Transcriptome analysis of Zymomonas mobilis ZM4 reveals mechanisms of tolerance and detoxification of phenolic aldehyde inhibitors from lignocellulose pretreatment

Background Phenolic aldehydes generated from lignocellulose pretreatment exhibited severe toxic inhibitions on microbial growth and fermentation. Numerous tolerance studies against furfural, 5-hydroxymethyl-2-furaldehyde (HMF), acetate, and ethanol were reported, but studies on inhibition of phenolic aldehyde inhibitors are rare. For ethanologenic strains, Zymomonas mobilis ZM4 is high in ethanol productivity and genetic manipulation feasibility, but sensitive to phenolic aldehyde inhibitors. Molecular mechanisms of tolerance for Z. mobilis toward phenolic aldehydes are not known. Results We took the first insight into genomic response of Z. mobilis ZM4 to the phenolic aldehyde inhibitors derived from lignocellulose pretreatment. The results suggest that the toxicity to cells is caused by the functional group of phenolic aldehyde, similar to furfural and HMF, rather than aromatic groups or phenolic hydroxyl groups. Transcriptome response against 4-hydroxybenzaldehyde, syringaldehyde, and vanillin, representing phenolic groups H, S, and G, respectively, was investigated. The atlas of the important genes responsible for significantly enhanced and repressed genes at the genomic level was illustrated. 272 genes with twofold greater expressions than non-treated controls and 36 gene clusters in response to challenges of these phenolic aldehydes were identified. Several reductases encoded by ZMO1116, ZMO1696, and ZMO1885 were found to play the key roles in reducing phenolic aldehydes into the corresponding phenolic alcohols. Reduction of phenolic aldehydes by overexpression of ZMO1116, ZMO1696, and ZMO1885 in Z. mobilis ZM4 resulted in the increased inhibitor conversion and ethanol productivity, especially for 4-hydroxybenzaldehyde and vanillin. Several transporter genes such as ZMO0282, ZMO0283, ZMO0798, ZMO0799, and ZMO0800 was also displayed significantly increased expressions against the phenolic aldehydes. Conclusions The genes encoding reductases are with potentials on phenolic aldehydes-tolerant genes contributing to the reduction of phenolic aldehydes into the corresponding phenolic alcohols forms for Z. mobilis ZM4. Overexpression of the key genes improved the conversion ratio and ethanol productivity of 4-hydroxybenzaldehyde and vanillin with high toxicity. New knowledge obtained from this research aids understanding the mechanisms of bacterial tolerance and the development of the next-generation biocatalysts for advanced biofuels production. Electronic supplementary material The online version of this article (doi:10.1186/s13068-015-0333-9) contains supplementary material, which is available to authorized users.

Ethanologenic strain Zymomonas mobilis ZM4 has demonstrated its potential in lignocellulose biorefinery applications for high ethanol productivity, ethanol tolerance, and genetic manipulation feasibility [5,6]. It tolerates high levels of phenolic acids [7], but is sensitive to phenolic aldehydes. Existence of minimum vanillin and syringaldehyde will lead to 1/3 decrease of ethanol yield for Z. mobilis [8] due to the disruptions of cell membrane and enzyme hydrophobic sites by phenolic compounds [1,2]. While numerous studies on tolerance of varied strains were reported against ethanol [9], acetate [10], furfural [11], studies on inhibition of phenolic aldehyde inhibitors are rare. Molecular mechanisms of tolerance for Z. mobilis toward phenolic aldehydes are not known.
In this study, we investigated the profiles of genome expression of Z. mobilis ZM4 using DNA microarray in response to three typical phenolic aldehydes, 4-hydroxybenzaldehyde representing phenolic group H, syringaldehyde for group S, and vanillin for group G. Candidate genes against varied phenolic aldehydes were identified from Z. mobilis ZM4 and were selectively expressed in Z. mobilis ZM4 for the confirmation of gene functions. This study provides the first insight into the genome response of Z. mobilis against phenolic aldehyde inhibitors. Tolerant genes identified in this study will serve as valuable resources for robust strain development for future biorefinery applications.

