A bacterial aromatic aldehyde dehydrogenase critical for the efficient catabolism of syringaldehyde

Vanillin and syringaldehyde obtained from lignin are essential intermediates for the production of basic chemicals using microbial cell factories. However, in contrast to vanillin, the microbial conversion of syringaldehyde is poorly understood. Here, we identified an aromatic aldehyde dehydrogenase (ALDH) gene responsible for syringaldehyde catabolism from 20 putative ALDH genes of Sphingobium sp. strain SYK-6. All these genes were expressed in Escherichia coli, and nine gene products, including previously characterized BzaA, BzaB, and vanillin dehydrogenase (LigV), exhibited oxidation activities for syringaldehyde to produce syringate. Among these genes, SLG_28320 (desV) and ligV were most highly and constitutively transcribed in the SYK-6 cells. Disruption of desV in SYK-6 resulted in a significant reduction in growth on syringaldehyde and in syringaldehyde oxidation activity. Furthermore, a desV ligV double mutant almost completely lost its ability to grow on syringaldehyde. Purified DesV showed similar kcat/Km values for syringaldehyde (2100 s−1·mM−1) and vanillin (1700 s−1·mM−1), whereas LigV substantially preferred vanillin (8800 s−1·mM−1) over syringaldehyde (1.4 s−1·mM−1). These results clearly demonstrate that desV plays a major role in syringaldehyde catabolism. Phylogenetic analyses showed that DesV-like ALDHs formed a distinct phylogenetic cluster separated from the vanillin dehydrogenase cluster.

. Lignin valorization through microbial catabolic functions. Vanillin (VN) and syringaldehyde (SN) are key intermediates for the production of industrially valued chemicals in microbial cell factories. VN and SN are initially oxidized to generate vanillate and syringate by VN and SN dehydrogenases, respectively. The VN dehydrogenase genes, ligV and vdh, are well characterized from various bacterial strains, whereas the SN dehydrogenase gene is not yet identified.
Scientific RepoRts | 7:44422 | DOI: 10.1038/srep44422 pathway and lower ring-cleavage pathways have been identified and characterized 5 . Among these genes, the VN dehydrogenase gene (ligV), which is essential for the catabolism of VN, was isolated by shotgun cloning 25 . The deduced amino acid sequence of ligV exhibited 35-53% identity with those of the known VN dehydrogenase genes of Pseudomonas 21 , Rhodococcus 27 , Corynebacterium 24 , and Amycolatopsis 18 . Even though LigV showed a broad range of activity against benzaldehyde derivatives, the activity for SN was considerably lower than that for VN. Further, disruption of ligV minimally affected the growth of SYK-6 on SN 25 . These results suggest that an alternative aromatic ALDH gene is involved in the catabolism of SN in SYK-6. Previously, we tried to isolate the SYK-6 SN dehydrogenase gene using the same procedure employed for the cloning of ligV. Consequently, bzaA and bzaB, whose products were capable of converting a wide range of benzaldehyde derivatives including SN, were isolated 28 . However, disruption of these genes in SYK-6 had only a slight impact on its growth on SN 28 .
In this study, in order to identify the SN dehydrogenase gene involved in SN catabolism in SYK-6, we carried out a genome-wide screening for the gene. Based on the SN oxidation activities of the products of 20 putative ALDH genes and their expression levels in the SYK-6 cells, a reduced number of candidate genes were examined further. Gene disruption experiments identified the actual SN dehydrogenase gene, and enzymatic properties of SN dehydrogenase and LigV were characterized.

Results and Discussion
Characterization of SN conversion by cell extracts of Sphingobium sp. SYK-6. To obtain information on properties of the enzymes involved in the conversion of SN in SYK-6, coenzyme requirements and induction profiles of the enzyme activity were examined. When an extract of SYK-6 cells grown in LB (300 μ g of protein/ml) was incubated with 300 μ M SN in the presence of 500 μ M NAD + , the extract converted SN into syringate with a specific activity of 34 ± 0.3 nmol·min −1 ·mg −1 . This activity was 7.0-and 14-fold higher than those obtained in the presence of NADP + and in the absence of a coenzyme, respectively. These results indicated that NAD + -dependent aromatic ALDHs are involved in the oxidation of SN in SYK-6.
