Bacterial strains and plasmids used in this study are listed in Table 1 and Supplementary Table S1. E. coli TOP10 was used for plasmid construction. The E. coli/C. glutamicum shuttle vectors pCES208 and pCXE50 were used for gene expression in C. glutamicum (Lee, 2014; Syukur Purwanto et al., 2018). The plasmid pK19mobsacB was used for constructing deletion and insertion mutants of C. glutamicum (Schäfer et al., 1994). DNA fragments were amplified by polymerase chain reaction (PCR) using TOPsimple™ DryMIX-Tenuto (Enzynomics, Daejeon, Korea). An EZ-fusion^TM HT Cloning Kit (Enzynomics) was used for cloning of multiple DNA fragments via Gibson assembly (Gibson et al., 2009). The sequence correctness of PCR products cloned in vectors was confirmed by sequencing. Transformation of C. glutamicum was performed by electroporation (Eggeling and Bott, 2005).

The primers for deletion and insertion of genes are listed in Supplementary Tables S2, S3. To construct gene knockout plasmids responsible for known or hypothesized ADHs, oxidoreductases, aldo/keto reductases, etc., the primer sets for amplifying two flanking regions of 20 genes are as follows: P1/P2 and P3/P4 for NCgl0186, P7/P8 and P9/P10 for NCgl0219, P13/P14 and P15/P16 for NCgl0313, P19/P20 and P21/P22 for NCgl0324, P25/P26 and P27/P28 for NCgl0503, P31/P32 and P33/P34 for NCgl0908, P37/P38 and P39/P40 for NCgl1213, P43/P44 and P45/P46 for NCgl1302, P49/P50 and P51/P52 for NCgl1459, P55/P56 and P57/P58 for NCgl1608, P61/P62 and P63/P64 for NCgl 1962, P67/P68 and P69/P70 for NCgl2122, P73/P74 and P75/P76 for NCgl2213, P79/P80 and P81/P82 for NCgl2277, P85/P86 and P87/P88 for NCgl2358, P91/P92 and P93/P94 for NCgl2382, P97/P98 and P99/P100 for NCgl2449, P103/P104 and P105/P106 for NCgl2582, P109/P110 and P111/P112 for NCgl2709, and P115/P116 and P117/P118 for NCgl2952. The two amplified fragments of each gene were cloned into pK19mobsacB/HindIII/EcoRI using Gibson assembly, resulting in the construction of pK19-Δ0186, pK19-Δ0219, pK19-Δ0313, pK19-Δ0324, pK19-Δ0503, pK19-Δ0908, pK19-Δ1213, pK19-Δ1302, pK19-Δ1459, pK19-Δ1608, pK19-Δ1962, pK19-Δ2122, pK19-Δ2213, pK19-Δ2277, pK19-Δ2358, pK19-Δ2382, pK19-Δ2449, pK19-Δ2582, pK19-Δ2709, and pK19-Δ2952 ( Table 1 ).

To restore pobA from APS963, pobA gene (1.2 kb) and upstream and downstream regions (1.2 kb) were obtained from genomic DNA of C. glutamicum ATCC 13032 via PCR using the primer sets P-pobA-F and P-pobA-R. The amplified fragment was cloned to HindIII/XbaI-treated pK19mobsacB using Gibson assembly to generate pK19-pobA. The plasmid pK19-ΔvanAB was constructed by fusion of PCR fragments obtained by using primer sets P-vanA-F/P-vanA-R and P-vanB-F/P-vanB-R with HindIII/EcoRI-digested pK19mobsacB ( Table 1 ).

