Accepted Manuscript. Propionic acid production in glycerol/glucose co-fermentation by Propionibacterium freudenreichii subsp.

Transcription

1 Accepted Manuscript Propionic acid production in glycerol/glucose co-fermentation by Propionibacterium freudenreichii subsp. shermanii Zhongqiang Wang, Shang-Tian Yang PII: S (13) DOI: Reference: BITE To appear in: Bioresource Technology Received Date: 14 January 2013 Revised Date: 2 March 2013 Accepted Date: 4 March 2013 Please cite this article as: Wang, Z., Yang, S-T., Propionic acid production in glycerol/glucose co-fermentation by Propionibacterium freudenreichii subsp. shermanii, Bioresource Technology (2013), doi: j.biortech This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

2 Propionic acid production in glycerol/glucose co-fermentation by Propionibacterium freudenreichii subsp. shermanii Zhongqiang Wang and Shang-Tian Yang* William G. Lowrie Department of Chemical &Biomolecular Engineering, The Ohio State University, 140 W 19th Ave, Columbus, OH 43210, USA *Corresponding author: phone: ; fax: ; yang.15@osu.edu 1

3 Abstract Propionibacterium freudenreichii subsp. shermanii can ferment glucose and glycerol to propionic acid with acetic and succinic acids as two by-products. Propionic acid production from glucose was relatively fast (0.19 g/l h) but gave low product yield (~0.39 g/g) and selectivity (P/A: ~2.6; P/S: ~4.8). In contrast, glycerol with a more reduced state gave a high propionic acid yield (~0.65 g/g) and selectivity (P/A: ~31; P/S: ~11) but low productivity (0.11 g/l h). On the other hand, co-fermentation of glycerol and glucose at an appropriate mass ratio gave both a high yield ( g/g) and productivity ( g/l h) with high product selectivity (P/A: ~14; P/S: ~10). The carbon flux distributions in the co-fermentation as affected by the ratio of glycerol/glucose were investigated. Finally, co-fermentation with cassava bagasse hydrolysate and crude glycerol in a fibrous-bed bioreactor was demonstrated, providing an efficient way for economic production of bio-based propionic acid. Keywords: Propionibacterium freudenreichii subsp. shermanii; Propionic acid; Glycerol; Cofermentation; Fibrous-bed bioreactor 2

4 1. Introduction Propionic acid is a C3 carboxylic acid with many industrial applications as a specialty chemical and its calcium, potassium and sodium salts are widely used as food and feed preservatives (Boyaval and Corre, 1995). Currently, propionic acid is produced almost exclusively via petrochemical processes, with an annual production capacity of ~400 million lbs in the US. As the crude oil prices had surpassed US$100 per barrel, there have been increasing interests in propionic acid production from renewable bioresources by fermentation using propionibacteria (Feng et al., 2010; Goswami and Srivastava, 2001; Jin and Yang, 1998; Martínez-Campos and de la Torre, 2002; Paik and Glatz, 1994; Rickert et al., 1998; Suwannakham et al., 2006; Wang et al., 2012; Zhang and Yang, 2009ab; Zhu et al., 2012), a group of gram-positive, facultative anaerobic, non-spore forming bacteria that have long been used in the production of Swiss-type cheese and vitamin B 12 (Thierry et al., 2011) and also recently recognized for their probiotic properties for human consumption. However, conventional propionic acid fermentation suffers from low productivity and yield due to strong end-product inhibition and the co-production of other byproducts, mainly acetic and succinic acids. To lower the product cost, recent efforts have focused on using industrial wastes or byproducts as low-cost renewable feedstocks for propionic acid fermentation (Feng et al., 2011; Liang et al., 2012; Zhu et al., 2012). With the fast growth of biodiesel production, a promising substitute for petroleum diesel, a large amount of crude glycerol, about 10% (w/w) of the biodiesel produced, is generated annually, making crude glycerol an economically feasible feedstock for industrial uses (da Silva et al., 2009). Several studies have shown that glycerol can be a good carbon source for propionic acid fermentation with a higher propionic acid yield and much lower acetic acid formation compared to glucose (Barbirato et al., 1997; Coral et al., 2008; Himmi et al., 2000; Ruhal and Choudhury, 2012; Zhang and Yang, 2009b; Zhu et al., 2010). Glycerol has a high reduction 3

