Experimental material

Okara samples were obtained from Nanjing Guoguo Food Co. Ltd. (Nanjing, China). The wet okara was dried at 80 °C to a constant weight and then pulverised using a pulveriser (HangZhou XuZhong Food Machinery Co., Ltd, Hangzhou, China) and filtered through a 40-mesh sieve for later use. Cellulase (130 FPU/g) (Cellic CTec2), β-glucosidase (45 FBG/g) (UltraFLO L), xylanase (500 FXU-S/g) (Shearzyme 500 L), and pectinase (3300 PGNU/g) (Pectinex UF) were purchased from Novozymes China (Beijing) Investment Company. Para-chloro-phenylalanine (p-Cl-Phe; Lot No. SHBC0245V) was purchased from Sigma-Aldrich (Nanjing, China). All analytical reagents were purchased from Nanjing Wanqing Chemical Glass Wear & Instrument Co. Ltd.

Strains and plasmids

The strains and plasmids used in this study are listed in Tables 1. E. coli DH5α and MG1655 were used for plasmid amplification and the source (preparation) of the galactose metabolism gene cluster galKTE, respectively. The host strain B. subtilis 168 and pheS* counter-screening markers were provided by Prof. Yan Xin of Nanjing Agricultural University (Nanjing, China). The E. coli-B. subtilis shuttle plasmid pMA5 was purchased from Wuhan Miaoling Biotechnology Co., Ltd. (Miaoling, Wuhan, China).

Table 1 Strain and plasmid used in studyPlasmid construction and transformation

The primers used in this study are listed in Additional file 1: Table S1. The B. subtilis 168 and E. coli MG1655 genomes were extracted using the TIANamp Bacteria DNA Kit (Tiangen, Beijing, China). The arabinose transporter-encoding gene araE and galactose-related gene cluster galKTE (Gene ID:945358,945357,945354) were amplified from the genomes of B. subtilis 168 and E. coli MG1655, respectively, using 2 × Phanta Max Master Mix DNA polymerase (Vazyme, Nanjing, China). The pMA5 plasmid was digested with restriction enzymes BamHI and MIuI (Takara, Beijing, China), and each gene was ligated into the vector using the ClonExpress Ultra One Step Cloning Kit (Vazyme, Nanjing, China). The recombinant plasmid was transformed into B. subtilis using the Spizizen method (Vojcic et al. 2012).

The knockdown procedure was based on previous work by Zhou et al. (2017) (Fig. 1b). Primers were used to amplify the upstream (~ 800 bp) (LF) and the downstream homologous arm (~ 800 bp) (RF) of the deleted fragment, repeat sequence DR (~ 500 bp), and PC cassette (Pbc-pheS* -cat cassette) (~ 1900 bp). LF, DR, PC, and RF were fused using overlapping PCR. Positive clones were screened on 5 μg/mL erythromycin plates, and subsequently cultured in a non-resistant LB medium until the OD600 reached 1 and coated with sterile water diluted 100-fold onto MGY-Cl (5 g/L glucose, 4 g/L yeast extract, 1 g/L NH4NO3, 0.5 g/L NaCl, 1.5 g/L K2HPO4, 0.5 g/L KH2PO4, 0.2 g/L MgSO4, 5 mM p-Cl-Phe, and 20 g/L agar powder) on solid medium. The knockout strains were verified using PCR and sequencing.

Fig. 1

a Overview of the production of TTMP from a mixture of glucose and galactose by B. subtilis engineered bacteria. Green words represent genes expressed on plasmid pMA5. Red words represent knockout genes. b pheS* counter-screening marker gene knockout process

Preparation of okara hydrolysateEnzymatic saccharification of acid-pretreated okara

The dried okara was treated with 500 mM sulphuric acid for 1.5 h at 121 °C, with a loading rate of 6 g/50 mL. After pretreatment, the pH was adjusted to 4.8 using 3M NaOH, and enzymatic hydrolysis was carried out for 36 h with cellulase loading at 8 FPU/mL, the supernatant was collected after centrifugation for 15 min at room temperature, 6000 rpm, and used for analysis.

Enzymatic hydrolysis of okara

The enzymatic hydrolysis was performed in a 250-mL conical flask with a working volume of 50 mL and a solid-to-liquid ratio of 6 g/50 mL. The pH of the enzymatic hydrolysis system (cellulose, β-glucosidase, xylanase, and pectinase) was controlled between 4.8 and 5.0 using a citrate-sodium citrate buffer (50 mM). The conical flask was placed in a water bath shaker (Henan Bainian Instrument Co., Ltd.) at 50 °C and 150 rpm for 36 h during the reaction. At the end of hydrolysis, the samples were autoclaved at 105 °C for 10 min to inactivate the enzymes. The supernatant was centrifuged (room temperature, 6000 rpm, 15 min) and used for analysis and fermentation. A rotary evaporator was used to concentrate the enzymatic hydrolysate for the bioreactor fermentation scale-up.