Our previous work suggested that SN conversion was constitutive in SYK-6 when LB was used as a non-inducing condition 25 . To confirm the inducibility of SN conversion, enzyme activities of extracts of SYK-6 cells grown in a minimal medium, Wx-SEMP, supplemented with or without 5 mM SN or VN, were evaluated. In the presence of NAD + , the SN oxidation activities of the extracts from cells grown in Wx-SEMP (32 ± 1.5 nmol· min −1 ·mg −1 ) and Wx-SEMP with VN (31 ± 2.9 nmol·min −1 ·mg −1 ) were almost identical to that of cells grown with SN (30 ± 2.2 nmol·min −1 ·mg −1 ). These results suggest that the genes responsible for the oxidation of SN are constitutively expressed.
Genome-wide screening of candidates for the SN dehydrogenase gene. The genome sequence of SYK-6 revealed the presence of 20 putative ALDH genes in addition to the previously characterized ligV, bzaA, and bzaB (Table S1). Phylogenetic analysis of 23 ALDHs in SYK-6 revealed that ligV, bzaA, and bzaB are placed into different clades (Fig. 2). bzaA clusters with SLG_07610 and SLG_07790 sharing 46 to 57% amino acid sequence identity, while bzaB clusters with SLG_07270 and SLG_28320 sharing 46 to 63% identity. To reduce the number of candidate genes, the presence of transcripts of all ALDH genes in SYK-6 were evaluated by reverse transcription (RT)-PCR analyses using total RNA prepared from SYK-6 cells grown with SN or VN. DNA fragments with expected sizes were amplified for 18 ALDH genes (Fig. 2). In contrast, no amplification products of SLG_31150, SLG_34940, SLG_38120, SLG_11410, and SLG_32240 were obtained from RNA isolated from cells grown in either culture condition (Fig. 2).
To investigate the ability of putative ALDH gene products to oxidize SN, all ALDH genes were PCR amplified and cloned into pET21a(+ ) and expressed in E. coli cells harboring the resultant plasmids. SDS-polyacrylamide gel electrophoresis (SDS-PAGE) analysis of cell extracts of the E. coli transformants showed successful expression of all the genes except SLG_32240 and SLG_34940 (Fig. S1). Cell extracts of the E. coli transformants (10-100 μ g of protein/ml) were incubated with 100 μ M SN or VN in the presence of 500 μ M NAD + to evaluate their SN and VN oxidation activities ( Fig. 2 and Fig. S2). To account for the different levels of expression of each ALDH gene, specific activities were normalized to their expression levels calculated from a gel image of SDS-PAGE (Table S2). As a result, eight gene products, including the products of SLG_28320, bzaB, SLG_07270, SLG_38120, SLG_11410, bzaA, SLG_07610, and SLG_07790, had specific activities toward SN that were the same as or higher than the gene product of ligV (80-620%). All these aromatic ALDH gene products converted SN into syringate (Fig. S2). Notably, the gene product of SLG_07610 had the highest SN dehydrogenase activity (1.11 ± 0.01 μ mol· min −1 ·mg −1 ). Most gene products showed lower oxidation activities toward VN (33-72%); however, the gene products of ligV and SLG_07270 had higher activity toward VN than to SN.
Based on the transcriptional level of ALDH genes in SYK-6 cells and the SN oxidation activities of the gene products, four genes, SLG_28320, SLG_07270, SLG_07610, and SLG_07790, were selected as candidates for SN dehydrogenase genes involved in SN catabolism in SYK-6 and were further characterized.