5 degree, which favors the production of more reduced metabolites (Ito et al., 2005; Malaviya et al., 2012; Zeng and Biebl, 2002) but can cause redox imbalance in metabolism, leading to reduced cell growth and productivity, when used as the sole carbon source in fermentation (Himmi et al., 2000; Zhang and Yang, 2009b). To overcome this problem, co-fermentation of glycerol with glucose has been proposed as an efficient process supporting both product formation and cell growth (Chen et al., 20; Liu et al., 2011; Xiu et al., 2007). The goal of this study was to evaluate the feasibility of producing propionic acid from crude glycerol present in biodiesel waste and glucose derived from cassava bagasse in a cofermentation process with Propionibacterium freudenreichii subsp. shermanii. Cassava is an important food crop in many Asian and Latin American countries with an annual production of more than 250 million tons in Industrial processing of cassava tuber for starch extraction yielded significant amounts of bagasse, which was usually used as animal feed or disposed into landfills, imposing serious environmental concerns (Pandey et al., 2000). Bioconversion of cassava bagasse has previously been studied for the production of fumaric and lactic acids (Carta et al., 1999; Thongchul et al., 2009), but never for propionic acid. In this study, co-fermentation of cassava bagasse hydrolysate and crude glycerol supplemented with corn steep liquor in a fibrous-bed bioreactor (FBB) was demonstrated as an efficient way for economic production of bio-based propionic acid. The effects of the glycerol/glucose mass ratio on NADH availability and carbon flux distributions in the co-fermentation were also investigated and are reported in this paper. This is the first report about the glycerol/glucose co-fermentation behavior of P. freudenreichii subsp. shermanii, which offers an environmentally friendly and sustainable route for propionic acid production with high product yield and productivity. 4

6 2. Materials and methods 2.1 Culture and media The stock culture of P. freudenreichii subsp. shermanii DSM 4902 (DSMZ, Germany) was cultivated anaerobically at 32 C in NLB medium containing (per liter) 10 g yeast extract, 10 g trypticase soy broth, and 10 g sodium lactate, in serum tubes and stored at 4 C. Unless otherwise noted, fermentation kinetics was studied in a synthetic medium containing (per liter) 10 g yeast extract, 5 g trypticase soy broth, 0.25 g K 2 HPO 4, 0.05g MnSO 4, 20 g CaCO 3, and 30 g carbon source (glucose, glycerol or glycerol/glucose mixture). All media were sparged with nitrogen gas, sealed in serum tubes or bottles, and autoclaved at 121 C for 30 min. 2.2 Preparation of cassava bagasse hydrolysate and crude glycerol as carbon sources Cassava bagasse (CB), which contained about 43% starch, 25% cellulose, 10% hemicellulose and 10% lignin on a dry weight basis, was obtained from a cassava-processing factory in Guangdong, China and was dried and milled to fine powder of μm in diameter. To prepare the CB hydrolysate, 100 g CB powder mixed with 900 ml distilled water in a 2-L flask were autoclaved at 121 C for 30 min. Then, commercial glucoamylase (Distillase L-400, activity: 350 GAU/g, Genencor, NY) at a loading of 0.06 g/g CB (on a dry solids basis) and cellulase (Accellerase1500, endoglucanase activity: CMC U/g, -glucosidase activity: pnpg U/g, Genencor, NY) at 0.1 ml/g CB (on a dry solids basis) were aseptically added into the flask to hydrolyze starch and cellulose, respectively, at 58 C, ph 4.3, 200 rpm for 48 h. HCl was used to adjust ph before enzymatic hydrolysis. After the enzyme treatments, the hydrolysate was centrifuged at 8,000 rpm for 10 min to remove insolubles and the supernatant was stored at 4 C for future use. The CB hydrolysate contained g/l glucose, 0.96 g/l xylose and trace amounts of arabinose and acetic acid. 5

7 Crude glycerol present in biodiesel wastewater from a biodiesel manufacturing plant was prepared together with corn steep liquor (CSL), as nitrogen source, from a corn wet-milling plant. Approximately 90 g crude glycerol solution containing ~40 g glycerol and 60 g CSL were mixed in distilled water to a final volume of 500 ml. The ph of this mixture was adjusted to 6.8 with ammonium hydroxide. After centrifugation at 8000 rpm for 15 min, the aqueous phase between a layer of fatty acids on the top and precipitates on the bottom was collected and autoclaved at 121 C for 30 min. This sterile solution was then aseptically added to the bioreactor containing CB hydrolysate for fermentation kinetics study described later. In addition to crude protein, amino acids and trace elements (metal ions and vitamins), CSL used in this study also contained g/g lactic acid as additional carbon source and trace amount of acetic acid and xylose. 2.3 Batch fermentation Batch fermentations with glucose, glycerol, and glycerol/glucose mixture, respectively, as carbon sources were studied in 125-ml serum bottles and 5-L bioreactors. Each serum bottle containing 50 ml of the medium was inoculated with 2.5 ml of a freshly prepared seed culture (OD 600 ~3.0) in NLB medium in a serum tube. The serum bottle cultures were incubated at 32 o C with ph buffered with 20 g/l CaCO 3, (initial ph 6.8) and samples were withdrawn periodically with 1-ml syringes. After centrifugation, clear broth samples were frozen at -20 C for future analysis. Unless otherwise noted, duplicate bottles were used for each condition studied. Batch fermentations were also carried out in a 5-L stirred-tank fermentor controlled at 32 C, ph 6.5 by adding 6 N NaOH, and agitation at 50 rpm. The fermentor containing ~900 ml of the basic medium without the carbon source and a concentrated substrate (glucose, glycerol, or glycerol/glucose mixture) solution (~50 ml) in a flask were autoclaved at 121 C for 30 min separately and then mixed aseptically in the fermentor. After sparging with N 2 for 45 min to 6

