metabolic engineering of e. coli. a sustainable industrial platform for bio-based chemical...

8/19/2019 Metabolic Engineering of E. Coli. a Sustainable Industrial Platform for Bio-based Chemical Production

1/24

Research review paper

Metabolic engineering of Escherichia coli: A sustainable industrialplatform for bio-based chemical production

Xianzhong Chen a,⁎, Li Zhou a, Kangming Tian a, Ashwani Kumar b, Suren Singh b,Bernard A. Prior c, Zhengxiang Wang d,⁎⁎a Key Laboratory of Industrial Biotechnology, Ministry of Education & School of Biotechnology, Jiangnan University, 1800 Lihu Avenue, Wuxi 214122, Chinab Department of Biotechnology & Food Technology, Faculty of Applied Sciences, Durban University of Technology, P.O. Box 1334, Durban, 4001, South Africac Department of Microbiology, University of Stellenbosch, Private Bag X1, Matieland, 7602, South Africad Key Laboratory of Industrial Fermentation Microbiology, Ministry of Education & The College of Biotechnology, Tianjin University of Science & Technology, Tianjin 300457, China

a b s t r a c ta r t i c l e i n f o

Article history:

Received 18 October 2012

Received in revised form 4 February 2013

Accepted 25 February 2013

Available online 6 March 2013

Keywords:

Escherichia coli

Metabolic engineering

Bioprocess

Biochemical product

Industrial platform

In order to decrease carbon emissions and negative environmental impacts of various pollutants, more bulk

and/or ne chemicals are produced by bioprocesses, replacing the traditional energy and fossil based inten-

sive route. The Gram-negative rod-shaped bacterium, Escherichia coli has been studied extensively on a

fundamental and applied level and has become a predominant host microorganism for industrial applications.

Furthermore, metabolic engineering of E. coli for the enhanced biochemical production has been signicantly

promoted by the integrated use of recent developments in systems biology, synthetic biology and evolutionary

engineering. In this review, we focus on recent efforts devoted to the use of genetically engineered E. coli as a

sustainable platform for the production of industrially important biochemicals such as biofuels, organic acids,

amino acids, sugar alcohols and biopolymers. In addition, representative secondary metabolites produced by

E. coli will be systematically discussed and the successful strategies for strainimprovements will be highlighted.

Moreover, this review presents guidelines for future developments in the bio-based chemical production using

E. coli as an industrial platform.

© 2013 Elsevier Inc. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1201

2. Biofuels production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1201

2.1. Hydrogen production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1201

2.2. Bioethanol production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1203

2.3. Advanced biofuels production using recombinant E. coli as ef cient biocatalyst . . . . . . . . . . . . . . . . . . . . . . . . . . 1205

2.3.1. 1-Butanol and 1-propanol production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1205

2.3.2. 2-Methyl-1-butanol and 3-methyl-1-butanol production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1205

2.3.3. Isopropanol and isobutanol production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1206

3. Organic acids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1206

3.1. Lactic acid production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1206

3.2. Succinic acid production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12093.3. Production of other organic acids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1210

4. Amino acids production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1210

4.1. L -Threonine production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1211

4.2. L -Valine production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1212

4.3. L -Phenylalanine production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1212

4.4. Production of other amino acids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1212

5. Sugar alcohols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1213

Biotechnology Advances 31 (2013) 1200–1223

⁎ Corresponding author. Tel./fax: +86 510 85918122.

⁎⁎ Corresponding author. Tel/fax: +86 022 60601958.

E-mail addresses: [email protected] (X. Chen), [email protected] (Z. Wang).

0734-9750/$ – see front matter © 2013 Elsevier Inc. All rights reserved.

http://dx.doi.org/10.1016/j.biotechadv.2013.02.009

Contents lists available at ScienceDirect

Biotechnology Advances

j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / b i o t e c h a d v

http://dx.doi.org/10.1016/j.biotechadv.2013.02.009http://dx.doi.org/10.1016/j.biotechadv.2013.02.009http://dx.doi.org/10.1016/j.biotechadv.2013.02.009mailto:[email protected]:[email protected]://dx.doi.org/10.1016/j.biotechadv.2013.02.009http://www.sciencedirect.com/science/journal/07349750http://crossmark.crossref.org/dialog/?doi=10.1016/j.biotechadv.2013.02.009&domain=pdfhttp://www.sciencedirect.com/science/journal/07349750http://dx.doi.org/10.1016/j.biotechadv.2013.02.009mailto:[email protected]:[email protected]://dx.doi.org/10.1016/j.biotechadv.2013.02.009


2/24

6. Biopolymer and monomers production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1213

6.1. 1,3-Propanediol and 1,2-propanediol . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1213

6.2. 1,4-Butanediol and 2,3-butanediol . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1214

6.3. Polyhydroxyalkanoates production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1214

6.4. Polylactic acid and copolymers production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1215

7. Biosynthesis of complex natural compounds in E. coli . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1215

7.1. Isoprenoid biosynthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1216

7.2. Nonribosomal peptides and polyketides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1216

7.3. Coenzyme Q10 production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1218

8. Conclusion and future directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1218Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1219

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1219

1. Introduction

Currently commodity chemicals and traditional fuels are predom-

inantly produced from fossil based resources and have played an

essential role in international social and economic development.

However, withthe rising concernsof the sustainability, carbon dioxide

release into the atmosphere and the negatively environmental impact

of traditional petrochemical based products and fossil fuels, the bio-

technological route of production from renewable carbon sources is

a desirable alternative to petrochemical-based production (Bozell

and Petersen, 2010; Bruschi et al., 2011). The economic potential of

biotechnology industry has become an important contribution to the

world economy in recent years: its global value is estimated about

500 billion dollars in 2011, compared to only 54 billion dollars in

1999 and 101 billion dollars in 2003 (Bruschi et al., 2011). Increasingly

bulk chemicals (such as organic acid, amino acid, and biofuels) andne

chemicals (antibiotics, vitamins, pharmacy chemicals, sweeteners, and

performance materials) have been produced by the intervention of bio-

technology at an industrial scale. Many industrial processes are based

on the catalytic activities of microorganisms and therefore strains

need to be developed and selected that grow rapidly to a high cell

density with high volumetric and specic productivities (activity per

fermentation volume or biomass), and yield products from substratesclose to the theoretical maximum. Additional properties of high stress

tolerance,and simple molecular manipulation based on availablegenetic

information are desirable (Huffer et al., 2012).

Extensive fundamental studies on Escherichia coli over the last

50 years have resulted in the bacterium becoming the prime prokary-

otic genetic model. Moreover, since the beginning of the modern

biotechnology era in the late 70s, E. coli has been used widely for mo-

lecular cloning methodologies and as a host to produce primary and

secondary metabolites. Representative biochemical products are

summarized in Table 1. E. coli possesses a number of excellent proper-

ties, such as a rapid doubling time and growth rate, ease of high-cell-

density fermentation, low production cost and, detailed knowledge of

the metabolism and most importantly, the availability of excellent

genetic tools for strain improvement. These characteristics makeE. coli an ideal candidate for both metabolic engineering and

commercial-scale production of desirable bioproducts (Table 1).

Traditionally, strain improvement was achieved mainly by multiple

rounds of random mutagenesis and selection, which are still very useful

nowadays (Portnoy et al., 2011; Sonderegger and Sauer, 2003). In the

latest decade, the development of gene deletion approaches (Bloor

and Cranenburgh, 2006; Datsenko and Wanner, 2000; Murphy et al.,

2000) enabled ef cient genome DNA inactivation and greatly improved

metabolic engineering of E. coli. A more systematic and integrated

approach for biotechnological process development and optimization

became prevalent. Furthermore, metabolic engineering of E. coli has

been employed to broaden the variety of available products (Table 1).

In view of the above, this review highlights the current trends towards

the bio-based production of chemicals using genetically modied E. coli

as a sustainable biocatalytic platform and the impact on industrial bio-

technology. Therefore, essential primary and secondary metabolites

produced by engineered E. coli are evaluated and discussed. In addition

we show that systematically engineeredE. coli will pave a broad avenue

for the development of a green replacement for petrochemical products

and play a critical role as a novel platform in industrial microorganism

and pharmaceutical biotechnology. The physiology, biochemistry,

genomics and methodology of this microorganism have been exten-

sively studied (Blattner et al., 1997; Bloor and Cranenburgh, 2006)

and are outside the scope of this review whereas strategies to improve

the catalytic ef ciency of E. coli are given more emphasis.

2. Biofuels production

Concerning availability and abundance of biomass resources, as

well as environmental pollution of fossil fuels, development of

bioenergy as an alternative fuels has gained more attention in recent

past. Generally, biofuelsincludes the hydrogen,bioethanol and advanced

fuels (Table 1) but the catabolic pathways to synthesize these products

are diverse (Fig. 1). In this section, we discuss the progress in biofuels

production by engineered E. coli and highlight some of successful strate-

gies for increasing catalytic ef ciency of cell.

