2007:12

LICENTIATE T H E SI S

Succinic acid production using metabolically engineered Escherichia coli

Christian Andersson

Luleå University of Technology Department of Chemical Engineering and Geosciences Division of Biochemmical and Chemical Process Engineering

2007:12|: -1757|: -lic -- 07⁄12 -- 

Succinic acid production using metabolically engineered Escherichia coli

Christian Andersson

Division of Biochemical and Chemical Process Engineering

Department of Chemical Engineering and Geosciences

Luleå University of Technoogy

S-971 87 Luleå

February 2007

Abstract

The prospects of peak oil, climate change and the dependency of fossil carbon have urged research and development of production methods for the manufacture of fuels and chemicals from renewable resources (biomass). To date, the primary emphasis has been placed on the replacement of oil for transportation fuels. A highly significant subset of petroleum usage is the production of chemicals, which represents 10-15% of the petroleum usage. White biotechnology, also called industrial biotechnology, is a fast evolving technology with a large potential to have a substantial impact on the industrial production of fuels and chemicals from biomass. This work addresses the issue of chemical production by investigating the production of bio-based succinic acid, which can be used in a wide range of applications to replace petroleum based chemicals. Succinic acid can be produced by fermentation of sugar by a number of organisms; one is Escherichia coli (E. coli). It is known that E. coli under anaerobic conditions produces a mixture of organic acids. In order to obtain a cost-effective production it is necessary to metabolically engineer the organism to produce succinic acid in greater yield than the other acids. In the current work, E. coli mutant AFP184 was used. AFP184 originates from a near wild type strain, the C600 (ATCC 23724), which can ferment both five and six carbon sugars and has mutations in the glucose specific phosphotransferase system (ptsG), the pyruvate formate lyase system (pfl) and in the fermentative lactate dehydrogenase system (ldh). The previous studies using different organisms have all used cultivation mediums supplemented to some degree with different nutrients like biotin, thiamine and yeast extract. In order to apply the technology to large scale, production must be cost-effective and it is important to minimise the use of additional supplements.

The first part of this work aimed to investigate the fermentation characteristics of AFP184 in a medium consisting of corn steep liquor, inorganic salts and different sugar sources without supplementation of other additional nutrients. It addresses questions regarding the effect of different sugars on succinic acid kinetics and yields in an industrially relevant medium. In order to gain a sustainable production of succinic acid from biomass feedstocks (sugar from biomass) it is important to investigate how well the organism can utilise different sugars in the biomass. The sugars studied were sucrose, glucose, fructose, xylose and equal mixtures of glucose- fructose and glucose-xylose at a total initial sugar concentration of 100 g L-1. AFP184 was able to utilise all sugars and sugar combinations except sucrose for biomass generation and succinate production. Using glucose resulted in the highest yield, 0.83 (g succinic acid per g sugar consumed anaerobically). Fructose resulted in a yield of 0.66 and xylose of 0.5. Using a high

i initial sugar concentration made it possible to obtain volumetric productivities of almost 3 g L- 1h-1, which is above estimated values for feasible economic production. Succinic acid production ceased at final concentrations greater than 40 g L-1. In order to further increase succinic acid concentrations, this inhibitory effect was studied in the second part of the present work. The inhibitory effects can be two-fold including pH-based inhibition and an anion specific effect on metabolism. It has been reported that high concentrations of ammonia inhibit E. coli growth and damage cell membranes. In order to limit toxic and inhibitory effects different neutralising agents were tested. First the use of NH4OH was optimised with respect to fermentation pH and it was found that the best results were obtained at pH 6.5-6.7.

Optimal pH was then used with NaOH, KOH, and Na2CO3 as neutralising agents and it was shown that NaOH, KOH, and Na2CO3 neutralised fermentations could reach succinic acid concentrations of 69 and 61 and 78 g L-1 respectively without any significant decrease in succinic acid productivity. It was observed that cells lost viability during the cause anaerobic phase. It resulted in decreasing succinic acid productivities. It is believed that the viability decrease is a combined effect of organic acids concentration and the osmolarity of the medium. The work done in this thesis is aimed towards increasing the economical feasibility of a biochemical succinic acid production.

Keywords: Industrial biotechnology, Fermentation, Succinic acid, Escherichia coli, Productivity

ii Acknowledgements

First of all I would like to thank my supervisor Professor Kris A. Berglund. Your constant good mood and undying enthusiasm is close to contagious. You are never too tired for a discussion be it politics or fermentation technology. Thank you for including me in your team.

My deepest gratitude and thankfulness I would like to extend to my secondary supervisor Dr. Ulrika Rova. You taught me the basics of microbiology. Without you there would be no thesis. Your care and resourcefulness are the foundation of our division. Thank you for all the work you have put in.

I extend a special thanks to my master thesis worker and fermentation engineer Jonas Helmerius. You came when I needed you the most. It has been a pleasure supervising you. Thank you for all your help.

I am also grateful to Josefine Enman, Magnus Sjöblom, David Hodge, and all my friends and colleges at the Department of Chemical Engineering and Geosciences. Working with you guys is great fun. My earlier master thesis workers Andreas Lennartsson, Ksenia Bolotova, and Mario Winkler are all acknowledged. You all did great jobs. I also thank all my students. Our discussions and your curiosity help me to develop.

Min närmste vän Ivan Kaic och min bror Fredrik Andersson, ni är speciellt tackade. Ert stöd och er vänskap är ovärderlig.

Mina föräldrar Jarl och Margaretha Andersson. Jag har inte ord att beskriva vad ni betyder. Ni är mina förebilder. Mitt rättesnöre.

Finally, my love Antonina Lobanova. You came to me. I still cannot believe it is true. You have changed everything. ə ɥɸɛɥɸ ɬɟɛɹ.

Luleå 2007, Christian Andersson

iii iv List of papers

This thesis is based on the work contained in the following papers, referred to in the text by Roman numerals.

I Effects of different carbon sources on the production of succinic acid using metabolically engineered Escherichia coli Christian Andersson, David Hodge, Kris A. Berglund, and Ulrika Rova Biotechnology Progress, In press

II Impact of neutralising agent and pH on the succinic acid production by metabolically engineered Escherichia coli in dual-phase fermentations Christian Andersson, Jonas Helmerius, David Hodge, Kris A. Berglund, and Ulrika Rova Manuscript

v vi Contents

INTRODUCTION ...... 1

PRODUCTION OF BIO-BASED FUELS AND CHEMICALS ...... 2

THE INTEGRATED FOREST BIOREFINERY (IFBR) ...... 4

CURRENT LEVEL OF SUCCINIC ACID PRODUCTION...... 7

THEORY ...... 11

BASIS FOR THE EXPERIMENTAL WORK ...... 11

SUGAR METABOLISM IN AFP184 ...... 11 Glucose...... 12 Fructose ...... 16 Xylose ...... 16

ACID RESISTANCE, INHIBITION, AND TOXICITY IN ESCHERICHIA COLI...... 16 Energy generation and the chemiosmotic theory ...... 16 Background to pH homeostasis...... 17 Organic acid, metabolic uncoupling and anion accumulation...... 18 Acid resistance systems...... 20

OSMOTIC STRESS IN ESCHERICHIA COLI ...... 23 Osmotic pressure, osmolarity and turgor pressure ...... 23 The role of compatible solutes in the response of Escherichia coli to osmotic stress ...... 25

RESULTS AND DISCUSSION ...... 29

SUGAR UTILISATION ...... 29

IMPACT OF THE NEUTRALISING AGENT ...... 33

CONCLUSIONS ...... 37

FUTURE WORK ...... 39

REFERENCES ...... 41

vii viii Introduction

The prospects of peak oil, climate change and the dependency of fossil carbon have urged research and development of production methods for the manufacture of fuels and chemicals from renewable resources (biomass). The main focus so far has been replacement of oil for transportation fuels, but the utilisation of petroleum for production of chemicals could represent 10-15% of the petroleum usage.

White biotechnology, also called industrial biotechnology, is a fast evolving technology with a large potential to have a substantial impact on the industrial production of fuels and chemicals from biomass. By definition, the technology uses the properties of living organisms such as yeast, moulds, bacteria and plants to make products from renewable resources. Since biotechnology processes utilise living systems the reactions involved generally occurs under mild conditions with high product specificity. One well-known example of white biotechnology is glucose fermentation by the yeast Saccharomyces cerevisiae for the production of ethanol.

Conventional, non-biological processes can be replaced by biochemical conversion of biomass resulting in reduction of greenhouse gas emissions and energy usage for the production of bio- degradable fuels and chemicals with reduced formation of undesirable/harmful by-products. As a first step, target fermentations producing chemicals that can compete with fossil based alternatives must be identified. Consequently, it is essential to develop fermentations that produce building blocks, i.e. molecules that can be converted into a number of high-value chemicals or materials. In a report from the U.S. Department of Energy (USDOE), succinic acid was considered as one of twelve top chemical building blocks manufactured from biomass (1). In addition to its own use as a food ingredient and chemical, succinic acid can be used to derive a wide range of products including: diesel fuel oxygenates for particulate emission reduction; biodegradable, glycol free, low corrosion deicing chemicals for airport runways; glycol free engine coolants; polybutylene succinate (PBS), a biodegradable polymer that can replace polyethylene and polypropylene; and non-toxic, environmentally safe solvents that can replace chlorinated and other VOC emitting solvents (2). Therefore bio-based production of succinic acid offers the opportunity for a widespread replacement of fossil fuel based chemicals. The key for a successful production of succinic acid based products is the development of an

1 economically feasible, efficient manufacturing process. The present work addresses different aspects of succinic acid production by fermentation using metabolically engineered Escherichia coli (E. coli).

Production of bio-based fuels and chemicals

It has been estimated that the annual global production of plant biomass exceeds 100×1012 kg dry matter (3). The generated biomass must however be shared between many interests, for example human food, forage crops and raw material for industry. With the depletion of oil- reserves, increased greenhouse-gas emissions and an interest in using locally available raw materials, attention has shifted from fossil fuel based technologies to alternative technologies using sun, wind, water and biomass to produce energy and chemicals. Even if the current interest for bio-based fuels and chemicals production increases the demand for usable biomass, the supply of biomass is certainly enough to sustain the present biomass consumers and the growing bio-based fuel and chemical sectors (3). Although today’s prices for biomass (e.g. agricultural residues and forest products) are much lower than the price of crude oil,(3) bio- based produced fuels and chemicals have problems competing with their oil-based counterparts. The main reasons for the lack of viability of the bio-based production methods are:

x High processing costs; the costs of converting the cheap biological feedstocks into molecules that can be further processed to fuels and chemicals are high compared to petrochemical alternatives.

x Low raw material usage; only parts of the raw material is used to produce the desired products.

x Lack of process integration; while using the same raw material or different parts of the same feedstock, the production sites are often not located close to each other.

x High capital costs; in order to produce bio-based products investments in new plants have to be made.

The situation in the petrochemical industry is the reverse. Even though the feedstock (crude oil) is traded at a higher price than agricultural residues or forest materials, the processing costs are lower due to well implemented and mature processes and technologies. The raw material

2 usage is very high and different processes refining various parts of the crude oil into high-value products are linked together in an oil refinery (3). Oil refineries are very large chemical plants in which the investments were done decades ago. Oil refineries offer a wide spectrum of products from a single feedstock, and if the biotechnical process industries are to be competitive they need to develop in the same direction. The solution is an integrated production of many bio-based products by what is called a biorefinery.

The main concept of an integrated biorefinery is efficient conversion of all feedstock components into value-added products by thermochemical or biochemical processes (Fig. 1). By product integration, the use of all feedstock components will be maximised including the use of by-products and waste streams. Today there exists a number of biorefineries or potential biorefineries like corn dry and wet milling plants and pulp and paper mills (3, 4). Biorefineries typically use regionally available biomass; in the US biorefineries are constructed around corn wet, and dry mills and in Brazil around sugar cane mills. In Sweden the main source of biomass is the forest. The pulp and paper industry is well established, has the infrastructure and labour force to handle and process forest biomass. For Sweden it would be a suitable path to generate a biorefinery around a pulp and paper mill.

Grains Agricultural residues Forest biomass Municipal solid waste

Feedstocks

Biochemical Thermo-chemical

Processing

Physical Chemical

Products

Fuels Chemicals Materials Foods

Figure 1. Principal sketch of the biorefinery concept.

3 The integrated forest biorefinery (IFBR)

Pulp and paper mills process forestry materials into value added products; pulp, paper, lignin, energy and could therefore already be considered as an example of a biorefinery. Besides the primary purpose of converting cellulose to pulp, other products from alkaline pulping processes include extractives such as tall oil and the generation of steam and electricity. The main concept of the integrated forest product refinery is continuous production of pulp and paper products in addition to manufacturing of fuels and chemicals from the raw material or waste streams that is currently underutilised.

The hemicellulose and lignin constitutes approximately 40-50% of the dry weight of wood (5). In the present kraft pulp mill hemicelluloses and lignin are combusted in the recovery boiler to generate steam and recover pulping chemicals. Based on a heating value of 13.6 MJ kg-1 the hemicellulose value has been calculated to be $50 per oven dry metric ton (ODMT) (6). Converting the biomass into bioproducts valued at $2000 ODMT-1 or more provides a great opportunity for a kraft mill to expand its business (6, 7). Due to the accumulating quantities, lignocellulosic materials could potentially provide a large and less expensive sugar source for bio-based industries in countries with large amount of wood compared to sugars from starch.

In the kraft pulp process (also called sulphate process or alkaline pulping) wood chips are boiled together with sodium hydroxide and sodium sulphide in large digesters. The aim is to separate lignin and hemicellulose from the cellulose fibres without degrading the fibres. After separating the fibres from the spent pulping liquor (black liquor) the fibres are processed into paper pulp and the liquor is concentrated by multi-effect evaporation and burned in the recovery boiler to generate steam and regenerate the pulping chemicals. If a present day kraft mill (Fig. 2) should be converted to an IFBR the hemicellulose fraction of the wood chips must be liberated and separated from the wood chips before pulping.

4 Evaporation

Black liquor

Steam Wood chips Flue gas Pulping chemicals

Digester

Recovery boiler Pulp

Caustisation

Figure 2. Recovery of pulping chemicals in the kraft process.

The hemicellulose can be separated from the wood chips prior to pulping by leaching the wood chips with a hot alkaline solution. The hemicellulose in softwoods is mainly acetyl- galactoglucomannan and during hot alkaline treatment the glucomannan is degraded by the peeling reaction (6, 8, 9). In hardwoods on the other hand, the main hemicellulose form is glucoronoxylan which is dissolved in oligomeric form when treated in the same way, and hot alkaline extraction would thus be a suitable method for extraction of hemicellulose from hardwood (6, 8, 9). If pure water was used as extractant at high temperature the acetic acid released from the hemicellulose would create acidic conditions and may cause unwanted cellulose degradation due to acid hydrolysis at the high temperatures used. Extracting the hemicellulose from the wood chips has a number of positive effects besides opening up possibilities for bio-based chemicals productions. It will increase the delignification rate in the digester, reduce the alkali consumption and thus also reduce the amount of produced black liquor (6). The most important question that has to be addressed is whether the pulp quality is negatively affected.

5 Further processing of the black liquor could be done by precipitating the lignin under acidic conditions. The lignin could then be used to produce carbon fibres, fuels and other chemicals as outlined in Figure 3.

Fischer-Tropsch fuels Methanol Syngas Hydrogen Ethanol Wood Pulp and paper

Carbon fibers Biorefinery Lignin Binders Fuel

Chemicals Power

Green chemicals Hemicellulose Bioplastics Ethanol

Figure 3. Products from an IFBR.

Present kraft mills use Tomlinson recovery boilers for regenerating pulping chemicals and producing energy and steam for other mill activities. The cost of the generated energy, due to low efficiency, is rather high (10). A significant increase in electricity production at a mill could be provided by black liquor gasification, where the black liquor after evaporation is gasified, generating green liquor containing the pulping chemicals and hot synthesis gas (10). A simplified block-flow diagram of a chemical recovery system for a kraft process employing hemicellulose extraction, lignin precipitation and black-liquor gasification is presented in Figure 4. One important product from black liquor gasification is synthesis gas. The produced synthesis gas is cooled in a gas cooler generating low and medium pressure steam (Fig. 4). The syngas can after cooling be further processed into liquid fuels or chemicals at a price comparable to that of gasoline (7). The next step before industrial implementation is to demonstrate the black liquor gasification technique at a commercial scale.

6 Evaporation Alkaline solution Water Syngas

Black liquor

Extractor Wood LP Steam chips Extracted Gas hemicellulose Acid cooler Lignin MP Steam precipitation Gasifier Digester Black Pulping liquor chemicals Lignin

Pulp

Recovered Cooling water Pulping Caustisation chemicals Green liquor

Figure 4. Chemical recycling in a kraft mill with hemicellulose extraction, lignin precipitation and black liquor gasification.

Current level of succinic acid production

Succinic acid, a dicarboxylic acid with the molecular formula C4H6O4, was first discovered in 1546 by Georgius Agricola when he dry distilled amber. Currently, succinic acid is manufactured through oxidation of n-butane or benzene followed hydrolysis and finally dehydrogenation. It is used in the production of esters, plastics and dyes. Succinic acid could also be produced using biochemical conversion of biomass by fungal or bacterial fermentations. Increased markets for bio-based production of succinic acid are expected to come from the synthesis of biodegradable polymers; polybutylene succinate (PBS) and polyamides (Nylon®x,4) and various green solvents. Different reaction pathways for succinic acid applications from fermentation are outlined in Figure 5.

7 New applications Fermentation Deicing chemicals Cleaners Runway Food and feed Commercial, additives markets residential, industrial Succinic Salts

Chelating agents Detergent additives Separations Polymerization Acidulants Water treatment Food and beverage Corrosion Agicultural inhibitors chemicals Succinic Acid Thermoset resins Herbicides Defoliants Adjuvants Esterification Dehydration

Diesel fuel additives Solvents/Surfactants Processing Succinate Esters Succinic Anhydride Cleaning Paint industry Cosmetics Condensation Dyes/pigments Polymer additives Specialty Chemicals Itaconic Acid Latex polymers Polymers

Figure 5. Value-added products that can be derived from succinic acid.

Production of succinic acid has been demonstrated in a number of organisms including Bacteroides ruminicola and Bacteroides amylophilus, Anaerobiospirillum succiniciproducens, Actinobacillus succinogenes, Mannheimia succiniciproducens, Corynebacterium glutamicum, and Escherichia coli (11-23). The work done so far have all used media supplemented to some degree with high-cost nutrient sources like tryptone, peptone and yeast extract.

Manufacturing costs of succinic acid are affected by succinic acid productivity and yield, raw material costs and utilisation and recovery methods. Hence, in order to develop a bio-based industrial production of succinic acid, three main issues need to be addressed. First of all a low cost medium must be used. Second, the producing organism must be able to utilise a wide range of sugar feedstocks and produce succinic acid in good yields in order to make use of the cheapest available raw material. The third and most important factor for economically feasible succinic acid economy is the volumetric productivity. To make succinic acid production feasible volumetric productivities above 2.5 g L-1h-1 will be necessary (1). In a recent review by Song and Lee the most promising of these organisms, including E. coli mutants, were compared (24).

8 The most investigated organism is A. succiniciproducens, a strict anaerobe. A succiniciproducens has been characterised in a number of studies both with regards to medium composition (25-29) and processing conditions (30, 31). One of the achievements with A. succiniciproducens was conversion of wood hydrolysates into succinate with a mass yield of 0.88 gram succinate per gram glucose demonstrated by Lee et al. (32). The main drawback wit the organism is that it does not seem to tolerate succinic acid concentrations higher than 30-35 g L-1.

Recently a facultative anaerobe, M. succiniciproducens, was isolated from bovine rumen and shown to produce succinate as its main fermentation product at yields in the order of 0.7 gram succinate per gram sugar consumed.(14). This organism has been demonstrated to ferment both glucose and xylose from wood hydrolysates (16). The highest succinic acid concentration obtained in M. succiniciproducens fermentations is 52.4 g L-1 (21). The concentration was achieved using LB growth medium. M. succiniciproducens seems promising, and future work should focus on development of mutants able to produce succinate in higher yields.