Cell growth and fermentation response of Z. mobilis ZM4 to phenolic aldehydes
The maximum cell growth of Z. mobilis ZM4 was depressed approximately 5.0, 1.4, and 3.1-folds when 4-hydroxybenzaldehyde, syringaldehyde, and vanillin were separately added, comparing to the non-phenolic aldehydes-treated culture (Fig. 1a). Z. mobilis ZM4 completely consumed glucose and reached the maximum ethanol titer at 8 h, but delayed about 12, 16, and 24 h and led to decrease of ethanol productivity from 0.73 to 0.24, 0.70, and 0.61 g/L/h when 4-hydroxybenzaldehyde, syringaldehyde, and vanillin were separately added (Fig. 1b, c). Apparently, 4-hydroxybenzaldehyde is the most potent inhibitory compound to Z. mobilis ZM4 among the three evaluated phenolic aldehydes. Conversion of phenolic aldehydes into the corresponding phenolic alcohols by Z. mobilis ZM4 was identified by GC-MS (Fig. 1d, e, Additional file 1). No further degradation of phenolic alcohols to other metabolites was detected. Approximately, 80 % of 4-hydroxybenzaldehyde was reduced into 4-hydroxybenzyl alcohol and 70 % of vanillin degraded into vanillyl alcohol after incubated for 36 h. On the other hand, syringaldehyde was relatively stable with only a minimum degradation in 36 h. The results were partially in agreement with the previous studies of Z. mobilis ZM4 converting 4-hydroxybenzaldehyde, syringaldehyde, and vanillin into their alcoholic forms [12].

Transcriptome response of Z. mobilis ZM4 against phenolic aldehydes
The microarray results revealed that a total of 2020 gene fragments were amplified. To confirm microarray expression data, 20 genes involving in central carbon metabolic pathways (Additional file 2) were selected for quantitative expression using qRT-PCR under the stress of the three phenolic aldehydes. Results were consistent in detection of the trend of up-or down-regulated expressions for both assays, except for only one gene showing a marginal variation (Fig. 2). As anticipated, quantitative results of qRT-PCR were more sensitive in general with greater fold changes than that of microarray.
Our hierarchical cluster analysis showed two clear distinguished expression patterns for 1908 genes in the genome of Z. mobilis ZM4 in response to challenges of 4-hydroxybenzaldehyde, syringaldehyde, and vanillin representing three groups of phenolic aldehyde inhibitors (Fig. 3). The genome response to syringaldehyde challenge was similar and very closely related to the non-treated control. In contrast, the expression patterns against 4-hydroxybenzaldehyde and vanillin were closely related and distinct from the control as well as the responses treated with syringaldehyde. Results from the replicated microarray experiments showed high levels of consistence with minimum variations within each treatment. Based on their expression responses, two major groups of genes were clustered separately. The control and the syringaldehyde-treated cultures shared a group of approximate 16 genes showing up-regulated response. In contrast, most genes in this group displayed a repressed or down-regulated expression in response to the 4-hydroxybenzaldehyde or vanillin treatment. On the other hand, many genes showing up-regulated expressions against 4-hydroxybenzaldehyde and vanillin displayed normal or down-regulated expressions for the syringaldehyde treated and the control. In response to challenge of a single phenolic aldehyde, 144 genes showed   between the DNA microarray and qRT-PCR. The gene expression ratios of both microarray data and qRT-PCR data for 20 genes were log transformed in base 2 (log 2 , treatment/control), and the microarray log 2 ratio values were plotted against the qRT-PCR log 2 values. The 20 selected genes included ZMO0152 (pyruvate kinase), ZMO0177 (16S rRNA), ZMO0179 (fructose-bisphosphate aldolase), ZMO0367 (glucose-6-phosphate 1-dehydrogenase), ZMO0368 (6-phosphogluconate dehydrogenase), ZMO0369 (glucokinase), ZMO0387 (hpcH/ hpaI aldolase), ZMO0543 (aconitate hydratase), ZMO0544 (isocitrate dehydrogenase), ZMO0567 (succinyl-CoA synthetase), ZMO0569 (succinate dehydrogenase), ZMO0997 (KDPG aldolase), ZMO1237 (d-isomer specific 2-hydroxyacid dehydrogenase), ZMO1307 (fumarase), ZMO1360 (pyruvate decarboxylase), ZMO1478 (6-phosphogluconolactonase), ZMO1496 (phosphoenolpyruvate carboxylase), ZMO1596 (Fe-containing alcohol dehydrogenase), ZMO1608 (phosphopyruvate hydratase), and ZMO1963 (citrate synthase). The primers for qRT-PCR of the genes are listed in Additional file 1 Fig. 3 Hierarchical cluster analysis of genome expression of Z. mobilis strain ZM4 in response to selective phenolic aldehydes. Gene expression levels were clustered based on log 2 transformation. Color code red, green, or black each represents up-, down-, or normally regulated expression, respectively. Letter H, G, and S, each stands for, phenolic group H represented by 4-hydroxybenzaldehyde, phenolic group G represented by vanillin, and phenolic group S represented by syringaldehyde, respectively. CK stands for a non-phenolic aldehyde-treated control. Replications of each treatment for the microarray experiment were marked in Roman numeral up-regulated expression and 299 genes repressed with a greater than twofold changes against 4-hydroxybenzaldehyde, a representative of p-hydroxyphenyl group (H), compared with the control (Additional file 3). For vanillin, a representative of guaiacyl group, 111 genes were up-regulated expressed and 198 genes down-regulated. For syringaldehyde representing syringyl group (S), only 17 genes were up-regulated and 63 down-regulated. We found 272 genes including 36 gene clusters showing upregulated expressions against these phenolic aldehydes distributed throughout the entire genome of Z. mobilis ZM4 (Fig. 4). Another 560 genes including 52 gene clusters were repressed significantly by three phenolic aldehydes. We identified 72 candidate tolerance genes showing significantly up-regulated expressions against challenges of at least two phenolic aldehydes examined in this study ( Table 1).
The results show that the transcriptome response to syringaldehyde stress for Z. mobilis was significantly different from that to 4-hydroxybenzaldehyde and vanillin stress. The possible reasons may come from the following two aspects: (1) low molecular weight. The molecular weights of 4-hydroxybenzaldehyde, syringaldehyde, and vanillin were 122.1, 182.2, and 152.2, respectively. Similar to S. cerevisiae to low molecular weights phenolic compounds [1], the low molecular weight syringaldehyde gave the least toxicity for Z. mobilis ZM4. (2) Substituent position. The methoxyl substituent at ortho position increases the toxicity of vanillins [13], while the methoxyl and hydroxyl substituents at meta and para positions or vice versa do not change the toxicity [1]. Vanillin has one more methoxyl group and syringaldehyde has two more methoxyl groups on 4-hydroxybenzaldehyde. The configuration of the two methoxyl groups may generate space shelling for the toxic methoxyl substituent at ortho position, and then reduced the toxicity of syringaldehyde.