Substrate range of the putative SN dehydrogenases. In order to characterize the substrate range of the candidates for SN dehydrogenase, the oxidation activities of cell extracts of E. coli carrying SLG_07270, SLG_07610, SLG_07790, and SLG_28320 along with ligV, bzaA, and bzaB were examined using a spectrophotometric lactate dehydrogenase-coupled assay with various aromatic aldehydes including SN, VN, benzaldehyde, p-hydroxybenzaldehyde, protocatechualdehyde, m-anisaldehyde, veratraldehyde, coniferyl aldehyde, salicylaldehyde, and m-hydroxybenzaldehyde (Fig. S3). The gene products of bzaA, SLG_07790, and SLG_28320 exhibited oxidation activities toward all of the substrates ( Table 1). The highest activities of LigV, BzaA, SLG_07610, and SLG_07790 were observed when m-hydroxybenzaldehyde was used as a substrate. In contrast, BzaB and SLG_07270 shared similar substrate ranges and showed the highest activities toward m-anisaldehyde. These two enzymes share relatively high amino acid sequence similarities (63% identity), implying a correlation between their substrate specificity and their amino acid sequence similarity.
Transcriptional levels of the putative SN dehydrogenase genes in SYK-6. Because the gene(s) responsible for the oxidation of SN may be constitutively expressed in SYK-6, we predicted that genes that are highly transcribed must be involved in SN catabolism. To accurately determine the transcriptional levels of the candidate genes, qRT-PCR analyses for ligV, bzaA, SLG_07270, SLG_07610, SLG_07790, and SLG_28320 were  (Table S2). Asterisks indicate activities lower than 0.01 μ mol·min −1 ·mg −1 .
Scientific RepoRts | 7:44422 | DOI: 10.1038/srep44422 performed using the total RNA isolated from SYK-6 cells grown with or without 5 mM of SN or VN. All six genes showed constitutive transcription, and ligV and SLG_28320 were transcribed at similar and much higher levels than other genes (Fig. 3). In contrast, mRNA levels of bzaA, SLG_07270, SLG_07610, and SLG_07790 in cells grown in SEMP were only 1.5 to 9.6% of that of ligV (Fig. 3). These results may suggest that SLG_28320 and ligV play a major role in the catabolism of SN.

Role of SLG_28320 and ligV in the catabolism of SN and VN.
To examine whether SLG_28320 and ligV are involved in the catabolism of SN, gene knockout mutants were created by the insertion of chloramphenicol or kanamycin resistance genes. These mutations were confirmed by Southern hybridization analyses (Fig. S4). The growth of SLG_28320 mutant (SME076), ligV mutant (DLV) 25 , and SLG_28320 ligV double mutant (SME 077) on 5 mM SN was compared to that of SYK-6 ( Fig. 4A). The growth rate of SME076 on SN was significantly reduced, and the final biomass yield was almost 65% that of the wild type. To determine if this growth defect was caused by the disruption of SLG_28320, pJB866 29 carrying SLG_28320 (pJB28320) was introduced into SME076 cells. The SME076 cells harboring pJB28320 grew on SN as well as the wild type (Fig. 4B). These results indicate that SLG_28320 is indeed involved in SN catabolism in SYK-6; therefore, this gene was designated desV. DLV showed a moderate growth reduction on SN as reported in our previous study 25 ; however, SME077 almost completely lost the ability to grow on SN (Fig. 4A). These results indicate that both desV and ligV are necessary for the efficient conversion of SN, and desV plays a dominant role in SN transformation. We also examined the involvement of desV in growth on VN (Fig. 4C). While a significant reduction in growth was observed in DLV, SME076 showed almost the same growth as the wild type. However, further growth reductions were observed in SME077, suggesting that desV also contributes, in part, to VN catabolism.
To further investigate the level of involvement of desV and ligV in the catabolism of SN and VN, the oxidation activities for SN and VN of SME076, DLV, and SME077 were determined using cell extracts (300 μ g of protein/ml) ( Fig. 4D and E). Whereas the SN oxidation activity of SME076 was significantly decreased (26% that of the wild type), DLV retained 72% of the original activity. SME077 showed only 15% of the wild-type activity. The drastic Substrate Specific activity (μmol·min −1 ·mg −1 of total protein) a   decrease in SN oxidation activity in SME076 caused by the disruption of desV appears to result in the slow growth of SME076 cells on SN and their low final biomass yield. In contrast, no substantial loss of VN oxidation activity was observed in SME076. However, SME077 exhibited a further decline in activity when compared to that of DLV. These results correspond with those obtained in the growth assays.