2.1. Hydrogen production

As an alternative for petroleum fuels, microbial production of

hydrogen as a future fuel is a promising possibility (Kim and Lee, 2010;

Panagiotopoulos et al., 2009). Hydrogen is a highly energy dense source

andits conversion to heat or power is simpleand clean when combusted

with oxygen as only water is formed without generation of pollutants

(Hoffmann, 2002). Biological production of hydrogen using biomass as

a feedstock is less energy-intensive, sustainable, eco-friendly, and con-

sidered to be neutral for CO2 emissions (Panagiotopoulos et al., 2009).

A number of newly isolated microbes i.e. Thermoanaerobacterium

thermosaccharolyticum, Clostridium beijerinckii and Sporoacetigenium

mesophilum have been reported in literature for hydrogen production,

but with limited production (Cai et al., 2011; Oh et al., 2011). The lowyield has been a major obstacle for the hydrogen production through

microbial fermentation of glucose. Kim et al. (2009) attempted to over-

come the limitations with natural isolates by metabolically engineering

E. coli strains for hydrogen production. Deletion of hyc A, a negative

regulator for formate hydrogen lyase and two uptake hydrogenases

(hya and hyb), resulted in carbon ux alterations to H2 pathway. H2yield was further improved to 2.11 mol/mol glucose fermented by

deletion of lactate dehydrogenase (ldhA) and fumarate reductase

( frdAB)underreduced H2 pressure in batchexperiments. Using a similar

genetic approach, H2 was produced with an improved yield by an

engineered E. coli strain (Maeda et al., 2008). A recent report showed a

high volumetric productivity of 2.4 H2/L/h using immobilized cells of a

genetically recombinant E. coli, which has deletion mutations in uptake

hydrogenases (ΔhyaAB), lactate dehydrogenase (ΔldhA) and fumarate

1201 X. Chen et al. / Biotechnology Advances 31 (2013) 1200–1223


3/24

Table 1

Representative bio-based chemicals produced by genetically engineered E. coli.

B io-b ased chemica l I ndu st rial

applications

Engin eerin g strategy Fe rmen tati on process Carbon

source

Titer and yield Refs

Biofuels Hydrogen Fuel and energy carrier Deletion of negative regulator and

competing carbon metabolic

pathways

Immobilized

recombinant cells

F ormate 1.0 mol H2/mol

formate; 2.4 l

H2/L/ha; formate.

Seol et al.

(2011)

Bioethanol Fuel, solvent, food,

beverage

Minimized metabolic functionality for

conversion of xylose and glucose intoethanol by multiple-gene knockout

10-L bioreactor batch

anaerobic cultivation

Xylose and

glucose

38.81 g/L;

0.49 g/g glucoseor xylose

Trinh et al.

(2008)

1-Propanol Gasoline additive,

allround solvent

Improving specic activity and

releasing feedback inhibition of the

key enzyme by directed evolution

Shake asks and IPTG

induction

Glucos e 3.5 g/L;

0.049 g/gbAtsumi and

Liao (2008a)

1-Butanol Bulk material, gasoline

additive or fuel

Construction of modied clostridial

1-butanol pathway in E. coli strain

and enhancement of driving forces

1-L Bioreactor with

aerobic–anaerobic dual

phase fermentation

Glucose 30 g/L; 88% of the

theoretical yield

Shen et al.

(2011)

3-Methyl-

1-butanol

Advanced fuel,

alternative gasoline

Random mutagenesis and selection

combined with overexpression of key

genes

Shake ask with

two-phase fermentation

to remove product

Glucose 9.5 g/L; 0.11 g/g

glucose

Connor et al.,

(2010)

Isopropanol Biodiesel, precursor of

polypropylene

Heterologous expression of target

product pathway from various

sources in E. coli host

Fed-batch fermentation,

gas-stripping-based

recovery process

G lu cose 1 43 g/L ; 67 .4 %

(mol/mol)

Inokuma et

al. (2010)

Isobutanol Gasoline blend stock,

precursor of butenes

Introducing non-fermentative

synthetic pathway in E. coli and

elimination of pathways competing

for pyruvate and cofactors

Screw-cap conical asks

and IPTG induction

Glucose 20 g/L; 86% of the

theoretical

maximum

Atsumi et al.

(2008b)

L -Lactic acid Food, beverages,

biopolymer

Overexpression of L -lactate

dehydrogenase and deletion of genes

responsible for by-products pathway

500-mL Flask with

automatic pH control,

batch fermentation

Glucose and

xylose

57.1 g/L; 0.83 g/g

of total sugar

Dien et al.

(2002)

D-Lactic acid Cosmetics,

pharmaceuticals

biopolymer

Disruption of the competing

pathways and expression of key

enzyme

7-l Bioreactor aerobic and

oxygen-limited fed-batch

fermentation

G lu cose 1 22 .8 g/L;

0.866 g/g

Zhou et al.

(2011b)

Succinic acid Bulk chemical, food,

agriculture

Inactivation of ptsG, p and ldhA,

combined overexpression of pyc gene

2.5-L Fermentor

aerobic-anaerobic

process fed-batch

G lu cose 9 9.2 g/L ; 1 10 % Vemuri et al.

(2002)

Pyruvate Food, pharmaceuticals,

and chemicals

Elevation of glycolytic ux to

pyruvate by multiple-gene knockout

2.5-L Fermentor

fed-batch process

G lu cose 9 0 g/L ; 0 .6 8 g/g Zhu et al.

(2008)

Acetate Food, plastics, solvents,

and de-icers

Genetic modication with blocking of

by-products pathways, oxidative

phosphorlation and TCA function

14-L Fermentor with

fed-batch process and DO

control

G lu cose 8 78 m M; 7 5% of

the theoretical

maximum

Causey et al.

(2003)

Amino

acids

L -Threonine Food, antibiotics,

pharmaceuticals, animal

feed

System metabolic engineering with

transcriptome proling and in silico

ux response analysis

6.6-L Fermentor pH-stat

fed-batch culture

Gluco se 82. 4 g/L;

0.393 g/g

Lee et al.

(2007)

L -Valin e Cosme ti cs,

pharmaceuticals, feed

additive

Systematically metabolic engineering

with transcriptome analysis and in

silico genome-scale simulation

Batch fermentation in

shake asks

Gluco se 7.55 g/L;

0.378 g/g

Park et al.

(2007)

L -Phenylalanine Food industry,

Sweetener aspartame

Reverse engineering with metabolic

transcription analysis, deletion of PTS

system and expression of key genes

Shake asks with IPTG

induction process

Gluco se 0.33 g/g Baez-Viveros

et al. (2007)

L -Tryptophan Food, animal feed,

pharmaceutical

industries

Implementation of temperature-

inducible expression system, elimina-

tion of feedback inhibitions and degra-

dation pathway of desired product

3-L Fermentor with

fed-batch fermentation,

temperature-switching

method

Glucose 13.3 g/L; 0.1 g/g Zhao et al.

(2011)

L -Tyrosine Precursor of drugs,

biopolymers, melanin

Expression of the key feedback

inhibition-insensitive enzyme from

Z.mobilis

1-L Bioreactors batch

fermentation

G lu cose 3 g/L; 6 6 mg/g Chavez-Bejar

et al. (2008)

Sugar

alcoholos

Xylitol Pha rma ceut ic al a nd

food industry

Heterologous overexpression of the

key enzymes in xyltiol biosynthetic

pathway and construction of cofactor

regeneration

Aerobic growth- anaero-

bic production, fed-batch

fermentation

Glucose and

xylose

8.52 g/L b;

4 mol/mol

Akinterinwa

and Cirino

(2011)

Mannitol Pharmaceutical andfood industry

Construction of cofactor cycle systemand heterologous expression of the

key enzymes

Whole-cellbiotransformation with

pH-static conditions

D-fructose 362 mM;0.84 mol/mol

Kaup et al.(2004)

Diols and

polymers

1 ,3 -PDO Monom er s f or t he

synthesis of polyester

Synergetic introduction of glycerol

pathway from S. cerevisiae and

1,3-PDO pathway from K. pneumonia

Fed-batch fermentation Glucose 130 g/L Emptage et

al. (2003)

1,2-PDO Polyester, antifreeze,

and deicer

Heterologous overexpression of the

key enzymes and disruption of

undesired by-products pathways

Fermentor containing

500-mL medium, batch

anaerobic fermentation

Glycerol 5.6 g/L; 0.213 g/g Clomburg

and Gonzalez

(2011)

1,4-BDO Polymers, ne

chemicals and solvents.

Implementation of synthetic biology

and systematic metabolic engineering,

and rational metabolic engineering

2-L Bioreactors with

aerobic-microaerobic

two phase fed-batch

Either of

glucose,

xylose or

sucrose

18 g/L Yim et al.

(2011)

(R,R)-2,3-BDO Polymers, food,

cosmetics, and

pharmaceuticals

Evaluation of various synthetic

operons consisting of the 2,3-BDO

biosynthesis pathway with different

secondary alcohol dehydrogenases

Shake asks with IPTG

induction process

Glucose 6.1 g/L; 0.31 g/g Yan et al.