Significant transporter genes for detoxification
Efflux system of living cells is an efficient mechanism for detoxification of external toxic compounds and internal damaging intermediates. We found at least four transporter genes were significantly up-regulated in response to the challenge of phenolic aldehydes. Two genes ZMO0799 and ZMO0800 belonging to adenosine triphosphate (ATP)-binding cassette (ABC) superfamily showed consistently greater expressions to all the three phenolic aldehydes (Table 1). Other two transporter genes, ZMO0282 and ZMO0798 belonging to resistancenodulation-cell division (RND) superfamily, showed up to sevenfold increased expression to the phenolic challenge. One gene ZMO1288 encoding a major facilitator superfamily (MFS) transporter protein was down-regulated more than twofold change. To limit the intracellular concentration of toxic inhibitors, Gram-negative bacteria can diminish its entry by developing a barrier of low permeability [14]. This non-specific phenomenon is also existing in Z. mobilis ZM4, such as the down-regulation of ZMO1288 encoding a major facilitator superfamily (MFS) transporter protein.

Overexpression and tolerance evaluation of the important genes in Z. mobilis ZM4
Genes with strong response to phenolic aldehydes on the transcriptional level were screened and overexpressed in Z. mobilis ZM4 to confirm their functions. Four genes, ZMO1116, ZMO1288, ZMO1696, and ZMO1885, were selected, including (1) significantly upregulated genes: ZMO1116 encoding an oxidoreductase with up-regulation by 2.53-, 2.47-, and 3.60-fold change, and ZMO1885 encoding a NADH:flavin oxidoreductase/NADH oxidase with 11.37-, 2.67-, and 3.33-fold change under the stress of 4-hydroxybenzaldehyde, syringaldehyde, and vanillin, respectively (data shown   (Fig. 6) were introduced into Z. mobilis ZM4 to generate the recombinants , and c syringaldehyde into the corresponding phenolic alcohols by candidate genes using NADH as a cofactor. Genes colored red indicate the most significant shared by all three phenolic aldehydes according to the protocol in Dong et al. [15]. The shuttle expression vector pHW20a, as well as pHW20a-gfp with gfp and a short linker were used as the two controls. The successful expression of ZMO1116, ZMO1696, and ZMO1885 were confirmed by vanillin degradation by the recombinants at the rate of 0.112, 0.125, and 0.023 μmol per minute, respectively. The expression of the transporter gene ZMO1288 was confirmed by the fluorescence intensity of 10,340 at 507 nm for the recombinant with the fused gfp, comparing to 9497 of the control containing only gfp reporter gene. The recombinants were cultured in RM medium separately amended with 5 mM of 4-hydroxybenzaldehyde, syringaldehyde, or vanillin. The cell growth, glucose consumption rate, ethanol productivity, and phenolic aldehyde inhibitors conversion rate were measured as shown in Table 2. Under the stress of 4-hydroxybenzaldehyde, the overexpression of ZMO1116 increased the cell growth, glucose consumption, ethanol productivity, and 4-hydroxybenzaldehyde conversion by 3 The low toxicity of syringaldehyde to Z. mobilis led to the limited inhibitory influence on cell growth and ethanol productivity, so the reduced level of syringaldehyde by the overexpression of the key genes also did not increase the cell growth and ethanol productivity. But the obvious improvements of cell growth and ethanol fermentation were obtained for the more toxic 4-hydroxybenzaldehyde and vanillin for inhibitor conversion. Overexpression of the reductase genes ZMO1116, ZMO1696, and ZMO1885 in Z. mobilis ZM4 significantly improved the tolerance to phenolic aldehydes, indicating the importance of reductases for phenolic aldehydes conversion. Müller et al. [16] found the reduction activity of NAD(P)H-dependent reductase encoded by ZMO1885 from Z. mobilis on the carbon double bond, and this study also verified the reduction activity of phenolic aldehydes into phenolic alcohols. Overexpression of ZMO1288 encoding a major facilitator superfamily (MFS) protein contributed to the highest inhibitor conversion, suggesting that the increase of toxin transport into intracellular space by enhancing ZMO1288 activity did not harm the intracellular activity. Instead, the enhanced transport of inhibitors to intracellular space facilitated the fast degradation and reduction of the toxic phenolic aldehydes.

