Strain- and bioprocess-design strategies to increase ...

Strain- and bioprocess-design strategies to increase production

of (R)-3-hydroxybutyrate by Escherichia coli

Mónica Alejandra Guevara Martínez

Doctoral Thesis in Biotechnology

KTH Royal Institute of Technology School of Engineering Sciences in Chemistry, Biotechnology and Health

Stockholm 2019

© Mónica Alejandra Guevara Martínez Stockholm 2019 KTH Royal Institute of Technology School of Engineering Sciences in Chemistry, Biotechnology and Health Department of Industrial Biotechnology AlbaNova University Center SE-10691 Stockholm Sweden Printed by Universitetsservice US-AB Drottning Kristinas väg 53B SE-11248 Stockholm Sweden ISBN 978-91-7873-205-0 TRITA-CBH-FOU-2019:25

To my parents

Abstract: Microbial bio-based processes have emerged as an alternative to replace fossil-based processes for the production of fuels and chemicals. (R)-3-hydroxybutyrate (3HB) is a medium-value chemical that has gained special attention as a precursor of antibiotics and vitamins, as a monomer for the synthesis of tailor-made polyesters and as a nutritional source for eukaryotic cells. By integrating strain and bioprocess-design strategies the work of this thesis has aimed to improve microbial 3HB production by the well-studied platform organism Escherichia coli (strain AF1000) expressing a thiolase and a reductase from Halomonas boliviensis. Uncoupling growth and product formation by NH4+- or PO43-- limited fed-batch cultivations allowed for 3HB titers of 4.1 and 6.8 g L-1 (Paper I). Increasing the NADPH supply by overexpression of glucose-6-phosphate dehydrogenase (zwf) resulted in 1.7 times higher 3HB yield compared to not overexpressing zwf in NH4+

depleted conditions (Paper II). To increase 3HB production in high-cell density cultures, strain BL21 was selected as a low acetate-forming, 3HB-producing platform. BL21 grown in NH4+ limited fed-batch cultivations resulted in 2.3 times higher 3HB titer (16.3 g L-1) compared to strain AF1000 (Paper III). Overexpression of the native E. coli thioesterase “yciA”, identified as the largest contributor in 3HB-CoA hydrolysis, resulted in 2.6 times higher 3HB yield compared to AF1000 not overexpressing yciA. Overexpressing zwf and yciA in NH4+depleted fed-batch experiments resulted in 2 times higher total 3HB yield (0.210 g g-1) compared to AF1000 only overexpressing zwf (Paper IV). Additionally, using 3HB as a model product, the bacterial artificial chromosome was presented as a simple platform for performing pathway design and optimization in E. coli (Paper V). While directly relevant for 3HB production, these findings also contribute to the knowledge on how to improve the production of a chemical for the development of robust and scalable processes. Keywords: E. coli, (R)-3-hydroxybutyrate, metabolic engineering, bioprocess design, NADPH, acetic acid, thioesterase, BAC

Sammanfattning Mikrobiella biobaserade processer har vuxit fram som ett alternativ till fossila processer för produktion av bränslen och kemikalier. (R)-3-hydroxybutyrat (3HB) är en mellanvärdeskemikalie som har fått uppmärksamhet för användningsområden inom produktion av antibiotika, vitaminer, som en monomer för skräddarsydda polyestrar och som en näringskälla för eukaryota celler. Målet för denna avhandling har varit att, genom att integrera stam- och bioprocessdesign, öka den mikrobiella produktionen av 3HB från den välstuderade plattformsorganismen Escherichia coli (stam AF1000) som uttrycker ett thiolas och ett reduktas från Halomonas boliviensis. Att koppla isär tillväxt och produktion via NH4+- eller PO43-- begränsade fed-batch odlingar gav 3HB halter på 4.1 och 6.8 g L-1 (Artikel I). Att, i NH4+ hämmade förhållanden, öka tillförseln av NADPH via överuttryck av glukos-6-fosfatdehydrogenas (zwf) resulterade i ett 1.7 gånger högre 3HB-utbyte jämfört med att inte överuttrycka zwf (Artikel II). För att öka 3HB produktion vid höga cellhalter valdes stammen BL21 med anledning av den låga ättiksyraproduktionen. BL21 odlad i NH4+ begränsade fed-batch processer gav 2.3 gånger högre 3HB halt (16.3 g L-1) jämfört med stammen AF1000 (Artikel III). Överutryck av det nativa thioesteraset ”yciA” från E. coli, vilket identifierats som den största bidragande faktorn till 3HB-CoA hydrolys, resulterade i ett 2.6 gånger högre 3HB-utbyte jämfört med AF1000 utan överuttryck av yciA. Kombinerat överuttryck av zwf och yciA i NH4+ begränsad fed-batch gav 2 gånger högre 3HB-utbyte (0.210 g g-1) jämfört med AF1000 som endast överuttryckte zwf (Artikel IV). Dessutom, med 3HB som modellprodukt, presenterades den bakteriella artificiella kromosomen som en enkel plattform för design och optimering av metaboliska vägar i E. coli (Artikel V). Även om de är direkt relevanta för 3HB-produktion, bidrar dessa resultat också till kunskapen om hur man förbättrar kemikalieproduktion för utveckling av robusta och skalbara processer.

List of publications and manuscripts This thesis is based on the following publications or manuscripts, which are referred to in the text by their roman numerals:

I. Guevara-Martínez M, Sjöberg Gällnö K, Sjöberg G, Jarmander J, Perez-Zabaleta M, Quillaguamán J, Larsson G (2015): Regulating the production of (R)-3- hydroxybutyrate in Escherichia coli by N or P limitation. Frontiers in Microbiology. 6:844

II. Perez-Zabaleta M, Sjöberg G., Guevara-Martínez M, Jarmander

J, Gustavsson M., Quillaguamán J, Larsson G (2016): Increasing the production of (R)-3-hydroxybutyrate in recombinant Escherichia coli by improved cofactor supply. Microbial cell Factories 15:91.

III. Perez-Zabaleta M, Guevara-Martínez M, Gustavsson M.,

Quillaguamán J, Larsson G, van Maris A. (2019): Comparison of engineered Escherichia coli AF1000 and BL21 strains for (R)-3-hydroxybutyrate production in fed-batch cultivation. Accepted for publication in Applied Microbiology and Biotechnology

IV. Guevara-Martínez M, Pérez-Zabaleta M, Gustavsson M., Quillaguamán J, Larsson G, van Maris A. (2019): The role of the acyl-CoA thioesterase “YciA” in the production of (R)-3-hydroxybutyrate by recombinant Escherichia coli. Applied Microbiology and Biotechnology 103:3693-3704

V. Sjöberg G, [1] Guevara-Martínez M[1], Gustavsson M., Quillaguamán J, van Maris A. (2019): Metabolic engineering applications of the Escherichia coli bacterial artificial chromosome. Submitted

[1]Shared first authorship.

Contributions to papers

I. Contributed to the planning and writing of the manuscript and performed the majority of the experiments.

II. Contributed to the manuscript and performed batch experiments related to the use of glutamate.

III. Contributed to the manuscript and the experiments

IV. Wrote the manuscript and performed the majority of the

experiments

V. Wrote the manuscript and planned and performed the majority of the experiments together with Sjöberg G.

Related publications not included in this thesis

I. Jarmander J, Belotserkovsky J, Sjöberg G., Guevara-Martínez M, Pérez-Zabaleta M, Jorge Quillaguamán J, Larsson G (2015): Cultivation strategies for production of (R)-3-hydroxybutyric acid from simultaneous consumption of glucose, xylose and arabinose by Escherichia coli. Microbial Cell Factories 14:51

Table of contents 1 Introduction.................................................................................. 1

1.1 Moving forward to a sustainable

bio-based economy....................................................................... 1 1.2 Biorefineries................................................................................... 3

1.2.1 Biorefinery classification....................................................... 4 1.3 Microbial production................................................................... 11

1.3.1 Microbial native producers................................................. 13 1.3.2 Microbial recombinant producers....................................... 18

1.4 Polyhydroxyalkanoates and (R)-3-hydroxybutyrate................ 22 1.4.1 Polyhydroxyalkanoates...................................................... 22 1.4.2 (R)-3-hydroxybutyrate........................................................ 25

1.5 Steering production..................................................................... 28 1.5.1 Steering production by strain choice and -design.............. 33

1.5.1.1 Genomic tools............................................................. 37 1.5.2 Steering production by bioprocess design......................... 39

2 Present Investigation.............................................................. 43

2.1 Aim and strategy......................................................................... 43 2.2 Escherichia coli as a model production organism.................. 44 2.3 Production of 3HB by E. coli expressing recombinant

genes from Halomonas boliviensis........................................... 45 2.4 Overview of methods used in this thesis................................... 46 2.5 Steering 3HB production by bioprocess design

(Paper I)........................................................................................ 47 2.5.1 Production of 3HB under ammonium or

phosphate depletion........................................................... 48 2.5.2 Ammonium- and phosphate-limited fed-batch

experiments for increased 3HB accumulation.................... 50 2.6 Steering 3HB production by engineering of the

redox-cofactor supply (Paper II) ................................................ 53 2.6.1 Determining cofactor specificity of..................................... 53

acetoacetyl-CoA reductase (rx).......................................... 53 2.6.2 Strategies to increase NADPH supply............................... 54 2.6.3 3HB production in fed-batch cultivations............................ 57

2.7 Steering 3HB production by decreasing production of the inhibitory byproduct acetic acid (Paper III) ................... 59

2.7.1 Eliminating genes responsible for acetic acid formation.......................................................... 59

2.7.2 Investigation of 3HB production in different E. coli strains...................................................................... 63

2.8 Steering 3HB production by interventions in the pathway to reduce competing pathways (Paper IV)................. 67

2.8.1 Specific contribution of native thioesterases to 3HB production........................................ 68

2.8.2 Overexpression of yciA for improved 3HB-CoA hydrolysis........................................................... 70

2.8.3 Impact of yciA overexpression on 3HB production in fed-batch cultivations.................................... 72

2.9 Implementation of a novel recombinant expression system for metabolite production (Paper V)............................. 76

2.9.1 Re-discovering the Bacterial Artificial chromosome......................................................... 76

2.9.2 Comparing the performance of the BAC with that of a multi-copy plasmid........................................ 77

2.9.3 Using the BAC for pathway evaluation............................... 78 2.9.4 The BAC mimics chromosomal expression….................... 81

3 Concluding remarks and Outlook..................................... 83

4 Acknowledgements................................................................. 86

5 Abbreviations............................................................................. 88 6 References.................................................................................. 90

Mónica Guevara Martínez |

1

1. Introduction

1.1 Moving forward to a sustainable bio-based economy

Society is currently highly dependent on many different products derived from petroleum sources. Not only is petroleum related to energy, but also many every-day-use products are derived from it. Plastics, ink and paint are just some examples of the many products derived from petroleum sources. The use of fossil-based resources for non-energy related purposes has become increasingly important, especially for production of value-added chemicals. In terms of production volumes about 10% of fossil resources are consumed for non-energy related purposes such as production of chemicals and 90% is used for energy related purposes such as fuels [1-4]. However, because of the much higher value of the non-energy related products of petroleum, they represent a larger share of the total value of the petroleum industry [3]. The globally-increased requirement of petroleum raises concerns that this is a finite source. Furthermore, the scale of exploitation of fossil fuels has highly increased the emission of carbon dioxide, the biggest contributor of greenhouse gas emissions [5]. A direct coupling has been observed between the increase in concentration of greenhouse gases over the years and the increase in the Earth’s “Radiative forcing”, a parameter that measures the difference between the energy absorbed by the earth from sunlight minus the energy radiated back to space [6]. The effect of greenhouse gas emissions is decreased energy radiation towards space and consequently an increased average temperature of the Earth’s atmosphere.

| Strain- and bioprocess-design strategies to increase production of (R)-3-hydroxybutyrate by Escherichia coli

2

Figure 1. Radiative forcing due to the greenhouse gases, relative to 1750 (zero radiative forcing). Data extracted from: The NOAA Annual Greenhouse Gas Index between 1979-2017 [6].

Thus, over the past decades the finite-supply of petroleum and growing public concern for global warming, have been increasing driving forces towards the development of more sustainable production processes for energy, materials and chemicals. It is therefore of much importance to find alternative sources not only for energy but also for chemical production. Because of both climate change and energy insecurity the EU and the US have implemented policies to lower their dependence on fossil fuels and make energy production and consumption more sustainable. In 2009 the EU’s Renewable energy directive has set a target that by 2020, 20% of their energy consumption should come from renewable sources. The directive specified individual renewable energy per country setting targets from 10% in Malta and 49% in Sweden. In 2014, the EU updated the renewable energy target to at least 27% by 2030 [7]. Although substitution of fossil fuels can be done by several ways, for example generation of electricity by wind and sun, these technologies do not directly provide the carbon mass for production of materials and chemicals compounds [8]. Carbon sources are essential for making products such as plastics, pharmaceuticals and other chemicals. This is where biomass, which refers to the mass derived from organic material such as trees, plants and agricultural and urban waste, can be a valuable alternative. The use of biomass has the potential to move us from a fossil-based economy (where products and fuels are based on fossil resources) towards a bio-based economy where we instead use renewable feedstocks. A bio-based economy is defined by the European Commission as: “The sustainable production of renewable biological resources


3

from land and sea – such as crops, forests, fish, animals and micro-organisms – to produce food, materials and energy” [9]. Sustainability refers to being able to meet today’s needs without compromising the needs of future generations [10]. Environment, economics and social aspects are three important components of sustainability and therefore for a bio-economy. In this line of thought, a transitioning towards a bio-based economy should include an assessment on sustainability of these three components. One benefit of the use of plant biomass is the reliance on crop cycles of at most a few years versus millions of years for geological formation of fossil resources. Additionally, compared to exploitation of fossil resources, which releases CO2 that has been sealed underground for tens of thousands of years into the atmosphere, the use of biomass has the potential to not increase the amount of CO2 released in the atmosphere, because during growth, plants take sunlight and CO2 from the atmosphere and convert it into oxygen, organic material and energy. Furthermore, the extensive global distribution of biomass, gives a bio-based economy a great potential for rural development, economic growth and generation of new markets and employment opportunities. 1.2 Biorefineries With the objective of moving from oil dependence towards a bio-based economy, the concept of biorefineries has emerged as an alternative to replace petroleum refineries. Biorefining is a concept that involves the process of conversion of biomass to fuels, energy, commodity chemicals and value-added chemicals. Many definitions exist for biorefinery. Two widely used definitions formulated by the intergovernmental organization “IEA” (International Energy Agency) in the Bioenergy Task 42 [11] and the United States’ National Renewable Energy Laboratory (NREL) [12] respectively are: - “Biorefining is the sustainable processing of biomass into a spectrum of

marketable products and energy.” - “A biorefinery is a facility that integrates biomass conversion processes and

equipment to produce fuels, power, and chemicals from biomass.”


4

Figure 2. The biorefinery concept.

Biorefineries involve a wide variety of processes (upstream processing, transformation, fractionation, downstream processing) that convert different forms of biomass (crops, organic residues, agro-residues, forest residues, wood, aquatic biomass) into one or many bio-based products or into their building blocks [13]. Similar to petroleum refineries, biorefineries should be able to convert raw material into a number of value-added products necessary for society with very little waste. This would maximize economic potential and minimize negative environmental aspects. Fossil-based resources and biomass differ in various aspects including their chemical composition, necessitating new approaches in research & development, conversion technology of biomass and production technology to facilitate a switch towards the use of biomass [14]. Advances and incorporation of biology, chemistry, agriculture, genetics, biotechnology, process chemistry and bio-process engineering have the potential of developing biorefineries able to make this switch to a sustainable bio-based economy [15]. 1.2.1 Biorefinery classification The range of biomass-derived raw materials for biorefineries includes cellulose, hemicellulose, lignin, starch, sucrose, vegetable oils, and fats [4]. The sugar containing polysaccharides are converted to smaller monosaccharides or disaccharides such as: glucose, fructose, cellobiose, galactose, xylose and/or arabinose [4]. These serve as intermediates for further conversion to bioproducts.


5

Hydrolysis of triglycerides from oils and fats give glycerol and fatty acids which can be used as platform chemicals [4]. Classification of biorefineries is done depending on the plant biomass and the fraction used of this biomass. There are already existing biorefineries that use conversion and separation technologies to separate a main product such as sugar and oil from biomass [8, 13]. Their main emphasis is to produce their main products and although side streams are valorized there are no large efforts invested in producing a broader spectrum of products. These are:

(a) sugar extraction from sugar-beet or -cane (b) pulp and paper industry using forest-based biomass (c) extraction of vegetable oil from soy or rape seed (d) processing of cereal grain- based biomass to starch, sugar syrups, ethanol,

cereal oil and/or fibers These biorefineries commonly do not add value to every part of the plant. As an alternative to try and use the whole plant, more advanced biorefineries are in the present being developed [15]. These are: (a) Whole crop biorefineries based on dry or wet milling of cereals as the raw

material. Feedstocks are for example: rye, wheat and maize. The first step in these biorefineries involves a mechanical separation of the grain fraction and the straw fraction from each other which respectively constitute around 20% and 80% of the whole crop [13, 15]. The straw can be further treated in a lignocellulose biorefinery or used as a starting material for production of syngas [15]. The grain fraction can be further used after grinding to meal or converted to starch [16]. Starch can be transformed by chemical modification, biotechnological conversion or plasticization. Meal can be treated further and extruded into binder, adhesives and filler [17]. Examples of products derived from whole crop biorefineries are: syngas, ethanol, methanol, sorbitol, bioplastics, etc. (Fig. 3).


6

Figure 3. Products derived from a whole crop biorefinery. Figure reprinted from Kamm B. et

al. (2004) [16] with permission from Springer Nature.

Figure 4. Products derived from a whole crop wet mill-based biorefinery. Figure reprinted from Kamm B. et al. (2016) [15] with permission from John Wiley and Sons


7

Wet milled based technology provides an alternative to utilize more of the whole crop. Here the whole grain is water soaked (usually also adding SO2) and the grain germ is pressed to release high-value oils [15]. The advantage is the good separation of natural structures such as starch, oils and proteins. Wet milling of corn results in oil, fiber and starch [15] (Fig. 4).

(b) Green Biorefineries. These biorefineries employ wet biomass, such as green grass, alfalfa, clover or immature cereal [16]. By wet fractionation the contents of a green crop biomass are isolated into a fiber-rich press cake and a nutrient-rich green juice [15]. The press cake contains valuable pigments, crude drugs, some cellulose and starch [15]. While the green juice contains proteins, organic acids, hormones and others [15]. Products such as organic acids, amino acids and proteins can be obtained. The residues of conversions can be used for production of biogas.

(c) Lignocellulose Feedstock Biorefinery. These biorefineries use renewable dry raw material called lignocellulose. Some examples are straw, wood, paper waste or reed which are significantly less expensive than cereals [18]. The composition of lignocellulose consists of three primary chemical fractions: two carbohydrates, hemicellulose and cellulose and lignin which is an aromatic polymer [15]. After fractionation, the C6 sugars (glucose) and C5 (xylose and arabinose) are used as feedstocks for biochemical conversion processes to produce biofuels or value-added chemicals. Lignin on the other hand can be applied for combined heat and power, or for production of value-added chemicals as phenolic components [15]. Lignocellulose feedstock biorefineries are expected to be very important in the future because of their low costs, large availability and broad spectrum of different products. There are however some inconvenient aspects of lignocellulose feedstock such as the biomass recalcitrance [15] and the lack of valorization options for lignin (since there exists no natural enzymes capable of converting lignin into its monomers). Examples of products derived from lignocellulose based biorefineries are fuels, organic acids, solvents, lubricants, etc. (Fig. 5). These biorefineries can also include aquatic biomass such as microalgae and macroalgae as feedstock. Using aquatic biomass has the potential of highly increasing the total biomass availability. Depending on the algae used, significant amounts of different bio-products can be obtained such as oils, starch and carbohydrates [13].


8

Figure 5. Products of a lignocellulosic feedstock biorefinery. Figure reprinted from Kamm B. et al. (2004) [16] with permission from Springer Nature.

To try and get a more homogeneous classification the IEA Bioenergy Task 42 [19] has developed an approach to distinguish four main features of biorefineries: the feedstock, the intermediate platform, type of product (energetic or non-energetic related) and the main process involved in the conversion. An example using this classification is: hydrolysis of corn starch crops to a one platform C6 sugar biorefinery for the production of the energetic product bio-ethanol by fermentation and with animal feed as a coproduct [19]. Although often used in combination, the main conversion processes can be separated in mechanical, thermochemical, chemical and biochemical methods [20]: a. Mechanical/physical biorefining. These include conversion processes such as

pressing, fiber separation, laminating and milling, which do not change the chemical structure of the components of the biomass [21]. The physical conversion of plant biomass mainly uses their tight physical structure to convert them into useful materials. They can benefit of using both timber (wood) and non-timber (bagasse, wheat, straw etc.) feedstocks [21]. This type of biorefining converts plant biomass into value-added materials such as lignocellulosic composites, sheets, construction materials, pulp and paper [21].


9

b. Thermochemical biorefining. These transform the biomass through the use of extreme conditions such as high temperatures or high pressures [19]. These processes offer an efficient and economical way of providing energy and preparing chemicals at high conversion rates. Examples of these technologies include combustion, gasification, steam gasification, pyrolysis and liquefaction. Gasification converts biomass at high temperatures, >700°C and low O2 levels and produces syngas (CO and H2), which can be used directly as a biofuel or as a intermediate platform for production of chemicals and fuels [22]. Steam gasification uses steam as the gasification agent resulting in a higher amount of H2 [23]. Pyrolysis uses more intermediate temperatures (between 400 and 800°C) with no oxygen and decomposes biomass into a carbon rich solid residue (charcoal) and a hydrocarbon rich gas or liquid mixture which is usually cooled down and known as pyrolysis liquid [24]. Liquefaction is a process in which in the presence of a suitable catalyst and high-pressure biomass is decomposed into small molecules, which are unstable and repolymerize into oily liquids. Hydro-thermolysis is a process in which degradation and hydrolysis of biomass takes place via acidic hydronium ions and basic hydroxide ions from water at elevated temperatures [25].

c. Chemical biorefining. These change the chemical structure by reacting with

other substances. Some common processes are hydrolysis, the Fischer-Tropsch process and transesterification. The Fischer-Tropsch process uses a metal catalyst for the formation of synthetic fuels such as paraffins and olefins from syngas (H2 and CO) [26]. Transesterification is done for example for the production of biodiesel and glycerin where vegetable oils react with an alkali catalyst in methanol to be converted into the esters of the fatty acids. Other chemical processes involve the formation of higher alcohols from syngas using chemical catalysts from syngas, or formation of H2 from water hydrolysis or water gas shift from syngas.

d. Biochemical biorefining. This uses enzymes and/or whole microorganisms to

convert biomass into bio-products. Examples of biochemical transformations are: aerobic respiration cultivations, anaerobic respiration cultivations, anaerobic cultivations (fermentation), anaerobic digestion (in which by absence of oxygen, biogas is produced by a collection of different natural occurring microorganisms in subsequent four key stages: hydrolysis, acidogenesis, acetogenesis and methanogenesis) and enzymatic bioprocesses. These conversions occur at lower temperatures and at lower reaction rates


10

than thermochemical processes, but don’t require much external energy [27]. Using microorganisms for the conversion process does not require any other solvents than water. Examples of biochemical conversion processes are the breaking down of cellulose and hemi-cellulose components to simple sugars and the conversion of sugars into final products by the use of whole cells or enzymes as catalysts. For the latter, a key challenge is to directly convert the sugars into biofuels or bioproducts in an efficient, economic and sustainable manner. Examples of products derived from biochemical biorefining are: penicillin, growth hormones, insulin, hydrogen, organic acids (pyruvate, lactate, oxalic acid, levulinic acid, citric acid), biogas, ethanol, acetone, butanol, 2,3-butanediol, 1,4-butanediol, isobutanol, xylitol and mannitol [21]. In 2004 a list of the 12 most promising bio-based platform molecules was published by the US Department of Energy [28] (Fig. 6) with the intention to identify potential building-block molecules for the production of value-added products and support the production of energy from biomass. Biochemical conversions account for the majority of routes from plant feedstocks to glucose to building blocks. Additional chemical transformations can be used for subsequent conversion of these building blocks to other value-added chemicals.

Figure 6. Conversion of glucose to 11 out of the 12 most promising bio-based platform molecules published by the US department of energy [28]. Xylitol/Arabinitol can be obtained from the hydrogenation of the sugars xylose and arabinose. Figure adapted and redrawn from Murzin et al. (2008) [29]

O OH

OHHO

OH

OH

Glucose

HO

OH

OH

OH

OH

OH

Sorbitol

Chemical transf.O

O

HO3-Hydroxybutyrolactone

Biochemicaltransf. & oxidation

OHHO

O O

NH2

Biochemical transf.

Biochemical transf.

OHHO

OO

O

HO OH

O

Biochemical transf.

Glutamic acid

Itaconic acid

HOOH

O

O NH2

Aspartic acid

3-Hydroxypropionic acid

Biochemical transf.

Biochemical transf.

HOOH

O

O

OH

O

O

HO

OH

Malic acid

Succinic acid

HOOH

O

OFumaric acid

OO HO

Chemical transf.

5-hydroxymethylfurfural

OO O

HO OH

2,5-Furandicarboxylic acid

HOOH

OH

OH

OH

OH O

Chemical transf.

Chemical transf.

O

Glucaric acid

HO

O

OLevulinic acid

Chemical transf.

HO

OH

OH

Biochemical transf.


11

1.3 Microbial production Biochemical biorefining that uses whole cells as their main process to convert biomass or its derivatives into final products, can be distinguished as microbial biorefining. Microbial biorefining has been further distinguished depending on the raw material and production technologies used [20, 30]. First generation biorefineries refer to those that use easily degradable and often edible raw material such as sugar, starch and vegetable oils. Their conversion processes are fairly simple and their technologies are well stablished. Second generation biorefineries on the other hand use non-edible raw materials, such as residues of food crops, straw, wood, dedicated energy crops or other waste materials. These raw materials contain a high amount of carbon though usually in the form of lignocellulose, which is made-up of, hemicellulose and lignin. The benefit of these biorefineries is that lignocellulosic material or waste are highly abundant. Third generation biorefineries use the idea of capturing carbon dioxide directly from the environment by using organisms capable of using CO2 as a carbon source and sunlight for energy, for example photosynthetic algae. In order to achieve sustainability, a biorefinery should attain to utilize biomass with low (or no) value in which the components are utilized optimally to produce both high and low value products. A key technology needed for the development of these kind of biorefining is industrial biotechnology. Biotechnology is defined as: “The broad area of biology involving living systems and organisms to develop or make products, or, any technological application that uses biological systems, living organisms, or derivatives thereof to make or modify products or processes for specific use” [31]. Before the 1950s, industrial biotechnology was used as a tool for production of many bulk products including fuels (ethanol and butanol) and organic acids (acetic acid, citric acid, lactic acid). Processes back then were mainly done by the cultivation of fungi or bacteria [32]. One important example was the production of acetone (used for production of cordite for use in gun powder) and butanol (used for automobile paints and as building block for esters) during World War I and II. The acetone-butanol-ethanol (ABE) fermentation process was developed from the isolated bacteria Clostridium acetobutylicum when petrochemical raw materials were hard to obtain [33]. With the development of the petroleum industry oil became very cheap and biotechnological processes were often no longer economical and lost much attention. In the 1970s, the first oil crisis led to much concern with regards of oil dependence and renewable substrates were considered again [32].


12

Although deprioritized when oil prices decreased again, this shows that the shortage of raw material and economic factors are powerful driving forces for the development of alternative sustainable processes. Despite the low oil prices, biotechnological processes have in recent years increasingly received attention because of the driving forces towards a sustainable bio-based economy [34]. Microbial biotechnology exploits the use of whole cells for synthesis of a variety of products. Microbial cells can be used as cell factories as they can use simple cheap sources of carbon and nutrients for both growth and synthesis of the product [35]. These cell factories allow the production of fuels and chemicals through multi-step pathways which further benefit of cofactor regeneration and high stereo- and chemo-selectivity at mild temperature and pressure [36]. Through their metabolism, they provide a vast network of pathways which are used to produce a wide variety of natural and synthetic products including antibodies, antibiotics, other pharmaceuticals, enzymes, proteins, biofuels, fibers, fabrics, organic acids, solvents, beverages, plastics fine chemicals and bioplastics [37]. Some examples of industrial production of bioproducts by the use of biorefineries which use microorganism for their main conversion step are shown in Figure 7.

Figure 7. Overview of companies using engineered microorganisms for the synthesis of biofuels or chemical compounds. Figure reprinted from Jullesson et al. (2000) [38] with permission from Elsevier.


