Metabolic engineering of a Saccharomyces cerevisiae strain capable ofsimultaneously utilizing glucose and galactose to produce enantiopure(2R,3R)-butanediol

Jiazhang Lian a,b, Ran Chao a,b, Huimin Zhao a,b,c,n

a Department of Chemical and Biomolecular Engineering, University of Illinois at Urbana-Champaign, Urbana, IL 61801, United Statesb Energy Biosciences Institute, Institute for Genomic Biology, University of Illinois at Urbana-Champaign, Urbana, IL 61801, United Statesc Departments of Chemistry, Biochemistry, and Bioengineering, University of Illinois at Urbana-Champaign, Urbana, IL 61801, United States

a r t i c l e i n f o

Article history:Received 25 June 2013Received in revised form21 January 2014Accepted 3 February 2014Available online 10 February 2014

Keywords:(2R,3R)-ButanediolGlucose derepressionSugar co-utilizationMetabolic engineeringPyruvate decarboxylase

a b s t r a c t

2,3-Butanediol (BDO) is an important chemical with broad industrial applications and can be naturallyproduced by many bacteria at high levels. However, the pathogenicity of these native producers is amajor obstacle for large scale production. Here we report the engineering of an industrially friendly host,Saccharomyces cerevisiae, to produce BDO at high titer and yield. By inactivation of pyruvate decarboxy-lases (PDCs) followed by overexpression of MTH1 and adaptive evolution, the resultant yeast grew onglucose as the sole carbon source with ethanol production completely eliminated. Moreover, thepdc- strain consumed glucose and galactose simultaneously, which to our knowledge is unprecedentedin S. cerevisiae strains. Subsequent introduction of a BDO biosynthetic pathway consisting of thecytosolic acetolactate synthase (cytoILV2), Bacillus subtilis acetolactate decarboxylase (BsAlsD), and theendogenous butanediol dehydrogenase (BDH1) resulted in the production of enantiopure (2R,3R)-butanediol (R-BDO). In shake flask fermentation, a yield over 70% of the theoretical value was achieved.Using fed-batch fermentation, more than 100 g/L R-BDO (1100 mM) was synthesized from a mixture ofglucose and galactose, two major carbohydrate components in red algae. The high titer and yield of theenantiopure R-BDO produced as well as the ability to co-ferment glucose and galactose make ourengineered yeast strain a superior host for cost-effective production of bio-based BDO from renewableresources.

& 2014 International Metabolic Engineering Society. Published by Elsevier Inc. All rights reserved.

1. Introduction

Due to increasing concerns on sustainability, energy security,and global warming, intensive research has been devoted tomicrobial production of fuels and chemicals from renewablefeedstocks such as lignocellulose biomass and marine macroalgae(Du et al., 2011; Nielsen et al., 2009). One such example is thebiological production of 2,3-butanediol (BDO), an important che-mical with extensive industrial applications (Celińska and Grajek,2009; Ji et al., 2011; Savakis et al., 2013; Syu, 2001). Due to its lowfreezing point, BDO can be used as an antifreeze agent. A moreimportant potential application is to produce 1,3-butanediene, themonomer of synthetic rubber. In addition, BDO derivatives, such asmethyl ethyl ketone and diacetyl, have found applications in fueland food industries (Celińska and Grajek, 2009). BDO exists inthree isomeric forms, (2S,3S)-butanediol (S-BDO), meso-butanediol

(meso-BDO), and (2R,3R)-butanediol (R-BDO). Many native hostssuch as Klebsiella and Enterobacter species can accumulate BDO tohigh levels. Nevertheless, these native producers are pathogenicand synthesize a mixture of BDO stereoisomers, which preventstheir commercial application (Ji et al., 2011). Therefore, develop-ment of an industrially friendly host for the production of BDOfrom renewable feedstock is highly desirable.

Saccharomyces cerevisiae has been successfully used in mod-ern fermentation industry to produce a wide variety of productsincluding ethanol, organic acids, amino acids, enzymes, andtherapeutic proteins (Chen et al., 2013; Hong and Nielsen, 2012;Runguphan and Keasling, 2014). Various metabolic engineeringstrategies have been applied to redirect the metabolic flux fromethanol to the desired products (Abbott et al., 2009), amongwhich the most straightforward strategy is to delete the alcoholdehydrogenases (ADHs, Fig. 1). Unfortunately, the resultant yeaststrains suffer from accumulation of the toxic intermediates,acetaldehyde and acetate, as well as the residual productionof ethanol as the major product due to the redundancy of ADHsin the S. cerevisiae genome (de Smidt et al., 2012; Ida et al., 2012).A recent effort to produce BDO by inactivating ADHs in

Contents lists available at ScienceDirect

journal homepage: www.elsevier.com/locate/ymben

Metabolic Engineering

http://dx.doi.org/10.1016/j.ymben.2014.02.0031096-7176 & 2014 International Metabolic Engineering Society. Published by Elsevier Inc. All rights reserved.

n Corresponding author at: Department of Chemical and Biomolecular Engineer-ing, University of Illinois at Urbana-Champaign, Urbana, IL 61801, United States.

E-mail address: zhao5@illinois.edu (H. Zhao).