Discussion
In this study, our transcriptome analysis using microarray revealed the first insight into genomic response of Z. mobilis ZM4 to the three groups of phenolic aldehyde inhibitors derived from lignocellulose pretreatment. We identified 272 up-regulated genes, including 36 gene clusters, against phenolic aldehydes as the potential candidate tolerance genes. We also identified significantly repressed genes and gene clusters to the phenolic aldehydes and provided their physical locations in the genome map. Furthermore, we illustrated mechanisms of bacterial tolerance and detoxification of these phenolic aldehydes through a group of genes encoding varied reductases and transporters. New knowledge obtained from this research aids understanding of the bacterial tolerance and the development of strong inhibitor tolerant fermenting strains.
Phenolic aldehydes are among the most toxic inhibitory compounds for microbial growth and fermentation. The toxicity is often thought caused by aromatic ring and the mechanism of inhibition is not clear. It is generally accepted that these phenolic compounds are classified into three groups of p-hydroxyphenyl group (H), guaiacyl group (G), and syringyl group (S) based on their methoxyl groups [1]. We used three phenolic aldehydes of 4-hydroxybenzaldehyde, vanillin, and syringaldehyde to represent each of these three different phenolic groups. We observed in the Z. mobilis ZM4 cultures, phenolic aldehydes concentrations were reduced and the corresponding alcohols were increased. As a result, when the phenolic aldehydes were reduced to certain lower levels, Z. mobilis ZM4 cells were able to accelerate the rate of glucose consumption and cell growth and eventually complete the fermentation. As demonstrated by our GC-MS analysis, 4-hydroxybenzaldehyde was converted into 4-hydroxybenzyl alcohol, and vanillin into vanillyl alcohol, and syringaldehyde into syringic alcohol. In these corresponding alcohols, a typical aromatic ring still existed but it apparently did not cause inhibition to the cell growth and metabolism of Z. mobilis ZM4. Functional group of aldehyde is apparently the toxic component of these phenolic aldehydes, regardless of their classification of phenolic group H, G, or S, based on the mass balance of phenolic aldehydes and its alcohol or acid derivatives, as well as the almost constant total phenolic concentrations. In response to the challenge of these phenolic aldehydes, the gene expression of ZMO0157, ZMO0788, ZMO0833, ZMO1116, ZMO1222, ZMO1254, ZMO1303, ZMO1399, ZMO1576, ZMO1696, ZMO1885, and ZMO1984 was significantly up-regulated under the stress of at least two phenolic aldehydes. These genes were found primarily encoding oxidoreductase, alcohol dehydrogenase, oxidoreductase/NADH oxidase, aldo/keto reductase, and a member of short-chain dehydrogenase/reductase (SDR), respectively. Our results suggest these genes are commonly involved in the conversion of phenolic aldehydes into the corresponding phenolic alcohols. However, since the complexity of the compound structure differs, the fine toned mechanism of the conversion pathway can be different. For example, syringaldehyde is more stable due to its complex structure and may require additional conversion pathways. It indicated that reduction and transport contributed to the conversion of the phenolic aldehydes into the corresponding phenolic alcohols. Similarly, furfural and HMF were conventionally called furan inhibitors that hindered studies on microbial tolerance to these common biomass fermentation inhibitors. He et al. [9,11] has found 127 genes and 433 genes differentially expressed in Z. mobilis in response to ethanol and furfural by DNA microarray, respectively. Metabolic conversion pathways of furfural and HMF into furanmethanol and furan-2,5-dimethanol by tolerant yeast Saccharomyces cerevisiae NRRL Y-50049 were presented and the toxic functional