Economical D-Lactic Acid Production From Agricultural Residues By Membrane Integrated Continuous Fermentation Coupled With B Vitamin Supplementation


 Background: D-lactic acid played an important role in the establishment of PLA as a substitute for petrochemical plastics. But, so far, the D-lactic acid production was limited in only pilot scale, which was definitely unable to meet the fast growing market demand. To achieve industrial scale D-lactic acid production, the cost-associated problems such as high-cost feedstock, expensive nutrient sources and fermentation technology need to be resolved to establish an economical fermentation process.Results: In the present study, the combined effect of B vitamin supplementation and membrane integrated continuous fermentation on D-lactic acid production from agricultural lignocellulosic biomass by Lactobacillus delbrueckii was investigated. The results indicated the specific addition of vitamins B1, B2, B3 and B5 could reduce the yeast extract (YE) addition from 10 g/l to 3 g/l without obvious influence on fermentation efficiency. By employing cell recycling system in 350 h continuous fermentation with B vitamin supplementation, YE addition was further reduced to 0.5 g/l, which resulted in nutrient source cost reduction of 86%. A maximum D-lactate productivity of 18.56 g/l/h and optical purity of 99.5% were achieved and higher than most recent reports. Conclusion: These findings suggested the novel fermentation strategy proposed could effectively reduce the production cost and improve fermentation efficiency, thus exhibiting great potential in promoting industrial scale D-lactic acid production from lignocellulosic biomass.

Nancib et al. [32] proved the possibility of partially substituting YE with inexpensive nitrogen sources and B vitamins in the study of lactate fermentation from date juice. However, the facts in these studies, such as the relatively low fermentation e ciency and large proportion of YE in mixed nutrient sources, are still a focus for further research.
The fermentation process is another important factor affecting the e ciency of biotechnological D-isomer production [33]. Comparative studies of various operation processes were performed including batch, fed-batch, repeated-batch (semi-continuous) and continuous fermentation by a wide range of homofermentative D-LAB [9]. Continuous techniques have advantage of reducing process costs through constantly feeding nutrients to keep their optimal concentration for cell growth [34,35]. Therefore, the productivity and yield of lactate obtained is signi cantly higher than other fermentation modes [9]. However, some inherent defects in conventional continuous techniques such as the e ux of non-utilized substrates and cell loss still need to be resolved [1]. The e cient membrane-integrated continuous fermentation (MICF) system has been developed and assessed in several experimental studies. The results declare that MICF system obviously achieved high-densities and high D-lactic acid productivity [8,36].
The main objective of this work is to minimize the YE requirement in D-lactic production by means of B vitamin addition and improve the fermentation e ciency through MICF process. First, the in uence of nine B vitamins on cell growth and D-lactic acid fermentation was investigated to screen the essential B vitamins for Lactobacillus delbrueckii. Then, the optimal amount of essential B vitamins and trace element and the in uence of fermentation parameters such as dilution rate were examined. Finally, the membrane-integrated continuous fermentation using rice straw hydrolysate as substrate was conducted under well-optimized nutrient and other experimental conditions to evaluate the effectiveness of suggested fermentation strategy on YE reduction and the improvement of D-lactic acid fermentation e ciency.

Materials
Rice straw (RS) was harvested from a local farm (Dalian, China). The dried RS was milled to ≤2 mm and stored in plastic bags for further pretreatment. A commercial enzyme cocktail (Cellic ® CTec2), acquired from Novozyme Inc. (A/S, USA) was used in enzymatic hydrolysis process. YE purchased from Merck KGaA. B vitamin bought from Sigma.

Pretreatment of rice straw hydrolysate (RSH)
Prewashing treatment and liquid hot water pretreatment (LHWP) The prewashing treatment of the RS was conducted at 10:1 (w/w) water/solid ratio in reactor with agitation rate of 50 rpm for 40 min. After prewashing process, the RS was separated by vacuum ltration and stored at 4 °C for subsequent liquid hot water pretreatment (LHWP).
LHWP of the prewashed RS was conducted in a pilot-scale high-pressure chemical reactor. LHWP was carried out at 185 °C and 50 rpm for 25 min with a 1:10 (w/w) solid/liquid (deionized water) ratio. After pretreatment, the reactor was immediately cooled to below 80 °C. The water-insoluble solids (WIS) and liquid fraction (prehydrolyzate) was then separated by vacuum ltration. The WIS was stored at 4 °C for use in the following experimental runs.
Enzymatic hydrolysis (EH) WIS (6 kg) was subjected to enzymatic hydrolysis by cellulase (Cellic ® CTec2) at dose of 15 FPU/g dry WIS. EH was carried out at 50 °C, 200 rpm for 72 h with 15% (w/w) of WIS loading in 50 mM citrate buffer (pH 5.0). After EH, the liquid fraction was centrifuged at 8000g for10 min, followed by ltration (MF, 0.22 μm/UF, MWCO 100 kDa). Finally, a hybrid membrane ltration process was carried out to simultaneously detoxify and concentrate the RSH [52]. The prepared RSH contained 50.85 g/l glucose, 4.37 g/l xylose and 0.82 g/l arabinose, with no other inhibitors detected.