Metabolic Engineering 23 (2014) 92–99

S. cerevisiae led to low productivity and yield (Ng et al., 2012).Another strategy is to inactivate the upstream enzymes, pyruvatedecarboxylases (PDCs, Fig. 1). The pyruvate decarboxylationactivity results from the expression of three structural genes,PDC1, PDC5, and PDC6, whose expression are dependent onthe transcription factor, PDC2 (Hohmann, 1993). However, the pdc-strain (either pdc2- or pdc1-pdc5-pdc6-) is notorious for its inability togrow on glucose as the sole carbon source and the requirement of C2(acetate or ethanol) supplementation to synthesize cytoplasmicacetyl-CoA (Flikweert et al., 1999). Inverse metabolic engineeringrevealed that an internal deletion in the MTH1 coding sequenceenabled the growth of pdc- strain on glucose (Oud et al., 2012; vanMaris et al., 2004). MTH1 is a transcription factor involved in glucosesensing which can inhibit the expression of hexose transporter genes(HXTs), thus the uptake of glucose. The internal deletion increased thestability of MTH1 by removing the putative sites related to proteindegradation (Oud et al., 2012), maintaining the intracellular glucoseconcentration to a low level and alleviating glucose repression.

Recently, the use of marine macroalgae as a renewable feedstockhas attracted increasing attention mainly because the lack of ligninmakes the hydrolysis of seaweed rather simple and straightforward(Wargacki et al., 2012; Wei et al., 2013). Among several differenttypes of marine macroalgae, the red algae, such as Gelidium amansii,are known for high carbohydrate content and abundance in nature.The major carbohydrate components of red algae are glucose andgalactose, which are mainly released from cellulose and agarose,respectively (Wi et al., 2009). To make a biorefinery process basedon a renewable feedstock economically feasible, the producing hostshould be able to utilize all sugars efficiently and simultaneously(Kim et al., 2012b). However, catabolite repression is widely foundin microorganisms, which means glucose is the preferred carbonsource and the utilization of other carbon sources will be inhibiteduntil glucose is depleted. As a result, sequential utilization or diauxic

growth is observed during mixed sugar fermentation, leading to lowyield and productivity of the final products (Vinuselvi et al., 2012). Acellobiose utilization pathway consisting of a cellodextrin transpor-ter and a β-glucosidase was recently introduced into S. cerevisiae toovercome glucose repression, which enabled the consumption ofcellobiose and galactose simultaneously (Ha et al., 2011). Never-theless, the cellobiose consumption rate is much slower than that ofglucose even after intensive pathway engineering (Du et al., 2012;Eriksen et al., 2013; Ha et al., 2013; Yuan and Zhao, 2013), makingco-utilization of glucose and galactose still a preferred process.

In the present study, S. cerevisiae was engineered to produceenantiopure R-BDO at high titer and yield. To redirect the fluxfrom ethanol, three PDC structural genes (PDC1-PDC5-PDC6) weredeleted and growth on glucose was restored through the intro-duction of the internal truncated MTH1 (MTH1T) followed byadaptive evolution. In the presence of a BDO biosynthetic path-way, efficient production of BDO was achieved by completelyeliminating ethanol production (Fig. 1). Moreover, the engineeredpdc- strain showed glucose-derepressed properties, enablingco-fermentation of glucose and galactose, two major carbohydratecomponents in red algae. Notably, this strain has the potential tobe further engineered for efficient and simultaneous utilization ofa mixture of sugars derived from other renewable resources, suchas brown macroalgae and cellulosic biomass, for cost-effectiveproduction of fuels and chemicals.

2. Materials and methods

2.1. Strains, media, and cultivation conditions

All engineered strains used in this study are based on S. cerevisiaeCEN.PK2-1C strain. Escherichia coli strain DH5αwas used to maintain

Fig. 1. Metabolic pathway for the synthesis of (2R,3R)-butanediol in the engineered Saccharomyces cerevisiae strain. The dashed arrows indicate several steps of enzymaticreactions. Genes in red represent the endogenous genes, which lead to the synthesis of major metabolites, such as ethanol and glycerol, and are targets to be inactivated toredirect the flux to BDO. As for the overexpressed BDO pathway, green genes are cloned from S. cerevisiae, while the blue one indicates the heterologous AlsD gene. (Forinterpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

J. Lian et al. / Metabolic Engineering 23 (2014) 92–99 93

and amplify plasmids. Yeast strains were cultivated in complex (YP)medium consisting of 2% peptone and 1% yeast extract supplementedwith the corresponding carbon source, such as glucose, galactose,acetate, and/or ethanol. Recombinant strains were grown on com-plete synthetic (SC) medium consisted of 0.17% yeast nitrogen base,0.5% ammonium sulfate, and the appropriate amino acid drop outmix (CSM-HIS, CSM-URA, or CSM-HIS-URA, MP Biomedicals, Solon,OH), supplemented with the corresponding carbon source. E. colistrains were cultured at 37 1C in Luria-Bertani broth containing100 μg/mL ampicillin. S. cerevisiae strains were cultured at 30 1Cand 250 rpm for aerobic growth, and 30 1C and 100 rpm for oxygenlimited fermentation. For anaerobic fermentation, anaerobic culturetubes with butyl rubber stoppers and aluminum seals (ChemglassLife Sciences, Vineland, NJ) were vacuumed and purged with nitro-gen to remove the residual oxygen and 420 mg/L Tween-80 and10 mg/L ergosterol were supplemented as anaerobic growth factors.All restriction enzymes, Q5 polymerase, and the E. coli–S. cerevisiaeshuttle plasmids, including pRS423 and pRS426, were purchasedfrom New England Biolabs (Ipswich, MA). All chemicals werepurchased from either Sigma-Aldrich or Fisher Scientific.