13

Cell factories are currently used in many different industrial sectors, though production of chemical and fuels is still governed by chemical processes from fossil feedstocks. Despite microorganisms being capable of naturally producing (native microorganisms) many interesting compounds, they are not necessarily able to do it in a sufficiently efficient manner to reach commercial industrial production. Production through naturally present metabolic pathways often results in low product concentrations and formation of by products, which in turn affect yield, productivity and the complexity of downstream processing. Additionally, microorganisms that naturally produce a product often have limitations concerning substrate utilization and stoichiometrically or kinetically suboptimal pathways. This gives researchers a challenge for developing efficient microbial cell factories and bio-processes that can overcome the mentioned limitations. A key step in the development of a bioprocess is to choose an appropriate producing strain [4]. In the past, a microorganism was chosen after it was identified to naturally produce an interesting product, but in the present, recombinant DNA technology allows the use of a host organism that does not necessarily naturally produce the product [4]. A key issue on either choice is to improve the properties of the selected strain. Before describing examples of engineered cell factories (1.3.2), below first a description of microbial native producers and some of their products is given. 1.3.1 Microbial native producers A microbial native producer is a microorganism capable of naturally producing specific compounds. These compounds can be found in nature and can be of high value. These can be both, primary or secondary metabolites. Native microorganisms can have enormous potential because of their unique biochemical pathways, such as their ability to: take up many different substrates, produce a variety of interesting valuable compounds and tolerate toxic products and byproducts. These characteristics can be unique for research and industry. Although generally improvements may be necessary to attain efficient production from these organisms. Classical strain improvements (CSI) methods such as random mutagenesis, directed evolution and dominant selection have been used for these organisms as a way of enhancing their production capacities. Below three examples of industrial metabolite production by native producers relevant for this thesis are shown. Two examples correspond to organic acids produced as primary metabolites and one of them can be further used as the monomer of a polymer. The


14

last example is a secondary metabolite that is often used to illustrate the improvement of production by both strain and a bio-process optimization. - Lactic acid (LA). This organic acid has industrial applications in chemical,

pharmaceutical, food, and textile industries [39]. It further gained attention as the monomer for polylactic acid (PLA) which has applications that range from packaging and fibers to foams and biomedical devices [40]. LA exists in both L and D enantiomeric configurations. In the food and pharmaceutical industry, the L configuration is preferred since it can be metabolized by humans [41] while the D configuration can be toxic to humans [42, 43]. The first LA commercial plant was built in 1881 and was based on microbial production [44]. In 1963, production via chemical synthesis was also done commercially. Microbial production of LA has advantages over chemical synthesis, such as: optically pure form instead of racemic mixtures, renewable sources as substrates and lower production temperatures. By 2016, 90 % of the annual 1220 kton global [45] LA production came from microbial cultivations [46]. Microbial production relies on bacteria called Lactic Acid producing Bacteria (LAB) [47]. Depending on the metabolic pathway LAB are classified in: homo- or hetero-fermentative (Fig. 8A-B). Homo-fermentative LAB are not able to use pentoses and convert 1 mol of glucose to 2 mols of LA using glycolysis. Hetero-fermentative LAB can be strictly heterofermentative and only use the 6-phosphogluconate/phosphoketolase (6-PG/PK) pathway to convert glucose to equimolar amounts of lactic acid, CO2 and ethanol (or acetate) [45] or they can be facultative and metabolize glucose by glycolysis and pentoses by the 6-PG/PK pathway. Submerged fermentations, cell recycling, immobilization and repeated batches are used to enhance productivities in industrial production [48, 49]. The cultivation involves the use of nutritional rich medium (because of LAB’s limited ability to synthesize B vitamins) [48], pH around 5-6 (maintained by the addition of CaCO3 or Ca(OH)2) and temperatures around 45°C. LAB have undergone classical strain improvement techniques to improve production. For example, by addition of the mutagen ethyl methane sulfonate in Lactobacillus delbrueckii, higher productivities and improved tolerance to higher acid concentrations were obtained [50]. By application of heat shock and UV mutagenesis in Lactobacillus isolates, undesirable traits such as antibiotic resistance and citrate metabolism were eliminated [51]. Finally, Lactobacillus strains with improvements in pH tolerance by either directed evolution in chemostats or


15

nitrose guanidine mutagenesis, were obtained by genome shuffling and produced three time more lactic acid than wild-type [52].

Figure 8. Schematic overview of the metabolic pathway for production of lactate by LAB. a) Homo-fermentative b) Hetero-fermentative. Figure redrawn and adapted from Okano et al. (2014) [53]

- Citric acid. This is an organic acid which naturally exists in fruits such as lemons. It´s worldwide production is over 1.6 million tons per year [54] and it has applications in the food, beverage, pharmaceutical and chemical industry [55]. Citric acid was first commercialized in 1826 in England by extracting it from lemons imported from Italy [55]. By 1880, Pfizer produced it in the same way until World War I interrupted the supply of lemons from Italy and new methods for producing the acid were needed. In 1917 James Currie discovered that Aspergillus niger was able to convert sugar to high amounts of citric acid. By 1919 Pfizer opened a pilot plant using his fermentation process [56]. To increase production, submerged cultivations were suggested and by mid 1920s the price of microbial citric acid was much lower than the one from lemon extraction [56]. Aspergillus niger is superior to others in the production of the acid since it is easy to handle, can take up different substrates and high yields are obtained. Furthermore, citrate accumulation is supported by a mitochondrial-membrane citrate-malate transporter capable of exchanging malate for citrate, an active cytosolic pyruvate carboxylase and a low activity α-ketoglutarate dehydrogenase (Fig. 9).

Glucose(A)

Fructose 6-P

Fructose 1,6-BP

Glyceraldehyde 3-P Dihydroxyacetone-P

2 Pyruvate

2 Lactate

Glucose

6-P-gluconate

Xylulose-5-P

Glyceraldehyde 3-P Acetyl-P

Pyruvate

Lactate

(B)

CO2

ldh ldh

Ethanol Acetate


16

Figure 9. Schematic overview of the metabolic pathway for production of citric acid by Aspergillus niger. Figure redrawn and adapted from Kubicek et al. (2011) [57]

The production is done under aerobic conditions and very high sugar concentration (>100 g L-1) with limitation of a metal such as Mn2+, Zn2+ or Fe3+, low pH (around 1-2) and 32°C. Processes can be run for more than 160 hours with a titer of 200 g L-1 and yields of citrate to sugar above 0.8 g g-1. Classical strain improvement has also been employed in this organism [58, 59]. For example, by UV and chemical mutagenesis hyperproducer strains were obtained [58, 60]. Furthermore, a strain with higher affinity towards sucrose assimilation and higher activities on both hexokinase and phospho-fructokinase was obtained by UV mutagenesis [61]. Finally, a more rational approach based on a model that suggested that sugar transport and phosphorylation constituted the most important step for controlling carbon flux towards citric acid [57, 62, 63], led to a study in which inhibition of hexokinase (by trehalose 6-phosphate synthase tpsA) was disrupted, resulting in a higher rate of citric acid production [64].

- Penicillin. This is a group of antibiotics which were among the first efficient medications used against many bacterial infections. The first penicillin was discovered by Alexander Fleming in 1928 when he found that the Staphylococcus plated petri-dish that he had mistakenly left open was contaminated with a mold (Penicillium notatum) that somehow killed

Glucose

Pyruvate

OAA

Malate

Acetate

Oxalate

Pyruvate

Mitochondrion

Cytosol

AcCoA

CitrateCitrate

OAA

Malate

CO2

Oxalate

Citrate

GlucoseFructose

Sucrose

Fructose

pyruvate carboxylase

malate dehydrogenase

malate dehydrogenase

citrate synthase


17

bacteria. In 1939, Howard Florey and a team of researchers from Oxford University purified very small quantities of penicillin and proved its potential [65]. One of the first design strategies to increase the titer was done in 1941 by changing the production media by adding corn-steep liquor resulting in a 10-fold increase. Aware that the fungus Penicillium notatum was not able to produce sufficient amounts of penicillin, Florey and his team, together with American Scientists in Peoria, searched for a more efficient organism. Mary Hunt, a laboratory assistant, found Penicillium chrysogenum. This mold was able to produce 200 times more penicillin then Penicillium notatum [66]. Furthermore, an X-ray mutated strain was capable of producing 1000 times more. In the 1940s an Anglo-American cooperative effort was made to find a way to produce high enough amounts of penicillin. Until then penicillin was produced in bed pans through surface fermentation with very low yields [67]. In 1943, Jasper Kane, a Pfizer representative, suggested to switch to deep-tank fermentation. Development of an efficient process, able to provide sterile air to the process and proper agitation was necessary. Within 6 weeks, a process was developed in which several 7500-gallon tanks were set up to produce industrial amounts of penicillin [67]. Over the years, Penicillium chrysogenum has undergone a comprehensive amount of mutagenesis including UV irradiation, X-ray and methyl-bis(β-chloroethyl) amine) which resulted in [68] enhanced expression of the genes that express enzymes of penicillin synthesis (ACV synthase, IPN synthase and acyltransferase) (Fig. 10) [69]. Further process development led to a fed-batch process with [70] : (i) a continuous addition of ammonia (pH control and avoid lysis of the mycelium) (ii) Intermittent/slow continuous addition of precursors iii) slow feed of glucose and corn steel liquor to assure low concentrations of nitrogen, carbon and phosphorus, since higher concentrations inhibit the synthesis of enzymes involved in penicillin pathway [70]. In 1939 the achievable final titer was of 0.001 g L-1, while today with all the improvements it is possible to produce > 50 g L-1 of penicillin.


18

Figure 10. Schematic overview of the metabolic pathway for production of penicillin-G by Penicillium chrysogenum. Figure redrawn and adapted from Kubicek et al. (2011) [57]

Although the case of penicillin illustrated the potential of classical strain improvement and process development, there are many natural products where intrinsic properties of the microorganism prevent economic viability. Native producers can for instance be difficult to culture, require complex media, or contain complex metabolic control. To convert them into efficient cell factories enhancement by genetic engineering tools may be needed. Though, these organisms often also suffer from lack of genetic accessibility and molecular biology tools, required for engineering of the production strain [71]. 1.3.2 Microbial recombinant producers An alternative approach to overcome the limitations of native producers is the use of recombinant DNA technology. This technology refers to the methodology of transferring genetic material from one organism to another [72].These genetic combinations open up a range of new opportunities for protein and metabolite production that are of high-value to the industry, science and medicine. By inserting foreign genes into the genome of another microorganism, the later can produce a protein or a metabolite it does not naturally produce, overproduce a

HO OHO

ONH2

+NH2

OHO + HS OH

NH2

O

α-amino adipic acid valine cysteine

ACV synthase pcbAB

NH2

NSH

NHO

H

O

IPN synthase pcbC

ACV

NH2

N

NO

H

O

HO

O

HO

O

OHO

S

OHOIsopenicillin-N

acyltransferasepenDE N

NO

H

O

S

OHOPenicillin-G


19

desired compound or obtain enhanced desired characteristics for a microorganism (e.g. improve their cellular physiology or to extend their substrate uptake range). Organisms used for this technology are often standard laboratory and/or industrial organisms, called “platform cell factories”. Two of the most well-studied platform organisms that have gained special attention for the production of chemicals and fuels are the bacteria Escherichia coli and the yeast Saccharomyces cerevisiae [73]. The advantages of using these organisms are [73] a) being well adapted to grow on a simple medium generally consisting only of a carbon source and simple salts for nutrients b) they are very well characterized with regards to genetics and physiology c) there are well established molecular genetic tools available for the modification. d) many gene expression tools are available (plasmids, promoters, terminators). Below are given two relevant examples of products developed through recombinant DNA technology. One of the scopes of this thesis involves the improvement of production of a low molecular weight compound by a recombinant cell using strain design. The first example (1,3 propanediol) is a low molecular weight compound derived from central metabolism. The second example (artemisinin) is derived from a longer and more complex pathway. Both examples are a paradigm for the impact of genetic engineering on pathway development. - 1,3 propanediol (PDO). This is a colorless and odorless biodegradable

chemical with low toxicity, which can be used directly as a solvent or an antifreeze agent. Furthermore, PDO is an intermediate molecule applicable for many purposes such as production of pharmaceuticals, coatings, solvents and polymers [74]. PDO is employed as a monomer for the production of the polymer polytrimethylene terephthalate (PTT) [75], which can be used to make fabrics, carpets, thermoplastics, as simple fiber or as construction material in the automotive industry [75]. PDO’s global demand in 2012 was > 60 000 ton/year and was expected to reach 150 000 ton/year by 2019 [76]. PDO can be produced both chemically and biochemically. Chemically it has been done both by the company “Shell” (from ethylene oxide with organometallic catalysts) and by the company “Degussa” (from acrolein by a catalyst-mediated reaction). Even though the chemically produced PDO was well established, the processes still had disadvantages, such as high temperature and pressure, expensive catalytic step and/or toxic feedstocks. Microbially produced PDO emerged as an alternative that utilizes renewable resources and less extreme conditions. PDO is naturally produced by fermentation of glycerol by some


20

native organisms like Citrobacter, Clostridium, Enterobacter, Klebsiella [77] (Fig. 11).

Figure 11. Schematic overview of the metabolic pathway for production of PDO from glycerol by native producers. Figure redrawn and adapted from Nakamura et al. (2003) [78]

To make a viable process the companies DuPont and Tate & Lyle worked together to develop a process with an efficient E. coli strain developed by Dupont and Genentech which converts a cost-effective carbon source directly to PDO. At that time (1995-1996) glycerol was to expensive [79], an alternative was glucose derived from renewable corn feedstocks derived from Tate & Lyle’s corn wet mill. These resulted in the creation of a recombinant engineered E. coli strain capable of converting glucose directly to PDO in an aerobic process. Details of the engineering will be explained in section “1.5.1 Steering production by strain choice and design”. Currently, DuPont and Tate & Lyle, are capable of producing 45 000 tons per year and are planning to expand production by 35 % by mid 2019 [80, 81].

- Artemisinin. This compound is a terpenoid with antimalarial properties

recommended as part of combination therapies by the World Health Organization. It is extracted from the Wormwood tree, Artemisia annua [82]. The extraction is very expensive and there are sharp fluctuations in its prices, generated mainly by demand and supply imbalances [83]. Microbial production might provide a more stable source of artemisinin would be attained. In 2003, Martin et al. [84] used metabolic engineering of E. coli to produce terpenoids resulting in 112 mg L-1 of the precursor “amorphadiene”. This was done by the recombinant expression of genes coming from a

OH OH

OHglycerol

OH

H

O

3-hydroxypropionaldehyde

OH OH

1,3-Propanediol

H2O

NADH

NAD+

Glycerol dehydratase (dhaB1-3)

1,3-Propanediol oxidoreductase (dhaT)


21

mevalonic acid pathway of S. cerevisiae expressed in two operons one from AcCoA to mavelonate (pMevT) and the other from mevalonate to farnesyl pyrophosphate FPP (pMKPMK) together with a codon optimized amorphadiene synthase “ADS” from A. annua (Fig. 12). This led to the origin of the Semi-Synthetic Artemisinin project, in which the University of California, together with the companies Amyris, PATH and Sanofi developed a bioprocess capable of viably producing Artemisinin.

Figure 12. Schematic overview of the metabolic pathway for production of amorphadiene by recombinant E. coli. Figure reprinted and modified from Martin et al. (2003) [84] with permission of Springer Nature.

A recombinant E. coli strain capable of producing 0.5 g L-1 of amorphadiene [85] was further engineered by the change of 2 genes from the first operon (AcCoA to mavelonate) and resulted in 25 g L-1 of amorphadiene in a novel developed fed-batch [86, 87]. However, E. coli was not able to convert amorphadiene to artemisinic acid since this reaction is catalyzed by a eukaryotic cytochrome P450 enzyme and E. coli is generally not a suitable host for expressing these enzymes. For this reason, a switch to Saccharomyces cerevisiae was made (Fig. 13). The identified P450 enzyme (CYP71AV1) and its cognate reductase (CPR1) in an amorphadiene producing strain of S. cerevisiae (S288C) enabled the production of 100 mg L-1 of artemisinic acid [88]. Incorporation of ADS and using a galactose-based fermentation process led to the production of 2.5 g L-1 of artemisinic acid [89]. By expression of CYP71AV1 (cytrochromeP450 enzyme) optimized with a low expressed reductase enzyme CPR1 and cytochrome b5 “CYB5” together with co-expression of A. annua’s aldehyde dehydrogenase and artemisinic aldehyde dehydrogenase resulted in a high titer of 25 g L-1 of artemisinic acid [82]. The transformation of artemisinic acid to artemisinin was done by a novel chemical process which involves an alternative high-efficiency photochemical


22

conversion process [90]. In the year 2013 commercialization of this chemical started by microbial production by the company Sanofi, although currently production has halted.

Figure 13. Schematic overview of the metabolic pathways for production of artemisinin by recombinant S. cerevisiae and E. coli. Figure redrawn from Paddon et al. (2014) [90]. 1.4 Polyhydroxyalkanoates and (R)-3-hydroxybutyrate

Two specific (classes of) products that are relevant for the context of this thesis and will be discussed in more detail below are polyhydroxyalkanoates (PHA) and (R)-3-hydroxybutirate (3HB). 1.4.1 Polyhydroxyalkanoates Polyhydroxyalkanoates (PHAs) are linear polyesters which are produced naturally by many different native microorganisms. Nevertheless, recombinant DNA


23

technology has been used to increase production, change the composition of the PHA chain and/or increase the length of the resulting polymers PHAs. These polyesters are mostly composed of chiral hydroxy-alkanoates in the R stereospecific configuration. The ester bond is formed between the carboxyl group of one monomer and the hydroxyl group of another monomer.

PHAs have properties similar to those of polypropylene, for which they are considered bio-plastics, which in addition are biodegradable and biocompatible [91-94]. Polyhydroxybutyrate (PHB), a PHA purely composed of (R)-3-hydroxybutyrate monomers, is the most widely studied and best characterized PHA, it was the first PHA identified intracellularly in the bacterium Bacillus megaterium in 1926 by Maurice Lemoigne [92]. Today, more than 150 different hydroxy-alkanoates monomers have been detected as constituents in different kinds of PHAs with molecular weights ranging from 50 000 to 3 000 000 Daltons [92].

Figure 14. General structure of a) polyhydroxyalkanoates and b) polyhydroxybutyrate. * Chiral center

Depending on their monomer composition, PHAs vary greatly in their chemical and physical properties from crystalline to elastic [92, 95]. PHAs composed of short-chain-length monomers with less than 5 carbon atoms are the most common, even though they are stiff and brittle [96]. On the other hand, PHAs composed of longer-chain-length monomers are more elastic and flexible. PHB, is highly crystalline, it has good thermoplastic properties, though it has poor mechanical properties making it brittle and stiff. Incorporation of a secondary monomer (into the PHB structure) such as 3-hydroxyvalerate (3HV) resulting in Poly(3HB-co-3HV) gives different properties, such as lower crystallinity, decreased stiffness and lower melting temperature. Although PHA has applications from every-day-use plastics, to photographic materials [97, 98], biocompatibility and other properties, have made it especially attractive for medical applications, such as: tissue engineering, bio-implant patches, drug delivery and surgical applications [99, 100].

O

O

OH

R

mH

n

O

O

OH

CH3H

n

a) b)

*


24

Industrial production of PHAs is mostly done by the use of microorganisms. Both prokaryotic and eukaryotic microorganisms [101] have been reported to naturally store PHA intracellularly as source of carbon, energy and redox material during conditions of nitrogen, phosphorus or oxygen limitation and excess of carbon [91-94]. Until now, more than 90 genera and 300 species of bacteria have been identified as natural PHA producers [102-104] such as: Pseudomonas sp., Cupriavidus necator, Alcaligenus latus, Cyanobacteria and Halomonas. [92]. Bacteria produce PHB from the central metabolite acetyl-CoA (Fig. 15) through a three steps pathway: (1) condensation of two molecules of acetyl-CoA to acetoacetyl-CoA catalyzed 3-keto-thiolase (phaA), (2) reduction of acetoacetyl-CoA to (R)-3-hydroxybutyrate-CoA catalyzed by acetoacetyl-CoA reductase (phaB), and (3) polymerization of (R)-3-hydroxybutyrate-CoA to PHB catalyzed by a PHA polymerase or PHA synthase (phaC). Different composition PHA polymers are obtained by polymerization of different (R)-hydroxyalkanoate-CoA intermediates, which can be obtained by the fatty acid biosynthesis metabolism or the fatty acid degradation metabolism and catalyzed by specific PHA synthases (Fig. 15) [92]. In the 1970s, Imperial Chemical Industries developed a bio-based process for the production of both, PHB and Poly(3HB-co-3HV) by Cupriavidus necator [105]. These polymers were trademarked with the name Biopol. The patents were later sold to Zeneca, then to Monsanto then to Metabolix and then in 2016 to CJ CheilJedang. Other companies, including Bio-On, Tianjin Green Bioscience, Mitsubishi; among others, have produced and/or are producing PHA at pilot scale [92, 97]. Over the years commercialization of PHA has shown several disadvantages that limit their competition with traditional synthetic plastics or their application as ideal biomaterials [106]. These disadvantages are: difficulty to achieve the desired mechanical properties, high production costs, limited functionalities and susceptibility to thermal degradation. Therefore, further modification and control of composition and polymer chain length is imperative to ensure improved performance or even lower costs for specific applications.


25

Figure 15. Schematic overview of the metabolic pathway for production of PHAs from different substrates. Figure redrawn and adapted from Yang et al. (2003) [107]

1.4.2 (R)-3-hydroxybutyrate To circumvent some of the limitations of direct microbial production of PHAs, novel tailor-made copolymers of different hydroxy-carboxylic acids (or their esters) can be made through chemical condensation or polymerization [108-113]. Not only would (R)-3-hydroxycarboxylic acids be a valuable monomer for this, but because of their chiral center and functional groups, (R)-hydroxycarboxylic acids have gained attention as chemical building blocks [108, 114]. These compounds for example have potential applications for chemical synthesis of many compounds such as antibiotics, vitamins and flavors. Table 1 shows some of the possible applications of using hydroxycarboxylic acids as building blocks for the production of pharmaceuticals.

Pathway 1Carbon Sources

(Sugars)

Acetyl-CoA

Acetoacetyl-CoATCAcycle

PhaA

(R)-3-hydroxybutyryl-CoA

PhaB

PhaC

PHA PhaC

(R)-3-hydroxyacyl-CoA

3-ketoacyl-CoA Enoyl-CoA

Pathway 2Fatty acid degradation

(ß-oxidation)Carbon Sources

(Fatty acids)

Acyl-CoA

Succinyl-CoA

4-hydroxybutyryl-CoA

PhaC

FadE

FadD

PhaJFabGPhaB

Pathway 3Fatty acid Biosinthesis

Carbon Sources (Fatty acids)Acetyl-CoAMalonyl-CoA

Malonyl-ACP

3-Ketoacyl-ACP

(R)-3-Hydroxyacyl-ACP

Enoyl-ACP

Acyl-CoA

PhaG


26

R-HAs Potential building block for

(R)-3-hydroxyundec-10-enoate

L-659,699 (inhibitor of cholesterol biosynthesis)

(R)-3-hydroxyundecanoate

Depsipeptides (antibiotic/antifungal)

(−)-tetrahydrolipstatin (anti-obesity drug) Globomycin (antibiotic) (R)-3-hydroxy-nonanoate Globomycin analogs (antibiotic) (R)-3-hydroxyoctanoate Simvastatin (antihypercholesterolemic)

(R)-3-hydroxyhept-6-enoate Sphingofungin D and F (antifungal) β-lactams for synthesis of carbacephems (class

of antibiotics) (R)-3-hydroxyheptanoate Anachelin (siderophore of Anabaena cylindrica) Pravastatin (atherosclerosis/hypercholesteremia

agent) (R)-3-hydroxyhexanoate Analogs of laulimalide (paclitaxel like

antimicrotubule agent)

Table 1. Potential applications of hydroxycarboxylic acids as building blocks for the production of pharmaceuticals. Table adapted and extracted from Ren et al. (2010) [108]

(R)-3-hydroxybutyrate (3HB), the focal product of the work of this thesis (Fig. 16) is a (R)-hydroxycarboxylic acid formed by a four-carbon chain and a hydroxy group on the chiral third carbon atom and is the monomer of the well-known PHB.

Figure 16. General structure of (R)-3-hydroxybutyrate.* Chiral center

This metabolite has gained special attention for its potential applications as:

a) Monomer in condensation reactions for production of tailor-made polyesters with high mole

b) cular weights or different monomer compositions. For example, Song et al. (2000) [115] polymerized monomers of both 3-hydroxypropionyl-CoA

O

OH

OH

*


27

(3HP-CoA) and 3-hydroxybutyryl-CoA (3HB-CoA) using Wautersia eutropha PHB synthase to make novel co-polyesters.

c) Building block for a range of valuable compounds such as antibiotics, aminoacids, vitamins, aromatics and pheromones [110, 116, 117]. For instance, 3HB esters have been used as building block for synthesis of an antiglaucoma drug developed by Merck called “Trusopt”[118, 119]. Another example is the use of 3HB as a precursor for the production of pharmaceuticals, such as captopril and β-lactams [120-122]. Additionally, 3HB is an important building block for the production of carbapenem antibiotics [109, 117, 123] and both (+) -colletediol [124] and di-O-methylelaiophylidene [110].

d) Nutritional source for eukaryotic cells. Also known as beta-hydroxybutyric acid, 3HB is a ketone body, which are molecules produced in the liver of mammalian cells for use when glucose is not available or during periods of starvation. The molecule’s signaling- and regulatory functions have implications for metabolic neurological diseases. For example, 3HB has been found to be an endogenous inhibitor of histone deacetylase enzymes, for which it is considered for clinical therapy for intractable epilepsy and Alzheimer [125, 126].

3HB has been produced via the following production methods:

a) Chemical synthesis. Because of the difficulty on synthesizing chiral (R)-3HB, very few reports are available on chemical synthesis of pure 3HB in the R configuration. Noyori et al. (1987) [127] were able to produce pure optical (R)-3HB by asymmetric hydrogenation of β -keto carboxylic esters. Other synthetic asymmetric routes for chemical production of enantiopure alpha and beta hydroxy acids use either expensive chiral metal catalysts, expensive substrates, extreme reaction conditions or have limited enantiomeric purity [128, 129].

b) Chemical hydrolysis of polyhydroxyalkanoates. Chemical degradation using sulfuric acid, of PHB and P(3HB-co-3HV) has been reported [130] . In addition alcoholysis with anhydrous hydrochloric acid of PHB has been reported as a method to hydrolyze 3HB [131]. Even though high chiral purity is obtainable, significant amounts of the by-product methylcrotonate are also formed. Furthermore, this approach requires the use of strong acids or bases and two consecutive main processes to arrive at the final product.


28

c) Enzymatic hydrolysis of polyhydroxyalkanoates. Enzymatic degradation can be done in vitro or in vivo. When done in vitro, PHAs first have to be purified and then a purified depolymerase degrades the polymer. One example is the use of the thermophilic depolymerase from Streptomyces sp. which has been reported to hydrolyze PHB to 3HB [132]. When done in vivo, PHAs are first produced by wild-type bacteria and then depolymerized by intracellular depolymerases. This concept was described for esters of 3HB produced by wild type Alcaligenes latus, where hydrolysis with its natural depolymerase was occurred after harvesting and resuspending of the cells into water (pH 4) [133]. Drawbacks with this approach are the low required pH (if the ester is recovered) and the requirement to run two consecutive processes to arrive at the final product. Furthermore, the depolymerized products could be further metabolized by 3HB dehydrogenase.

d) Microbial synthesis by genetically engineered strains from carbon sources such as glucose, sucrose or sugars derived from other renewable raw material. This approach is promising since it avoids the extreme conditions used in chemical synthesis or depolymerization and avoids the use of two consecutive processes when 3HB is obtained from PHB hydrolysis. Initially there were reports on metabolic engineering of E. coli expressing heterologous genes from Cupriavidus necator (phaA, phaB, and phbC) (Fig. 15) together with a depolymerase gene to do the hydrolysis of PHB to produce 3HB ([134, 135]. It was later discovered that 3HB could also be produced without the polymerization and hydrolysis steps and instead using only phbA and phbB and a thioesterase to hydrolyze 3HB from 3HB-CoA [123, 134, 136-138].