Purification of DesV and LigV. The coding regions of desV and ligV were each cloned into pET-16b, and His-tag fused desV and ligV were expressed in E. coli BL21(DE3). SDS-PAGE analyses revealed the production of proteins 57 kDa and 51 kDa in size (Fig. S5A), which are similar to the values calculated from the deduced amino acid sequences of His-tag fused desV (M r . 55,173) and ligV (M r . 52,879), respectively. DesV and LigV were purified to near homogeneity by Ni affinity chromatography. Using size exclusion chromatography, the native molecular masses of purified DesV and LigV were estimated to be 120 kDa and 220 kDa, respectively (Fig. S5B). Furthermore, similar results were obtained by native-PAGE (Fig. S5C). Based on the size of the monomers, DesV and LigV were deduced to be a dimer and a tetramer, respectively. Previously, a dimer and a tetramer of VN dehydrogenases have been reported from Burkholderia cepacia TM1 and Micrococcus sp. TA1, respectively 26 .

Enzymatic properties of DesV and LigV.
Purified DesV and LigV showed the same optimal temperature (50 °C) and optimal pH (10.0) (Fig. S6). A similar optimal temperature and pH were reported from VN dehydrogenases of B. cepacia TM1 and Micrococcus sp. TA1 26 .
DesV could oxidize SN in the presence of either NAD + or NADP + ; however, the specific activity was 7.3-fold higher when NAD + was used (0.99 μ mol·min −1 mg −1 ). In contrast, LigV specifically required NAD + (0.33 μ mol·min −1 ·mg −1 with NAD + and no activity with NADP + ). The UV-visible spectra of both DesV and LigV showed no absorbance related to bound flavin cofactors (data not shown).
The kinetic parameters of the purified DesV and LigV for SN and VN were determined (Table 2 and Fig. S7). These kinetic data clearly demonstrate that LigV functions as a VN dehydrogenase, while DesV has the potential to function as a dehydrogenase for both SN and VN. To date, kinetic parameters of aromatic ALDHs toward SN have not been reported, while the parameters toward VN were determined for VN dehydrogenases from Micrococcus sp. TA1 26 , B. cepacia TM1 26 , and Corynebacterium glutamicum ATCC 13032 24 , p-hydroxy benzaldehyde dehydrogenase from Acinetobacter sp. ADP1 30 , and ALDH1A1 and ALDH3A1 from humans 31 (Table S3). LigV and DesV exhibited the lowest K m and the highest k cat /K m values (Table S3). These results raise the question of why the SN oxidation activity of DLV was significantly reduced despite LigV exhibiting significantly low catalytic efficiency toward SN ( Fig. 4 and Table 2). LigV may have a role in the conversion of SN when its cellular concentration is high because ligV and desV showed similar expression levels in the SYK-6 cells, and little difference exists between the k cat values of DesV and LigV for SN (0.99 s −1 and 0.59 s −1 ), although the K m value of LigV for SN is very high (412 μ M). In contrast, a double mutant of desV and ligV (SME077) still retained a poor ability to grow on and convert SN and VN (Fig. 4). Moreover, the specific activities for the conversion of SN (5.2 nmol·min −1 ·mg −1 ) and VN (7.1 nmol·min −1 ·mg −1 ) with NAD + in SME077 were significantly higher than those of the wild type in the absence of NAD + (SN, 2.3 nmol·min −1 ·mg −1 ; VN, 0.8 nmol·min −1 ·mg −1 ). These results suggest that the remaining activities for the conversion of SN and VN in SME077 were derived from other NAD + -dependent aromatic ALDHs.