8 anaerobiosis, the fermentor was inoculated with 50 ml of an overnight culture (OD 600 ~2.0). Samples were withdrawn at regular time intervals to monitor cell growth and fermentation kinetics. 2.4 Repeated batch fermentations in a fibrous-bed bioreactor Repeated batch fermentations were studied in a 350-ml fibrous-bed bioreactor (FBB) connected with a recirculation loop to a 5-L stirred-tank fermentor for temperature and ph controls. The FBB was made of a glass column packed with a spirally wound cotton cloth laminated with a corrugated stainless steel wire mesh. Detailed description of the FBB system can be found elsewhere (Suwannakham and Yang, 2005). After 2448 h incubation, the fermentation broth with cells in the 5-L fermentor was recirculated through the FBB for cell immobilization in the fibrous bed for 2436 h until the cell density in the broth no longer decreased. The old broth was then drained and replaced with a fresh medium to allow the cells in the FBB to continue to grow. This process was repeated several times to obtain a stable and high cell density in the reactor system. Then, the fermentation kinetics with glycerol/glucose at a mass ratio of 2 was studied with three consecutive batches, followed with a batch with crude glycerol and CB hydrolysate as substrates. The total liquid volume in each batch was ~1.5 L, including ~350 ml in the FBB. 2.5 Stoichiometric analysis of carbon flux distribution Carbon flux distributions among various metabolites and cell biomass in the metabolic pathway (see Fig. S1 in Supplemental Materials) of propionibacteria were analyzed using a stoichiometric model (see Table 1) and batch fermentation kinetics data from 5-L bioreactor. The model involves 6 reactions for glucose fermentation, 5 reactions for glycerol fermentation and 7 reactions for co-fermentation. The metabolic fluxes were determined based on several 7

9 assumptions. First, pseudo-steady-state hypothesis was applied to the intermediate metabolite, pyruvate. The net formation rate of pyruvate was set to zero so that there was no accumulation during fermentation. Second, the system was NADH balanced so that the production and consumption rates of this reducing co-factor were equal. Third, the system was energy sufficient so that ATP produced in the oxidation of carbon sources and acetate pathway could meet the needs of metabolite and biomass synthesis. The carbon flux distributions were estimated based on the experimental data on substrates (glucose and glycerol) consumption and metabolites production, and the fluxes at the pyruvate node were normalized to show the mole percentage of pyruvate formed or consumed in each branch pathway. 2.6 Analytical methods Cell growth was monitored by measuring the optical density (OD) at 600 nm in a 1.5-ml cuvette using a spectrophotometer (Shimadzu, UV-16-1). Broth samples with suspended cells were diluted to an OD reading of less than 0.8 with distilled water. Glycerol, glucose, and organic acids (acetic, succinic, and propionic acids) were quantified by using high performance liquid chromatography (Shimadzu) with an organic acid analysis column (HPX-87H, Bio-Rad) operated at 45 C with M H 2 SO 4 as the mobile phase at 0.6 ml/min. 3. Results and discussion 3.1 Glucose and glycerol fermentations Batch fermentation kinetics with glucose and glycerol as carbon source, respectively, were studied in serum bottles and 5-l bioreactors. Figure 1 shows batch fermentation kinetics of glucose and glycerol in bioreactor with ph controlled at 6.5. In general, the fermentation was faster with glucose than with glycerol as the substrate, but more propionic acid was produced from glycerol on the same weight basis. Theoretically, one mol glucose produces 4/3 mol 8

10 propionic acid and 2/3 mol acetic acid via the EMP pathway, as shown in the following equation (Playne, 1985). 1.5 Glucose 2 Propionic acid + Acetic acid + CO 2 + H 2 O So the theoretical yield for propionic acid production from glucose is 0.55 g/g. In contrast, one mol glycerol produces one mol propionic acid and no acetic acid via the EMP pathway, with a theoretical propionic acid yield of 0.80 g/g. Glycerol Propionic acid + H 2 O However, the actual propionic acid yield could be lower due to a fraction of the substrate carbon was used for cell biomass or higher if the HMP pathway was used in glycolysis, which was affected by the growth conditions. As expected, glycerol fermentation gave a higher propionic acid yield than that in glucose fermentation (see Tables 2 and 3). Compared to glucose, glycerol with a more reductive state gave a much higher propionic acid/acetic acid (P/A) product ratio for balancing the intracellular NADH/NAD +. Glycerol also gave a higher propionic acid/succinic acid (P/S) ratio than that in glucose fermentation in the bioreactor at ph 6.5, but the P/S ratio was lower in serum bottles without ph control (ph dropped from 6.8 to 4.8) because of stronger propionic acid inhibition at the lower ph resulting in more succinic acid accumulation. It is noted that the propionic acid yield from glucose in serum bottles was higher than that in the bioreactor because cell growth was inhibited in serum bottles, due to the lower ph, and thus more substrate carbon was converted to propionic acid. However, as a result of the lower cell biomass and ph, the propionic acid productivity was also lower in serum bottles. For cells grown in the bioreactor at ph 6.5, the specific growth rate was unexpectedly higher in glycerol fermentation than in glucose fermentation although the final cell density (OD) was lower with glycerol as carbon source. Clearly, P. shermanii can use glycerol to support good cell 9

