Biotechnology Letters

, Volume 32, Issue 3, pp 405–411

Production of ethanol from thin stillage by metabolically engineered Escherichia coli


    • Department of Chemical and Biomolecular EngineeringRice University
    • Department of BioengineeringRice University
  • Paul Campbell
    • Glycos Biotechnologies Inc.
  • Matthew Wong
    • Glycos Biotechnologies Inc.
Original Research Paper

DOI: 10.1007/s10529-009-0159-2

Cite this article as:
Gonzalez, R., Campbell, P. & Wong, M. Biotechnol Lett (2010) 32: 405. doi:10.1007/s10529-009-0159-2


Thin stillage is a by-product generated in large amounts during the production of ethanol that is rich in carbon sources like glycerol, glucose and maltose. Unfortunately, the fermentation of thin stillage results in a mixture of organic acids and ethanol and minimum utilization of glycerol, the latter a compound that can represent up to 80% of the available substrates in this stream. We report here the efficient production of ethanol from thin stillage by a metabolically engineered strain of Escherichia coli. Simultaneous utilization of glycerol and sugars was achieved by overexpressing either the fermentative or the respiratory glycerol-utilization pathway. However, amplification of the fermentative pathway (encoded by gldA and dhaKLM) led to more efficient consumption of glycerol and promoted the synthesis of reduced products, including ethanol. A previously constructed strain, EH05, containing mutations that prevented the accumulation of competing by-products (i.e. lactate, acetate, and succinate) and overexpressing the fermentative pathway for glycerol utilization [i.e. strain EH05 (pZSKLMgldA)], efficiently converted thin stillage supplemented with only mineral salts to ethanol at yields close to 85% of the theoretical maximum. Ethanol accounted for about 90% (w/w) of the product mixture. These results, along with the comparable performance of strain EH05 (pZSKLMgldA) in 0.5 and 5 l fermenters, indicate a great potential for the adoption of this process by the biofuels industry.


BiofuelsMetabolic engineeringGlycerol fermentationEscherichia coliThin stillageEthanol


The commercial production of alternative transportation fuels such as ethanol is an important and growing industry. A variety of feedstocks currently used or under development include corn, sugarcane, sugar beets, and lignocellulosic biomass (Sanchez and Cardona 2008). As the scale of production increases, ethanol producers are faced with the growing challenge of improving the yield of ethanol from the same quantity of feedstock while reducing costs associated with the production process. In a typical ethanol production facility, sugars (e.g. glucose) are converted into ethanol, CO2, glycerol, and small amounts of other chemicals. During the distillation process, the ethanol is removed and further processed according to the manufacturer’s need. The remaining material (often referred to as the column bottoms, vinasse, or whole stillage, depending on the feedstock) may be further processed as a means of recovering value from the by-products (Rausch and Belyea 2006). For ethanol production from corn, the whole stillage is primarily utilized as livestock feed, a route that provides ethanol producers with a substantial revenue source and significantly increases the profitability of the process. However, with the growth of ethanol production in recent years, it is imperative to identify new outlets for these by-product streams in order to maintain the economic viability of this industry (Khanal et al. 2008).