group of aldehyde, not furan group, was clarified [17][18][19]. It appeared there might be an aldehyde reductase gene family exist [20,21]. In this study, we observed multiple enzymes were involved in the conversion of phenolic aldehydes into corresponding alcohols that is consistent with previous reported. The overexpression of oxidoreductase (ZMO1116), zinc-binding alcohol dehydrogenase (ZMO1696), and NADH:flavin oxidoreductase/NADH oxidase (ZMO1885) with significantly enhanced conversion of phenolic aldehydes into phenolic alcohols gave further supports the multiple gene involvement in microbe tolerance to aldehyde inhibitors.
In our study, we also observed expression levels of at least eight transporter genes were significantly increased in response to varied phenolic aldehydes, including ZMO0143, ZMO0799, and ZMO0800 from ATP-binding cassette (ABC) superfamily; ZMO0282, ZMO0798, and ZMO1529 from RND (resistance-nodulation-cell division) superfamily; and ZMO0283 and ZMO0965 from efflux system proteins. In yeast, numerous transporter genes were identified to be involved in response to aldehyde including ABC transporter genes and other drug and toxin transporter genes [22]. Efflux pumps from RND family are considered one of the most efficient mechanisms for tolerance to toxin in Gram-negative bacteria [23]. Results of this study in Z. mobilis ZM4 in response to phenolic aldehyde inhibitors concur with the conclusion obtained from S. cerevisiae.
Our comprehensive analysis of the genome expression in Z. mobilis ZM4 provided the first important gene atlas for significantly enhanced and repressed genes at the genomic level. This illustrated distribution of potential candidate tolerance genes and gene clusters at the genomic landscape can serve as a valuable resource for genetic engineering efforts in genomic manipulation of Z. mobilis ZM4 for the continued tolerant strain development. In general, enhanced expressed genes under the inhibitor challenge pressure are considered as tolerance genes. However, this has to be in the context of consistent dynamics over a time course. In this study, we took merely a snapshot in the mid of the logarithm growth phase, thus, a more conserved approach should be taken in selection of candidate tolerance genes. On the other hand, it needs to point out that repressed genes are also important, particularly for those recovered from the early repression. Such genes often contribute to adaptation, redirecting gene interactions, and rewired networks for survival and tolerance development as demonstrated in S. cerevisiae [24,25].

Conclusions
The ethanologenic bacterium Z. mobilis ZM4 degrades phenolic aldehydes into their corresponding phenolic alcohols forms, and the toxicity of phenolic aldehydes to cell growth and ethanol fermentation is caused by the functional group of phenolic aldehyde rather than the phenolic aromatic ring. Among 272 up-regulated genes, including 36 gene clusters, reductases are the potential candidate tolerance genes for the reduction of phenolic aldehydes. Tolerant genes identified in this study will serve as valuable resources for robust strain development for future biorefinery applications.

Cell growth conditions
Cells of Z. mobilis ZM4 were cultured and maintained in RM medium containing 20 g/L glucose, 2 g/L KH 2 PO 4 , and 10 g/L yeast extract. The initial culture was incubated at 30 °C without agitation. RM medium was amended separately by adding 5 mM each of 4-hydroxybenzaldehyde, syringaldehyde, or vanillin for testing the inhibitor tolerance. A non-phenolic aldehyde-treated culture was served as control. Samples were taken at a 4 h interval till 36 h. For DNA microarray, 10 mL culture was transferred into 100 mL fresh RM medium when the culture was 2.0 at the optical density at 600 nm. Cells were harvested at 4 h for RNA extraction, DNA microarray, and quantitative real-time PCR (qRT-PCR).