1.5 Steering production

When an interesting microbial product has been identified there are a number of aspects to consider or improve before an economically viable industrial process can be launched [4]. Decisions for capital investment will be based on a preliminary analysis of the efficiency of the bio-process, which includes both the cultivation and the following downstream process (not included in the scope of this thesis.). An economically ideal bioprocess should be able to produce large amounts of the product in short amounts of time with little byproducts and be run under conditions that will lead to reduced purification costs. A primary requirement is that the yield of product on substrate is sufficiently high since this yield, together


29

with the difference in value between the substrate and the product will determine the profitability of the production process. The estimation of costs is based on the results of mass and energy balances and equipment sizing (which in turn is calculated based on the mass balances). An intrinsic part of mass balances is the rate of production or consumption of a compound in a cultivation. For cellular transformations the individual cell production/consumption rates better quantify the performance of the strain. These parameters are called biomass specific rates. In a microbial cultivation, where biomass is represented by a subscript x, which for instance gives the cell mass in grams as gx. Analogously, for a compound ‘i’ the biomass specific rate it is represented by 𝑞" # $%

$&∙)*. Specific rates can be defined for any compound: 𝑞+ for

product formation, 𝑞, for substrate consumption, 𝑞-. for oxygen consumption, 𝑞/-. for CO2 formation. For the specific case of the specific cell mass production rate, µ # $&

$&∙)* is commonly used instead of 𝑞1and is called growth rate. The maximum

growth rate is dependent on the genetic background of the specific cell, the temperature and the composition of the medium. The actual growth rate depends on the capacity of the uptake systems for substrates in the cell membrane of an organism, which is dependent on the concentration of the substrate in the medium. This situation is described by a model of saturation kinetics and is represented by a mathematical expression called “Monod equation” [139] shown below. Where 𝐶, 3$45 6is the variable substrate concentration and 𝐾,3$45 6 is the substrate saturation constant which refers to the substrate concentration at which the µ = 9:;&

..

µ = µ<=1 ∙𝐶,

𝐾, + 𝐶,

In order to evaluate the rate at which a determined compound is either produced or consumed per volume of broth the volumetric rate is used. This parameter is important for capital investment costs (lower volumetric productivities will require increased bioreactor volumes in order to reach specific demands). For a compound “i”, the volumetric rate is represented by 𝑟"3 $%5@)6. The total rate of production or consumption of a compound (𝑅"3$%) 6) directly reflects the amount of product produced per amount of time which directly reflects the capacity of production of the cultivation part of a bioprocess in a production plant. The rate parameters are related as shown by the equations below. The concentration of cells is represented by 𝐶13$&5 6 and the total volume of broth contained in the reactor is represented by 𝑉D[𝐿].


30

𝑟" = 𝑞" ∙ 𝐶1; 𝑅" = 𝑞" ∙ 𝐶1 ∙ 𝑉D

In a bioreactor, for any compound “i” the mass balance can be written as:

𝑑(VD · C")𝑑𝑡 =𝑅" + 𝐹"P ∙ 𝐶","P − 𝐹STU ∙ 𝐶",STU + 𝑇W,"

Where the inflow/outflow of medium is represented by 𝐹 and the rate at which a component is transferred from/into another phase by 𝑇W,". Depending on the operation mode the mass balance can be resolved and specific rates can be determined as well as subsequent volumetric and total rates. Using the common modes of operation for the liquid components of the bioreactor: - Batch. This cultivation is initiated by an inoculation of cells to the bioreactor

containing cultivation broth to give an initial cell mass concentration of 𝐶1,S. The initial substrate concentration is 𝐶,,S and there is no liquid flow going in or out from the bioreactor. Assuming a constant volume, this gives:

𝑞" =1𝐶1∙𝑑𝐶"𝑑𝑡 → µ =

𝑑𝐶1𝑑𝑡 ∙

1𝐶1;𝑞, = −

𝑑𝐶,𝑑𝑡 ∙

1𝐶1;𝑞+ =

𝑑𝐶Z𝑑𝑡 ∙

1𝐶1

- Fed-batch. This type of cultivation was developed as a way to increase biomass

yield in the production of baker’s yeast [140]. By controlled addition of carbon, the issue of overflow metabolism is overcome and specific oxygen consumption is controlled. The fed-batch is typically started after a batch in which biomass concentration has increased to a desired level and usually the level of the selected limited substrate has reached very low levels. During this period no liquid broth is extracted from the bioreactor which means VD is a function of time.

𝑞" =𝑑(VD · c")

𝑑𝑡 − 𝐹"P ∙ 𝐶","PV𝐿 ∙ C𝑥

µ =𝑑(VD ∙ C1)

𝑑𝑡 ∙1

VD ∙ C1; 𝑞, =

𝐹"P ∙ 𝐶,,"P −𝑑(VD ∙ C,)

𝑑𝑡VD ∙ C1

; 𝑞+ =𝑑(VD ∙ C+)

𝑑𝑡 ∙1

VD ∙ C+

- Continuous cultivation (steady state). This cultivation is generally started after

a short batch phase. The batch phase contains low concentrations of a substrate to assure it becomes limiting before other substrates in the medium, which gives a growth restriction inside the bioreactor. Whereas the rate of


31

medium addition determines the specific growth rate, the concentration of this growth-limiting component in the medium will determine the cell concentration. The attainment of a steady state requires that the broth volume is balanced and maintained by regulating the outflow rate. Cells and products dissolved in the broth leave the reactor through the liquid effluent. Assuming 𝐹STU = 𝐹"P = 𝐹, 𝑉D is constant and steady state (“](^5·_%)

]U” = 0), this gives:

𝑞" =𝐹 ∙ (𝐶",STU − 𝐶","P)

𝐶1 ∙ 𝑉D→ µ =

𝐹𝑉D= 𝐷;𝑞, =

𝐷 ∙ (𝐶,,"P − 𝐶,)𝐶1

;𝑞+ =𝐷 ∙ 𝐶+𝐶1

The efficiency of a bioprocess is determined by specific production parameters. The criteria of optimization of these parameters highly depends on the type of product. The biotechnological products are of very diverse kinds, at one extreme there are multi-ton products such as beer which have lower prices and at the other extreme there are products that are required only in few kilograms quantities such as special pharmaceuticals which have a much higher price (Fig. 17).

Figure 17. Plot of the price of biotechnology products vs. their annual production volume. Data extracted from Schmid R. (2000) [141]

For products closer to the first extreme “high volume/ low value (HVLV)” the most important design parameters are titer, productivity and yield [4]: - Final product titer 3$%5 6. This parameter indicates the final concentration of

product attained in a cultivation process. For extracellular products this


32

parameter will directly affect production costs since this parameter directly affects the operating and capital cost of especially the downstream processes (DSP) [4]. If the final titer is very low it may be very difficult and expensive to extract and purify from the cultivation medium with good enough yields. Higher titers, on the other hand, can result in easier DSP processes since less volume is sent to process and there is less water to remove.

- Rate or Productivity. This parameter indicates how much product can be produced per time unit. For most processes it is an economical benefit if the production process is short since production demands must be satisfied and if in a process the volumetric productivity is not high enough then new and/or larger reactors have to be installed which will directly affect capital investment costs. During process development strains are optimized and the specific production rate directly reflects the performance of the microorganism. It is beneficial if the specific production rate is high enough to assure efficient working cells able to produce product in specific amount of time.

- Yield #$%$a*. The yield, also referred to as the efficiency factor, is defined as the

quotient between two rates:

𝑌"c = d𝑞"𝑞cd = d

𝑟"𝑟cd = d

𝑅"𝑅cd

The suffixes i and j refer to two different compounds in the process. The yield is always positive and relates any of the compounds existent in the cultivation (cells, substrate, product, overflow metabolite, etc.) with another. For high-volume low-cost products, it is a benefit if the yield is maintained close to the theoretical yield since substrate costs can account for more than 80 % of the total costs, which means that special attention should be paid to make sure that substrates are not converted to a byproduct or leave the cultivation process as waste. When talking about an excreted metabolite the yield 𝑌ef is also of special interest since it is beneficial if less substrate is directed towards cell production and more substrate towards product formation.

As the product moves more towards the other extreme “low volume/high value added (LVHV)” quality gains more importance relative to the other parameters [4]. - Quality. This parameter indicates [142, 143] the totality of features and

characteristics of a product to standards that achieve uniformity in order to satisfy specific costumer (or user) or legal requirements. This parameter is always strictly defined by the specific product. Quality is important since products need to maintain a certain purity and characteristics that could


33

drastically change if the process would be changed or process deviation occurred.

Another important parameter for all processes is robustness: - Robustness. This parameter refers to that a production process should be able

to operate in the same way within boundaries and ranges of variated conditions, materials and equipment without compromising titer, yield, quality or productivity. A decrease in the stability of a cell, such as mutations, is a common problem in long cultivations or continuous cultivations.

Once targets for process development parameters titer, rate, yield, quality and robustness (TRYQR) have been defined it is necessary to investigate if they can be met through implementation of strain- or bioprocess-design. 1.5.1 Steering production by strain choice and -design

When a commercially interesting product has been identified and the commercially relevant ranges of the process development parameters TRYQR have been determined by conceptual process design, a next key strategy is to choose and/or design the specific microbial strain so that it performs the conversion process in an efficient manner. All the electrons and carbon in the substrate will end up in the desired product, biomass and/or byproducts which come from a number of diverging pathways. The goal is to understand how carbon flow can be regulated in the complicated metabolic network of a cell. By engineering the cellular pathways, changes will be made in the genome of the organism that may allow for enhanced production of the compound. Traditionally, strain improvement has been done by classical strain improvement (CSI). The efficacy of CSI is clearly illustrated by for example the industrial production of penicillin, citric acid and lactic acid (as described above). These techniques are powerful means of improving the properties of production strains and are still highly used. Nonetheless with the use of model host organisms and recombinant DNA technology it is now possible to use “metabolic engineering” as a more direct and rational approach towards strain improvement. Metabolic engineering is defined by James Bailey as: “The improvement of cellular activities by manipulation of enzymatic, transport, and regulatory functions of the cell with the use of recombinant DNA technology” [144]. Through metabolic engineering we can modify the cellular activities and the metabolism of a cell factory and improve production process parameters. Furthermore, by introduction


34

of heterologous genes, metabolic engineering enables the creation of novel cell factories leading to new products. An illustrative example of introduction of heterologous genes and improvement of process parameters is the heterologous production of 1,3-propanediol (PDO) by Escherichia coli developed by DuPont and Genentech (as described above). To make a viable process, an E. coli K12 strain capable of efficiently converting glucose to PDO was designed (Fig. 18). This conversion requires the combination of two pathways: glucose to glycerol and glycerol to PDO. Since yeast are known to have efficient pathways for production of glycerol from dihydroxyacetone phosphate (DHAP), genes encoding glycerol 3-phosphate dehydrogenase (G3PDH) and glycerol 3-phosphate phosphatase and (G3PP), DAR1 and GPP2 respectively, were cloned from the yeast S. cerevisiae. The genes: dhaB1-3 (encoding glycerol dehydratase) together with its reactivating factors (dhaBX, orfX) which catalyze the conversion of glycerol to 3-hydroxypropionaldehyde and dhaT (encoding 1,3 propanediol dehydrogenase) which catalyzes the reduction of the aldehyde to PDO were cloned from the native PDO producer K. pneumoniae. Although recombinant expression of these genes enabled production of PDO, further engineering (explained lines below) of E. coli was required to improve titer, productivity, yield, and quality and achieve efficient commercial production of PDO [78, 79].

Figure 18. Schematic overview of the pathway for production of PDO by recombinant E. coli. Figure redrawn and adapted and redrawn from Nakamura et al. (2003) [78].

- Yield. By influencing the rates of translation and transcription of DAR1 and

GPP2 and their relative position, enzyme expression levels were varied to

glycerol


1,3-propanediol

H2O

NADH

NAD+

dhaB1-3

dhaT

DHAP GAP

Glucose

To TCA cycleand respiration

tpi

PEP

PEP

NADH

NAD+

G3P

DAR1

GPP2

Glucose-6-P


35

obtain optimal conversion of glucose to glycerol. Since E. coli is capable of re-consuming glycerol, the genes encoding glycerolkinase (glpK) and glycerol dehydrogenase (gldA) were deleted from the chromosome (Fig. 19). In a first step the isomerase tpi was removed to force a 50:50 split of carbon flow to glycerol production and to the TCA cycle. This was later reversed since it created an artificial roof to the yield to a maximum of 50 % of the carbon. Instead reduction of the expression of glyceraldehyde 3-phosphate dehydrogenase (gapA) was done by replacing its natural promoter with promoter variants that decreased its expression. Glucose transport and phosphorylation using the phosphotransferase system (PTS) was another limitation to the yield since it requires an equivalent amount of PEP per mol of glucose. This means that each mol of glucose will provide carbon for a maximum of 1 mol of PDO, thereby limiting the yield. To overcome this limitation, genes of the PTS system were deleted. Glucose transport was then restored by enabling facilitated diffusion through promoter replacement of the galactose permease. Subsequent phosphorylation of glucose was accomplished by replacement of the native promoter of the ATP dependent glucokinase by a trc constitutive promoter.

Figure 19. Schematic overview of the metabolic engineering done in recombinant E. coli to enhance PDO production by DuPont and Tate&Lyle.

- Titer. The maximum PDO titer achieved with the previously improved strain

was lower than desired for a final process, while glycerol production continued. To solve this an effort was made to explore alternative enzymes for the conversion of glycerol to PDO. As control, a strain was constructed which did not carry the gene dhaT (PDO oxidoreductase). Predicted not to produce PDO,

glycerol


1,3-propanediol

dhaB1-3, B12 K. pneumaniae

yqhDE. coli

DHAP GAP

Glucose

tpi

G3P

GPP2 (S. cerevisiae)

gap

glpK

DAR1(S. cerevisiae)

NADH

NAD+

NADPH

NADP+

H2O

gldA

DHA

To TCA cycleand respiration

PEP

GlucoseATP

glucokinase


36

this strain surprisingly did not only produce PDO but even increased the previous PDO titer with 66%. It was later discovered that a native E. coli NADPH dependent oxidoreductase (yqhD) was induced in the absence of dhaT (Fig. 19). The higher NADPH/NADP+ ratio (compared to the NADH/NAD+

ratio) is likely the cause of the higher titer since it pushed the equilibrium of this reaction towards PDO and NADP+ formation.

- Productivity. Acetate accumulation is unquestionably one of the major problems during high density cultivation of E.coli, since it inhibits cells and takes a significant part of the carbon flux. For this reason, expression of PEP carboxylase (ppc) was increased to improve the performance of the TCA cycle and reduce acetate accumulation. Furthermore, the methylglyoxal synthase gene (mgsA) which converts DHAP to methylglyoxal was deleted to avoid introduction of DHAP into central metabolism. These modifications allowed for increased volumetric rate since no control of by-product formation by glucose-limited fed-batches were needed [145]. The resulting strain was able to produce PDO with high yield, rate and titer of 0.9 molPDO mol Glc, 3.5 g L-1 h-

1, and 135 g L-1 respectively.

- Quality. The compounds glycerol, acetate and 3-hydroxypropionic acid are potential by-products in the aerobic cultivation of recombinant E. coli for production of PDO. These compounds are low molecular by-products that can be hard or expensive to remove in the subsequent downstream processing. PDO is a monomer that is condensed with terephtelate to form the polymer PTT. If instead either acetate, 3-hydroxypropionic acid were added in the growing chain of the polymer, the condensation reaction would stop because of the lack of hydroxyl groups. This would result in changed polymer composition and compromise quality. Native genes encoding phosphotransacetylase (pta) and acetate kinase (ackA) are responsible for formation of acetate from acetyl-CoA. Native genes encoding aldehyde dehydrogenase (aldA and aldB) are responsible for 3-hydroxypropionic acid formation from 3-HPA. These four genes were deleted from the chromosome of E. coli to avoid the formation of the carboxylic acids that could potentially compromise quality.

Strain design can also be used to influence the process development parameter robustness. Below an example is shown an example of how this has or can be done for another product and strain.


37

- Robustness. With over 1.95 billion hl produced annually [146], beer is one of the most widely produced and consumed biotechnological product. It is one of the oldest known alcoholic beverages. The largest producers are China, Germany, Brazil, Japan, Russia and the United States [141]. Traditionally yeast is used for the production of beer from maltose hydrolyzed from malt in large-scale bioreactors. Despite maltose being the main product of the hydrolysis, small amounts of glucose and sucrose may sometimes be present in the culture medium. This results in intermittent lag phases between the consumption of glucose and other sugars, which may have a costly impact in the fermentation process. Decreasing the sensitivity to medium variation has been suggested by alleviation of glucose control (repression) on sucrose and maltose metabolism [147]. Deletion of genes encoding transcriptional repressor proteins or overexpression of genes encoding positive transcriptional activators have been used in both laboratory and industrial strains [148-151]. These modified strains could be the solution if GMO would be acceptable for beer production in the future or if mutants with similar phenotype were obtained by CSI.

1.5.1.1 Genomic tools. In order to design or engineer the metabolic pathway in a strain, changes are made in the genome of the organism that may allow for enhanced production of a compound. This section will present some of the genomic tools available for increasing the process parameters (TRYQR). (1) Whole genome sequencing (WGS), which refers to the analysis of the entire

genomic DNA, has dramatically improved the potential of metabolic engineering approaches since it enables to relate variances of phenotype and associate them with changes of genotype. A challenge in the design of metabolic engineering strategies is identifying flux controlling steps of a metabolic pathway of an organism. To overcome this challenge, Inverse metabolic engineering (IME) emerged as a convenient framework that integrates evolutionary strategies or CSI with metabolic engineering strategies. The details and definition of Inverse metabolic engineering were first explained by Bailey et al (1995) [152] as: “the elucidation of a metabolic engineering strategy by: first, identifying, constructing, or calculating a desired phenotype; second, determining the genetic or the particular environmental factors conferring that phenotype; and third, endowing that phenotype on another strain or organism by directed genetic or environmental


38

manipulation.” This technique has successfully been used to guide metabolic engineering strategies [153, 154].

(2) Genome-scale modeling. The complexity of metabolic networks together with the development of whole genome sequencing has led to the development of genome scale metabolic models [155]. These models contain the stoichiometric mathematical representation of the complex cellular metabolism showing the thousands of metabolic reactions and metabolites that occur in a cell. This information is extracted from databases such as KEGG and MetaCyc which are available for a number of host organisms. With these models, Constraint-based reconstruction and analysis (COBRA) methods such as MFA (metabolic flux analysis) are used to help predict metabolic flux distributions on the basis of simulating for example genetic modifications (new pathways or knock out of genes) or the use of different various substrates [156]. MFA can guide metabolic engineering strategies since it allows us to predict which modifications can help us increase bioprocess development parameters.

(3) Genetic engineering Metabolic engineering relies in the regulation of metabolic pathways by: changes in amino acid sequence, decreased expression of a gene, elimination of native genes, expression of heterologous genes and overexpression of native genes. Regulation of enzyme expression levels (decreased/increased) is generally done by using inducible promoters, varying promoter strengths, codon optimization and variation of the strength of ribosomal binding sites. Expression of foreign genes or overexpression of native ones has relied in recombinant DNA technology by the use of multi copy plasmids or by chromosomal integration. The employed plasmids can require selection markers, have different copy numbers and can impose metabolic burden. Furthermore, they often possess limited capacity for harboring genes. Chromosomal integration has the potential to circumvent with plasmid’s issues and result in more robust strains with higher genetic stability. Because of this, chromosomal integration is the preferred tool in industry. Nevertheless, for prokaryotic organisms, this procedure is not as simple as plasmid cloning and transformation for which the facility to modify and transfer DNA is lost. CRISPR/Cas9 systems (clustered regularly interspaced short palindromic repeats/CRISPR-associated proteins) and other similar systems are the basis for new genome editing methods that have gained enormous attention since 2013 [157]. By synthesizing synthetic CRISPR arrays for targeted genomic editing these systems can contribute as genomic tools for metabolic engineering. CRISPR/Cas9 has not been used in the present


39

investigation since the methodology of this thesis was designed before the establishment of CRISPR/Cas9 as a genomic editing tool.

1.5.2 Steering production by bioprocess design

In addition to strain engineering, also bioprocess design plays an important role in the economic optimization of a biotechnological product process. Bioprocess design is defined as [158]: “A part of chemical engineering that deals with the design and the development of the equipment and processes for the manufacturing of products, fuels and treatment of waste water by using microbial, animal or plant cells”. In the section below some examples are shown of how bioprocess design is or can be used to influence the 5 process development parameters titer, rate, yield, quality and robustness in the bioprocessing of different compounds. - Yield and titer. The use of yeast for making bread from flour has been

documented since Ancient Egypt. Until the 19th century bread bakers obtained their yeast from beer brewers. Somewhere around the 1870s Louis Pasteur introduced a more advanced method of producing baker’s yeast with high yield and little formation of ethanol by using strong aeration [141]. Saccharomyces cerevisiae strains have been used for long because of their dough-leavening characteristics. This yeast is a facultative anaerobic organism thus it can grow both with and without oxygen. Compared to growth in aerobic conditions, in anaerobic conditions smaller yields are observed since less ATP is available to build up cells and the product of fermentation ethanol is increased as a way to oxidize back the NADH produced in glycolysis. Baker’s yeast is a low-price product which means that the production yield with respect to substrate should be high. Due to its low costs and high content in sugar and nitrogen sources, vitamins and minerals, molasses are preferably used as feedstocks for the production of baker’s yeast. The aerated glucose limited fed-batch process was developed in the 1920s as way to increase biomass yield in the production of baker’s yeast [140]. To obtain high yield #$&$4*and titers 3$&5 6 of cell mass (the

main product) 4 factors are essential for this fed-batch process: (1) strong aeration during the cultivation to avoid ethanol formation. (2) limited glucose feed with strict control of its concentration in the medium. If glucose concentration exceeds 100 mg L-1 ethanol can be formed because of the Crabtree effect or respiro-fermentative metabolism. (3) To assure that the aeration provided in the reactor (oxygen transfer rate) is sufficient for a full aerobic cultivation (no oxygen limitation), the glucose feed rate should be


40

maintained low enough to be able to control the growth rate and consequently the specific oxygen consumption rate. (4) To assure a robust process the growth rate should be maintained sufficiently below the calculated maximum growth rate based on the oxygen transfer rate of the bioreactor and the residual glucose concentration in the bioreactor.

- Titer. Bioethanol is a bio-based fuel that was introduced into the transport fuel supply in the 1970s in Brazil [159]. Bioethanol may be used as an additive for petroleum-based fuels or as a complete replacement. This biofuel has emerged as one of the strongest candidates to move away from the dependency of fossil-based fuels [159]. The US and Brazil together account for more than 94 billion liters of ethanol produced per year which constitutes for around 85% of the global production [160]. Traditionally bioethanol is produced by Saccharomyces cerevisiae by fermentation of sucrose syrup or sugar cane molasses in Brazil or of sugar syrup from hydrolysis of corn starch in the US. This yeast is tolerant to ethanol accumulation in the early stages of accumulation but as the ethanol titer increases ethanol becomes more and more toxic to the cells to a point in which ethanol production ceases. S. cerevisiae strains are capable of growth over a wide range of temperatures from 5-40°C [161, 162]. Under industrial production, temperatures between 30-35°C have been used, since the metabolic activities of the strain are highest [163, 164]. For application of very high gravity technology in which mashes with over 27 g of dissolved solid/100 g are used the ethanol toxicity doesn’t allow for high titers to be obtained which results in unconsumed sugars. To circumvent this limitation Thomas et al. (1993) [164] and Jones et al. (1994) [165] suggested decreasing the temperatures to alleviate the effects of toxicity and reactivate the metabolic activities for ethanol formation. In laboratory scale with a mash of 38% of dissolved solids and observed after 192h, 22.4% (v/v) of ethanol was obtained at 20°C compared to 18.2% (v/v) obtained at 30°C .

- Productivity. Besides the fact that in Brazil bioethanol fermentation is done from sucrose syrup or sugar cane molasses and in the US it is done from sugar syrup from hydrolysis of corn starch, another remarkable difference is the recycling of cells that is employed in Brazil’s cultivation processes. This recycling technique was developed for increasing productivities and it was patented in 1937 by Firmino Boinot in France [160]. After the end of each fermentation the broth is centrifuged and cells are separated from the broth.


41

The supernatant goes to subsequent distillation while the cells return to the fermentation tanks for a new fermentation cycle. Due to high concentration of solids (>30%) recirculation is not possible in processes run in the US. In both the US and Brazil high ethanol yields are obtained 85-90% and 90-92% respectively. Although the processes run in Brazil have a higher productivity since 7-12% (v/v) of ethanol titer is obtained in 6-12 hours while in the US 12-18% (v/v) of ethanol titer is obtained in cultivations that lasts 45-60h.

- Quality. Escherichia coli has been widely studied for the recombinant production of proteins [166, 167]. The fed-batch technique used to obtain high cell density cultures of E. coli has allowed high yields and productivities of protein [166]. Quality of proteins can be compromised by two common limitations in recombinant protein production by E. coli, the formation of insoluble protein aggregates (inclusion bodies) and the low stability of the recombinant proteins due to proteolytic degradation [168]. Sandén et al. (2005) [168] suggested a hypothesis to minimize these limitations for the production of a type of proteins which translocate to the periplasm. In these cases, the formation of soluble protein is believed to depend on the rate of synthesis of messenger RNA (rate of transcription). They suggest that the rate of transcription can be influenced by changes in the inducer lever and by the level of substrate feed. In their study they took the model proteins Zb-Ma1E and Zb-Ma1E31 which are based on isoforms of the Maltose Binding Protein. They show that both proteolysis and inclusion body formation was highly reduced by lower substrate feeds and moderately reduced by lower inducer concentration. These results show that by designing low feed profiles, control of both proteolysis and inclusion body formation is possible.

- Robustness. Commercial baker’s yeast is either packaged as refrigerated yeast cream or active dried and vacuum packed. In either case yeast must be sufficiently robust to tolerate the storage and still be active for subsequent baking applications. The fermentative capacity after storage is best maintained in dry yeast with high glycogen and trehalose content (a carbon storage molecule) [169]. In industrial baker’s yeast production towards the end of the cultivation process the specific growth rate is decreased then the process terminates by starving the cells for nitrogen [170]. Using these bioprocess design techniques, it is possible to obtain glycogen and trehalose accumulation in the cells allowing for more robust cells.


42

Challenges and opportunities The state of the art of biorefineries as an alternative to replace petroleum refineries was discussed above, with a specific focus on the use of microbial-cell-factory-based processes for sustainable production of fuels and chemicals. Despite the start-up and success of many industrial biochemical processes that use cell factories over the past decade, multiple challenges remain for the bio-based economy. These challenges are often directly related to the overall process, such as high production costs compared to fossil-based processes, lack of scalability, and time from proof-of-principle to commercial production. In turn the production costs are influenced by the performance and limitations of the cell-factory: low yield, limited substrate range, product inhibition, by-product formation, or use of energetically inefficient pathways. To overcome these limitations there is a need to develop and implement scalable robust bioprocesses with increased yield, titer and productivity. At currently low oil prices, it is beneficial for process economy to make products that are not only sustainable, but also add functionality to the scope of chemicals commonly produced by petrochemical refineries. Additionally, robust, scalable, sustainable and efficient processes for production of value-added products maintain their competitive edge independent of future oil-price development.


43

2. Present Investigation

2.1 Aim and strategy

The aim of this thesis was to contribute knowledge on how to improve production of a medium-value chemical relevant for a biorefinery, by using bioprocess- and strain-design strategies. Because of its commercially valuable applications the product selected was the chiral compound (R)-3-hydroxybutyrate (3HB). The main focus of the project was the development of a robust and scalable bioprocess with improved titer, yield and productivity of 3HB. As a model system an E. coli host strain expressing heterologous enzymes of the 3HB-CoA pathway of Halomonas boliviensis was chosen. In this thesis, four different strategies for improved, robust and scalable 3HB production were investigated. In the first strategy the focus was to uncouple growth and product formation by bioprocess design strategies that are inspired by natural PHB production, which mainly occurs under nutrient limitation or depletion (Paper I). The second strategy focused on strain- or bioprocess-design strategies involving the engineering of the redox-cofactor supply for 3HB production (Paper II). The third strategy focused on strain design strategies to increase yield, rate and titer by reducing or eliminating competing pathways towards formation of inhibitory byproducts (Paper III,IV). Using 3HB as a model product, the fourth strategy focused on the implementation of a new recombinant expression system, which has the potential to be used as a screening system and/or to increase yield, titer and rate (Paper V).


44

2.2 Escherichia coli as a model production organism

Because of their relatively simple structure bacteria have extensively been used to study many fundamentals in biochemistry and in molecular biology. E. coli is a gram-negative, rod shaped, facultative anaerobic bacterium which was discovered by Theodor Escherich in 1885 [171, 172]. The first complete genome of E. coli was published in 1997 and corresponded to laboratory strain K-12 derivative of MG1655 [173]. Its DNA is around 4.6 million base pairs in length and contains 4288 annotated protein-coding genes. In 1973 Stanley Cohen and Herbert Boyer developed for the first time DNA cloning and recombinant DNA, using E. coli host cells [174], opening the possibility of producing different proteins by microbial cultivation. The first recombinant protein “Insulin” was industrially produced with recombinant E. coli in 1982 [174]. The advantages of using this organism are: a) Well adapted to grow on many different carbon sources with the addition of

only simple mineral salts for nutrients and very high cell densities are easily achieved [175, 176].

b) Many E. coli strains are capable of taking up five carbon sugars such as xylose and arabinose, both contained in hemicellulose a major constituent of lignocellulose.

c) Under optimal growth conditions doubling times can be as short as 20 min, compared to for instance 1-2 h for Saccharomyces cerevisiae and >24 h for mammalian cells.

d) Extensive information is available about metabolism, physiology and genomics of this bacterium, which facilitates further metabolic design and genetic manipulation.

e) Very well-established genetic engineering tools available for DNA modification as well as very high transformation efficiencies, which facilitate cloning and expression of recombinant genes.

