The wild-type Z. mobilis ZM4 (ATCC 31821) strain and its derivative strains were cultured in Rich Medium (RM: 10 g/L yeast extract, 1 g/L KH₂PO₄, 1 g/L K₂HPO₄) with 50 g/L glucose (RMG5, pH 5.80) or 120 g/L glucose (RMG12, pH 5.80) at 30 °C, 100 rpm. Escherichia coli DH5α, used for plasmid construction, was grown in Luria-Bertani medium (LB: 10 g/L NaCl, 10 g/L tryptone, 5 g/L yeast extract, and 1.5% agar for solid) at 37°C, 250 rpm. The antibiotics of spectinomycin (100 μg/mL) and chloramphenicol (100 μg/mL) were used for E. coli or Z. mobilis when required, respectively.

The 20% beet molasses (BM) used in this study contains 77.1 g/L sucrose, 4.8 g/L glucose, 6.8 g/L fructose, and 6.3 g/L lactate, pH 5.93. The original corncob residue hydrolysate (CRH) was provided by ZheJiang HuaKang Pharmaceutical Co., Ltd. (Zhejiang, China), which contains 150 g/L glucose, 19.5 g/L xylose, and 2.1 g/L acetic acid, pH 4.35. The CRH used in present study was supplemented with 10 g/L yeast extract, 1 g/L KH₂PO₄ and 1 g/L K₂HPO₄.

A heterologous gene LmldhA (AB233384.1) encoding D-lactate dehydrogenase derived from L. mesenteroides was synthesized from GenScript (Nanjing, China). And the 370-bp sequence in front of the alcohol dehydrogenase gene (ZMO1596) was chose as the constitutive promoter PadhB. Then, LmldhA and PadhB were amplified with the primers of LmldhA-F/R and PadhB-F/R, respectively. The LmldhA driven by PadhB was constructed as a 1366-bp LmldhA-expressing cassette through overlapping extension PCR and further cloned into the shuttle vector pEZ15Asp (Yang et al., 2016) to generate plasmid pEZ-ldh. Lastly, the plasmid pEZ-ldh was electroporated into ZM4 using a Bio-Rad Gene Pulser (Bio-Rad, CA, United States). Electroporated cells were recovered in RMG5 for 3 h and then spread on RMG5 agar plates containing 100 μg/mL spectinomycin for 2-days incubation at 30°C. Transformants were selected by colony PCR with primers of pEZ15A-F/R, and confirmed by Sanger sequencing at Sangon Biotech (Shanghai, China) to obtain recombinant strain ZM4 (pEZ-ldh).

The chromosomal locus of ZMO0038 and ZMO1650 were selected as gene integration location since their deletions do not affect the growth of Z. mobilis (Yang et al., 2016; Qiu et al., 2020). Here, the integration of ZMO0038 with LmldhA was presented as an example. Briefly, the 32-bp spacer sequences (gRNA-0038-F and gRNA-0038-R) were designed and ordered from TsingKe Biotechnology Co., Ltd. (Beijing, China). The oligonucleotides of spacers were annealed and ligated into the linearized pL2R plasmid (Zheng et al., 2019) with Bsa I digestion, and the resulting plasmid was named pL2R-g0038. Subsequently, donor DNA fragments, each containing 800-bp upstream and downstream sequences of ZMO0038 were amplified with the primers of up-0038-F/R and down-0038-F/R. And pL2R-g0038 was amplified with pL2R-FK-F/R for linearization. Then, the donor DNA fragments and LmLdhA-expressing cassette were cloned into linearized pL2R-g0038 by T5 exonuclease (NEB, WA, United States) (Zheng et al., 2019). The resultant ZMO0038-replacing editing plasmid was PCR confirmed with pEZ15A-F/R, and the correct plasmid was named as pRep-0038 (ldh). Similarly, editing plasmids of pRep-1650 (ldh), pRep-1650 (pdc) and pRep-1360 (ldh) were constructed for replacement of ZMO1650 with LmldhA, replacement of ZMO1650 with ZMO1360 (pdc), and replacement of ZMO1360 (pdc) with LmldhA, respectively.

The editing plasmid pRep-0038 (ldh) was then electroporated into Z. mobilis ZM4. Electroporated cells were then spread on RMG5 agar plates containing 100 μg/mL spectinomycin and incubated at 30 °C for 2–3 days. Single colonies were selected based on colony PCR results using the primers of Chk-0038-F/R, and confirmed by Sanger sequencing at Sangon Biotech (Shanghai, China).