Microorganism and growth conditions
Lactobacillus delbrueckii subsp delbrueckii (NBRC 3202) from the NITE Biological Resource Centre (NBRC, Japan) was used in this study. For inoculum, the cells were transferred to 100 ml sterile MRS broth then cultured at 40 °C with shaking rate of 130 rpm on a rotary shaker. After 24 h cultivation, the cells were harvested by centrifugation and then washed twice. An inoculation size of 10% (v/v) was used for fermentation.

Continuous fermentation
Continuous fermentation was performed in a BIOSTAT® B-plus 2-l bioreactor (Sartorius, Germany) coupled with a hollow-ber micro ltration module (MICROZA PSP 103, 42 mm × 347 mm; Asahi Kasei, Tokyo, Japan) for cell recycling. Fig. 6 presents a schematic diagram of the membrane-integrated continuous fermentation (MICF) system. The module composed polyethylene hollow-bers (ID, 0.7 mm) with a pore diameter of 0.1 μm and an effective surface area of 0.17 m 2 . The fermentation changed from initial batch mode to continuous mode after complete consumption of glucose with the agitation speed increased from 300 rpm to 800 rpm at 40 °C with N 2 -ow rate of 0.2 l/min. The broth was sustained at 1 l by supplying fresh RSH containing approximately 55 g/l reducing sugar, 38.5 mg/l Mg 2+ , 3 g/l YE and four essential B vitamins with enrichment factor of α = 1.0.
The in uence of dilution rate (D) on fermentation was examined in the range of 0.1 h -1 to 0.7 h -1 . Sampling was performed ve times under each dilution rate after fermentation reached steady state. Fermentation without cell-recycle was simultaneously performed as control.

Analytical methods
Cell growth was measured periodically by monitoring the optical density (OD) at 660 nm using a spectrophotometer (UV-1600, Shimadzu, Japan) and calculating the dry cell weight (DCW) by an appropriate calibration curve. The sugars and metabolites were quanti ed using HPLC (Agilent Technologies 1200 Series, USA) equipped with a refractive index detector using an Aminex HPX-87H column (Bio-Rad, Richmond, CA, USA) as described previously [52]. The optical purity of D-lactic acid was assayed by HPLC using a chiral column (50 mm × 4.6 mm ID; MCI GEL CRS15W; Mitsubishi Chemical Co., Tokyo, Japan) at 254 nm. The optical purity of D-lactic acid was calculated with the following equation: Optical purity (%) = 100 × (D-lactic acid concentration -L-lactic acid concentration)/( D-lactic acid concentration + L-lactic acid concentration).