(2009)



4/24

dehydrogenase (Δ frdAB) (Seol et al., 2011). Although, biological H2 pro-

duction was signicantly improved by genetically modied E. coli strain,

many critical barriers involved in the yield, productivity and metabolic

robustness are still not at a level that would allow commercialization.

2.2. Bioethanol production

Bioethanol is the predominant renewable liquid energy source

capturing 90% of the current world biofuel market ( Antoni et al.,

2007), with annual production of more than 105 billion liters in

2011. Production by Saccharomyces cerevisiae of starch-based ethanol

and sugar cane-based ethanol are now mature industries but thesebiomass sources compete with food and feed (Geddes et al., 2011b),

which leads to a global increase in the price and demand for food

crops. Ethanol production from xylose, glucose or mixture sugars

(maincompounds of plant lignocellulose residues) has been intensively

investigatedfor more than 20 years (Alterthumand Ingram, 1989; Ohta

et al., 1990) and many companies are attempting to commercialize the

process. The recalcitrance of lignocellulose (cellulose and hemicellu-

lose) to direct microbial fermentation to ethanol has resulted in the

requirement for complicated pretreatment processes (Himmel et al.,

2007). Ethanologenic E. coli strains have the advantage that they are

able to ferment pretreated biomass directly (Alterthum and Ingram,

1989; Moniruzzaman et al., 1997; Zaldivar et al., 2000; Zheng et al.,

2012). For example, E. coli strains which are adapted to phosphoric

acid hydrolysates, can ferment hemicellulose and cellulose-derivedsugars together in a single 80-l bioreactor vessel, termed simultaneous

saccharication and co-fermentation (SScF), and an ethanol yield of

0.27 g/g bagasse was obtained (Nieveset al., 2011). In another example,

deletion of mgsA encoding methylglyoxal synthase (and methylglyoxal)

resulted in the co-metabolism of glucose and xylose, and increased

the fermentation rate of ethanologenic E. coli by accelerating the co-

metabolism of hexose and pentose sugars to ethanol (Yomano et al.,

2009).

However, inhibitors present in lignocellulose hydrolysates such as

furfural, 5-hydroxymethyl furfural, acetate and soluble products

released from dilute acid pretreatment are toxic for most organisms

including E. coli and reduce ethanol yield drastically (Mills et al.,

2009). The toxicity problem can be partially overcome by increasing

tolerance of strains to hydrolysate inhibitors. In recent reports,

some E. coli metabolic regulators involved in resistance to furfural

have been identied (Miller et al., 2009a, 2009b; Turner et al., 2011;

Wang et al., 2011c). Miller et al. (2009a, 2009b) found that silencing

of yqhD, an NADPH-dependent furfural oxidoreductase induced by

furfural, can result in resistance of E. coli cells to low concentrations

of furfural. Furthermore, the enzyme of YqhD has unusually low Km

values for NADPH that would allow to compete with biosynthesis for

NADPH, which appears to be the primary basis for growth inhibition

by furfural (Miller et al., 2009b). The combination of deletion of yqhD

and overexpression of fucO, an NADH-dependent, L -1,2-propanediol

reductase, can provide a further benet for furfural tolerance (Wang

et al., 2011c). In addition, Wang et al. (2012) found that plasmidexpres-sion of ucpA, a putative oxidoreductasesgene, improved the furan toler-

ance in both the native W strain and ethanologenic strain LY180.

Deletion of the chromosomal ucpA decreased furfural tolerance, provid-

ing a clear phenotype for this cryptic gene.

On the other hand, hydrolysate resistant E. coli strains, which were

generated by either adaptive evolution or metabolic engineering

(Miller et al., 2009b; Mills et al., 2009; Wang et al., 2012), combined in

a SScF process resulted in both higheryields andless process complexity

(Geddes et al., 2011a). Accordingly, Ingram and coworkers (Wang et al.,

2011c) found that the overexpression of NADH-dependent propanediol

oxidoreductase can improve the furfural tolerance of E. coli cells, which

provided a new approach to improve furfural tolerance. In a recent ex-

ample, Edwards et al. (2011) constructed a recombinant ethanologenic

E. coli strain by introducing the Klebsiella oxytoca cellobiosephosphotransferase genes, a pectate lyase with secretion capability

and oligogalacturonide lyasegene from Erwinia chrysanthemi. Thismod-

ied strainef ciently convertedpolygalacturonate to monomeric sugars

and an increased ethanol production of 45% was achieved compared to

the control strain. In our studies, we have created a recombinant E. coli

strain, which can covert xylose to ethanol by introducing pdc and adhB

genes from Zymomonas mobilis (Sun et al., 2004). Synchronous expres-

sion of pdc and adhB genes encoding pyruvate decarboxylase and

alcohol dehydrogenase, respectively, redirected the carbon ux into

the ethanol production and the resulting recombinant strain produced

ethanol up to 1.28% (v/v) using xylose as carbon source within 36 h

fermentation. Similar approaches were implemented to engineer

E. coli for ethanol production from xylose, glucose or mixed sugars

(Sanny et al., 2010; Wang et al., 2008). Interestingly, E. coli with a

Table 1 (continued)

Bio-b ased c hemica l Indu str ia l

applications

E ngineering strategy Fermentation process Carbon

source

Titer and yield Refs

PHA Biopolyesters and

biofuel

Constitutive expression of an operon

responsible for PHA biosynthesis

pathway of A. latus in E. coli

6.6-L Fermentor

fed-batch culture with

pH-control

Glucose 141.6 g/L; 4.63 g

of PHB/L/h.cChoi et al.

(1998)

Taxadiene Taxol precursor, a

potent anticancer drug

Synthetic biology and

implementation of the

multivariate-modular approach to

optimally balance the two pathway

modules

Fed-batch and liquid–

liquid two-phase cultiva-

tion carried out in 1-L

bioreactors

Glucose 1.02 g/L Ajikumar et

al. (2010)

Echinomycin Antitumor, antibacterial

and antiviral activity

Cloning and heterologous expression

in E. coli of a monocistronic

reconstituted form of gene cluster

responsible for echinomycin

biosynthetic pathway

Fed-batch fermentation

carried out in 1-L biore-

actors and minimal

medium

Glucose 0.6 mg /L Watanabe et

al. (2006)

Anthracyclines Bacterial aromatic

polyketides, antibiotics

and anticancer drugs

The total biosynthesis of

anthracyclines by dissection and

reassembly of fungal polyketide

synthase into a synthetic PKS

High-cell density,

fed-batch fermentation in

2-L fermentor, IPTG

induction

Glucose 3 mg/L Zhang et al.

(2008)

Co Q10 Medicin e, foods and

cosmetics

Improvement of IPP supply by intro-

ducing a foreign mevalonate pathway

and coexpression of decaprenyl di-

phosphate synthasegene and IPP

isomerase

Fermentation with a

rotary shaking incubator

and optimization of pH

Glycerol 2700 μ g/g DCWd Zahiri et al.

(2006b)

a Proudction rate of H2.b

Calculated based on the references data.c PHB productivity.d The result was obtained when supplemented with exogenous mevalonate of 3 mM.



5/24

minimal metabolic functionality was rationally designed and

constructed for ethanol production based on metabolic pathway analy-

sis for cellular network (Trinh et al., 2008). With the disruption of

nonessential pathways, the strain produced 39 g/L ethanol through

simultaneous utilization of pentoses and hexoses with a highest yield

close to the theoretical maximum. This rational approach may be useful

for strain development to enhance different products. Conclusively,

E. coli, theworkhorse of modernbiotechnology, hasbecome a promising

host for the microbial production of biofuels due to the ease of genetic

manipulation, high yields, capability of utilizing a wide variety of

substrates, and high growth and metabolic rates (Ingram et al., 1987;

Ohta et al., 1991; Wang et al., 2008). However, it should be noted that

compared to S. cerevisiae, its ability to produce high concentrations of

ethanol, and tolerance to ethanol and other inhibitory compounds of

lignocellulosic biomass is lower and not commercially viable for E. coli

strains, pointing to the need to employ systems metabolic engineering

for strain development in future.

In addition to glucose andxylose,E. coliis capable of utilizing glycerol.

These substrates are the most readily available renewable feedstock.