All these advantages make E. coli an attractive host for the production of value-added products. This is further exemplified by the industrial use of E. coli for production of chemicals, such as succinic acid produced by Myriant or 1,3-propanediol by the joint venture Dupont Tate & Lyle.


45

2.3 Production of 3HB by E. coli expressing recombinant genes from Halomonas boliviensis

Halophiles are organisms that can be found as archaea, bacteria and eukarya which have evolved to cope with high salt concentrations allowing them to grow in saline or hypersaline environments [177]. Because of their biotechnological potential, many screening programs have been developed in saline habitats to isolate and characterize novel enzymatic activities [114]. Applications of enzymes derived from these organisms have been explored for their biotechnological potential in the production of diverse metabolites [178]. A recent example is the ability to produce polyhydroxyalkanoates. An example of a native PHA producer relevant for this thesis is Halomonas boliviensisis, a halophilic, gram negative bacteria which can tolerate NaCl concentrations ranging from 0-250 g L-1. In 2004, this bacterium was isolated from a soil sample around lake Laguna Colorada at 4300 meters above sea level in Bolivia. It is known to accumulate PHB at high yield and at high productivity (81% of the cell dry weight and 1.1 g L-1 h-1 respectively) [179, 180]. However, a problem with the use of halophiles is the high salt concentration requirement which can damage industrial bioreactors. Furthermore, the lack of available information about metabolism, physiology and genomics of H. boliviensis hinders genetic manipulation and metabolic engineering. Nevertheless, because of H. boliviensis’ efficient PHB production, this bacterium is an attractive source of enzymes for the production of PHB and 3HB. To overcome the challenges of halophilic wildtype organisms, recombinant expression of enzymes of the metabolic pathway of PHA in a model host organism is proposed. Because of the advantages of using E. coli as a host organism and the commercially valuable applications of (R)-3-hydroxybutyrate, the enzymes of the PHB pathway of H. boliviensis were previously used for construction of a (R)-3-hydroxybutyrate (3HB) producing E. coli strain [181]. This strain harbors recombinant genes from H. boliviensis to form a three-step pathway for 3HB production (Fig. 20): i) Condensation of two molecules of acetyl-CoA to acetoacetyl-CoA catalyzed by a 3-keto-thiolase. ii) Reduction of acetoacetyl-CoA to (R)-3-hydroxybutyrate-CoA catalyzed by an acetoacetyl-CoA reductase. iii) hydrolysis of the thioester bond of (R)-3-hydroxybutyrate-CoA to result in (R)-3-hydroxybutyrate catalyzed by a native E. coli thioesterase and subsequent secretion to the medium. For this purpose, the genes for 3-keto-thiolase (t3) and acetoacetyl-CoA reductase (rx) from H. boliviensis were cloned in the order t3-rx into the low-copy number plasmid PACYC184 under control of the lacUV5 promoter with lacI repressor resulting in the plasmid pJBGT3RX, which was subsequently transformed into the E. coli


46

laboratory strain AF1000 [181]. Assuming sufficient activity of a native thioesterase catalyzing the conversion of 3-hydroxybutyrate-CoA to 3HB [182], no heterologous gene encoding a thioesterase was introduced [181].

Figure 20. Schematic overview of the pathway for production of (R)-3-hydroxybutyrate by recombinant E. coli. 2.4 Overview of methods used in this thesis Unless stated otherwise, in this thesis, the following materials and methods were used for heterologous 3HB production by E. coli. Strains and genetic background The E. coli strain background was AF1000 (MC4100 relA1+). Recombinant gene expression was done by the use of low copy plasmids (~15) with inducible promoters. For 3HB production, a pACYC184-derived plasmid “pJBG” harboring genes t3 and rx in the same operon under control of a lacI repressed lacUV5

glucose

glycolysis

acetyl-CoA

zwf

NADPH

acetoacetyl-CoAPi

acetyl-P (R)-3HB-CoA

NAD(P)H

thioesteraseATP

Pi

NH4+

biomass (R)-3-hydroxybutyrateacetic acid

t3

rx

acs

ackA

pta

pyruvate

poxB

ATP

isocitrate

malate

glyoxylate α-ketoglutarate

glutamate

NADPH

PPP

iclR

succinate

acetyl-CoAcitrate

synthase

TCA-cycle


47

promoter was used. Overexpression of other enzymes was based on a pBAD/hisC (Invitrogen) derived plasmid. Cultivation design The cultivation medium was based on a heat-sterilized minimal salt medium with glucose as the sole carbon source, NH4+ based salts as nitrogen source and PO43- based salts as the phosphorus source. For all experiments, recombinant E. coli was inoculated from a glycerol stock at -80°C to sterile shake flasks. Cells in shake flasks were grown overnight in orbital shakers and subsequently inoculated into bioreactors to give a starting OD600 of o.1-0.2. Recombinant gene expression was induced by addition of 200 µM isopropyl b-D-1-thiogalactopyranoside (IPTG) for pJBG based plasmids or 0.002% (w/w) L-arabinose for pBAD based plasmids in early stages of exponential growth (OD600=0.2). Bioreactor cultivations were maintained at 37 °C, pH 7 and DOT values above 20% of saturation. Quantification and Analysis of data For papers I and II, quantification of glucose, 3HB and acetic acid used commercially available enzymatic kits (Boehringer Mannheim Cat No. 716251), (R-BIOPHARM acetic acid Kit Art. No.10148261035, Megazyme D-3-hydroxybutyric acid assay kit Cat No. K-HDBA). For papers III-V quantification of both substrates and metabolites was done by using high performance liquid chromatography (HPLC) (Alliance Waters 2695) equipped with an Aminex HPX-87H organic acid column (Bio-Rad, Hercules, CA) and quantified using either a RI detector for glucose or a UV detector (Waters) at 210 nm for organic acids.The following operating conditions were used to generate peak separation: 0.5 mL/min flow rate, 0.008 N H2SO4 mobile phase, column temperature 20°C. For all studies ammonium and phosphate concentrations were determined using commercially available enzymatic kits: Megazyme Ammonia Kit Cat No. K-AMIAR (megazyme) and abcam Phosphate Assay Kit (Colorimetric) Cat No. ab65622). 2.5 Steering 3HB production by bioprocess design (Paper

I)

In previous research, 3HB production by E. coli AF1000 pJBGT3RX was tested in batch mode with a defined minimal medium under nutrient excess conditions resulting in a 3HB yield on glucose of 0.057 g g-1 and an average specific productivity of 0.068 g g-1 h-1 [181]. Having an E. coli strain capable of producing


48

3HB, the next step was to investigate if production could be improved by bioprocess design strategies. The common strategy for PHB production by native organisms is based on repeated batch procedures in which after some cell growth a selected nutrient depletes from the broth while excess of carbon is maintained by addition of sugar to the cultivation [183, 184]. This results in high intracellular NADH concentrations, which consequently inhibits citrate synthase. This enzyme together with phosphotransacetylase (pta) are the main competitors of 3-keto-thiolase (t3) for the common substrate acetyl-CoA (Fig. 20) [185]. The citrate synthase inhibition together with the depletion of essential biosynthetic building blocks causes a redistribution of the carbon flux from cell growth and respiration to PHB storage, which together with nutrient depletion allows for an uncoupling between cell growth and product formation. Here we investigate whether the cultivation principle used for PHB production (depletion of a nutrient while maintaining glucose in excess) could also be employed for 3HB production. For this reason, as a first step, the effects of ammonium- or phosphate depleted batch cultivations on 3HB production were investigated (2.5.1). As an alternative cultivation strategy, fed-batch cultivations allow for control of cellular growth rate and cultivation rate parameters by continuous addition of limited amounts of a selected nutrient (that for 3HB production should not be the carbon source). In addition, nutrient limitation has the potential to avoid severe effects on cellular metabolism imposed by complete starvation of a nutrient and might thus result in positive effects on productivities. Furthermore, nutrient-limited fed-batches are common cultivation strategies used in academia and industry, to improve product formation [186, 187]. In this regard, effects of ammonium- or phosphate- limited fed-batch cultivations on 3HB production were studied as well (2.5.2). 2.5.1 Production of 3HB under ammonium or phosphate

depletion

Before nutrient limitation was investigated, batch experiments with nutrient low content were studied as a first step. Nutrient-depleted batch cultivations were divided in two phases: a first exponential-growth phase in which all nutrients including glucose were in excess, and a second depleted phase that started after complete consumption (depletion) of either ammonium or phosphate while glucose was maintained in excess (Fig. 21). Contrary to, stationary phases upon glucose depletion, both nutrient depleted cultivations showed linear increase in cell dry weight (CDW) in the starvation phase, with higher growth observed for the phosphate depleted experiment (Fig. 21 B). This seems to be related to the


49

differences in intracellular reserves. The intracellular reserves of phosphate in form of polyphosphates, rRNA, and phospholipids can be used by the cells when phosphate is scarce, while there is no readily available intracellular reserve of nitrogen for situations when nitrogen is scarce [188]. The nitrogen for this purpose is believed to come from degradation of intracellular nitrogen-rich compounds such as proteins.

E. coli AF1000 cultivated under ammonium-depleted conditions resulted in a 75% lower 3HB specific productivity (q3HB) and a 50% higher 3HB yield on glucose (Y3HB/Glc) compared to E. coli AF1000 grown in a batch ending with glucose depletion (Y3HB/Glc = 0.057 g g-1 and q3HB=0.068 g g-1 h-1) [181]. Production of acetic acid (HAc), the overflow metabolite commonly produced by E. coli, was also observed in both phases and followed a similar pattern as 3HB in the depleted phase. E. coli AF1000 cultivated under phosphate-depleted conditions resulted in a slightly lower 3HB specific productivity (q3HB) and 2.7 times higher 3HB yield on glucose (Y3HB/Glc) compared to E. coli AF1000 grown in a batch ending with glucose depletion [181]. Higher 3HB specific productivity (q3HB) and yield of 3HB with respect to glucose (Y3HB/Glc), were observed for the depleted-phase of the phosphate- compared to that of the ammonium-experiment (Table 2). Furthermore, production of the overflow metabolite acetic acid (HAc), was also lower for the phosphate-depletion experiment and uncoupled to 3HB production in the second phase (Fig. 21 B). Compared to growth in a batch ending with glucose depletion, nutrient-depleted batch cultivations result in higher yields of 3HB. Nevertheless, nutrient-depleted cultivations also have the disadvantage of lower specific productivities due to that the cells highly reduce growth.


50

Figure 21. 3HB produced in batch cultivations of AF1000 pJBGT3RX with depletion of either (A) ammonium or (B) phosphate.

2.5.2 Ammonium- and phosphate-limited fed-batch experiments for increased 3HB accumulation

To alleviate the negative effects on cellular metabolism (lower q3HB) imposed by complete starvation of a nutrient and to allow control of cultivation rate parameters, the next step was to perform fed-batch experiments in which glucose was maintained in excess while ammonium or phosphate were the limiting substrates that controlled growth. The cultivations were designed so that three sequential different phases were studied (Fig. 22): (1) a batch phase with reduced ammonium or phosphate content that would run out before any other component in the cultivation medium and set to reach an OD600 value of approximately 8, (2) an exponential feed phase described by equation 2.1 where the feed was set to give a specific growth rate (µ) of 0.35 h-1 until an OD600 value of 40, and (3) a linear feed phase described by equation (2.2) in which the feed was set to give a ]g

]Uof 0.1 L h-2.

(2.1) 𝐹 = 𝐹S ∙ 𝑒9∙i (2.2) 𝐹 = 𝐹S +]g]U∙ 𝑡 (2.3) 𝐹S =

9∙j∙^f∙k&4

𝐹S is the feed rate at phase start, defined by equation (2.3) where x is CDW at feed-phase start, V is volume of broth, S is the concentration of either ammonium or phosphate in the feed and Yxs is the yield of cells on ammonium or phosphate.

3HB

and

HAc

(g L

-1)

0

0.5

1.0

1.5

Glc

(g L

-1)

0

5

10

15

CDW

(g L

-1)

0

1

2

3

Time(h)0 2.5 5.0 7.5 10.0

Time(h)0 2.5 5.0 7.5 10.0

0

5

10

15

0

0.5

1.0

1.5

0

0.1

0.2

0.3

NH+4 depletionExp. growth

(A) N-depletionPO3-

4 depletion(B) P-depletion

µ (h-1

)NH

+ 4 and

10x

PO3- 4

(mM

)

CDWµ

GlcNH+

4

3HBHAc

CDWµ

GlcPO3-

4

3HBHAc

Exp. growth


51

Figure 22. Fed-batch cultivations of AF1000 pJBGT3RX with limitation of A) ammonium for production of 3HB, B) phosphate for induced 3HB production (Black) and non-induced 3HB production (Red). The dashed lines mark the shift between phases.

The exponential phase of the ammonium-limited (Fig. 22 A) and the phosphate-limited (Black in Fig. 22 B) cultivations resulted in constant growth rates of 0.26 h−1 and 0.25 h−1 respectively. Unexpected accumulation of ammonium and

(A) N-limitation

00.1

0.2

0.3

0.4

050100150200250

02.5

5.0

7.5

µ (h-1

)NH

+ 4(m

M)

Batch Exponential NH+4 feed Linear NH+

4 feed

CDWµ

GlcNH+

4

HAc3HB

q3HB

q3H

B(m

g g

-1 h

-1)

0

25

50

75

3HB

and

HAc

(g L

-1)

02.5

5.0

7.5

Glc

(g L

-1)

05

10152025

CDW

(g L

-1)

05

10

15

20

Time (h)0 2.5 5.0 7.5 10.0 12.5

100 10075

50

25

0

00.10.20.30.40.5

012345

02.5

5.0

7.510.0

µ (h-1

)PO

3- 4(m

M)

Linear PO3-4 feedExponential PO3-

4 feedBatchCDWµ

GlcPO3-

4

HAc3HB

q3HB

0

50

100

q3H

B(m

g g-1

h-1

)

0

50

100

3HB

and

HAc

(g L

-1)

02.5

5.0

7.510.0

Glc

(g L

-1)

05

10152025

CDW

(g L

-1)

05

10152025

Time(h)0 2.5 5.0 7.5 10.0 12.5

25

75 75

25

(B) P-limitation


52

phosphate in the respective experiments were observed at the end of the exponential feed phases, although growth rates remained low (Fig. 22). The 3HB specific productivities in the phosphate-limited cultivation were almost 2 times higher in both feed phases than those in the ammonium-limited experiment (Table 2), which resulted in higher final 3HB titers of 6.8 g L-1 compared to 4.1 g L-1 respectively. Nevertheless, in the exponential feed phase the 3HB yield on glucose for the ammonium-limited experiment was only slightly lower than the phosphate-limited experiment (Table 2).

Table 2. Comparison of 3HB production by AF1000 pJBGT3RX grown under different cultivation modes. The rates and yields in the table correspond to the depleted/limitation phases only, excluding prior exponential-growth phases.

q3HB

(g g-1 h-1) r3HB

(g L-1 h-1) 3HB titer

(g L-1) CDW (g L-1)

Y3HB/Glc

(g g-1) Ammonium

Batch Depletion 0.017 0.037 0.50 2.54 0.075

Fed-Batch Exponential

feed 0.035 0.120-0.510 4.10 20.40

0.046

Linear feed 0.025 0.460 0.056 Phosphate

Batch Depletion 0.060 0.170 1.25 3.27 0.150

Fed-Batch Exponential

feed 0.060 0.200-1.500 6.81 21.27 0.054

Linear feed 0.050 1.000 0.100

Volumetric productivities, r3HB, increased when switching from nutrient starvation to depletion conditions (Table 2). The main factor here is the CDW that increases over time over the feed phases compared to depletion phases. Furthermore, in the ammonium case, there are higher specific productivities in limitation compared to depletion. Although the phosphate-limited cultivation had higher 3HB production parameters then the ammonium case, a considerable amount of carbon was directed to the by-product acetic acid (Fig. 22). Acetic acid formation is undesirable since it competes with 3HB for the precursor AcCoA (decreasing the yield) and also because it has toxic effects for the cells. A replicate phosphate-limited fed-batch experiment without 3HB producing gene expression (Red in Fig. 22 B) showed similar trend in the first two phases of the experiment, but in the linear feed phase HAc was consumed even when glucose was in excess. These observations indicated that in the linear phase, HAc formation is not due to the phosphate limitation but instead due to the expressed 3HB pathway. This was not the case for phosphate-depletion conditions (Fig. 21 B), where the HAc concentration remained fairly constant. This might reflect the impact of phosphate-depletion on intracellular


53

phosphorus homeostasis and thereby on the activity of acetate producing enzymes phosphotransacetylase (pta) and acetate kinase (ack) (Fig. 20) Ammonium- and phosphate-limited fed-batch cultivations allow for higher 3HB titers and higher productivities compared to ammonium- and phosphate- depleted batch cultivations, though highest yields were observed under ammonium- or phosphate-depletion conditions (Table 2). Although improving cultivation modes and conditions is a key strategy for uncoupling growth and product formation for improvement of 3HB production by E. coli, nevertheless some issues remained to resolve for optimized 3HB production. 2.6 Steering 3HB production by engineering of the redox-

cofactor supply (Paper II)

2.6.1 Determining cofactor specificity of acetoacetyl-CoA reductase (rx)

Despite the increase in 3HB titers and productivities when uncoupling growth and product formation by using nutrient-limited fed-batch cultivations, there were still some issues remaining that bioprocess design strategies alone could not solve. These issues are related to thermodynamics and kinetic regulation of the 3HB pathway and could potentially be solved by strain design strategies. Starting from the central metabolite acetyl-CoA, 3HB production involves three catalytic steps (Fig. 20), two of which are catalyzed by recombinant enzymes. Although both the reaction catalyzed by the heterologous rx (acetoacetyl-CoA to 3HB-CoA) and the reaction catalyzed by the native thioesterase (3HB-CoA to 3HB) are thermodynamically favorable, the reaction catalyzed by the t3 (acetyl-CoA to acetoacetyl-CoA) is not, which means that in order to push the carbon flux towards 3HB formation the intracellular acetyl-CoA concentration has to be kept high and/or the acetoacetyl-CoA must be kept low. To assure that the acetoacetyl-CoA concentrations are maintained low, the stereospecific reduction of acetoacetyl-CoA to (R)-3HB-CoA catalyzed by a NAD(P)H dependent acetoacetyl-CoA reductase (Fig. 20) must be efficient. A critical issue in recombinant metabolite production is the supply and regeneration of redox-cofactors. For this reason, the cofactor availability of the reductase catalyzed step was considered to be crucial. Before applying genetic engineering strategies for Paper III, metabolic flux analysis was applied to help predict the effect of different metabolic pathways on 3HB production based on the redox cofactor specificity of acetoacetyl-CoA reductase.


54

The model suggested that to optimize 3HB production, carbon flux should be predominantly steered through the glycolysis pathway if the reductase was NADH-dependent, while if it was NADPH-dependent then carbon-flux should be steered towards the pentose phosphate pathway or Entner-Doudoroff pathway instead. Many acetoacetyl-CoA reductases involved in the production of PHB (composed of 3HB monomers in the R stereospecific configuration) are known to be NADPH-specific. A phylogenetic study that compared H. boliviensis’ rx to other reductase genes with known cofactor specificity was inconclusive. From an in vitro kinetic study in which rx was used to catalyze the reduction of acetoacetyl-CoA using either NADH or NADPH as cofactors, it was determined that rx accepts both redox cofactors although an 800% higher maximum reaction rate was observed when using NADPH, indicating a strong preference towards NADPH (Table 3).

Table 3. Kinetic parameters of the acetoacetyl-CoA reductase rx Cofactor Vmax KM kcat kcat/KM

(μmol mg-1 min-1) (μM) (s-1) (s-1 μM-1) NADPH 1.6 ± 0.2 70 ± 6 780 ± 100 11.1 ± 0.3

NADH 0.2 ± 0.02 24 ± 6 100 ± 7 4.2 ± 0.2

In E. coli the NADPH/NADP+ ratio is higher than the NADH/NAD+ ratio, which is beneficial since a high NADPH/NADP+ ratio provides an efficient thermodynamic push towards 3HB-CoA formation and or maintains a low acetoacetyl-CoA. In E. coli NADPH is mostly produced for growth by the first two steps of the pentose phosphate pathway and in the TCA cycle by the NADP+-dependent isocitrate dehydrogenase [189]. The amount of NADPH normally provided for growth is likely not sufficient to support a high q3HB through acetoacetyl-CoA reductase (rx). For this reason, it was hypothesized that to increase 3HB production the following strategies should focus on strain- or bioprocess-design strategies to improve the NADPH supply in 3HB-producing E. coli. 2.6.2 Strategies to increase NADPH supply

To increase NADPH supply for the reductase step in the E. coli strain expressing the 3HB pathway, two strategies were investigated. The first strategy was to reduce NADPH consumption in cell growth. Ammonium assimilation by E. coli converts α-ketoglutarate and ammonium into glutamate and glutamine. These two compounds then supply the nitrogen for all intracellular nitrogen-containing compounds [190]. Glutamate for example provides alpha-amino groups for all the amino-acids and half the nitrogen for pyrimidine, purine [190]. Glutamine for


55

example provides the nitrogen for amino sugars, NAD(P), and the remaining nitrogen for purines and pyrimidines[190]. The ammonia assimilatory pathway involves three enzymatic conversions[190]:

For each assimilated molecule of ammonia, the cell spends one equivalent of NADPH for producing glutamate from ammonia and α-ketoglutarate [190]. By supplementing glutamate into the cultivation medium, we speculated that the reaction catalyzed by glutamate dehydrogenase could be bypassed, decreasing NADPH consumption for cell growth. The second strategy was to increase the availability or ability to regenerate intracellular NADPH. The pentose phosphate pathway starts with the oxidation of glucose-6-phosphate to 6-phosphogluconate by a NADP+-dependent “glucose-6-phosphate dehydrogenase” (encoded by zwf) [189]. Then, the also NADP+-dependent enzyme 6-phosphogluconate further oxidizes 6-phosphogluconate to ribulose-5-phosphate and CO2 (with 3-keto-6-phosphogluconate as an intermediate) [189]. In E. coli, these 2 steps together with the NADP+-dependent isocitrate dehydrogenase (TCA cycle) provide the majority of NADPH needed for growth. Ribulose-5-phosphate gets transformed by a series of steps of the pentose phosphate pathway into glyceraldehyde-3-P and fructose-6-P and carbon flux is channeled back to the subsequent steps of glycolysis [189]. By overexpression of the zwf gene we presumed that carbon flux would predominately be steered to the pentose phosphate pathway and consequently, by the first 2 steps in this pathway the availability of NADPH could by increased. To test these hypotheses, two similar experimental setups were designed. The first was based on a batch ending with glucose depletion. The second was based on nutrient-depleted batch experiments that were divided in two different phases: (1) exponential growth and (2) nutrient-depleted phase (Fig. 23). For the first setup (a batch ending with glucose depletion), final CDWs of 3.7 and 3.8 g L-1 were obtained respectively for glutamate addition and overexpression of zwf. These values are slightly lower than the CDW obtained when cultivating AF1000 expressing t3 and rx under similar conditions (4.18 g L-1). Furthermore, glutamate supplementation and overexpression of zwf resulted in respectively 70 and 10% higher 3HB specific productivities, 14 and 28% higher 3HB yields on

NH3 + glutamate + ATPglutaminesynthetase

Mg 2+ glutamine + ADP + Pi

glutamine + α-ketoglutarate + NADPH 2 Glutamate + NADP+glutamatesynthase

NH3 + α-ketoglutarate + NADPH glutamate + NADP+ (1)

(3)

(2)

glutamatedehydrogenase


56

substrate, and 33 and 43% higher titers compared to a reference experiment (Table 4) in which E. coli expressing recombinant t3 and rx was grown in minimal medium [181]. The experiment which had access to glutamate had the highest specific 3HB production rate (0.116 g g-1 h-1), though the highest 3HB yield on glucose (0.073 g g-1) and 3HB titer (0.73 g L-1), were obtained by the strain expressing zwf. Both strategies (glucose supplementation and overexpression of zwf) resulted in higher 3HB production parameters. These results indicated that both strategies for improving intracellular NADPH were promising. To investigate if the engineering strategies for improved NADPH supply were compatible with the previously shown benefit of nutrient-depletion/limitation, as a first step we designed nutrient-depleted batch cultivations for both strategies: glutamate supplementation and overexpression of zwf. Phosphate-depleted batch cultivations had previously (Paper I) resulted in the highest improvements in yield and in titer without heavily compromising productivity. Nevertheless, a replicate of the phosphate-depleted batch experiment (Fig. 21 B) in which E. coli in addition of recombinantly expressing t3 and rx also overexpressed zwf, showed no improvement in yield, titer or productivity when compared to the strain in which zwf was not overexpressed (data not shown). For this reason, phosphate-depleted/limited cultivations were no longer pursued in this thesis. Since ammonium-depleted batch cultivations had also previously resulted in higher 3HB yields compared to batch cultivations in exponential growth phase (batch cultivations ending with glucose depletion) (Table 2), they were used as experimental setups for both strategies to improve NADPH supply (glutamate supply and overexpressing zwf) (Fig. 23). Ammonium-depleted batch cultivations with either glutamate supply or overexpression of zwf resulted in respectively 2 and 3 times higher specific productivities. In addition, the overexpression of zwf resulted in 73% higher 3HB yield on glucose and 2 times higher 3HB titer compared to the reference experiment without overexpressing zwf (Fig. 21 A) cultivated in an ammonium-depleted batch (Table 4). Furthermore, the overexpression of zwf allowed for a high productivity of 0.057 g g-1 h-1 in the ammonium-depleted phase, which is comparable to previous productivities obtained for exponential growth in minimal medium batch cultivations without zwf overexpression (0.068 g g-1 h-1) [181]. Although glutamate supply and overexpression of zwf led to higher production parameters in the ammonium-depleted phase, they also resulted in almost 28% higher acetic acid production (Fig.23) when compared to the reference cultivation (Fig. 21 A).


57

Table 4. 3HB production for different cultivation strategies for AF1000 pJBGT3RX (and pBADzwf). For ammonium-depleted batch experiments, the rates and yields in the table correspond to the depleted phases only, excluding prior batch phases.1= exponential growth in batch cultivations ending with glucose. 2 = ammonium-depleted batch cultivations. * Control experiment performed in [181].**Control experiment performed for Paper I of this thesis. The rates and yields in the table correspond to the depleted phases only.

Setup Genes expressed

Nitrogen source

Time q3HB 3HB CDW Y3HB/Glc (h) (g g-1 h-1) (g L-1) (g L-1) (g g-1)

1 t3, rx * NH4+ 5.5 0.068 0.51 4.18 0.057 1 t3, rx NH4++ glut. 6.5 0.116 0.68 3.84 0.065 1 t3, rx, zwf NH4+ 7.0 0.075 0.73 3.69 0.073 2 t3, rx ** NH4+ 11.8 0.017 0.50 2.54 0.075 2 t3, rx NH4++ glut. 11.9 0.035 0.55 2.39 0.075 2 t3, rx, zwf NH4+ 11.8 0.057 1.06 2.00 0.130

Figure 23. Ammonium-depleted batch cultivations of AF1000 producing 3HB. (A) Strategy 1: pJBGT3RX with excess of glutamate in the medium. (B) Strategy 2: pJBGT3RX, pBADzwf

2.6.3 3HB production in fed-batch cultivations

In this thesis we aimed at obtaining increased 3HB production parameters (yield, productivities and titer) with a special focus on the robustness and scalability of the process. Although yields and specific productivities can be improved in batch cultivations, fed-batch cultivations with higher cell density are necessary to further increase titers and volumetric productivities for a scalable process.

3HB,

HAc

(g

L-1

)

0.5

1.0

1.5

Glc

, Glu

(g L

-1)

5

10

15

CD

W(g

L-1

)

1

2

3

Time(h)

0 2.5 5.0 7.5 10.0Time(h)

0 2.5 5.0 7.5 10.0

5

10

15

NH

+ 4(m

M)

CDW

GlcGluNH+

4

3HBHAc

(A) Glutamate supplementation (B) zwf overexpression Exp. phase NH+

4 depletion +4 depletion Exp. phase NH


58

Since the strategy of increasing NADPH availability by overexpression of zwf resulted in higher production parameters in ammonium-depleted batch cultivations than in phosphate-depleted batch cultivations, an ammonium-limited fed-batch with delayed induction of the 3HB pathway was designed to obtain higher volumetric productivities and titers of 3HB (Fig. 24). The experiment was designed so that two sequential phases were distinguished (1) A repeated batch phase in which high accumulation of cell mass was obtained (CDW=20 g L-1) by repeatedly adding glucose and nutrient salts to the broth while maintaining high enough glucose concentrations so that the ammonium content would run out before any other component in the cultivation medium. (2) An ammonium-limited feed phase in which glucose was maintained above 10 g L-1 by feeding it to the medium or adding it manually.