Involvement of multiple ALDH genes for the conversion of VN was also shown in Pseudomonas putida KT2440 32 . Although P. putida KT2440 has a vdh gene (PP_3357), disruption of this gene does not affect growth on VN. In contrast, proteomic analyses indicated that the production of other ALDHs, PP_0545, PP_1948, PP_2680, PP_3151, PP_5120, and PP_5258, increased in response to VN. Additional disruption of PP_0545, PP_1948, and PP_2680 in vdh mutant enhanced productivity of VN from ferulate 23 . Another example of the involvement of multiple ALDH genes in the assimilation of VN was reported in C. glutamicum ATCC 13032 24 . In this strain, vdh plays an important role in the degradation of VN. However, catabolism of VN was still observed in a vdh mutant, suggesting the presence of alternative ALDH genes for the oxidation of VN. In addition, the disruption of vdh showed no effect on growth on SN. In the genome sequence of C. glutamicum ATCC 13032 (NC_003450), we found seven putative ALDH genes in addition to vdh. Subsequent phylogenetic analysis indicated the presence of an ALDH gene (NCgl0523) whose deduced amino acid sequence shared 38% identity with that of desV (Fig. S8). NCgl0523 may be involved in the conversion of VN and SN in C. glutamicum ATCC 13032.
Aromatic ALDH genes affiliated with the DesV cluster, BzaA cluster, and LigV/Vdh cluster in living organisms were surveyed using the co-occurrence tool of the STRING database (Fig. S9). The genes which showed the highest amino acid sequence identities with DesV (31-65%), BzaA (39-64%), and LigV (29-84%) were found from in 37 taxa of bacteria, eukaryotes, and archaea (Table S5). The phylogenetic tree constructed using the amino acid sequences of the above genes showed that almost half of the extracted genes belonged to one of the three clusters  Table S4. ALDHs from SYK-6 are shown in blue, and ALDHs which had SN oxidation activity are indicated in bold blue.
Scientific RepoRts | 7:44422 | DOI: 10.1038/srep44422 (Fig. S10). When compared to the DesV and BzaA types of aromatic ALDH genes, the LigV/Vdh-type aromatic ALDH genes are more broadly distributed among bacteria. The aromatic ALDH genes in the DesV and BzaA clusters were found not only in the SYK-6-related Sphingomonad strains but also in other Proteobacteria and Actinobacteria. These observations may imply that, in addition to the LigV/Vdh cluster genes, aromatic ALDH genes classified into the DesV and BzaA clusters also generally participate in the catabolism of lignin-derived aromatic compounds. For example, Xanthomonas oryzae 41 , a gammaproteobacterial pathogen of rice, has a DesV ortholog, XOC_0933 which exhibited 65% amino acid sequence identity with DesV (Fig. S10). Interestingly, its proximal gene, XOC_0934, showed 61% amino acid sequence identity with the feruloyl-coenzyme A (CoA) hydratase/lyase gene (ferB) of SYK-6 that is responsible for the conversion of feruloyl-CoA to VN 42 . In addition, the gene set of XOC_0933-XOC_0934 was also found in several members of Xanthomonas including Xanthomonas axonopodis, Xanthomonas campestris, and Xanthomonas citri. These findings may suggest that the DesV-type aromatic ALDH gene is involved in the catabolism of plant-derived aromatic compounds such as ferulate in Xanthomonas.

Conclusions
The goal of this study was to identify the aromatic ALDH gene responsible for the conversion of SN from 20 putative ALDH genes in Sphingobium sp. SYK-6. For the first time, we were able to successfully identify desV as the SN dehydrogenase gene mainly involved in SN catabolism. Detailed information on a number of aromatic ALDH genes obtained in this study will be useful for application to biological lignin valorization. To establish economically viable biofuel production, increasing the value of lignin is essential. Recently, some approaches combining the chemical deconstruction of lignin and the microbial conversion of the resultant heterologous aromatic compounds have been attempted 17,43 . Since VN and SN are generally major intermediates of both microbial catabolism and chemical decomposition, aromatic ALDHs with a high catalytic efficiency toward these aromatic aldehydes, such as LigV and DesV, are extremely valuable for the production of chemicals from lignin.