11 growth. In contrast, P. acidipropionici ATCC 4875 did not grow well with glycerol as the sole carbon source. When 40 g/l glycerol was used as carbon source, glycerol could not be completely used by P. acidipropionici after an extended culturing period and the final OD was only 3, much lower than that in glucose fermentation (Zhang, 2009). Nevertheless, compared to glucose, propionic acid productivity with glycerol as sole carbon source for P. shermanii was still low even though glycerol could support good cell growth with high propionic acid yield and P/A ratio. 3.2 Co-fermentation of glycerol and glucose Co-fermentation of glycerol and glucose at various mass ratios of 1, 2, 3, 4 and 5 was first investigated in serum bottles and the results are summarized and compared in Table 2. In general, increasing the glycerol/glucose mass ratio also increased the ratio of glycerol consumption rate to glucose consumption rate from ~1.0 to 1.9, suggesting that glycerol became increasingly a more favorable substrate than glucose in the co-fermentation, which also increased the propionic acid yield to reach the maximum value of ~0.65 g/g. However, the P/A ratio remained relatively stable at ~6, which was more than 2-fold of that in glucose fermentation but lower than that with glycerol as sole carbon source (9.6). Clearly, glucose as the co-substrate allowed a significant amount of pyruvate to be converted to acetate, which generated more ATP and resulted in faster fermentation while still maintained a high propionic acid yield. Meanwhile, the P/S ratio also increased to 710, which was comparable to that with glucose (9.2) and much higher than that with glycerol (3.1) as sole carbon source. Since the ratio of glycerol consumption rate to glucose consumption rate obtained in serum bottles was between 1 and ~1.9, the glycerol/glucose mass ratios of 1, 1.5, 2, and 3 in the co-fermentation were further studied in 5-L bioreactors at ph 6.5. In general, glycerol was 10

12 consumed faster than glucose at all mass ratios studied (Figure 2A; also see Fig. S2 in Supplemental Materials). However, the fermentation became significantly slower when the glycerol/glucose mass ratio increased to 3, although the specific growth rate was not significantly different at all conditions studied (see Table 3). The mass ratio of 2 gave the highest propionic acid productivity of ~0.23 g/l h, which was higher than that with glucose (0.19 g/ L h) and about double of that with glycerol (0.11 g/l h) as sole carbon source. The propionic acid yield ( g/g) in the co-fermentation was much higher than that of glucose fermentation (0.39 g/g) but lower than that of glycerol fermentation (0.65 g/g), except at the higher mass ratio of 3. Also, both P/A and P/S ratios in the co-fermentation were much higher than those in the glucose fermentation. It is thus clear that the fermentation with glycerol and glucose as co-substrates was advantageous for propionic acid production. Batch fermentation was then studied with crude glycerol, CB hydrolysate and CSL as low-cost carbon and nitrogen sources, with glycerol/(glucose + lactate) mass ratio of ~2. Lactic acid, which has the same reductance degree as glucose, was present in CSL and also used as carbon source by propionibacteria. The results are shown in Figure 2B. In general, the fermentation kinetics was similar to that with 2 glycerol/glucose in the synthetic medium, with slightly higher propionic acid yield and productivity and lower P/S and P/A ratios. The results showed that the propionibacteria used these inexpensive feedstocks as efficiently as the more expensive pure glycerol, glucose, yeast extract and trypticase for propionic acid production. The results also suggested no significant inhibition from impurities present in the crude glycerol as most of the fatty acids and methanol should have been removed during media preparation. 3.3 Effects of co-fermentation on carbon flux distributions 11

13 Propionibacteria use the dicarboxylic acid pathway (see Fig. S1 in Supplemental Materials), in which the substrate or carbon source is first oxidized to pyruvate via. NADH-generating glycolysis pathways. Carbon source with a lower oxidation state, such as glycerol, can generate more NADH for the same amount of pyruvate produced. From pyruvate, two mol NADH are oxidized with the formation of one mol propionic acid, while one mol NADH is produced with the synthesis of one mol acetic acid. Partitioning carbon fluxes between these two pathways renders propionibacteria great flexibility to use a broad spectrum of substrates with various oxidation states to maintain NADH balance. Since glycolysis cannot provide enough NADH for propionic acid production, acetic acid is formed as a compensating metabolite providing extra reducing power to maintain redox balance. Consequently, to produce propionic acid from glucose, which has a lower redox state (reductance degree = 4) than propionic acid (reductance degree = 4.67), requires the co-production of a more oxidized metabolite acetic acid (reductance degree = 4). Therefore, propionic acid production from glucose is tightly coupled with and limited by acetic acid production. In contrast, glycerol with the same redox state as propionic acid (reductance degree = 4.67) has more reducing power than glucose and its conversion to pyruvate yields sufficient NADH for propionic acid biosynthesis without requiring the coproduction of acetic acid to provide additional NADH. When glycerol and glucose were used as co-substrates in propionic acid fermentation, they were consumed simultaneously with glycerol mainly used for propionic acid biosynthesis and glucose as a hydrogen donor substrate for the supply of reducing equivalents and ATP for cell biomass synthesis (Liu et al., 2011). Pyruvate is an important node in propionibacteria metabolic pathways because propionic acid, acetic acid and succinic acid, as well as biomass, are all formed from pyruvate. Metabolic flux analysis was performed to elucidate the carbon flux distributions at this node for different 12