2.2. DNA manipulation

The yeast homologous recombination based DNA assemblermethod was used to construct the recombinant plasmids (Shaoet al., 2009). Briefly, DNA fragments sharing homologous regions toadjacent DNA fragments were generated via polymerase chainreaction (PCR). The desired DNA fragments were purified with aQIAquick Gel Extraction Kit (Qiagen, Valencia, CA) and co-transformed into S. cerevisiae along with the linearized backboneto assemble several elements in a single step manner. Oligonucleo-tides used in this study were listed in Supplementary Table S1. TheMTH1 gene was amplified from the genomic DNA of CEN.PK2-1Cstrain. To construct the internal truncated MTH1 (MTH1T) (Oudet al., 2012), the N-terminal fragment (1–168 bp) and C-terminalfragment (394–1302 bp) were amplified separately and piecedtogether using the DNA assembler method. For the construction ofthe BDO pathway (Fig. 1), the endogenous acetolactate synthasegene without the mitochondrial targeting sequence (cytoILV2,271–2064 bp, predicted by MITOPROT (Claros and Vincens, 1996)),the Bacillus subtilis acetolactate decarboxylase gene (BsAlsD, NCBIAccession no. NP_391481.1), and the endogenous butanediol dehy-drogenase (BDH1) gene were PCR-amplified from their correspond-ing genomic DNAs and assembled into the 2μ multiple copyplasmid, pRS426. Wizard Genomic DNA Purification Kit (Promega,Madison, WI) was used to extract the genomic DNA from bothbacteria and yeast, according to the manufacturer's protocol. Toconfirm the correct clones, yeast plasmids were isolated using aZymoprep Yeast Plasmid Miniprep II Kit (Zymo Research, Irvine, CA)and re-transformed into E. coli. Plasmids were isolated using aQIAprep Spin Miniprep Kit (Qiagen) and confirmed using bothdiagnostic PCR and restriction digestion. Yeast strains were trans-formed using the LiAc/SS carrier DNA/PEG method (Gietz andSchiestl, 2007), and transformants were selected on the appropriateSC plates.

2.3. Strain construction

All the strains and plasmids used in this study are listed inTable 1. For the construction of pdc- strain, the loxP-KanMX-loxPmethod (Hegemann et al., 2006) was used for successive deletion ofall three structural genes in the order of PDC5-PDC6-PDC1. Theresultant strain (JL0178) was maintained in mediumwith ethanol asthe sole carbon source. After restoring the growth on glucose, theMTH1 containing strains (JL0273 and JL0274) were maintained onlow concentration glucose (5 g/L). To make the strain more suitable

for practical use, adaptive evolution by serial transfer in 20 g/Lglucose was carried out to improve the glucose consumption rate.The resultant strain (JL0304) was transformed with the plasmidharboring the BDO biosynthetic pathway to construct strain JL0432.

2.4. Flask and fed-batch fermentation

The recombinant yeast strains were pre-cultured in the appro-priate SC medium, washed twice in ddH2O, and then inoculatedinto YP medium supplemented with glucose and galactose with aninitial OD of 5, unless specified otherwise. To test the effect ofacetate on the growth and product formation in the engineeredpdc- strain, acetate was added into the medium at the concentra-tion of 5% (16.7 mM), 10% (33.3 mM), or 15% (50 mM) of totalsugar, respectively. For the engineered strains without the hetero-logous pathway, cells were cultured under aerobic condition,while oxygen limited condition was used for BDO fermentation.

Fed-batch fermentation was carried out in the Multifors Multi-ple Benchtop Bioreactors (Infors-HT, Bottmingen, Switzerland)with 0.5 L working volume. The temperature was controlled at30 1C and the pH of the culture media was maintained at 5.5 bycontrolled addition of 4 N NaOH or 4 N H2SO4. The dissolvedoxygen concentration in the bioreactor was maintained at around10–15% by controlling the air flow rate and agitation speed. StrainJL0432 was pre-cultured in SC-His-Ura medium and inoculatedinto 500 mL sterile YP medium supplemented with 10 g/L glucoseand 10 g/L galactose with an initial OD of 5. After 16 h offermentation, glucose and galactose were fed simultaneously fromsugar mixture stock containing 200 g/L glucose and 200 g/Lgalactose. Sugar feeding rate was micro-adjusted based on thesugar consumption rate, maintaining the sugar concentration inthe fermentor at around 10 g/L.

2.5. Analytical methods

Cell growth was determined by measuring the absorbance at600 nm using a Biotek Synergy 2 Multi-Mode Microplate Reader(Winooski, VT). For flask fermentation, samples were taken every4 h after inoculation. For fed-batch fermentation, samples weretaken every 2–4 h in the early stage to monitor sugar concentra-tion in the fermentor and every 12 h after reaching steady state.Extracellular metabolites, including glucose, galactose, acetate,and butanediol, were detected and quantified using ShimadzuHPLC (Columbia, MD) equipped with an Aminex HPX-87H column(Bio-Rad, Hercules, CA) and Shimadzu RID-10A refractive indexdetector. For the detection of pyruvate, the Shimadzu SPD-20Aprominence diode array detector was used. The column was keptat 65 1C and 0.5 mM sulfuric acid solution was used as a mobilephase with a constant flow rate of 0.6 mL/min. Each data pointrepresents the mean of at least duplicate samples.

2.6. Genome sequencing and single nucleotide variation (SNV)analyses

Next-generation sequencing of the whole genomes of JL0274,JL0274e1 (JL0304), JL0274e2, and JL0274e3 was performed by theRoy J. Carver Biotechnology Center at the University of Illinois atUrbana-Champaign. The barcoded shotgun genomic DNA librarieswere constructed with the TruSeq DNA Sample Preparation Kit(Illumina, San Diego, CA), and quantitated by qPCR. Then thelibraries were sequenced for 101 cycles from one end of thefragments (single-reads) for 100 nt on an Illumina HiSeq2000using a TruSeq SBS Sequencing Kit version 3. FASTQ files weregenerated with Casava 1.8 (Illumina).