A. succinogenes is also a facultative anaerobe inhabiting the bovine rumen. The organism has been shown to produce succinic acid to impressive concentrations (105.8 g L-1) in good yields (approximately 0.85 gram per gram glucose) (33-35). The fermentations that have reached

-1 succinate concentrations of more than 80 g L have been neutralised with MgCO3. When -1 using NH4OH or sodium alkali final concentrations were in the range of 60 g L , which is lower than the concentration some E. coli strains have been shown to produce (36). Finally, the fermentations have been carried out in either vial flasks or 1 L reactors. Before applying the organism to large-scale production, it is necessary to demonstrate the process at a production scale achieving the excellent concentrations obtained in laboratory experiments.

Under anaerobic conditions E. coli is known to produce a mixture of organic acids and ethanol (37). Typical yields from such fermentations are 0.8 moles ethanol, 1.2 moles formic acid, 0.1- 0.2 moles lactic acid, and 0.3-0.4 moles succinic acid per mol glucose consumed. If the objective is to produce succinic acid, the yield of succinic acid relative the other acids must be increased. In the late 1990s USDOE initiated the Alternative Feedstock Program (AFP) with the aim to develop a number of metabolically engineered E. coli strains with increased succinic acid production. The two most interesting mutants developed by the program were AFP111 and AFP184 (13, 22, 38). AFP111 is a spontaneous mutant with mutations in the glucose specific phosphotransferase system (ptsG), the pyruvate formate lyase system (pfl) and in the

9 fermentative lactate dehydrogenase system (ldh) (13). The mutations resulted in increased succinic acid yields (1 mol succinic acid per mol glucose) (22). AFP184 is a metabolically engineered strain where the three mutations described above were deliberately inserted into the E. coli strain C600 (ATCC 23724), which can ferment both five and six carbon sugars and has strong growth characteristics (38). Beside these organisms, a number of different E. coli mutants have been developed and tested for succinate production including pioneering work to produce succinate under aerobic conditions (17, 18, 20, 23, 39-43). In order to improve succinate yield other studies used recombinant AFP111 overexpressing heterologous pyruvate carboxylase in complex media with the main components tryptone (20 g L-1) and yeast extract (10 g L-1), which resulted in mass yields of 1.10 and productivities of 1.3 g L-1h-1 (15, 36).

10 Theory

Basis for the experimental work

The specific experimental details in this work are elucidated in the papers I and II. Here an overview of the conditions for the fermentation procedures is presented.

The medium used in this thesis was developed to be cost effective and relevant for use in industrial production of succinic acid. Corn steep liquor is a by-product from the corn wet- milling industry and is an inexpensive source of vitamins and trace elements as opposed to yeast extract and peptone. The composition of corn steep liquor can vary widely mainly depending on the variables involved in starch processing and the type and condition of the corn, but averages can be found in literature (44-46). In the present work AFP184 was used in 1-10 L laboratory fermentations using a medium based on corn steep liquor (50% solids) supplemented with inorganic salts and a sugar solution of glucose, fructose, or xylose or mixtures of glucose/fructose or glucose/xylose. The initial sugar concentration in the fermentations was 100 g L-1.

The fermentations were conducted in dual-phase mode, where each fermentation consisted of an aerobic phase and an anaerobic phase. This way of running fermentations takes advantage of E. coli being a facultative anaerobe; i.e. can grow both in the prescence or absence of oxygen. Aerobic cultivation of E. coli results in high growth rates and a fast increase in cell concentration. When the desired cell concentration is reached, the fermentation is shifted to anaerobic conditions by shutting of the air supply to the reactor and instead sparging the medium with CO2. The cells grown aerobically now consume sugar anaerobically and produce a mixture of organic acids, with succinic acid as the main product for AFP184.

Sugar metabolism in AFP184

In E. coli different sugars are metabolised in more or less different ways depending on structural similarities between the sugars and the properties of the enzyme systems responsible for uptake and degradation. As outlined above biological succinic acid production must be able to utilise

11 different sugars. The present work deals with the glucose, fructose, and xylose metabolism of AFP184. Glucose and xylose are the most abundant sugars in lignocellulosic biomass, while fructose and glucose are the components of sucrose, a widely available disaccharide. In the following section the metabolism of each of the sugars will be described briefly. The impacts of the mutations in AFP184 will be discussed in the next section.

Glucose

The first step in the glucose metabolism of E. coli is glycolysis (also called Embdern-Meyerhof- Parnas (EMP) pathway). In glycolysis one mole of glucose is taken up by the cell, phosphorylated and degraded into 2 moles of pyruvate generating small amounts of energy (ATP, NADH). In the energy metabolism of E. coli the fate of pyruvate depends on the environmental conditions. In the presence of oxygen, pyruvate molecules enter the citric acid cycle (TCA-cycle or Kreb’s cycle) where together with oxidative phosphorylation the main part of the cellular energy is produced. Under anaerobic conditions E. coli will instead undergo mixed acid fermentation and produce a mixture of organic acids and ethanol (37).

However, when metabolising glucose under anaerobic conditions the mutations in the pfl and ldh genes significantly influence the ratio of the products formed. It has been long known that E. coli under anaerobic conditions produced a mixture of organic acids and ethanol (37). The mutations in AFP184 direct the metabolic fluxes so that the only end-products that accumulate in any significant amounts are succinic acid and acetic acid. The mutations in the ldh and pfl genes limit the generation of the by-products lactic acid, formic acid, ethanol and acetic acid, although a small amount of pyruvic acid also accumulates. The sugar metabolism of AFP184 for glucose, fructose, and xylose is outlined in Figure 6.

12 Xylose

Glucose Xylulose PEP ATP ATP Pyruvate ADP ADP Glucose 6-P Xylulose 5-P Ribose 5-P

PEP Pyruvate

Fructose Fructose 6-P ATP

ADP Glyceraldehyde 3-P Sedoheptulose 7-P Glyceraldehyde 3-P

2H, ATP

PEP CO2 Erythrose 4-P Fructose 6-P ATP 2H, CO2 ATP Pyruvate Acetate 2H ADP 2H, CO2 Glyceraldehyde 3-P

Lactate Formate 4H Oxaloacetate Acetyl-CoA 2H

Malate Glyoxylate Citrate Ethanol

Fumarate Isocitrate

2H Succinate

Figure 6. Anaerobic glucose, fructose and xylose metabolism of AFP184 for maximal succinate production. Note that some metabolites are excluded. The reactions blocked by the mutations in AFP184 are indicated by crosses.

The main product from anaerobic fermentations by AFP184 is succinic acid. It is also the aim of the present work to investigate and improve the succinic acid production using strain

AFP184. Anaerobic succinic acid production by AFP184 requires CO2. The enzyme phosphoenolpyruvate carboxylase directs the carbon flow from PEP to oxaloacetic acid by fixing CO2. Oxaloacetate is then reduced to malate, fumarate, and finally succinate via the reductive arm of the TCA-cycle. From a carbon balance a yield of 2 moles of succinic acid for

13 every mole of glucose consumed is obtained. Carrying out a redox balance shows that ~15 % of the carbon must be committed to the glyoxylate shunt to generate the necessary amount of reducing equivalents (NADH). The maximum theoretical yield of succinate from one mole of glucose is thus 1.71 (1.12 gram per gram glucose). The anaerobic activity of the glyoxylate shunt has been demonstrated for AFP111 after aerobically inducing it (15). Based on the similarities between AFP184 and AFP111 it is assumed in the present study that glyoxylate shunt also is active in AFP184. The third mutation in AFP184 affects the uptake and phosphorylation of glucose.

E. coli has two hydrophobic membranes surrounding its cytoplasm, the cytoplasmic (or inner) membrane and the outer membrane. The outer membrane contains special proteins, porins, which allow glucose and other small solutes to diffuse through the membrane into the periplasm (the volume between the inner and outer membrane). The two main porins for glucose uptake during concentrations above 0.2 mM are OmpC and OmpF (47), as described under the section Osmotic stress in Escherichia coli. Normally in E. coli glucose is taken up and phosphorylated by the phosphotransferase system (PTS), which is a system of transport proteins (48, 49) that internalise glucose (Fig. 7). The diffusion of glucose across the outer membrane into the cytoplasm is a passive process and the rate of transport is determined by the activity of the transport system.

Cytoplasm Periplasm Outside Glucose 6-P cell

IICDMan IIABMan OmpC

P Glucose Glucose PEP EI HPr

IIBGlc Pyruvate EI HPr P IICGlc IIAGlc P OmpF Glucose 6-P

+ Glucokinase + H Glucose 6-P Glucose + H GalP

ADP ATP ADP ATP Glucokinase Glucose 6-P Glucose MglC MglA MglB

ADP ATP

Figure 7. Glucose uptake systems in E. coli.

14 The process starts with Enzyme I (EI) taking a phosphate group from glycolytic intermediate phosphoenolpyruvate (PEP), generating phosphorylated EI and pyruvate. EI is said to be non- sugar specific, meaning that its activity is not linked to a specific sugar substrate, instead it will assist in the uptake of for example both fructose, glucose, and mannose. The phosphate group is then transported to a sugar specific enzyme II complex via the phosphohistidine carrier protein (HPr). In E. coli there are 21 different enyme II complexes (48). The ones responsible for glucose uptake and phosphorylation are IIGlc and IIMan where the later complex has lower glucose uptake rate than IIGlc. The IIGlc complex is made up of the sugar-specific enzymes, IIAGlc and IIBCGlc . IIAGlc is a soluble enzyme and IIBCGlc is an integral membrane protein permease.

The gene encoding the glucose-specific permease IIBCGlc normally represses the genes encoding the enzymes in the IIMan complex. AFP184 was metabolically engineered to have the glucose-specific permease knocked out, resulting in an organism where the genes for the IIMan complex is no longer repressed and glucose uptake can commence through the complexes IIABMan and IICDMan. The affinity and capacity of the mannose permease system is somewhat lower for glucose than the glucose specific uptake system and lower growth rates has been reported in mutants relying on the IIMan complex (50).

Two other systems that are important for glucose internalisation in E. coli, which lacks a functional glucose-specific permease, are the GalP and Mgl systems. GalP is a galactose:H+ symport that has the ability to transfer both galactose and glucose into the cytoplasm. However the rate for internalisation is substantially lower than by the PTS enzymes. The Mgl system consists of a galactose/glucose periplasmic binding protein (MglB), an integral membrane transporter protein (MglC), and an ATP-binding protein (MglA) (47). Glucose imported via GalP or Mgl are not phosphorylated. The phosphorylation necessary before entering glycolysis is carried out by the cytoplasmic enzyme glucokinase. Instead of using PEP as a phosphate donor, this phosphorylation is ATP dependent. Uptake and phosphorylation by glucokinase has been reported as slower than when carried out by the PTS (51). Once phosphorylated the aerobic metabolism of AFP184 is essentially the same as in wild-type E. coli.

15 Fructose

The fructose metabolism of AFP184 (Fig. 6) is similar to glucose, but with the main difference that the PTS for fructose is intact and fructose is thus internalised and phosphorylated by the fructose specific PTS and not by the mannose PTS or glucokinase (52-54). The result is a higher aerobic growth rate, but a lower succinate yield since instead of generating succinate more PEP is converted to pyruvate during uptake and phosphorylation of fructose . An electron balance for succinate production by fructose metabolism gives a maximum theoretical molar yield of 1.20 (mass yield 0.79).

Xylose

Xylose metabolism differs from the two other sugars since xylose is not a PTS substrate. Uptake of xylose is instead governed by chemiosmotic effects (55, 56). An initial increase of pH has been reported as xylose is consumed by E. coli (56). This increase was interpreted as an influx of protons (or efflux of hydroxyl groups) accompanying a protonmotive force driven transport of xylose into the cells Once inside the cell, xylose is isomerised by xylose isomerase to xylulose which is phosphorylated by xylulose kinase to xylulose 5-phosphate (55, 57). Xylulose 5-phosphate then enters the pentose phosphate pathway and can after conversion to either fructose 6-phosphate or glyceraldehyde 3-phosphate enter the glycolytic pathway (Fig. 6). The maximum theoretical yield of succinate from xylose calculated by a redox balance is 1.43 moles per mole xylose (mass yield 1.12).

Acid resistance, inhibition, and toxicity in Escherichia coli

It is well known that a reduction in the pH of foods helps to prevent microbial contamination and spoilage (58). This preservation utilises the impact of three major factors influencing cell physiology, function and viability during acid conditions. These are energy generation, intracellular pH homeostasis and protection of proteins and DNA (59).

Energy generation and the chemiosmotic theory

At the time when Mitchell proposed his chemiosmotic theory it was known that oxidation of a substrate is coupled to the phosphorylation of ADP (60). The chemiosmotic theory outlines

16 that a proton concentration gradient across the cytoplasmic membrane is generated and used as a pool of energy for ATP synthesis (60). An essential component for generating the protonmotive force is the relative impermeability of the cytoplasmic membrane to protons. Protons are channeled into the cell by active transport via the membrane bound enzyme ATP synthase. This enzyme concomitantly catalyses the phosphorylation of ADP from the energy obtained by the movement of protons along the proton gradient.

The energy stored in the proton concentration gradient is called the protonmotive force (PMF) and is formed by two contributors; the difference in proton concentration and the difference in charge between the inner and outer sides of the membrane. The difference in proton concentration across the membrane is simply the difference in pH between the cytoplasm and the surrounding medium (ǻpH). The charge difference is called membrane potential (ǻȌ) and is given in V or mV.

Using R = 8.315 J K-1 mol-1, F = 96.48 kJ V-1 mol-1, and a standard temperature of 298.15 K the protonmotive force can be written as:

PMF '<  0.058'pH (1)

The protonmotive force in E. coli actively undergoing cell division, is in the range of -140 to - 200 mV (61). If ǻpH increases the membrane potential is reduced to keep a stable PMF. A very large PMF would create a massive pull on extracellular protons and result in increased influx and acidification of the cytoplasm.

Background to pH homeostasis

The cytoplasmic pH plays an important role in bacterial cell physiology. The membrane surrounding the cytoplasm is more or less impermeable to protons that instead are taken up by the cells via active transport systems, in particular ATP synthase. There are two main factors that affect the pH homeostasis system in E. coli; the pH of the cultivation medium and the concentraion of organic acids produced during fermentation. In the advent of a reduction of the extracellular pH, bacteria respond by changing the activity of the ion transport systems responsible for bringing protons into the cytoplasm (62). E. coli is a neutrophile and thus grows best at near neutral pH. The cytoplasm usually has a pH somewhat above that of the

17 surrounding medium. For growth of E. coli in a near neutral medium the internal pH is in the range of 7.4-7.9 (63, 64). When E. coli is subjected to extreme acid stress the pH gradient across the membrane increases and creates a strong influx force of protons. The cytoplasmic membrane of E. coli are close to impermeable to protons, but at large ǻpH the net influx of protons increases nevertheless. It is hypothesised that protons travel across the membranes by using protein channels or damaged parts of the lipid bilayers (61). It has been shown that E. coli cannot maintain a ǻpH of more than 2 pH units (61). The result is more or less a severe acidification of the cytoplasm. At low cytoplasmic pH, the activity and function of cellular enzymes will be affected. E. coli has been demonstrated to survive acid stress down to pH 2 for several hours (63, 65). However, due to improper enzyme function, protein denaturation and damage on the purine bases of DNA, the low pH environment will eventually kill the cells.

The second source of cytoplasmic pH perturbations, weak acids, works differently. Organic acids, like acetate, lactate, and succinate are all produced during anaerobic fermentation of e.g. glucose. Those acids have pKa values in the range of 4 and in a cultivation medium at pH 7, which is a common cultivation pH for E. coli, the acids dissociate and lower the pH. This mechanism has been used hundreds of years as a method for food preservation. In an industrial fermentation the pH is kept constant at optimal levels by addition of acid and base. If the pH of the medium is lowered to values under the pKa of the organic acids, the acids remain undissociated and hence uncharged. An uncharged organic acid can travel freely across the cell membrane by passive diffusion (66). Once inside the slightly alkaline cytoplasm the acids dissociate increasing the proton concentration inside the cell, lowering the cytoplasmic pH and dissipating the trans-membrane proton gradient and the protonmotive force (67). The effect of a pH reduction is thus most severe when present together with organic acids.

Organic acid, metabolic uncoupling and anion accumulation

Acetate negatively affects growth of E. coli cultures (68). By using the ethanologenic E. coli strain LY01 the effects of different organic acids from hemicellulose hydrolysates on growth has been demonstrated (69). It was shown that the acids reduced growth more at lower pH than at neutral or slightly alkaline. The effect of organic acids on pH is not only due to the lowering of the cytoplasmic pH, but also includes anion specific effects (70). It has been shown that as acetate anions accumulate, glutamate is released from the E. coli cells. In E. coli the turgor pressure (pressure on the cell wall from the cytoplasm) is produced by accumulation of mainly

18 potassium glutamate (71). E. coli uses the release of glutamate to keep the turgor pressure constant during acetate accumulation (70). The growth inhibition caused by acetate anions is proposed to be linked to methionine biosynthesis (72). It was suggested that the inhibition caused by acetate is an effect of the anion affecting an enzyme in the methionine biosynthesis pathway leading to accumulation of homocysteine, a metabolite that has been shown to inhibit E. coli growth. The effects of the fermentation acids anions are thus not fully understood, although some interesting theories and contributions have been made.

As explained above, the protons generated by dissociation of organic acids in the cytoplasm affects the organisms both by reducing the cytoplasmic pH and thus damaging protein structures and DNA, and by reducing ǻpH and disrupting the protonmotive force. The reduction of ǻpH can be mediated by what is called an uncoupler. Synthetic uncouplers are lipophilic compounds that travel across the cell membranes in both undissociated as well as in dissociated form (Fig. 8). The protonated form of the species crosses the membrane and releases its proton upon entry into the alkaline cytoplasm. The deprotonated form is then pushed to the external side of the membrane by the electrical gradient. The anion is again protonated, diffuses into the cytoplasm and releases another proton. Protons however are not able to cross the membranes and remain in the cytoplasm. The cycle continues until the protonmotive force is completely disrupted.

Inside cell Outside cell

HX HX

+ H H+ X- X-

Figure 8. Mechanism of uncoupling. HX represents the undissociated acid.

There has been a debate if organic acids are able to act as true uncouplers. Russel argues against organic acids as true uncouplers by pointing out their inability to diffuse through the cell membranes, instead they will accumulate in the cytoplasm like protons (58). He is contradicted by Axe and Bailey who propose that both acetate and lactate permeates through the

19 membranes at comparable rates in both undissociated and dissociated form and in this way they catalytically dissipate the protonmotive force (73). This model also purports that since acetate could freely traverse the cell membrane it would not accumulate in the cytoplasm. In their work it was observed that the intracellular acetate concentrations did not reach levels predicted by ǻpH. In contrast, Diez-Gonzalez and Russel have shown that acetate accumulation increases with increasing external acetate concentration and an equilibrium between the intracellular acetate and ǻpH occurs when the external acetate concentrations is 60 mM or more (74). They also found that acetate accumulates as long as the intracellular pH and thus the pH gradient is kept high (74, 75) However, as intracellular pH decreased acetate accumulation ceased. The effects of organic acids can thus be two-fold. Even though it cannot be said that organic acids act as uncouplers per se, the growth inhibiting effects seems to originate both from acidification of the cytoplasm and of anion accumulation. In E. coli K-12 even low (<50 mM) extracellular acetate concentrations will result in decreased levels of intracellular ATP (75). The effect can be attributed to the ATP requirements for the reversible activity of ATP synthases to pump protons out of the cytoplasm against the proton gradient in order to keep a slightly alkaline intracellular pH. Acidification of the cytoplasm and the following reduction in intracellular ATP levels as an effect proton extrusion is a typically observed in the presence of an uncoupler (75).