14 fermentation conditions, and the results are shown in Figure 3. The flux distribution was expressed as the percentage of pyruvate formed or consumed in each pathway. For glucose fermentation, more than 90% of pyruvate was obtained through the EMP pathway and less than 10% was through the HMP pathway. In the co-fermentation, glycerol contributed to ~65% of pyruvate while glucose only accounted for ~35% (30% via EMP and 5% via HMP), regardless of the different glycerol/glucose mass ratios (between 1 and 3) in the fermentation (Fig. 3A). For the fluxes from pyruvate to various products, about ~75% went to propionic acid, ~12% to biomass, ~6.5% to succinic acid, and ~6.5% to acetic acid for all co-fermentations (Fig. 3B). The flux to propionic acid was slightly higher at ~80% with glycerol and much lower at ~52% with glucose as sole carbon source, whereas the flux to acetate showed an opposite trend, 3% with glycerol and 25.6% with glucose as sole carbon source. In general, the flux toward cell biomass decreased slightly as more glycerol and less glucose were present in the fermentation, which was consistent with the final cell density obtained in the fermentation. The flux toward succinic acid did not seem to be affected by the carbon substrate used in these fermentations. These results showed that the flux redistribution for redox balance was more robust in the co-fermentation with sufficient acetate and ATP biosynthesis to support good cell growth and faster fermentation as compared to glycerol fermentation. 3.4 Repeated-batch fermentations in the FBB Repeated batch fermentations with glycerol and glucose as co-substrates at 2:1 mass ratio were studied with cells immobilized in a fibrous-bed bioreactor (FBB). After a high cell density had been immobilized in the FBB, three consecutive batches were performed with glycerol and glucose in the synthetic medium followed with a fourth batch with crude glycerol, CB hydrolysate, and CSL as the substrates. The fermentation kinetics is shown in Figure 4. In 13

15 general, similar fermentation kinetics was obtained for all four batches (Fig. 4A), suggesting that the FBB was stable for continued production of propionic acid under the repeated batch mode. In fact, there was a slight increase in both the propionic acid yield (from 0.52 g/g to 0.58 g/g) and volumetric productivity (from 0.44 g/l h to 0.58 g/l h) from the first batch to the third batch (Fig. 4B), a result of increased cell density in the FBB due to continued cell growth and cell adaptation (Liang et al., 2012; Suwannakham et al., 2005; Zhang et al., 2009ab). It is noted that the OD, an indicative of the density of free cells in the fermentation broth, in each batch was significantly lower than that in free-cell fermentation (up to ~10 vs. >15) but the propionic acid productivity was more than two-fold of that in comparable free-cell fermentations because of the higher density of cells immobilized in the FBB. Based on the FBB working volume, the reactor productivity was as high as 2.7 g/l h, which was much higher than that in free-cell fermentation at a comparable propionic acid concentration (Fig. 4C). Comparable propionic acid yield and productivity were obtained with crude glycerol, CB hydrolysate, and CSL as the substrates in the fourth batch, confirming that these low-cost feedstocks can be used efficiently for propionic acid production. 3.5 Comparison to other studies Several studies on propionic acid fermentation with glycerol as sole carbon source or with a cosubstrate have been reported and are summarized in Table 4 for comparison. The highest propionic acid yield of 0.72 g/g from glycerol as sole carbon source was reported with P. acidipropionici ATCC 4875 but the productivity was low, only 0.07 g/l h (Zhang, 2009). Himmi et al. (2000) reported a good propionic acid yield of 0.64 g/g and productivity of 0.42 g/l h from glycerol with P. acidipropionici ATCC Ruhal and Choudhury (2012) reported the production of propionic acid and trehalose from crude glycerol using P. freudenreichii subsp. 14

16 shermanii, achieving a propionic acid yield of 0.42 g/g with a significant amount of lactic acid also produced at a yield of 0.3 g/g. Apparently, the fermentation performance would be species and strain dependent. For the same strain, co-fermentation of glycerol with glucose usually gave a higher productivity although the propionic acid yield would be slightly reduced, as shown in the present study. Recently, Liu et al. (2011) reported propionic acid yield and productivity of 0.57 g/g and 0.15 g/l h, respectively, from glycerol/glucose at a mass ratio of ~2 with P. acidipropionici ATCC In our study with the same species and similar glycerol/glucose mixture as co-substrates, we obtained comparable propionic acid yield but a 50% higher productivity of 0.23 g/l h. A much higher propionic acid productivity of 0.58 g/l h based on total liquid volume (>2.5 g/l h based on reactor working volume) was achieved in the cofermentation with the FBB, demonstrating the advantages of the immobilized-cell fermentation for long-term continuous production of propionic acid in a repeated batch mode. It is noted that the propionic acid yield, productivity, and P/A and P/S ratios could be significantly affected by the medium ph. The pk a values of succinic acid, propionic acid and acetic acid are 5.6, 4.87 and 4.76, respectively. Most of these acids are present in the form of dissociated acids or ions at ph 6.5 while a substantial fraction of them would be present in the undissociated form at or near their pk a values. In general, a lower ph would increase the P/A ratio because of the reduced cell growth and acetate biosynthesis. On the other hand, the P/S ratio was lower in serum bottles without ph control (ph dropped from 6.8 to 4.8) because of stronger propionic acid inhibition at the lower ph resulting in more succinic acid accumulation. Nevertheless, the glycerol/glucose co-fermentation effects on cell growth and propionic acid production were consistent in both the bioreactor with ph controlled at 6.5 and serum bottles without ph control (ph 6.8 to 4.8). It should be noted, however, that the observed co- 15

