Introduction

1-Deoxynojirimycin (DNJ), a structural analogue of glucose, possesses α-glucosidase inhibitory activity, which creates its anti-diabetic effect. Development of DNJ-rich products show potential in the prevention and treatment of diabetes (Liu et al. 2016). DNJ can be synthesized by Bacillus spp. and Streptomyces species (Kang et al. 2011; Onose et al. 2013; Seo et al. 2013). A DNJ productive strain, Bacillus amyloliquefaciens HZ-12 has been used to prepare DNJ-rich product by solid fermentation of soybeans (Cai et al. 2016). However, a heavily unpleasant odor is formed during solid fermentation which has a serious influence on food sensory quality and environment. Thus the problem of unpleasant odor has to be solved.

Branched-chain short fatty acids (BCFAs), mainly including isobutyric acid and isovaleric acid, are the main odor ingredients in fermented soybeans (Takemura et al. 2000). Biochemically, BCFAs are synthesized by the branched-chain amino acid degradation pathway (Fig. 1), mainly catalyzed by branched-chain amino acid aminotransferases, leucine dehydrogenase, branched-chain α-keto acid dehydrogenase complex, phosphotransbutyrylase, acyl kinase, branched-chain α-keto acid decarboxylase and aldehyde dehydrogenase (Sirobhushanam et al. 2016; Smit et al. 2009). In Bacillus subtilis, deletion of leucine dehydrogenase gene (bcd) could eliminate 99% of BCFAs concentration (Takemura et al. 2000). In Clostridium acetobutylicum and Clostridium tyrobutyricum, phosphate butryltransferase encoded by ptb gene was confirmed to be a rate-limiting enzyme, and knockout of ptb then inhibited butyric acid synthesis (Yu et al. 2011). Herein, the genes of ptb and bcd were deleted from B. amyloliquefaciens HZ-12 by a markerless knockout method, and the concentrations of BCFAs and DNJ were determined to evaluate the gene-deficient candidates comprehensively.

Fig. 1
figure 1

Biosynthesis pathway of BCFAs in bacteria (Smit et al. 2009). Branched-chain amino acids aminotransferase (1), leucine dehydrogenase (2), branched-chain α-keto acid dehydrogenase complex (3), phosphotransbutyrylase (4), acyl kinase (5), branched-chain α-keto acid decarboxylase (6) and aldehyde dehydrogenase (7)

Full size image

Materials and methods

Strains, plasmids, primers and growth media

Strains and plasmids involved in this study are listed in Supplementary Table 1. The primers were designed according to the B. amyloliquefaciens genome sequence (Supplementary Table 2). LB medium (10 g peptone l−1, 5 g yeast extract l−1 and 10 g NaCl l−1) was used for culture of Escherichia coli DH5α and B. amyloliquefaciens HZ-12 (CCTCC M2015234).

Table 1 BCFAs concentrations of different strains
Full size table

Chemicals

T4 DNA ligase, restriction enzymes and DNA marker were from Takara Bio (Dalian, China). TransStart Fast Pfu DNA polymerase was purchased from TransGen Biotech (Beijing, China). All other chemicals were supplied by Sinopharm Chemical Reagent (Shanghai, China).

Construction of knockout plasmids

Gene knockout vectors were constructed according to Qi et al. (2014) and Qiu et al. (2014). Firstly, two homologous arms (about 500 bp) around ptb gene were amplified from the total DNA of B. amyloliquefaciens HZ-12 with primers of ptbF1/ptbR1 and ptbF2/ptbR2. Secondly, two homologous arms were fused through the Splicing-with-Overlapping-Extension PCR method (SOE-PCR) using the primers of ptbF1 and ptbR2. The fused fragment finally was cloned into T2(2)-ori vector at the restriction sites of XbaI and BamHI, further verified through gene sequencing performed by Beijing Genomics institution (China), and named as T2(2)-oriΔptb. The knockout plasmid of T2(2)-oriΔbcd was obtained by the same method.

Construction of gene-deficient strains

Gene knockout was performed by homologous recombination (Qi et al. 2014). Taking ptb gene knockout as an example, B. amyloliquefaciens HZ-12 competent cells were electrotransformed with T2(2)-oriΔptb, then the transformants were picked by kanamycin-resistance screening (20 μg ml−1), verified by PCR with primers of T2F and T2R. Positive transformants were transferred onto LB medium containing 20 μg kanamycin ml−1, and incubated at 45 °C and 180 rpm for 8 h to obtain single-crossover strains with kanamycin resistance, further verified by PCR with Δptb single crossover primers of T2F and ptbYR. Single-crossover colonies were inoculated into LB liquid medium, cultured at 37 °C for 8 h several times, then sprayed onto LB plates to screen the kanamycin-sensitive cells, which were verified by PCR with primers of ptbYF and ptbYR to obtain the ptb knockout strain of HZ-12Δptb. The double gene deficient strain of HZ-12ΔptbΔbcd was constructed by above method based on HZ-12Δptb.