Acid resistance systems

E. coli have a number of ways to control acid charge. The pH homeostasis of E. coli can be divided into a passive and an active homeostatic system (76). The main contributions to the passive system are the relative impermeability of the lipid bilayer to protons and ions, and the cytoplasmic buffering capacity. The buffering system uses phosphate groups and carboxylated metabolic substances to obtain a buffering capacity in the range of 50-100 nmol H+ per pH unit and mg cell protein (62). The cell membranes are an essential part of the passive pH homeostasis. E. coli exposed to acidic conditions have been shown to increase the amount of cyclopropane fatty acids in the cytoplasmic membrane. The higher concentration of cyclopropane fatty acids is supposed to increase the survival rate of the cells (77, 78).

The active homeostasic system controls the flow of ions (mainly sodium, potassium, and protons) across the cell membranes. It has been demonstrated that the role for Na+/H+ antiporters is central in the sodium regulation, but not for cytoplasmic pH control (79, 80).

20 Potassium uptake on the other hand has a strong correlation with the regulation and maintenance of an alkaline cytoplasmic pH. Kroll and Booth showed that E. coli cells suspended in potassium-free medium quickly reduced their intracellular pH. Upon introduction of potassium in the medium alkaline pH in the cytoplasm was restored (81, 82).

The uptake and accumulation of K+ is an important part of pH homeostasis at close to neutral pH, under extreme acid stress E. coli also use other systems to ensure survival. Today four well characterised acid resistance systems (AR, Fig. 9) are known (61, 83-86). AR1 is activated when cells are grown aerobically in LB medium at pH 5.5. This system is dependent on the alternative sigma factor (rpoS, part of the RNA polymerase holoenzyme necessary for DNA transcription), and the cAMP receptor protein (63, 83). The AR1 system is repressed by glucose and does not need any extracellular amino acids present in the medium. The system provides stationary phase acid resistance down to pH 2.5 (85).

Two other stationary phase acid resistance systems were discovered when E. coli was cultivated in glucose containing media (87). These systems depend on extracellular glutamate (for AR2) and arginine (for AR3) (87-90). Both systems confer acid resistance down to pH 2 and comprise an amino acid decarboxylase and an antiporter. The decarboxylases (GadA or GadB for glutamate and AdiA for arginine) substitute a carboxyl group on the amino acid with a proton from the cytoplasm. The reaction products are the decarboxylated amino acid (Ȗ-amino butyric acid (GABA) from glutamate and agmatine from arginine) and CO2. GABA or agmatine is excreted from the cells through their corresponding antiport (GadC for glutamate and AdiC for arginine) and new amino acids are taken up. AR2 is induced in aerobically grown cultures in the presence of glucose and extracellular glutamate. AR3 also requires glucose, but differs from AR2 in that it is activated under anaerobic cultivation. Both systems generate internal pH values in close proximity with the pH optima of the decarboxylases (pH 4 for GadA and GadB and pH 5 for AdiA). If the pH increases above the optimum for the respective carboxylases, activity will decrease and pH will drop. A fourth acid resistance system (AR4) was recently discovered (89). It is similar to AR2 and AR3, but is dependent on lysine and its efficiency is low. The low efficiency could be explained by a higher pH optima for lysine decarboxylase enzyme activity. During extreme acid stress conditions, the extracellular pH initially lowers the intracellular pH too much and as a consequence the decarboxylase looses activity and is not able to meet the proton charge and raise the internal pH (61).

21 H+

GABA rpoS (AR1) GABA

+ Glutamate decarboxylase (AR2) Glutamate H Glutamate

Agmatine Agmatine

+ Arginine decarboxylase (AR3) + Arginine H H Arginine Cadaverine Cadaverine

+ Lysine decarboxylase (AR4) Lysine H Lysine

Figure 9. Acid resistance system in E. coli.

When E. coli are grown at near neutral pH, ǻȌ is maintained at roughly -90 mV. In the stationary phase the membrane potential decreases to approximately -50 mV. Transferring the culture to acidic pH (pH < 3) reduces ǻȌ to 0. The effect is expected to be caused by a loss of membrane integrity. Without a functioning membrane it is no longer possible to generate and maintain ion gradients and extracellular and cytoplasmic concentrations of ions are evened out, hence reducing the membrane potential. In an investigation of AR2 and AR3 it was found that if glutamate or arginine was present in the medium, E. coli reverts its membrane potential in the same way as some acidophilic organisms do. With glutamate a membrane potential of +30 mV was achieved and with arginine +80 mV (91). The proposed mechanism of protection would be that the positive membrane potential should repel protons and thus relieve some of the acid stress.

Finally E. coli possess systems for the protection of vital protein. In order to function properly, cells need to maintain the activity of proteins, including those linked to cellular membranes and the cell wall. HdeAB, a heterodimer expressed in the periplasm, has been shown to dissociates into monomers at acid pH. The dissociated subunits bind to unfolded proteins in the periplasm and prevents unwanted protein aggregation (92). It is not known what happens to the proteins once the acid stress is relieved, but mutants lacking hdeAB showed a dramatic

22 loss of viability at acid pH (93). The protection of periplasmic proteins thus seems crucial for cell survival at low pH.

Osmotic stress in Escherichia coli

Osmotic pressure, osmolarity and turgor pressure

In the same way as the lipid membranes of E. coli are vital to pH homeostasis they also restrict the transport of other ionic compounds, thus generating concentration gradients between the cytoplasm and the medium.

The phenomenon of osmotic pressure is well described by a vessel divided into two compartments by a semipermeable membrane. The membrane restricts the transport of solutes between the two compartments, but permits the flow of water. If compartment 1 is filled with pure water and compartment 2 with a solution having a certain concentration of for example an ionic compound the water activity (or concentration of water) in compartment 2 is lower than in compartment 1. As a result, water will diffuse from compartment 1 across the semipermeable membrane into compartment 2 and raise the liquid level. Water will continue to diffuse into compartment 2, lowering the liquid level in compartment 1 and further raising it in compartment 2 creating a hydrostatic pressure difference. The influx of water to compartment 2 will continue until the hydrostatic pressure on compartment 2 can balance the tendency of water to diffuse into the compartment. The equilibrium pressure achieved is termed the osmotic pressure. The osmotic pressure can be quantified using the van’t Hoff equation:

3 icB RT (2) where i is the van’t Hoff factor that describes the degree of dissociation of a solute into ions,

+ - e.g. NaCl into Na and Cl , cB is the total solute concentration in moles per liter, R is the universal gas constrant and T is the temperature in Kelvin. Equation (2) holds for ideal dilute solutions.

23 At higher solute concentration the quantity 3 RT diverges from cB. Instead the unit osmolarity (Osm, effective molar concentration of the solute in solution) is introduced and defined as:

3  ln aw Osm m (3) RT Vw

m where,aw is the water (solvent) activity and Vw is the molar volume of water. The complete derivation of the equations is given elsewhere (94, 95). Osmolarity is measured in osmoles per liter solvent (an osmole is the number of moles of a substance that contributes to a solutions osmotic pressure).

E. coli has been shown to be able to grow in osmolarities from 0.015 Osm to approximately 1.9 Osm (minimal media) or 3.0 Osm (rich media) (96-98). An E. coli cell can also be described in terms of the two compartment model. The interior of the cell responds to increasing or decreasing concentration in the medium by adjusting the water or solute content in the cytoplasm. In most cases the solute concentration of the cytoplasm is higher than that of the surrounding medium and water has a tendency to diffuse into the cell, a state where the cells are said to be in positive water balance (99).

The hydrostatic pressure difference between the cytoplasm and the cell’s environment is called turgor pressure (ǻȆ).

RT aw,out '3 3 cyto  3 ex RT Osmcyto  Osmex m ln (4) Vw aw,in

The turgor pressure calculated by equation (4) has a major impact on the functionality of the cells. Proteins, enzymes and other macromolecules have a range of water activity and ionic concentrations in which they retain proper function. In media that force osmolarities of the cytoplasm outside of this range. E. coli cells could lose important cellular functions and mechanisms (100). When cells are subjected to hyper- or hypoosmotic shock they will respond by fast influx or efflux of cytoplasmic water. Hypoosmotic shock means that bacteria are subjected to very low osmolarities. The result is an influx of water and simultaneous swelling of the cell. Gram-negative cell walls can withstand pressures of up to 100 atm and hypoosmotic treatment has limited effect on cells (101, 102). Hyperosmotic shock is the opposite. Cells are

24 subjected to solutions with very high salt concentrations. This treatment causes water efflux and shrinkage of the cytoplasmic volume. Severe hyperosmotic shock will lead to plasmolysis; the cytoplasmic membrane being retracted from the cell wall leaving a gap between cell wall and cytoplasm. Hyperosmotic environments will negatively affect cell functions causing inhibition of DNA replication, decreased uptake of nutrients, reduced synthesis of macromolecules and subsequent ATP accumulation (103-106).

The role of compatible solutes in the response of Escherichia coli to osmotic stress

Cells exposed to high medium osmolarity export intracellular water resulting in a reduction in turgor pressure and cytoplasmic volume. The concentrations of intracellular metabolites increase with the decrease of the cytoplasmic volume. This passive adaptation to the surrounding osmolarity is often not adequate and could result in inhibitory or toxic concentrations of the intracellular metabolites.. Instead E. coli exposed to hyper- or hypoosmotic stress adapt by accumulating or releasing certain solutes like potassium ions, amino acids (glutamate, proline), quaternary amines (e.g. glycinebetaine), and sugars (e.g. sucrose, trehalose). Theses solutes are called compatible solutes because they can be accumulated to high levels through uptake from the surrounding environment or by de novo synthesis without negatively affecting most cytoplasmic enzymes (100, 107). Common for most compatible solutes is that they cannot traverse the cellular membranes easily, but has to be transported mainly by active transport mechanisms. Most compatible solutes are uncharged, with the exception of the most important compatible solutes, K+ and glutamate-, and will thus not interfere with cellular macromolecules. Some compatible solutes do not only help the cell to survive osmotic stress, but will also positively affect growth rate. These are called osmoprotectants (107).

Potassium is the major intracellular cation and is accumulated as the first response of E. coli to osmotic upshock (71, 107). Potassium is taken up via the transport systems Kdp and Trk. Trk is a constitutive low affinity, high capacity uptake system and Kdp is a high affinity, low capacity system that is induced by low turgor pressure when Trk cannot maintain the turgor pressure (108, 109). Closely coupled with potassium accumulation is the increased de novo synthesis of glutamate. Glutamate synthesis is strongly dependent of K+ and if a medium deprived of K+ is used, glutamate synthesis is reduced to approximately 10 % of the synthesis

25 rate in the presence of K+ (71). A second response after initial K+ accumulation is the displacement of K+ and glutamate by neutral compatible solutes that can accumulate to higher concentrations without inhibiting vital cellular functions. The uptake of neutral species may occur simultaneously with K+ uptake, but uptake via Trk is more rapid and will be most important for the osmotolerance in the early stages of an osmotic upshock (107).

Glycinebetaine and proline are accumulated following initial uptake of K+. These compatible solutes interfere less with the cytoplasmic constituents than K+ and glutamate and during uptake of glycinebetaine and proline K+ is excreted from the cells. Glycinebetaine was shown by May et al. to be mediated by the transport systems ProP and ProU. The same systems are used for transport of proline, but the affinity of t ProU is higher for glycinebetaine (110, 111). Glycinebetaine also offers higher osmotolerance to E. coli than proline. ProP uses the energy from the protonmotive force for active transport and ProU uses energy derived from hydrolysis of ATP (107, 111). The first system to respond to an osmotic upshock is the ProP, which is a low affinity transport system. The transport system is activated within seconds after the cells are subjected to the high osmolarity environment. ProP is also the physiologically more important system at high osmolarities, since cells under high osmolarity conditions without this system are unable to take up proline and glycinebetaine from the medium. ProU, on the other hand, is first activated after several minutes (107). It is a periplasmic high affinity transport system that because of its high affinity for its substrates plays an important part in osmotic adaptation when glycinebetaine and proline are present in low concentrations. Both systems however only accumulate glycinebetaine under osmotic stress (111). Roth et al. demonstrated that cells grown in a medium lacking compatible solutes lost their colony-forming activity and regained it when glycinebetaine was added (112).

Most bacteria are unable to synthesise glycinebetaine de novo and to use glycinebetaine as an osmoprotectants they are dependent on transport of this compound for accumulation (100). Even though glycinebetaine cannot be synthesised by E. coli from glucose it has been shown that choline under osmotic stress is converted to glycinebetaine (113). Therefore choline is also an osmoprotectants for E. coli. In E. coli one enzyme is responsible for the conversion of choline to glycinebetaine and its activity is dependent on electron transport and requires thus some terminal electron acceptor, e.g. O2. Choline can therefore only be used as an osmoprotectants during aerobic conditions (111, 113).

26 Like glycinebetaine, proline cannot be synthesised in E. coli and other gram-negative bacteria and high intracellular concentrations of proline during osmotic stress can only be obtained by increased uptake. It has been found that both proline and glycinebetaine uptake from exogenous sources is proportional to the osmotic strength of the medium (100, 114). Nagata et al. studied the combined effects of proline together with K+ and found that the co-existence of K+ and proline in high osmolarity media resulted in greater recovery of growth than in the presence of proline alone (115).

Ectoine, another amino acid that has been identified as a compatible solute in halophiles, was shown to improve E. coli growth in high osmolarity media (116). Ectoine accumulates in E. coli and has similar osmoprotective abilities as glycinebetaine. Both the ProP and ProU systems are responsible for ecotine uptake, where ProP is the main uptake system (116).

Trehalose is accumulated when other compatible solutes are absent from the cultivation medium. It is only accumulated via synthesis since uptake of the sugar from the medium will result in the phosphorylated form of the sugar, which is not a compatible solute and thus does little to relieve the bacterium during osmotic stress (117). Synthesis of trehalose is under control of the rpoS sigma factor, which accumulates when cells are grown at high osmolarity. Trehalose can accumulate to intracellular concentrations equal to approximately 20 % of the osmolar concentration of solutes in the cultivation medium (100) and synthesis also leads to potassium efflux.

A last part of the osmoregulation of E. coli is the outer membrane porins. Porins are proton channels incorporated in the outer membrane of E. coli that transport small hydrophilic molecules into the periplasm by passive diffusion. For osmotic regulation the two most important porins are OmpF and OmpC. Their production is regulated by a number of environmental conditions such as osmolarity, carbons source and temperature. The total cellular concentration of the porins is more or less constant, but the ratio between them varies. High osmolarity, temperature and good carbon sources increase the levels of OmpC and concomitantly reduce the levels of OmpF and vice versa. The pore size of ompF is larger than of ompC, which is an explanation to why OmpC is preferentially synthesised in high salinity media where the nutrients are plentiful the temperature high; conditions which makes diffusion into the cell easy. OmpF on the other hand is needed when nutrients are scarce and it is important for the bacterium to scavenge the medium for nutrients (100).

27 Finally it should be noted that the rate of the hyperosmotic shock also is important for viability and growth. Results from studies with osmotic upshock have shown that if cells are subjected to moderate osmotic stress their growth at higher osmolarities is not equally impaired. Osmotolerance can thus be induced by treatment prior to osmotic shock (118). Furthermore the survival of E. coli in increased osmolarity media also depends on time. A sudden osmolarity increase by NaCl reduced the viability of the cultures more than if the increase is carried out during 20 minutes (119). The results imply that cells are able to respond to the increase in osmolarity if the increase is slow.

28 Results and discussion

In this section a brief overview of the experiments carried out in Papers I and II will be given together with a discussion of the main results. For full descriptions of experimental procedures and results, please refer to Papers I and II. Common for both studies are the use of a low-cost medium, relevant for use in industrial succinic acid fermentations and the use of a high initial sugar concentration. Since the fermentations are carried out with a high sugar concentration the organism is not metabolically limited by the carbon source and it is possible to achieve high succinic acid productivity. The importance of the volumetric productivity on succinic acid production was stressed under ‘Current level of succinic acid productivity’ and should preferably be at least 2.5 g L-1 h-1 (1).

Sugar utilisation

The first study conducted addresses the effect of different sugars on succinic acid kinetics and yields during fermentations with AFP184. Glucose and xylose are the most abundant sugars in lignocellulosic biomass, while fructose and glucose are the components of sucrose, a widely available disaccharide. In line with the concept of a biorefinery, i.e. to make use of the cheapest available raw material, the producing organisms must be able to produce succinic acid in good yields from a wide range of sugar feedstocks. In this study glucose, fructose, xylose and equal mixtures of glucose:fructose and glucose:xylose were used as feedstocks in dual-phase fermentations. The fermentation profiles are shown in Figure 10.

29 120 16 120 16

Glucose ab14 Fructose 14 100 100

12 12 ) ) -1 -1 80 80 9 10 10

60 8 60 8

6 6 Cells/mL x 10 40 40 Concentration (g L (g Concentration Dry cell weight (g L 4 4

20 20 2 2

0 0 0 0 0 5 10 15 20 25 30 0 5 10 15 20 25 30 Time (hours) Time (hours)

120 16 60 16 Xylose c Glucose:Fructose d 14 14 100 50

12 12 ) ) -1 -1 80 40 9 10 10

60 8 30 8

6 6 Cells/mL x10 40 20 Concentration (g L (g Concentration Dry cell weight (g L 4 4

20 10 2 2

0 0 0 0 0 5 10 15 20 25 30 0 5 10 15 20 25 30 Time (hours) Time (hours)

60 16

Glucose:Xylose e 14 50

12 ) 40 -1 9 10

30 8

6 Cells/mL x 10 20 Concentration (g/L) Concentration Dry cellweight (g L 4 10 2

0 0 0 5 10 15 20 25 30 Time (hours)

Figure 10. Sugar, product and cell concentration profiles for (a) glucose, (b) fructose, (c) xylose, (d) glucose:fructose, and (e) glucose:xylose fermentations. Product and sugar concentrations are given in g L-1, dry cell weights in g L-1 and viable cells in (cells mL-1)×109. Symbols used: glucose (ǻ), fructose (¹), xylose (R), succinic acid (Ɣ), pyruvic acid („), viable cells (*), dry cell weight (×). Staggered line indicates time of switch to the anaerobic phase.

The most important results are summarised in Table 1. It is concluded that glucose is the preferred sugar, resulting in the highest productivities, final titres, and yields. The higher yields

30 obtained when glucose was used as the carbon source is due to the mutation in the PTS (described under ‘Sugar metabolism in AFP184’). Fermentation of fructose uses PEP for phosphorylation committing a larger fraction of the carbon to pyruvate (Fig. 6) and thus reducing the succinate yield. Xylose metabolism uses the pentose phosphate pathway (Fig. 6) and, as described under ‘Sugar metabolism’, the theoretical yield on xylose is somewhat lower than on glucose, but higher than on fructose. However, xylose fermentation resulted in the lowest yield achieved. A carbon balance can only account for approximately 60 % of the carbon. Further work is required to identify the remaining carbon and improve xylose fermentation.

Table 1. Summary of fermentation results for AFP184a.