17 fermentation benefits might be species or even strain dependent as glucose as a co-substrate did not improve propionic acid production from glycerol by P. acidipropionici ATCC 4875 (Zhang, 2009). 4. Conclusions Glucose fermentation produced considerable cell biomass and acetate, leading to a relatively low propionate yield, whereas glycerol fermentation had higher propionate yield and selectivity, but suffered from low productivity. When glycerol and glucose were co-fermented, propionate productivity was greatly improved with higher yield and selectivity comparable to those of the glycerol fermentation. Metabolic flux analysis confirmed that the flux redistribution for redox balance was more robust in the co-fermentation with sufficient acetate and ATP biosynthesis to support cell growth and faster fermentation. Finally, propionate production from crude glycerol, cassava bagasse, and corn steep liquor as low-cost feedstocks was demonstrated. Acknowledgements This study was supported in part by a research grant from The Dow Chemical Company. References 1. Barbirato, F., Chedaille, D., Bories, A. (1997). Propionic acid fermentation from glycerol : comparison with conventional substrate. Appl. Microbiol. Biotechnol.47, Berríos-Rivera SJ, San K-Y, Bennett GN. (2003). The effect of carbon sources and lactate dehydrogenase deletion on 1,2-propanediol production in Escherichia coli. J. Ind. Microbiol. Biotechnol. 30, Boyaval, P., Corre, C. (1995). Production of propionic acid. Lait. 75: Carta, F.S., Soccol, C.R., Ramos, L.P., Fontana, J.D., Production of fumaric acid by fermentation of enzymatic hydrolysates derived from cassava bagasse. Bioresour. Technol. 16

18 68, Chen, X. S., Ren, X.D., Dong, N., Li, S., Li, F., Zhao, F.L., Tang, L., Zhang, J.H., Mao, Z.G. (2012). Culture medium containing glucose and glycerol as a mixed carbon source improves -poly-l-lysine production by Streptomyces sp. M-Z18. Bioprocess. Biosyst. Eng. 35, Coral, J., Karp, S.G., Porto de Souza Vandenberghe, L., Parada, J.L., Pandey, A., Soccol, C.R. (2008). Batch fermentation model of propionic acid production by Propionibacterium acidipropionici in different carbon sources. Appl. Biochem. Biotechnol. 151, da Silva, G. P., Mack, M., Contiero, J. (2009). Glycerol: a promising and abundant carbon source for industrial microbiology. Biotechnol. Adv. 27, Feng, X., Chen, F., Xu, H., Wu, B., Li, H., Li, S., Ouyang, P. (2011). Green and economical production of propionic acid by Propionibacterium freudenreichii CCTCC M in plant fibrous-bed bioreactor. Bioresour. Technol. 102, Feng, X., Xu, H., Yao, J., Li, S., Zhu, H., Ouyang, P. (2010). Kinetic analysis and ph-shift control strategy for propionic acid production with Propionibacterium freudenreichii CCTCC M Appl. Biochem. Biotechnol.160, Goswami, V., Srivastava, A. K. (2001). Propionic acid production in an in situ cell retention bioreactor. Appl. Microbiol. Biotechnol. 56, Himmi, E. H., Bories, A., Boussaid, A., Hassani, L. (2000). Propionic acid fermentation of glycerol and glucose by Propionibacterium acidipropionici and Propionibacterium freudenreichii ssp. shermanii. Appl. Microbiol. Biotechnol.53, Ito, T., Nakashimada, Y., Senba, K., Matsui, T., Nishio, N. (2005). Hydrogen and ethanol production from glycerol-containing wastes discharged after biodiesel manufacturing 17