Figure 24. Ammonium-limited fed-batch cultivation of AF1000 pJBGT3RX pBADzwf

Although glucose was maintained in excess and no oxygen limitation was observed during the cultivation, growth decreased after the first 10 h of cultivation. This was attributed to the toxic effects of the high acetic acid accumulation which had a value of 3 g L-1 at 10 h of the cultivation and reached 26 g L -1 at the end of the cultivation. A high 3HB titer of 12.7 g L-1 and high total volumetric productivity of 0.42 g L-1 h-

1 (of the whole experiment) were obtained after 30 h of cultivation. The 3HB yield on glucose in ammonium-limited conditions decreased compared the that of the ammonium-depleted conditions (0.090 and 0.130 g g-1h-1 respectively), which was attributed to fact that contrary to depleted-conditions in ammonium-limited conditions carbon-flux is also directed to growth. Although ammonium-limited conditions had previously resulted in higher 3HB specific productivities compared to ammonium-depleted conditions (Paper I, Table 2 in this thesis), in this experiment the 3HB specific productivity in the limited-phase was severely

CDW 3HB r3HB

0.1

0.2

0.3

0.4

0.5

0.6

r 3

HB

(g L

-1 h

-1)

Repeated batch Constant feed

CD

W, 3

HB

(g

L-1

)

5

10

15

20

25

30

Time (h)0 5 10 15 20 25 30


59

hampered (0.009 g g-1 h-1) dropping to ¼ of the amount obtained by the same construct in the ammonium-depleted phase (Table 4), which was attributed to the inhibition of high acetic acid concentration. Overexpression of zwf in the E. coli strain expressing the 3HB pathway (t3 and rx) resulted in higher yield, titers and specific productivities for both exponential-growth and ammonium-depleted conditions in batch cultivations (Table 4), though when switching to high cell density cultivations by ammonium-limited fed-batch cultivations, especially high undesired formation of the inhibitory byproduct acetic acid was present. In this sense and as a next logical target the subsequent parts of this thesis (Papers III and IV) simultaneously focused on strain design strategies to increase 3HB yield, rate and titer by reduction or elimination of the production of the inhibitory and competing byproduct “acetic acid”. 2.7 Steering 3HB production by decreasing production of

the inhibitory byproduct acetic acid (Paper III) The inhibitory effects of high acetate concentrations prevented higher 3HB specific productivities and further reduced growth. In addition, since both acetate and 3HB are derived from acetyl-CoA, acetate formation pulls carbon and electrons away from 3HB. There are two main pathways for acetic acid formation in E. coli, the first one starts from acetyl-CoA and uses two-step pathway catalyzed by phosphotransacetylase (pta)-acetate kinase(ackA) and the second one catalyzes the transformation of pyruvate to acetate directly by pyruvate oxidase (poxB) (Fig. 20). In this regard, it was hypothesized that 3HB production could be improved for scalable high cell density fed-batch cultivations by reducing acetic acid formation with metabolic engineering or strain design. 2.7.1 Eliminating genes responsible for acetic acid formation

Metabolic engineering is a tool that has extensively been used to decrease by-product formation “See section 1.5.1 Steering production by strain choice and -design” [73]. Chromosomal deletion of the genes encoding the first enzymes in the pathways towards acetate formation pta and poxB (Fig. 20) has shown to increase PHB production by recombinant E. coli [191, 192]. It was speculated that the same effect could be obtained in 3HB production. For this reason, 3HB production by engineered AF1000 mutants (Δpta, ΔpoxB, ΔptaΔpoxB) harboring plasmids pJBGT3RX and pBADzwf was studied. Low acetate production by E. coli strains


60

BL21 and B has been attributed to their low isocitrate lyase regulator (iclR) expression (Fig. 20) [193-195]. For this reason, 3HB production by an iclR mutant of AF1000 harboring plasmids pJBGT3RX and pBADzwf was also studied. In previous research, (Papers I and II of this thesis) it was shown that compared to exponential growth in batch cultivations, ammonium-depleted batch cultivations allowed for increased 3HB yields in the depletion phase, which made this experimental setup a good screening method. In this regard, four different knockout (Δpta, ΔpoxB, ΔptaΔpoxB, ΔiclR) strains derived from strain AF1000 together with the control strain (AF1000,Ref) were screened for 3HB production when harboring plasmids pJBGT3RX and pBADzwf in ammonium-depleted batch experiments (Table 5). To increase the resolution of the experiment, the experiments were performed for a total of 24 h (with approximately 7h of exponential phase and approximately 17 h of ammonium depleted phase) Table 5. Calculated parameters of the screening of different knockout variant strains derived from E. coli AF1000 (reference strain). All strains were harboring pJBGT3RX and pBADzwf and investigated in ammonium-depleted batch cultivations.

Mutation μ Y3HB/Glc YHAc/Glc YPyr/Glc

(g g-1) (g g-1) (g g-1)

Exp. Ref. 0.54 ± 0.01 0.077 ±0.010 0.104 ±0.009 0.003 ±0.000

Exp. Δpta 0.46 ± 0.01 0.078 ±0.006 0.016 ±0.003 0.073 ±0.011

Exp. ΔpoxB 0.47 ± 0.02 0.071 ±0.012 0.107 ±0.017 0.011 ±0.003

Exp. ΔptaΔpoxB 0.40 ±0.02 0.101 ±0.016 0.020 ±0.007 0.069 ±0.001

Exp. ΔiclR 0.52 ±0.02 0.090 ±0.010 0.126 ±0.020 0.004 ±0.001

NH4+ dep Ref. - 0.149 ±0.010 0.130 ±0.010 n.d.

NH4+ dep Δpta - 0.206 ±0.010 0.220 ±0.001 0.380 ±0.074

NH4+ dep ΔpoxB - 0.208 ±0.012 0.194 ±0.010 0.391 ±0.059

NH4+ dep ΔptaΔpoxB - 0.120 ±0.016 0.103 ±0.003 0.290 ±0.033

NH4+ dep ΔiclR - 0.142 ±0.015 0.135 ±0.005 n.d.

n.d. no product detected. Mean deviation was calculated for duplicate experiments for all strains.

During exponential growth, the knockout variant ΔiclR only slightly decreased its specific growth rate compared to the reference strain while the variants Δpta, ΔpoxB, ΔptaΔpoxB respectively reduced their growth rate by 15, 14 and 25%. The variants Δpta and ΔpoxB had a similar yield of 3HB on glucose then the reference strain (~0.075 g g-1) while the ΔiclR and ΔptaΔpoxB variants increased the yield of 3HB on glucose by 17 and 31% compared to the reference strain. The two variants


61

with pta deletion further decreased the yield of acetic acid on glucose by approximately 80%. However, these variants started to accumulate pyruvate (pyr) instead. Neither iclR deletion nor the poxB deletion alone resulted in reduced acetic acid on glucose yield compared to the reference strain (Table 5). During the ammonium-depleted phase the deletion strains with pta and poxB deletions increased their 3HB on glucose yields with almost 40% compared to the reference strain. Nevertheless, both strains (Δpta and ΔpoxB) almost doubled their acetic acid to glucose yield (compared to the control strain). The double knock-out strain (ΔptaΔpoxB) decreased the acetic acid to glucose yield by 20% compared to the control strain in the ammonium-depleted phase while the variant with the iclR deletion did not seem to have any effect on any of the calculated yields in this phase. Additionally, during the ammonium-depleted phase the pyruvate production on variants Δpta, ΔpoxB and ΔptaΔpoxB actually exceeded their respective acetate production. Compared to acetate, pyruvate has both a lower pKa and a lower oleyl-water partitioning coefficient. In this line of thought, we presumed that compared to the reference strain, a less toxic effect would be obtained by these knockout strains harboring plasmids pJBGT3RX and pBADzwf in high-cell density fed-batches. For this reason, the reference AF1000 strain, and both variants (AF1000 Δpta and AF1000 ΔptaΔpoxB) were evaluated in high cell density ammonium-limited fed-batch cultivations consisting of two different phases: (1) A repeated batch phase in which high accumulation of cell mass was obtained (~CDW=20 g L-1) by repeatedly adding glucose and nutrient salts to the broth while maintaining high enough glucose concentrations so that the ammonium content would run out before any other component in the cultivation medium. (2) An ammonium-limited feed phase in which glucose was maintained above 10 g L-1 by feeding it to the medium or adding it manually (Fig. 25). During the repeated batch-phase, acetic acid formation reduced for both knockout variants (Δpta and ΔptaΔpoxB) variants with the knockouts. At the end of this phase acetate concentration was decreased by 73% for the Δpta variant (2.73 g L-1) and by 78% for the ΔptaΔpoxB variant (2.27 g L-1) compared to the reference strain (10.14 g L -1). The slower increase of acetate formation allowed for higher specific growth rates and shorter duration of this phase (Fig. 25 B-C) compared to the reference strain (Fig. 25 A). However, the specific growth rate of both knockout variants Δpta and ΔptaΔpoxB reduced towards the end of the repeated batch phase


62

even though acetate concentration was not considerably high (~2 g L-1). The reduction in growth rate was attributed to pyruvate accumulation which resulted in 2.84 g L-1 for the Δpta variant and 3.37 g L- for the ΔptaΔpoxB variant. During the ammonium-limited feed phase, the CDW of all strains increased only slightly (Fig. 25), which was attributed to the inhibitory effects of acetate. Upon completion of the experiments, variants Δpta and ΔptaΔpoxB produced 37% and 62% less acetic acid compared to the reference strain. Nevertheless, for both knockout variants, carbon flux was not redirected to 3HB formation and instead the pyruvate concentration highly increased in both phases of the experiment. Although there was a decreased acetic acid formation for Δpta and ΔptaΔpoxB variants, there was no increase in 3HB production in high-cell-density ammonium-limited fed batch cultivations. The apparently limited capacity of the 3HB production pathway together with a decreased acetate formation might have resulted in increased acetyl-CoA concentrations. High acetyl-CoA concentrations can decrease the activity of the pyruvate dehydrogenase complex due to an allosteric inhibition of the transacetylase component [191, 196], which may result in accumulation and further excretion of pyruvate. Pyruvate formation was especially high under ammonium-depleted or ammonium-limited conditions in which the demand for pyruvate and ATP is lower resulting in decreased flux through the TCA cycle and probably also decreases the demand for acetyl-CoA. Increased pyruvate levels result in inhibition of the phosphotransferase system [197], which explained the decreased growth rates observed during the repeated batch-phases of both Δpta and ΔptaΔpoxB variants


63

Figure 25. Ammonium-limited fed-batch cultivations of knockout variants of E. coli AF1000 harboring plasmids pJBGT3RX pBADzwf. (A) AF1000. (B) AF1000 Δpta. (C) AF1000 ΔptaΔpoxB. The dashed lines mark the shift between phases.

2.7.2 Investigation of 3HB production in different E. coli

strains In view of the unsuccessful by eliminating enzymes responsible for acetic acid formation, the next approach focused instead on investigating if other E. coli strains would be more suitable for 3HB production while producing less acetic acid. E. coli commercial safe strains belong to groups B,K-12, C and W [198]. These strains have been widely used as laboratory strains and studied for industrial production of chemicals, fuels and proteins [198]. AF1000 derived from strain MC41000 [199] which derives from strain K-12 [173, 198]. K12 is the most studied E. coli strain in basic biology, medicine and biotechnology [200, 201]. For this reason, three additional K-12 strains were selected for the study: MG1655 [202], W3110 [202] and BW25113 [203]. E. coli strains belonging to group B including

r 3H

B, r H

Ac (

g L-1

h-1)

0

0.5

1.0

1.5

3HB,

HAc

, Pyr

(g L-1

)

0

4

8

12

16(N

H4)

2SO

4, C

DW

(g

L-1)

010

20

30

40

50

Time (h)8 12 16 20 24

Time (h)8 12 16 20 24

Time (h)8 12 16 20 24

(A): AF1000-T3Rxzwf Rep Batch Fed-batch

(B): AF1000 pta -T3Rxzwf Rep Batch Fed-batch

(C): AF1000 pta poxB-T3Rxzwf Rep Batch Fed-batch

GlucoseCDW(NH4)2SO4

[3HB][HAc]µ[Pyr]

r3HBrHAcq3HBqHAc

040

80

120

160

200

0

0.2

0.4

0.6

0.8

0

0.1

0.2

0.3 q3H

B , qH

Ac (g g-1 h

-1)G

lucose consumed (g L

-1) µ (h

-1)


64

BL21 and B are well known to grow with reduced acetic acid formation compared to K12 strains [195]. This has been attributed to that these strains possess: (i) low expression of the isocitrate lyase regulator (iclR) which regulates carbon-flux towards the glyoxylate shunt (which is highly repressed on growth on glucose as the sole carbon), and (ii) low expression of pyruvate oxidase [193-195]. For this reason, both strains B and BL21 were also selected. W strain was also selected since this is a newly sequenced strain which has a high specific growth rate on glucose and produces low amounts of acetate [204]. Strains from group C are generally used for expression of toxic genes [198] for which they were not considered in this study. The six promising E. coli strains (without harboring plasmids pJBGT3RX nor pBADzwf) were then screened for growth and acetate production in minimal medium batch cultivations for a total of 6 h approximately, using the background strain AF1000 as control (Table 6).The strains AF1000, W3110 and MG1655 had the highest yields of acetate on glucose and highest specific acetate productivities. Strains BW25513 and B had both approximately 64% lower acetate on glucose yield and 73% lower acetate specific productivity compared to the reference strain AF1000. While strain W had a smaller acetate on glucose yield reduction (56%) it had 75% less specific acetate productivity (compared to strain AF1000), in addition, it had the highest growth rate (1.04 -1) of all the screened strains. Out of all the screened strains, BL21 had the lowest acetate yields and acetate specific productivities, corresponding to 94% and 96% (respectively) lower parameters compared to the reference strain. Table 6. Calculated parameters of the screening of different E. coli strains for growth and acetate production, investigated under minimal medium batch cultivations for a total of 6 h.

Strain µ (h-1)

YHAc/Glc (g g-1)

qHAc (g g-1h-1)

AF1000 0.84 ± 0.03 0.112 ± 0.003 0.187 ± 0.002

W3110 0.67 ± 0.01 0.082 ± 0.002 0.112 ± 0.002

MG1655 0.65 ± 0.02 0.068 ± 0.002 0.073 ± 0.002

W 1.00 ± 0.04 0.049 ± 0.003 0.046 ± 0.003

BW25113 0.80 ± 0.02 0.040 ± 0.002 0.050 ± 0.002

B 0.89 ± 0.01 0.038 ± 0.002 0.050 ± 0.003

BL21 0.71 ± 0.08 0.006 ± 0.001 0.007 ± 0.002

Mean deviation was calculated for duplicate experiments for all strains.


65

Based on this first screening (Table 6), strains BL21 (lowest acetate parameters), BW25513 (K-12 strain with low acetate production), and W (highest growth rate and low acetate) were selected to be evaluated for 3HB production when harboring the heterologous genes t3 and rx. In this regard, the three different strains and the control background strain (AF1000) were evaluated in ammonium-depleted batch experiments (Table 7), which allowed for two different phases: exponential growth phase and ammonium-depleted phase. To allow for a better screen on the depleted phase, after the exponential growth phase ended (~4. 5-7 h) the experiments were left in the depleted-phase for a total of 24 h of experiment (Table 7). Table 7. Calculated 3HB production parameters of different E. coli strains. All strains were harboring pJBGT3RX and investigated under ammonium-depleted batch cultivations. The parameters are calculated for the whole experiment.

Strain μ (h-1)

Y3HB/Glc (g g-1)

YHAc/Glc (g g-1)

Y3HB/HAc (g g-1)

AF1000 0.62 ± 0.04 0.149 ± 0.003 0.139 ± 0.009 1.1 ±0.1

W 0.86 ± 0.08 0.103 ± 0.017 0.034 ± 0.003 3.1 ± 0.9

BW25113 0.52 ± 0.01 0.138 ± 0.006 0.075 ± 0.002 1.8 ± 0.0

BL21 0.65 ± 0.01 0.093 ± 0.002 0.017 ± 0.002 5.6 ± 0.6

Mean deviation was calculated for duplicate experiments for all strains.

Control strain AF1000 resulted in the highest 3HB on glucose yields and acetate on glucose yields (Table 7). Strains W and BW25113 had 31% and 7% respectively lower acetate on glucose yields compared to the reference strain AF1000 and an intermediate value 3HB yield on glucose of all the strains. Strain BL21 had the lowest acetate on glucose yield of 0.017 g g-1. Furthermore, it had the highest 3HB on acetate yield of all the evaluated strains. This later parameter must be as high as possible to avoid acetate accumulation and inhibition in high-cell density ammonium-limited fed-batch cultivations which aim to attain high 3HB titers. Because of its higher yield of 3HB on acetate, BL21 was selected to be further cultivated when harboring plasmids pJBGT3RX and pBADzwf to obtain high 3HB titers under high cell density ammonium-limited fed-batch cultivations (Fig. 26). The cultivations were designed so that two sequential phases were distinguished (1) A repeated batch phase in which high accumulation of cell mass was obtained (~CDW=20 g L-1) by repeatedly adding glucose and salts to the broth while maintaining high enough glucose concentrations so that the ammonium content would run out before any other component in the cultivation medium. (2) An


66

ammonium-limited feed phase in which glucose was maintained above 10 g L-1 by feeding it to the medium or adding it manually (Fig. 26).

Figure 26. Ammonium-limited fed-batch cultivations of E. coli BL21 harboring plasmids pJBGT3RX pBADzwf. The dashed lines mark the shift between phases.

When the repeated batch phase ended, the acetate concentration was 0.86 g L-1, corresponding to twelve times less acetate produced then when growing AF1000 under the same conditions (Fig. 25A). Furthermore, compared to the knockout variants Δpta and ΔptaΔpoxB cultivated under the same conditions (Fig. 25 B-C) no pyruvate was accumulated. Compared to AF1000 variants cultivated under the same conditions (Fig. 25 A-C), which dropped their growth rate to practically zero throughout the repeated-batch phase, BL21 was able to maintain a constant growth rate of 0.43 h-1 in this phase. Consequently, the BL21 cultivation had a much shorter repeated-batch phase (11.5 h) compared to that of AF1000 (19.5 h). This was a direct consequence of the decreased metabolic-inhibitory effects of acetate or avoidance of the inhibition of the phosphotransferase system by high pyruvate accumulation. Furthermore, compared with the AF1000 cultivation, the lower

r 3H

B, r H

Ac (g

L-1

h-1)

0

0.5

1.0

1.5

3HB,

HAc

(g L

-1)

0

4

8

12

16

(NH

4)2S

O4,

CD

W (

g L-1

)0

10

20

30

40

50

Time (h)4 8 12 16 20 24

0

40

80

120

160

200

0

0.2

0.4

0.6

0.8

0

0.1

0.2

0.3

Glucose consum

ed (g L-1)

µ (h-1)

q3H

B , qH

Ac (g g-1 h

-1)

GlucoseCDW(NH4)2SO4

[3HB][HAc]µ

r3HBrHAcq3HBqHAc

BL21-T3Rxzwf Rep Batch Fedbatch


67

acetate accumulation for the BL21 cultivation allowed for a 2.3 times higher final cell dry weight (47.14 g L-1 vs. 20.1 g L-1), a 2.3 times higher final 3HB titer (16.31 g L-1 vs. 7.04 g L-1) and a 3 times higher volumetric productivity in the ammonium-limited phase (1.27 g L-1h-1 vs. 0.42 g L-1 h-1). These results demonstrate the importance of careful selection of strain background for increased 3HB production parameters with focus on the robustness and scalability of the process. 2.8 Steering 3HB production by interventions in the

pathway to reduce competing pathways (Paper IV) As an alternative strategy to increase 3HB yield, rate and titer by reduction or elimination of the production of acetic acid, this current section of the thesis (2.8, Paper IV) was done in parallel to the previous section (2.7, Paper III). Although 3HB yield, titer and specific productivity had been improved by both ammonium-depleted cultivation strategies and engineering of the redox cofactor supply a remaining problem was the high acetic acid production in high cell density cultivations. The present section focused on regulating carbon-flux towards 3HB formation (hence reducing carbon-flux towards HAc) by interventions on the 3HB pathway in the original background strain E. coli AF1000. Starting from the central metabolite acetyl-CoA, 3HB production by recombinant E. coli involves three catalytic steps (Fig. 20) reviewed on section “2.3 Production of 3HB by E. coli expressing recombinant genes from Halomonas boliviensis” of this thesis. The last step of the 3-step production involves the highly thermodynamically favorable hydrolysis of (R)-3HB-CoA to (R)-3HB and CoA catalyzed by a native E. coli thioesterase (Fig. 20). Thioesterases are a group of enzymes that catalyze the hydrolysis of the thioester bond between a carbonyl group and a sulfur atom [205]. Acyl-coA thioesterases specifically catalyze the hydrolysis of acyl-CoA intermediates into the respective carboxylic acid and CoA [206, 207]. E. coli harbors a series of several native thioesterases, of which TesA, TesB, FadM, YciA, YdiI and YbgC have been reported to catalyze the hydrolysis of acyl-CoA intermediates. These enzymes are of course central to the metabolic engineering of a carboxylic acid production strain. In the previous sections of this thesis (2.5-2.7), no heterologous gene encoding a thioesterase was introduced in the E. coli, assuming sufficient activity of unspecified native thioesterases. However, with the achieved improvements in 3HB production, it cannot be excluded that thioesterase activity became limiting. Therefore, it was investigated if engineering thioesterase activity can provide the thermodynamic


68

pull needed to redirect acetyl-CoA from acetate to 3HB, especially in the context of previously achieved improvements through implementation of ammonium-depletion/limitation and/or increased NADPH supply (zwf). 2.8.1 Specific contribution of native thioesterases to 3HB

production Before the effect of overexpression of native thioesterases on 3HB production by E. coli could be investigated, we first aimed at identifying the native thioesterase(s) responsible for the hydrolysis of 3HB-CoA to 3HB and CoA. As a first step, four out of the six native thioesterases known to hydrolyze acyl-CoA intermediates, were selected as possible candidates because of their broad substrate specificity towards carboxyl-CoA intermediates of short and medium carbon length. First, thioesterase TesA, this enzyme is located in the periplasmic space of E. coli and had been reported to increase aliphatic carboxylic acid production [208, 209]. Second, thioesterase TesB, this enzyme had been reported to hydrolyze β-hydroxyacyl-CoA thioesters [210, 211] and to play an important role in 3-hydroxydecanoyl-CoA hydrolysis [212]. In addition, overexpression of tesB in E. coli enhanced production of both enantiomers of 3HB [136-138, 213]. Third, thioesterase FadM, this enzyme was reported to be involved in the ß-oxidation of oleic acid by hydrolyzing the minor side product 3,5-tetradecadienoyl-CoA [214] [215]. Fourth, thioesterase YciA, this enzyme had been reported to have an effect on 4 carbon carboxylic acid production [182, 216]. Furthermore, it was shown to have catalytic efficiency towards acyl-CoA intermediates, such as, acetyl-CoA, acetoacetyl-CoA and 3HB-CoA [182]. The other two thioesterases were not considered as candidates in this study because of their different reported substrate specificity. Thioesterase YdiI had been reported to hydrolyze aromatic-CoA intermediates [217] and YbgC had been reported to be more specific towards ACP-CoA intermediates [218]. The next step was to investigate the individual contribution of the four selected thioesterases to 3HB-CoA hydrolysis. For this reason, the four genes encoding individual thioesterases (tesA, tesB, yciA and fadM) were individually deleted from the AF1000 background strain. The resulting knockout variants (with the reference strain, AF1000) harboring the plasmid pJBGT3RX were subsequently screened for 3HB production in ammonium-depleted batch cultivations (Fig. 27). Batches for starvation studies were divided in two phases: a first exponential-phase in which all nutrients including glucose were in excess (~5h), and a second ammonium-


69

depleted phase. The total duration of the experiment was of approximately 9.5 h (Fig. 27).

Figure 27. Assessment of the quantitative impact on 3HB production by E. coli AF1000 harboring plasmid pJBGT3RX of deletion of four thioesterase-encoding genes. Parameters calculated over (A) Exponential phase. (B) Ammonium-depleted phase. Mean deviations were calculated for all strains from duplicate experiments.

During the exponential phase all four knockout variants showed a similar growth rate of ~0.67 h-1 compared to the control strain (AF100o). In this phase, the control strain had a yield of 3HB on glucose of 0.077 g g-1 at a specific production rate of 0.138 g g-1h-1. In the ammonium-depleted phase, the control strain had a yield of 3HB on glucose of 0.150 g g-1 accompanied by a 3HB specific production rate of 0.044 g g-1 h-1. The variant ΔtesA resulted in no significant difference (compared to the control) in the calculated 3HB production parameters in either of the phases, which was attributed to the enzyme’s periplasmic localization [219]. In the exponential phase, the variants ΔtesB and ΔfadM both showed approximately only 3% lower 3HB yield on glucose and 10% lower specific productivity compared to the reference strain. In the depletion phase ΔtesB showed a 15% lower 3HB yield

q3HB Y3HB/Glc

q3HB [g g -1 h -1]

0

0.01

0.02

0.03

0.04

0.05

Y3H

B/Glc [g3H

B g -1Glc ]

0

0.05

0.10

0.15

0.20

control tesA tesB yciA fadM

(A)

(B)

q3HB Y3HB/Glc

Gro

wth

rate

[h-1

]

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

q3HB [g g -1 h -1]

0

0.02

0.04

0.06

0.08

0.10

0.12

0.14

0.16

Y3H

B/Glc [g3H

B g -1Glc ]

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.10

control tesA tesB yciA fadM


70

on glucose and a 10% lower 3HB specific productivity (compared to the reference), while ΔfadM showed no difference in 3HB yield on glucose and only a 5% lower 3HB specific productivity compared to the reference strain. The variant ΔyciA resulted in the biggest impact on 3HB production in both phases. In the exponential phase, the ΔyciA variant resulted in 16% lower specific productivity and 13% lower 3HB glucose on yield, compared to the reference strain. The biggest difference was presented in the ammonium-depleted phase with 36% and 43% lower 3HB yield on glucose and 3HB specific productivity respectively. None of the four individual knockouts abolished 3HB-CoA hydrolysis completely, which meant that the remaining thioesterases (possibly including YdiI and YbgC) also contribute to 3HB-CoA hydrolysis. Deletion of yciA showed the biggest decrease in 3HB production, indicating that enzyme YciA was the biggest contributor to 3HB-CoA hydrolysis. For this reason, YciA was selected for further investigation on the next steps of this section (2.8). 2.8.2 Overexpression of yciA for improved 3HB-CoA

hydrolysis Once the largest contributing thioesterase was identified (YciA), the next step was to investigate whether overexpression of yciA would improve 3HB production in the background E. coli strain AF1000 expressing genes t3 and rx (Fig. 20). For this reason, AF1000 harboring plasmids pBAD-(Km)-yciA and pJBGT3RX was screened together with the control construct AF1000 harboring the empty pBAD-(Km)-Blank plasmid and pJBGT3RX in ammonium-depleted batch cultivations identical to those of the previous section (2.8.1) (Fig. 28 A-B). The variant recombinantly expressing t3-rx-yciA had a slightly lower growth rate of 0.52 h-1 compared to the variant recombinantly expressing t3-rx (0.6 h-1). Overexpression of yciA (together with t3-rx) in AF1000 doubled the specific productivity and 3HB yield on glucose in both phases compared to the control variant (t3-rx) (Table 8). Coinciding with the higher 3HB production parameters, acetic acid production was lower when overexpressing yciA compared to the control strain (Fig. 28 A-B). Furthermore, in the ammonium-depleted phase, the t3-rx-yciA variant consumed some of the acetic acid previously formed (Fig. 28B).


71

Figure 28. Ammonium-depleted batch cultivations of E. coli AF1000 engineered for production of 3HB (A) pJBGT3RX, pBAD-(Km)-Blank (B) pJBGT3RX, pBAD-(Km)-yciA (C) pJBGT3RX, pBAD-(Km)-zwf-yciA. The dashed lines mark the shift between phases.

In Paper II it was shown that an increase in the NADPH supply by overexpression of the glucose-6-phosphate dehydrogenase (zwf) improved 3HB production. To study if the positive effect of yciA overexpression on 3HB production would be accumulative to the effect of zwf overexpression, a strain harboring plasmids pBAD-(Km)-zwf-yciA and pJBGT3RX was cultivated under the same ammonium-depleted batch conditions (Fig. 28 C). This variant (t3-rx-zwf-yciA) had a further decreased growth rate of 0.48 h-1. Expression of pBAD-(Km)-zwf-yciA further increased the specific productivity by approximately 33% in both phases compared to the variant t3-rx-yciA (Table 8). Although the 3HB yield on glucose increased by 42% in the exponential phase no increase was shown in the ammonium-depleted phase (Table 8). Contributing to high 3HB production, acetate was almost completely consumed in the depletion phase.