To construct a plasmid for the overexpression of the native NCgl0324 gene in C. glutamicum, the NCgl0324 ORF (open reading frame) from C. glutamicum was amplified via PCR using primer sets P-0324-F and P-0324-R and C. glutamicum genomic DNA as a template and then assembled with EcoRI- and HindIII-digested pCXE50, yielding pXT-0324 ( Table 1 ). The plasmid for purification of NCgl0324 protein in E. coli was constructed as follows. The NCgl0324 gene was achieved via PCR using the primer sets P-0324-NF and P-0324-XR and cloned in pET-24a(+) cut with NdeI and XhoI. The resulting plasmid was designated pET-0324 ( Table 1 ). The plasmid construction for expressing a mutated comt gene from Rattus norvegicus encoding the soluble catechol O-methyltransferase mutant (Y200L COMT; accession number NP_036,663) is as follows (Law et al., 2016): a codon-optimized comt (0.666 kb) from Rattus norvegicus was synthesized from Integrated DNA Technologies (IDT, Iowa, United States; Supplementary Table S4) and cloned in pCXE50/EcoRI/HindIII, resulting in pYL200 ( Table 1 ). To overexpress this gene in C. glutamicum, a plasmid with insertional mutation in the TIR (translation initiation region) in front of comt ORF was constructed. The amplified tuf promoter by PCR using primers P-X48-F and P-Ptuf-mR was digested with XbaI and EcoRI and ligated with pYL200/XbaI/EcoRI to gain the pYL230 plasmid. In addition, the plasmid pYL250 with the truncated rrnB transcriptional terminator (T_t-rrnB ) was constructed by insertion of the 0.23-kb truncated T_t-rrnB (primers P-TrrnB-tF and P-X48-R) into the pYL200/HindIII/KpnI site. In order to express comt and car genes in different compatible vectors, the 1.2-kb fragment with tuf promoter, comt ORF, and T_t-rrnB in pYL250 was subcloned into the pCES208/NotI/KpnI site, yielding the pYLS2250 plasmid ( Table 1 ).

The plasmids pK19-Δ0186, pK19-Δ0219, pK19-Δ0313, pK19-Δ0324, pK19-Δ0503, pK19-Δ0908, pK19-Δ1213, pK19-Δ1302, pK19-Δ1459, pK19-Δ1608, pK19-Δ1962, pK19-Δ2122, pK19-Δ2213, pK19-Δ2277, pK19-Δ2358, pK19-Δ2382, pK19-Δ2449, pK19-Δ2582, pK19-Δ2709, and pK19-Δ2952 were transformed into GAS355, followed by two-step homologous recombination. The deletions of the corresponding gene fragment were confirmed via colony PCR using primer sets P-C5/P-C6, P-C11/P-C12, P-C17/P-C18, P-C23/P-C24, P-C29/P-C30, P-C35/P-C36, P-C41/P-C42, P-C47/P-C48, P-C53/P-C54, P-C59/P-C60, P-C65/P-C66, P-C71/P-C72, P-C77/P-C78, P-C83/P-C84, P-C89/P-C90, P-C95/P-C96, P-C101/P-C102, P-C107/P-C108, P-C113/P-C114, and P-C119/P-C120, respectively. The resulting mutant strains were named HB-Δ0186, HB-Δ0219, HB-Δ0313, HB-Δ0324, HB-Δ0503, HB-Δ0908, HB-Δ1213, HB-Δ1302, HB-Δ1459, HB-Δ1608, HB-Δ1962, HB-Δ2122, HB-Δ2213, HB-Δ2277, HB-Δ2358, HB-Δ2382, HB-Δ2449, HB-Δ2582, HB-Δ2709, and HB-Δ2952, respectively ( Supplementary Table S1 ).

To restore the pobA gene from the pobA-deleted 4-HBA–producing strain, APS963, pK19-pobA was transformed into APS963 and followed by two-step homologous recombination, yielding the MA183 strain ( Table 1 ). Verification of pobA recovery was confirmed through PCR using P-pobA-CF and P-pobA-CR primers. To gain a strain harboring partial deletion of pcaHG and insertion of ubiC ^pr, pK19-ΔpcaHG::P _sod -ubiC ^pr was transformed into the mother strain, MA183, and finally the MA225 strain was obtained ( Table 1 ). The deletion of the 0.82 kb pcaHG locus and insertion of 1 kb ubiC ^pr were confirmed by using primers P-pca-CF and P-pca-CR. The strain MA303 with partial deletion of the vanAB fragment was constructed by transformation of pK19-ΔvanAB into the mother strain, MA225 ( Table 1 ). The partial deletion of the vanAB fragment was confirmed by PCR using primers P-van-CF and P-van-CR. The NCgl0324-deleted strain, PV-Δ0324, was constructed by the transformation of pK19-Δ0324 into MA303 ( Table 1 ).