Each sample yielded approximately 25 million reads, and theaverage quality scores per base were higher than 30. High

J. Lian et al. / Metabolic Engineering 23 (2014) 92–9994

throughout sequencing data were analyzed using CLC GenomicsWorkbench version 6.5 (Qiagen). Reads were trimmed based onquality scores, which were then mapped to an S288C referencesequence. SNVs were identified using the Probabilistic VariantDetection function. The Filter against Control Reads tool was usedto remove the variants found in JL0274, generating the SNVsunique to JL0274e1 (JL0304), JL0274e2, and JL0274e3.

3. Results

3.1. Construction of a pdc- strain to redirect the metabolicflux from ethanol

To redirect the flux from ethanol to the desired product, thethree structural genes encoding pyruvate decarboxylase, PDC1,PDC5, and PDC6, were knocked out consecutively using the classicloxP-KanMX-loxP method (Hegemann et al., 2006). As expected, thepdc- strain (JL0178) cannot grow on the medium with hexose (suchas glucose or galactose) as the sole carbon source, while the growthon ethanol was not affected (Supplementary Fig. S1). The introduc-tion of an internal truncated MTH1 was reported to enable thegrowth of pdc- strain on glucose (Oud et al., 2012; van Maris et al.,2004). Since the truncated MTH1 was claimed to have increasedintracellular stability and abundance to better repress glucoseuptake, a similar strategy was carried out by over-expressingMTH1 on a multi-copy plasmid in this study. Accordingly, theoverexpression of either the full length MTH1 (JL0273) or thetruncated version (MTH1T, JL0274) could rescue the growth in themediumwith low glucose concentration, where JL0274 grew slightlybetter (Fig. 2A). These results in turn supported the assumption thatthe internal deletion in MTH1 alleviated glucose repression byinterfering degradation of this transcriptional regulator protein.Nevertheless, the growth on higher sugar concentration was still

poor. To make this strain more suitable for industrial applications,adaptive evolution was carried out to improve the cell growth rateon high glucose concentrations. After 10 rounds of serial transfer inSC medium containing 20 g/L glucose, three single clones, JL0274e1(JL0304), JL0274e2, and JL0274e3, were isolated and the cell growthrate of the evolved strain was significantly improved (Fig. 2B). Thewhole genomes of the parent strain JL0274 and the three evolvedstrains were sequenced to identify the mutations that might beresponsible for the improved growth on high concentrations ofglucose. Approximately 30 SNVs unique to each of the evolved strainrelative to the parent strain were identified, and a majority of theSNVs were intergenic (non-coding regions) (Table S2). Four out offive non-synonymous SNVs were shared by all the evolved strains(Table 2). YHR094C (HXT1) encodes a low affinity hexose transporterinduced only at high glucose concentration, and the loss of functionwould decrease the glucose uptake rate when its concentration washigh (Ozcan and Johnston, 1995; Scalcinati et al., 2012). YML088W(UFO1) encodes a subunit of the SCF ubiquitin ligase complexresponsible for the ubiquitination and subsequent degradation ofregulatory and signaling proteins (Ivantsiv et al., 2006), and theinactivation might enhance the intracellular stability of MTH1. The

Table 1Strains and plasmids used in this study.

Strain Genotype Plasmids Source

CEN.PK2-1C MATa ura3-52 trp1-289 leu2-3,112 his3Δ1 MAL2-8C SUC2 – EuroscarfJL0178 CEN.PK2-1C Δpdc5::loxP Δpdc6::loxP Δpdc1::KanMX – This studyJL0273 CEN.PK2-1C Δpdc5::loxP Δpdc6::loxP Δpdc1::KanMX pRS423-PGK1p-MTH1-HXT7t This studyJL0274 CEN.PK2-1C Δpdc5::loxP Δpdc6::loxP Δpdc1::KanMX pRS423-PGK1p-MTH1T-HXT7t This studyJL0304 CEN.PK2-1C Δpdc5::loxP Δpdc6::loxP Δpdc1::KanMX,

evolved for better growth on glucosepRS423-PGK1p-MTH1T-HXT7t This study

JL0432 CEN.PK2-1C Δpdc5::loxP Δpdc6::loxP Δpdc1::KanMX,evolved for better growth on glucose

pRS423-PGK1p-MTH1T-HXT7t This studypRS426-ENO2p-cytoILV2-PGK1t-TPI1p-BsAlsD-TPI1t-TEF1p-BDH1-TEF1t

Fig. 2. Construction of a pdc- strain that can grow on glucose as the sole carbon source. (A) Restoring the cell growth on glucose via the introduction of MTH1 (JL0273) andMTH1T (JL0274). (B) Engineering better growth of strain JL0274 on glucose via adaptive evolution. Yeast cells were pre-cultured on ethanol, inoculated into SC mediumcontaining either 5 g/L (A) or 20 g/L (B) glucose with an initial OD around 0.05, and cultured under aerobic condition.

Table 2SNVs identified in the evolved strains as compared with JL0274.

Non-synonymous SNVs in CDS JL0274e1 JL0274e2 JL0274e3

YER046W (SPO73) Frame shift after Thr13YHR094C (HXT1) Frame shift after Pro420YLR412C Frame shift after Gly46YML088W (UFO1) Stop codon after Glu600YNR038W (DBP6) – – Leu605SerSNVs in CDS 4 4 5SNVs in mitochondrial DNA 8 7 8Total SNVs 32 34 38

J. Lian et al. / Metabolic Engineering 23 (2014) 92–99 95

mechanisms of YER046W and YLR412C mutations in the evolvedphenotype remain unclear. Notably, about 25% of the SNVs werefound in the mitochondrial DNA (Table 2), which agreed withthe requirement of mitochondrial activity for cell growth in thepdc- strain. Since the phenotypes of the three evolved strains weresimilar, JL0274e1 (JL0304) was chosen for further studies. As shownin Supplementary Fig. S2, pyruvate was secreted as the majorproduct under aerobic condition.