-1 -1 -1 -1 -1 -1 Sugar P (h ) YP/S (g g ) Final titer QP (g L h )QP (g L h )

-1 d d anaerobic (g L ) anaerobic, T22 anaerobic, T32

Glucose 0.42 0.83 40.6 2.89 1.76

Fructose 0.50 0.66 32.3 1.79 1.26

Xylose 0.57 0.50 24.9 1.49 1.08

Glu:Frub - 0.58 25.2 1.67 1.05

Glu:Xylc - 0.60 27.6 1.48 1.08

a Specific growth rates were obtained from fermentations carried out in 1 L bioreactors. YP/S is

the mass yield of succinic acid based on the glucose consumed in the anaerobic phase and QP is the volymetric productivity of succinic acid during the anaerobic phase. b Glu:Fru is an abbreviation for glucose:fructose. c Glu:Xyl is an abbreviation for glucose:xylose. d anaerobic productivity at 22 and 32 hours total fermentation time respectively

A series of fermentations were also carried out to determine aerobic growth rates for the three sugars (Table 1). AFP184 grew on all sugars and sugar combinations. However specific growth rate on glucose was slower than for growth on xylose or fructose. This result is most likely an effect of the mutation in the PTS of AFP184. There is also a phosphotransferase permease for fructose (48, 52), but in strain AFP184 this permease is not affected by any mutations. Apart from the PTS, E. coli can also transport glucose by galactose permease and phosphorylate it by

31 the enzyme glucokinase (120). Glucokinase is not subjected to any mutations in AFP184. According to literature glucose uptake and phosphorylation by glucokinase is slower than by the PTS (51), which explains why the growth rate on fructose is higher than the growth rate on glucose. Unlike fructose, xylose is not transported into the cells via the PTS. Instead xylose uptake is facilitated by chemiosmotic effects (55, 56). Interestingly growth on xylose was initially very slow, while the use of glucose and fructose rapidly initiated exponential growth. Since the seed cultures were grown in a glucose medium, glucose and fructose fermentations may already have the enzymes required for utilisation of the respective sugars, while in the case of xylose expression of the enzymes first needs to be induced before exponential growth can start.

When glucose is mixed with other sugars E. coli usually consumes it first and after all glucose has been metabolised utilisation of other sugars is initiated, a phenomenon called catabolite repression. From the sugar consumption (Fig. 10) it was observed that when fermenting the sugar mixtures, fructose was metabolised slightly faster during the aerobic phase. This effect was expected due to the mutation strain AFP184 has in the glucose PTS. More interesting is the observation that when glucose and xylose were mixed, the sugars were fermented simultaneously, but xylose was consumed at a substantially higher rate. The effect is attributed to a mutation in the PTS, which in turn prevents glucose from repressing uptake of other sugars (120-123).

Organic acid concentrations (including both succinic and other acids) above approximately 30 g L-1 resulted in drastically decreased productivities per viable cell. The general inhibiting effects of organic acids is well known and describes in the section ‘Acid resistance, inhibition, and toxicity in Escherichia coli’. The fermentations in this study were carried out with ammonium hydroxide as base for neutralisation of end products and it has been reported that ammonium in concentrations above 3 g L-1 inhibits growth (124, 125). In the present work ammonium concentrations of approximately 10 g L-1 were reached at the end of the fermentations with high succinate production. In other studies changes in the growth characteristics of E. coli under different ammonium sulphate concentrations was investigated and ammonium sulphate concentrations of 5 % had to be used before any severe inhibition was observed (126). When the fermentations carried out in the present work showed a productivity decreases they were in the anaerobic phase and only very little growth would take place even without any inhibition. Therefore it is not clear if productivity would be affected

32 by the achieved ammonium concentrations. To better understand the causes of the possible organic acid and ammonia inhibition was further investigated in Paper II.

Impact of the neutralising agent

In this study the aim was to address the problems associated with decreased succinic acid productivity and limited final succinic acid concentrations. If the technology is to be applied to a large scale production facility it is important to identify and minimise the factors that negatively influence and limit productivity and final titre. Succinic acid production by AFP184 was studied when NH4OH, KOH, NaOH or Na2CO3 were used as neutralising agents. In Paper I a decrease in succinic acid production was observed when the fermentations accumulated organic acids. Organic acids have a pKa in the order of 4. The fermentations in Paper I were carried out at a pH of pH 6.6-6.7. At this pH the produced acids are dissociated and the cytoplasmic membrane should be relatively impermeable to the acid anions and the protons (62, 66). However, other studies indicate that acid anions can traverse the cytoplasmic membrane (73). If dissociated in the medium, the effect would be accumulation of the acid anions in the cytoplasm. Acetate anions are known to negatively affect growth in E. coli by affecting the methionine biosynthesis (72). The metabolic effects of succinate anions on anaerobe succinic acid production are unknown. The neutralising agent used in Paper I,

-1 NH4OH, is known to cause growth inhibition in E. coli at concentrations >3 g L (124, 125). The aim of Paper II was to study the possible effects of changing neutralising agent in order to improve the succinic acid productivity and final titres obtained in Paper I. Fermentations with different bases are summarised in Table 2.

Common for all fermentations was the loss of viable cells during the anaerobic phase of the fermentations. The rate of cell death was significantly higher when using NH4OH than any of the other neutralising agents. When using NH4OH succinate production and glucose consumption completely ceased after 32 hours of total fermentation time. For potassium and sodium bases succinate production never stopped, but the volumetric productivity decreased during the course of the fermentations (Fig. 11). The reason could be both due to the toxicity of ammonia (127) and due to lower glucose consumption and thus resulting in a decreased energy generation as the fermentations proceed. Productivities per viable cell were calculated

33 and neither sodium nor potassium bases showed any significant reduction in productivity per viable cell (see Paper II) and the decrease in productivity was attributed to a loss of viable cells.

Table 2. Summary of fermentation results for different basesa

-1 -1 -1 -1 -1 -1 -1 Base YP/S (g g ) QP (g L h ) QP (g L h ) QP (g L h )

-1 b anaerobic anaerobic, T20 anaerobic, T40 anaerobic, (max g L )

NH4OH 0.74 2.91 1.38 1.85*

KOH 0.76 2.61 1.33 0.65

NaOH 0.71 2.45 1.52 0.72

Na2CO3 0.77 3.31 2.05 0.89 a YP/S is the mass yield of succinic acid based on the glucose consumed in the anaerobic phase, and QP is the volumetric productivity of succinic acid during the anaerobic phase after 20 h b (T20) or 40 h (T40) total fermentation time. Anaerobic productivity calculated either when fermentations are terminated or maximum succinic acid concentration is achieved. *maximum concentration was achieved after 32 hours

Comparing the bases used in this study showed that Na2CO3 generated the highest productivities and final succinate concentrations. The reason why Na2CO3 neutralisation results in a higher succinate productivity than NaOH could be explained by an incomplete saturation of the medium with CO2. Addition of extra carbonate from the neutralising agent

 might increase the available levels of HCO 3 resulting in a higher amount of carbon that can be channelled through the reductive arm of the TCA.-cycle instead of being committed to pyruvate. Using Na2CO3 also resulted in a higher succinate yield than NaOH. The differences in the yield can be explained by an increased carbon flux down the reductive arm of the TCA-cycle. However it does not explain the generally low yields (0.7-0.8 gram succinate per gram glucose consumed anaerobically) achieved. The theoretical yield of succinic acid from one gram of glucose during anaerobic conditions is 1.12 (128).

34 100 a 100 b ) 80 ) 80 -1 -1 9 9

60 60

Cells/mL10 x 40 Cells/mL x 10 40 ConcentrationL (g Concentration (g L

20 20

0 0 0 20406080100 0 20406080100 Time (hours) Time (hours)

100 c 100 d ) 80 ) 80 -1 -1 9 9

60 60 Cells/mL x 10 40 Cells/mL x 10 40 Concentration (g L (g Concentration Concentration (g L

20 20

0 0 0 20406080100 0 5 10 15 20 25 30 35 40 Time (hours) Time (hours)

Figure 11. Succinic acid and glucose concentrations and viable cell density for fermentations neutralised with KOH (a), NaOH (b), Na2CO3 (c), and NH4OH (d). Symbols used in the Figure: Ƈ Succinic acid (g L-1), ¹ Glucose (g L-1), and Ɣ Viable cells (cells mL-1 x 109).

The productivity is strongly correlated to the concentration of viable cells and to further increase succinate productivity by AFP184 it is essential to reduce the loss of viability during the anaerobic phase. The reason the cells lose viability is not clear. Elevated organic acid concentration in the medium and possible accumulation of organic acid anions in the cytoplasm may be important factors. Increased osmolarity due to organic acid production and addition of neutralising agent is also assumed to have a negative effect on viability. During high osmolarity E. coli accumulates compatible solutes, preferably glycinebetaine, proline, and ectoine. Corn steep liquor does not contain these osmoprotectants and in the absence of these, E. coli is known to synthesise and accumulate trehalose. Synthesis of trehalose and accumulation of succinate could be part of the answer to why the yields are diverging from the theoretical. A study elucidating intracellular concentrations of metabolites, in particular succinate and trehalose, and the effect of supplementation of the medium with osmoprotectants like glycinebetaine is being undertaken.

35 36 Conclusions

In Paper I we have demonstrated growth and succinate production of E. coli strain AFP184 in a medium containing corn steep liquor, salts, and different monosaccharides. The reduction of complex components in the medium has a positive effect on the final price of the produced succinate due to lower cost of raw materials; however final concentrations are in the range of 30-40 g L-1 after 32 hours and need to be improved for the process to become more economical.

In Paper II we have demonstrated that replacing the neutralising agent can provide substantial process improvements in the form of increased duration of high volumetric productivity and increased final titre. Compared to fermentations neutralised with NH4OH it was possible to achieve an almost 100 % increase in final succinic acid concentration using Na2CO3. It was also observed that the cell viability decreased during the anaerobic phase irrespective of the base used. The viable cell decrease is attributed to increased acid concentration coupled with possible cytoplasmic succinate accumulation, high medium osmolarity and long time exposure to a low energy generating environment and high maintenance demands during the course of acid production in the anaerobic phase.

37 38 Future work

The two studies done in this work has greatly improved to possible production of biological succinic acid. They have also generated a number of potential future studies.

The most important aspect to address is the loss of cell viability during anaerobic succinate production. To unravel the probable reasons studies investigating intracellular concentrations of metabolites such as succinate and trehalose should be undertaken. The impact of addition of different osmoprotectants to the growth medium should also be investigated in conjunction with determination of intracellular metabolite concentrations.

A thorough study of the fermentation parameters (pH, temperature, agitation, etc…) for the anaerobic phase should be carried out to determine optimal conditions for succinic acid production.

AFP184 has been shown to be able to utilise a number of different monosaccharides. In order to fit into a biorefinery concept it is important to demonstrate and investigate the fermentation characteristics of AFP184 in industrial hydrolysates from e.g. hemicellulose extracted prior to alkaline pulping in a kraft pulp mill or hydrolysed corn stover.

Following fermentation succinic acid must be separated from the fermentation broth. Very little work has been done in this area and more attention needs to be directed to this field in order to improve the economical feasibility of biochemical succinic acid production.

39 40 References

1. T. Werpy and G. Petersen, Top value added chemicals from biomass, volume I. 2004. 2. J.G. Zeikus, M.K. Jain, and P. Elankovan, Biotechnology of succinic acid production and markets for derived industrial products. Applied Microbiology and Biotechnology, 1999. 51(5): p. 545-552. 3. B.E. Dale, 'Greening' the chemical industry: research and development priorities for biobased industrial products. Journal of Chemical Technology and Biotechnology, 2003. 78(10): p. 1093-1103. 4. B. Kamm and M. Kamm, Principles of biorefineries. Applied Microbiology and Biotechnology, 2004. 64(2): p. 137-145. 5. H. Theliander, M. Paulsson, and H. Brelid, Introduktion till massa- och pappersframställning. 4th ed. 2002, Göteborg: Chalmers university of technology. 6. A. Van Heiningen, Converting a kraft pulp mill into an integrated forest biorefinery. Pulp & Paper-Canada, 2006. 107(6): p. 38-43. 7. U. Wising and P. Stuart, Identifying the Canadian forest biorefinery. Pulp & Paper- Canada, 2006. 107(6): p. 25-30. 8. R.C. Pettersen, The chemical composition of wood. Advances in chemistry series, 1984. 207: p. 57-126. 9. T.E. Timell, Recent progress in the chemistry of wood hemicelluloses. Wood science and technology, 1967. 1(1): p. 45. 10. E.D. Larson, Preliminary Economics of Black Liquor Gasifier/Gas Turbine Cogeneration at Pulp and Paper Mills. Journal of engineering for gas turbines and power, 2000. 122: p. 255. 11. M.J. Van der Werf, M.V. Guettler, M.K. Jain, and J.G. Zeikus, Environmental and physiological factors affecting the succinate product ratio during carbohydrate fermentation by Actinobacillus sp. 130Z. Archives of Microbiology, 1997. 167(6): p. 332- 342. 12. M.V. Guettler, D. Rumler, and M.K. Jain, Actinobacillus succinogenes sp. nov., a novel succinic-acid-producing strain from the bovine rumen. International Journal of Systematic Bacteriology, 1999. 49: p. 207-216.

41 13. R. Chatterjee, C.S. Millard, K. Champion, D.P. Clark, and M.I. Donnelly, Mutation of the ptsG gene results in increased production of succinate in fermentation of glucose by Escherichia coli. Applied and Environmental Microbiology, 2001. 67(1): p. 148-154. 14. P.C. Lee, S.Y. Lee, S.H. Hong, and H.N. Chang, Isolation and characterization of a new succinic acid-producing bacterium, Mannheimia succiniciproducens MBEL55E, from bovine rumen. Applied Microbiology and Biotechnology, 2002. 58(5): p. 663-668. 15. G.N. Vemuri, M.A. Eiteman, and E. Altman, Effects of growth mode and pyruvate carboxylase on succinic acid production by metabolically engineered strains of Escherichia coli. Applied and Environmental Microbiology, 2002. 68(4): p. 1715-1727. 16. D.Y. Kim, S.C. Yim, P.C. Lee, W.G. Lee, S.Y. Lee, and H.N. Chang, Batch and continuous fermentation of succinic acid from wood hydrolysate by Mannheimia succiniciproducens MBEL55E. Enzyme and Microbial Technology, 2004. 35(6-7): p. 648- 653. 17. H. Lin, G.N. Bennett, and K.Y. San, Fed-batch culture of a metabolically engineered Escherichia coli strain designed for high-level succinate production and yield under aerobic conditions. Biotechnology and Bioengineering, 2005. 90(6): p. 775-779. 18. H. Lin, G.N. Bennett, and K.Y. San, Genetic reconstruction of the aerobic central metabolism in Escherichia coli for the absolute aerobic production of succinate. Biotechnology and Bioengineering, 2005. 89(2): p. 148-156. 19. S. Okino, M. Inui, and H. Yukawa, Production of organic acids by Corynebacterium glutamicum under oxygen deprivation. Applied Microbiology and Biotechnology, 2005. 68(4): p. 475-480. 20. A.M. Sanchez, G.N. Bennett, and K.Y. San, Efficient succinic acid production from glucose through overexpression of pyruvate carboxylase in an Escherichia coli alcohol dehydrogenase and lactate dehydrogenase mutant. Biotechnology Progress, 2005. 21(2): p. 358-365. 21. S.J. Lee, H. Song, and S.Y. Lee, Genome-based metabolic engineering of Mannheimia succiniciproducens for succinic acid production. Applied and Environmental Microbiology, 2006. 72(3): p. 1939-1948. 22. M.I. Donnelly, C.S. Millard, D.P. Clark, M.J. Chen, and J.W. Rathke, A novel fermentation pathway in an Escherichia coli mutant producing succinic acid, acetic acid, and ethanol. Applied Biochemistry and Biotechnology, 1998. 70-2: p. 187-198.

42 23. A.M. Sanchez, G.N. Bennett, and K.Y. San, Novel pathway engineering design of the anaerobic central metabolic pathway in Escherichia coli to increase succinate yield and productivity. Metabolic Engineering, 2005. 7(3): p. 229-239. 24. H. Song and S.Y. Lee, Production of succinic acid by bacterial fermentation. Enzyme and Microbial Technology, 2006. 39(3): p. 352-361. 25. N.S. Samuelov, R. Datta, M.K. Jain, and J.G. Zeikus, Whey fermentation by Anaerobiospirillum succiniciproducens for production of a succinate-based animal feed additive. Applied and Environmental Microbiology, 1999. 65(5): p. 2260-2263. 26. N.P. Nghiem, B.H. Davison, J.E. Thompson, B.E. Suttle, and G.R. Richardson, The effect of biotin on the production of succinic acid by Anaerobiospirillum succiniciproducens. Applied Biochemistry and Biotechnology, 1996. 57-8: p. 633-638. 27. P.C. Lee, W.G. Lee, S.Y. Lee, and H.N. Chang, Succinic acid production with reduced by-product formation in the fermentation of Anaerobiospirillum succiniciproducens using glycerol as a carbon source. Biotechnology and Bioengineering, 2001. 72(1): p. 41-48. 28. P.C. Lee, W.G. Lee, S.Y. Lee, H.N. Chang, and Y.K. Chang, Fermentative production of succinic acid from glucose and corn steep liquor by Anaerobiospirillum succiniciproducens. Biotechnol. Bioprocess Eng., 2000. 5: p. 379-381. 29. P.C. Lee, W.G. Lee, S.Y. Lee, and H.N. Chang, Effects of medium components on the growth of Anaerobiospirillum succiniciproducens and succinic acid production. Process Biochemistry, 1999. 35(1-2): p. 49-55.

30. N.S. Samuelov, R. Lamed, S. Lowe, and J.G. Zeikus, Influence of CO2-HCO3- Levels and pH on Growth, Succinate Production, and Enzyme-Activities of Anaerobiospirillum- Succiniciproducens. Applied and Environmental Microbiology, 1991. 57(10): p. 3013-3019. 31. N.P. Nghiem, B.H. Davison, B.E. Suttle, and G.R. Richardson, Production of succinic acid by Anaerobiospirillum succiniciproducens. Applied Biochemistry and Biotechnology, 1997. 63-5: p. 565-576. 32. P.C. Lee, S.Y. Lee, S.H. Hong, H.N. Chang, and S.C. Park, Biological conversion of wood hydrolysate to succinic acid by Anaerobiospirillum succiniciproducens. Biotechnology Letters, 2003. 25(2): p. 111-114. 33. M.V. Guettler, M.K. Jain, and D. Rumler, Method for making succinic acid, bacterial variants for use in the process, and methods for obtaining variants. 1996. U.S. Patent 5,573,931

43 34. M.V. Guettler, M.K. Jain, and B.K. Soni, Process for making succinic acid, microorganisms for use in the process and methods of obtaining the microorganisms. 1996. U.S. Patent 5,504,004 35. M.V. Guettler, M.K. Jain, and B.K. Soni, Process for making succinic acid, microorganisms for use in the process and methods for obtaining the microorganisms. 1998. U.S. Patent 5,723,322 36. G.N. Vemuri, M.A. Eiteman, and E. Altman, Succinate production in dual-phase Escherichia coli fermentations depends on the time of transition from aerobic to anaerobic conditions. Journal of Industrial Microbiology & Biotechnology, 2002. 28(6): p. 325-332. 37. J.L. Stokes, Fermentation of glucose by suspensions of Escherichia coli. Journal of Bacteriology, 1948. 57(2): p. 147-158. 38. M.I. Donnelly, C.Y. Sanville-Millard, and N.P. Nghiem, Method to produce succinic acid from raw hydrolysates. 2004. U.S. Patent 6,743,610 39. H. Lin, G.N. Bennett, and K.Y. San, Chemostat culture characterization of Escherichia coli mutant strains metabolically engineered for aerobic succinate production: A study of the modified metabolic network based on metabolite profile, enzyme activity, and gene expression profile. Metabolic Engineering, 2005. 7(5-6): p. 337-352. 40. H. Lin, G.N. Bennett, and K.Y. San, Effect of carbon sources differing in oxidation state and transport route on succinate production in metabolically engineered Escherichia coli. Journal of Industrial Microbiology & Biotechnology, 2005. 32(3): p. 87-93. 41. H. Lin, G.N. Bennett, and K.Y. San, Metabolic engineering of aerobic succinate production systems in Escherichia coli to improve process productivity and achieve the maximum theoretical succinate yield. Metabolic Engineering, 2005. 7(2): p. 116-127. 42. H. Lin, K.Y. San, and G.N. Bennett, Effect of Sorghum vulgare phosphoenolpyruvate carboxylase and Lactococcus lactis pyruvate carboxylase coexpression on succinate production in mutant strains of Escherichia coli. Applied Microbiology and Biotechnology, 2005. 67(4): p. 515-523. 43. N.R. Yun, K.Y. San, and G.N. Bennett, Enhancement of lactate and succinate formation in adhE or pta-ackA mutants of NADH dehydrogenase-deficient Escherichia coli. Journal of Applied Microbiology, 2005. 99(6): p. 1404-1412. 44. R.W. Liggett and H. Koffler, Corn steep liquor in microbiology. Microbiology and molecular biology reviews, 1948. 12: p. 297-311.