Transformants with correct PCR results were cultivated in RMG5 plates and screened using primers pEZ15A-F/R to cure the editing plasmid. The loss of the editing plasmids for recombinant strains ZML, ZML-ldh, ZML-pdc, and ZML-pdc-ldh were further confirmed by colony PCR. All primers used in this work were provided in Supplementary Table S1.

Strains were cultured in a 50 mL flask containing 40 mL RMG5 at 30 °C in an orbital incubator shaker at a speed of 100 rpm with an initial OD_{600 nm} value of 0.1. Batch fermentation was conducted in 1-L bioreactor (T&J Bio-engineering Co., Ltd., Shanghai, China) with 0.6 L RMG5, RMG12, BM, or CRH. The temperature and agitation were set at 30 °C and 100 rpm, respectively. pH was set at a constant value of 5.8 using 4 M KOH during batch fermentation for RM and BM. As for batch fermentation using CRH, 10 g/L CaCO₃ was supplemented in the medium at the beginning for pH adjustment. The initial OD_{600 nm} of 0.1 was used for fermentation using RM, and initial OD_{600 nm} of 0.5 for fermentation using BM and CRH, respectively.

During fermentation, the cultures were sampled at different time points of post-inoculation to monitor cell growth and concentrations of sucrose, glucose, fructose, ethanol, and lactate. Cell growth in terms of its optical density at 600 nm was monitored with a UV–visible spectrophotometer UV-1800 (AoYi Instrument Co., Ltd., Shanghai, China). Samples from the shake flasks or bioreactors were centrifuged at 13,000 rpm for 2 min and then the supernatants were filtered through a 0.2-μm syringe filter into high-performance liquid chromatography (HPLC) vials. Concentrations of sucrose, glucose, fructose, ethanol, and lactate in the supernatants were then detected by HPLC (Shimadzu, Japan) equipped with a refractive index detector (RID) and a column (300 × 7.8 mm) of Bio-Rad Aminex HPX-87H (Hercules, CA, USA) with 5 mM H₂SO₄ as the mobile phase at a flow rate of 0.5 mL/min, column temperature of 65 °C, and an injection volume at 20 μL as previously described (Hu et al., 2021).

The total consumed sugar (C _Total ) was calculated as follows (Hu et al., 2021): C T o t a l = S   *   0.526 + G + S   *   0.526 + F

In this equation, “S” means sucrose consumed, “G” means glucose consumed, and “F” means fructose consumed. “0.526” is the theoretical yield from sucrose into glucose and fructose.

The sugar consumed for ethanol and lactate production are calculated according to the following formulas, respectively: C e t h a n o l = g e t h a n o l / 0.511 C L A = g L A / 1

“0.511” and “1” represent the theoretical yields of ethanol and lactate from glucose, respectively (Liu et al., 2020).

The carbon conversion efficiency (η_c, %) represents the efficiency to convert all sugars into lactate and ethanol, which can be calculated as follows (Lian et al., 2021): η c = C e t h a n o l + C L A /   C T o t a l

Data presented in the graphs were performed with the mean ± SD and T-tests value using the GraphPad Prism statistical software (version 8.0.1). P < 0.05 was considered with statistically significant difference.

LmldhA gene from L. mesenteroides subsp. mesenteroides ATCC 8293 was selected for constructing lactate-producing strain of Z. mobilis in this study, which had a high specific activity when expressed in Saccharomyces cerevisiae (Li et al., 2012; Baek et al., 2016). LmldhA gene driven by the native strong promoter PadhB of Z. mobilis was cloned into the shuttle vector pEZ15Asp to obtain the plasmid pEZ-ldh, which was then transformed into ZM4 to generate ZM4 (pEZ-ldh) (Figure 1). ZM4 (pEZ-ldh) produced 6.0 ± 0.0 g/L D-lactate in medium containing 50 g/L glucose, which was significantly higher than that of 0.5 ± 0.1 g/L lactate in the parental strain ZM4 (Table 1). However, the growth rate of 0.29 ± 0.00 h^-1 in ZM4 (pEZ-ldh) was lower than that of 0.39 ± 0.01 h^-1 in ZM4. The lactate accumulation in recombinant strain ZM4 (pEZ-ldh) may confront with several negative effects, such as the oxidative stress from the accumulated lactate and the acidic medium (Peetermans et al., 2021).