3.2. Simultaneous utilization of glucose and non-glucose carbonsources by the engineered pdc- strain

Since the overexpression of MTH1 released glucose repression,we hypothesized that this engineered strain (JL0304) might enablethe utilization of glucose and non-glucose sugars simultaneously.To test this hypothesis, co-fermentation of glucose and galactose,which are the major carbohydrate components of red algae, wascarried out. Although S. cerevisiae can consume galactose efficientlyusing the endogenous Leloir pathway, its utilization is completelyinhibited with the presence of glucose in the wild-type strain(Ha et al., 2011). Interestingly, glucose and galactose were con-sumed simultaneously in the pdc- strain, without any glucoserepression phenomena observed (Fig. 3A). A 10 g/L glucose and10 g/L galactose were metabolized in 20 h after inoculation, andthe specific consumption rate of glucose and galactose was rathersimilar.

Although acetate was not reported to be present in the hydro-lysate of red seaweed at high concentration, acetic acid hydrolysiswas proposed to be a better method than the traditional sulfuricacid hydrolytic technology (Ha et al., 2013; Kim et al., 2012a).Therefore, co-fermentation of glucose and galactose was alsocarried out in the presence acetate. As shown in Fig. 3B, all threecomponents, 10 g/L glucose, 10 g/L galactose, and 2 g/L acetate, wereconsumed nearly at the same rate. Interestingly, the sugar con-sumption rate was even higher with the presence of acetate, whichmay serve as a source to generate cytosolic acetyl-CoA for lipid andhormone biosynthesis. Similar growth and sugar consumption ratewere observed if acetate was supplemented to the medium in therange of 5–15% of total sugar (data not shown). In other words,acetate, the previously reported inhibitor (Himmel and Bayer,2009), could unexpectedly enhance the cell growth and sugarconsumption rate of the engineered strain at the concentrationrange typically found in the biomass hydrolysate (Sedlak and Ho,2004).

3.3. Co-fermentation of glucose and galactose to produceenantiopure (2R,3R)-butanediol at high titer and yield

By introducing a BDO biosynthetic pathway into the pdc- strain,glycolytic flux was redirected into BDO fermentation (Fig. 1).Notably, better cell growth and enhanced sugar consumption wereobserved in the strain containing the BDO pathway (JL0432) whenthe oxygen level was low. Particularly, BDO fermentation enabledthe growth of the engineered pdc- strain under anaerobic condi-tion, in both YP media (Fig. 4) and SC media (SupplementaryFig. S3). In other words, the introduction of the BDO pathwaycould partially restore anaerobic growth of the pdc- yeast strain.Acetoin was accumulated to a high level under aerobic condition,while glycerol was the major product under anaerobic condition(Supplementary Fig. S4 and Supplementary Table S3). In addition,higher BDO yield and productivity were achieved in YP mediumthan those in SC medium (Supplementary Table S3). Therefore,subsequent BDO fermentation studies were carried out underoxygen-limited condition in a rich medium. In YP mediumsupplemented with 10 g/L glucose and 10 g/L galactose, the BDOyield was higher than 70% of the theoretical value (Fig. 5A andSupplementary Table S3).

Interestingly, only one peak for BDO was detected in theHPLC chromatogram of the fermentation broth, indicating theproduction of enantiopure BDO in strain JL0432. To determinethe chirality of the BDO produced, enantiopure R-BDO andmeso-BDO were loaded to the column. Using the HPLC methodmentioned above, acetoin, meso-BDO and R-BDO can be separated(Supplementary Fig. S4A–D). Owing to the stereospecificityof the yeast BDH, R-BDO and meso-BDO are synthesized fromR-acetoin and S-acetoin, respectively (Gonzalez et al., 2000).Thus, S-BDO was not included in this study. Accordingly, it couldbe concluded that only R-BDO, with R-acetoin as the intermediate,was synthesized in the pdc- yeast strain (SupplementaryFig. S4E–G). As a comparison, in our previous work to redirectthe flux from ethanol to BDO by inactivating ADHs, a mixtureof R-BDO and meso-BDO were produced at low yield and produc-tivity, and the production of ethanol as the major product as wellas the accumulation of acetate were observed (SupplementaryFig. S4H).

Similar to previously reported results (Oud et al., 2012; vanMariset al., 2004), sugar utilization was severely impaired after completedisruption of PDCs. Although the pdc- strain was engineered forbetter growth on glucose via introduction of MTH1 and adaptiveevolution, the sugar consumption rate and BDO productivity weredecreased at higher sugar concentration (Supplementary Table S4).

Fig. 3. Co-utilization of glucose and non-glucose carbon sources in the engineered pdc- yeast strain. (A) Simultaneous consumption of glucose and galactose.(B) Co-utilization of glucose and galactose with the presence of acetate at the concentration of 10% of total sugar.

J. Lian et al. / Metabolic Engineering 23 (2014) 92–9996

Thus fed-batch fermentation was carried out to produce more BDOin a fermentor. By feeding the same concentration of glucose andgalactose continuously to the bioreactor, BDO continued to besynthesized in up to 300 h, and more than 100 g/L (roughly1100 mM) was produced at the end of fermentation (Fig. 5B).Moreover, the concentration of glucose and galactose in thebioreactor was maintained relatively constant, indicating the simul-taneous utilization of glucose and galactose during the wholefermentation process.

