMZM4 and RM532. Based on previous experiment, the concentrations of acetic acid and furfural combinations were set at 5.0 g/L and 3.0 g/L to study the responses of ZM4 and mutant ZM532 [10]. The cells at exponential growth phase of ZM532 and ZM4 were collected. The cell pellets collected were used for subsequent experiments. Reads mapping to the reference genome and quanti cation of gene expression Raw data (raw reads) were interpreted via in-house perl scripts and clean data was extracted by eliminating reads comprising adapters. Then, the clean data of Q20, Q30, and GC content were computed.
Complete genome annotation les downloaded from the genome website Bowtie2-2.2.3 (ftp://ftp.ncbi.nlm.nih.gov/genomes/bacteria/Zymomonas_mobilis/) were used to construct a reference genome index and match clean reads to the reference genome [104]. Novel genes, operon and transcription start sites were identi ed by Rockhopper [22]. Then, extracted the 5'UTR (3'UTR) sequences. Then, RBS nder [105] and TransTermHP ([106] were used to predict SD sequences and terminator sequences, respectively. IntaRNA was used to predict the sRNA targets. And then we used RNAfold to predict RNA secondary structures [107,108]. Mapping of clean reads to each gene was counted using HTSeq v0.6.1. The fragments per kilobase of exon per million fragments mapped (FPKM) reads of every single gene were determined as described earlier [109].

Differentially expression genes and functional analyses
The read counts were modi ed for each sequenced library by edger software package via one standardized scaling factor. Differentially expression genes (DEGs) analyses of two conditions were performed using the DESeq package in R (1.18.0) [110]. Then using Benjamini & Hochberg approach, the p-values were adjusted. Genes with fold change (FC) > 1.5 and a false discovery rate (FDR; < 0.05) were considered as DEGs. Gene Ontology (GO) and pathway enrichment analyses of the DEGs were implemented with GOseq package in R [111], respectively. GO terminologies with adjusted p-value less than 0.05 were identi ed as signi cantly enriched.

Validation of differentially expressed genes by quantitative PCR
A quantitative PCR (qPCR) was conducted as previously described [112]. A total of 10 DEGs were used for qPCR and were chosen from the RNA-seq data on the basis of their differential expression patterns in Total protein extraction and protein quality test Samples (2 strains × 2 conditions × 3 replications) were independently ground in liquid nitrogen and lysed with a lysis buffer (consisting of 6 M Urea and 0.2 % SDS, 100 mM NH 4 HCO 3 ,pH 8.0), accompanied by 5 min of ice ultrasound. At 12000 g for 15 min at 4°C, the lysate was centrifuged and the supernatants transmitted to a clean tube. The extracts from each sample were reduced to 10 mM DTT for 1 h at 56°C and alkylated with iodoacetamide under dark room temperature for 1 hour. Samples were thoroughly vortexed with 4x the volume of precooled acetone and incubated at -20°C for at least 2 hours. Samples were then centrifuged and precipitated. They were washed twice with cold acetone and pellets were dissolved with a dissolution buffer of 0.1 M triethylammonium bicarbonate (TEAB, pH 8.5) and 6 M urea [113][114][115].
Label-free quantitative protein analysis  , China). Similarly, the plasmids Pmini-T-ZMO-RS06525 plasmid were constructed using the same approach (Additional File 3: Table S3). All DNA manipulation, such as transformation of E. coli, preparation of plasmid from E. coli, ligation, digestion of restriction enzyme, and agarose gel electrophoresis were conducted according to standard protocols [119]. Cell growth, ethanol production and glucose consumption by recombinant strains were calculated under furfural (3.0 g/L) and acetic acid (5.0 g/L) stress conditions.

Analytical methods
Concentrations of ethanol production and glucose consumption were determined using the Highperformance liquid chromatograph (HPLC, Agilent 1200) with column (HPX-87H) while UV Spectrophotometer was used to estimate the cell density at OD600. Fresh cultures were incubated at 30 ℃. At speci c time periods, 1uL of culture was harvested by centrifugation at 4,500 g for 2 mins, and the extracts collected and diluted 10 times. HPLC (Agilent 1200) was used to estimate ethanol production and glucose consumption at 0.6 mL/min ow rate with 5 mM H2SO4, and 35°C column temperature with 20.0 µL volume of injection, respectively.

Statistical analysis
According to the statistics, the data was considered signi cant as the value obtained was p < 0.05 and the data expression is mean ± SD (Mean ± SD).     Establishment of the Type I-F CRISPR-based genome engineering system for Z. mobilis (A) A selftargeting plasmid contained an arti cial CRISPR locus (B) Design of the self-targeting CRISPR and the donor DNA in knockout plasmids (C) deletion of mutants by screening of colony PCR (D) Con rmation by Sanger sequencing