19 process. J. Biosci. Bioeng.100, Jin, Z., Yang, S.T. (1998). Extractive fermentation for enhanced propionic acid production from lactose by Propionibacterium acidipropionici. Biotechnol. Prog. 14, Liang, Z., Li, L., Li, S., Cai, Y., Yang, S.T., Wang, J. (2012). Enhanced propionic acid production from Jerusalem artichoke hydrolysate by immobilized Propionibacterium acidipropionici in a fibrous-bed bioreactor. Bioprocess. Biosyst. Eng. 35, Liu, Y., Zhang, Y.G., Zhang, R.B., Zhang, F., Zhu, J. (2011). Glycerol/glucose cofermentation: one more proficient process to produce propionic acid by Propionibacterium acidipropionici. Curr. Microbiol. 62, Malaviya, A., Jang, Y.S., Lee, S.Y. (2012). Continuous butanol production with reduced byproducts formation from glycerol by a hyper producing mutant of Clostridium pasteurianum. Appl. Microbiol. Biotechnol. 93, Martínez-Campos, R., de la Torre, M. (2002). Production of propionate by fed-batch fermentation of Propionibacterium acidipropionici using mixed feed of lactate and glucose. Biotechnol. Lett. 24, Paik, H.D. and Glatz, B.A. (1994). Propionic acid production by immobilized cells of a propionate-tolerant strain of Propionibacterium acidipropionici. Appl. Microb. Biotechnol. 42, Pandey, A., Soccol, C.R., Nigam, P., Soccol, V.T., Vandenberghe, L.P.S., Mohan, R. (2000). Biotechnological potential of agro-industrial residues. II: Cassava bagasse. Bioresour. Technol. 74, Papoutsakis, E.T. and Meyer, C.L. (1985). Fermentation equations for propionic acid bacteria and production of assorted oxychemicals from various sugars. Biotechnol. Bioeng. 18

20 27, Rickert, D. A., Glatz, C.E., Glatz, B.A. (1998). Improved organic acid production by calcium alginate-immobilized propionibacteria. Enz. Microb. Technol. 22, Ruhal, R., Choudhury, B. (2012). Use of an osmotically sensitive mutant of Propionibacterium freudenreichii subspp. shermanii for the simultaneous productions of organic acids and trehalose from biodiesel waste based crude glycerol. Bioresour. Technol. 109, San, K.-Y., Bennett, G.N., Berríos-Rivera, S.J., Vadali, R.V., Yang, Y.-T., Horton, E., Rudolph, F.B., Sariyar, B., Blackwood, K. (2002). Metabolic engineering through cofactor manipulation and its effects on metabolic flux redistribution in Escherichia coli. Metab. Eng.4, Suwannakham, S., Yang, S.T. (2005). Enhanced propionic acid fermentation by Propionibacterium acidipropionici mutant obtained by adaptation in a fibrous-bed bioreactor. Biotechnol. Bioeng. 91, Suwannakham, S., Huang, Y., Yang, S.T. (2006). Construction and characterization of ack knock-out mutants of Propionibacterium acidipropionici for enhanced propionic acid fermentation. Biotechnol. Bioeng. 94, Thierry, A., Deutsch, S.M., Falentin, H., Dalmasso, M., Cousin, F.J., Jan, G. (2011). New insights into physiology and metabolism of Propionibacterium freudenreichii. Int. J. Food. Microbiol. 149, Thongchul, N., Navankasattusas S., Yang, S.T., Production of lactic acid and ethanol by Rhizopus oryzae integrated with cassava pulp hydrolysis. Bioprocess Biosyst. Eng. 33,

21 28. Wang, P., Wang, Y., Su, Z. (2012). Microbial production of propionic acid with Propionibacterium freudenreichii using an anion exchanger-based in situ product recovery (ISPR) process with direct and indirect contact of cells. Appl. Biochem. Biotechnol. 166, Xiu, Z. L., Chen, X., Sun, Y.Q., Zhang, D.J. (2007). Stoichiometric analysis and experimental investigation of glycerol-glucose co-fermentation in Klebsiella pneumoniae under microaerobic conditions. Biochem. Eng. J. 33, Zeng, A. P., Biebl, H. (2002). Bulk chemicals from biotechnology: the case of 1,3- propanediol production and the new trends. Adv. Biochem. Eng. Biotechnol. 74, Zhang, A. (2009). "Metabolic engineering and process development for enhanced propionic acid production by Propionibacterium acidipropionici." Dissertation, The Ohio State University. 32. Zhang, A., Yang, S.T. (2009a). Engineering Propionibacterium acidipropionici for enhanced propionic acid tolerance and fermentation. Biotechnol. Bioeng. 104, Zhang, A., Yang, S.T. (2009b). Propionic acid production from glycerol by metabolically engineered Propionibacterium acidipropionici. Process Biochem. 44, Zhu, L., Wei, P., Cai, J., Zhu, X., Wang, Z., Huang, L., Xu, Z. (2012). Improving the productivity of propionic acid with FBB-immobilized cells of an adapted acid-tolerant Propionibacterium acidipropionici. Bioresour. Technol. 112, Zhu, Y., Li, J., Tan, M., Liu, L., Jiang, L., Sun, J., Lee, P., Du, G., Chen, J. (2010). Optimization and scale-up of propionic acid production by propionic acid-tolerant Propionibacterium acidipropionici with glycerol as the carbon source. Bioresour. Technol. 101,