E. coli BL21 (DE3) harboring pET-0324 was cultured at 37°C until OD_600nm reached 0.5 in 300 mL LB broth supplemented with 50 mg/L kanamycin and induced with 1 mM isopropyl β-D-1-thiogalactopyranoside. After cultivation at 18°C for 16 h, the cells were harvested by centrifugation at 5,478 x g for 5 min, washed with PBS buffer, and lysed by using Novagen BugBuster. After centrifugation at 25,155 x g for 1 h, the supernatant was loaded onto an affinity column containing Ni-NTA agarose gel (Bio-Works, Uppsala, Sweden) equilibrated with column buffer (20 mM Tris-HCl, 300 mM NaCl, and pH 7.4). The column was washed with the wash buffer (20 mM Tris-HCl, 300 mM NaCl, 10 mM imidazole, and pH 7.4), and then eluted with the elution buffer (20 mM Tris-HCl, 300 mM NaCl, 300 mM imidazole, and pH 7.4). The purified NCgl0324-His tag protein was dialyzed (20 mM potassium phosphate and pH 7.6) and concentrated using a spin column and stored at -72°C until use.

The reaction mixture, 1 mL, for measuring aromatic aldehyde reductase activity of NCgl0324-His tag consisted of 0.1 M potassium phosphate (pH 7.6), 0.4 mM NADPH, and 5 mM aromatic aldehyde. Enzyme reaction was started by adding NADPH at 25°C, stopped by adding 5 μL of 2 N HCl to 200 μL, and neutralized by adding 5 μL of 2 N NaOH. The concentrations of 4-HB alcohol, PC alcohol, and vanillyl alcohol were measured using high-performance liquid chromatography (HPLC) at 280 nm, and then reductase activity for aromatic aldehydes, that is, 4-HB aldehyde, PC aldehyde, and vanillin was defined as the amount of enzyme required to produce micromoles of the corresponding aromatic alcohol per min per milligram of protein. The reaction mixture, 1 mL, for measuring aromatic alcohol dehydrogenase activity of NCgl0324-His tag consisted of 0.1 M potassium phosphate (pH 7.6), 0.4 mM NADP⁺, and 5 mM aromatic alcohol. The enzyme reaction was started by adding NADP⁺ at 25°C, stopped by adding 5 μL of 2 N HCl to 200 μL, and neutralized by adding 5 μL of 2 N NaOH. The concentrations of 4-HB aldehyde, PC aldehyde, and vanillin were measured using HPLC at 280 nm, and then dehydrogenase activity for aromatic alcohols, that is, 4-HB alcohol, PC alcohol, and vanillyl alcohol was defined as the amount of enzyme required to produce micromoles of corresponding aldehyde per min per milligram of protein.

C. glutamicum strains were routinely grown at 32°C in the Luria–Bertani (LB) medium (tryptone 10 g/L, yeast extract 5 g/L, and NaCl 10 g/L), supplemented with 50 mg/L kanamycin and/or 4.5 mg/L chloramphenicol (if necessary). To test the toxic effect of aromatic aldehydes on wild-type C. glutamicum, the CGXII medium with 40 g/L d-glucose and different ranges of 4-HB aldehyde, PC aldehyde, and vanillin were used (Eggeling and Bott, 2005). In order to produce aromatic aldehydes and alcohols in 250-mL baffled flasks, strains grown on LB agar plates for 36–48 h were suspended in saline and inoculated into 25 mL of the flask fermentation medium (Syukur Purwanto et al., 2018), followed by cultivation for 42–48 h with vigorous shaking at 240 rpm. Cell growth was measured by using the spectrophotometer (UV-2550, Shimadzu, Japan) and expressed as CDW (cell dry weight, g/L) by multiplying the measured OD_600nm by 0.25. When the color of the culture broth turned brown, the OD was measured after washing two to three times with distilled water. All the experiments were performed in triplicate, and mean and standard deviation were given.