The liquid fraction of whole stillage (i.e. thin stillage) contains a complex mixture of chemicals, including low concentrations of various sugars (e.g. maltose, glucose) and up to 2% glycerol (Rausch and Belyea 2006). If all bioethanol plants operating in U.S. were to recover the byproduct glycerol, the amount produced would be similar to the current world production of this compound (Zeng and Biebl 2002; Claude et al. 2000). Although conversion of thin stillage to valuable products via microbial fermentation is highly desirable, the complexity of this mixture poses significant challenges for its efficient use, namely: (i) the inability of industrial organisms such as Saccharomyces cerevisiae to ferment glycerol, a compound that represents a large fraction of the available carbon in thin stillage, and (ii) the catabolite repression exerted by the small amounts of glucose on the consumption of glycerol and other carbon sources. While significant effort has been devoted to avoid the preferential utilization of sugars in the case of sugar mixtures (e.g. lignocellulosic sugars) (Dien et al. 2002; Nichols et al. 2001; Hernandez-Montalvo et al. 2001), no strain has been reported that can simultaneously consume the substrates present in thin stillage. In this work we address the above issues by using an engineered Escherichia coli that efficiently converts the carbon sources present in thin stillage to ethanol. To do so, we took advantage of the knowledge base created by our previous studies on the fermentation of glycerol by E. coli under anaerobic and microaerobic conditions (Durnin et al. 2009; Gonzalez et al. 2008; Murarka et al. 2008; Yazdani and Gonzalez 2007; Dharmadi et al. 2006). A previously engineered strain (Durnin et al. 2009; Yazdani and Gonzalez 2007), containing the genetic modifications shown in Fig. 1, consumed sugars and glycerol simultaneously and produced ethanol as the main fermentation product. Thin stillage required only mineral salts as supplementation and the process was shown to be efficient in 0.5 and 5 l fermenters, all demonstrating a great potential for the successful adoption of this technology by the biofuels industry.
Fig. 1

Summary of pathways involved in the metabolism of glycerol, glucose, and maltose in E. coli. Relevant reactions are represented by the names of the corresponding genes: ackA, acetate kinase; adhE, acetaldehyde/alcohol dehydrogenase; dhaKLM, dihydroxyacetone kinase; fdhF, formate dehydrogenase; frdABCD, fumarate reductase; gldA, glycerol dehydrogenase; glpD, aerobic glycerol-3-P dehydrogenase; glpK, glycerol kinase; hycB-I, hydrogenase; ldhA, lactate dehydrogenase; pflB, pyruvate formate lyase; pta, phosphotransacetylase. Genetic modifications in strain EH05 (pZSKLMgldA) are indicated by thicker lines (overexpression) or double bars (disruption). Broken lines indicate multiple steps. Abbreviations: Gal-3-P glyceraldehyde 3-phosphate, Gly-3-P glycerol 3-phosphate, PEP phosphoenolpyruvate, PYR pyruvate, QH2 reduced quinones


Strains and plasmids

Wild-type E. coli K12 strain MG1655 (F- λ- ilvG- rfb-50 rph-1) was obtained from the University of Wisconsin E. coli Genome Project ( (Kang et al. 2004). Strain EH05 is a derivative of MG1655 previously constructed by inactivating fumarate reductase (ΔfrdA), phosphate acetyltransferase (Δpta), and lactate dehydrogenase (ΔldhA) (Fig. 1) (Durnin et al. 2009). Plasmids pZSKLMgldA (Yazdani and Gonzalez 2007) and pZSglpKglpD (Durnin et al. 2009) were used to overexpress the fermentative and respiratory pathways, respectively, involved on glycerol utilization (Fig. 1). Manufacturer’s protocols (Qiagen, CA, USA) and standard methods (Miller 1992; Sambrook et al. 1989) were followed for DNA purification, plasmid isolation and electroporation. The strains were kept in 32.5% (v/w) glycerol stocks at −80°C. Plates were prepared using LB medium containing 1.5% (w/v) agar with 34 μg chloramphenicol/ml and 50 μg kanamycin/ml.

Culture medium and cultivation conditions

Culture media were prepared by adding mineral salts to autoclaved thin stillage at the concentrations reported by Neidhardt et al. (1974). MOPS was excluded from the formulation when indicated. Thin stillage generated in a dry grind milling production process was kindly provided by White Energy Holding Company, LLC (Dallas, TX) and had the following characteristics: pH 4.4, Brix 7.6, 8.8% solids, 3.5 g maltose/l, 1.5 g glucose/l, and 24.6 g glycerol/l. Chemicals were obtained from Fisher Scientific (Pittsburgh, PA) and Sigma–Aldrich Co.

Fermentations were conducted in either a 0.5 l (working volume) fermentation system (Ward’s Natural Science, Rochester, NY) or a 5 l (working volume) Biostat A+ from Sartorius Stedim North America Inc. (Bohemia, NY). Both systems have independent control of temperature (37°C), pH (6.3), and stirrer speed.