4. Discussion

Unlike many BDO native producers, S. cerevisiae is an indust-rially friendly cell factory for the production of bio-based fuels

and chemicals (Hong and Nielsen, 2012). During normal alcoholicfermentation, several endogenous pathways in S. cerevisiae,mainly the PDC-dependent pathway and acetolactate synthase-dependent pathway, lead to the formation of a small amount ofBDO (Gonzalez et al., 2000). Although at a low concentration ofaround 1 mM, the BDO produced by the wild-type yeast consti-tutes a mixture of 67% R-BDO and 33% meso-BDO, derived from(3R)-acetoin and (3S)-acetoin, respectively (Gonzalez et al.,2000). Metabolic engineering, which has been widely used toredirect the metabolic flux to the desired products in varioushosts (Atsumi et al., 2008; Cooksley et al., 2012; Lan and Liao,2011; Yu et al., 2011), was also applied to the production of BDOin S. cerevisiae. By disrupting the pyruvate decarboxylases andintroducing a BDO biosynthetic pathway (Fig. 1), BDO wasproduced at high titer and yield. Moreover, only one enantiopure

Fig. 4. Comparison of the growth and sugar consumption of the evolved pdc- strain with or without the BDO pathway. The wild-type strain (CEN.PK2), pdc-/MTH1 strain(JL0274), evolved pdc-/MTH1 strain (JL0304), and the evolved strain with the BDO pathway (JL0432) were cultured under aerobic (A), oxygen-limited (B), and anaerobicconditions (C). The amount of glucose consumed at the end of fermentation under these conditions was also compared (D). Yeast cells were pre-cultured on ethanol andinoculated into YPD media with an initial OD around 0.05.

Fig. 5. Co-fermentation of glucose and galactose for R-BDO production. (A) R-BDO fermentation profile in shake flask with 10 g/L glucose and 10 g/L galactose under oxygen-limited condition. (B) Fed-batch fermentation for R-BDO production. A mixture of sugars consisting of the same concentration of glucose and galactose was continuously fedto the bioreactor. A representative fed-batch fermentation profile was provided, and the variation of final titer and yield was less than 10%.

J. Lian et al. / Metabolic Engineering 23 (2014) 92–99 97

form (R-BDO) was produced in the pdc- strain (SupplementaryFig. S4), indicating the important role of PDCs in the biosynthesisof meso-BDO. For certain applications where enantiopure BDO isneeded, e.g. the use of R-BDO as an antifreeze agent, the yeaststrain constructed in this study should be advantageous com-pared to other reported engineered yeast strains and the nativeBDO producers.

Many renewable feedstocks contain different sugar compo-nents, such as glucose and xylose from cellulosic biomass,glucose and galactose from red algae, and glucose and mannitolfrom brown algae. To increase the productivity of the fermenta-tion process, it is desirable to engineer a yeast strain capableof consuming a mixture of sugars simultaneously. Althoughintensive studies have contributed to understanding the mechan-isms of catabolite repression, the construction of a glucose-derepressed yeast strain has never been achieved, probably dueto the complex regulation of the glucose signaling network(Rolland et al., 2002; Vinuselvi et al., 2012). In the present work,the pdc- strain was found to enable co-utilization of glucose andgalactose, with or without the presence of acetate (Fig. 3).Therefore, the constructed strain would be an ideal host toproduce bio-based fuels and chemicals from red algae-derivedsugar mixtures. To use lignocellulose as a feedstock, a yeast strainthat can consume glucose and xylose efficiently and simulta-neously needs to be engineered, which is a more challengingtask. However, our previously developed COMPACTER method(Du et al., 2012), which allows rapid optimization of host-specificmetabolic pathways by fine-tuning the expression levels ofmultiple genes simultaneously, can be applied to construct anefficient xylose utilization pathway in the pdc- strain. Simulta-neous utilization of glucose and xylose may be achieved in theglucose derepressed host with an optimized xylose utilizationpathway.

Due to the generation of one net NADH molecule in the BDObiosynthetic pathway, glycerol is produced as the major bypro-duct (Fig. 1 and Supplementary Fig. S4E), especially under lowoxygen level or anaerobic conditions. Although the deletion ofGPD1 and GPD2 was able to completely abolish glycerol produc-tion, the host was reported to have no growth under anaerobicconditions because of the lack of an electron sink to re-oxidizethe generated NADH molecule (Guadalupe Medina et al., 2010).Recently, acetate was proposed to be used as an electron sink inS. cerevisiae by introducing the E. coli MhpF gene, an acetylatingNADþ-dependent acetaldehyde dehydrogenase, which convertsacetyl-CoA to acetaldehyde then to ethanol by the endogenousADHs (Guadalupe Medina et al., 2010). Compared with theglycerol sink, there are several advantages to use acetate as theelectron acceptor. First of all, acetate is a cheap substrate andwidely existed in biomass hydrolysates (Himmel and Bayer, 2009;Sedlak and Ho, 2004). In addition, two NADH molecules will bere-oxidized per ethanol molecule produced from acetate, whileonly one NADH molecule will be re-oxidized per glycerol mole-cule produced from glucose.

In summary, a pdc- yeast strain was constructed by completedisruption of pyruvate decarboxylase activity, and the growthon glucose was restored by the overexpression of MTH1 followedby adaptive evolution. The engineered strain was found to consumeglucose and non-glucose carbon sources simultaneously, such asgalactose and galactose/acetate. After the introduction of the BDObiosynthetic pathway, enantiopure R-BDO was produced at hightiter and yield from a mixture of glucose and galactose. By feedingglucose and galactose continuously, more than 100 g/L R-BDO wasproduced, which is the highest BDO titer ever reported in yeast.To our knowledge, this is the first report to achieve simultaneousutilization of glucose and galactose and the production of enantio-pure R-BDO in S. cerevisiae.