The concentrations of 4-HBA, 4-HB aldehyde, 4-HB alcohol, PCA, PC aldehyde, PC alcohol, vanillate, vanillin, and vanillyl alcohol were determined using HPLC equipped with a UV detector and an Eclipse XDB-C18 (4.6 × 150 mm, 5 µm, Agilent) column. The analysis of 4-HBA, 4-HB aldehyde, and 4-HB alcohol from C. glutamicum was performed as previously described (Kim et al., 2020). For analysis of PCA, PC aldehyde, PC alcohol, vanillate, vanillin, and vanillyl alcohol, 1% acetic acid and acetonitrile (85:15, v/v) were used as a mobile phase at a flow rate of 0.8 mL/min, and metabolites were detected at 280 nm. To obtain crude extracts from C. glutamicum, the cells were disrupted with glass beads (BioSpec Products, Oklahoma, United States), and the supernatant was obtained by centrifugation at 11,400 x g for 30 min. Ten micrograms of denatured proteins were loaded on 10% gel in the SDS-PAGE (sodium dodecyl sulfate-polyacrylamide gel electrophoresis) experiment. The protein concentration was determined using the Pierce BCA protein assay kit (Thermo Fisher Scientific Inc., Massachusetts;, United States).

Reduction of aromatic aldehydes can be catalyzed by various ADH/aldehyde reductase superfamilies. To date, functionally identified genes and enzymes involved in these reactions are little known in C. glutamicum. We preliminarily considered 19 candidate genes annotated as ADHs, oxidoreductases related to aryl–alcohol dehydrogenases, aldo–keto reductases, and short-chain dehydrogenases from bioinformatics analysis of C. glutamicum ATCC 13032 using UniProt and CoryneRegNet websites ( Supplementary Table S5 ). In addition, six SDR (short-chain dehydrogenase/reductases) and five MDR (medium-chain dehydrogenase/reductase) families harboring the common cofactor binding site “TGXXXGXG” and the zinc-containing ADH signature “GHEX₂GX₅(G,A)X₂(I,V,A,C,S),” respectively, were mined from the C. glutamicum genome database (de Smidt et al., 2008; Kavanagha et al., 2008; Knoll and Pleiss, 2008; Persson et al., 2009). Moreover, since deletion of adhP, fucO, eutG, yjgB, yahK, ybbO, gldA, dkgA, yghA, and/or yqhD in E. coli led to improved accumulation of several aldehydes (Rodriguez and Atsumi, 2012; Kunjapur et al., 2014; Rodriguez and Atsumi, 2014), BLAST search was performed by using these 10 proteins from E. coli as a query within the C. glutamicum genome. Ten proteins, that is, NCgl0219, NCgl0313, NCgl0324, NCgl0503, NCgl1003, NCgl1112, NCgl2053, NCgl2277, NCgl2709, and NCgl2952 were classified as candidates that showed a BLASTP expect value less than E⁻²⁰. Likewise, three ADHs, that is, NCgl0219, NCgl0324, and NCgl2709 were chosen using ADH6-encoded ADH from S. cerevisiae as a query (Hansen et al., 2009). FudC encoded by NCgl0324 from C. glutamicum, which was identified as a dehydrogenase with higher specificity toward furfural, was included (Tsuge et al., 2016). As some genes, such as CGS9114_RS10340, CGS9114_RS09230, CGS9114_RS09375, CGS9114_RS11565, CGS9114_RS06005, and CGS9114_RS01115, were significantly expressed in C. glutamicum S9114 in response to biotransformation of furaldehydes and benzaldehydes (Zhou et al., 2019), and the corresponding proteins within ATCC 13032, that is, NCgl0168, NCgl2277, NCgl0908, NCgl 1962, NCgl1608, and NCgl2709, respectively, were included in the screening. Of the 28 explored genes, NCgl1112 was already deleted in the model GAS355 strain (Kim et al., 2020). Overall, 27 genes were systematically analyzed for identification of enzymes that reduce aromatic aldehydes to corresponding aromatic alcohols in C. glutamicum ( Supplementary Table S5 ).