In the 0.5 l fermenter, a pH controller was fitted using 2 M NaOH for pH control. The stirrer speed was maintained at 200 rpm. The temperature was maintained at 37°C. The dissolved O2 concentration was measured with a DO-BTA dissolved O2 sensor. In the 5 l fermenter, the temperature, pH, agitation (100 rpm) and air flow rate were controlled using manufacturer’s software package (Sartorius Stedim North America Inc., Bohemia, NY).

Microaerobic cultures were established by bubbling air at 1 v.v.m. Operation under these conditions resulted in decreasing dissolved O2 concentrations that fell below the detection limits after 2 h cultivation (i.e., undetectable dissolved O2 concentrations during almost the entire course of the fermentation). The volumetric oxygen transfer coefficient (kLa) under the above operational conditions was 4 h−1 in both fermentors.

Prior to use, the cultures (stored as glycerol stocks at −80°C) were streaked onto LB plates and incubated overnight at 37°C. Ten colonies were used to inoculate 250 ml Pyrex bottles containing 175 ml minimal medium supplemented with 10 g glycerol/l. The bottles were incubated at 37°C and shaken at 150 rpm until an OD600 of ~0.3 was reached. An appropriate volume of this actively growing pre-culture was centrifuged, and the pellet was washed and used to inoculate 350 ml medium in each fermenter, with an initial OD600 of 0.05.

Analytical methods and calculation of fermentation parameters

After centrifugation, supernatants were stored at −20°C for HPLC analysis. Glycerol, sugars, organic acids, and ethanol were measured by HPLC as previously described (Dharmadi et al. 2006; Dharmadi and Gonzalez 2005). The transfer of O2 in microaerobic cultures was characterized by measuring the volumetric O2 transfer coefficient (kLa, h−1) as previously described (Durnin et al. 2009). Data for glycerol consumption and product synthesis were used to calculate average product yields (g product/g glycerol) for 72 h fermentations as previously described (Durnin et al. 2009; Yazdani and Gonzalez 2007).

Results and discussion

Substrate utilization and product synthesis during the fermentation of thin stillage by wild-type E. coli

Wild-type E. coli K12 metabolized sugars present in thin stillage but the consumption of glycerol, the main carbon source, was delayed until glucose was depleted from the medium (Fig. 2). Consequently, only small amount of glycerol was consumed after 72 h fermentation. The observed pattern of sequential consumption of sugars and glycerol is due to the repression that more favorable carbon sources such as glucose exert on glycerol, a phenomenon known as carbon catabolite repression (CCR) (Goerke and Stulke 2008).
Fig. 2

Fermentation of thin stillage supplemented with mineral salts by E. coli K12 strain MG1655 (pZSblank) under microaerobic conditions and initial OD600 = 0.05. Concentration of glycerol (open square), maltose (open diamond), glucose (open circle), organic acids (filled circle), and ethanol (filled triangle)

The metabolism of thin stillage generated a mixture of fermentation products, primarily lactate, acetate and succinate with only minor amounts of ethanol (Fig. 2; Table 1). Given the amounts of organic acids generated, these products clearly originated from the metabolism of sugars (e.g. glucose and maltose). Ethanol, on the other hand, represents less than 10% (w/w) of the product mixture and could have been generated from either sugars or glycerol, although glycerol consumption was very low.
Table 1

Product synthesis for 72-h fermentations using various strains of E. coli (see “Methods” and Fig. 1)a


Culture conditionsb


Organic acidsd










MG1655 (pZSBlank)










MG1655 (pZSKLMgldA)










MG1655 (pZSglpKglpD)




















EH05 (pZSKLMgldA)










EH05 (pZSKLMgldA)









EH05 (pZSKLMgldA)