44 45. H.C. Vogel and C.L. Tadaro, Fermentation and biochemical engineering handbook - Principles, process design, and equipment. 2nd ed. 1997, New Jersey: William Andrew Publishing/Noyes. 46. E.V. Cardinal, Microbiological assay od corn steep liquor for amino acid content. The Journal of biological chemistry, 1948. 172(2): p. 609. 47. G. Gosset, Improvement of Escherichia coli production strains by modification of the phosphoenolpyruvate : sugar phosphotransferase system. Microbial Cell Factories, 2005. 4. 48. J.H. Tchieu, V. Norris, J.S. Edwards, and M.H. Saier, The complete phosphotransferase system in Escherichia coli. Journal of Molecular Microbiology and Biotechnology, 2001. 3(3): p. 329-346. 49. P.W. Postma, J.W. Lengeler, and G.R. Jacobson, Phosphoenolpyruvate - Carbohydrate Phosphotransferase Systems of Bacteria. Microbiological Reviews, 1993. 57(3): p. 543-594. 50. C.H. Chou, G.N. Bennett, and K.Y. San, Effect of modulated glucose uptake on high- level recombinant protein production in a dense Escherichia coli culture. Biotechnology progress, 1994. 10(6): p. 644-7. 51. S.J. Curtis and W. Epstein, Phosphorylation of d-glucose in Escherichia coli mutants defective in glucosephosphotransferase, mannosephosphotransferase, and glucokinase. Journal of Bacteriology, 1975. 122: p. 1189-1199. 52. H.L. Kornberg, Fructose Transport by Escherichia-Coli. Philosophical Transactions of the Royal Society of London Series B-Biological Sciences, 1990. 326(1236): p. 505-513. 53. H.L. Kornberg, L.T.M. Lambourne, and A.A. Sproul, Facilitated diffusion of fructose via the phosphoenolpyruvate/glucose phosphotransferase system of Escherichia coli. Proceedings of the National Academy of Sciences of the United States of America, 2000. 97(4): p. 1808-1812. 54. H. Kornberg, If at first you don't succeed... fructose utilization by Escherichia coli, in Advances in Enzyme Regulation, Vol 42, Proceedings. 2002. p. 349-360. 55. T.W. Jeffries, Utilization of xylose by bacteria, yeasts, and fungi. Advances in Biochemical Engineering/Biotechnology, 1983. 27: p. 1-32. 56. V.M.S. Lam, K.R. Daruwalla, P.J.F. Henderson, and M.C. Jones-Mortimer, Proton- linked d-xylose transport in Escherichia coli. Journal of Bacteriology, 1980. 143(1): p. 396- 402. 57. J.D. David and H. Wiesmeyer, Control of xylose metabolism in Escherichia coli. Biochimica et biophysica acta, 1970. 201(3): p. 497-9.

45 58. J.B. Russell, Another Explanation for the Toxicity of Fermentation Acids at Low Ph - Anion Accumulation Versus Uncoupling. Journal of Applied Bacteriology, 1992. 73(5): p. 363-370. 59. I.R. Booth, P. Cash, and C. O'Byrne, Sensing and adapting to acid stress. Antonie Van Leeuwenhoek International Journal of General and Molecular Microbiology, 2002. 81(1-4): p. 33-42. 60. P.J.E. Mitchell, Coupling of phosphorylation to electron and proton transfer by a chemiosmotic type of mechanism. Nature, 1961. 191: p. 144. 61. J.W. Foster, Escherichia coli acid resistance: Tales of an amateur acidophile. Nature Reviews Microbiology, 2004. 2(11): p. 898-907. 62. I.R. Booth, Regulation of cytoplasmic pH in bacteria. Microbiological reviews, 1985. 49: p. 359-378. 63. P. Small, Acid and base resistance in Escherichia coli and Shigella flexneri: role of rpoS and growth pH. Journal of bacteriology, 1994. 176(6): p. 1729. 64. D. Zilberstein, Escherichia coli intracellular pH, membrane potential, and cell growth. Journal of bacteriology, 1984. 158(1): p. 246. 65. J.H.M. Gorden, Acid resistance in enteric bacteria. Infection and immunity, 1993. 61(1): p. 364. 66. T. Warnecke and R.T. Gill, Organic acid toxicity, tolerance, and production in Escherichia coli biorefining applications. Microbial Cell Factories, 2005. 4. 67. E.R. Kashket, Bioenergetics of Lactic-Acid Bacteria - Cytoplasmic Ph and Osmotolerance. Fems Microbiology Reviews, 1987. 46(3): p. 233-244. 68. G.W. Luli and W.R. Strohl, Comparison of Growth, Acetate Production, and Acetate Inhibition of Escherichia-Coli Strains in Batch and Fed-Batch Fermentations. Applied and Environmental Microbiology, 1990. 56(4): p. 1004-1011. 69. J. Zaldivar and L.O. Ingram, Effect of organic acids on the growth and fermentation of ethanologenic Escherichia coli LY01. Biotechnology and Bioengineering, 1999. 66(4): p. 203- 210. 70. A.J. Roe, Perturbation of Anion Balance during Inhibition of Growth of Escherichia coli by Weak Acids. Journal of bacteriology, 1998. 180(4): p. 767. 71. D. McLaggan, J. Naprstek, E.T. Buurman, and W. Epstein, Interdependence of K+ and Glutamate Accumulation During Osmotic Adaptation of Escherichia-Coli. Journal of Biological Chemistry, 1994. 269(3): p. 1911-1917.

46 72. A.J. Roe, C. O'Byrne, D. McLaggan, and I.R. Booth, Inhibition of Escherichia coli growth by acetic acid: a problem with methionine biosynthesis and homocysteine toxicity. Microbiology-Sgm, 2002. 148: p. 2215-2222. 73. D.D. Axe and J.E. Bailey, Transport of Lactate and Acetate through the Energized Cytoplasmic Membrane of Escherichia-Coli. Biotechnology and Bioengineering, 1995. 47(1): p. 8-19. 74. F. Diez-Gonzalez and J.B. Russell, The ability of Escherichia coli O157:H7 to decrease its intracellular pH and resist the toxicity of acetic acid. Microbiology, 1997. 143 ( Pt 4): p. 1175-80. 75. F. Diez-Gonzalez and J.B. Russell, Effects of carbonylcyanide-m-chlorophenyl- hydrazone (CCCP) and acetate on Escherichia coli O157:H7 and K-12: uncoupling versus anion accumulation. FEMS microbiology letters, 1997. 151(1): p. 71-6. 76. R.K. Poole, Bacterial responses to pH - Summary, in Bacterial Response to Ph. 1999. p. 251- 255. 77. J.L. Brown, T. Ross, T.A. McMeekin, and P.D. Nichols, Acid habituation of Escherichia coli and the potential role of cyclopropane fatty acids in low pH tolerance. International Journal of Food Microbiology, 1997. 37(2-3): p. 163-173. 78. Y.Y. Chang and J.E. Cronan, Membrane cyclopropane fatty acid content is a major factor in acid resistance of Escherichia coli. Molecular microbiology, 1999. 33(2): p. 249-59. 79. E. Padan and S. Schuldiner, Molecular physiology of Na+/H+ antiporters, key transporters in circulation of Na+ and H+ in cells. Biochimica et biophysica acta, 1994. 1185(2): p. 129-51. 80. E. Padan, Y. Gerchman, A. Rimon, A. Rothman, N. Dover, and O. Carmel-Harel, The molecular mechanism of the regulation of the NhaA Na+/H+ antiporter of Escherichia coli, a key transporter of in the adaptation to Na+ and H+. Novartis Foundation symposium, 1999. 221: p. 183-199. 81. R.G. Kroll and I.R. Booth, The relationship between intracellular pH, the pH gradient and potassium transport in Escherichia coli. Biochemical Journal, 1983. 216: p. 709-716. 82. R.R.G. Kroll and I.I.R. Booth, The role of potassium transport in the generation of a pH gradient in Escherichia coli. The Biochemical journal, 1981. 198(3): p. 691-8. 83. M.P. Castanie-Cornet, T.A. Penfound, D. Smith, J.F. Elliott, and J.W. Foster, Control of acid resistance in Escherichia coli. Journal of Bacteriology, 1999. 181(11): p. 3525-3535. 84. H.T. Richard and J.W. Foster, Acid resistance in Escherichia coli, in Advances in Applied Microbiology, Vol 52. 2003. p. 167-186.

47 85. J.S. Lin, M.P. Smith, K.C. Chapin, H.S. Baik, G.N. Bennett, and J.W. Foster, Mechanisms of acid resistance in enterohemorrhagic Escherichia coli. Applied and Environmental Microbiology, 1996. 62(9): p. 3094-3100. 86. S. Bearson, Acid stress responses in enterobacteria. FEMS microbiology letters, 1997. 147(2): p. 173-180. 87. J. Lin, I.S. Lee, J. Frey, J.L. Slonczewski, and J.W. Foster, Comparative analysis of extreme acid survival in Salmonella typhimurium, Shigella flexneri, and Escherichia coli. Journal of bacteriology, 1995. 177(14): p. 4097-104. 88. B.O. Hersh, A glutamate-dependent acid resistance gene in Escherichia coli. Journal of bacteriology, 1996. 178(13): p. 3978. 89. R. Iyer, Arginine-Agmatine Antiporter in Extreme Acid Resistance in Escherichia coli. Journal of bacteriology, 2003. 185(22): p. 6556. 90. S. Gong, YjdE (AdiC) Is the Arginine: Agmatine Antiporter Essential for Arginine- Dependent Acid Resistance in Escherichia coli. Journal of bacteriology, 2003. 185(15): p. 4402. 91. H. Richard and J.W. Foster, Escherichia coli glutamate- and arginine-dependent acid resistance systems increase internal pH and reverse transmembrane potential. Journal of Bacteriology, 2004. 186(18): p. 6032-6041. 92. K.S. Gajiwala and S.K. Burley, HDEA, a periplasmic protein that supports acid resistance in pathogenic enteric bacteria. Journal of molecular biology, 2000. 295(3): p. 605-12. 93. S.R. Waterman and P.L. Small, Identification of sigma S-dependent genes associated with the stationary-phase acid-resistance phenotype of Shigella flexneri. Molecular microbiology, 1996. 21(5): p. 925-40. 94. P.W. Atkins, Physical chemistry. 6th ed. 1999, Oxford: Oxford university press. 95. D.J. Shaw, Colloid & surface chemistry. 4th ed. 1992, Oxford: Butterworth- Heinemann. 96. W.W. Baldwin, R. Myer, T. Kung, E. Anderson, and A.L. Koch, Growth and buoyant density of Escherichia coli at very low osmolarities. Journal of bacteriology, 1995. 177(1): p. 235-7. 97. S. Cayley, B.A. Lewis, H.J. Guttman, and M.T. Record, Characterization of the cytoplasm of Escherichia coli K-12 as a function of external osmolarity. Implications for protein-DNA interactions in vivo. Journal of molecular biology, 1991. 222(2): p. 281-300.

48 98. M.T. Record, E.S. Courtenay, D.S. Cayley, and H.J. Guttman, Responses of E. coli to osmotic stress: large changes in amounts of cytoplasmic solutes and water. Trends in biochemical sciences, 1998. 23(4): p. 143-8. 99. M.T. Madigan and J.M. Martino, Brock Biology of microorganisms. 11th ed. 2006, Upper Saddle River: Pearson Prentice Hall. 100. L.N. Csonka, Physiological and Genetic Responses of Bacteria to Osmotic-Stress. Microbiological Reviews, 1989. 53(1): p. 121-147. 101. N.C. Carpita, Tensile Strength of Cell Walls of Living Cells. Plant physiology, 1985. 79(2): p. 485-488. 102. J.B. Stock, B. Rauch, and S. Roseman, Periplasmic space in Salmonella typhimurium and Escherichia coli. The Journal of biological chemistry, 1977. 252(21): p. 7850-61. 103. W.G. Roth, M.P. Leckie, and D.N. Dietzler, Osmotic stress drastically inhibits active transport of carbohydrates by Escherichia coli. Biochemical and biophysical research communications, 1985. 126(1): p. 434-41. 104. J. Meury, Glycine betaine reverses the effects of osmotic stress on DNA replication and cellular division in Escherichia coli. Archives of microbiology, 1988. 149(3): p. 232-9. 105. T. Ohwada and S. Sagisaka, An immediate and steep increase in ATP concentration in response to reduced turgor pressure in Escherichia coli B. Archives of biochemistry and biophysics, 1987. 259(1): p. 157-63. 106. W.G. Roth, S.E. Porter, M.P. Leckie, B.E. Porter, and D.N. Dietzler, Restoration of cell volume and the reversal of carbohydrate transport and growth inhibition of osmotically upshocked Escherichia coli. Biochemical and biophysical research communications, 1985. 126(1): p. 442-9. 107. B. Poolman and E. Glaasker, Regulation of compatible solute accumulation in bacteria. Molecular microbiology, 1998. 29(2): p. 397-407. 108. W.S. Epstein, Osmoregulation by potassium transport in Escherichia coli. FEMS microbiology reviews, 1986. 39: p. 73. 109. L.A. Laimins, D.B. Rhoads, and W. Epstein, Osmotic control of kdp operon expression in Escherichia coli. Proceedings of the National Academy of Sciences of the United States of America, 1981. 78(1): p. 464-8. 110. G. May, E. Faatz, M. Villarejo, and E. Bremer, Binding protein dependent transport of glycine betaine and its osmotic regulation in Escherichia coli K12. Molecular & general genetics, 1986. 205(2): p. 225-33.

49 111. I.R. Booth, J. Cairney, L. Sutherland, and C.F. Higgin, Enteric bacteria and osmotic stress: an integrated homeostatic system. Symposium series, 1988. 17: p. 35-49S. 112. W.G. Roth, Restoration of colony-forming activity in osmotically stressed Escherichia coli by betaine. Applied and environmental microbiology, 1988. 54(12): p. 3142. 113. B. Landfald, Choline-glycine betaine pathway confers a high level of osmotic tolerance in Escherichia coli. Journal of bacteriology, 1986. 165(3): p. 849. 114. B. Perroud and D.L. Rudulier, Glycine betaine transport in Escherichia coli: osmotic modulation. Journal of bacteriology, 1985. 161(1): p. 393. 115. S. Nagata, H. Sasaki, A. Oshima, S. Takeda, Y. Hashimoto, and A. Ishida, Effect of proline and K+ on the stimulation of cellular activities in Escherichia coli K-12 under high salinity. Bioscience, biotechnology, and biochemistry, 2005. 69(4): p. 740-6. 116. M. Jebbar, Osmoprotection of Escherichia coli by ectoine: uptake and accumulation characteristics. Journal of bacteriology, 1992. 174(15): p. 5027. 117. C.P. O'Byrne and I.R. Booth, Osmoregulation and its importance to food-borne microorganisms. International journal of food microbiology, 2002. 74(3): p. 203-16. 118. A. Ishida, Y. Kawatake, and N. Ono, Osmotic-Stress Conditioning for Induction of Acquired Osmotolerance in Escherichia-Coli. Journal of General and Applied Microbiology, 1994. 40(1): p. 35-42. 119. I. Poirier, P.A. Marechal, C. Evrard, and P. Gervais, Escherichia coli and Lactobacillus plantarum responses to osmotic stress. Applied Microbiology and Biotechnology, 1998. 50(6): p. 704-709. 120. B.S. Dien, N.N. Nichols, and R.J. Bothast, Fermentation of sugar mixtures using Escherichia coli catabolite repression mutants engineered for production of L-lactic acid. Journal of Industrial Microbiology & Biotechnology, 2002. 29(5): p. 221-227. 121. V. Hernandez-Montalvo, F. Valle, F. Bolivar, and G. Gosset, Characterization of sugar mixtures utilization by an Escherichia coli mutant devoid of the phosphotransferase system. Applied Microbiology and Biotechnology, 2001. 57(1-2): p. 186-191. 122. J. Stulke and W. Hillen, Carbon catabolite repression in bacteria. Current Opinion in Microbiology, 1999. 2(2): p. 195-201. 123. N.N. Nichols, B.S. Dien, and R.J. Bothast, Use of catabolite repression mutants for fermentation of sugar mixtures to ethanol. Applied Microbiology and Biotechnology, 2001. 56(1-2): p. 120-125. 124. J. Shiloach and R. Fass, Growing E-coli to high cell density - A historical perspective on method development. Biotechnology Advances, 2005. 23(5): p. 345-357.

50 125. D. Riesenberg, High-Cell-Density Cultivation of Escherichia-Coli. Current Opinion in Biotechnology, 1991. 2(3): p. 380-384. 126. Y.C. Liu and K.Y. Cheng, Effect of ammonium concentration on the cultivation of recombinant Escherichia coli producing penicillin G acylase. Journal of the Chinese Institute of Chemical Engineers, 2000. 31(6): p. 601-607. 127. G. Naundorf and N.G. Aumen, Ammonia-Induced Cell-Envelope Injury in Escherichia-Coli and Enterobacter-Aerogenes. Canadian Journal of Microbiology, 1990. 36(8): p. 525-529. 128. C. Andersson, D. Hodge, K.A. Berglund, and U. Rova, Effect of Different Carbon Sources on the Production of Succinic Acid Using Metabolically Engineered Escherichia coli. Biotechnol. Prog., 2007.

51 52 Paper I

Effect of Different Carbon Sources on the Production of Succinic Acid Using Metabolically Engineered Escherichia coli

Christian Andersson, David Hodge, Kris A. Berglund, and Ulrika Rova*

Division of Biochemical and Chemical Process Engineering, Luleå University of Technology, SE-971 87 Luleå, Sweden

Succinic acid (SA) is an important platform molecule in the synthesis of a number of commodity and specialty chemicals. In the present work, dual-phase batch fermentations with the E. coli strain AFP184 were performed using a medium suited for large-scale industrial production of SA. The ability of the strain to ferment different sugars was investigated. The sugars studied were sucrose, glucose, fructose, xylose, and equal mixtures of glucose and fructose and glucose and xylose at a total initial sugar concentration of 100 g L-1. AFP184 was able to utilize all sugars and sugar combinations except sucrose for biomass generation and succinate production. For sucrose as a substrate no succinic acid was produced and none of the sucrose was metabolized. The succinic acid yield from glucose (0.83 g succinic acid per gram glucose consumed anaerobically) was higher than the yield from fructose (0.66 g g-1). When using xylose as a carbon source, a yield of 0.50 g g-1 was obtained. In the mixed-sugar fermentations no catabolite repression was detected. Mixtures of glucose and xylose resulted in higher yields (0.60 g g-1) than use of xylose alone. Fermenting glucose mixed with fructose gave a lower yield (0.58 g g-1) than fructose used as the sole carbon source. The reason is an increased pyruvate production. The pyruvate concentration decreased later in the fermentation. Final succinic acid concentrations were in the range of 25-40gL-1. Acetic and pyruvic acid were the only other products detected and accumulated to concentrations of 2.7-6.7 and 0-2.7gL-1. Production of succinic acid decreased when organic acid concentrations reached approximately 30 g L-1. This study demonstrates that E. coli strain AFP184 is able to produce succinic acid in a low cost medium from a variety of sugars with only small amounts of byproducts formed.