With the aim of screening genes encoding aromatic alcohol dehydrogenases or aromatic aldehyde reductases (AARs), C. glutamicum GAS355 producing 4-HB alcohol was used as the model strain (Kim et al., 2020; Table 1 and Figure 1A). Upon transformation of GAS355 with the constructed knockout vectors and two-step homologous recombination, 20 gene deletion mutants among selected 27 candidates were successfully constructed, and the deletion of corresponding genes was verified through PCR analysis ( Supplementary Figure S1 ). Using these mutants, production of 4-HB aldehyde and 4-HB alcohol and the ratio of 4-HB aldehyde to 4-HB alcohol were investigated in flask cultures with 80 g/L d-glucose ( Figure 2 ). The parental strain GAS355 produced 2.31 ± 0.28 g/L of 4-HB alcohol as a main product and 0.47 ± 0.09 g/L of 4-HB aldehyde as a minor product, which means endogenous aldehyde reductase(s) readily converted 4-HB aldehyde into 4-HB alcohol. Nineteen mutants mainly produced 4-HB alcohol, but after 48 h of cultivation, the deletion mutants did not accumulate higher titers of 4-HB aldehyde than the control strain. By contrast, the NCgl0324 deletion mutant, HB-Δ0324, accumulated 1.36 ± 0.09 g/L of 4-HB aldehyde as the dominant product (2.9-fold more than the parental strain) and 0.48 ± 0.03 g/L of 4-HB alcohol (4.8-fold less than the control). Thus, in the absence of the NCgl0324 enzyme, less 4-HB aldehyde was converted to 4-HB alcohol. Next, genetic complementation by plasmid-borne expression of NCgl0324 (pXT-0324) in strain HB-Δ0324 was analyzed to exclude effects by non-intentional secondary mutations. It was found that the 4-HB aldehyde titer in flask culture was decreased, while the 4-HB alcohol titer was increased, which indicated that the NCgl0324 enzyme complemented the deletion effect of NCgl0324 ( Figure 2 ). Accordingly, we identified NCgl0324 as the primary gene responsible for conversion of 4-HB aldehyde to 4-HB alcohol in C. glutamicum.

In the previous study, Tsuge et al. (2016) demonstrated that NCgl0324 (designated FudC) is an NADPH-dependent dehydrogenase that preferentially reduces furfural to furfuryl alcohol in C. glutamicum. Since it was unknown if NCgl0324 shows activity toward aromatic aldehydes, His-tagged NCgl0324 protein was purified from E. coli ( Supplementary Figure S2 ), and its specific activity was determined using 4-HB aldehyde, PC aldehyde, or vanillin as a substrate with NADPH as a cofactor. The specific activity of NCgl0324 toward 4-HB aldehyde, PC aldehyde, and vanillin was 2.57 ± 0.14 µmol/min/mg, 3.15 ± 0.29 µmol/min/mg, and 4.59 ± 0.17 µmol/min/mg, respectively ( Table 2 ), representing that NCgl0324 is characterized by aromatic aldehyde reductase (AAR) activity. To determine whether NCgl0324 expresses activity for the reverse oxidation reaction, its activity was measured using aromatic alcohols as substrates and NADP⁺ as a cofactor. The specific activities of NCgl0324 toward 4-HB alcohol, PC alcohol, and vanillyl alcohol were 0.50 ± 0.08 µmol/min/mg, 0.53 ± 0.10 µmol/min/mg, and 0.91 ± 0.06 µmol/min/mg, respectively ( Table 2 ), which exhibit about 19.5, 16.8, and 19.8% activities compared to those toward the corresponding aromatic aldehyde. These results revealed that NCgl0324 has a higher activity for the reduction of aromatic aldehydes than for the oxidation of aromatic alcohols. Thus, our enzyme activity data indicate that NCgl0324 has a broad range of substrates and catalyzes the reduction of aromatic aldehydes to the corresponding aromatic alcohols, along with reduction activity toward furfural.