EH05 (pZSKLMgldA)









aThe coefficient of variation (standard deviation/mean * 100) was below 8% in all cases

bCells were cultivated using the medium and conditions described in the “Methods” section. Fermenter working volume (l), inoculum size (OD600), and presence (+) or absence (−) of MOPS are indicated. A kla of 4 h−1 was used in all experiments

cEthanol titer (g/l), yield (g ethanol/g substrate metabolized), and percentage in the product mixture (% w/w) are shown

dLactate (Lac), acetate (Acet) and succinate (Succ) produced (g/l)

Effect of overexpression of glycerol-dissimilating pathways on substrate utilization and product synthesis

In E. coli, CCR of catabolic genes is mediated by the combined action of global and operon-specific regulatory mechanisms. The major players in the global pathway are the signal metabolite cAMP, the transcription activator CRP (cAMP receptor protein), the enzyme adenylate cyclase, and the IIA component of the glucose-specific phosphoenolpyruvate:carbohydrate phosphotransferase system (PTS) (Goerke and Stulke 2008). Several strategies have been reported to eliminate the sequential and inefficient metabolism of sugar mixtures, all of them based on engineering the aforementioned global regulatory pathways or other components of the PTS (Dien et al. 2002; Nichols et al. 2001; Hernandez-Montalvo et al. 2001). In contrast to previously reported strategies for sugar mixtures, we sought to achieve simultaneous consumption of glycerol and sugars present in thin stillage by circumventing operon-specific regulatory mechanisms: i.e. by directly amplifying the pathways involved on glycerol utilization under microaerobic conditions. These pathways were recently characterized (Durnin et al. 2009) and are shown in Fig. 1.

Overexpression of the fermentative pathway (gldA and dhaKLM) eliminated CCR, thus leading to simultaneous utilization of glycerol and sugars (Fig. 3). These results indicate that repression of the gldA and dhaKLM operons contribute to the CCR exerted by glucose (and/or other sugars) on glycerol utilization. It is noteworthy that, although CCR of the dhaKLM operon has been documented (Bachler et al. 2005), there is no report in the literature on the regulation of gldA via CCR. Amplification of the respiratory pathway (glpK and glpD) also led to a derepressed phenotype, although glycerol utilization was slower than in the case of gldAdhaKLM overexpression (Fig. 3).
Fig. 3

Utilization of glycerol (squares), maltose (diamonds), and glucose (circles) in strains overexpressing glycerol-dissimilation pathways. Open and closed symbols represent the results for strains MG1655 (pZSKLMgldA) and MG1655 (pZSglpKglpD), respectively. An initial OD600 = 0.05 was used

The overexpression of glycerol-utilization pathways also had a profound effect in the product mixture generated during the fermentation of thin stillage (Table 1). The effect, however, was specific to the pathway being overexpressed. Amplification of the fermentative pathway (gldA and dhaKLM) led to 6- and 10-fold increase in ethanol yield and titer, respectively (Table 1). The relative proportion of ethanol to other fermentation products also increased by about eight times. Interestingly, lactate and succinate production increased as well, while acetate synthesis decreased (Table 1). Overall, reduced products ethanol, lactate, and succinate accounted for 88% (w/w) of the product mixture (i.e. acetate represented only 12%, w/w). These changes are likely due to the fact that glycerol utilization through the gldAdhaKLM pathway generates reducing equivalents in the form of NADH, which then can be used in the synthesis of ethanol, lactate or succinate (Fig. 1).

Although ethanol production also increased upon overexpression of the respiratory pathway (glpK and glpD), the changes were less pronounced: i.e. 3- and 6-fold increases in ethanol yield and titer, respectively (Table 1). Moreover, the effect on the synthesis of organic acids was opposite to that caused by overexpression of the gldAdhaKLM pathway: i.e. overexpression of glpK–glpD resulted in higher levels of acetate and lower levels of lactate and succinate (Table 1). Overall, acetate accounted for 40% (w/w) of the product mixture, a 3.4-fold increase over that observed in the case of gldAdhaKLM overexpression. Taken together, these results indicate that overexpression of the respiratory pathway favors the synthesis of oxidized products (acetate) while overexpression of the fermentative pathway leads to an increased accumulation of reduced products (ethanol, lactate, succinate).