With the increasing international biodiesel production, a surplus of

crude glycerol has been generated resulting in a signicant price reduc-

tion. Value-added chemical production using glycerol as feedstock has

become attractive recently. In addition to low cost and abundance, the

higher degree of reduction compared to sugars such as glucose and

xylose makes glycerol be an excellentpotential carbon source to produce

chemicals with high yield (Clomburg and Gonzalez, 2010). Previously, it

Glucose

Pyruvate

2-Ketoisolvalerate

Isopropanol

Isobutanol

2-Methyl-1-butanol

1-PropanolL-Threonine

2-Ketobutyrate

2-Keto-3-

methyl-valerate

2-Ketovalerate

1-Butanol

2-Keto-4-methyl-

pentanoate

3-Methyl-1-butanol

Kivd ( Ll)

adh2 (Sc)

Valine

biosynthesis

Isoleucine

biosynthesis

Leucine biosynthesis

Norvaline

biosynthesis

Nonfermentative

pathways

Pyruvate

Acetyl-CoA Acetaldehyde

Acetate

Acetoacetyl-CoA

1-Butanol

Acetoacetate

3-Hydroxybutyl-CoACrotonyl-CoA

Butyryl-CoA

Butyraldehyde

Fermentative

pathways

atoB ( Ec)

thl (Ca)hbd (Ca)bcd-etfBA (Ca)

crt (Ca)

adhE2 (Ca)

adhE2 (Ca)

atoAD (EC) ctfAB (Ca)

Acetoneadc (Ca) adh (Cb)

adhE ( Ec)adhE ( Ec)

Ethanol

pdc ( Zm)

adhB ( Zm)

Kivd ( Ll)

adh2 (Sc)

Kivd ( Ll)

adh2 (Sc)

Kivd ( Ll)

adh2 (Sc)

kivd (Ll)

adh2 (Sc)

Fig. 1. Fermentative pathways and nonfermentative pathways in E. coli for the production of a number of biofuels (Clomburg and Gonzalez, 2010; Shen et al., 2011). Various syn-

thetic and metabolic strategies have been successfully employed for the enhanced production of candidate biofuel compounds (shown in the brown shaded boxes). Relevant re-

actions are represented by the name of the genes and enzymes. Abbreviations of genes and enzymes: adc , acetoacetate dehydrogenase; adh, secondary alcohol dehydrogenase;

adhB, alcohol dehydrogenase; adhE , acetaldehyde/alcohol dehydrogenase; adhE2, secondary alcohol dehydrogenase; atoAD, acetyl-CoA:acetoacetyl-CoA transferase; atoB,

acetyl-CoA acyltransferase; bcd, butyryl-CoA dehydrogenase; crt , crotonase; ctfAB, acetoacetyl-CoA transferase; etfBA, electrotransfer avor protein; hbd, β-hydroxy butyryl-CoA dehy-

drogenase; pdc , pyruvate decarboxylase; thl, acetyl-CoA acyltransferase; kivd, ketoisovalerate decarboxylase; adh2, alcohol dehydrogenase; Cb, C. boidinii; Ca, C. acetobutylicum; Ec , E.

coli. Ll, L. lactis; Sc , S. cerevisiae.



6/24

was thought that wild type E. coli cannot convert glycerol into ethanol in

chemically dened medium under anaerobic condition because of lack

of electron acceptors, which hampers the potential use of glycerol as

carbon source in bioprocesses (Booth et al., 2005; Shams Yazdani and

Gonzalez, 2008). However, recombinant strains producing fuels and

other high-value reduced chemicals using glycerol as substrate without

requiring external electron acceptors have been reported by inducing

the dormant, native 1,2-propanediol fermentative pathway in E. coli

(Dharmadi et al., 2006; Gonzalez et al., 2008; Murarka et al., 2008).Dharmadi et al. (2006) demonstrated that glycerol can be converted

into reduced compounds such as ethanol and succinate by E. coli

under anaerobic and acidic conditions, and the activity of the formate

hydrogen-lyase and F(0)F(1)-ATPase systems were also found to facili-

tate the fermentative metabolism of glycerol for this process (Gonzalez

et al., 2008). Based on this investigation, they engineered E. coli for the

ef cient conversion of crude glycerol to ethanol, (Shams Yazdani and

Gonzalez, 2008). Recombinant E. coli overexpressing glycerol dehydro-

genase and dihydroxyacetone kinase and deleting fumarate reductase

and phosphate acetyltransferase produced ethanol-hydrogen from

unrened glycerol at yields exceeding 95% of the theoretical maximum

and specic rates inthe order of15–30 mmol/g cell/h. These resultsare

superior to previous reported for the conversion of glycerol to ethanol-

H2 or ethanol-formate by other organisms and equivalent to those

achieved in the production of ethanol from sugar using engineered

E. coli strains (Shams Yazdani and Gonzalez, 2008). Another example

involved in converting glycerol to ethanol was reported with a novel

approach by employing oxygen as the electron acceptor in well dened

microaerobic culture conditions (Trinh and Srienc, 2009). In this study,

rational design with minimized metabolic functionality tailored to

ef ciently convert glycerol to ethanol using elementary node analysis

and adaptive laboratory evolution were implemented sequentially. As

a result, the strain converted 40 g/L glycerol to ethanol in 48 h with

90% of the theoretical ethanol yield (Trinh and Srienc, 2009).

2.3. Advanced biofuels production using recombinant E. coli as ef cient

biocatalyst

Ethanol, as a traditional bulk chemical manufactured from variousfeedstocks, is at present the main biofuel by volume as mentioned

above. However ethanol has a number of limitations as a fuel because

of its high hygroscopicity, low energy density, corrosiveness and

incompatibility with existing fuel infrastructure (Atsumi and Liao,

2008b; Zhang et al., 2011). Fortunately,other advancedbiofuels, typically

higher alcohols (C4 and C5) such as butanol and isopentanol, terpenes

and fatty acid ethyl esters, have similar properties to current petroleum

based transportation fuels which could overcome the problems associ-

ated withthe utilizationof bioethanol (Atsumi and Liao, 2008b). Never-

theless, advanced fuels production by biological processes is still not

commercially feasible, except 1-butanol produced by Clostridium spe-

cies ( Jones and Woods, 1986), because of the absence of robust and

ef cient biocatalysts. Therefore, much more research attention has

recently been dedicated to the development of microbial advancedfuels. For example the fermentative and 2-keto acid pathways of

E. coli were employed to develop a catalystfor advanced fuels production

(Fig. 1) with 2-keto acid route being paid more attention for medium-

chain alcohols production (Atsumi and Liao, 2008b; Zhang et al., 2011).

Signicant progress in microbial advanced fuels production through

metabolic engineering of E. coli is described below.

2.3.1. 1-Butanol and 1-propanol production

E. coli lacks some of the relevant pathways for advanced fuels

production and metabolic engineering has afforded the opportunity

to produce non-traditional biofuels through the construction of non-

native biosynthesis pathways (Atsumi and Liao, 2008b; Atsumi et al.,

2008b). 1-Butanol is attractive as an alternative biofuel with high

energy density and favorable compatibility with the potential to

completely replace gasoline. Traditionally, 1-butanol was produced by

C. acetobutylicum together with acetone, butyrate and ethanol ( Jones

and Woods, 1986). However, the complex physiology and absence of

genetic tools has limited the ability to improve the bacterium to a level

that application in a modern industrial process is feasible (Atsumi

et al., 2008a). Dueto accumulatedknowledge of the1-butanol metabolic

pathway, it has become feasible to reconstruction of butanol pathway in

E. coli. In an example, Atsumi et al. (2008a) constructed a fermentatively

producing 1-butanol E. coli strain through transferring a set of six genes(thl, hbd, crt , bcd, et fAB, adhE2) from C. acetobutylicum1-butanol pathway

and deletion of the competingpathway.This engineered strain produced

1-butanol up to 552 mg/L from a rich medium under semi-anaerobic

conditions. While the product titer was low, the achievement has

paved a novel way for butanol production in a non-native host. To

achieve a higher titer and productivity, the substitution of two key

genes involved in 1-butanol pathway has been explored (Shen et al.,

2011). The replacement of the putative NADH-independent butyryl-

CoA dehydrogenase with the irreversible NADH-dependent trans-

enoyl-CoA reductase, and Clostridium acetoacetyl-CoA thiolase with the

E. coli acetyl-CoA acetyltransferase, and deletion of the genes involved

in mixed-acid fermentation reactions, resulted in a strain that produced

30 g/L butanol with 70 to 88% of the theoretical yield when grown

aerobically followed by anaerobic fermentation. Similar approaches of

gene replacement and balancing of redox cofactors were employed by

Bond-Watts et al. (2011) to redirect more ux towards butanol produc-

tion. This strain produced 4.65 g/L butanol with a yield of 28% from

glucose.

Recently alternative systematic approaches were implemented to

produce 1-butanol and 1-propanol by utilizing the amino-acid biosyn-

thetic pathways have been reported (Fig. 1; Atsumi and Liao, 2008a;

Shen and Liao, 2008). 1-Propanol, a potential gasoline substitute, is a

general solvent with many industrial applications and can also be

converted to diesel fuels and propylene. However, it is dif cult to

produce 1-propanol in signicant quantities using native organisms

(Shen and Liao, 2008). 2-Ketobutyrate, a precursor of isoleucine, not

only acts as a precursor of 1-propanol, but also can be converted to

1-butanol and 2-methyl-1-butanol via multi-steps enzymes reactions.