Introduction the cheapest available raw material. Glucose and xylose are the most abundant sugars in lignocellulosic biomass, while fructose Sustainable production of fuels and chemicals is driven by and glucose are the components of sucrose, a widely available the need to control atmospheric concentrations of greenhouse disaccharide. The third and most important factor for the gases such as carbon dioxide, depletion of existing oil reserves, succinic acid economy, as described in the USDOE report, is and increasing oil prices. In a report from the U.S. Department the volumetric productivity. To make succinic acid production of Energy (USDOE), succinic acid was considered as one of feasible, volumetric productivities above 2.5 g L-1 h-1 will be 12 top chemical building blocks manufactured from biomass necessary (1). (1). Succinic acid has a wide range of applications and can be In 1949 Stokes demonstrated that E. coli under anaerobic a platform for deriving many commodity and specialty chemi- conditions produces a mixture of organic acids and ethanol (11). cals, for example, diesel fuel additives, deicers, biodegradable Typical yields from such fermentations are 0.8 mol ethanol, polymers and solvents, and detergent builders (2). 1.2 mol formic acid, 0.1-0.2 mol lactic acid, and 0.3-0.4 mol Succinic acid can be produced by a number of organisms succinic acid per mole glucose consumed. If the objective is to including Bacteroides ruminicola and Bacteroides amylophilus, produce succinic acid, the yield of succinic acid relative the Anaerobiospirillum succiniciproducens, Actinobacillus succi- other acids must be increased. The USDOE initiated in the late nogenes, Mannheimia succiniciproducens, and Escherichia coli 1990s the Alternative Feedstock Program (AFP) with the aim - (3 9). Corynebacterium glutamicum has also been shown to to develop a number of metabolically engineered E. coil strains produce mixtures of lactic and succinic acid (10). The work with increased succinic acid production. The two most interest- done so far have all used media supplemented to some degree ing mutants developed by the program were AFP111 and with high-cost nutrient sources such as tryptone, peptone, and AFP184 (9, 12, 13). AFP111 is a spontaneous mutant with yeast extract. mutations in the glucose specific phosphotransferase system In order to develop a bio-based industrial production of (ptsG), the pyruvate formate lyase system (pfl), and the succinic acid, three main issues need to be addressed. First, a fermentative lactate dehydrogenase system (ldh)(12). The low cost medium must be used. Second, the producing organism mutations resulted in increased succinic acid yields (1 mol must be able to utilize a wide range of sugar feedstocks, succinic acid per mole glucose) (13). AFP184 is a metabolically producing succinic acid in good yields in order to make use of engineered strain where the three mutations described above were deliberately inserted into the E. coli strain C600 (ATCC * To whom correspondence should be addressed. Ph: +46920491315. 23724), which can ferment both five- and six-carbon sugars and Fax: +46920491199. Email: [email protected]. has strong growth characteristics (9). Beside these organisms,

10.1021/bp060301y CCC: $37.00 © xxxx American Chemical Society and American Institute of Chemical Engineers Published on Web 01/25/2007 PAGE EST: 7.2 B a number of different E. coli mutants have been developed and Biotechnology, The Netherlands) to determine if the strain could tested for succinate production including pioneering work to grow and produce succinic acid when using sucrose as the produce succinate under aerobic conditions (14-22). In order carbon source. The fermentation procedure was the same as in to improve succinate yield, other studies used recombinant the 12 L bioreactor. Three fermentations in the 1 L bioreactors AFP111 overexpressing heterologous pyruvate carboxylase in were also carried out to determine the specific growth rates of complex media with the main components tryptone (20 g L-1) AFP184 growing on glucose, fructose, and xylose, respectively. and yeast extract (10 g L-1), which resulted in mass yields of Samples were taken during the aerobic phase and analyzed for 1.10 and productivities of 1.3 g L-1 h-1 (23, 24). In the present optical density. No anaerobic phase was carried out. work the aim was to investigate the fermentation characteristics Analysis. Cell Growth. Cell growth was monitored by of AFP184 in a medium consisting of corn steep liquor, measuring the optical density at 550 nm (OD550). Optical inorganic salts, and different sugars without supplementation densities were then converted to dry cell weight (DCW, g L-1) of other additional nutrients. Corn steep liquor is a byproduct using the following OD-DCW calibration curve developed in from the corn wet-milling industry, but as opposed to yeast our laboratory: extract and peptone it is an inexpensive source of vitamins and ) + trace elements. The composition of corn steep liquor can vary DCW 0.324OD550 1.687 widely depending mainly on the variables involved in starch processing and the type and condition of the corn, but averages To establish the number of viable cells, 10-fold serial dilutions can be found in the literature (25, 26). For the purpose of this of the fermentation samples were plated on tryptone soy agar ° study sucrose, glucose, fructose, xylose, and mixtures of glucose plates and incubated overnight at 37 C. The number of colonies and fructose and glucose and xylose were used as carbon was calculated, and the number of viable cells was expressed sources. This work addresses the effect of different sugars on as cells per milliliter fermentation broth. succinic acid kinetics and yields in an industrially relevant Organic Acid and Sugar Analysis. The organic acid con- medium. centrations were measured by HPLC (Series 200 Quaternary LC pump and UV-vis detector, Perkin-Elmer) equipped with Materials and Methods a C18 column (Ultra Aqueous, 5 μm, 150 mm × 4.6 mm, Strain and Inoculum Preparation. The E. coli strain Restek) using a 50 mM KH2PO4 buffer with 2% acetonitrile - AFP184, which lacks functional genes coding for pyruvate (pH 2.5 adjusted by HCl) at a flow rate of 0.35 mL min 1 as formate lyase, fermentative lactate dehydrogenase, and the the mobile phase, with 20 μL of sample injected manually for glucose phosphotransferase system (9), was used in this study. analysis with UV detection at 210 nm. Samples for acid analysis Cultures were stored at -80 °C as 30% glycerol stocks. The were centrifuged at 10 000 rpm for 10 min at 4 °C. The inoculum was prepared by inoculating four 500-mL shake flasks supernatant was diluted with the mobile phase and 37% containing 100 mL sterile medium (same medium as used in hydrochloric acid (volume sample:buffer:HCl ) 0.1:0.8:0.1 mL) the biorector; see below) with 200 μL of the glycerol stock and filtered through a 0.2 μm syringe filter. Sugar concentra- culture. The inoculated flasks were incubated for 16 h at 37 °C tions were determined using the same HPLC system as above (200 rpm). but equipped with a Series 200 refractive index (RI) detector Fermentations. Batch fermentations were conducted in a 12 (Perkin-Elmer), guard column, and an ion exchange column L bioreactor (BR12, Belach Bioteknik AB, Sweden) with a total (Aminex HPX87-P, BioRad). The column was kept at 85 °Cin starting volume of 8 L (including 0.4 L inoculum and 2 L sugar a column oven for optimal performance. Prior to analysis, the solution). The medium used for strain AFP184 contained the samples were centrifuged at 10 000 rpm for 10 min at 4 °C. -1 following components (g L ): K2HPO4, 1.4; KH2 PO4, 0.6; The supernatant was diluted with water and filtered through a μ μ (NH4)2SO4, 3.3; MgSO4‚7H2O, 0.4; corn steep liquor (50% 0.2 m syringe filter, and 20 L of sample was injected - solids, Sigma-Aldrich), 15; and 3 mL antifoam agent (Antifoam manually. Water at a flow rate of 0.6 mL min 1 was used as 204, Sigma-Aldrich). The medium was sterilized in the fer- the mobile phase. All data was collected and processed using menter at 121 °C for 20 min; thereafter2Lofaseparately Perkin-Elmer’s TotalChrom analytical software. Peak areas sterilized sugar solution (400 g L-1) and 400 mL inoculum were from the chromatograms were evaluated by comparison to aseptically added to a final volume of 8 L and a total sugar standard curves prepared from solutions with known concen- concentration of 100 g L-1. The fermentation temperature was trations of organic acids (formic, pyruvic, malic, lactic, acetic, controlled at 37 °C, and the pH was maintained between 6.6 succinic, and fumaric acid), sucrose, glucose, fructose, and and 6.7 by automatic addition of NH4OH (15% NH3 solution). xylose. The dissolved oxygen concentration (%DO), measured by a pO2 electrode, was regulated by varying the agitation speed between Results 500 and 1000 rpm. The total fermentation time was 32 h and Fermentations with strain AFP184 using sucrose, glucose, consisted of an aerobic growth phase and an anaerobic produc- fructose, xylose, and equal mixtures of glucose/fructose and tion phase. During the aerobic growth phase (7-9 h) the culture glucose/xylose were performed and analyzed. medium was aerated with an air flow of 10 L min-1. When the Single-Sugar Fermentations. AFP184 was able to use optical density at 550 nm had reached a value of 30-35, the glucose, fructose, and xylose, but not sucrose, as a carbon source anaerobic production phase was initiated by withdrawing the for biomass and succinic acid production (Figure 1a-c). During air supply and sparging the culture medium with CO2 at a flow sucrose fermentation no sugar was metabolized, and sucrose rate of3Lmin-1. The agitation speed was set to 500 rpm for was therefore excluded from further analysis. Specific growth the anaerobic phase. Succinic acid was produced during this rates during the aerobic phase were calculated using data anaerobic production phase, which continued for the remainder collected from experiments carried out with glucose, fructose, of the fermentation. During the fermentations, samples were and xylose in 1 L bioreactors (Table 1). It was observed that in aseptically withdrawn for analysis of optical density, viable cells, fermentations using glucose and fructose exponential growth and sugars and organic acids concentrations. Sucrose was tested was initiated within 1 h after inoculation, but when xylose was with strain AFP184 ina1Lbioreactor (Biobundle 1L, Applikon used exponential growth did not start until after 4 h. Once C

Figure 1. Sugar, product, and cell concentration profiles for (a) glucose, (b) fructose, (c) xylose, (d) glucose/fructose, and (e) glucose/xylose fermentations. Product and sugar concentrations are given in g L-1, dry cell weights in g L-1 and viable cells in (cells mL-1) × 109. Symbols used: glucose (4), fructose ()), xylose (O), succinic acid (b), pyruvic acid (9), viable cells (/), and dry cell weight (×). Staggered line indicates time of switch to the anaerobic phase.

Table 1. Summary of Fermentation Parameters for AFP184a -1 -1 -1 -1 -1 -1 d -1 -1 d sugar μ (h ) YP/S (g g ) anaerobic QP (g L h ) overall QP (g L h ) anaerobic, T22 QP (g L h ) anaerobic, T32 glucose 0.42 0.83 1.27 2.89 1.76 fructose 0.50 0.66 1.01 1.79 1.26 xylose 0.57 0.50 0.78 1.49 1.08 glu/frub 0.58 0.79 1.67 1.05 glu/xylc 0.60 0.86 1.48 1.08

a Specific growth rates were obtained from fermentations carried out in 1 L bioreactors. YP/S is the mass yield of succinic acid based on the glucose b consumed in the anaerobic phase, and QP is the volumetric productivity of succinic acid during the anaerobic phase. glu/fru is an abbreviation for glucose/ fructose. c glu/xyl is an abbreviation for glucose/xylose. d Anaerobic productivity at 22 and 32 h total fermentation time, respectively. growing exponentially xylose resulted in the highest specific increased slightly as an effect of continued growth. After growth rate and glucose in the lowest. It should be noted that approximately 20 h the viable cell density started to decrease. when glucose was used a slow increase in dry cell mass was Succinic, acetic, and pyruvic were the only acids produced observed during the first7hoftheanaerobic phase. in concentrations over1gL-1. The highest succinic acid Viable cell counts were made for all of the different concentration, 40 g L-1, was detected after 32 h when glucose fermentations (Figure 1a-c). During the first hours of the was used as the carbon source. After 32 h the succinate anaerobic phase the number of viable cells was constant or production did not increase significantly (data not shown), and D

Figure 2. Succinic acid productivities for single-sugar fermentations in grams succinate per viable cell as a function of (a) time and (b) succinic acid concentration. Symbols used in the figure: glucose fermentation (4), fructose fermentation (]), and xylose fermentation (O). hence 32 h was selected as the final fermentation time. When centrations of approximately 1-3gL-1. Acetate levels increased fructose and xylose were used, the final succinic acid concentra- slowly during both fermentations. Less acetate was formed when tions reached values of 32 and 25 g L-1, respectively. These fructose was used, 2.7 g L-1 compared to 6.7 g L-1 when xylose fermentations also resulted in lower acetate concentrations but was used. produced pyruvate. The highest yield, 0.83 g succinate per gram The viable cell densities increased during the aerobic phases sugar consumed in the anaerobic phase, was obtained when and were approximately constant during the anaerobic phases glucose was used as a carbon source. The lowest yield was up until 20 h of the fermentations had passed, where after they obtained from xylose (0.50 g g-1). The yield from fructose was decreased (Figure 1d and e). 0.66 g g-1. Productivities during mixed-sugar fermentation are presented The overall volumetric productivities (QP) are presented in in the same way as for single-sugar fermentations (Figure 3). Table 1. Productivities per viable cell were calculated as When using gluose/fructose the productivity per viable cell as functions of time and succinic acid concentrations, respectively a function of time was initially high (Figure 3a). A major (Figure 2). Productivities per viable cell as function of time for productivity decrease occurred at a succinate concentration of fermentations using fructose and xylose were initially high but 20gL-1 (Figure 3b). For glucose and xylose the productivity decreased substantially during the first hours of anaerobic was constant throughout the anaerobic phase (Figure 3a and conditions (Figure 2a). During the remainder of the fermenta- b). tions productivities remained relatively constant. Glucose fermentations can be distinguished from fermentations on the Discussion other sugars as the productivity per viable cell showed only a The aim of the current work was to investigate the fermenta- marginal productivity decrease up until 16 h, after which it tion characteristics of E. coli strain AFP184 in a corn steep decreased significantly. When investigating productivities as a liquor based medium using different mono- and disaccharides function of succinic acid concentration, the fructose and xylose as carbon source. From the cell densities obtained after8hof fermentations resulted in, after a strong initial productivity aerobic growth and under the conditions used, it could be decrease, slightly decreasing productivities as the concentration concluded that AFP184 grew on all sugars and sugar combina- of succinic acid increased (Figure 2b). Glucose on the other tions except sucrose. However, specific growth rate on glucose hand gave a constant productivity until a succinate concentration was slower than for growth on xylose or fructose. This result is - of approximately 30 g L 1 was reached, after which the most likely an effect of the mutation in the phosphotransferase productivity decreased in an almost linear manner. system (PTS) of AFP184. The PTS is a system of enzymes Mixed-Sugar Fermentations. Strain AFP184 grew well in responsible for uptake and phosphorylation of different sugars both glucose/fructose and glucose/xylose mixtures. No increase in bacteria (27, 28). The PTS is sugar-specific, and the mutation in dry cell weight was observed during the anaerobic phase in AFP184 is affecting EIICBglc, a glucose specific permease (Figure 1d and e). An interesting observation is that the substrate in the PTS (12). There is also a phosphotransferase permease consumption rate was higher for both fructose and xylose than for fructose (27, 29), but in strain AFP184 this permease is not for glucose in the aerobic phase when AFP184 was grown on affected by any mutations. Apart from the PTS, E. coli can also mixed sugars. In the anaerobic phase glucose and fructose were transport glucose by galactose permease and phosphorylate it metabolized at similar rates, but when glucose was mixed with by the enzyme glucokinase (30). Glucokinase is not subjected xylose, xylose was metabolized faster during the first 16 h of to any mutations in AFP184. According to the literature glucose the fermentation. After 16 h the consumption rates of glucose uptake and phosphorylation by glucokinase is slower than by and xylose were approximately equal. The succinic acid yields the PTS (31), which explains why the growth rate on fructose in the anaerobic phase were 0.58 and 0.60 g succinic acid per is higher than the growth rate on glucose. Unlike fructose, xylose gram sugar consumed for glucose/fructose and glucose/xylose is not transported into the cells via the PTS. Instead xylose mixtures, respectively (Table 1). The final succinic acid uptake is facilitated by chemiosmotic effects (32, 33). Once concentrations were in the range of 25-27gL-1, and during inside the cell it is isomerized by xylose isomerase to xylulose both mixed-sugar fermentations between 8 and 12 g L-1 of and phosphorylated by xylulose kinase to xylulose 5-phosphate. pyruvate was formed (Figure 1d and e). Most of the pyruvate Xylulose 5-phosphate then enters the pentose phosphate pathway was excreted during the first 2-4 h of the anaerobic phase and and can enter the glycolytic pathway after conversion to either at a rate similar to the succinic acid production rate. The fructose 6-phosphate or glyceraldehyde 3-phosphate. Interest- pyruvate concentration decreased in both cases to final con- ingly, growth on xylose was initially very slow, whereas the E

Figure 3. Succinic acid productivities for mixed-sugar fermentations in grams succinate per viable cells as a function of (a) time and (b) succinic acid concentration. Symbols used in the figure: Glucose/fructose fermentation (]), and glucose/xylose fermentation (O). use of glucose and fructose rapidly initiated exponential growth. phoenolpyruvate (PEP) is converted to pyruvate by the PTS Because the inocula were grown in a glucose medium, glucose (29, 37), with 1.0 mol fructose yielding 1.0 mol PEP and 1.0 and fructose fermentations may already have the enzymes mol pyruvate due to transport requirements. This metabolite required for utilization of the respective sugars, whereas in the imbalance prevents a balanced redistribution of reducing case of xylose expression of the enzymes first needs to be equivalents, and as a result pyruvate will accumulate and be induced before exponential growth can start. excreted (38) as was observed in Figure 1b. Glucose mixed with When glucose is mixed with other sugars, E. coli usually fructose should provide an increased pool of PEP to balance consumes it first, and after all glucose has been metabolized the pyruvate generated by fructose transport and result in higher the utilization of other sugars is initiated, a phenomenon called yields than fructose alone. However, experimental results do catabolite repression. From the sugar consumption (Figure 1) not confirm this hypothesis (Table 1), although much more it was observed that in glucose/fructose mixtures, fructose was pyruvic acid was excreted initially with fructose/glucose metabolized slightly faster during the aerobic phase. This effect mixtures relative to fermentation of pure fructose (12.4 vs 5.2 was expected as a result of the mutation strain AFP184 has in gL-1). The accumulation of higher concentrations of pyruvic the glucose PTS. More interesting is the observation that when acid for the mixed substrate fermentation is suspected to be due glucose and xylose were mixed, the sugars were fermented to the more rapid uptake of both sugars, resulting in a faster simultaneously but xylose was consumed at a substantially intracellular accumulation of pyruvate with no available electron higher rate. Other studies of glucose/xylose fermentations by sink for further metabolism. For both fermentations, most of E. coli PTS mutants show similar results where glucose and this excreted pyruvic acid was taken up again, requiring the xylose are co-metabolized. The effect is attributed to a mutation expenditure of reducing equivalents for active transport (39). in the PTS affecting EIICBglc, which in turn prevents glucose This excretion and reutilization of pyruvate can explain both from repressing uptake of other sugars (30, 34-36). the low yields and the lower rates found for fructose/glucose To estimate maximum yields and optimal metabolic flows fermentation. between intermediates, redox balances were carried out. The A redox balance for xylose gives a maximum theoretical yield anaerobic glucose, fructose, and xylose metabolism of AFP184 of 1.43 mol succinate per mole xylose. In the case of xylose for maximal succinate production is presented in simplified form only about 45% of the theoretical yield was achieved (0.64 mol in Figure 4. From a redox balance for the anaerobic phase the mol-1). By carrying out a carbon balance, 60% of the carbon maximum succinate yield from 1 mol of glucose is 1.71 mol can be accounted for. Further work is required to identify the (1.12 g succinate per gram glucose), which is in accordance remaining carbon. When xylose was mixed with glucose, the with previous results (23). yield increased to 0.83 mol succinate per mole sugar, which The redox balance assumes activity of the enzyme isocitrate was expected because glucose gives higher yields. lysase, a key enzyme in the glyoxylate shunt. It has previously When glucose was used as the carbon source, the productivity been reported that AFP111 shows isocitrate lyase activity after per viable cell decreased drastically after 16 h (Figure 2a). aerobic growth (23). The yield obtained in this study from Succinic acid productivity as a function of succinic acid glucose was 1.27 mol succinate per mole glucose, which is 74% concentration showed a sharp decrease in productivity at of the theoretical. If generation of new cells during the anaerobic approximately 30 g L-1 (Figure 2b). The succinic acid concen- phase is considered, a carbon balance can account for more than tration at 16 h corresponds to almost 30 g L-1. From this result 95% of the carbon. it can be concluded that as long as the cells are viable, they The theoretical yield from fructose is 1.20 mol succinate per retain their productivity until the concentration of succinic acid mole fructose, which corresponds to a mass yield of 0.79, is not too high. Fructose and xylose did not show any drastic although this requires another electron acceptor such as ethanol, decreases in productivity (Figure 2). Neither of these fermenta- which was not detected. A redox balance without considering tions achieved succinate concentrations of more than 20-25 g ethanol yields 1 mol succinate and 1 mol pyruvate per mole L-1 during the first 20 h of the fermentations (Figure 1b and fructose. A molar yield of 1.00 was achieved (mass yield 0.66) c). In the mixed-sugar fermentations glucose/xylose mixtures for succinate from pure fructose. Redox balances under anaero- did not result in very high succinate concentrations and no bic conditions indicate that 2.5 times greater flux through the decrease in productivity was observed (Figure 3c and d). When reductive arm is required to match the flux through the oxidative glucose and fructose were mixed and fermented, a sharp TCA plus the glyoxylate shunt for either glucose or fructose as productivity decrease was observed after 16 h (Figure 3a) when a substrate. During fructose fermentation by AFP184, phos- the concentration of succinate was approximately 20 g L-1 F

Figure 4. Anaerobic glucose, fructose, and xylose metabolism of AFP184 for maximal succinate production. Note that some metabolites are excluded. The reactions blocked by the mutations in AFP184 are indicated by crosses.