3HB,

HAc

(g

L-1

)

0

0.5

1.0

1.5

2.0

Glc

NH

+ 4(g

L-1

) (m

M)

0

5

10

15

20C

DW

(g L

-1)

0

1

2

3

Time (h)0 2.5 5.0 7.5 10.0

Time (h)0 2.5 5.0 7.5 10.0

Time (h)0 2.5 5.0 7.5 10.0

0

5

10

15

20

Glc N

H+4

(g L -1) (mM

)

(A)+4 depleted Exp. growth NH+

4 depleted Exp.growth NH+4 depleted

0

1

2

3

0

0.5

1.0

1.5

2.0

3HB, H

Ac (g L -1)

CD

W(g L -1)

(B)

(C) t3 rx t3 rx yciA t3 rx zwf yciA

Exp. growth NH

CDW

GlcNH+

4

3HBHAc


72

Table 8. Calculated 3HB production parameters of different E. coli constructs. All variants and investigated under ammonium-depleted batch cultivations. The parameters are calculated for the whole experiment.

Phase Recombinantly

expressed genes

μ (h-1)

Y3HB/Glc (g g-1)

q3HB (g g-1h-1)

Exp. t3-rx 0.60 ± 0.01 0.060 ± 0.003 0.102 ± 0.003

Exp. t3-rx-yciA 0.52 ± 0.00 0.120 ± 0.096 0.205 ± 0.011

Exp. t3-rx-zwf-yciA 0.48 ± 0.00 0.170 ± 0.012 0.270 ± 0.008

NH4+ dep t3-rx - 0.162 ± 0.010 0.045 ± 0.002

NH4+ dep t3-rx-yciA - 0.302 ± 0.019 0.115 ± 0.006

NH4+ dep t3-rx-zwf-yciA - 0.302 ± 0.007 0.153 ± 0.003 Mean deviations were calculated for all strains from triplicate or duplicate experiments.

In a different study it was shown that YciA exhibited twice as much catalytic efficiency towards acetyl-CoA than to 3HB-CoA [182]. Furthermore, in that study [182] overexpression of yciA together with overexpression of the reverse β-oxidation enzymes (which are responsible for 3HB-CoA formation in the S configuration), resulted in increased acetate formation and actually eliminated (S)-3HB formation. This meant that in order to increase 3HB production by overexpression of yciA, a highly efficient 3HB-pathway capable of pulling carbon-flux from the acetyl -CoA pull was required. Although having twice as much catalytic efficiency towards acetyl-CoA than to 3HB-CoA, the selected YciA has shown to have a positive effect on 3HB production. The high observed 3HB yield on glucose on the depleted phase (0.302 g g-1) corresponds to 52% of the maximum biochemical theoretical yield. In addition, the efficient pull by the 3HB pathway on the acetyl-CoA pool in combination with an efficient 3HB-CoA hydrolysis (which allowed for a reduced 3HB-CoA concentrations) was reflected by the higher 3HB production and acetate re-consumption. These results clearly showed the importance of an efficient pathway and of the understanding and selection of the thioesterase for the production of other compounds that require thioesterase activity. 2.8.3 Impact of yciA overexpression on 3HB production in

fed-batch cultivations The potential of using fed-batch cultivations with higher cell density to increase titers and volumetric productivities has been shown in sections 2.6.3 and 2.7 of this thesis. The next step was to investigate if the improvements in 3HB production by


73

overexpression of yciA obtained in batch cultivations translate to fed-batch cultivations. As a first step, ammonium-limited fed-batch experiments with a total duration of 19.5 h and consisting of two phases were attempted (Fig. 29): (1) A batch phase to allow rapid growth until a CDW of approximately 5 g L-1 was reached. and (2) An ammonium-limited feed phase.

Figure 29. Ammonium-reduced fed-batch cultivations of E. coli AF1000 harboring plasmids:

(A) pJBGT3RX pBAD-(Km)-zwf. (B) pJBGT3RX pBAD-(Km)-zwf-yciA. The dashed lines mark the shift between phases. The arrow indicates the time of induction.

As a reference construct, AF1000 recombinantly expressing t3, rx and zwf was used (Fig. 29 A). During the first hours of feed, CDW increased with a lower expected linear rate for both variants (t3-rx-zwf and t3-rx-zwf-yciA) until finally cell growth almost completely stopped after 6 hours and high amounts of ammonium were accumulated in the broth (Fig. 29 A-B). Compared to the t3-rx-zwf variant, the t3-rx-zwf-yciA variant showed an almost 2.3 fold increase in the specific 3HB

q 3H

B(g

g-1

h-1)

00.050.100.150.200.25

3HB,

HAc

(g L

-1)

0369

1215

Glc

(g L

-1)

01020304050

CD

W(g

L-1

)

0

5

10

15

Time (h)0 5 10 15 20

Time (h)0 5 10 15 20

03691215

NH

+4(m

M)

CDW

GlcNH+

4

3HBHAc

t3 rx zwf t3 rx zwf yciA Batch NH+

4 reduced Batch NH+4 reduced

0306090120150

00.20.40.60.81.0

r3HB

(g L -1h -1)

q3HB q3HBd r3HB r3HBd

3HB, H

Ac (g L -1)

0

5

10

15

CD

W(g L -1)

(A)

(B)


74

productivity (0.085 vs 0.036 g g-1 h-1), the 3HB yield on glucose (0.14 vs 0.06 g g-1 ) and the 3HB titer (8.9 vs 4.1 g L-1) in the feed phase. Although in this setup higher 3HB production parameters were obtained with the yciA variants, ammonium-limited conditions were not achieved due to heavily decreased growth (Fig. 29). This was attributed to the toxic effects of acetic acid and/or the metabolic burden imposed by the induced pathway or proteins [220]. Furthermore, 3HB yield on glucose was negatively affected in the attempted ammonium-limited phase (0.14 g g-1), since it was not only lower than in the ammonium-depleted phase of the batch experiments (0.302 g g-1), but it was also lower than that of the exponential growth phase of the batch experiments(0.170 g g-1). In this regard, as a next step a different experimental set-up for fed-batch cultivations was designed. Contrary to ammonium-depleted batch cultivations of the t3-rx-zwf-yciA variant, the ammonia-limited fed-batch cultivations attempt of this same variant continuously accumulated acetic acid (Fig. 29). For this reason, fed-batch cultivations with ammonium-depleted conditions were designed with a total duration of 24 h and two phases (Fig. 30): (1) A growth phase to allow rapid growth until a CDW of approximately 9.5 g L-1 was reached. (2) An ammonium-depleted phase in which glucose was fed constantly fed. Furthermore, to avoid metabolic pathway burden during the growth phase, a delayed induction (OD600 9 instead of 0.2) of both recombinant plasmids was employed. As a reference construct, AF1000 recombinantly expressing t3, rx and zwf was also used (Fig. 30 A). In line with the results from ammonium-depleted batch experiments, acetate was almost completely re-consumed in the depleted phase of the strain expressing t3-rx-zwf-yciA (Fig. 30 B). As a result of yciA overexpression almost 2 times higher 3HB specific productivities (0.066 vs 0.036 g g-1 h-1) and 3HB yield on glucose (0.24 vs 0.13 g g-1) and almost 2.6 time higher 3HB titer (14.3 vs 5.4 g L-1) were obtained in the depleted-phase of this setup (compared to the reference strain expression t3-rx-zwf).


75

Figure 30. Ammonium-depleted fed-batch cultivations of E. coli AF1000 harboring plasmids: (A) pJBGT3RX pBAD-(Km)-zwf. (B) pJBGT3RX pBAD-(Km)-zwf-yciA. The dashed lines mark the shift between phases.The arrow indicates the time of induction. Arrow indicates moment of induction

Overexpression of yciA in ammonium-depleted fed-batch cultivations, successfully prevented acetic acid accumulation and a final 3HB titer of 14.3 g L-1 was obtained. Although the average specific productivity (0.066 g g h-1) was lower compared to the exponential phase (0.270 g g h-1) due to the impact of starvation on cellular metabolism, a relatively high 3HB yield on glucose of 0.24 g g-1 was obtained in the extended depleted phase. The positive impact on 3HB production of overexpression of yciA in this paper (IV), which is opposite to previous results where yciA overexpression together with overexpression of the reverse ß-oxidation pathway resulted in elimination of (S)-3HB formation [182], shows the importance of evaluating thioesterases in the context of specific pathways and specific strains and of having a highly efficient pathway preceding the reaction catalyzed by the thioesterase capable of pulling carbon-flux from the acetyl-CoA pool.

q 3H

B(g

g-1

h-1)

00.050.100.150.200.25

3HB,

HAc

(g L

-1)

0369

1215

Glc

(g L

-1)

01020304050

CD

W(g

L-1

)

0

5

10

15

Time (h)0 5 10 15 20 25

Time (h)0 5 10 15 20 25

03691215

NH

+4(m

M)

0306090120150

00.20.40.60.81.0

r3HB

(g L -1h -1)3H

B, HAc

(g L -1)

0

5

10

15

CD

W(g L -1)

(A) (B) t3 rx zwf t3 rx zwf yciA

+4 depleted Batch NH+

4 depleted Batch NH


76

2.9 Implementation of a novel recombinant expression system for metabolite production (Paper V)

2.9.1 Re-discovering the Bacterial artificial chromosome The previous sections of this thesis (2.5-2.8) have relied on the use of multi-copy plasmids for the expression or overexpression of genes to optimize the recombinant pathway for 3HB production. In Section “1.5.1.1 Genome Tools” of this thesis, the advantages and disadvantages of using multi copy plasmids and/or chromosomal integration for the expression of foreign genes or overexpression of native genes in prokaryotic organisms has been discussed. Although chromosomal integration is the preferred tool for industry, the level of difficulty of the procedure hampers its use as a screening tool. Their easier handling makes multi-copy plasmids a favorite tool for strain screening. Nevertheless, because of the varying copy number, the screening results obtained by the use of multi-copy plasmids is highly context dependent. This means that for scale-up the results obtained with multi-copy plasmids will not necessarily mimic chromosomal integration. The Bacterial artificial chromosome (BAC) (or the F plasmid) is a plasmid which was discovered in E. coli by Esther Lederberg in the year 1952 [221]. The BAC has the ability to carry DNA fragments of sizes greater than 300 kb [222]. For this reason, it was used extensively in the Human Genome Project [223, 224]. Furthermore, the BAC has an origin of replication which ensures a copy number of 1-2 per chromosome in the cell [225, 226] and a segregation system that ensures that each daughter cell gets a copy of the plasmid during cell division [226]. In addition, previous studies [226] have shown that the BAC could be stably maintained for over 150 generations (without the use of a resistance marker) and that by using ß-galactosidase as the recombinant product the growth rate using the BAC was 10% higher than for a multicopy plasmid. Taking all these benefits into account, by using 3HB as a model product this section of the thesis focuses on the investigation of the BAC as a screening tool for pathway design and optimization in Escherichia coli before chromosomal integration of the expression operon to a final production strain. This was investigated in 3 steps: (1) By recombinant expression of the genes encoding the enzymes (t3 and rx) in the 3HB-pathway (Fig. 20) the performance of BAC was compared to that of a multi-copy plasmid for the production of metabolites. (2) A strategy for pathway evaluation for the production of 3HB by modulating the translation of t3, rx and


77

zwf (since overexpression of zwf improved 3HB production, See section 2.6 of this thesis). (3) Comparing growth phenotypes by expression of whole operons on the BAC compared to chromosomal expression by native E. coli. 2.9.2 Comparing the performance of the BAC with that of a

multi-copy plasmid To investigate how 3HB production is affected by reducing the copy number, 3 different plasmids carrying the 3HB-pathway were constructed based on pJBGT3RX (Fig. 31). First a BAC based plasmid was constructed that carried the 3HB expression cassette (lacI repression, lacUV5 promoter and genes t3 and rx) from pJBGT3RX, resulting in placUV5BACT3RX. Since p15A ORI (origin of replication of pJBGT3RX) has a plasmid copy number between 15-20, while the BAC has a copy number between 1-2, an additional BAC based plasmid (ptrcBACT3RX) was constructed where genes t3 and rx were expressed using the stronger “trc” promoter [227]. Furthermore, a p15A ORI (copy number ~15) based plasmid with the trc promoter expressing t3 and rx was also constructed for this comparison “ptrcJBGT3RX”. To investigate their performance, 4 E. coli AF1000 constructs (harboring the respective plasmids) were grown in ammonium-depleted batch cultivations identical to those of the previous section (2.8.1) (Fig. 32)

Figure 31. Schematic plasmid maps of evaluated plasmid constructs. Plasmids derived from either p15A or BAC plasmids with either lacUV5 or trc promoters.

The combination of the strong (trc) promoter with the multi-copy plasmid (p15A) resulting in ptrcJBGT3RX had the lowest 3HB production parameters as well as the lowest growth rate and cell yield on glucose. The growth rate of placUV5BACT3RX was 10% higher than that of pJBGT3RX, showing the lower metabolic burden imposed by the BAC plasmid. The variant harboring placUV5BACT3RX resulted in only 30% lower 3HB yield on glucose and 3HB specific productivity compared to the pJGBT3RX (p15A ORI), although there is an approximately 10 times lower copy number of the BAC (compared to the p15A ORI), which may be explained by pJBGT3RX producing an excess of transcript

p15A

p15A

BAC

BAC

t3trc rx

t3lacUV5 rx

t3lacUV5 rx

t3 rxtrc

lacI

lacI

lacI

lacI

pJBGT3Rx

ptrcJBGT3Rx

placUV5BACT3Rx

ptrcBACT3Rx


78

while the carbon-flux is dependent by many other factors such as thermodynamics or cofactor supply. Furthermore, when changing to a stronger promoter in the BAC (from lacUV5 to trc), the 3HB specific productivity and the yield were respectively 18 and 12% higher than those of pJBGT3RX. These results indicated that the BAC could perform as well as (or better than) multi-copy plasmids and impose lower metabolic burden.

Figure 32. Comparison of the quantitative impact on 3HB production between multi-copy plasmids (p15A) and BAC at different promoter strengths (lacUV5 and trc). Parameters calculated over (A) Exponential phase. (B) Ammonium-depleted phase. Mean deviations were calculated for all strains from duplicate experiments. *ammonium-depletion not achieved due to low growth rate in exponential phase.

2.9.3 Using the BAC for pathway evaluation Given the positive results of the comparison of the performance of BAC plasmids and multi-copy plasmids together with the fact that the BAC has the ability to carry DNA fragments of sizes greater than 300 kb, the next step was to investigate if the BAC could be used for pathway evaluation by only modulating translation (and not transcription) of expressed genes. First a promoter capable of high transcription read-through was sought. The T7 system is composed of the T7 specific polymerase and the T7 promoter [228]. It has shown to give higher transcription (between 1-2 orders of magnitude) compared to the trc promoter [227], it has an efficient readthrough (transcription continues even when terminators are present), and is very well characterized and robust. Although the T7 system has been extensively used for the production of proteins in multi-copy plasmids, its strong transcription is often excessive for optimal production of metabolites [229]. This issue could potentially be overcome by the use of the BAC. For this reason, two plasmid

pJBGT3RxptrcJBGT3RxplacUV5BACT3RxptrcBACT3Rx

µ (h

-1),

3HB

(g L

-1) a

nd Y

X/G

lc (g

g-1

)

0

0.25

0.50

0.75

1.00

µ YX/Glc Y3HB/Glc q3HB

Y3H

B/G

lc (g g-1) and q

3HB (g g

-1 h-1)

0

0.05

0.10

0.15

0.20

3HB Y3HB/Glc q3HB

(A) (B)

* *


79

variants of the T7 system harboring the 3HB pathway (t3 and rx) were constructed (Fig. 33): (1) The first variant (pT7BACT3RX) expressed the T7 polymerase with a strong ribosomal binding site (RBS) (90%) and used an unrepressed T7 promoter for the expression of the 3HB pathway. (2) Due to the risk of excessive transcription of the T7 system a second variant (pR_T7BACT3RX) which used a weak RBS to express the T7 polymerase and a synthetic lacI repressed T7 promoter was constructed.

Figure 33. Schematic plasmid maps of BAC plasmids using T7 as promoter.

Figure 34. Investigation of the suitability of the T7 polymerase for 3HB production from BAC, with the two different T7 promoters. Parameters calculated over (A) Exponential phase. (B) Ammonium-depleted phase. Mean deviations were calculated for all strains from duplicate experiments.

The two T7-based plasmid variants together with the original pJBGT3RX multi-copy plasmid (as reference) were evaluated under the same ammonium-depleted batch conditions as in section 2.10.2 (Fig. 34). During exponential growth the two T7 variants resulted in similar cell yield on glucose and 10% higher growth rates than that of the reference strain. The unrepressed variant of the T7 plasmid had similar 3HB titers, 3HB yields on glucose and 3HB specific productivities compared to the reference variant. The T7 repressed variant (pR_T7BACT3RX) resulted in approximately 20% reduced 3HB production parameters compared to the reference variant. Once a BAC plasmid with a T7 repressed expression system (limited by the transcription rate) was constructed, the next step was to investigate if it could be

BAC t7lacUV5 t3 rxR_T7

BAC t7

lacI

lacI lacUV5 t3 rxT7pT7BACT3Rx

pR_T7BACT3Rx

pJBGT3RxpT7BACT3RxpR_T7BACT3Rx

µ (h

-1),

3HB

(g L

-1) a

nd Y

X/G

lc (g

g-1

)

0

0.25

0.50

0.75

1.00

µ YX/Glc Y3HB/Glc q3HB

Y3H

B/G

lc (g g-1) and q

3HB (g g

-1 h-1)

0

0.05

0.10

0.15

0.20

3HB Y3HB/Glc q3HB

(A) (B)


80

used as tool for modulating the level of translation of genes encoding enzymes in the 3HB pathway. EMOPEC [230] is a tool that helps predict and modulate the translation of any gene in Escherichia coli by varying the strength of ribosomal binding sites (RBS) of expressed genes. To assess the use of the BAC with the T7 expression system (pR_T7BAC) as a tool for screening production parameters, different strains with variated RBS strengths (predicted by EMOPEC) of expressed genes sitting in the BAC. For these reasons, constructs based on the pR_T7BAC system were created by varying the RBS’s (25% or 90% strength) of 3 genes involved in the 3HB-pathway (t3, rx, zwf) (Fig. 35 A). The 8 variants were transformed into E. coli AF1000 and subsequently screened for 3HB production with pR_T7BACT3RX as a reference (Fig. 35 B) in shake flasks grown for 24 h with minimal medium and low content of ammonium to allow 2 phases: exponential growth and ammonium depleted phase. The growth rate of all the constructs including the reference was ~ 0.75 h-1, meaning that the addition of an extra gene in the expression cassette did not result in extra metabolic burden or toxicity of the overexpressed zwf. To adequately evaluate optimal expression level of the genes in the 3HB-pathway, a factorial design was created, with predicted RBS strength as an input and 3HB yield on glucose as an outcome. The 3HB yield on glucose of the eight different strains spanned approximately a 5-fold range, correlating with the ~4-fold range of the predicted strengths of the RBSs. To assess the effect of each RBS on the 3HB yield on glucose, a multiple linear regression model was used (Fig. 35 C-D). The model suggested that genes t3 and zwf had the largest effect on 3HB yield, since even a weak RBS for rx allowed for a high 3HB yield. Furthermore, between t3 and zwf, t3 had a higher effect than zwf on 3HB yield. In addition, if the RBS of both t3 and zwf was weak an increased RBS of rx could improve the 3HB yield. These results show that the BAC can successfully be used as a tool for pathway evaluation and screening for optimal expression level of genes involved in the production pathway of a metabolite by only variating the strength of RBSs of expressed genes.


81

Figure 35. Study of pathway optimization of the genes encoding enzymes in the 3HB-pathway (t3, rx and zwf) in the BAC. (A) Schematic representation of the BAC with variation of strength of RBSs strong (90%) or weak (25%). (B) Screening performance of BAC variants using pR_T7BACT3RX as reference. (C) A model fitted using the software MODDE (Umetrics, Umeå,Sweden) (excluding pR_T7BACT3RX) of the dependence of the 3HB yield on the RBS strength. (D) Values predicted by the model (solid line) plotted against experimental values for the Y3HB/Glc. The dashed line indicates the area of two standard deviations above and below the line. * = Removed outlier.

2.9.4 The BAC mimics chromosomal expression Given that the BAC allowed for a simple strategy for pathway optimization, the next step was to investigate how well the expression of a whole operon on the BAC and in a native chromosome correlated. For this reason, as a model operon the complete 9.5 kb native glycolic acid utilization operon (glcC-A) was selected (Fig. 36 A). To investigate the correlation, first the glcC-A operon was deleted from the chromosome of E. coli W3110 resulting in strain W3110 ΔglcC-A. Then this strain was transformed with a BAC plasmid harboring the same deleted operon (pBAC-GA). The 3 resulting strains, (1) W3100, (2) W3110 ΔglcC-A, and (3) W3110 ΔglcC-A pBAC-GA were grown in a minimal medium using glycolic acid as the sole carbon source (Fig. 36 B). As expected, the deletion strain (W3110 ΔglcC-A) was unable to

rx RBS

t3 RBSzwf RBS

90%

90% 90%

25%

25%

25%

Y3HB/Glc (mg g-1)

0

8

16

24

32

40

Expe

rimen

tal Y

3HB/

Glc (

mg

g-1)

0

10

20

30

40

Predicted Y3HB/Glc (mg g-1)0 10 20 30 40

A

CB

StrongorWeak

StrongorWeak

StrongorWeak

BAC RBS RBS RBSzwft7lacI t3 rxlacUV5 R_T7

D

zwf RBSt3 RBSrx RBS

Y X/G

lc (m

g g-1

)

0

50

100

150

200

Y HAc

/Glc

(mg

g-1)

0

50

100

150

200

Y 3HB

/Glc

(mg

g-1)

0

10

20

30

40

-9080

909090

259090

902590

909025

252590

259025

902525

252525


82

grow. The original strain (W3110) had a growth rate of 0.220 h-1. The strain W3110 ΔglcC-A pBAC-GA had similar cell yield on glycolic acid (compared to the wild type strain) and a growth rate of 0.207 h-1, which corresponded to 94% of that of the wild type strain. A 10% lower glycolic specific consumption rate was also observed, which seemed to explain the decrease in the growth rate. Although slightly lower growth rates and glycolic acid specific consumption rates were observed the differences were small. These results show the ability of the BAC to mimic chromosomal expression.

Figure 36. A) glcC-A operon from E. coli W3110. Picture adapted from ecocyc.org. B) Comparison of strains carrying the glcC-A operon either on the chromosome or on BAC. Standard deviations were calculated for all strains from triplicate experiments. *: No growth.

A

BglcA glcB glcG glcF glcE glcD

glcCPdhR ArcA

GlcCGlcC

CraCrpIHF

Fis

W3110W3110 glcC-AW3110 glcC-A pBAC-GA

***

µ (h

-1) a

nd Y

X/G

A (g

g-1

)

0

0.05

0.10

0.15

0.20

0.25

0.30q

GA (g g

-1 h-1)

0

0.25

0.50

0.75

1.00

1.25

1.50

µ YX/GA qGA


83

3. Concluding remarks and outlook The ultimate goal for microbial-cell-factory-based processes is sustainable and profitable production of fuels and chemicals. To develop and implement scalable robust bioprocesses requires integration of strain and bioprocess-design. In this thesis several integrated bioprocess- and strain-design strategies were studied to improve 3HB production parameters for robust and scalable bioprocesses. A particular emphasis was put on the robustness and scalability of the process. For this reason, strains and cultivation techniques that have the potential to be scaled were designed and investigated. Although yields and specific productivities can and were also improved in batch cultivations, fed-batch cultivations with higher cell density were necessary to further increase titers and volumetric productivities. The table below shows a summary of the improvements in scalable fed-batch cultivations. Table 9. Calculated production parameters of different metabolic engineering strategies for E. coli strain AF1000 or BL21 cultivated in different fed-batch cultivation set-ups.

Setup Strain Phase Y3HB/Glc (g g-1)

Total Y3HB/Glc (g g-1)

q3HB (g g-1 h-1)

3HB titer

(g L-1) CDW (g L-1)

NH4+-limited Exp. + Linear

Feed (NH4++Glc.)

AF1000 t3 rx

Exp. 0.057 0.053

0.068 4.1

(13h) 21 Lim. 0.050 0.030

NH4+-limited Constant Feed

(NH4++Glc.)

AF1000 t3 rx zwf

Exp. 0.072 0.0724

0.048 7.04 (24 h) 20.1

Lim. 0.076 0.022 NH4+-limited

Constant Feed (NH4++Glc.)

BL21 t3 rx zwf

Exp. 0.042 0.102

0.048 16.3 (24h) 47

Lim. 0.138 0.040 NH4+-depleted Constant Feed

(Glc.)

AF1000 t3 rx zwf

Exp. - 0.110

- 5.4 (24h) 9.25

Dep. 0.130 0.035 NH4+-depleted Constant Feed

(Glc.)

AF1000 t3 rx zwf

yciA

Exp. - 0.210

- 14.3 (24 h) 9.9 Dep. 0.24 0.066

The second highest 3HB specific productivity (0.066 g g-1 h-1) and highest yield of 3HB on glucose (0.21 g g-1) were obtained with the strain AF1000 expressing t3-rx-zwf-yciA. This yield corresponds to an almost 270% improvement compared to the initial yield obtained by E. coli AF1000 pJBGT3RX in exponential growth of a batch ending with glucose depletion of 0.057 g g-1 [181]. Nevertheless, the 0.21 g g-1 3HB yield on glucose still only corresponds to 27% of the maximum theoretical yield (defined as all available electrons ending up in the product and no growth) or 36% of the maximum biochemical yield with two acetyl-CoA derived from pyruvate


84

converted to 3HB. Despite successfully avoiding acetic acid accumulation by AF1000 t3-rx-zwf-yciA grown in ammonium-depleted fed-batch cultures, this effect was not as beneficial for ammonium-limited fed-batch cultures, that in theory would allow higher cell density cultivations with higher productivities and titers. With the BL21 strain expressing t3-rx-zwf the highest cell densities were obtained which allowed for both the highest volumetric productivity of 3HB (1.27 g L-1 h-1) and the highest 3HB titer (16.3 g L-1) described in literature by recombinant E. coli, although the 3HB yields on glucose (0.103 g g-1) and specific productivities (0.040 g g-1 h-1) are still considerably low. If 3HB was considered for industrial production a more detailed analysis of the efficiency and economics of the bio-process would be necessary to determine the minimum yield of product on substrate necessary for profitable production, as well as the possible trade-off between volumetric rates and yields. Although no detailed economic analysis was made in this thesis, commercially viable production of a medium value product is likely to require a volumetric productivity of at least 2 g L-1 h-1 , > 50% of theoretical yield and a final product titer >100 g L-1. Despite the improvements achieved in this thesis, this means that further improvements in production parameters are still required. One possible improvement would be further optimization of the uncoupling between growth and product formation. Ammonium depletion resulted in the highest product yield coinciding with the second highest biomass specific 3HB production rate. However, due to relatively low biomass concentrations, the volumetric productivity was also low. Prolonged growth in the absence of product formation and/or recycling of the cells, could result in increased biomass concentrations and thereby result in increased volumetric productivities. Such a cultivation design would further benefit from decreasing the expression of for instance the TCA cycle by CRISPRi during the production phase. Additionally, further improvements are possible by integrating improved strain design: - To improve the thermodynamic pull of the 3HB pathway, the

thermodynamically unfavorable thiolase reaction, could be replaced by combined ATP-dependent carboxylation of acetyl-CoA into malonyl-CoA and subsequent condensation/decarboxylation of malonyl-CoA and acetyl-CoA to acetoacetyl-CoA [231].


85

- Protein-design/engineering of a thioesterase with increased substrate specificity for (R)-3HB-CoA to decrease acetate formation and thereby increase the potential to achieve high 3HB titers.

- The redox co-factor use of the Entner Doudoroff pathway, would allow

formation of 1 mol of 3HB per 1 mol of glucose, compared to 0.8 mol of 3HB per 1 mol of glucose when the pentose phosphate pathway is used for NADPH regeneration. Enabling the use of this pathway would require deregulation of the transcription of this pathway in E. coli [232].

- Alternatively, especially under non-growing conditions, replacement of the

native NAD+-GAPDH (glyceraldehyde 3-phosphate dehydrogenase) (EC:1.2.1.12) with an NADP+ dependent 3-phophoglycerate yielding reaction (EC:1.2.1.9) [233, 234], might improve the availability of NADPH for 3HB production, as has previously been shown for other NADPH dependent pathways [233, 235].