Medium and culture conditions

Seed liquid was cultured in LB medium at a volume of 100/500 mL. Fermentation experiments were performed in a 50/250-mL conical flask at 37 °C using M9 medium and acetoin production meduium (carbon source, 15 g/L corn dry power, 3 g/L urea, pH 6.8 after sterilization) for strain fermentation with a rotating speed of 100 rpm.

Batch fermentation of individual sugars

Batch fermentations were carried out in a conical flask with 10 g/L of commercial monosaccharides (glucose, xylose, galactose, and arabinose) as the carbon source, 15 g/L corn dry powder, and 3 g/L urea.

Acetoin production using conical flask fermentation

The effect of an initial reducing sugar concentration in the hydrolysate (with a range of 8 g/L to 41 g/L and supplemented with 1.5% corn dry powder and 0.3% urea) on acetoin production was investigated by conical flask fermentation. Subsequently, optimization of corn dry powder with a concentration range of 0–3% and 29 g/L reducing sugar containing hydrolysate as the carbon source was carried out for acetoin production.

Bioreactor culture

After optimization of the shake flask culture (29 g/L reducing sugar containing hydrolysate, 0.3% urea), batch and fed-batch fermentations were performed in a 7.5-L fermenter with a working volume of 3 L. The pH, temperature, and rotation speed were 6.8, 37 °C, and 400 rpm, respectively. The inoculum volume was 5% (v/v), and the aeration ratio was 1 vvm (air volume/culture volume/min).

TTMP synthesis

(NH4)2HPO4 and acetoin were mixed in a molar ratio of 3:1 and reacted at 105 °C for 3 h to obtain TTMP as previously described (Peng et al. 2020).

Scanning electron microscopy (SEM)

A Hitachi S-4800 (Japan) scanning electron microscope (SEM) was used to photograph the surface morphological changes of okara before and after enzymatic hydrolysis.

Analysis method

The fundamental components of okara, cellulose, hemicellulose, and lignin were determined using the Van Soest method (Syaftika and Matsumura 2018). The crude protein content was determined using the Kjeldahl method as per the Chinese national standard GB 5009.5–2016. Pectin and ash were measured as per the Chinese national standard GB 25533-2010 and GB/T 5505-2008, respectively. Organic acids (succinic, acetic, and lactic acids), acetoin, and 2,3-butanediol were detected by liquid chromatography on an Agilent 1290 Infinity (Agilent Technologies, Waldbronn, Germany) instrument. The column model was an Aminex HPX-87H (Bio-Rad), injection volume was 20 μL, mobile phase was 5 mM H2SO4, flow rate was 0.6 mL/min, column temperature was 60 °C, and a refractive index detector was used (Souza et al. 2021). TTMP was determined as per Peng et al. (2020). The mobile phase included 0.1% formic acid and acetonitrile mixed at a ratio of 8:2 (v/v), the flow rate was 0.8 mL/min, injection volume was 10 μL, column temperature was 40 °C, and column used was an XBridge C18 (5 μm, 4.6 × 250 mm). The detector wavelength was 278 nm. Monosaccharides were determined using a Shodex SUGAR SP-G 6B (6.0 mm I.D. × 50 mm) column; the mobile phase was double distilled water, flow rate was 0.5 mL/min, column temperature was 80 °C, injection volume was 10 μL, and a refractive index detector was used (Ding et al. 2019). Amino acid content was determined using PITC pre-column derivatisation (Hao et al. 2016; Li et al. 2021), and the conditions were as follows: column model Hedera ODS-2 column (4.6 mm × 250 mm, 5 μm, Hanbon Sci. & Tech., Jiangsu, China); UV detector wavelength 254 nm; column temperature 40 °C, flow rate 1 mL/min, injection volume 10 μL; mobile phase A, 0.1 mol/L sodium acetate pH 6.5; mobile phase B, 80% acetonitrile; gradient elution, (5 min 3%B, 14 min 11%B, 17 min 21% B, 29 min 34% B, 41 min 100% B, 43 min 100% A, 47 min 100% A). Biomass was determined by a spectrophotometer (756S, China) at 600 nm. The cell dry weight using the empirical formula 1 OD600 = 0.352 DCW (g/L) (Hu et al. 2020).