Acknowledgments

This work was supported by the Energy Biosciences Institute.JL also acknowledges the support of the 3M Graduate ResearchFellowship.

Appendix A. Supplementary material

Supplementary data associated with this article can be found inthe online version at http://dx.doi.org/10.1016/j.ymben.2014.02.003.

References

Abbott, D.A., Zelle, R.M., Pronk, J.T., van Maris, A.J.A., 2009. Metabolic engineering ofSaccharomyces cerevisiae for production of carboxylic acids: current status andchallenges. FEMS Yeast Res. 9, 1123–1136.

Atsumi, S., Cann, A.F., Connor, M.R., Shen, C.R., Smith, K.M., Brynildsen, M.P., Chou,K.J., Hanai, T., Liao, J.C., 2008. Metabolic engineering of Escherichia coli for1-butanol production. Metab. Eng. 10, 305–311.

Celińska, E., Grajek, W., 2009. Biotechnological production of 2,3-butanediol-current state and prospects. Biotechnol. Adv. 27, 715–725.

Chen, Y., Daviet, L., Schalk, M., Siewers, V., Nielsen, J., 2013. Establishing a platformcell factory through engineering of yeast acetyl-CoA metabolism. Metab. Eng.15, 48–54.

Claros, M.G., Vincens, P., 1996. Computational method to predict mitochondriallyimported proteins and their targeting sequences. Eur. J. Biochem. 241, 779–786.

Cooksley, C.M., Zhang, Y., Wang, H., Redl, S., Winzer, K., Minton, N.P., 2012. Targetedmutagenesis of the Clostridium acetobutylicum acetone–butanol–ethanol fer-mentation pathway. Metab. Eng. 14, 630–641.

de Smidt, O., du Preez, J.C., Albertyn, J., 2012. Molecular and physiological aspects ofalcohol dehydrogenases in the ethanol metabolism of Saccharomyces cerevisiae.FEMS Yeast Res. 12, 33–47.

Du, J., Shao, Z., Zhao, H., 2011. Engineering microbial factories for synthesis of value-added products. J. Ind. Microbiol. Biotechnol. 38, 873–890.

Du, J., Yuan, Y., Si, T., Lian, J., Zhao, H., 2012. Customized optimization of metabolicpathways by combinatorial transcriptional engineering. Nucleic Acids Res. 40,e142.

Eriksen, D.T., Hsieh, P.C.H., Lynn, P., Zhao, H., 2013. Directed evolution of a cellobioseutilization pathway in Saccharomyces cervisiae by simultaneously engineeringmultiple proteins. Microb. Cell Fact. 12, 61.

Flikweert, M.T., de Swaaf, M., van Dijken, J.P., Pronk, J.T., 1999. Growth requirementsof pyruvate-decarboxylase-negative Saccharomyces cerevisiae. FEMS Microbiol.Lett. 174, 73–79.

Gietz, R.D., Schiestl, R.H., 2007. High-efficiency yeast transformation using the LiAc/SS carrier DNA/PEG method. Nat. Protoc. 2, 31–34.

Gonzalez, E., Fernandez, M., Larroy, C., Sola, L., Pericas, M., Pares, X., Biosca, J., 2000.Characterization of a (2R,3R)-2,3-butanediol dehydrogenase as the Saccharo-myces cerevisiae YAL060W gene product. J. Biol. Chem. 275, 35876–35885.

Guadalupe Medina, V., Almering, M.J., van Maris, A.J., Pronk, J.T., 2010. Eliminationof glycerol production in anaerobic cultures of a Saccharomyces cerevisiae strainengineered to use acetic acid as an electron acceptor. Appl. Environ. Microbiol.76, 190–195.

Ha, S.J., Galazka, J.M., Joong, Oh, E., Kordic, V., Kim, H., Jin, Y.S., Cate, J.H., 2013.Energetic benefits and rapid cellobiose fermentation by Saccharomyces cerevi-siae expressing cellobiose phosphorylase and mutant cellodextrin transporters.Metab. Eng. 15, 134–143.

Ha, S.J., Wei, Q., Kim, S.R., Galazka, J.M., Cate, J., Jin, Y.S., 2011. Cofermentation ofcellobiose and galactose by an engineered Saccharomyces cerevisiae strain. Appl.Environ. Microbiol. 77, 5822–5825.

Hegemann, J.H., Guldener, U., Kohler, G.J., 2006. Gene disruption in the buddingyeast Saccharomyces cerevisiae. Methods Mol. Biol. 313, 129–144.

Himmel, M.E., Bayer, E.A., 2009. Lignocellulose conversion to biofuels: currentchallenges, global perspectives. Curr. Opin. Biotechnol. 20, 316–317.

Hohmann, S., 1993. Characterisation of PDC2, a gene necessary for high levelexpression of pyruvate decarboxylase structural genes in Saccharomyces cere-visiae. Mol. Gen. Genet. 241, 657–666.

Hong, K.K., Nielsen, J., 2012. Metabolic engineering of Saccharomyces cerevisiae: akey cell factory platform for future biorefineries. Cell. Mol. Life Sci. 69,2671–2690.

Ida, Y., Furusawa, C., Hirasawa, T., Shimizu, H., 2012. Stable disruption of ethanolproduction by deletion of the genes encoding alcohol dehydrogenase isozymesin Saccharomyces cerevisiae. J. Biosci. Bioeng. 113, 192–195.

Ivantsiv, Y., Kaplun, L., Tzirkin-Goldin, R., Shabek, N., Raveh, D., 2006. Unique rolefor the UbL–UbA protein Ddi1 in turnover of SCFUfo1 complexes. Mol. Cell. Biol.26, 1579–1588.