(Figure 1d). When glucose was used as the carbon source, the achieved ammonium concentrations. Possible ammonia productivities did not decline until concentrations of 30 g L-1 inhibition is being further investigated in studies where different were reached. In the other fermentations the concentrations of bases are compared to better understand the causes of the end products other than succinate were low, but in this observed productivity decrease. fermentation the concentration of pyruvic acid was 10 g L-1 Finally, a comment should be made regarding the volumetric after 16 h and the total concentration of end products exceeded productivities achieved. As mentioned in the Introduction, 30gL-1 (Figure 1d). From these results it seems likely that biological production of succinic acid should preferably reach high final concentrations of end products inhibit productivity. volumetric productivities of 2.5 g L-1 h-1 to be feasible. In Many investigations regarding the inhibition effects by organic this study volumetric productivities during anaerobic production acids on E. coli growth and productivity have been carried out in the range of 1.5-2.9gL-1 h-1 were demonstrated. The (40-42). The effects of organic acid inhibition is greatest at reason the productivities reach almost3gL-1 h-1 is because low pH when the acids are undissociated, but even at neutral the cells after the switch to the anaerobic phase are not sugar- pH high concentrations of the acids have shown inhibitory limited. A high initial sugar load was used, but the same results effects (42). The fermentations in this study were carried out could be achieved by feeding a high concentration sugar stream with ammonium hydroxide as base for neutralization of end after the end of the aerobic phase. The total volumetric products, and it has been reported that ammonium in concentra- productivity is directly dependent on the cell density, and to tions above3gL-1 inhibits growth (43, 44). In the present achieve a high cell density requires the use of substantial work ammonium concentrations of approximately 10 g L-1 were amounts of the carbon source. This of course could compromise obtained at the end of the fermentations with high succinate the overall process economics, and the necessary cell density production. In other studies changes in the growth characteristics for feasible production of succinic acid must be considered if of E. coli under different ammonium sulfate concentrations was the technology is to be applied to a large scale production investigated, and ammonium sulfate concentrations of 5% had facility. to be used before any severe inhibition was observed (45). When the fermentations carried out in the present work showed Conclusions productivity decreases, they were in the anaerobic phase and In this paper we have demonstrated growth and succinate only very little growth would take place even without any production of E. coli strain AFP184 in a medium containing inhibition; it is not clear if productivity would be affected by corn steep liquor, salts, and different mono- and disaccharides. G

The reduction of complex components in the medium has a (17) Lin, H.; Bennett, G. N.; San, K. Y. Fed-batch culture of a positive effect on the final price of the produced succinate due metabolically engineered Escherichia coli strain designed for high- to lower cost of raw materials; however, final concentrations level succinate production and yield under aerobic conditions. Biotechnol. Bioeng. 2005, 90, 775-779. are in the range of 30-40gL-1 after 32 h and need to be (18) San, H.; Bennett, G. N.; San, K. Y. Genetic reconstruction of the improved for the process to become more economical. Studies aerobic central metabolism in Escherichia coli for the absolute will be directed to increase the final titer and to further improve aerobic production of succinate. Biotechnol. Bioeng. 2005, 89, 148- the process. 156. (19) Lin, H.; Bennett, G. N.; San, K. Y. Metabolic engineering of Acknowledgment aerobic succinate prod-action systems in Escherichia coli to improve process productivity and achieve the maximum theoretical succinate This research was supported by Diversified Natural Products yield. Metab. Eng. 2005, 7, 116-127. Inc. and the Research Council of Norrbotten. Equipment was (20) Lin, H.; San, K. Y.; Bennett, G. N. Effect of Sorghum vulgare provided by the Kempe Foundation. phosphoenolpyruvate carboxylase and Lactococcus lactis pyruvate carboxylase coexpression on succinate production in mutant strains of Escherichia coli. Appl. Microbiol. Biotechnol. 2005, 67, 515- References and Notes 523. (1) Top Value Added Chemicals from Biomass; Werpy, T., Petersen, (21) Yun, N. R.; San, K. Y.; Bennett, G. N. Enhancement of lactate G., Eds.; USDOE: Washington, DC, 2004; Vol. I. and succinate formation in adhE or pta-ackA mutants of NADH dehydrogenase-deficient Escherichia coli. J. Appl. Microbiol. 2005, (2) Zeikus, J. G.; Jain, M. K.; Elankovan, P. Biotechnology of succinic 99, 1404-1412. acid production and markets for derived industrial products. Appl. Microbiol. Biotechnol. 1999, 51, 545-552. (22) Sanchez, A. M.; Bennett, G. N.; San, K. Y. Novel pathway engineering design of the anaerobic central metabolic pathway in (3) VanderWerf, M. J.; Guettler, M. V.; Jain, M. K.; Zeikus, J. G. Escherichia coli to increase succinate yield and productivity. Metab. Environmental and physiological factors affecting the succinate Eng. 2005, 7, 229-239. product ratio during carbohydrate fermentation by Actinobacillus sp. (23) Vemuri, G. N.; Eiteman, M. A.; Altman, E. Effects of growth 130Z. Arch. Microbiol. 1997, 167, 332-342. mode and pyruvate carboxylase on succinic acid production by (4) Guettler, M. V.; Rumler, D.; Jain, M. K. Actinobacillus succino- metabolically engineered strains of Escherichia coli. Appl. EnViron. genes sp. nov., a novel succinic-acid-producing strain from the bovine Microbiol. 2002, 68, 1715-1727. - rumen. Int. J. System. Bacteriol. 1999, 49, 207 216. (24) Vemuri, G. N.; Eiteman, M. A.; Altman, E. Succinate production (5) Lee, P. C.; Lee, S. Y.; Hong, S. H.; Chang, H. N. Isolation and in dual-phase Escherichia coli fermentations depends on the time characterization of a new succinic acid-producing bacterium, Man- of transition from aerobic to anaerobic conditions. J. Ind. Microbiol. nheimia succiniciproducens MBEL55E, from bovine rumen. Appl. Biotechnol. 2002, 28, 325-332. Microbiol. Biotechnol. 2002, 58, 663-668. (25) Liggett, R. W.; Koffler, H. Corn steep liquor in microbiology. (6) Kim D. Y.; Yim, S. C.; Lee, P. C.; Lee, W. G.; Lee, S. Y.; Chang, Microbiol. Mol. Biol ReV. 1948, 12, 297-311. H. N. Batch and continuous fermentation of succinic acid from wood (26) Fermentation and Biochemical Engineering Handbook-Prin- hydrolysate by Mannheimia succiniciproducens MBEL55E. Enzyme ciples, Process Design, and Equipment, 2nd ed.; Vogel, H. C., Microb. Technol. 2004, 35, 648-653. Todaro, C. L., Eds.; William Andrew Publishing/Noyes: Westwood, (7) Nghiem, N. P.; Donnelly, M. I.; Sanville-Millard, C. Y.; Stols, L. NJ, 1997. Method for the production of dicarboxylic acids. U.S. Patent (27) Tchieu, J. H.; Norris, V.; Edwards, J. S.; Saier, M. H. The 5,869,301, 1999. complete phosphotransferase system in Escherichia coli. J. Mol. (8) Lee, S. J.; Song, H.; Lee, S. Y. Genome-based metabolic engineer- Microbiol. Biotechnol. 2001, 3, 329-346. ing of Mannheimia succiniciproducens for succinic acid production. (28) Postma, P. W.; Lengeler, J. W.; Jacobson, G. R. Phospho- Appl. EnViron. Microbiol. 2006, 72, 1939-1948. enolpyruvate-carbohydrate phosphotransferase systems of bacteria. V - (9) Donnelly, M. I.; Sanville-Millard, C. Y.; Nghiem, N. P. Method Microbiol. Re . 1993, 57, 543 594. to produce succinic acid from raw hydrolysates. U.S. Patent (29) Kornberg, H. L. Fructose transport by Escherichia coli. Philos. - 6,743,610, 2004. Trans. R. Soc. London, Ser. B 1990, 326, 505 513. (10) Okino, S.; Inui, M.; Yukawa, H. Production of organic acids by (30) Dien, B. S.; Nichols, N. N.; Bothast, R. J. Fermentation of sugar Corynebacterium glutamicum under oxygen deprivation. Appl. mixtures using Escherichia coli catabolite repression mutants Microbiol. Biotechnol. 2005, 68, 475-480. engineered for production of L-lactic acid. J. Ind. Microbiol. Biotechnol. 2002, 29, 221-227. (11) Stokes, J. L. Fermentation of glucose by suspensions of Escheri- (31) Curtis S. J.; Epstein, W. Phosphorylation of D-glucose in Escheri- chia coli. J. Bacteriol. 1948, 57, 147-158. chia coli mutants defective in glucosephosphotransferase, mannose- (12) Chatterjee, R.; Millard, C. S.; Champion, K.; Clark, D. P.; phosphotransferase, and glucokinase. J. Bacteriol. 1975, 122, 1189- Donnelly, M. I. Mutation of the ptsG gene results in increased 1199. production of succinate in fermentation of glucose by Escherichia V - (32) Jeffries, T. W. Utilization of xylose by bacteria, yeasts, and fungi. coli. Appl. En iron. Microbiol. 2001, 67, 148 154. AdV. Biochem. Eng./Biotechnol. 1983, 27,1-32. (13) Donnelly, M. I.; Millard, C. S.; Clark, D. P.; Chen, M. J.; Rathke, (33) Lam, V. M. S.; Daruwalla, K. R.; Henderson, P. J. F.; Jones- J. W. A novel fermentation pathway in an Escherichia coli mutant Mortimer, M. C. Proton-linked d-xylose transport in Escherichia coli. producing succinic acid acetic acid and ethanol. Appl. Biochem. J. Bacteriol. 1980, 143, 396-402. - Biotechnol. 1998, 70, 187 198. (34) Hernandez-Montalvo, V.; Valle, F.; Bolivar, F.; Gosset, G. (14) Sanchez, A. M.; Bennett, G. N.; San, K. Y. Efficient succinic Characterization of sugar mixtures utilization by an Escherichia coli acid production from glucose through overexpression of pyruvate mutant devoid of the phosphotransferase system. Appl Microbiol. carboxylase in an Escherichia coli alcohol dehydrogenase and lactate Biotechnol. 2001, 57, 186-191. dehydrogenase mutant. Biotechnol. Prog. 2005, 21, 358-365. (35) Stulke, J.; Hillen, W. Carbon catabolite repression in bacteria. (15) Lin, H.; Bennett, G. N.; San, K. Y. Chemostat culture character- Curr. Opin. Microbiol. 1999, 2, 195-201. ization of Escherichia coli mutant strains metabolically engineered (36) Nichols, N. N.; Dien, B. S.; Bothast, R. J. Use of catabolite for aerobic succinate production: A study of the modified metabolic repression mutants for fermentation of sugar mixtures to ethanol. network based on metabolite profile, enzyme activity and gene Appl. Microbiol. Biotechnol. 2001, 56, 120-125. expression profile. Metab. Eng. 2005, 7, 337-352. (37) Kornberg, H. If at first you don’t succeed... fructose utilization (16) Lin, H.; Bennett, G. N.; San, K. Y. Effect of carbon sources by Escherichia coli.AdV. Enzyme Regul. 2002, 42, 349-360. differing in oxidation state and transport route on succinate produc- (38) Causey, T. B.; Shanmugam, K. T.; Yomano, L. P.; Ingram, L. O. tion in metabolically engineered Escherichia coli. J. Ind. Microbiol. Engineering Escherichia coli for efficient conversion of glucose to Biotechnol. 2005, 32,87-93. pyruvate. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 2235-2240. H PAGE EST: 7.2

(39) Lang, V. J.; Leystralantz, C.; Cook, R. A. Characterization of the (43) Shiloach, J.; Fass, R. Growing E. coli to high cell density-A specific pyruvate transport-system in Escherichia coli K-12.J. historical perspective on method development. Biotechnol. AdV. Bacteriol. 1987, 169, 380-385. 2005, 23, 345-357. (40) Zaldivar, J.; Ingram, L. O. Effect of organic acids on the growth (44) Riesenberg, D. High-cell-density cultivation of Escherichia coli. and fermentation of ethanologenic Escherichia coli LY01. Biotech- Curr. Opin. Biotechnol. 1991, 2, 380-384. nol. Bioeng. 1999, 66, 203-210. (41) Warnecke, T.; Gill, R. T. Organic acid toxicity, tolerance, and (45) Liu, Y. C.; Cheng, K. Y. Effect of ammonium concentration on production in Escherichia coli biorefining applications. Microb. Cell the cultivation of recombinant Escherichia coli producing penicillin Fact. 2005, 4 G acylase. J. Chin. Inst. Chem. Eng. 2000, 31, 601-607. (42) Luli, G. W.; Strohl, W. R. Comparison of growth, acetate production, and acetate inhibition of Escherichia coli strains in batch Received October 2, 2006. Accepted December 21, 2006. and fed-batch fermentations. Appl. EnViron. Microbiol. 1990, 56, 1004-1011. BP060301Y Paper II

Impact of neutralising agent and pH on the succinic acid production by metabolically engineered Escherichia coli in dual-phase fermentations

Christian Andersson, Jonas Helmerius, David Hodge, Kris A. Berglund, and Ulrika Rova* Division of Biochemical and Chemical Process Engineering, Luleå University of Technology, SE-971 87 Luleå

Abstract

Escherichia coli AFP184 has previously been reported to produce succinic acid in a low-cost medium achieving volumetric productivities close to 3 g L-1 h-1. At total organic acid concentrations of 30 g L-1 the productivity decreased drastically resulting in final succinate concentrations of 40 g L-1. In order to improve the economical viability of biochemical succinic acid production the final succinic acid concentration must be increased and the volumetric productivity should be maintained at elevated values (>2.5 g L-1 h-1) for an extended period of time. In the present work the fermentation pH was optimised and the effects of different neutralising agents (NH4OH, KOH, NaOH, and Na2CO3) on succinic acid production by AFP184 were investigated. It was found that AFP184 was rather sensitive to the fermentation pH and an optimum was found in the range of 6.5-6.7. Highest concentration -1 of succinic acid was obtained with Na2CO3, 78 g L . It was also found that the productivity per viable cell for the K- and Na-bases was rather constant during the anaerobic phase. However, loss of cell viability occurred, and decreased the volumetric productivities. In this study it was demonstrated that by altering the neutralising agent it was possible to increase the final succinic acid concentration by almost 100 %.

Keywords: Succinic acid, E. coli fermentation, Neutralising agent, Cell viability

* To whom all correspondence should be addressed Ph: +46920491315. Fax: +46920491199. Email: [email protected]

1 Introduction

Increased costs of petroleum have motivated the search for cost effective alternatives for transforming relatively inexpensive biomass into fuels and chemicals. The key to success in the development of profitable industrial biotechnology is the choice of target fermentations that can compete with the efficiency of the petrochemical industry. It is essential to develop fermentations that produce molecular building blocks which can be used as precursors to produce a number of high-value chemicals or materials. This building block concept follows much of the same strategy used by the petrochemical industry, i.e. production of high-value chemicals from a limited number of chemical intermediates. In 2004, U.S. Department of Energy (USDOE) identified twelve sugar-derived high-value chemicals that could be produced from both lignocellulose and starch and serve as an economic driver for a biorefinery (1). Succinic acid is considered as one of the most interesting building blocks to be produced in a biorefinery. In addition to its own use as a food ingredient and chemical, succinic acid has the potential to produce a wide range of products and derivatives; diesel fuel oxygenates for particulate emission reduction; biodegradable, glycol free, low corrosion deicing chemicals for airport runways; glycol free engine coolants; polybutylene succinate (PBS), a biodegradable polymer that can replace polyethylene and polypropylene; and non- toxic, environmentally safe solvents that can replace chlorinated and other VOC emitting solvents. Succinate is traditionally manufactured from maleic anhydride through n-butane using petroleum as raw material. Succinic acid could also be produced using biochemical conversion of biomass by fungal or bacterial fermentations. Production of succinic acid has been demonstrated in a number of microorganisms including Bacteroides ruminicola and Bacteroides amylophilus, Anaerobiospirillum succiniciproducens, Actinobacillus succinogenes, Mannheimia succiniciproducens, Corynebacterium glutamicum, and Escherichia coli (2-14). The Alternative Feedstock Program (AFP) initiated by the USDOE resulted in the development of two very interesting metabolically engineered E. coli strains with increased succinic acid production, AFP111 and AFP184 (4, 13, 15). AFP111 is a spontaneous mutant with mutations in the glucose specific phosphotransferase system (ptsG), the pyruvate formate lyase system (pfl) and in the fermentative lactate dehydrogenase system (ldh) (4). In this strain the mutations resulted in succinic acid yields in the order of 1 mol succinic acid per mol glucose (13). AFP184 is a metabolically engineered strain where the three mutations described above were deliberately inserted into the E. coli strain C600 (ATCC 23724), which can ferment both five and six carbon sugars (15). A number of different E. coli

2 mutants have been developed by different groups and tested for succinate production (8, 9, 11, 14, 16-20). Fermentations using recombinant AFP111 overexpressing heterologous pyruvate carboxylase conducted in a complex media with the main components tryptone (20 g L-1) and yeast extract (10 g L-1), resulted in improved succinate mass yield of 1.10 and productivities (6, 21).

In a previous study AFP184 was used to produce succinic acid to final concentrations of 25- 40 g L-1 and productivities in the range of 1.5-3 g L-1 h-1 from glucose, fructose, and xylose in a low-cost industrially relevant medium (22). The productivities, although initially very high, declined as the organic acid concentration increased. It was suggested that the loss of succinic acid productivity and cell viability were due to inhibition and toxicity of the produced organic acids or the increased addition of the neutralising agent. Organic acid toxicity resulting in inhibition of cell growth and product formation is a well known phenomenon in E. coli (23- 26). The inhibitory effects can be both pH-based and include anion specific effects on metabolism (27-29). Acetate for example has been shown to inhibit enzymes in the methionine pathway leading to accumulation of homocysteine, which is inhibitory to growth (25). E. coli cell membranes possess an inherently low proton permeability, which is of importance both for pH homeostasis and for creating and sustaining the protonmotive force (30, 31). The undissociated form of the organic acids can however diffuse freely over the cell membrane of E. coli. At low pH the extracellular concentration of undissociated form of the organic acid will increase, resulting in increased diffusion into to cytosol. Since the intracellular pH (pHi) is slightly alkaline (~7.5) the acids, having pKa values of around 4, will dissociate, resulting in an increased proton concentration in the cytoplasm (32). Organic acids could be considered as protonophores which shuttles protons into the cytoplasm. Eventually, accumulation of protons will result in a disruption of the pHi and hence an uncoupling of the membrane pH gradient (ǻpH) (24) which, during aerobic growth is essential for maintenance of the protonmotive force and the cellular energy generation. However, during anaerobic succinate production the protonmotive force does not play a major role, but a lowering of the cytoplasmic pH can have detrimental effects on the function of cellular proteins and enzymes as well as affecting the integrity of the DNA. E. coli counter the imposed acid stress by different acid resistance (AR) systems described in detail elsewhere (33-43). The effects of the acid anions have been investigated for some acids, e.g. acetate (25), but the specific effects of the succinate anion is not known.