Results And Discussion
Preliminary experiment: the in uence of different nutrient supplementation on D-lactic acid production The effects of supplementation with different nutrient sources (YE, meat extract (ME), peptone (PEP) and corn steep liquor (CSL)) on D-lactic acid fermentation by L. delbrueckii were investigated in model fermentation broth. The amount of each nutrient source added was equivalent to a nitrogen dose of 10 g/l YE (corresponding to 0.1% nitrogen). The D-lactic acid produced after 18 h fermentation was used for comparison.
The highest D-lactic acid concentration was yielded by YE, con rming that YE was the most appropriate supplement for D-lactate production using L. delbrueckii (data not shown). To determine the optimum YE dose for fermentation, 1, 3, 5, 10 and 20 g/l YE was individually added to the model medium (abbreviated YE1, YE3, YE5, YE10 and YE20), and the in uence of YE amount on cell mass, D-lactic acid productivity and yield and byproduct formation was depicted in Fig. 1. A linear relationship between byproduct and YE dose was observed. The byproduct increased from 0.21 to 3.75 g/l as the initial YE dose varied from 1 to 20 g/l. However, the productivity, yield and cell mass changed logarithmically, the maximum D-lactate yield of 0.95 g/g did not increase at dose higher than YE10, and more YE yielded no obvious enhancement in cell mass and productivity. The observations indicated that high level of YE addition induced higher titer byproduct, which led to inhibition of cell growth and fermentative ability. On the other hand, excess byproduct might cause the additional cost of product recovery and downstream processing. Therefore, the optimal dose of YE for D-lactic acid fermentation by L. delbrueckii was 10 g/l.
Screening essential B vitamin for D-lactic acid production B vitamins are integral nutrient for LAB growth and lactate fermentation. To understand the in uence of individual B vitamins on D-lactic acid production by L. delbrueckii, screening experiments were conducted by means of excluding single B vitamins from the synthetic fermentation medium containing YE3. Nine B vitamins presented in YE were selected, and the individual supplementation amount was equivalent to that of YE10. Moreover, fermentation in medium comprising YE3, YE10 and YE3 coupled with full B vitamins was performed as a control. As Fig. 2 de ned, the maximum and minimum cell mass and D-lactic acid titer was obtained from the control medium of YE10 and YE3, respectively. By supplementing full B vitamins in YE3 medium, the cell mass and D-lactate titer were signi cantly improved to levels close to those of YE10, implying the combined supplementation of YE3 and B vitamins could ful ll the nutritional requirement of L. delbrueckii, consequently overcoming the inhibition of cell growth and fermentation capacities caused by YE reduction, and maintaining the cell mass and D-lactic acid production at a high level. The elimination of vitamins B 6 , B 7 , B 8 , B 9 and B 12 gave no obvious in uence on cell growth and fermentation abilities. The initial glucose of 50.6 g/l was almost consumed after 18 h fermentation with the maximum cell mass of 5.23, 5.17, 5.14 and 5.03 g/l and D-lactic acid titer of 45.5, 45.1, 43.9 and 42.8 g/l, close to that of control medium supplemented with YE3 and full B vitamins (5.27 and 46.3 g/l). In contrast in the case of vitamins B 1 , B 2 , B 3 and B 5 absence, the cell growth was similar to that of control medium with YE3, presenting a signi cant inhibitory pro le. Accordingly, glucose consumption was severely delayed, over 20 g/l glucose left after 18 h fermentation, leading to a considerable reduction in D-lactic acid titers (28.1, 25.3, 26.0 and 23.1 g/l). The observations revealed that VB 1 , VB 2 , VB 3 and VB 5 were essential for L. delbrueckii growth and fermentation, which coincided with the published report showing that the four B vitamins played important roles in the regulatory mechanisms governing LAB growth and fermentation abilities. Klotz et al. [37] reported that VB 5 was a component of coenzyme A (CoA) and acyl-carrier-protein (CAP) involved in protein and fatty acid synthesis of LAB. VB 2 and VB 3 were the precursors of cofactors such as FAD and NAD. NADH directly impacts lactic acid production as a cofactor of lactate dehydrogenase (LDH) [37,38]. In addition, some studies gured out that B vitamin supplementation enhanced the activities of key enzymes of LA biosynthetic pathway in terms of phosphofructokinase (PFK) and L(+)-LDH. VB 1 promoted glycolysis rate while VB 2 accelerated electron transfer, thus increasing the metabolic ux of EMP pathway [38,39].
Effect of trace elements on L. delbrueckii D-lactic acid production It is already known that trace elements have positive effect on lactate fermentation by promoting the cell growth of LAB and regulating the activities of the key enzymes involved in metabolism [40]. To elucidate the in uence of trace elements in terms of Mn 2+ , Mg 2+ , P 3+ and Fe 3+ on L. delbrueckii, the individual constituents were excluded one-by-one in synthetic medium with four B vitamin and YE3 addition. D-lactate accumulation and optical purity had no uctuations in the absence of Mn 2+ , P 3+ and Fe 3+ , but signi cantly decreased in the absence of Mg 2+ , indicating that Mg 2+ was necessary for D-lactate production by L. delbrueckii (Table 1). Moreover, the optimum Mg 2+ concentration for fermentation was also studied. The maximum D-lactate titer of 47.9 g/l and optical purity of 99.5% were achieved under Mg 2+ concentration of 38.5 mg/l. Nevertheless, the relatively high dose of Mg 2+ decreased the lactate titer, which might be contributed to the inhibitory in uence on cell growth.