To construct a stable lactate-producing strain, ZMO0038 of Z. mobilis ZM4 was replaced by PadhB-LmldhA, a construct containing LmldhA gene driven by PadhB promoter, using the native type I-F CRISPR-Cas genome editing system to generate a lactate-producing recombinant strain ZML (Figure 1). Our result exhibited that ZML can produce 8.1 ± 0.0 g/L D-lactate in medium containing 50 g/L glucose, which was 1.35 times higher than that of ZM4 (pEZ-ldh) with 6.0 ± 0.0 g/L D-lactate (Table 1). It took ZML 25 h to consume all glucose compared to that of 36 h for ZM4 (pEZ-ldh) (Table 1), which indicated that ZML finished lactate fermentation faster than ZM4 (pEZ-ldh).

To further compete with ethanol production for high lactate production, two recombinant strains were constructed by increasing the copy number of LmldhA gene. ZML (pEZ-ldh) was constructed by introducing the pEZ-ldh plasmid into ZML, and ZML-ldh was constructed by integrating another copy of LmldhA gene to replace the chromosomal gene ZMO1650 (Figure 1). However, ZML (pEZ-ldh) and ZML-ldh produced 7.3 ± 1.3 g/L and 7.0 ± 0.7 g/L D-lactate, which were lower than that of ZML with a titer of 8.1 ± 0.0 g/L D-lactate (Table 1). More importantly, the ethanol production was unaffected. Thus, the carbon flux cannot be diverted to produce lactate by increasing the copy number of heterologous D-lactate dehydrogenase LmldhA gene alone in this study.

We then attempted to redirect carbon flux from ethanol to lactate by replacing the pyruvate decarboxylase gene ZMO1360 (pdc) driven by its strong promoter with an inducible promoter Ptet from the vector Ptet_Dual_Spe (MW812440) (Yang et al., 2019). Pdc gene driven by Ptet was first integrated into the locus of chromosomal gene ZMO1650 to generate the resultant strain ZML-pdc (Figure 1), which produced 6.9 ± 0.5 g/L D-lactate compared to 8.1 ± 0.0 g/L in ZML. Subsequently, pdc gene driven by its native strong promoter was replaced by LmldhA under the control of PadhB promoter to generate the recombinant strain ZML-pdc-ldh (Figure 1). ZML-pdc-ldh produced 13.8 ± 0.2 g/L D-lactate after all 50 g/L glucose was consumed 44 h post-inoculation (Table 1). In addition, less ethanol was produced in ZML-pdc-ldh (16.9 ± 0.3 g/L) compared with ZML (21.0 ± 0.0) and Z. mobilis ZM4 (25.1 ± 0.1 g/L) in flasks, respectively (Table 1). It thus demonstrated that the strategy of diverting the carbon into ethanol was effective with about 20–30% ethanol production reduced for lactate production.

A recombinant strain Zmo-LdhA was reported in a previous study, which can produce 14 g/L lactate in pH-controlled bioreactor when Pdc was repressed in the absence of IPTG and LdhA was induced with 400 nM tetracycline (Liu et al., 2020). In current study, no extra inducer was utilized, and ZML-pdc-ldh can produce 13.8 ± 0.2 g/L D-lactate in flasks without pH control during fermentation. Thus, ZML-pdc-ldh was selected for further experiments.

Compared with the growth rate of 0.29 h^-1 for ZM4 (pEZ-ldh), the growth rate of ZML-pdc-ldh was only 0.18 h^-1, which might be attributed to the acidic environment resulting from lactate accumulation that have been reported in other microorganisms such as Lactobacillus lactis and S. cerevisiae (Yang et al., 2015; Peetermans et al., 2021). The acidic pH environment has an impact on cell functions including DNA and RNA synthesis and many metabolic processes (Warnecke and Gill, 2005; Yang Q. et al., 2020a; Peetermans et al., 2021), and the addition of neutralizer like CaCO₃ to balance the pH is usually applied in lactate fermentation (Yen et al., 2010).

Therefore, the influence of CaCO₃ addition on D-lactate production in strain ZML-pdc-ldh was investigated. First, 10 g/L CaCO₃ was supplemented in the medium at the beginning of the fermentation, and the changes of pH values in the medium of wild-type strain ZM4 and ZML-pdc-ldh were monitored. The pH values of the media in both strains dropped without the addition CaCO₃. And the pH was kept at 4.6 for ZM4, while it sharply declined to pH value lower than 4.0 after 10 h post-inoculation, and continuous to drop to nearly 3.0 at the end of the fermentation for ZML-pdc-ldh (Figure 2A).