- The pathway from glucose to 3HB as used in this thesis yields an excess of

NADH. When ATP is required for growth it can to some extent be beneficial to use respiration to regenerate NAD+. However, to get closer to the maximum theoretical yield, especially under non-growing conditions, recombinant expression of phosphoribulokinase (PRK) and ribulose-1,5-bisphosphate carboxylase (Rubisco) might be beneficial to redirect excess NADH and carbon dioxide [236, 237], and thereby increase the yield of 3HB on sugar.

The strategies for improvement of 3HB production that were investigated in this thesis, resulted in improved yields, titers and productivities. These findings are not only relevant for 3HB production but also contribute to the knowledge in the general industrial biotechnology field on how to employ design strategies to improve the production of a metabolite of medium value.


86

4. Acknowledgements

I would like to thank all the people that in one way or another have helped or supported me during the years of completing this PhD. Thank you, Gen Larsson, my main supervisor, for believing in me and accepting me as a PhD student. I am extremely grateful for your kindness, guidance and support over this years. Special thanks Ton van Maris, my co-supervisor, for taking this challenge in the last years of my Ph.D. Thank you for your availability, support, discussions and immeasurable guidance. Thank you Jorge Quillaguamán, my other co-supervisor, for such enthusiasm, dedication to this project and for your support through these years. Thank you to the SIDA Cooperation for the financial support for the realization of this PhD. Special thanks to Teresa Soop, Veronica Melander and Aksana Mushkavets for their kindness and understanding through this years. Thank you FORMAS for the financial support. Special thanks to Martin Hedberg for managing projects and organizing all the interesting meetings. Thank you, Andres Veide for accepting to be my internal advance reviewer and for your kindness during these years. Thank you, Gustav Sjöberg for introducing me to many things in the lab, for all the discussions, encouragement, friendship and all your help through these years. Thank you, Mariel Perez, for all the significant collaboration and cooperation during this PhD. Thank you as well for the friendship during this years. Thank you, Johan Jarmander and Jaroslav Belotserkovsky, for the nice discussions and introducing me to many things in the lab in the beginning of my PhD. Thank you, Karin Sjöberg-Gällno for your kindness and for all the assistance and your help for Paper I.


87

Thank you, Martin Gustavsson for all the contributions, discussions and support over this years. Thank you, David Hörnström for the scientific discussions, for the conversations, for the nice atmosphere and for becoming a friend. Thank you, Johannes Yayo for helping me in my experimental work for some time (and also for your help with Swedish). Thanks also for your kindness and for becoming a friend. Thank you Teun Kuil and Magnus Lindroos for your sense of humor for the nice conversations, good mood in the office and friendship. Thank you to all the colleagues, support staff and students in the Department of Industrial Biotechnology not only for the inspiration but also for your assistance, advice and specially for your friendship! Special thanks to Caroline B., Ye, Lina, Johan N., Sandhya, Atefeh, Erika R., Erika H.,Veronique, Guna, Kaj, Anna Klara, Berndt, Nils, Hubert, Liang, Caijuan, Brigita, Meeri, Lubna, Emil and Martin D. Thanks to all the Bolivian SIDA Ph.D. students of my generation for the push and nice friendship made during these years. Thank you to the Center of Biotechnology and the Bioprocess center at UMSS. Thank you Papá y Mamá. Papi y Mami muchas gracias por todo el amor, el apoyo, empuje, dirección y el tiempo otorgado por todos estos años. Sin ustedes no estaría acá y es a ustedes a quienes les debo todo. Papi gracias por siempre escucharme, por enseñarme a resolver mis cosas y ver las cosas desde otra perspectiva. Gracias mami por toda la bondad de tu corazón, gracias por siempre creer en mi y ser mi motor. Thank you Jorge, Dani and Señora Lucy. Gracias Jorgito por todo el cariño, empuje y la comprensión que me das. Gracias mi Dani por siempre alentarme y empujarme. Gracias Señora Lucy por su cariño y apoyo en estos años. Thank you to my dear loved husband Alejandro. Gracias mi amor hermoso por apoyarme en todos mis sueños no puedo agradecerte suficiente por toda la ayuda y paciencia que me das y por estar junto a mi en todo este proceso.


88

5. Abbreviations

(S)-3HB (S)-3-hydroxybutyrate (S)-3HB-CoA (S)-3-hydroxybutyrate-CoA 3HB (R)-3-hydroxybutyrate 3HB-CoA (R)-3-hydroxybutyrate-CoA 3HP 3-hydroxypropionyl 3HP-CoA 3-hydroxypropionyl-CoA 3HV 3-hydroxyvalerate 6-PG/PK 6-phosphogluconate/phosphoketolase ABE Acetone-butanol-ethanol AcAcCoA Acetoacetyl-CoA AcCoA Acetyl-CoA BAC Bacterial artificial chromosome CDW Cell dry weight COBRA Constraint based reconstruction and analysis CRISPR Clustered regularly interspaced short palindromic repeats CRISPRi CRISPR interference Cs Substrate concentration CSI Classical strain improvement DHAP Dihydroxyacetone phosphate DNA Deoxyribonucleic acid DSP Downstream process E. coli Escherichia coli F Inflow/outflow rate G3PDH Glycerol 3-phosphate dehydrogenase G3PP Glycerol 3-phosphate phosphatase GA Glycolic acid GAPDH Glyceraldehyde 3-phosphate dehydrogenase Glc Glucose H. boliviensis Halomonas boliviensis HAc Acetic acid HPLC High performance liquid chromatography HVLV High volume/ low value IEA International Energy Agency IME Inverse metabolic engineering IPTG Isopropyl b-D-1-thiogalactopyranoside Kcat Enzyme turnover number KM Michaelis-Menten constant- Substrate saturation constant Ks Substrate saturation constant LA Lactic acid


89

LAB Lactic acid producing bacteria LVHV Low volume/ high value MFA Metabolic flux analysis NAD Nicotinamide adenine dinucleotide NADP Nicotinamide adenine dinucleotide phosphate NREL United States’ National Renewable Energy OD600 Optical density at 600 nm PDO 1,3 propanediol PHA Polyhydroxyalkanoate PHB Polyhydroxybutyrate PLA Polylactic acid PTS Phosphotransferase system PTT Polytrimethylene terephthalate Pyr Pyruvate q Specific rate q3HB (R)-3-hydroxybutyrate specific production rate qGA Glycolic acid specific consumption rate qGlc Glucose specific consumption rate qHAc Acetic acid specific production rate r3HB (R)-3-hydroxybutyrate volumetric production rate RBS Ribosomal binding site ri Volumetric rate Ri Total rate RNA Ribonucleic acid rRNA Ribosomal RNA rx H. boliviensis’ gene encoding acetoacetyl-CoA reductase S Concentration of nutrient S. cerevisiae Saccharomyces cerevisiae t3 H. boliviensis’ gene encoding a 3-ketothiolase TCA cycle Tricarboxylic acid TN Phase transfer rate TRYQR Titer, rate, yield, quality and robustness UV Ultraviolet Vmax Maximum reaction rate WGS Whole genome sequencing X Cell mass Y Yield Y3HB/Glc (R)-3-hydroxybutyrate yield on glucose Y3HB/HAc (R)-3-hydroxybutyrate yield on acetate YHAc/Glc Acetic acid yield on glucose YX/GA Cell mass yield on glycolic acid µ Growth rate


90

6. References 1. The European Commission: Eurostat statistics explained. Oil and petroleum products -

A statistical overview [https://ec.europa.eu/eurostat/statistics-explained/index.php/Oil_and_petroleum_products_-_a_statistical_overview#Use_of_petroleum_products]

2. US Energy Information Administration: Today in energy [https://www.eia.gov/todayinenergy/detail.php?id=35672]

3. Clews RJ: Chapter 11 - The petrochemicals industry. In Project finance for the international petroleum industry. Edited by Clews RJ. San Diego: Academic Press; 2016: 187-203.

4. Villadsen J, Nielsen J, Lidén G: Chemicals from metabolic pathways. In Bioreaction Engineering Principles. Boston, MA: Springer US; 2011: 7-62.

5. Earth Science Communications Team at NASA’s Jet Propulsion Laboratory: Global climate change: Vital signs of the planet [https://climate.nasa.gov/]

6. US National Oceanic and Atmospheric Administration (NOAA) Earth System Research Laboratory: The NOAA annual greenhouse gas index (AGGI) [https://www.esrl.noaa.gov/gmd/aggi/aggi.html]

7. The European Commission: Energy: Renewable energy directive [https://ec.europa.eu/energy/en/topics/renewable-energy/renewable-energy-directive]

8. Kamm B, Kamm M, Gruber PR, Kromus S: Biorefinery systems – An overview. In Biorefineries-Industrial Processes and Products. Edited by Kamm B, Gruber PR, Kamm M. Weinheim: WILEY-VCH; 2006.

9. The European Commission: Research and innovation: Bioeconomy [https://ec.europa.eu/research/bioeconomy/index.cfm]

10. Brundtland G, Khalid M, Agnelli S, Al-Athel S, Chidzero B, Fadika L, Hauff V, Lang I, Shijun M, Morino de Botero M, et al: Our Common Future ('Brundtland report'). Oxford University Press, USA; 1987.

11. International energy agency: Task 42: Biorefineries [https://www.iea-bioenergy.task42-biorefineries.com/upload_mm/5/8/2/a47bd297-2ace-44d0-92bb-cb7cc02f75de_Brochure_Totaal_definitief_webs.pdf]

12. U.S National Renewable Energy Laboratory (NREL). What Is a Biorefinery? [https://www.nrel.gov/biomass/biorefinery.html]

13. Ree Rv, Annevelink E: Status report biorefinery 2007. Wageningen: Agrotechnology & Food Sciences Group; 2007.

14. Naik SN, Goud VV, Rout PK, Dalai AK: Production of first and second generation biofuels: A comprehensive review. Renewable & Sustainable Energy Reviews 2010, 14:578-597.

15. Kamm B, Gruber PR, Kamm M: Biorefineries–Industrial processes and products. In Ullmann's Encyclopedia of Industrial Chemistry. 2016.

16. Kamm B, Kamm M: Principles of biorefineries. Applied Microbiology and Biotechnology 2004, 64:137-145.

17. Thoen J, Busch R: Industrial chemicals from biomass – Industrial concepts. In Biorefineries-Industrial Processes and Products. Edited by Kamm B, Gruber PR, Kamm M. Weinheim: WILEY-VCH; 2006.

18. Kamm B, Kamm M, Schmidt M, Hirth T, Schulze M: Lignocellulose-based chemical products and product family trees. In Biorefineries-Industrial Processes and Products. Edited by Kamm B, R. GP, M. K2008.

19. Cherubini F, Jungmeier G, Wellisch M, Willke T, Skiadas I, Van Ree R, de Jong E: Toward a common classification approach for biorefinery systems. Biofuels Bioproducts & Biorefining-Biofpr 2009, 3:534-546.

20. Cherubini F: The biorefinery concept: Using biomass instead of oil for producing energy and chemicals. Energy Conversion and Management 2010, 51:1412-1421.

21. Chen H, Wang L: Chapter 1 - Introduction. In Technologies for Biochemical Conversion of Biomass. Edited by Chen H, Wang L. Oxford: Academic Press; 2017: 1-10.

22. International energy agency: Task 33: Gasification of biomass and waste [http://task33.ieabioenergy.com/content/taks_description]

23. Parthasarathy P, Narayanan KS: Hydrogen production from steam gasification of biomass: Influence of process parameters on hydrogen yield – A review. Renewable Energy 2014, 66:570-579.


91

24. Sharma A, Pareek V, Zhang D: Biomass pyrolysis—A review of modelling, process parameters and catalytic studies. Renewable and Sustainable Energy Reviews 2015, 50:1081-1096.

25. Bonn G, Concin R, Bobleter O: Hydrothermolysis - a new process for the utilization of biomass. Wood Science and Technology 1983, 17:195-202.

26. Hu J, Yu F, Lu YW: Application of Fischer-Tropsch Synthesis in Biomass to Liquid Conversion. Catalysts 2012, 2:303-326.

27. Demirbas A: Biochemical processes. In Biorefineries: For Biomass Upgrading Facilities. London: Springer London; 2010: 193-209.

28. Werpy T, Petersen G: Top value added chemicals from biomass: Volume I -- Results of screening for potential candidates from sugars and synthesis gas. ; National Renewable Energy Lab., Golden, CO (US); 2004.

29. Murzin DY, Leino R: Sustainable chemical technology through catalytic multistep reactions. Chemical Engineering Research and Design 2008, 86:1002-1010.

30. Demirbas A: Biorefinery. In Biorefineries: For Biomass Upgrading Facilities. Edited by Demirbas A. London: Springer London; 2010: 75-92.

31. UN Convention on Biological Diversity: Article 2. Use of terms [https://www.cbd.int/convention/articles/default.shtml?a=cbd-02]

32. Willke T, Vorlop KD: Industrial bioconversion of renewable resources as an alternative to conventional chemistry. Applied Microbiology and Biotechnology 2004, 66:131-142.

33. Jones DT, Woods DR: Acetone-butanol fermentation revisited. Microbiological Reviews 1986, 50:484-524.

34. Clark JH, Farmer TJ, Hunt AJ, Sherwood J: Opportunities for bio-based solvents created as petrochemical and fuel products transition towards renewable resources. International Journal of Molecular Sciences 2015, 16:17101-17159.

35. Hara KY, Araki M, Okai N, Wakai S, Hasunuma T, Kondo A: Development of bio-based fine chemical production through synthetic bioengineering. Microbial Cell Factories 2014, 13:173.

36. de Carvalho CCCR: Whole cell biocatalysts: essential workers from Nature to the industry. Microbial Biotechnology 2017, 10:250-263.

37. Vandamme EJ: The search for novel microbial fine chemicals, agrochemicals and biopharmaceuticals. Journal of Biotechnology 1994, 37:89-108.

38. Jullesson D, David F, Pfleger B, Nielsen J: Impact of synthetic biology and metabolic engineering on industrial production of fine chemicals. Biotechnology Advances 2015, 33:1395-1402.

39. Ouyang J, Ma R, Zheng ZJ, Cal C, Zhang M, Jiang T: Open fermentative production of L-lactic acid by Bacillus sp strain NLO1 using lignocellulosic hydrolyzates as low-cost raw material. Bioresource Technology 2013, 135:475-480.

40. Shi D, Hua J, Zhang L, Chen M: Synthesis of bio-based poly(lactic acid-co-10-hydroxy decanoate) copolymers with high thermal stability and ductility. Polymers 2015, 7:468.

41. Cubas-Cano E, Gonzalez-Fernandez C, Ballesteros M, Tomas-Pejo E: Biotechnological advances in lactic acid production by lactic acid bacteria: lignocellulose as novel substrate. Biofuels Bioproducts & Biorefining-Biofpr 2018, 12:290-303.

42. Datta R, Tsai S-P, Bonsignore P, Moon S-H, Frank JR: Technological and economic potential of poly(lactic acid) and lactic acid derivatives. FEMS Microbiology Reviews 1995, 16:221-231.

43. de Oliveira RA, Maciel R, Rossell CEV: High lactic acid production from molasses and hydrolysed sugarcane bagasse. 2nd International Conference on Biomass (Iconbm 2016) 2016, 50:307-312.

44. Rodrigues C, Vandenberghe LPS, Woiciechowski AL, de Oliveira J, Letti LAJ, Soccol CR: 24 - Production and application of lactic acid. In Current Developments in Biotechnology and Bioengineering. Edited by Pandey A, Negi S, Soccol CR: Elsevier; 2017: 543-556.

45. Alves de Oliveira R, Komesu A, Vaz Rossell CE, Maciel Filho R: Challenges and opportunities in lactic acid bioprocess design—From economic to production aspects. Biochemical Engineering Journal 2018, 133:219-239.

46. Kadam SR, Patil SS, Bastawde KB, Khire JM, Gokhale DV: Strain improvement of Lactobacillus delbrueckii NCIM 2365 for lactic acid production. Process Biochemistry 2006, 41:120-126.

47. Miller C, Fosmer A, Rush B, McMullin T, Beacom D, Suominen P: 3.17 - Industrial production of lactic acid. In Comprehensive Biotechnology (Second Edition). Edited by Moo-Young M. Burlington: Academic Press; 2011: 179-188.


92

48. Komesu A, de Oliveira JAR, Martins LHD, Maciel MRW, Maciel R: Lactic acid production to purification: a review. Bioresources 2017, 12:4364-4383.

49. Hamdan A, Sonomoto K: Production of optically pure lactic acid for bioplastics. In Lactic Acid Bacteria: Current Progress in Advanced Research. Edited by Sonomoto K, Yokota A. Norfolk; 2011: 211-220.

50. Demirci A, Pometto AL: Enhanced production of D(-)-lactic acid by mutants of Lactobacillus delbrueckii ATCC-9649. Journal of Industrial Microbiology 1992, 11:23-28.

51. Derkx PM, Janzen T, Sørensen KI, Christensen JE, Stuer-Lauridsen B, Johansen E: The art of strain improvement of industrial lactic acid bacteria without the use of recombinant DNA technology. Microbial Cell Factories 2014, 13:S5.

52. Patnaik R, Louie S, Gavrilovic V, Perry K, Stemmer WP, Ryan CM, del Cardayre S: Genome shuffling of Lactobacillus for improved acid tolerance. Nature Biotechnology 2002, 20:707-712.

53. Okano K, Tanaka T, Kondo A: Lactic acid. In Bioprocessing of Renewable Resources to Commodity Bioproducts. Edited by Bisaria VS, Kondo A. New Jersey, US: John Wiley & Sons; 2014: 353-380.

54. Anastassiadis S, Morgunov IG, Kamzolova SV, Finogenova TV: Citric acid production patent review. Recent Patents on Biotechnology 2008, 2:107-123.

55. Max B, Salgado JM, Rodriguez N, Cortes S, Converti A, Dominguez JM: Biotechnological production of citric acid. Brazilian Journal of Microbiology 2010, 41:862-875.

56. American Chemical Society: Discover chemistry: Pfizer’s work on penicillin for World War II becomes a National Historic Chemical Landmark [https://www.acs.org/content/acs/en/pressroom/newsreleases/2008/june/pfizers-work-on-penicillin-for-world-war-ii-becomes-a-national-historic-chemical-landmark.html]

57. Kubicek CP, Punt P, Visser J: Production of organic acids by filamentous fungi. In Industrial Applications The Mycota (A Comprehensive Treatise on Fungi as Experimental Systems for Basic and Applied Research). Volume 10. Edited by Hofrichter M. Berlin, Heidelberg: Springer Berlin Heidelberg; 2011: 215-234.

58. Oladele KO, Siew QY, Aziz Zakry FA, Lan JC-W, Ling TC: Overview of citric acid production from Aspergillus niger Frontiers in Life Science 2015, 8:271-283.

59. Kirimura K, Honda Y, Hattori T: 3.13 - Citric acid. In Comprehensive Biotechnology (Second Edition). Edited by Moo-Young M. Burlington: Academic Press; 2011: 135-142.

60. Kubicek CP, Karaffa L: Organic acids. In Basic Biotechnology. 3 edition. Edited by Kristiansen B, Ratledge C. Cambridge: Cambridge University Press; 2006: 359-380.

61. Schreferl-Kunar G, Grotz M, Röhr M, Kubicek CP: Increased citric acid production by mutants of Aspergillus niger with increased glycolytic capacity. FEMS Microbiology Letters 1989, 59:297-300.

62. Torres NV: Modeling approach to control of carbohydrate metabolism during citric acid accumulation by Aspergillus niger: II. Sensitivity analysis. Biotechnology and Bioengineering 1994, 44:112-118.

63. Torres NV: Modeling approach to control of carbohydrate metabolism during citric acid accumulation by Aspergillus niger: I. Model definition and stability of the steady state. Biotechnology and Bioengineering 1994, 44:104-111.

64. Arisan-Atac I, Wolschek MF, Kubicek CP: Trehalose-6-phosphate synthase A affects citrate accumulation by Aspergillus niger under conditions of high glycolytic flux. FEMS Microbiol Lett 1996, 140:77-83.

65. American Chemical Society: Chemical landmarks: Penicillin production through deep-tank fermentation [https://www.acs.org/content/acs/en/education/whatischemistry/landmarks/penicillin.html#penicillin-production-by-deep-tank-fermentation]

66. Public Broadcasting Service: Health: The real story behind penicillin [https://www.pbs.org/newshour/health/the-real-story-behind-the-worlds-first-antibiotic]

67. The Chemical Engineer from The Institution of Chemical Engineers (IChemE) Magazine and Journals: Chemical engineers who changed the world: Pfizer's penicillin pioneers – Jasper Kane and John McKeen [https://www.thechemicalengineer.com/features/cewctw-pfizers-penicillin-pioneers-jasper-kane-and-john-mckeen/]

68. Salo OV, Ries M, Medema MH, Lankhorst PP, Vreeken RJ, Bovenberg RA, Driessen AJ: Genomic mutational analysis of the impact of the classical strain improvement program on beta-lactam producing Penicillium chrysogenum. BMC Genomics 2015, 16:937.

69. Adrio JL, Demain AL: Genetic improvement of processes yielding microbial products. FEMS Microbiology Reviews 2006, 30:187-214.


93

70. Schmidt F-R: The β-Lactam antibiotics: current situation and future prospects in manufacture and therapy. In Industrial Applications The Mycota (A Comprehensive Treatise on Fungi as Experimental Systems for Basic and Applied Research). Edited by Hofrichter M. Berlin, Heidelberg: Springer Berlin Heidelberg; 2011: 101-121.

71. Alper H, Stephanopoulos G: Engineering for biofuels: exploiting innate microbial capacity or importing biosynthetic potential? Nature Reviews Microbiology 2009, 7:715-723.

72. Chedrese PJ: Recombinant DNA technology. In Reproductive Endocrinology: A Molecular Approach. Edited by Chedrese PJ. Boston, MA: Springer US; 2009: 75-82.

73. Nielsen J, Keasling JD: Engineering cellular metabolism. Cell 2016, 164:1185-1197. 74. Liu H, Xu Y, Zheng Z, Liu D: 1,3-propanediol and its copolymers: research, development

and industrialization. Biotechnology Journal 2010, 5:1137-1148. 75. Kraus GA: Synthetic methods for the preparation of 1,3-propanediol. CLEAN – Soil, Air,

Water 2008, 36:648-651. 76. Vivek N, Binod P: Biosynthesis of 1,3-Propanediol: Genetics and Applications. In

Biosynthetic Technology and Environmental Challenges. Edited by Varjani SJ, Parameswaran B, Kumar S, Khare SK. Singapore: Springer Singapore; 2018: 143-165.

77. Zeng AP, Biebl H: Bulk chemicals from biotechnology: the case of 1,3-propanediol production and the new trends. Advances in Biochemical Engineering/Biotechnology 2002, 74:239-259.

78. Nakamura CE, Whited GM: Metabolic engineering for the microbial production of 1,3-propanediol. Current Opinion in Biotechnology 2003, 14:454-459.

79. Saltzberg MA, Byrne AM, Jackson EN, Miller ES, Nelson MJ, Tyreus BD, Zhu Q: DuPont: Biorenewables at E.I. DU Pont DE Nemours & Co. In Industrial Biorenewables. Edited by Domínguez de María P2016.

80. Straathof AJJ: Transformation of biomass into commodity chemicals using enzymes or cells. Chemical Reviews 2014, 114:1871-1908.

81. DuPont Tate & Lyle BioProducts: DuPont Tate & Lyle bio products celebrates 10 years of bio-based performance [http://www.duponttateandlyle.com/DuPont-Tate-and-Lyle-Bio-Products-10%20Years-Bio-based-Performance-1-3-Propanediol]

82. Paddon CJ, Westfall PJ, Pitera DJ, Benjamin K, Fisher K, McPhee D, Leavell MD, Tai A, Main A, Eng D, et al: High-level semi-synthetic production of the potent antimalarial artemisinin. Nature 2013, 496:528.

83. Adrio JL, Demain AL: Recombinant organisms for production of industrial products. Bioengineered Bugs 2010, 1:116-131.

84. Martin VJ, Pitera DJ, Withers ST, Newman JD, Keasling JD: Engineering a mevalonate pathway in Escherichia coli for production of terpenoids. Nature Biotechnology 2003, 21:796-802.

85. Newman JD, Marshall J, Chang M, Nowroozi F, Paradise E, Pitera D, Newman KL, Keasling JD: High-level production of amorpha-4,11-diene in a two-phase partitioning bioreactor of metabolically engineered Escherichia coli. Biotechnology and Bioengineering 2006, 95:684-691.

86. Pitera DJ, Paddon CJ, Newman JD, Keasling JD: Balancing a heterologous mevalonate pathway for improved isoprenoid production in Escherichia coli. Metabolic Engineering 2007, 9:193-207.

87. Tsuruta H, Paddon CJ, Eng D, Lenihan JR, Horning T, Anthony LC, Regentin R, Keasling JD, Renninger NS, Newman JD: High-level production of amorpha-4,11-diene, a precursor of the antimalarial agent artemisinin, in Escherichia coli. PLOS ONE 2009, 4:e4489.

88. Ro D-K, Paradise EM, Ouellet M, Fisher KJ, Newman KL, Ndungu JM, Ho KA, Eachus RA, Ham TS, Kirby J, et al: Production of the antimalarial drug precursor artemisinic acid in engineered yeast. Nature 2006, 440:940.

89. Lenihan JR, Tsuruta H, Diola D, Renninger NS, Regentin R: Developing an industrial artemisinic acid fermentation process to support the cost-effective production of antimalarial artemisinin-based combination therapies. Biotechnology Progress 2008, 24:1026-1032.

90. Paddon CJ, Keasling JD: Semi-synthetic artemisinin: a model for the use of synthetic biology in pharmaceutical development. Nature Reviews Microbiology 2014, 12:355.

91. Laycock B, Halley P, Pratt S, Werker A, Lant P: The chemomechanical properties of microbial polyhydroxyalkanoates. Progress in Polymer Science 2013, 38:536-583.

92. Anjum A, Zuber M, Zia KM, Noreen A, Anjum MN, Tabasum S: Microbial production of polyhydroxyalkanoates (PHAs) and its copolymers: A review of recent advancements. International Journal of Biological Macromolecules 2016, 89:161-174.


94

93. Croxen MA, Law RJ, Scholz R, Keeney KM, Wlodarska M, Finlay BB: Recent advances in understanding enteric pathogenic Escherichia coli. Clinical Microbiology Reviews 2013, 26:822-880.

94. Sagong HY, Son HF, Choi SY, Lee SY, Kim KJ: Structural insights into polyhydroxyalkanoates biosynthesis. Trends in Biochemical Sciences 2018, 43:790-805.

95. Crank M, Patel M, Marscheider-Weidemann F, Schleich J, Huesing B, Angerer G: Techno-economic feasibility of large-scale production of bio-based polymers in Europe (PRO-BIP) Final report. Netherlands, 2004.

96. Możejko-Ciesielska J, Kiewisz R: Bacterial polyhydroxyalkanoates: Still fabulous? Microbiological Research 2016, 192:271-282.

97. Chen GQ: A microbial polyhydroxyalkanoates (PHA) based bio- and materials industry. Chemical Society Reviews 2009, 38:2434-2446.

98. Orts WJ, Nobes GAR, Kawada J, Nguyen S, Yu G-e, Ravenelle F: Poly(hydroxyalkanoates): Biorefinery polymers with a whole range of applications. The work of Robert H. Marchessault. Canadian Journal of Chemistry 2008, 86:628-640.

99. Raza ZA, Abid S, Banat IM: Polyhydroxyalkanoates: Characteristics, production, recent developments and applications. International Biodeterioration & Biodegradation 2018, 126:45-56.

100. Zinn M, Witholt B, Egli T: Occurrence, synthesis and medical application of bacterial polyhydroxyalkanoate. Advanced Drug Delivery Reviews 2001, 53:5-21.

101. Chen G-Q: Plastics completely synthesized by bacteria: polyhydroxyalkanoates. In Plastics from Bacteria: Natural Functions and Applications. Edited by Chen GG-Q. Berlin, Heidelberg: Springer Berlin Heidelberg; 2010: 17-37.

102. Kosseva MR, Rusbandi E: Trends in the biomanufacture of polyhydroxyalkanoates with focus on downstream processing. International Journal of Biological Macromolecules 2018, 107:762-778.

103. Chanprateep S: Current trends in biodegradable polyhydroxyalkanoates. Journal of Bioscience and Bioengineering 2010, 110:621-632.

104. Kim DY, Kim HW, Chung MG, Rhee YH: Biosynthesis, modification, and biodegradation of bacterial medium-chain-length polyhydroxyalkanoates. Journal of Microbiology 2007, 45:87-97.

105. Reddy CSK, Ghai R, Rashmi, Kalia VC: Polyhydroxyalkanoates: an overview. Bioresource Technology 2003, 87:137-146.

106. Li ZB, Yang J, Loh XJ: Polyhydroxyalkanoates: opening doors for a sustainable future. Npg Asia Materials 2016, 8.

107. Yang JE, Choi SY, Shin JH, Park SJ, Lee SY: Microbial production of lactate-containing polyesters. Microbial biotechnology 2013, 6:621-636.