Optimization of vitamin supplementation in biomass hydrolysate
To determine the optimal dose of four B vitamins for D-lactic acid production by L. delbrueckii, fermentations were carried out in RSH medium with YE3 and varied enrichment factors of B vitamins (α = 0.5, 1.0, 1.5 and 2.0), respectively ( Fig. 3 and Table 2). Fermentation in YE3 medium (α = 0) was conducted as the performance reference. The results showed that the increasing amount of B vitamins continuously stimulated the growth of L. delbrueckii, nally reaching the maximum cell mass of 5.61 g/l at α 2.0. Glucose was unable to be completely converted into lactate at α 0.5 after 18 h fermentation, resulting in a poor concentration of 30.7 g/l. When B vitamin addition was improved to α 1.0, the glucose was fully depleted, achieving the maximum lactate titer, productivity, yield and optical purity of 47.6 g/l, 2.64 g/l/h, 0.95 g/g and 99.5%, respectively. The lactate concentration was enhanced by 134.5% in comparison to that of control medium. Nevertheless, additional B vitamin supplementation led to slight decrease in D-lactate concentration and optical purity. Singhvi et al. [41] reported that lactate dehydrogenase expression in lactic acid producing bacteria was in uenced by nitrogen sources, resulting in the enhancement of D-lactic acid production. Therefore, it was speculated that the increased supplementation of B vitamin in medium might cause the variation of lactate dehydrogenase involved in the isomerization reaction, consequently improving the amount of L-lactic acid during fermentation. However, the hypothesis needed further investigation. In conclusion, α 1.0 was considered as the optimal amount of B vitamin for the subsequent experiments.