Therefore, E. coli was engineered to co-produce 1-butanol and 1-propanol through this alternative route via 2-ketobutyrate. By over-

expression of kivd (L. lactis), adh2 (S. cereviasiae), and the E. coli ilvA,

leuABCD, thrA fbBC , followed by deletion of the competing genes includ-

ing tdh, metA, ilvI , ilvB, and adhE , an engineered strainwas created (Shen

and Liao, 2008). This strain produced 2 g/Lbutanol and propanol with a

titer of nearly1:1 ratio but with an undesirable concentration, of

fermentative by-products. In another example, a more direct route to

synthesize 2-ketobutyrate via citramalate pathway was engineered by

directed evolution of Methanococcus jannaschii citramalate synthase

(CimA) to produce 1-propanol and 1-butanol (Atsumi and Liao,

2008a). Error-prone PCRand DNAshuf ingwere carried out to enhance

the CimA specic activity and the resulting engineered E. coli strain

harboring the best CimA variant produced more than 3.5 g/L propanol

and 524 mg/L butanol. However production of other side products toa signicant level remains a problem.

2.3.2. 2-Methyl-1-butanol and 3-methyl-1-butanol production

2-Methyl-1-butanol and 3-methyl-1-butanol, ve-carbon alcohols,

possess similar properties of lower vapor pressure and high energy

density. Compared to ethanol, they are more compatible with existing

fuel infrastructure and more suitable to replace gasoline. S. cerevisiae

can naturally produce 2-methyl-1-butanol and 3-methyl-1-butanol in

minor amounts as by-products, and some metabolic approaches have

been attempted to increase 2-methyl-1-butanol production in yeast

(Abe and Horikoshi, 2005). However, production of these ve-carbon

alcohols by E. coli is thought to be a more desirable process. Thus,

E. coli was engineered to produce 3-methyl-1-butanol and 2-methyl-

1-butanol through manipulation of the valine and leucine biosynthesis



7/24

pathways, and the isoleucinebiosynthetic pathway, respectively (Fig. 1)

(Cann and Liao, 2008). Using the isoleucine biosynthesis pathway, a

recombinant E. coli overexpressing native thrABC of the isoleucine path-

way and the non-native ilvGM , ilvA, kivd and adh2 with the competing

pathways deleted resulted in a strain that produced 1.25 g/L 2-

methyl-1-butanol from glucose in 24 h. With a similar strategy,

Connor and Liao (2008) constructed a 3-methyl-1-butanol producing

E. coli strain via the native amino acid biosynthetic pathways. Upon

the removal of feedback inhibition and competing pathways, this strainproduced 3-methyl-1-butanol with a titer of 1.28 g/L from glucose in

28 h. Furthermore, this recombinant strain was employed to enhanced

3-methyl-1-butanol production by random mutagenesis and selection

(Connor et al., 2010). The resulting strain produced 9.5 g/L 3-methyl-

1-butanol with a yield of 0.11 g/g glucose when a two-phase fermenta-

tion process was used.

2.3.3. Isopropanol and isobutanol production

Isopropanol, one of the simplest secondary alcohols extensively

used in various applications, can be naturally produced by organisms

such as Clostridium. Although some attempts have been made to

improve the production ability of this strain, low titer and product

inhibition remain obstacles. Therefore, metabolic engineering of E. coli

for isopropanol production become a potential alternative industrial

route.Recently, Hanai et al. (2007) reported recombinantE. coli produc-

tion of isopropanol using non-native fermentative pathways (Fig. 1).

Upon introduction of a set of genes in optimal combination from differ-

ent sources, the pathway from acetyl-CoA via acetone to isopropanol

was established in an E. coli strain that produced 4.9 g/L isopropanol.

Moreover, this research group signicantly increased the titer up to

143 g/L with a molar yield of 67% (isopropanol/glucose) by the imple-

mentation of a gas trapping process, which promptly removed

isopropanol from culture broths to alleviate product toxicity to cell

(Inokuma et al., 2010). Notwithstanding the long fermentation time of

240 h, the recombinant E. coli produced the highest titer to date and is

much higher than the native clostridial strains. Introducing a similar

biosynthetic pathway described above, Jojima et al. (2008) constructed

a genetically engineered strain of E. coli that produced a 13.6 g/L

isopropanol from glucose under vigorous aerobic culture conditions.Isobutanol has similar physical properties to isopropanol, but

possesses a higher octane number. As an intermediate in valine

biosynthesis, E. coli production of isobutanol was also engineered

through the synthetic nonfermentative pathways via 2-ketoisovalerate

as precursor (Fig. 1) (Atsumi et al., 2008b; Savrasova et al., 2011). To

increase 2-ketoisovalerate pool, the Bacillus subtilis alsS gene encoding

acetolactate synthase, and E. coli ilvCD gene encoding acetohydroxy

acid isomeroreductase, and dihydroxy-acid dehydratase, were over-

expressed, and those pathways competing for pyruvate and cofactors

were deleted. Combined with overexpression of L. lactis kivd encoding

2-ketoacid decarboxylase, and S. cerevisiae adh2 encoding alcohol dehy-

drogenase, this engineered strain produced up to 22 g/L isobutanol at

86% of the theoretical yield after 110 h under micro-aerobic conditions

(Atsumi et al., 2008b). Comparative studies on the alcohol dehydroge-nases (S. cerevisiae adh2, L. lactis adhA and E. coli yqhD) to convert

isobutyraldehyde to isobutanol in E. coli, both of which indicated that

overexpression of yqhD or adhA showed better results than adh2 in

isobutanolproduction(Atsumiet al., 2010). With a differentculture pro-

cess, Bastian et al. (2011) created an E. coli strain that produced 13.4 g/L

isobutanol with 100% theoretical yield under anaerobic conditions. This

strain was manipulated by overexpression of recombinant ketol-acid

reductoisomerase and alcohol dehydrogenase, with only NADH cofactor

specicity to achieve a redox balance under anaerobic conditions.

3. Organic acids

Organic aids are an important component of the building-block

chemicals that can be produced by microbial processes. Most

microbial organic acids are end-products, or at least natural interme-

diates in major metabolic pathways. Because of their functional

groups, organic acids extremely useful as starting materials for the

chemical, food and feed industries (Sauer et al., 2008). However, a

low ef ciency of bio-conversion of sugar to the desired product

together with a low yield and productivity limits the current market

for some organic acids dominated by the petrochemical industry.

Once a competitive fermentation technology based on strain improve-

ment for these acids is established, the market for microbial producedacids should increase (Bozell and Petersen, 2010). In this section, recent

critical advances in strain development for organic acid production

by metabolic engineering approaches using E. coli as a platform

are discussed with emphasis on lactic acid, succinic acid and 3-

hydroxypropionic acid as examples.

3.1. Lactic acid production

Lactate (D- or L -), an important organic acid, is a bulk chemical

increasingly used in the food and pharmaceutical industries in products

such as food, beverages and biodegradable polymers and currently

produced at 150,000 tons per year (Sauer et al., 2008). Previously,

Lactobacilli were commonly employed forindustrialproduction of lactic

acid but had drawbacks of requiring complex growth nutrients, an

optically impure product, a mixed acid fermentation, and poor ability

to utilize pentoses (Wendisch et al., 2006). In order to expand the use

of lactic acid especially in biopolymers, it is necessary to improve the

optical purity of the acid required for polylactide use in bio-based plas-

tics (Narayanan et al., 2004; Wee et al., 2006), and to reduce production

costs. Therefore,E. coli strains have beendeveloped for effective produc-

tion of D- and L -lactic acid using a metabolic engineering approach.

In E. coli, D-lactate is produced by homofermentative or

heterofermentative pathways together with NADH consumption in

order to achieve a redox balance under anaerobic culture conditions

(Fig. 2). However, in E. coli the pathway for production of L -lactate is

absent. E. coli has been successfully manipulated to produce optically

pure lactate with the potential for large scale commercial production

(Table 1; see reviews by Okano et al., 2010; Sauer et al., 2008;

Wendisch et al., 2006; Yu et al., 2011). In a rst successful example of D- or L -lactate production by engineered E. coli, Chang et al. (1999)

created two strains, one which produced D-lactate with a titer of

62.2 g/L and one which produce L -lactate production with a concentra-

tion of 45 g/L. Two genes pta and ldhA were deleted and L -lactate dehy-

drogenase gene from Lactobacillus casei was overexpressed in the

L -lactate producing strain described above. Dien et al. (2002)engineered

E. coli strains to produce L -lactate with utilizing the mixtures of glucose

and xylose as substrates. A ptsG-negative but glucose-positive E. coli

strain was employed to produce L -lactic acid by deletion of the p and

d-ldh, with heterologous expression a L -lactate dehydrogenase gene.