108. Ren Q, Ruth K, Thony-Meyer L, Zinn M: Enatiomerically pure hydroxycarboxylic acids: current approaches and future perspectives. Applied Microbiology and Biotechnology 2010, 87:41-52.

109. Chiba T, Nakai T: A new synthetic approach to the carbapenem antibiotic Ps-5 from ethyl (S)-3-hydroxybutanoate. Chemistry Letters 1987:2187-2188.

110. Seebach D, Chow HF, Jackson RFW, Sutter MA, Thaisrivongs S, Zimmermann J: (+)-11,11’-di-O-methylelaiophylidene - preparation from elaiophylin and total synthesis from (R)-3-hydroxybutyrate and (S)-malate. Liebigs Annalen Der Chemie 1986:1281-1308.

111. Lee SY, Park SH, Lee Y, Lee SH: Production of chiral and other valuable compounds from microbial polyesters. In Biopolymers Online. Volume 4. Edited by Doi Y, Steinbüchel A. Weinheim, Germany: Wiley-VCH; 2002: 375-387.

112. Seidel W, Seebach D: Crahamimycin A1 synthesis and determination of configuration and chirality. Tetrahedron Letters 1982, 23:159-162.

113. Sutter MA, Seebach D: Synthese von (2E,4E,6S,7R,10E,12E,14S,15R)-6,7,14,15-tetramethyl-8,16-dioxa-2,4,10,12-cyclohexadecatetraen-1,9-dion. Ein modellsystem für elaiophylin. Liebigs Annalen der Chemie 1983, 1983:939-949.

114. Hasegawa J, Nagashima N: Production of chiral β-Hydroxy acids and its application in organic synthesis In Stereoselective Biocatalysis. Edited by Patel RN. New York, USA: Marcel Dekker; 2000: 343-364.

115. Song JJ, Zhang S, Lenz RW, Goodwin S: In vitro polymerization and copolymerization of 3-hydroxypropionyl−CoA with the PHB synthase from Ralstonia eutropha. Biomacromolecules 2000, 1:433-439.

116. Park SH, Lee SH, Lee SY: Preparation of optically active beta-amino acids from microbial polyester polyhydroxyalkanoates. Journal of Chemical Research-S 2001:498-499.


95

117. Chiba T, Nakai T: A synthetic approach to (+)-thienamycin from methyl (R)-3-hydroxybutanoate - a new entry to (3r,4r)-3-[(R)-1-hydroxyethyl]-4-acetoxy-2-azetidinone. Chemistry Letters 1985:651-654.

118. Misra S, Srivastava AK, Raghuwanshi S, Sharma V, Bisen PS: 12 - Medical grade biodegradable polymers: A perspective from gram-positive bacteria. In Fundamental Biomaterials: Polymers. Edited by Thomas S, Balakrishnan P, Sreekala MS: Woodhead Publishing; 2018: 267-286.

119. Blacker AJ, Holt RA: Development of a multi-stage chemical and biological process for an optically active intermediate for an anti-glaucoma drug. In Chirality in Industry II: Developments in the Manufacture and Applications of Optically Active Compounds. Edited by Collins A, Sheldrake G, Crosby J. West Sussex, England: John Wiley & Sons; 1997: 245-262.

120. Ohashi T, J. H: D-(−)-β-hydroxycarboxylic acids as raw materials for captopril and beta lactams. In Chirality in Industry. Edited by Collins A, Sheldrake G, Crosby J. New York: John Wiley & Sons; 1992: 269-278.

121. Iimori T, Shibasaki M: Simple, stereocontrolled synthesis of 1β-methylcarbapenem antibiotics from 3(R)-hydroxybutyric acid. Tetrahedron Letters 1986, 27:2149-2152.

122. Ohashi T, J. H: New preparative methods for optically active β -hydroxycarboxylic acids. In Chirality in Industry. Edited by Collins A, Sheldrake G, Crosby J. New York: John Wiley & Sons; 1992: 249-268.

123. Chen GQ, Wu Q: Microbial production and applications of chiral hydroxyalkanoates. Applied Microbiology and Biotechnology 2005, 67:592-599.

124. Schnurrenberger P, Hungerbühler E, Seebach D: Total synthesis of (+)-colletodiol from (S,S)-tartrate and (R)-3-hydroxybutanoate. Liebigs Annalen der Chemie 1987, 1987:733-744.

125. Hartman AL, Rho JM: The new ketone alphabet soup: BHB, HCA, and HDAC. Epilepsy currents 2014, 14:355-357.

126. Henderson ST: Ketone bodies as a therapeutic for Alzheimer's disease. Neurotherapeutics 2008, 5:470-480.

127. Noyori R, Ohkuma T, Kitamura M, Takaya H, Sayo N, Kumobayashi H, Akutagawa S: Asymmetric hydrogenation of beta-keto carboxylic esters - a practical, purely chemical access to beta-hydroxy esters in high enantiomeric purity. Journal of the American Chemical Society 1987, 109:5856-5858.

128. Jaipuri FA, Jofre MF, Schwarz KA, Pohl NL: Microwave-assisted cleavage of Weinreb amide for carboxylate protection in the synthesis of a (R)-3-hydroxyalkanoic acid. Tetrahedron Letters 2004, 45:4149-4152.

129. Spengler J, Albericio F: Asymmetric synthesis of alpha-unsubstituted beta-hydroxy acids. Current Organic Synthesis 2008, 5:151-161.

130. Seebach D, Beck AK, Breitschuh R, Job K: Direct degradation of the biopolymer poly[(R)-3-hydroxybutyric acid] to (R)-3-hydroxybutanoic acid and its methyl ester. In Organic Syntheses. 2003.

131. Lee Y, Park SH, Lim IT, Han K, Lee SY: Preparation of alkyl (R)-(-)-3-hydroxybutyrate by acidic alcoholysis of poly-(R)-(-)-3-hydroxybutyrate. Enzyme and Microbial Technology 2000, 27:33-36.

132. Calabia BP, Tokiwa Y: A novel PHB depolymerase from a thermophilic Streptomyces Sp. Biotechnology Letters 2006, 28:383-388.

133. Lee SY, Lee Y, Wang F: Chiral compounds from bacterial polyesters: sugars to plastics to fine chemicals. Biotechnology and Bioengineering 1999, 65:363-368.

134. Lee SY, Lee Y: Metabolic engineering of Escherichia coli for production of enantiomerically pure (R)-(-)-hydroxycarboxylic acids. Applied and Environmental Microbiology 2003, 69:3421-3426.

135. Park SJ, Lee SY, Lee Y: Biosynthesis of R-3-hydroxyalkanoic acids by metabolically engineered Escherichia coli. Applied Biochemistry and Biotechnology 2004, 113-116:373-379.

136. Liu Q, Ouyang SP, Chung A, Wu Q, Chen GQ: Microbial production of R-3-hydroxybutyric acid by recombinant E. coli harboring genes of phbA, phbB, and tesB. Applied Microbiology and Biotechnology 2007, 76:811-818.

137. Tseng HC, Martin CH, Nielsen DR, Prather KL: Metabolic engineering of Escherichia coli for enhanced production of (R)- and (S)-3-hydroxybutyrate. Applied and Environmental Microbiology 2009, 75:3137-3145.

138. Gao HJ, Wu Q, Chen GQ: Enhanced production of D-(-)-3-hydroxybutyric acid by recombinant Escherichia coli. FEMS Microbiology Letters 2002, 213:59-65.


96

139. Monod J: Recherches sur la croissance des cultures bacteriennes. Paris: Hermann et Cie; 1942.

140. Villadsen J, Nielsen J, Lidén G: Design of fermentation processes. In Bioreaction Engineering Principles. Boston, MA: Springer US; 2011: 383-458.

141. Schmid RD: Pocket guide to Biotechnology and Genetic Engineering. Weinheim: Wiley-VCH; 2003.

142. Crosby PB: Quality is Free: The Art of Making Quality Certain. New York: McGraw-Hill; 1979. 143. International Organizational for Standardization: ISO 9000. In Quality management

systems - Fundamentals and Vocabulary. Genève, Switzerland, 2015. 144. Bailey JE: Toward a science of metabolic engineering. Science 1991, 252:1668-1675. 145. Saltzberg MA, Byrne AM, Jackson EN, Miller ESJ, Nelson MJ, Tyreus BD, Zhu Q: DuPont:

Biorenewables at E.I. DU Pont DE Nemours & Co. In Industrial Biorenewables. Edited by Domínguez de María P. New Jersey: John Wiley & Sons; 2016: 175-217.

146. The statistics portal: Beer production worldwide from 1998 to 2017 [https://www.statista.com/statistics/270275/worldwide-beer-production/]

147. Ostergaard S, Olsson L, Nielsen J: Metabolic engineering of Saccharomyces cerevisiae. Microbiology and Molecular Biology Reviews 2000, 64:34-50.

148. Klein CJL, Olsson L, Ronnow B, Mikkelsen JD, Nielsen J: Alleviation of glucose repression of maltose metabolism by MIG1 disruption in Saccharomyces cerevisiae. Applied and Environmental Microbiology 1996, 62:4441-4449.

149. Klein CJL, Olsson L, Ronnow B, Mikkelsen JD, Nielsen J: Glucose and maltose metabolism in MIG1-disrupted and MAL-constitutive strains of Saccharomyces cerevisiae. Food Technology and Biotechnology 1997, 35:287-292.

150. Klein CJL, Rasmussen JJ, Ronnow B, Olsson L, Nielsen J: Investigation of the impact of MIG1 and MIG2 on the physiology of Saccharomyces cerevisiae. Journal of Biotechnology 1999, 68:197-212.

151. Olsson L, Larsen ME, Ronnow B, Mikkelsen JD, Nielsen J: Silencing MIG1 in Saccharomyces cerevisiae: Effects of antisense MIG1 expression and MIG1 gene disruption. Applied and Environmental Microbiology 1997, 63:2366-2371.

152. Bailey JE, Shurlati A, Hatzimanikatis V, Lee K, Renner WA, Tsai PS: Inverse metabolic engineering: A strategy for directed genetic engineering of useful phenotypes. Biotechnology and Bioengineering 1996, 52:109-121.

153. Ohnishi J, Mitsuhashi S, Hayashi M, Ando S, Yokoi H, Ochiai K, Ikeda M: A novel methodology employing Corynebacterium glutamicum genome information to generate a new L-lysine-producing mutant. Applied Microbiology and Biotechnology 2002, 58:217-223.

154. Otero JM, Vongsangnak W, Asadollahi MA, Olivares-Hernandes R, Maury J, Farinelli L, Barlocher L, Osteras M, Schalk M, Clark A, Nielsen J: Whole genome sequencing of Saccharomyces cerevisiae: from genotype to phenotype for improved metabolic engineering applications. Bmc Genomics 2010, 11.

155. Reed JL: Genome-scale metabolic modeling and Its application to microbial communities. Chemistry of Microbiomes 2017:85-91.

156. O'Brien EJ, Monk JM, Palsson BO: Using genome-scale models to predict biological capabilities. Cell 2015, 161:971-987.

157. Sander JD, Joung JK: CRISPR-Cas systems for editing, regulating and targeting genomes. Nature Biotechnology 2014, 32:347-355.

158. Shuler ML, Kargi F: Bioprocess engineering: basic concepts. Upper Saddle River, NJ: Prentice Hall; 2002.

159. Otero JM, Panagiotou G, Olsson L: Fueling industrial biotechnology growth with bioethanol. In Biofuels. Edited by Olsson L. Berlin, Heidelberg: Springer Berlin Heidelberg; 2007: 1-40.

160. Lopes ML, Paulillo SCdL, Godoy A, Cherubin RA, Lorenzi MS, Giometti FHC, Bernardino CD, Amorim Neto HBd, Amorim HVd: Ethanol production in Brazil: a bridge between science and industry. Brazilian journal of microbiology : [publication of the Brazilian Society for Microbiology] 2016, 47 Suppl 1:64-76.

161. Russell I: Understanding yeast fundamentals. In The Alcohol Textbook. Edited by Jackes KA, Lyons TP, Kelsall DR. Nottingham: Nottingham University Press; 2003.

162. Walsh RM, Martin PA: Growth of Saccharomyces cerevisiae and Saccharomyces uvarum in a temperature-gradient incubator. Journal of the Institute of Brewing 1977, 83:169-172.


97

163. Della-Bianca BE, Basso TO, Stambuk BU, Basso LC, Gombert AK: What do we know about the yeast strains from the Brazilian fuel ethanol industry? Applied Microbiology and Biotechnology 2013, 97:979-991.

164. Thomas KC, Hynes SH, Jones AM, Ingledew WM: Production of fuel alcohol from wheat by vhg technology - Effect of sugar concentration and fermentation temperature. Applied Biochemistry and Biotechnology 1993, 43:211-226.

165. Jones AM, Ingledew WM: Fuel alcohol production: optimization of temperature for efficient very-high-gravity fermentation. Applied and Environmental Microbiology 1994, 60:1048-1051.

166. Choi JH, Keum KC, Lee SY: Production of recombinant proteins by high cell density culture of Escherichia coli. Chemical Engineering Science 2006, 61:876-885.

167. Choi JH, Lee SY: Secretory and extracellular production of recombinant proteins using Escherichia coli. Applied Microbiology and Biotechnology 2004, 64:625-635.

168. Sandén AM, Bostrom M, Markland K, Larsson G: Solubility and proteolysis of the Zb-MalE and Zb-MalE31 proteins during overproduction in Escherichia coli. Biotechnol Bioeng 2005, 90:239-247.

169. Jorgensen H, Olsson L, Ronnow B, Palmqvist EA: Fed-batch cultivation of baker's yeast followed by nitrogen or carbon starvation: effects on fermentative capacity and content of trehalose and glycogen. Applied Microbiology and Biotechnology 2002, 59:310-317.

170. Reed G, Nagodawithana TW: Baker’s yeast production. In Yeast Technology. Dordrecht: Springer Netherlands; 1990: 261-314.

171. Escherich T: Die Darmbakterien des Neugeborenen und Sauglings. Fortschritte der Medizin 1885, 3:515-522.

172. Escherich T: The Intestinal Bacteria of the Neonate and Breast-Fed Infant. Reviews of Infectious Diseases 1988, 10:1220-1225.

173. Blattner FR, Plunkett G, 3rd, Bloch CA, Perna NT, Burland V, Riley M, Collado-Vides J, Glasner JD, Rode CK, Mayhew GF, et al: The complete genome sequence of Escherichia coli K-12. Science 1997, 277:1453-1462.

174. Russo E: The birth of biotechnology. Nature 2003, 421:456-457. 175. Cooper G: Cells as experimental models. In The Cell: A Molecular Approach. 2nd edition:

Sunderland (MA): Sinauer Associates; 2000. 176. Nordberg Karlsson E, Johansson L, Holst O, Lidén G: Escherichia coli as a well-developed

host for metabolic engineering. In The Metabolic Pathway Engineering Handbook-Fundamentals. Edited by Smolke C. Florida: CRC Press; 2009.

177. Quillaguamán J, Guzmán H, Van-Thuoc D, Hatti-Kaul R: Synthesis and production of polyhydroxyalkanoates by halophiles: current potential and future prospects. Applied Microbiology and Biotechnology 2010, 85:1687-1696.

178. de Lourdes Moreno M, Perez D, Garcia MT, Mellado E: Halophilic bacteria as a source of novel hydrolytic enzymes. Life (Basel) 2013, 3:38-51.

179. Quillaguamán J, Hatti-Kaul R, Mattiasson B, Alvarez MT, Delgado O: Halomonas boliviensis sp. nov., an alkalitolerant, moderate halophile isolated from soil around a Bolivian hypersaline lake. International Journal of Systematic and Evolutionary Microbiology 2004, 54:721-725.

180. Quillaguamán J, Doan-Van T, Guzmán H, Guzmán D, Martin J, Everest A, Hatti-Kaul R: Poly(3-hydroxybutyrate) production by Halomonas boliviensis in fed-batch culture. Applied Microbiology and Biotechnology 2008, 78:227-232.

181. Jarmander J, Belotserkovsky J, Sjöberg G, Guevara-Martinez M, Perez-Zabaleta M, Quillaguaman J, Larsson G: Cultivation strategies for production of (R)-3-hydroxybutyric acid from simultaneous consumption of glucose, xylose and arabinose by Escherichia coli. Microbial Cell Factories 2015, 14:51.

182. Clomburg JM, Vick JE, Blankschien MD, Rodriguez-Moya M, Gonzalez R: A synthetic biology approach to engineer a functional reversal of the beta-oxidation cycle. ACS Synthetic Biology 2012, 1:541-554.

183. Doi Y: Microbial polyesters. New York, N.Y.: Wiley-VCH; 1990. 184. Kim BS, Lee SC, Lee SY, Chang HN, Chang YK, Woo SI: Production of poly(3-

hydroxybutyric acid) by fed-batch culture of Alcaligenes eutrophus with glucose concentration control. Biotechnology and Bioengineering 1994, 43:892-898.

185. Weitzman PDJ: Regulation of citrate synthase activity in Escherichia coli. Biochimica et Biophysica Acta (BBA) - Enzymology and Biological Oxidation 1966, 128:213-215.

186. Luzier WD: Materials derived from biomass/biodegradable materials. Proceedings of the National Academy of Sciences of the United States of America 1992, 89:839-842.


98

187. Yamanè T, Shimizu S: Fed-batch techniques in microbial processes. In Bioprocess Parameter Control Advances in Biochemical Engineering/Biotechnology. Volume 30. Berlin, Heidelberg: Springer; 1984.

188. Wanner U, Egli T: Dynamics of microbial growth and cell composition in batch culture. FEMS Microbiology Letters 1990, 75:19-43.

189. Gottschalk G: Biosynthesis of Escherichia coli cells from glucose. In Bacterial Metabolism. New York: Springer 1985.

190. Reitzer LJ: Ammonia assimilation and the biosynthesis of glutamine, glutamate, aspartate, asparagine, L-alanine, and D-alanine. In Escherichia coli and Salmonella : cellular and molecular biology. Volume 1. Edited by Curtiss R. III IJL, Lin E.C.C., Low B., Magasanik B., Reznikoff W.S., Riley M., Schaechter M., Umbarger E.E. Washington, D.C.: ASM Press; 1996.

191. Chang DE, Shin S, Rhee JS, Pan JG: Acetate metabolism in a pta mutant of Escherichia coli W3110: importance of maintaining acetyl coenzyme A flux for growth and survival. Journal of Bacteriology 1999, 181:6656-6663.

192. Jian J, Zhang SQ, Shi ZY, Wang W, Chen GQ, Wu Q: Production of polyhydroxyalkanoates by Escherichia coli mutants with defected mixed acid fermentation pathways. Applied Microbiology and Biotechnology 2010, 87:2247-2256.

193. Phue JN, Noronha SB, Hattacharyya R, Wolfe AJ, Shiloach J: Glucose metabolism at high density growth of E. coli B and E. coli K: differences in metabolic pathways are responsible for efficient glucose utilization in E. coli B as determined by microarrays and Northern blot analyses. Biotechnology and Bioengineering 2005, 90:805-820.

194. van de Walle M, Shiloach J: Proposed mechanism of acetate accumulation in two recombinant Escherichia coli strains during high density fermentation. Biotechnology and Bioengineering 1998, 57:71-78.

195. Daegelen P, Studier FW, Lenski RE, Cure S, Kim JF: Tracing ancestors and relatives of Escherichia coli B, and the derivation of B strains REL606 and BL21(DE3). Journal of Molecular Biology 2009, 394:634-643.

196. Sanwal BD: Allosteric controls of amphilbolic pathways in bacteria. Bacteriological reviews 1970, 34:20-39.

197. Deutscher J, Francke C, Postma PW: How phosphotransferase system-related protein phosphorylation regulates carbohydrate metabolism in bacteria. Microbiology and Molecular Biology Reviews 2006, 70:939-1031.

198. Bauer AP, Dieckmann SM, Ludwig W, Schleifer KH: Rapid identification of Escherichia coli safety and laboratory strain lineages based on Multiplex-PCR. FEMS Microbiology Letters 2007, 269:36-40.

199. Sandén AM, Prytz I, Tubulekas I, Forberg C, Le H, Hektor A, Neubauer P, Pragai Z, Harwood C, Ward A, et al: Limiting factors in Escherichia coli fed-batch production of recombinant proteins. Biotechnology and Bioengineering 2003, 81:158-166.

200. Kuhnert P, Nicolet J, Frey J: Rapid and accurate identification of Escherichia coli K-12 strains. Applied and Environmental Microbiology 1995, 61:4135-4139.

201. Schneider D, Duperchy E, Depeyrot J, Coursange E, Lenski R, Blot M: Genomic comparisons among Escherichia coli strains B, K-12, and O157:H7 using IS elements as molecular markers. BMC Microbiology 2002, 2:18.

202. Hayashi K, Morooka N, Yamamoto Y, Fujita K, Isono K, Choi S, Ohtsubo E, Baba T, Wanner BL, Mori H, Horiuchi T: Highly accurate genome sequences of Escherichia coli K-12 strains MG1655 and W3110. Molecular Systems Biology 2006, 2:2006.0007.

203. Grenier F, Matteau D, Baby V, Rodrigue S: Complete Genome Sequence of Escherichia coli BW25113. Genome Announcements 2014, 2.

204. Archer CT, Kim JF, Jeong H, Park JH, Vickers CE, Lee SY, Nielsen LK: The genome sequence of E. coli W (ATCC 9637): comparative genome analysis and an improved genome-scale reconstruction of E. coli. BMC Genomics 2011, 12:9.

205. Cantu DC, Chen YF, Reilly PJ: Thioesterases: A new perspective based on their primary and tertiary structures. Protein Science 2010, 19:1281-1295.

206. Hunt MC, Alexson SE: The role acyl-CoA thioesterases play in mediating intracellular lipid metabolism. Progress in Lipid Research 2002, 41:99-130.

207. Tillander V, Alexson SEH, Cohen DE: Deactivating fatty acids: acyl-CoA thioesterase-mediated control of lipid metabolism. Trends in Endocrinology Metababolism 2017, 28:473-484.

208. Liu T, Vora H, Khosla C: Quantitative analysis and engineering of fatty acid biosynthesis in E. coli. Metabolic Engineering 2010, 12:378-386.


99

209. Klinke S, Ren Q, Witholt B, Kessler B: Production of medium-chain-length poly(3-hydroxyalkanoates) from gluconate by recombinant Escherichia coli. Applied and Environmental Microbiology 1999, 65:540-548.

210. Barnes EM, Jr., Wakil SJ: Studies on the mechanism of fatty acid synthesis. XIX. Preparation and general properties of palmityl thioesterase. Journal of Biological Chemistry 1968, 243:2955-2962.

211. Barnes EM, Jr., Wakil SJ, Swindell AC: Purification and properties of a palmityl thioesterase II from Escherichia coli. Journal of Biological Chemistry 1970, 245:3122-3128.

212. Zheng Z, Gong Q, Liu T, Deng Y, Chen JC, Chen GQ: Thioesterase II of Escherichia coli plays an important role in 3-hydroxydecanoic acid production. Applied and Environmental Microbiology 2004, 70:3807-3813.

213. Gulevich AY, Skorokhodova AY, Sukhozhenko AV, Debabov VG: Biosynthesis of enantiopure (S)-3-hydroxybutyrate from glucose through the inverted fatty acid beta-oxidation pathway by metabolically engineered Escherichia coli. Journal of Biotechnology 2017, 244:16-24.

214. Ren Y, Aguirre J, Ntamack AG, Chu C, Schulz H: An alternative pathway of oleate beta-oxidation in Escherichia coli involving the hydrolysis of a dead end intermediate by a thioesterase. Journal of Biological Chemistry 2004, 279:11042-11050.

215. Nie L, Ren Y, Schulz H: Identification and characterization of Escherichia coli thioesterase III that functions in fatty acid beta-oxidation. Biochemistry 2008, 47:7744-7751.

216. Volker AR, Gogerty DS, Bartholomay C, Hennen-Bierwagen T, Zhu H, Bobik TA: Fermentative production of short-chain fatty acids in Escherichia coli. Microbiology 2014, 160:1513-1522.

217. Chen M, Ma X, Chen X, Jiang M, Song H, Guo Z: Identification of a hotdog fold thioesterase involved in the biosynthesis of menaquinone in Escherichia coli. Journal of Bacteriology 2013, 195:2768-2775.

218. Gully D, Bouveret E: A protein network for phospholipid synthesis uncovered by a variant of the tandem affinity purification method in Escherichia coli. Proteomics 2006, 6:282-293.

219. Cho H, Cronan JE, Jr.: Escherichia coli thioesterase I, molecular cloning and sequencing of the structural gene and identification as a periplasmic enzyme. Journal of Biological Chemistry 1993, 268:9238-9245.

220. Jones KL, Kim SW, Keasling JD: Low-copy plasmids can perform as well as or better than high-copy plasmids for metabolic engineering of bacteria. Metabolic Engineering 2000, 2:328-338.

221. Lederberg J, Cavalli LL, Lederberg EM: Sex Compatibility in Escherichia coli. Genetics 1952, 37:720-730.

222. Shizuya H, Birren B, Kim UJ, Mancino V, Slepak T, Tachiiri Y, Simon M: Cloning and stable maintenance of 300-kilobase-pair fragments of human DNA in Escherichia coli using an F-factor-based vector. Proceedings of the National Academy of Sciences of the United States of America 1992, 89:8794-8797.

223. Lander ES, Linton LM, Birren B, Nusbaum C, Zody MC, Baldwin J, Devon K, Dewar K, Doyle M, FitzHugh W, et al: Initial sequencing and analysis of the human genome. Nature 2001, 409:860-921.

224. Venter JC, Adams MD, Myers EW, Li PW, Mural RJ, Sutton GG, Smith HO, Yandell M, Evans CA, Holt RA, et al: The Sequence of the Human Genome. Science 2001, 291:1304-1351.

225. Diaz Ricci JC, Hernandez ME: Plasmid effects on Escherichia coli metabolism. Critical Reviews in Biotechnology 2000, 20:79-108.

226. Jones KL, Keasling JD: Construction and characterization of F plasmid-based expression vectors. Biotechnology and Bioengineering 1998, 59:659-665.

227. Tegel H, Ottosson J, Hober S: Enhancing the protein production levels in Escherichia coli with a strong promoter. Febs Journal 2011, 278:729-739.

228. Studier FW, Rosenberg AH, Dunn JJ, Dubendorff JW: Use of T7 RNA polymerase to direct expression of cloned genes. Methods in Enzymology 1990, 185:60-89.

229. Jones JA, Vernacchio VR, Lachance DM, Lebovich M, Fu L, Shirke AN, Schultz VL, Cress B, Linhardt RJ, Koffas MAG: ePathOptimize: A Combinatorial Approach for Transcriptional Balancing of Metabolic Pathways. Scientific Reports 2015, 5:11301.

230. Bonde MT, Pedersen M, Klausen MS, Jensen SI, Wulff T, Harrison S, Nielsen AT, Herrgård MJ, Sommer MOA: Predictable tuning of protein expression in bacteria. Nature Methods 2016, 13:233.


100

231. Lan EI, Liao JC: ATP drives direct photosynthetic production of 1-butanol in cyanobacteria. Proceedings of the National Academy of Sciences of the United States of America 2012, 109:6018-6023.

232. Egan SE, Fliege R, Tong S, Shibata A, Wolf RE, Jr., Conway T: Molecular characterization of the Entner-Doudoroff pathway in Escherichia coli: sequence analysis and localization of promoters for the edd-eda operon. Journal of Bacteriology 1992, 174:4638-4646.

233. Martinez I, Zhu J, Lin H, Bennett GN, San KY: Replacing Escherichia coli NAD-dependent glyceraldehyde 3-phosphate dehydrogenase (GAPDH) with a NADP-dependent enzyme from Clostridium acetobutylicum facilitates NADPH dependent pathways. Metabolic Engineering 2008, 10:352-359.

234. Verho R, Richard P, Jonson PH, Sundqvist L, Londesborough J, Penttila M: Identification of the first fungal NADP-GAPDH from Kluyveromyces lactis. Biochemistry 2002, 41:13833-13838.

235. Jiang LY, Zhang YY, Li Z, Liu JZ: Metabolic engineering of Corynebacterium glutamicum for increasing the production of L-ornithine by increasing NADPH availability. Journal of Industrial Microbiology and Biotechnology 2013, 40:1143-1151.

236. Guadalupe-Medina V, Wisselink HW, Luttik MA, de Hulster E, Daran J-M, Pronk JT, van Maris AJ: Carbon dioxide fixation by Calvin-Cycle enzymes improves ethanol yield in yeast. Biotechnology for Biofuels 2013, 6:125.

237. Kerfeld CA: Rewiring Escherichia coli for carbon-dioxide fixation. Nature Biotechnology 2016, 34:1035.

Strain- and bioprocess-design strategies to increase ...

Documents

Transcript of Strain- and bioprocess-design strategies to increase ...