The finding that NCgl0324 is active as an aromatic aldehyde reductase with PC aldehyde as the substrate prompted us to investigate whether NCgl0324 knockout could block the conversion of this aromatic aldehyde to PC alcohol. Therefore, the PC aldehyde–producing strain was engineered from 4-HBA–producing APS963 (Kim et al., 2020; Figure 1B). First, hydroxylation of 4-HBA to PCA was engineered. Since the hydroxylase gene pobA was deleted in the 4-HBA–producing strain APS963, it was restored using pK19-pobA to yield the strain MA183 ( Supplementary Figure S3A ). We reasoned that blockage of PCA degradation is required for PCA production as PCA is degraded to β-carboxymuconate by protocatechuate dioxygenase encoded by pcaHG. Using pK19-ΔpcaHG,::P _sod -ubiC ^pr, the strain MA225 was constructed through a partial deletion of the pcaHG locus and simultaneous insertion of the mutated ubiC, which is needed to convert chorismate to 4-HBA ( Supplementary Figure S3B ). The strain MA225 produced 4.5 ± 0.36 g/L of PCA with no accumulation of 4-HBA in flask culture after 42 h ( Figure 3 ). In vanillate catabolism of C. glutamicum, the vanAB gene products, vanillate demethylase subunits A and B, convert vanillate to PCA (Merkens et al., 2005). Deletion of vanAB genes in MA225 using pK19-ΔvanAB yielded the strain MA303, which was also used as the base strain for production of the related aromatic aldehyde, vanillin ( Supplementary Figure S3C ).

Next, we aimed to achieve PC aldehyde biosynthesis by reduction of PCA by a plasmid-borne expression of the heterologous car gene encoding the carboxylic acid reductase from Nocardia iowensis ( Figure 1B ). The control strain MA303 produced 4.63 ± 0.69 g/L of PCA, whereas the strain carrying the plasmid pICA4335 for car expression produced 1.33 ± 0.07 g/L of PCA, 0.25 ± 0.03 g/L of PC aldehyde, and 1.69 ± 0.14 g/L of PC alcohol in flask cultures, but no detection of 4-HBA, 4-HB aldehyde, and 4-HB alcohol ( Figure 3 ) was observed. In spite of incomplete conversion of PCA into PC aldehyde and alcohol, it is evident that PC aldehyde was accumulated from PCA by the action of carboxylic acid reductase, followed by accumulation of a high titer of PC alcohol from PC aldehyde by the action of endogenous aldehyde reductase(s). Finally, in order to increase accumulation of PC aldehyde, a PV-Δ0324 strain was constructed by deletion of NCgl0324 from MA303 ( Figure 1B; Table 1 ). In flask culture, the PV-Δ0324 strain with pICA4335 enabled production of 1.18 ± 0.15 g/L of PC aldehyde and 0.22 ± 0.03 g/L of PC alcohol, which corresponded to a 4.8-fold increase of PC aldehyde and a 7.5-fold decrease of PC alcohol in comparison to the NCgl0324-positive control strain carrying pICA4335 ( Figure 3 ). These results revealed that NCgl0324 knockout not only benefitted 4-HB aldehyde production but also led to higher accumulation of PC aldehyde by the engineered C. glutamicum PV-Δ0324.