The resulting recombinant strains produced L -lactate with a yield of

0.77 g/g sugar mixtures and low amounts of by-products. Similarly, a

series of D-lactate over-producing E. coli strains were created by many

investigators. Zhou et al. (2003) constructed derivatives of E. coliW3110 for D-lactate production by elimination of competing pathways

including mutation genes of frdABCD, adhE , p B, and ackA. The resulting

strains produced D-lactate with the theoretical maximum yield of

2 mol/mol glucose. Moreover, the chemical purity and the optical purity

of above acid was up to 98% and 99%, respectively. However, these

strains showed a defect of incomplete metabolism of high sugar concen-

trations (more than 50 g/L). Accordingly, they constructed SZ132 strain

from an ethanologenic E. coli KO11, which rapidly fermented 100 g/L

sugars to produce D-lactate over 90 g/L in rich medium containing beta-

ine as a protective osmolyte (Zhou et al., 2005). Because by-products

were increasingly produced and the poor performance in mineral salts

medium by the SZ132 strain, the strategies of metabolic evolution in

conjunction with genetic manipulations were carried out to yield an

over-producing strain E. coli SZ194 (Zhou et al., 2006). This strain



8/24

produced D-lactate with a titer of 110 g/L, a yield of 0.95 g/g glucose

but only trace amounts of co-products in mineral salts medium

supplemented with 1 mM betaine. However, the chiral impurity of 5%

implicated that purication will be required adding to production cost.

A higher titer of D-lactate with 138 g/L was achieved by metabolic

engineering strain E. coli ALS974 (aceEF , p , poxB, pps, frdABCD) using a

dened mediumand dual-phases process(aerobicgrowth and anaerobic

production) (Zhu et al., 2007). Overall productivity of 3.5 g/L/h, and an

overall yield of 0.86 g/g glucose were reported in a feed-batch fermenta-

tion. Noteworthy was that this strain necessitated the use of acetate to

form pyruvate during aerobic phase and that the reduction in the forma-

tion of succinic acid depended on the precise control of the acetate

concentration.

Zhou et al. (2012a) manipulated a wild-type E. coli B0013 by

deletion of eight genes (ackA-pta, pps, p B, dld, poxB, adhE , and frdA) to

over-produce D-lactate. Coupling lactate fermentation with cell growth

was employed and the cells in the bioreactor yielded an overall volu-

metric productivity of 5.5 g/L/h and a yield of 86 g lactic acid/100 g

glucose which were 66% higher and at the same level compared to

that of the aerobic and oxygen-limited two-phase fermentation, respec-

tively. These results have revealed an approach for improving produc-

tion of fermentative products in E. coli. To the knowledge of the

authors, this was among the highest D-lactate levels reported for

engineered E. coli and shows favorable comparison with production by

lactate bacteria. Noteworthy is that strain B0013 was selected for its

properties of rapid growth, high tolerance to lactate, and ease of genetic

glpABC

glpK

glpD

glucose

ptsG

PEP

pyruvateglucose 6-P

galP

glk

2PEP

pyruvate

pykA pykF

acetyl CoA

pps

pdh pflB

formate

lactate

ldhA

poxB

acetate pta-ackA

ethanol

adhE

o x a l o a c e t a t e

i s o c i t r a t

e

c i t

r a t e

2 - o x o g l u t a r a t e

s

u c c i n y l - C o A

s u c c i n

a t e

f u m a

r a t e

malate glyoxyate aceA aceB

c i t

i

t

y l -

ppc

pyc

s d h A

B

pck

m d h

f u m A B C

f r d

A B C D

glycerol

gldA

dhaKLM

DHAP

pyruvate

PEP

m a

e A

m a e B

glpF

Fig. 2. Lactic acid and succinic acid pathways from glucose and glycerol in E. coli and enzymes involved in metabolic regulation (Blankschien et al., 2010; Wendisch et al., 2006;

Zhang et al., 2010; Zhou et al., 2011b). The three pathways for succinic acid production are indicated by the thick blue, red, and purple arrows, respectively. Relevant biochemical

reactions are represented by the names of the gene(s) coding for the enzymes (all E. coli genes unless otherwise specied): aceA: isocitrate lyase; aceB: malate synthase; ackA,

acetate kinase; adhE , acetaldehyde/alcohol dehydrogenase; dhaKLM , dihydroxyacetonekinase; frdABCD, fumarate reductase; fumABC , three isoenzymes of fumarases; gldA, glycerol

dehydrogenase; glk, glucose kinase; glpABC , anaerobic glycerol-3-phosphate dehydrogenase; glpF , glycerol facilitator; glpK , glycerol kinase; ldhA, D-lactate dehydrogenase; maeA/

maeB, NADH/NADPH malice enzymes, respectively; mdh, malate dehydrogenase; pck, phosphoenolpyruvate carboxykinase (E. coli or A. succinogenes); pdh, pyruvate dehydrogenase

complex; p B, pyruvate formate-lyase; poxB, pyruvate oxidase; ppc , phosphoenolpyruvate carboxylase; pps, PEP synthase; pta, phosphoacetyl transferase; ptsG, phosphotransferase

system; poxB, pyruvate oxidase; pyc , L. lactis pyruvatecarboxylase; pykA/ pykF , pyruvate kinase.



9/24

manipulation compared to other standard strainssuch as E. coli K12andE. coli 3110. More recently, a genetic switching approach was developed

for highly ef cient D-lactate production via ne-tuning of the D-lactate

dehydrogenase gene in which the chromosomal native promoter was

replaced with temperature-sensitive λ pR and pL promoters (Fig. 3)

(Zhou et al., 2012b). By combining the elimination of competing path-

ways for by-product and cofactor formation and temperature-shifting

the fed-batch fermentation process, a pure D-lactate titer of 122.8 g/L

from glucose was achieved with overall volumetric productivity of 4.32 g/L/h and the oxygen-limited productivity of 6.73 g/L/h. These

values are amongst the highest levels produced by a recombinant

E. coli strain without adding special substrates and obligatory anaerobic

conditions. (Zhou et al., 2012b). Recently,the fermentation performance

of this strain has been evaluated for D-lactate production in a 1000 L-

fermentor and satisfactory results similar to those at laboratory scale

were obtained (unpublished data).

As mentioned above, crude glycerol has become an ideal feedstock

for the production of bio-based chemicals. Therefore, engineering

E. coli strains to produce lactate using glycerol as carbon source has

received attention. Mazumdar et al. (2010) created a recombinantE. coli strain by overexpressing pathways involved in the conversion

of glycerol to D-lactate and the elimination those leading to the synthe-

sis of competing by-products. Thus, an engineered homofermentative

E. coli strain was developed that converted 40 g/L glycerol to 32 g/L D-lactate with a yield of 85% of the theoretical maximum. Similarly a

metabolically engineered E. coli strain was created to produce D-lactate

from glycerol with a minimum of by-products in our lab (unpublished

data). With the elimination of by-product (formate, acetate, ethanol

and succinate etc) pathways and overexpression of lactate dehydroge-

nase gene, the resulting strain produced 114.4 g/L D-lactate with an av-

erage productivity of 3.95 g/L/h and yield of 0.78 g/g glycerol in 7-l

bioreactor with a modi

ed two-phase fermentation process. Moreover,only 1.8 g/L acetate was accumulated and no detectable succinate,

pyruvate, formate and ethanol were found (unpublished data). In addi-

tion, to expand the substrate range, Eiteman et al. (2008, 2009)

constructed two substrate-selective E. coli strains, one of which is

unable to consume glucoseand one which is unable to consume xylose.

These two recombinant strains were used to simultaneously convert

xylose and glucose sugar mixtures to lactateby a co-fermentation strat-

egy in a single process and the two-sugar mixture was ef ciently

converted into 37 g/L lactate with a lactate-sugar yield of 0.88 g/g.

Other products such succinic acid and acetate generated with lower

concentration during anaerobic conditions (Eiteman et al., 2009).

Although strain improvement of E. coli as a extensive platform was

achieved through genetic manipulation, a number of hurdles still need

to be crossed before the commercial production of D-lactate by an

P R P LActive cI ts ldhA

Repression

Inactive cI ts

Glucose

Pyruvate Lactic acid

Biomass

TCA

Byproducts

Glucose

Pyruvate Lactic acid

Biomass

TCA

Byproducts

P R P L ldhA

Transcription

Aerobic, 33 Oxygen-limited, 42

Fig. 3. An ef cient strategy of D-lactate production through metabolic control approach in engineered E. coli with 30 g/L glucose in the initial M9 medium (Zhou et al., 2012a,

2012b). Lactate biosynthesis was inhibited in cells with competing pathways deleted when grown at 33 °C under aerobic conditions due to inactive P R & P L promoters. When

cells were shifted to oxygen-limited and 42 °C conditions, the promoters are activated and D-lactate production increases signicantly. Less substrate was used in TCA cycle and

by-product formation, and much more of the carbon ux was directed to the D-lactate pathway. In the top part, the dotted arrow shows the time when the culture was shifted

from 33 °C to 42 °C. The blue line indicates the time when the culture was shifted from the aerobic cultivation to the oxygen-limited production phase. Additions of glucose of

136.2 (A), 136.2 (B), 150.9 (C) and 136.2 g (D) were made to the bioreactor as indicated by arrows. In the bottom part, the red and grey arrows indicate the active pathways

and inactive pathways, respectively.



10/24

engineered E. coli strain can be applied. For example, the operating cost

including of separation and purication processes is one of bottlenecks

for the biotechnological production of optically pure lactic acid.