22 Table 1. Stoichiometric equations used in metabolic flux analysis Reaction Stoichiometric equation Glucose oxidation EMP pathway HMP pathway Glucose + 2 ADP + 2 NAD + 2 Pyruvate + 2 ATP +2 NADH (Eq. 1) 3 Glucose + 5 ADP +11 NAD + 5 Pyruvate + 3 CO ATP + 11 NADH (Eq. 2) Glycerol oxidation Glycerol + ADP + 2 NAD + Pyruvate + ATP + 2 NADH (Eq. 3) Organic acids formation from pyruvate Pyruvate + CO NADH Succinate + 2 NAD + (Eq. 4) Pyruvate + ADP + NAD + Acetate + ATP + NADH + CO 2 (Eq. 5) Pyruvate + 2 NADH + ADP Propionate + 2 NAD + + ATP (Eq. 6) Biomass formation 4 Pyruvate NADH ATP Biomass NAD ADP (Eq. 7) Equations originally proposed by Papoutsakis and Meyer (1985) 21

23 Table 2. Kinetics of propionic acid fermentation by P. shermanii in serum bottles. Substrate Yield (g/g) Productivity (g/l h) P/A ratio (g/g) P/S ratio (g/g) Gly/Glu ratio (g/g) Glucose 0.43± ± ± ± Glycerol 0.64± ± ± ± Glycerol/Glucose 0.52± ± ± ± ± Glycerol/Glucose 0.58± ± ± ± ± Glycerol/Glucose 0.61± ± ± ± ± Glycerol/Glucose 0.65± ± ± ± ± Glycerol/Glucose 0.64± ± ± ± ±0.00 Gly/Glu: glycerol consumption rate/glucose consumption rate; The medium was buffered with 20 g/l CaCO 3. During the fermentation, the ph dropped from the initial value of ~6.5 to the final value of ~4.8. The initial total substrate concentration was 30 g/l. Each condition was run in duplicated bottles and the average and standard error are reported. 22

24 Table 3. Kinetics of propionic acid fermentation by P. shermanii in 5-L bioreactor at ph 6.5. Carbon source Propionate yield (g/g) Productivity (g/l h) P/A ratio (g/g) P/S ratio (g/g) Sp. growth rate μ (h -1 ) Glucose 0.390± ± ± ± ±0.002 Glycerol 0.647± ± ± ± ± Glycerol/Glucose 0.537± ± ± ± ± Glycerol/Glucose 0.566± ± ± ± ± Glycerol/Glucose 0.536± ± ± ± ± Glycerol/Glucose 0.645± ± ± ± ± Glycerol/Glucose 0.566± ± ± ± ±0.004 Glucose and glycerol were used as carbon sources in a synthetic medium with an initial total substrate concentration of 30 g/l. The last fermentation was with crude glycerol, CB hydrolysate, and CSL as substrates. For each fermentation, duplicated samples were analyzed and the average and standard error are reported. 23

25 Table 4. Comparison of propionic acid production from glycerol as sole carbon source and glycerol/glucose as co-substrates. Strain Co-fermentation P. acidipropionici ATCC4965 P. acidipropionici ATCC4875(ack) P. shermanii DSM4902 Glycerol P. acidipropionici ATCC4875 P. acidipropionici ATCC4875 (ack) Substrate Propionate yield (g/g) Productivity (g/l h) P/A ratio (g/g) μ (h -1 ) Reference 2 Gly/Glu Liu et al., Gly/Glu Zhang, Gly/Glu Gly/Glu Crude glycerol + CB hydrolysate + CSL This study Glycerol Zhang, > Zhang et al., 2009b P. acidipropionici Coral et al., 2008 ATCC 4965 P. acidipropionici ~ Barbirato et al., 1997 ATCC P. acidipropionici ~ ~ * Himmi et al., 2000 ATCC P. freudenreichii ATCC 9614 ~ ~ * P. shermanii Crude glycerol Ruhal and Choudhury, 2012 *: biomass production rate (g/l h) 24

26 List of Figures Figure 1. Batch fermentation kinetics of P. shermanii with glucose (A) or glycerol (B) as sole carbon source in 5-L bioreactors at ph 6.5, 32 o C. Figure 2. Batch fermentation kinetics of P. shermanii with glycerol/glucose mixture as carbon source at a mass ratio of 2 in synthetic media (A) or with crude glycerol and cassava bagasse hydrolysate as carbon source and corn steep liquor as nitrogen source (B) in a 5- L bioreactor at ph 6.5, 32 o C. Figure 3. Metabolic flux distributions in glucose fermentation, glycerol fermentation, and glycerol/glucose co-fermentation by P. shermanii. Figure 4. Kinetics of repeated-batch fermentations in the FBB; (A) Time course data; (B) Propionic acid yield and productivity; (C) Effects of propionic acid titer on volumetric productivity. Glycerol and glucose at a mass ratio of 2 was used in the first three batches; crude glycerol and CB hydrolysate with corn steep liquor were used in the last batch. 25

27 A B Figure 1 26

28 A B Figure 2 27

29 Glycerol GlucoseEMP A Propionicacid Aceticacid Biomass Succinicacid B Figure 3 28

30 A B C Figure 4 29

31 Highlights Glucose gave considerable cell biomass and acetate, with a low propionate yield Glycerol gave higher propionate yield and selectivity, but low productivity Co-fermentation of glycerol and glucose improved propionate productivity and yield Co-fermentation showed robust flux redistribution for redox balance and cell growth Propionate can be produced from low-cost crude glycerol and cassava bagasse 30