After having shown the beneficial effect of NCgl0324 knockout on the production of two aromatic aldehydes without the methoxy group, we attempted to produce another important aromatic aldehyde, vanillin, which possesses a methoxy group. To enable vanillin production, an artificial biosynthetic pathway expressing a mutated comt gene encoding murine catechol O-methyltransferase (COMT, Y200L) and the car gene, was introduced ( Figure 1B ). Soluble COMT from Rattus norvegicus is known to catalyze O-methylation of PCA and PC aldehyde to yield vanillate and vanillin, respectively, using S-adenosyl-l-methionine (SAM) as a methyl donor. Moreover, enzyme activity and substrate specificity toward 3′-OH of PCA/PC aldehyde were enhanced by enzyme engineering of COMT, yielding the Y200L mutant (Law et al., 2016). Therefore, a codon-optimized mutant form of the comt ^m gene was synthesized and cloned into the expression vector pCXE50 ( Supplementary Table S4 ). MA225 did not produce vanillate, while cells bearing the resulting pYL200 plasmid yielded 0.12 ± 0.01 g/L of vanillate, which demonstrated the functional expression of comt ^m from R. norvegicus in C. glutamicum ( Table 3 ). To increase comt ^m expression and, as a result, vanillate production, we constructed plasmids pYL230 or pYL250 either by adding an extra sequence to TIR of the tuf promoter or by removing 0.18 kb fragment of the transcriptional terminator region from pYL200, respectively ( Supplementary Figure S4 ). MA225-harboring pYL250 represented a noticeable production of COMT protein compared to cells with pYL200 or pYL230 ( Figure 4 ). Consequently, a vanillate titer of 0.17 ± 0.02 g/L was obtained using MA225 with pYL250, which was 43% higher than that by MA225 with pYL200 ( Table 3 ). It is evident that the deletion of an unnecessary fragment of the terminator yielded increased production of the COMT enzyme and vanillate titer. After having shown successful production of vanillate by the base strain MA225, pYL250 was used to transform the vanAB-deleted strain MA303 and the resultant strain produced 0.24 ± 0.03 g/L of vanillate. This indicated that the blockage of vanillate demethylation is significant for production of vanillate ( Figure 1B ). Upon addition of 0.5 g/L l-methionine to the culture medium in order to increase the availability of SAM for methylation reaction, the highest vanillate product titer of 0.38 ± 0.04 g/L was produced.

With the aim of vanillin and vanillyl alcohol production from PCA-producing strains, plasmids pICA4335 and pYLS2250 for the expressions of car and comt were used to transform strain MA303 and then cultivated for 48 h using 80 g/L d-glucose supplemented with 0.5 g/L l-methionine. If provisions of NADPH and SAM are efficient, PCA can be converted to various products by reaction sequences cascading CAR, COMT, and AARs, generating PC aldehyde (by CAR), PC alcohol (by CAR and AARs), vanillate (by COMT), vanillin (by CAR and COMT), or vanillyl alcohol (by CAR, COMT, and AARs). Consequently, 0.99 ± 0.14 g/L PCA, 0.11 ± 0.01 g/L PC aldehyde, 1.00 ± 0.08 g/L PC alcohol, and 0.11 ± 0.01 g/L vanillyl alcohol were produced after 48 h cultivation in baffled flasks ( Figure 5A ). The finding that vanillin was not detected in the culture broth indicated that vanillin is rapidly reduced to vanillyl alcohol by endogenous AAR(s). Poor methylation of PCA and PC aldehyde may likely explain that the PC alcohol titer was 9.6-fold higher than the titer for vanillyl alcohol. When the NCgl0324-deleted strain PV-Δ0324 was transformed with plasmids for co-expression of car and comt genes, the vanillin titer reached to 0.31 ± 0.03 g/L, while vanillyl alcohol was significantly decreased to 0.03 ± 0.00 g/L ( Figure 5B ). Similarly, PC aldehyde was increased to 0.22 ± 0.03 g/L and PC alcohol was decreased to 0.07 ± 0.01 g/L. Therefore, we demonstrated that NCgl0324 knockout led to higher production of vanillin by recombinant C. glutamicum.

Aromatic aldehydes are known to be detrimental to cell growth, which hampers their production (Zaldivar et al., 1999; Kunjapur et al., 2014). To monitor growth and production profiles of engineered C. glutamicum strains in response to biosynthesis of aromatic aldehydes and alcohols in more detail, flask cultivations were performed using minimal medium with 80 g/L d-glucose for 48 h. As expected, the cultivation of GAS355 showed that 4-HB alcohol production was steadily increased, reaching 2.35 ± 0.26 g/L after 48 h, but 4-HB aldehyde production did not exceed 0.30 ± 0.01 g/L due to the activity of endogenous AAR(s) ( Figure 6B ). Conversely, HB-Δ0324 displayed a gradual increase of 4-HB aldehyde production (1.02 ± 0.05 g/L) but a sharp decrease of 4-HB alcohol production (0.34 ± 0.02 g/L) as a consequence of NCgl0324 gene deletion ( Figure 6C ). Although the production of more toxic 4HB aldehyde in HB-Δ0324 was 3-fold higher than that in GAS355, the strains GAS355 and HB-Δ0324 showed comparable biomass and d-glucose consumption profiles with specific growth rates of 0.28 ± 0.01 /h and 0.27 ± 0.00 /h, respectively ( Figure 6A ).