Weusthuis et al. (2011) suggested a solution where lactate is

co-produced with ethanol in an existing ethanol factory and separated

during distillation with minimal extra costs. The poor tolerance of lactic

acid producing E. coli strains to low pH requires application of large

amounts of Ca(OH)2, CaCO3 or other bases to maintain a neutral pH

during fermentation process and in some cases (special for calciumions) this could hasbecomea big waste disposal problem in commercial

organic acids distillation. Development of acid tolerant E. coli strains by

directed evolution or systematic metabolic engineering would be pref-

erable for commercial production.

3.2. Succinic acid production

Succinic acid was identied as one of the top 12 building block

chemicals by the U.S. Department of Energy (Bozell and Petersen,

2010). It is not only used in many elds such as agriculture, chemical,

food, and pharmaceutical industries, but also can offer a huge potential

alternative as a C4-dicarboxylic acid platform to be converted to other

chemicals such as 1,4-butanediol, tetrahydrofuran, γ-butyrolactone,

N -methyl pyrrolidinone, and 2-pyrrolidinone (Delhomme et al., 2008;

Sauer et al., 2008; Wendisch et al., 2006). The market potential for

succinic acid and its immediate derivatives was estimated to be up to

245,000 tons per year (Bozell and Petersen, 2010) but most commer-

cially available succinic acid is presently produced by a petrochemical

process which generated negatively economicaland ecological impacts.

Excitingly, the signicant progress in the development of an economi-

cally competitive route of succinic acid production from renewable re-

sources has been made. Commercial succinic acid production via the

industrial biotechnology route has been reported by the Requette

(http://www.dsm.com/en_US/downloads/media/12e_09_dsm_and_

roquette_commercialize_bio_based_succinic_acid.pdf .) and Bioamber

(http://www.bio-amber.com/press_releases.php.) companies. Many

literature reports on succinic acid production from renewable sub-

strates have been published including a number of processes based on

engineered E. coli strains ( Jantama et al., 2008a; Lee et al., 2005).E. coli naturally produces succinic acid in a minor amount as an

intermediate of the central metabolism or as a fermentation end-

product via the reductive tricarboxylic acid (TCA) branch (Fig. 2). There

are three alternative routes to form succinic acid: the PEP-pyruvate-

oxaloacetate node (the reductive TCA branch), the oxidative TCA branch

and the glyoxylate shunt, respectively (Fig. 2), each of which is involved

in redox balance and energy regeneration. The formation of 1 mol of

succinic acid via the reductive pathway is calculated to consume 2 mol

of NADH and 1 mole of CO2, whereas 1 mol of glucose can furnish only

2 mol of NADH through the glycolytic pathway (Fig. 2). Therefore, the

redox balance and high succinic acid yield are closely tied to the pool of

NADH and these existing complex metabolic networks and regulation

mechanisms requires alternative strategies of metabolic engineering

of E. coli for succinic acid production to be considered.A feasible alternative strategy is to engineer E. coli to anaerobically

overproduce succinic acid by amplication of phosphoenolpyruvate

carboxylase (PPC) which catalyzes PEP to oxaloacetate (OAA) with

CO2 consumption (Kim et al., 2004; Millard et al., 1996). Stols and

Donnelly (1997) investigated whether the amplication of the malic

enzyme gene would restore fermentative metabolism of glucose and

produced succinic acid as a major fermentation product in E. coli

NZN111, a strain unable to ferment glucose because of inactivation of

the genes ldhA and p . When recombinant E. coli NZN111 harboring

malic enzymegene was cultured in LB medium containing 20 g/L sorbi-

tol as a more reduced carbon source under a CO2 atmosphere, 10 g/L of

succinic acid was produced (Hong and Lee, 2002). These results were

consistent with the prediction of in silico metabolic ux analysis (Lee

et al., 2002), which indicated that redox balancing was important for

the enhanced production of succinic acid. Subsequently, inactivation

of ptsG gene, encoding the membrane-bound protein involved in glu-

cose transport, not only partly restored the ability of NZN111 strain to

grow fermentatively on glucose, but also allowed this mutant strain to

yield higher succinic acid concentrations under anaerobic conditions

(Chatterjee et al., 2001). Using an optimized dual-phase fermentation,

the recombinant strain NZN111 ( ptsG inactived, pyc gene over-

expressed) produced 99.2 g/L succinic acid with an overall yield of

110% on glucose and productivity of 1.3 g/L/h ( Vemuri et al., 2002),which is the highest titer produced by a recombinant E. coli reported

in the literature. Furthermore, overproduction of succinic acid was

investigated by inhibition of byproduct formation (Lee et al., 2005),

diversion of NADH and provision of additional precursor for succinic

acid synthesis (Sanchez et al., 2005), as well as optimization of the

fermentation process ( Jiang et al., 2010; Wu et al., 2007). Interestingly,

various studies indicate that CO2, which is incorporated into thecarbon

backbone to form OAA from PEP via carboxylation of PPC, affected

succinic acid accumulation through the reductive TCA branch under

anaerobic fermentation process (Lu et al., 2009).

Systems metabolicengineeringwas appliedto improvesuccinicacid

production. Using the combination of genome-scale in silico optimiza-

tion and metabolic ux analysis, remarkable improvement of succinic

acid production in E. coli was achieved with a yield of 1.29 mol/mol

glucose (Wang et al., 2006). In another example, Lee et al. (2005)

compared the genomes of E. coli and succinic acid-overproducing

Mannheimia succiniciproducens, and constraints-based ux analysis

was carried out to improve the succinic acid production of E. coli,

which strategy increased the succinic acid production by more than

sevenfold and the ratio of succinic acid to fermentation products by

ninefold. Although succinic acid yield was not signicantly increased,

the results suggest that this approach can be an ef cient way of devel-

oping strategies for strain improvement.,

Recently, directed evolution in combination with genetic engi-

neering was carried out to achieve a succinic acid overproducing

strain with energy-conserving pathways, in which pck was cloned and

evolved to the major carboxylation pathway (PEP + CO2 + ADP =

oxalaacetate + ATP) for succinate production with the CO2 xation

and generation of ATP. In addition the native PEP-dependentphosphotransferase system (PTS) for glucose uptake was inactivated

( Jantama et al., 2008a; Zhanget al., 2009). Owing to no additional reduc-

ing equivalent consumed from glucose to succinic acid, this evolved

strain not only increased the pool of PEP available for redox balancing

but also increased energy ef ciency, thus leading to enhanced succinic

acid production with a ATP-generating pathway, which naturally exists

in succinic acid-producing rumen bacteria (Lee et al., 2006). Moreover,

this evolved strain was further engineered to improve succinic acid

production by deletion of the genes encoding the acetate, malate, and

pyruvate pathways. The strain produced 700 mM succinic acid with a

high yield of 1.5 mol/mol glucose ( Jantama et al., 2008b), which are

comparable to the best natural succinic acid-producing rumen bacteria

such as Actinobacillus succinogenes. In another similar example, to main-

tain the redox and ATP balance, Singh et al. (2011) cloned the geneencoding the ATP-generating PEPCK enzyme in combination with elim-

ination of ethanol and acetate forming to improve succinic acidsynthesis

in the ldhA, p B, ptsG mutant strain, and the resulting E. coli strains

produced succinic acid with a 60% increase compared to the a control

strain.

Another strategy of metabolic engineering has been carried out to

overproduce succinic acid under aerobic conditions (Lin et al., 2005a,

2005b, 2005c). In a successful example, they constructed two aerobic

central metabolism routes for succinic acid production, one of which

is a glyoxylate shunt and the other is an oxidative TCA branch for

aerobic succinic acid production, by inactivation of genes sdhAB, iclR, poxB, and ackA-pta (Lin et al., 2005c). The resultant strain was further

engineered by deletion of ptsG and overexpression of pepc gene from

Sorghum vulgare, to give an over-producing strain. Using a fed-batch


http://www.dsm.com/en_US/downloads/media/12e_09_dsm_and_roquette_commercialize_bio_based_succinic_acid.pdfhttp://www.dsm.com/en_US/downloads/media/12e_09_dsm_and_roquette_commercialize_bio_based_succinic_acid.pdfhttp://www.bio-amber.com/press_releases.phphttp://www.bio-amber.com/press_releases.phphttp://www.dsm.com/en_US/downloads/media/12e_09_dsm_and_roquette_commercialize_bio_based_succinic_acid.pdfhttp://www.dsm.com/en_US/downloads/media/12e_09_dsm_and_roquette_commercialize_bio_based_succinic_acid.pdf