Methods
Bacterial strains, plasmids, and culture conditions. The bacterial strains and plasmids used in this study are listed in Table S6. Sphingobium sp. SYK-6 and its mutant derivatives were routinely grown at 30 °C in Lysogeny broth (LB) or Wx minimal salt medium 44 containing 5 mM SN, 5 mM VN, or SEMP (10 mM sucrose, 10 mM glutamate, 0.13 mM methionine, and 10 mM proline). When necessary, 50 mg of kanamycin (Km)/liter, 30 mg of chloramphenicol (Cm)/liter, and 12.5 mg of tetracycline (Tc)/liter were added to the cultures. E. coli strains were grown in LB at 37 °C. For cultures of cells carrying antibiotic resistance markers, the media for E. coli transformants were supplemented with 100 mg of ampicillin/liter, 25 mg of Km/liter, and 12.5 mg of Tc/liter. Enzyme assays using cell extracts of SYK-6 and its mutants. Cells of SYK-6 and its mutants (DLV, SME076, and SME077) grown in LB were inoculated into the same medium (final concentration, 1%) and further incubated for 12 h. The resultant cells were washed twice with 100 mM KH 2 PO 4 -K 2 HPO 4 buffer (pH 7.0, buffer A). Cells resuspended in the same buffer were then broken by an ultrasonic disintegrator (UD-201; Tomy Seiko Co.). After centrifugation (19,000× g for 15 min at 4 °C), the supernatants were obtained as cell extracts. The protein concentration was determined by the Bradford method with bovine serum albumin as the standard. Cell extracts (300 μ g of protein/ml) were incubated with 300 μ M SN or 300 μ M VN in the presence and absence of 500 μ M of NAD + or NADP + at 30 °C. After incubation for 0.5 and 5.0 min, portions of the mixture were collected, and reactions were terminated by mixing them with the same volume of 0.2 N HCl. Supernatants obtained by centrifugation (19,000× g for 15 min at 4 °C) were filtrated and analyzed by high-performance liquid chromatography (HPLC; Acquity UPLC system; Waters) using a TSKgel ODS-140HTP column (2.1 by 100 mm; Tosoh) as described previously 45 . The mobile phase of the HPLC system was a mixture of water (75%) and acetonitrile (25%) containing formic acid (0.1%) at a flow rate of 0.3 ml/min. SN and VN were detected at 279 and 308 nm, respectively. The specific activity was expressed in moles of SN and VN converted per min per milligram of protein. For the determination of cofactor requirements, a centrifugal filtration of cell extracts was carried out with Amicon Ultra 3k (Millipore) to remove cofactors, and then the resultant fractions were used for the enzyme reaction. To examine the induction profile, cells of SYK-6 grown in LB were washed with Wx medium and resuspended in Wx-SEMP medium to an optical density at 600 nm (OD 600 ) of 0.2. Once cultures reached an OD 600 of 0.5 to 0.6, 5 mM SN or 5 mM VN was added to the cultures. After 6 h of further incubation, cell extracts were prepared and used for the enzyme assay. Genome search of putative ALDH genes. For the first step to search ALDH genes in the SYK-6 genome (AP012222 and AP012223), protein BLAST (BLASTP) was carried out using the deduced amino acid sequences of ligV, bzaA, and bzaB in the NCBI database. The BLASTP searches were then repeated using the homologous sequences found above to obtain more candidate genes. Pairwise alignments were performed using the EMBOSS Needle program through EMBL-EBI server (http://www.ebi.ac.uk/services) 46 . For phylogenetic analysis, multiple alignments were performed using the Clustal W program in MEGA software 47 , and then phylogenetic trees were constructed using the neighbor-joining algorithm of MEGA 7, employing 1000 bootstrap replicates.