DNA microarray
Total RNA of Z. mobilis ZM4 cells was extracted using a Trizol reagent kit from Invitrogen (Carlsbad, CA, USA). Purity and concentrations of the RNA samples were measured by a ratio of OD 260 / 280 readings using a Nan-oDrop ND-1000 spectrophotometer. DNA microarray was performed by CapitalBio Co., Beijing, China. cDNA fragments were hybridized with the probes and labeled with Cy3-dCTP fluorescent dye [26]. Data were analyzed using GeneSpring V12 Agilent Technologies (Santa Clara, CA, USA) with proper normalization and quality control procedures. Raw data were transformed into algorism phase based on Log 2 and statistical analysis was carried out using software CLUSTER 3.0 [27]. Significance of differential expression levels compared with the control was examined using False Discovery Rate (FDR) test (p < 0.05) [28] and analysis of variance (ANOVA). A threshold of fluorescence intensity with more or less than twofold changes was regarded as significantly differentially expressed genes. Finally, a hierarchical clustering analysis was performed using Java Treeview (Stanford University, Stanford, CA, USA).

qRT-PCR assays
To confirm expression data from microarray, the quantitative mRNA expressions were performed using qRT-PCR on a CFX96TM Real-Time System with C10000TM Thermal Cycler from Bio-Rad Laboratories (Hercules, CA, USA). Primers for qRT-PCR of the selected genes are shown in Additional file 2. The first strand of cDNA was synthesized using a cDNA synthesis kit (Torobo Co., Osaka, Japan). PCR amplification was prepared using an SYBR Green Real-time PCR Master Mix. PCR amplification profile used was follows: 94 °C for 5 min, then 35 cycles at 94 °C for 2 min and 62 °C for 30 s, and 72 °C for 30 s. 16S rRNA gene ZMOr003 was used as an internal control for data acquisition and normalization.

Construction of Z. mobilis recombinants
Genomic DNA of Z. mobilis ZM4 was extracted using Omega-Biotek Bacterial DNA kit (Norcross, GA, USA). Primers for amplification of ZMO1116, ZMO1288, ZMO1696, and ZMO1885 genes were listed in Additional file 3. Construction of the plasmids harboring ZMO1116, ZMO1288, ZMO1696, and ZMO1885 genes is illustrated in Fig. 6. The transporter gene ZMO1288 was fused with a reporter gene gfp which was amplified from the plasmid pMG36E [29], with a short linker encoding seven glycines at the C-terminal of gfp between gfp and ZMO1288 for confirmation of its expression. The plasmids were introduced into E. coli S17-1 λπ, and then into Z. mobilis ZM4 through biparental transconjugation method [15].

Evaluation of Z. mobilis ZM4 recombinants
The oxidoreductase activity of the recombinants was evaluated in 200 μL of reaction mixture containing 0.1 M potassium phosphate buffer (pH 7.5), 0.329 mM vanillin, 2.25 mM NADH, and 20 μL cell-free extract [30] at 30 °C for 30 min, and then stopped by addition of 10 μL of 3 M trichloroacetic acid solution and measured vanillin reduction by HPLC. One oxidoreductase unit (U) was defined as the amount of the crude enzyme required to reduce 1 μmol of vanillin per min at pH 7.5 and 25 °C. The expression of transporter gene ZMO1288 fused with gfp gene was detected by monitoring fluorescence using Synergy H1 Hybrid Microplate Reader at 507 nm (Biotek Instruments, Winooski, VT, USA). Fermentation performance of the newly generated Z. mobilis recombinants was evaluated as described above using RM medium separately amended with 5 mM each of 4-hydroxybenzaldehyde, syringaldehyde, or vanillin. Samples were periodically taken from the culture broth and supernatant obtained by centrifuging at 12,000g for 5 min and filtering through 0.22-μm filters.

HPLC and GC-MS analysis
Glucose and ethanol were analyzed using HPLC equipped with a refractive index detector RID-10A (Shimadzu, Kyoto, Japan) with an Aminex HPX-87H column (Biorad, Hercules, CA, USA) at 65 °C using a mobile phase of 5 mM H 2 SO 4 at a rate of 0.6 mL/min. 4-Hydroxybenzaldehyde, syringaldehyde, and vanillin were analyzed using reverse-phase HPLC (SPD-20A, Shimadzu, Kyoto, Japan) equipped with a YMC-Pack ODS-A column (Tokyo,