Solid fermentation

Cells of HZ-12, HZ-12Δptb, and HZ-12ΔptbΔbcd were inoculated into 50 ml LB medium, cultured at 37 °C and 180 rpm until the OD600 of broth reached 4.2. Cells were transferred into 50 g sterile soybeans with 60% (v/w) water, cultured at 37 °C for 48 h. Samples were collected, added with distilled water, and rotated at 180 rpm to obtain the extracts, which were used to measure cell density, DNJ and BCFAs concentrations.

Analytical methods

Cells were counted by the plate colony-counting method. The DNJ concentration was determined by HPLC (Yatsunami et al. 2008). BCFAs were determined by GC (Zhou et al. 2016).

Results and discussion

Identification of HZ-12Δptb and HZ-12ΔptbΔbcd

Using the primers of ptbYF and ptbYR, the amplified DNA product from assumed ptb-deficient strain was about 1551 bp, while the DNA fragment amplified from HZ-12 was about 2304 bp, which harbored the ptb gene (Fig. 2a). Moreover, the DNA product amplified from ptb knockout strain was sequenced and analyzed and no other mutation occurred except the deleted ptb gene fragment (data not shown), confirming that the ptb gene was deleted, named as HZ-12Δptb. Deletion of leucine dehydrogenase gene from B. subtilis eliminated 99% of BCFAs concentration during natto fermentation (Takemura et al. 2000). Therefore, the leucine dehydrogenase gene (bcd) was further deleted from HZ-12Δptb. Similarly, the amplified DNA product (about 1730 bp) from assumed ΔptbΔbcd strain was shorter than that of Δptb strain (about 2569 bp), and the difference in DNA length was identical to that of bcd gene (Fig. 2b), indicating that the ΔptbΔbcd strain was constructed.

Fig. 2
figure 2

Confirmation of defective mutants by PCR amplification. a DL5000 marker (M) and PCR products amplified from strain HZ-12 (Lane 1) and HZ-12△ptb (Lane 2) using primers of ptbYF and ptbYR; b DL5000 marker (M) and PCR products amplified from strain HZ-12△ptb (Lane 1) and HZ-12△ptb△bcd (Lane 2) using primers of bcdYF and bcdYR

Full size image

Determination of BCFAs concentrations

Table 1 shows the BCFAs concentrations produced from different strains. 0.82 g isobutyric acid kg−1 and 1.45 g isovaleric acid kg−1 were detected in fermented soybeans with strain HZ-12, which resulted in an unpleasant odor. Similar results have been reported for natto fermented by B. subtilis (Takemura et al. 2000). After knockout of ptb, the total content of BCFAs produced by HZ-12Δptb (1.01 g BCFAs kg−1) was 56% lower than that of HZ-12 (2.27 g BCFAs kg−1). Moreover, no isobutyric acid was detected and the unpleasant smell was decreased significantly. After further deletion of bcd, no isobutyric acid or isovaleric acid was detected from HZ-12ΔptbΔbcd (Table 1), and similar results were reported in fermented soybeans with B. subtilis (Takemura et al. 2000). It is the first report that single knockout of ptb or double deletion of ptb and bcd can decrease BCFAs production using B. amyloliquefaciens. The results confirm that phosphotransbutyrylase and leucine dehydrogenase were key enzymes involved in BCFAs biosynthesis in B. amyloliquefaciens.

DNJ production and cell growth during solid fermentation

DNJ fermentation processes of HZ-12, HZ-12Δptb, and HZ-12ΔptbΔbcd were compared. Deletion of ptb gene showed no significant influence on DNJ yield (Fig. 3a), and the maximum DNJ yield of HZ-12Δptb reached 0.7 g kg−1, which was as high as previous reports (Cho et al. 2008; Paek et al. 1997; Seo et al. 2013). On the other hand, knockout of ptb was beneficial to the cell growth compared with HZ-12 (Fig. 3b). However, double deletion of ptb and bcd significantly reduced the DNJ yield, and the maximum DNJ yield of HZ-12ΔptbΔbcd decreased by 26% compared with that of HZ-12, which might be due to the decreased viable cells (Fig. 3b). To balance the DNJ yield and BCFAs content, HZ-12Δptb is considered to be more suitable for preparation of DNJ-rich products with low odor.

Fig. 3
figure 3

Effects of gene deletion on DNJ production process. Time profiles of DNJ biosynthesis (a) and cell growth (b) from different strains during the fermentation process. Data were expressed as mean ± standard errors of three replicates

Full size image