RT-PCR and qRT-PCR analysis. After 6 h of incubation of SYK-6 cells in the presence and absence of 5 mM SN or VN, 2 ml of cultures were harvested. Total RNA was isolated using Isogen II (Nippon Gene Co., Ltd.), followed by treatment with RNase-free DNase I (Roche). PrimeScript reverse transcriptase (Takara Bio Inc.) was used to synthesize cDNA from 2 μ g of total RNA with random hexamer primer. A control PCR was performed with reverse transcriptase-negative samples to verify the absence of genomic DNA contamination. RT-PCR was performed with the resultant cDNA, specific primers, and ExTaq DNA polymerase (Takara Bio Inc.). qRT-PCR Scientific RepoRts | 7:44422 | DOI: 10.1038/srep44422 was carried out with a Fast SYBR green master mix (Applied Biosystems) and StepOne Real-time PCR System (Applied Biosystems). To normalize the amount of RNA in each sample, 16S rRNA was used as an internal standard. The primers used for the analyses are listed in Table S7.
Expression of SYK-6 ALDH genes. bzaA, bzaB, and other putative ALDH genes were amplified by PCR using PrimeSTAR GXL DNA Polymerase (Takara Bio Inc.) and primer sets listed in Table S7. The amplified fragments were cloned into pBluescript II KS(+ ) or pT7Blue, and then the nucleotide sequences were determined. The 1.4 to 1.5-kb NdeI-BamHI or NdeI-XhoI fragments from the resulting plasmids were cloned into the corresponding sites of pET21a(+ ) to yield the expression plasmids. E. coli BL21(DE3) cells harboring each expression plasmid were grown in LB at 30 °C, and the expression of the genes was induced for 4 h at 30 °C by adding 1 mM isopropyl-β -D-thiogalactopyranoside (IPTG) when the OD 600 of the culture reached 0.5. The resultant cultures were washed twice with buffer A, and then the cells resuspended in the same buffer were broken by an ultrasonic disintegrator. After centrifugation (19,000× g for 15 min at 4 °C), the supernatants were obtained as cell extracts. The expression of the genes was confirmed using SDS-12% PAGE. Cell extracts of the E. coli transformants (10-100 μ g of protein/ml) were incubated with 100 μ M SN or VN in the presence of 500 μ M NAD + at 30 °C. After incubation for 0.5 and 5.0 min, portions of the mixture were collected and the amounts of substrates were measured using HPLC. For the calculation of the specific activities of the ALDHs for the oxidations of SN and VN, the signal intensities of each band of the ALDH in SDS-PAGE were quantified using a LumiVision image analyzer (Aisin Seiki Co., Ltd). Specific activities were expressed in moles of SN and VN converted per min per milligram of protein, which was normalized to the expression level of LigV.
Substrate range. Enzyme assays to examine the substrate range of the gene products of ligV, bzaA, bzaB, SLG_07270, SLG_07610, SLG_07790, and SLG_28320 (desV) were performed according to the method described previously 25,37 . Cell extracts of the E. coli transformants (100-500 μ g of protein/ml) were incubated with 100 μ M substrate (Fig. S3), 500 μ M NAD + , 1.2 mM pyruvate, 1.0 U lactate dehydrogenase in buffer A at 30 °C. The decrease in the amount of substrates was measured spectrophotometrically using a spectrophotometer

Construction of mutants.
For the construction of a desV-disruption plasmid, a 1.5-kb HindIII-XbaI fragment carrying desV from pKS2832 was cloned into the same sites of pK18mobsacB with the insertion of a Cm resistance gene. The resulting plasmids, pK18-2832Cm was introduced into SYK-6 cells by electroporation, and the candidates for desV mutant (SME076) were isolated as described previously 48 . Similarly, the candidates for desV ligV double mutant (SME077) were obtained by introduction of a ligV-disruption plasmid, pIK34D 25 into SME076 cells. The disruption of each gene was confirmed by Southern hybridization analysis using digoxigenin system (Roche). The growth of the resulting mutant cells in Wx medium containing 5 mM SN or VN were examined by monitoring of OD 660 automatically every 4 h using Bio-Photorecorder (TVS062CA, Advantec). A complementary plasmid, pJB28320 was constructed by cloning a DNA fragment carrying desV into pJB866 29 . The resulting plasmid was introduced into the cells of SME076 by electroporation, and the growth of the transformant in Wx medium containing 5 mM SN was examined.