Membrane integrated continuous fermentation
Continuous fermentation with and without cell recycling was conducted by RSH using four B vitamin together with YE3 as nutrient sources. To assess the optimal dilution rate for fermentation, the experiments were performed under dilution rates range of 0.1 h −1 to 0.7 h −1 . The concentration of cell mass, residual sugar, D-lactic acid and productivity as a function of dilution rate were depicted in Fig. 4. In the case of MICF system (Fig 4B), the initial glucose was thoroughly converted to D-lactic acid under the dilution rate 0.3 h −1 , reaching D-lactate titer of 46.9 g/l. The cell mass progressively increased with the change of dilution rate and the cell density of 13.45 g/l was attained at the dilution rate of 0.7 h −1 .The maximum productivity of 18.61 g/l/h was achieved when the dilution rate was set at 0.4 h −1 , and then decreased with the increased dilution rate. In the case of fermentation without cell-recycle, the cell mass presented a liner decreasing trend with the variation of dilution rate, suggesting serious cell loss during fermentation ( Fig 4A). Accordingly, more residual sugars and a decrease in lactate concentration were observed. The maximum cell density and productivity were 4.98 g/l and 9.43 g/l/h, which were almost half of those obtained from MICF system. The observations indicated that the speci c growth rate of cells was equal to the dilution rate during the lactate fermentation with cell recycling. In comparison to a free cell continuous fermentation, the membrane integrated fermentation effectively prevented cell loss via medium exchange and maintained cells growth at its maximum rate, thereby greatly enhancing the D-lactic acid concentration and productivity. Moreover, although the increasing dilution rate promoted the progressive cell growth due to the rapid delivery of fresh medium, but the more unconsumed sugar detected in fermenter because the fermentation broth had insu cient time to ferment. Overall, taking account the few residual sugars, high cell density and D-lactic acid concentration and productivity, a dilution rate of 0.4 h −1 was considered applicable for subsequent experiments.
In addition, the optimal nutrient conditions determined in batch fermentation were proven to be suitable for MICF system using RSH as substrate. With the supplementation of YE3 and four B vitamin, the D-lactic acid concentration attained at dilution rate of 0.4 h −1 was 46.4 g/l, slightly higher than that obtained in synthetic medium with YE10 (45.5 g/l) under batch fermentation mode.
A 350 h continuous fermentation with cell recycling was conducted by gradually reducing YE from initial 3 g/l (phase I) to 0.5 g/l (phase III). As Fig. 5 exhibited, the cell mass, D-lactate concentration and productivity presented a sharp increase during the rst 18 h, then D-lactate production and productivity entered a steady state while the cell mass increased until fermentation was completed. Reducing the YE content in mixed nutrient source at different phases gave no in uence on cell growth and lactate accumulation, the average D-lactate titer and productivity obtained under the three phases were almost identical, at approximately 46.6 g/l and 18.56 g/l/h, respectively. Such stable fermentation performance further demonstrated the positive joint effect of B vitamin supplementation and MICF on YE reduction and improvement in fermentation e ciency. The continuous fermentation with cell recycling under appropriate dilution rate could greatly reduce the cell washout and maintain the high cell density by B vitamin supplementation during the long term operation, thereby, the constant lactate titer and productivity could be achieved even with the continuous reduction of YE. In addition, no signs membrane fouling was observed in our fermentation system. The concentration of main byproducts in terms of acetic acid and ethanol examined after fermentation was < 0.7 g/l. The low concentration of byproducts led to a high chemical purity of D-lactate. Table 3 summarized the recently published works on microbial D-lactic acid production from lignocellulosic wastes. Except for this study, all other reports were conducted in batch or fed-batch fermentation mode with different microbes. Regarding the nutrient sources, some agricultural wastes such as barley extract, whey protein hydrolysate, ME, soybean meal, cottonseed meal were used as single nutrients or supplemented with YE or YE and PEP. It was notable that the fermentation e ciencies in these studies were unsatis ed. The highest productivity value of 2.80 g/l/h was reported in the literature of fermentation from pulp mill residue using a mixed nutrient, which was far lower than that of our study of 18.56 g/l/h. Although various studies had been performed using MICF to improve L-lactic acid productivity, its application for D-lactic acid fermentation from lignocellulosic biomass was little. Hence, our experimental results provided important reference information for developing an economically e cient process for Dlactic acid production. Furthermore, the enantiomeric purity of D-lactate obtained in our work was 99.5%, which was preferred since high enantiomeric purities of more than 99.0% were needed to form heat-resistant PLA.
Finally, the YE content in our mixed nutrient source was 3 g/l in batch fermentation and an average of 1.5 g/l in continuous fermentation, far less than other studies, in which at least 5 g/l YE was required. Additionally, the supplementation amount of four B vitamins was minor, 13.8 mg/l in total. The unit prices of YE, VB 1 , VB 2 , VB 3 and VB 5 were $3.5, $36.6, $10.9, $9.1 and $9.3 per kilogram (≥99% purity, www. 1688.com, China). Under batch fermentation, when YE10 was used as nutrient source, 222 kg YE was necessary to produce one ton of D-lactic acid product, adding $777.0 in cost. By substitution of YE10 with YE3 and four B vitamins, the YE cost could be reduced to $233.3; together with the cost of $3.5 of four B vitamins, the total nutrient cost could become $236.8, yielding a saving of 69.5%. During 350 h continuous fermentation with a membrane, the YE content was further reduced to 1 g/l and 0.5 g/l at phase II and phase III, respectively, which resulted in nutrient cost of $109.0, saving up to 86.0%.The results clearly de ned the suggested strategy of four B vitamins supplementation combined with MICF system could signi cantly reduce the nutrient cost in fermentation by L. delbrueckii while also greatly improving the fermentation e ciency.
In conclusion, the research of using inexpensive nutrient sources to substitute YE provided a new insight into economical D-lactic acid production from lignocellulosic feedstock. However, there were still challenges required to be addressed to improve the process. One of the focuses for future work was to develop an effective pretreatment method to obtain the inherent B vitamin in lignocellulosic feedstocks, which would reduce the use of commercial B vitamin and thus further reducing the nutrient cost. The other important aspect was the selection of new D-lactic acid producing bacterial strains with the characteristics of highly productive, substrate and lactic acid resistant as well as capable of consuming all kinds of carbon sources presented in cheap feedstocks. Additionally, the improvement of lactic acid puri cation process e.g. simplifying the operation steps and eliminating the generation of byproduct to achieve production cost reduction was also expected.

Conclusion
The observations indicated that B vitamins were a satisfying partial substitute for YE in D-lactic acid production from RSH by L. delbrueckii. The speci c supplementation of four B vitamins together with YE could ful ll the nutritional requirements of L. delbrueckii. And the adoption of MICF could further reduce the YE amount and greatly improve D-lactate productivity. Hence, the novel fermentation strategy we proposed would be a promising solution for reducing the total manufacturing costs, consequently accelerating the industrial-scale production. b Values in parentheses indicated the concentration of total magnesium content calculated using mg/l as unit.  a Yield refers to initial substrate amount.
b Optical purity of D-lactic acid. Figure 1 The relationships of D-lactic acid productivity, cell density and byproduct with various yeast extract doses in batch fermentation.   Effect of dilution rates (D) on D-lactic acid production during continuous fermentation with or without cell recycling by L. delbrueckii.