Ji, X.J., Huang, H., Ouyang, P.K., 2011. Microbial 2,3-butanediol production: a state-of-the-art review. Biotechnol. Adv. 29, 351–364.

Kim, H.T., Lee, S., Kim, K.H., Choi, I.G., 2012a. The complete enzymatic saccharifica-tion of agarose and its application to simultaneous saccharification andfermentation of agarose for ethanol production. Bioresour. Technol. 107,301–306.

J. Lian et al. / Metabolic Engineering 23 (2014) 92–9998

Kim, S.R., Ha, S.J., Wei, N., Oh, E.J., Jin, Y.S., 2012b. Simultaneous co-fermentation ofmixed sugars: a promising strategy for producing cellulosic ethanol. TrendsBiotechnol. 30, 274–282.

Lan, E.I., Liao, J.C., 2011. Metabolic engineering of cyanobacteria for 1-butanolproduction from carbon dioxide. Metab. Eng. 13, 353–363.

Ng, C.Y., Jung, M.Y., Lee, J., Oh, M.K., 2012. Production of 2,3-butanediol inSaccharomyces cerevisiae by in silico aided metabolic engineering. Microb. CellFact. 11, 68.

Nielsen, D.R., Leonard, E., Yoon, S.H., Tseng, H.C., Yuan, C., Prather, K.L., 2009.Engineering alternative butanol production platforms in heterologous bacteria.Metab. Eng. 11, 262–273.

Oud, B., Flores, C.L., Gancedo, C., Zhang, X., Trueheart, J., Daran, J.M., Pronk, J.T.,van Maris, A.J., 2012. An internal deletion inMTH1 enables growth on glucose ofpyruvate-decarboxylase negative, non-fermentative Saccharomyces cerevisiae.Microb. Cell Fact. 11, 131.

Ozcan, S., Johnston, M., 1995. Three different regulatory mechanisms enable yeasthexose transporter (HXT) genes to be induced by different levels of glucose.Mol. Cell. Biol. 15, 1564–1572.

Rolland, F., Winderickx, J., Thevelein, J.M., 2002. Glucose-sensing and -signallingmechanisms in yeast. FEMS Yeast Res. 2, 183–201.

Runguphan, W., Keasling, J.D., 2014. Metabolic engineering of Saccharomycescerevisiae for production of fatty acid-derived biofuels and chemicals. Metab.Eng. 21, 103–113.

Savakis, P.E., Angermayr, S.A., Hellingwerf, K.J., 2013. Synthesis of 2,3-butanediol bySynechocystis sp. PCC6803 via heterologous expression of a catabolic pathwayfrom lactic acid- and enterobacteria. Metab. Eng. 20, 121–130.

Scalcinati, G., Knuf, C., Partow, S., Chen, Y., Maury, J., Schalk, M., Daviet, L., Nielsen, J.,Siewers, V., 2012. Dynamic control of gene expression in Saccharomycescerevisiae engineered for the production of plant sesquitepene alpha-santalene in a fed-batch mode. Metab. Eng. 14, 91–103.

Sedlak, M., Ho, N.W., 2004. Production of ethanol from cellulosic biomass hydro-lysates using genetically engineered Saccharomyces yeast capable of coferment-ing glucose and xylose. Appl. Biochem. Biotechnol. 114, 403–416.

Shao, Z., Zhao, H., Zhao, H., 2009. DNA assembler, an in vivo genetic method forrapid construction of biochemical pathways. Nucleic Acids Res. 37, e16.

Syu, M.J., 2001. Biological production of 2,3-butanediol. Appl. Microbiol. Biotechnol.55, 10–18.

van Maris, A.J.A., Geertman, J.M.A., Vermeulen, A., Groothuizen, M.K., Winkler, A.A.,Piper, M.D.W., van Dijken, J.P., Pronk, J.T., 2004. Directed evolution of pyruvatedecarboxylase-negative Saccharomyces cerevisiae, yielding a C2-independent,glucose-tolerant, and pyruvate-hyperproducing yeast. Appl. Environ. Microbiol.70, 159–166.

Vinuselvi, P., Kim, M.K., Lee, S.K., Ghim, C.M., 2012. Rewiring carbon cataboliterepression for microbial cell factory. BMB Rep. 45, 59–70.

Wargacki, A.J., Leonard, E., Win, M.N., Regitsky, D.D., Santos, C.N., Kim, P.B., Cooper,S.R., Raisner, R.M., Herman, A., Sivitz, A.B., Lakshmanaswamy, A., Kashiyama, Y.,Baker, D., Yoshikuni, Y., 2012. An engineered microbial platform for directbiofuel production from brown macroalgae. Science 335, 308–313.

Wei, N., Quarterman, J., Jin, Y.S., 2013. Marine macroalgae: an untapped resource forproducing fuels and chemicals. Trends Biotechnol. 31, 70–77.

Wi, S.G., Kim, H.J., Mahadevan, S.A., Yang, D.J., Bae, H.J., 2009. The potential value ofthe seaweed Ceylon moss (Gelidium amansii) as an alternative bioenergyresource. Bioresour. Technol. 100, 6658–6660.

Yu, M., Zhang, Y., Tang, I.C., Yang, S.T., 2011. Metabolic engineering of Clostridiumtyrobutyricum for n-butanol production. Metab. Eng. 13, 373–382.

Yuan, Y., Zhao, H., 2013. Directed evolution of a highly efficient cellobiose utilizingpathway in an industrial Saccharomyces cerevisiae strain. Biotechnol. Bioeng.110, 2874–2881.

J. Lian et al. / Metabolic Engineering 23 (2014) 92–99 99