Elimination of byproducts competing with ethanol

Since lactate, acetate, and succinate are major products of the fermentation of thin stillage (Fig. 2; Table 1), we sought to minimize their accumulation by blocking the metabolic pathways responsible for their synthesis. For this purpose, the genes encoding lactate dehydrogenase (ldhA), phosphotransacetylase (pta), and fumarate reductase (frdA), three key enzymes involved in the synthesis of lactate, acetate, and succinate, respectively (Sawers and Clark 2004; Gonzalez et al. 2008; Murarka et al. 2008) (Fig. 1), were disrupted. In the resulting strain, EH05, the synthesis of lactate was completely eliminated and the production of acetate and succinate significantly reduced (Table 1). While maltose and glucose consumption were reduced, glycerol utilization remained at the same low levels observed in the wild type. Although the yield and proportion of ethanol relative to other products increased significantly in strain EH05, the amount accumulated in the fermentation broth was still small (Table 1). The latter suggests that ethanol synthesis may be limited by the poor utilization of glycerol in wild-type and EH05 strains.

Efficient conversion of thin stillage to ethanol

Given the results discussed in previous sections, we evaluated the performance of an integrated biocatalyst for the production of ethanol: i.e., overexpression of gldA and dhaKLM in strain EH05, the latter a triple mutant containing disruptions of the genes frdA, pta, and ldhA (Fig. 1). Amplification of the gldAdhaKLM pathway was chosen over glpKglpD because of the more positive effect of the former on glycerol utilization and ethanol synthesis (see previous sections). Strain EH05 (pZSKLMgldA) exhibited all the desired characteristics of a biocatalyst to convert thin stillage into ethanol: i.e., simultaneous utilization of glycerol and sugars and synthesis of ethanol as the primary product (Fig. 4; Table 1). Ethanol yield was 0.38 g ethanol/g substrate consumed (i.e. 76% of the theoretical maximum) and this product accounted for 82% of the product mixture.
Fig. 4

Conversion of thin stillage to ethanol by strain EH05 (pZSKLMgldA) in a 0.5 l fermenter using an initial OD600 = 0.05. Data are shown for concentration of glycerol (open square), maltose (open diamond), glucose (open circle), organic acids (filled circle), and ethanol (filled triangle)

The use of an industrial medium is of great relevance for the biocatalyst developed in this work. Although based only on mineral salts, the medium used in our fermentations contains a costly component: MOPS. However, MOPS was omitted from the medium formulations without significantly affecting the performance of strain EH05 (pZSKLMgldA) (Table 1). To demonstrate the feasibility of improving the kinetics of ethanol production from thin stillage, we used a higher inoculum size: 0.5 OD600 instead of 0.05 OD600 in previous cultures. While this cell density is much lower than those used in the industrial production of ethanol, efficient conversion of thin stillage to ethanol was achieved by strain EH05 (pZSKLMgldA) (Fig. 5; Table 1). The ethanol yield increased to 0.42 g ethanol/g substrate consumed (i.e. 84% of the maximum theoretical) and this product accounted for 88% (w/w) of the total product mixture. Finally, to demonstrate the scalability of the process developed here, an experiment was conducted in a 5 l fermenter using an initial inoculum of 0.05 OD600. As shown in Table 1, the performance of strain EH05 (pZSKLMgldA) matched that observed in the smaller fermentation system. Taken together, the results reported here indicate a great potential for the adoption of this process by the biofuels industry.
Fig. 5

Production of ethanol from thin stillage in a 0.5 l fermenter using strain EH05 (pZSKLMgldA), a medium formulation lacking MOPS and an initial OD600 = 0.5. Concentration of glycerol (open square), maltose (open diamond), glucose (open circle), organic acids (filled circle), and ethanol (filled triangle) are shown


This work was partially supported by a grant from the National Science Foundation (CBET-0645188).

Copyright information

© Springer Science+Business Media B.V. 2009