11/24

process, the above strain produced 58.3 g/L succinic acid with a yield

of 0.94 mol/mol glucose and a productivity of 1.08 g/L/h under strict-

ly aerobic conditions (Lin et al., 2005a), which pointed to a promising

system for large-scale succinic acid production. Alternatively, an

engineered E. coli strain, using sucrose as carbon source to aerobically

produce succinic acid with a mol yield up to of 1.90, was achieved by

overexpression of scr genes (scrK , Y , A, B, and R genes converting su-

crose to β-D-fructose and α-D-glucose 6-phosphate) and Lactococcus

lactis pyc gene (Wang et al., 2011b).To expand the range of available substrates and ef ciently convert

inexpensive feedstock to high value bio-based products such as

succinic acid, glycerol, sucrose, fructose and/or a mixture sugar were

tested to produce succinic acid by various genetically modied E. coli

strains (Blankschien et al., 2010; Kang et al., 2011; Wang et al., 2011a,

2011b). In a successful example, Blankschien et al. (2010) created an

E. coli strain by overexpression of the L. lactis pyruvate carboxylase

gene and by blocking pathways for the synthesis of by-products,

which produced succinic acid with a yield of about 0.69 g/g glycerol

from 20 g/L glycerol in 72 h, on par with the use of glucose as a feed-

stock. Similarly, Zhang et al. (2010) created an E. coli strain able to

ef ciently convert glycerol to succinic acid without introduction of

non-native E. coli genes. Upon mutational activation of phosphoenol-

pyruvate carboxykinase gene ( pck*), and inactivation of ptsI (encoding

PtsI of the phosphorelay system) and p B gene (encoding pyruvate

formate-lyase), the resulting engineered strain converted 128 mM

glycerol to 102 mM succinic acid with 80% of the maximum theoretical

yield during anaerobic fermentation in mineral salts medium. Never-

theless, low growth, poor fermentation performance and long culture

period are major limitations for these strains. In another example,

21.07 g/L succinic acid and 0.54 g/L polyhydroxyalkanoate (PHA) was

produced simultaneously from a mixture of glycerol and fatty acid by

a modied E. coli strain created by Kang et al. (2011). However, similar

to the problem for lactate production mentioned above, the need for

medium neutralization process and the poor acid tolerance of the cell

are major disadvantages for succinic acid production by metabolically

engineered E. coli strains.

3.3. Production of other organic acids

In addition to lactic acid and succinic acid as discussed above,

other organic acids such as pyruvate, acetate and malate are also

important bulk chemicals and have a wide range of industrial applica-

tions. Currently, production of these acids is mainly by traditional

petrochemical-based routes and/or biotechnological processes by wild

type organisms. However, some investigations have recently reported

that metabolically engineered E. coli strains have been employed to

overproduce these acids In one example, Causey et al. (2003) created

an excellent E. coli strain by sequential genetic modication, which

can ef ciently convert sugar to acetate as the primary product. Upon

introduction of the ldhA, p B, frdBC , atpFH , adhE and sucA genes muta-

tions together with the blocking of native fermentation pathways,

oxidative phosphorylation and TCA function, the resulting strainnamed E. coli TC36 produced up to 878 mM acetate with 75% of the

maximum theoretical yield from a mineral salts medium containing

glucose as carbon source. Moreover, the same group further genetically

manipulated the TC36 strain by deletion of poxB and ackA genes to yield

pyruvate strain E. coli TC44 that accumulated up to 749 mM pyruvate

with a yield of 0.75 g/g glucose (Causey et al., 2004). All major nones-

sential pathways consuming pyruvate in this multiple mutated strain

were absentand the utilization of pyruvate for cell growth wasreduced.

In an another example, Tomar et al. (2003) constructed an over-

producing pyruvate E. coli strain by deletion of the gene encoding the

E2p subunit of the PDHC and ADH encoding gene in a PEP carboxylase

decient strain. Using this genetically engineered strain, 425 mM

pyruvate was obtained with a yield of about 0.60 to 0.74 g/g glucose

but some byproducts such as acetate and lactate were also detected.

Therefore, metabolic engineering was further carried out to direct

more carbon to pyruvate by elimination of intracellular conversion

of pyruvate to acetyl-CoA, PEP, acetate and lactate. Hence an

overproducing E. coli was achieved by which 1 mol glucose was

converted to 1.78 mol pyruvate, with titers as high as 110 g/L (Zelic

et al., 2004). Zhu et al. (2008) created an overproducing pyruvate

E. coli strain ALS1059 by mutations in the aceEF , p , poxB, pps, ldhA,

atpFH and arcA genes which encode, respectively, the pyruvate

dehydrogenase complex, pyruvate formate lyase, pyruvate oxidase,phosphoenolpyruvate synthase, and lactate dehydrogenase. Using this

strain, 90 g/L (more than 1 M) of pyruvate production with an overall

productivity of 2.1 g/L/h and yield of 0.68 g/g was achieved with the

fed-batch process.

Furthermore, E. coli also has been engineered to produce some

high-value organic acids such as shikinic acid (Escalante et al., 2010;

Johansson et al., 2005), glucaric acid (Moon et al., 2009, 2010) and 3-

hydroxypropanoic acid (3-HP) (Mohan Raj et al., 2009; Rathnasingh

et al., 2009), which generally are secondary metabolites that are pro-

duced in much lower amounts than the previously mentioned acids.

These organic acids are important compounds used in many elds

such the pharmaceutical and chemical industries.

4. Amino acids production

Amino acids are important bioproducts with extensive industrial

applications in the antibiotic, pharmaceutical, animal feed, and cosmetic

industries as well as in food additives. More than two million tons of

amino acids are produced annually (Park and Lee, 2008) with annual

growth rate of 5–7% (Lee et al., 2007) including more than 500 tons of L -valine, 8000 tons of L -phenylalanine, 1,000,000 tons of L -glutamate

and, and over 4000 tons of L -threonine by fermentation (Ikeda, 2003).

Traditional strategies of strain development for amino acid production

were multi-round random mutation and selection procedures. Because

of the complicated and highly regulated metabolic network, these

approaches have played the major role in development of industrial

amino acid producers in recent decades. However, such approaches

inevitably or unintentionally result in unwanted mutations not directly

related to the target metabolite which have undesirable effects on thephysiology of the organism and retard growth (Park et al., 2007).

Targeted metabolic engineering of E. coli and other organisms has

been applied to strain improvement with the purpose of modifying

specic genes and pathways to enhance production of a desired product

(Clomburg and Gonzalez, 2011; Lee et al., 2004; Nikel et al., 2010; Wang

et al., 2011a; Yi et al., 2002). These rational engineering strategies have

played a critical role in microbial improvement but the local pathways

or bioreactions of engineered cell are usually emphasized and

genome-wide, global metabolic network is often neglected. In recent

years, with theadvancesof omics technology andgenome-scale compu-

tational biology, systems metabolic engineering (systems biology

combined with metabolic engineering), has been applied in strain

improvement. By integrating information of entire metabolic and regu-

latory networks (Kumar and Shimizu, 2011; Lee et al., 2007; Park andLee, 2008; Park et al., 2007, 2011; Schaub et al., 2008), the problems

associated with random mutation and the dif culty of rationally engi-

neering the complex and highly regulated metabolic network can be

overcome.

Currently, E. coli has been extensively used for the production of

various amino acids (Ikeda, 2003) (Table 1), such as L -valine (Park

et al., 2007, 2011), L -phenylalanine (Baez-Viveros et al., 2004, 2007;

Gerigk et al., 2002), L -alanine (Lee et al., 2004), and L -threonine (Dong

et al., 2011; Lee et al., 2003a, 2009; Song et al., 2000). The metabolic

pathway of important amino acids and their regulator mechanism in

E. coli, are shown in Fig. 4. Here, some recent representative examples

are highlighted with particular emphasis on the advances of rational

metabolic engineering and system-level engineering approaches to

improve amino acid producing E. coli strains.



12/24


13/24

overproduction of L -threonine. In a similar example, an overproducingL -threonine E. coli strain was developed with a systems metabolic

engineering approach (Lee et al., 2007). Initially, negative regulation

including feedback inhibition and transcriptional attenuation regula-

tion was removed by targeted metabolic engineering. Then, the target

genes to be engineered were identied by transcriptomeproling com-

bined with in silico ux response analysis, and their desirable expres-

sion levels were manipulated accordingly. The nal engineered E. coli

strain was able to produce L

-threonine with a signi

cant high yield of 0.393 g L -threonine/g glucose, and 82.4 g/L L -threonine by fed-batch

fermentation. Subsequently, Lee et al. (2009) reported reengineering

of a reduced-genome E. coli for L -threonine production. Using strain

MDS42 lacking 14.3% of its chromosome as host (Posfai et al., 2006), a

feedback-resistant L -threonine operon was over-expressed, the genes

tdh encoding L -threonine dehydrogenase, tdcC and sstT encoding

L -threonine transporters, respectively, were deleted, and a L -threonine

exporter encoded by the mutated gene rhtA23 was introduced. The

resulting strain MDS-205 showed an 83% increase in L -threonine pro-

duction, compared to a wild-type E. coli strain MG1655 engineered

withthe absolutely same modications. Moreover, transcriptional anal-

ysis revealed that elimination of nonessential genes can increase the

productivity of an industrial strain, by reducing the metabolic burden

and improving the metabolic ef ciency of cells. These examples also

indicated that a systems metabolic engineering approach can be

employed to obtain an industrially competitive strain.

4.2. L-Valine production

L -valine, a branch-chainamino acid, hasbeen used in pharmaceuticals,

cosmetics, and feed-additive industries and produced

metabolic engineering of e. coli. a sustainable industrial platform for bio-based chemical...

Documents