When MA303 harboring pICA4335 (designated MA-I303 strain) was cultured in baffled flasks, the biomass (cell dry weight, CDW) was rapidly increased up to a maximum of 16.31 ± 0.43 g/L at 30 h with a specific growth rate of 0.29 ± 0.01 /h and then declined to 12.36 ± 1.26 g/L after 48 h ( Figure 6D ). By contrast, the biomass of PV-Δ0324 harboring pICA4335 (designated PV-IΔ0324 strain) was slowly increased to a maximum of 6.30 ± 0.43 g/L at 24 h and declined to 2.7 ± 0.33 g/L after 48 h. The specific growth rate of cells was 0.22 ± 0.01 /h, which was 33% slower than that of MA-I303. The added 80 g/L d-glucose was completely consumed during the 36-h cultivation in MA-I303, whereas 25 g/L d-glucose remained in PV-IΔ0324 during 48 h. PC alcohol and PC aldehyde were accumulated as major products in strains MA-I303 and PV-IΔ0324, respectively. In MA-I303, PC alcohol was rapidly accumulated as a major product (2.12 ± 0.21 g/L) until 36 h and then decreased to 1.69 ± 0.27 g/L after complete depletion of d-glucose ( Figure 6E ). Similar to cultivation of HB-Δ0324, the PC aldehyde titer in PV-IΔ0324 was gradually increased to 1.07 ± 0.18 g/L over 48 h, whereas PC alcohol production was 7.9-fold lower than the production in MA-I303 ( Figure 6F ). Overall, compared to the MA-I303 strain, slower cell growth and d-glucose consumption in PV-IΔ0324 were probably associated with a higher accumulation of the more toxic PC aldehyde.

The production performance of MA303 and PV-Δ0324 strains harboring pICA4335 and pYLS2250 was also evaluated in flasks culture supplemented with 0.5 g/L l-methionine. Cell growth and d-glucose consumption of both strains were significantly delayed during cultures compared to those of 4-HB aldehyde/alcohol or PC-aldehyde/alcohol producing strains ( Figure 6G ). Cultivation of MA303-harboring pICA4335 and pYLS2250 (designated MA-IY303 strain) reduced the specific growth rate (0.208 ± 0.01 /h) and final biomass (6.35 ± 1.01 g/L) during 48 h. In addition, cultivation of PV-Δ0324 harboring pICA4335 and pYLS2250 (designated PV-IYΔ0324 strain) led to significant reduction of the specific growth rate (0.14 ± 0.01 /h) and final biomass (2.55 ± 0.47 g/L) during the 48-h cultivation. At the same time, d-glucose consumption of both strains was also severely retarded by accumulation of various aromatic aldehydes and alcohols, including vanillyl alcohol and vanillin ( Figure 6G ). Even though the accumulated PC alcohol titer (0.95 ± 0.03 g/L) in the MA-IY303 strain was 55% reduced in comparison to that in MA-I303 ( Figure 6H ), d-glucose consumption and cell growth were remarkably impeded, which might be caused by accumulation of toxic vanillyl alcohol (0.11 ± 0.01 g/L). Cultivation of PV-IYΔ0324 displayed production of 0.31 ± 0.02 g/L vanillin, 0.03 ± 0.00 g/L vanillyl alcohol, and 0.22 ± 0.03 g/L PC aldehyde (Figure 6I) and the lowest specific growth rate among tested strains, implying the highest toxicity of vanillin toward C. glutamicum among accumulated aromatic aldehydes and alcohols.