With the addition of CaCO₃, the pH of ZML-pdc-ldh medium decreased at first 14 h and then increased from 4.0 to 5.7, which was consistent with the result observed in yeast Candida sonorensis (Ilmén et al., 2013). In addition, the glucose consumption, lactate, and ethanol production were determined during the fermentation process with CaCO₃ addition. When glucose was completely consumed 22 h post-inoculation, 21.5 ± 0.2 g/L D-lactate and 13.6 ± 0.1 g/L ethanol produced (Figure 2B). Compared with the result of 13.8 ± 0.2 g/L D-lactate and 16.9 ± 0.3 g/L ethanol produced in RMG5 without CaCO₃ for 44 h cultivation, it demonstrated that more carbon can be diverted into D-lactate in ZML-pdc-ldh with the addition of CaCO₃ as neutralizing agent to balance pH in culture medium (Figure 2B).

Lactate fermentation in pH-controlled fermenters was further tested for ZML-pdc-ldh. Batch fermentation with different glucose concentration (RMG5: 50 g/L and RMG12: 120 g/L) in 1-L bioreactor at a constant pH of 5.8 was conducted. After glucose was completely consumed within 19 h for RMG5 (Figure 2C) and 34 h for RMG12 (Figure 2D), 24.2 ± 0.6 g/L and 36.2 ± 1.0 g/L D-lactate accumulated, and 12.9 ± 0.8 g/L and 40.3 ± 0.3 g/L ethanol were generated, resulting in a total carbon conversion rate of 98.3% ± 2.5% and 96.2% ± 0.1% with the final product productivity of 1.9 ± 0.0 g/L/h and 2.2 ± 0.0 g/L/h in RMG5 and RMG12, respectively (Table 2).

20% beet molasses (BM) without yeast extract supplementation and sterilization process was conducted anaerobically using ZML-pdc-ldh in a 1-L bioreactor at pH 5.8. As demonstrated in Figure 3A, the concentration of total sugar including sucrose, glucose, and fructose was dramatically decreased at first 45 h from 92.9 ± 0.3 g/L to 24.5 ± 0.8 g/L. After 110 h fermentation, 32.9 ± 0.1 g/L D-lactate and 27.7 ± 0.2 g/L ethanol were produced by ZML-pdc-ldh (Figure 3A). Although the productivity of final product of 0.6 ± 0.0 g/L/h in 20% BM was significantly decreased due to the long time for sugars utilization, a high carbon conversion rate of 97.1% ± 0.0% was achieved (Table 2).

However, it is notable that the time for sugar consumption of ZML-pdc-ldh in 20% molasses significantly increased for more than 5.8 folds or 3.2 folds when compared with the conditions under RMG5 (19 h) or RMG12 (34 h), respectively. It might be ascribed to that sucrose is the main sugar composition of beet molasses, accounting for 81.16%. Sucrose is a disaccharide that needs to be hydrolyzed into monosaccharides of glucose and fructose for utilization during fermentation. More importantly, uptakes of these two sugars are usually competed, and both can be consumed as major substrate and converted to glucose 6-phosphate for Z. mobilis with different fermentation pathways (Palmonari et al., 2020; Braga et al., 2021). Therefore, considering the time for sucrose hydrolysis as well as transportation and catabolism of different sugars, longer time is needed than that of using monosaccharides. Additionally, mixed sugar monomers and oligomers with low quantities, organic acids (i.e., lactic, acetic, butyric, propionic, citric, and malic acids), and other components (i.e., sulfates, phosphates, chlorides, and nitrates) existing in beet molasses may also have negative effects on efficient sugar consumption (Dumbrepatil et al., 2008; Palmonari et al., 2020).

As for corncob residue hydrolysate (CRH) containing 150 g/L glucose, 19.5 g/L xylose, and 2.1 g/L acetic acid, glucose was completely utilized by ZML-pdc-ldh in a 1-L bioreactor after 45 h post-inoculation to produce 42.8 ± 0.0 g/L D-lactate and 53.1 ± 0.7 g/L ethanol (Figure 3B). The total carbon conversion rate of 99.1% ± 0.8% in CRH was the highest among all tested substrates in this study. And the value of 2.1 g/L/h final product productivity in CRH was comparable with the results under glucose fermentation, which was 1.9 g/L/h or 2.2 g/L/h in RMG5 or RMG12, respectively (Table 2).

Moreover, the batch fermentation using CRH by ZML-pdc-ldh was simple and convenient without sterilization and aeration. It also did not need to supplement nutrients such as amino acids, nucleotides and/or vitamins that are usually needed by several lactic acid bacteria for fermentation (Kylä-Nikkilä et al., 2000; Abedi and Hashemi, 2020; Ma et al., 2022). All these results demonstrated that Z. mobilis can utilize corncob residue hydrolysates as suitable carbon resource for economic lactate production for sustainable bioeconomy and environment protection.