Purification of DesV and LigV. The 1.5-kb NdeI-BamHI fragment carrying desV and the 1.9-kb NdeI-XhoI fragment carrying ligV from pT21-2832 and pLVH, respectively, were ligated into the corresponding sites of pET-16b. The resultant plasmids, pT16-desV and pT16-ligV, were independently introduced into E. coli BL21(DE3), and then the His tag-fused desV and ligV were expressed. For purification, cell extracts were applied to His GraviTrap TALON columns (GE Healthcare). Purified fractions were subjected to desalting and centrifugal filtration with Amicon Ultra 30k (Millipore), and stored at − 30 °C until use. The purity of the preparations was examined by SDS-12% PAGE.
Determination of molecular mass. Purified DesV (100 μ g/26 μ l) and LigV (100 μ g/32 μ l) were subjected to size exclusion chromatography on a Superdex200 10/300GL column (GE Healthcare) eluted with 50 mM KH 2 PO 4 -K 2 HPO 4 buffer (pH 7.0) containing 150 mM NaCl at a flow rate of 0.5 ml/min as described previously 45 . Native PAGE was performed using a 5-20% polyacrylamide gradient gel with a high-molecular-weight calibration kit for native electrophoresis (GE Healthcare).
Enzyme characterization. The enzyme reaction was typically carried out in a 100 μ l reaction mixture containing buffer A, DesV (20 μ g/ml of protein, 180 nM dimer) or LigV (20 μ g/ml of protein, 95 nM tetramer), and 500 μ M NAD + or NADP + . After the incubation (25 sec for the kinetic analysis and 60 sec for the determination of optimal pH and temperature, and coenzyme requirement), the reaction was terminated by the addition of the same volume of 0.2 N HCl. The amounts of the substrates and products were measured using HPLC. Specific activities were expressed in moles of syringate or vanillate produced per min per milligram of protein at 30 °C. The optimal temperatures were determined using buffer A (15 to 70 °C), and the optimal pH were examined using 50 mM GTA buffer (50 mM 3,3-dimethylglutarate, 50 mM Tris, and 50 mM 2-amino-2-methyl-1,3-propanediol; pH 5.0 to 9.0), 50 mM N-cyclohexyl-2-aminoethanesulfonate (pH 9.0 to 10.0), and 50 mM N-cyclohexyl-3-aminopropanesulfonate (pH 9.7 to 11.0) at 30 °C. The K m and k cat values were obtained from Hanes-Woolf plots and expressed as mean ± standard deviation from at least three independent experiments. Kinetic parameters of Scientific RepoRts | 7:44422 | DOI: 10.1038/srep44422 DesV and LigV were determined using the following substrate concentration ranges: DesV (0.4 μ g/ml of protein) for SN, 0.25 to 5.0 μ M; DesV (0.5 μ g/ml of protein) for VN, 0.28 to 10.0 μ M; LigV (20 μ g/ml of protein) for SN, 25 to 3,200 μ M; and LigV (0.2 μ g/ml of protein) for VN, 0.25 to 5.0 μ M.
Sequence analyses of ALDHs. The deduced amino acid sequences of previously reported bacterial ALDH genes (Table S4) were obtained from NCBI database. To examine the presence of aromatic ALDH genes affiliated with the DesV, BzaA, and LigV/Vdh clusters in the living organism, a co-occurrence analysis of desV, bzaA, and ligV was conducted using the STRING database (http://string-db.org/) version 10 49 . Putative aromatic ALDH genes, the deduced amino acid sequences of which exhibited the highest similarity scores with desV, bzaA, and ligV, were picked up from 37 selected taxa of bacteria, eukaryotes, and archaea shown in Table S5. Construction of phylogenetic trees was done as described above. All the alignments are available upon request.