3 As fermentation acids are produced the pH of the medium decreases. It is countered by addition of a neutralising agent. In earlier work NH4OH was used to neutralise the fermentations (22). The neutralising agent, in addition to the organic acid being produced, may negatively affect the fermentations. It has been demonstrated that ammonia challenged the integrity of the outer membrane of E. coli reducing growth (44) and at concentration >3 g L-1 ammonia is known to inhibit growth (45, 46). The present study aims to address the problems associated with decreased succinic acid productivity and limited final succinic acid concentrations. Maintaining a high volumetric productivity and achieving higher final succinic acid concentrations is very important for applying the technology to a large-scale industrial facility. Different neutralising chemicals were tested in order to outline and limit possible toxic and inhibitory effects. The fermentation pH was first optimised for succinate production using NH4OH. To investigate the differences between ammonia, alkali hydroxides and carbonates, fermentations at the optimal pH was conducted, using KOH, NaOH, and

Na2CO3 as neutralising agents.

Materials and Methods

Strain and seed culture preparation The E. coli strain AFP184, which lacks functional genes coding for pyruvate formate lyase, fermentative lactate dehydrogenase, and the glucose phosphotransferase system (15), was used in this study. Cultures were stored at í80 °C as 30% glycerol stocks. Seed cultures were prepared by inoculating 100 mL sterile medium (same medium as used for the batch fermentation, see below) in a 500 mL shake-flask with 200 ȝL of the glycerol stock culture. The seed culture was incubated at room temperature, 200 rpm, for 16 h.

Fermentations Batch fermentations with an aerobic growth phase and an anaerobic production phase were conducted in 1 L bioreactors (Biobundle 1L, Applikon Biotechnology, the Netherlands) with a total starting volume of 700 mL (including 35 mL seed culture and 200 mL glucose -1 solution). The growth medium contained the following components in g L : K2HPO4, 1.4;

KH2PO4, 0.6; (NH4)2SO4, 3.3; MgSO4×7H2O, 0.4; corn steep liquor (CSL; 50% solids, Sigma-Aldrich), 15; and 1 mL antifoam agent (Antifoam 204, SigmaAldrich). The bioreactor was sterilised with the medium at 121 °C for 15 min; thereafter 200 mL of a separately

4 sterilised glucose solution (350 g L-1) and 35 mL seed culture were aseptically added, resulting in a final volume of 700 mL and a total glucose concentration of 100 g L-1. The fermentation temperature was controlled at 37 °C and the pH during the aerobic phase was controlled at 6.65 by automatic addition of NH4OH (15% NH3 solution). The dissolved oxygen concentration (%DO), measured by a pO2 electrode, was controlled during the aerobic phase by varying the agitation speed between 500 and 1000 rpm. During the aerobic growth phase (8 h) the culture medium was aerated with an air flow of 5 L min-1. When the optical density at 550 nm reached a value of 30-35 the anaerobic production phase was initiated by withdrawing the air supply and sparging the culture medium with CO2 at a flow rate of 0.8 L min-1. The agitation speed was set to 500 rpm for the anaerobic phase. Succinic acid was produced during the anaerobic phase and to sustain acid production 50 mL sterilised glucose solution (500 g L-1) was added when the glucose concentration was less than 20 g L-1, typically after 20 h of total fermentation time. The anaerobic phase proceeded for approximately 100 h, or until the succinic acid productivity ceased. During the fermentations, samples were aseptically withdrawn for analysis of optical density, viable cells, sugars and organic acids concentrations. pH optimum

To establish pH optimum, fermentations were carried out with NH4OH as neutralising agent at a controlled anaerobic pH of 5.5, 6.0, 6.5, 6.65, 7.0 and 7.5, respectively.

Impact of the neutralising agent In order to evaluate the effect of a fast increase in succinate concentration on the subsequent succinate productivity and cell viability, succinic acid was externally added at the onset of the anaerobic phase to a final concentration of 50 g L-1. The added succinic acid solution, 400 g -1 L , was neutralised with KOH, NaOH or NH4OH respectively. To compare the effects of altering the neutralising agents dual-phase fermentations were carried out with KOH (10 M), -1 NaOH (10 M) or Na2CO3 (300 g L ). The pH was controlled at 6.5 – 6.7.

5 Analysis Cell Growth

Cell growth was monitored by measuring the optical density at 550 nm (OD550). Optical densities were then converted to dry cell weight (DCW, g L-1) using the following equation derived from an OD-DCW calibration curve:

0.3447OD550  0.8574

To establish the number of viable cells, ten-fold serial dilutions of the fermentation samples were plated on tryptone soy agar plates and incubated overnight at 37 °C. The number of colonies was calculated and the number of viable cells was expressed as cells per millilitre fermentation broth. All counts were made in duplicate.

Organic acid and sugar analysis Organic acid and sugars were detected and quantified by HPLC as previously described (22).

Results pH optimum In order to initiate the anaerobic phase at a constant cell density all fermentations were carried out under the same aerobic conditions at pH 6.65. To determine the optimum pH for succinic acid production, the anaerobic phase was controlled at pH 5.5, 6.0, 6.5, 6.65, 7.0, and 7.5, respectively using NH4OH as neutraliser. The final concentration of succinic acid was highest at pH 6.5 and pH 6.65 (Fig. 1a). The maximum succinic acid concentration obtained was 43 g L-1, which corresponds well with previously reported results for this organism (22). In contrast, fermentations conducted at pH 5.5 and 7.5 resulted in very low succinate concentration (< 5 g L-1). At pH 6.0 the final succinate concentration was in the same range as obtained for pH 6.65, but the productivity was considerably lower (Fig. 1b). Apart from succinate the only other product formed was acetate with the highest concentration obtained at pH 6.65 (5.3 g L-1). Acetate was not detected at pH 7.5 and thereafter the acetate concentration was lowest at pH 5.5 (1.4 g L-1), followed by pH 6.0 (3.6 g L-1), and pH 7.0 (3.9 g L-1). For the following fermentations it was decided to use a pH between 6.5 and 6.7.

6 From the measurements of viable cells it was observed that at the cell viability at the two higher pH values (pH 7.0 and 7.5) drastically decreased (Fig. 1c). The cell death was most severe for pH 7.5. For pH 7.0 the cell death started first after 20 hours of total fermentation time, whereas for pH 7.5 it started at the onset of the anaerobic phase. Fermentations at pH 6.65 and 6.0 showed similar decrease in cell viability as previously observed (22). At the higher pH values the loss in cell viability was accompanied by a decrease in optical density and hence cell lysis is a reasonable explanation (Fig. 1d).

50 50

45 a 45 b

40 40 ) -1

35 ) 35 -1

30 30

25 25

20 20

15 Succinic acid (gL 15 Final succinic acid L (g 10 10

5 5

0 0 55.566.577.58 01020304050 pH Time (hours)

50 45

45 c 40 d

40 35 35 30 9 30 25 25 550

OD 20 20 Cells/mL x 10 15 15 10 10

5 5

0 0 01020304050 0 1020304050 Time (hours) Time (hours)

Figure 1. Fermentation characteristics for pH optimisation during succinic acid production. Figure a) shows how the final succinic acid concentration depends on fermentation pH. Figure b) shows succinic acid concentration as a function of time. Figure c) and d) show the viable cell concentration and dry cell weight respectively. Symbols used in b), c), and d): Ɣ pH 6, Ƈ pH 6.65, Ÿ pH 7, Ŷ pH 7.5

Impact of the neutralising agent

Succinic acid neutralised with NH4OH, KOH, or NaOH, was added to the bioreactors to a final succinate concentration of 50 g L-1. After addition no succinate was produced and the cell viability decreased by approximately 50% within 2 hours after the addition. All three

7 fermentations gave similar results and no distinction considering cell viability could be observed.

Standard, dual-phase fermentations were carried out with KOH, NaOH, or Na2CO3 as neutralising agents. In order to maximise productivity and final titre the pH was controlled between 6.5 and 6.7. All three bases resulted in similar fermentation profiles (Fig. 2) where -1 the highest final succinic acid concentration, 78 g L , was achieved when Na2CO3 was used as base. Using NaOH resulted in 69 g L-1 and KOH in 61 g L-1. Again, the only by-product -1 formed was acetic acid and the lowest concentration, 3.7 g L , was obtained when Na2CO3 was used. Fermentations with NaOH and KOH resulted in acetic acid concentrations of 5.7 g L-1 and 6.0 g L-1 respectively. Yields achieved were in the range of 0.75 gram succinic acid per gram glucose consumed in the anaerobic phase (1.14 mol moles-1) for all fermentations. It should be pointed out that glucose was not limiting in any of the fermentations as glucose (25 g) was added when the glucose concentration was less than 20 g L-1.

Table 1. Summary of fermentation results for different basesa -1 -1 -1 -1 -1 -1 -1 Base YP/S (g g ) QP (g L h ) QP (g L h ) QP (g L h ) -1 b anaerobic anaerobic, T20 anaerobic, T40 anaerobic, (max g L )

NH4OH 0.74 2.91 1.38 1.85* KOH 0.76 2.61 1.33 0.65 NaOH 0.71 2.45 1.52 0.72

Na2CO3 0.77 3.31 2.05 0.89 a YP/S is the mass yield of succinic acid based on the glucose consumed in the anaerobic phase, and QP is the volumetric productivity of succinic acid during the anaerobic phase after b 20 h (T20) or 40 h (T40) total fermentation time. Anaerobic productivity calculated either when fermentations are terminated or maximum succinic acid concentration is achieved. *maximum concentration was achieved after 32 hours

Viable cell concentrations for fermentations using NaOH or KOH as the pH regulator resulted in similar trends. During the first 20 hours of anaerobic conditions the viability of the cultures decreased significantly, but for the remainder of the fermentations only small losses in viability occurred (Fig. 2a and b). No significant decrease in optical density was observed during either of the fermentations (data not shown). Using Na2CO3 resulted in viable cell loss

8 first after a total fermentation time of 20 hours, but like the fermentations with NaOH and KOH the decrease in cell density was slow during the remaining time of the fermentation (Fig. 2c).

100 a 100 b ) 80 ) 80 -1 -1 9 9

60 60 Cells/mL x 10

Cells/mL x 10 40 40 Concentration (g L Concentration (g L (g Concentration

20 20

0 0 0 20406080100 0 20406080100 Time (hours) Time (hours)

100 c

) 80 -1 9

60

Cells/mL x 10 40 Concentration (g L (g Concentration

20

0 0 20406080100 Time (hours)

Figure 2. Succinic acid and glucose concentrations and viable cell density for fermentations neutralised with KOH (a), NaOH (b), and Na2CO3 (c). Symbols used in the Figure: Ƈ Succinic acid (g L-1), ¸ Glucose (g L-1), and Ɣ Viable cells (cells mL-1 x 109).

The use of sodium based neutralisers gave higher productivity than potassium based, and

Na2CO3 gave higher productivity than NaOH. Productivities are presented as grams succinic acid produced per viable cell as functions of fermentation time for fermentations neutralised with NH4OH (pH 6.65), KOH, NaOH, and Na2CO3 (Fig. 3a). In general, productivities were initially high, but decreased in the beginning of the anaerobic phase. When using NH4OH the succinate productivity completely stopped after 32 hours, whereas the other fermentations showed stable productivities which remained constant or increased slightly during the rest of the fermentations. At the onset of the anaerobic phase, volumetric productivities reached initial values of more than 3-3.5 g L-1 h-1, but decreased significantly after approximately 15

9 hours of total fermentation time (Fig. 3b). After roughly 30 hours of total fermentation time the volumetric productivity only decreased slowly.

16 4.5 a b 14 4

3.5 12

) 3 -1 10 ) -1 h -1 h 2.5 -1 8

(g L 2 P

6 Q QP (g cell (g QP 1.5

4 1

2 0.5

0 0 0 102030405060708090100 0 102030405060708090100 Time (hours) Time (hours)

Figure 3. Productivity per viable cell x 1014 (a) and volumetric productivity (b) as functions of time. Symbols used in the Figure: ŸNH4OH, Ƈ KOH, Ŷ NaOH, and Ɣ Na2CO3. Viable counts were made in duplicates and the difference is shown by the error bars. Where no error bar is present the error was small enough to be covered by the Figure symbols.

Discussion pH optimum

From NH4OH neutralised fermentations at different pH it can be concluded that pH has a strong impact on succinic acid production. The optimal pH interval is also very narrow around 6.6 and diverging to higher or lower pH results in a significant reduction of succinic acid productivity and considerably decreased final titres. It is well-documented that enzymes have an optimal pH range outside which its activity is reduced (47, 48). It seems reasonable to assume that the decrease in productivity and cell viability observed at higher and lower pH values are correlated to generally impaired enzymatic activity. In case of cell viability, the cells were significantly more sensitive to the higher than the lower pH values investigated. Considerable loss in cell viability was observed during the anaerobic phase at pH 7.5. It is known that E. coli cells lyse at alkaline pH (pH 9-12) and that ammonia concentrations in the range of 80-100 mM damage the outer membrane and increase the leakage of periplasmic proteins to the medium (44, 49). However, a pH of 7.5 is only slightly alkaline and it is possible to grow E. coli aerobically to high cell density at that pH. Anaerobic acid production imposes limits on the cells ability to generate energy. At pH 7.5 the glucose consumption was

10 very low compared to that of the other pH values investigated (data not shown) and cells would be limited in their ability to divide and maintain viability. The slightly elevated pH might also weaken the outer and cytoplasmic membranes (49), this together with moderate agitation and the ammonia concentration could compromise the ability of the thin peptidoglycan cell wall to stabilise the cells. The result would be disruption of the cytoplasmic membrane resulting in cell lysis and death.

Organic acid production and inhibition The major aim of the present study was to elucidate the effects of different neutralising agents on succinic acid productivity. From the fermentation profiles (Fig. 2) it is clear that using potassium or sodium bases for neutralisation would be beneficial for the final succinic acid concentration. Compared to fermentations neutralised with NH4OH fermentations neutralised with potassium or sodium bases obtained increased final titres by at least 50%. It should be noted that unlike NH4OH fermentations succinate production in alkali fermentations never ceased. With regards to productivity and final acid concentrations the difference between using NaOH or KOH was small (Fig. 2 and 3). However, using Na2CO3 resulted in increased productivity and succinate concentration concurrently with a decreased acetic acid production. The increased productivity might caused by an increased availability of hydrogen carbonate

 (HCO 3 ) from the addition of the base chemical. The enzyme phosphoenolpyruvate (PEP)

 carboxylase catalysing the carboxylation of PEP to oxaloacetic acid uses HCO3 as a substrate for the reaction (50). The higher productivity during neutralisation of Na2CO3 indicates that the growth medium is not saturated by the CO2 from the sparging. The additional availability

 ofHCO 3 might help to direct the metabolic flow towards succinate which can explain the lower acetate formation. The yield when using Na2CO3 was 0.77 gram succinic acid produced per gram glucose consumed in the anaerobic phase. The anaerobe succinate yield when using NaOH was 0.71 g g-1. The difference in yield between the two sodium bases can be explained by the increased succinate concentration as a result of a reduced flow of carbon to acetate. The theoretical yield of succinic acid from glucose during anaerobic succinic acid production is 1.12 g g-1 (22). Synthesis and accumulation of compatible solutes as a response to osmotic stress in the high osmolarity environment generated by the succinate production and base addition could explain the divergence between the achieved and theoretical yield (see Viable cells below for a more thorough discussion).

11 Productivities per viable cell as a function of time showed that the production of succinate stopped after 32 hours when NH4OH was used as neutraliser. Fermentations using any of the other three bases did not show similar results, instead a productivity decrease was observed after approximately 60 hours for NaOH and KOH, and after 35 hours for Na2CO3 (Fig. 3). This corresponds to succinic acid concentration around 55-60 g L-1 for the sodium bases and around 50 g L-1 for KOH. For the remaining time of the fermentations productivity per viable cell remained constant or increased slightly. Given that the productivities per viable cell remain more or less constant from approximately 50 hours of total fermentation time for potassium and sodium base neutralised fermentations it seems like neither of these bases nor the succinic acid concentration negatively affects or inhibits the succinic acid productivity of AFP184 in the medium used. However, the viable cell concentration decreases throughout the anaerobic phase and reduces the total succinic acid productivity, which also reflects the volumetric productivities decrease during the anaerobic phase (Fig. 3b). This might be attributed to the sodium/potassium or succinic acid concentration (see discussion regarding viable cells below). When using NH4OH, productivity completely ceased after 32 hours. At this time the viable cell count was still high and it is concluded that ammonium at the given concentrations (~10 g L-1) severely inhibit succinic acid productivity in AFP184. The same productivity repression was observed in a previous study (22).

It is clear from the results that the preferred neutralising agent would be Na2CO3, since it generated the highest productivities, final titres and had the lowest by-product formation.

Viable cells From the results with neutralisation by potassium or sodium base the cell viability seems to be the limiting step in succinic acid production when using AFP184.

Organic acid toxicity is a well studied phenomena since it impairs cell growth and productivity (24, 25, 27-29, 51). The most obvious aspect to consider is the total organic acid load. Organic acids in their undissociated form can diffuse over the cell membranes. Once inside the cytoplasm they dissociate and lower the slightly alkaline cytoplasmic pH, which can result in protein denaturation and DNA damages (23, 28, 32, 33). The impact of different bases on succinic acid productivity was studied at conditions where the external pH was controlled between 6.5 and 6.7. At this pH interval succinic acid with pKa of 4.19 and 5.57 would be dissociated and not able to freely traverse the cell membranes. It has been indicated that acetate and lactate can cross the cytoplasmic membrane in the dissociated form (24). This

12 implies that the anions of succinic acid might accumulate in the cytoplasm. Increased cytoplasmic concentrations of the anions can have different metabolic affects depending on which specific organic acid being studied (23, 25, 52). Experiments to quantify intracellular succinate concentrations are currently being undertaken.

Another effect of the increased acid concentration is a subsequent increase in medium osmolarity, originating both from organic acids and from the added neutralising agent. There are numerous studies and reviews on osmotic stress and osmotolerance in bacteria (53-56). It could be proposed that the viable cell loss could be due to the high osmolarity of the medium that is generated during the anaerobic phase (>1.2 osmoles per litre from neutralised succinate alone). When succinic acid (50 g L-1) was added at the beginning of the anaerobic phase, cell viability decreased by approximately 50% irrespective of neutralising agent. However, when the osmolarity was due to cellular succinate production the cells adapted to the conditions and no sudden decreases in viability were observed. Even at high osmolarities (from e.g. 30-80 hours) viable counts remained relatively constant.

E. coli subjected to osmotic stress first respond by accumulating K+. At higher osmolarities K+ accumulation in the cytoplasm interfere with cellular functions and cells excrete K+ and instead accumulate other compatible solutes, like glycinebetaine, proline, and trehalose (57, 58). The solutes offering the highest osmotolerance in E. coli are glycinebetaine and proline. Since glycinebetaine cannot be synthesise de novo in E. coli, it must be accumulated from the medium. CSL does not contain glycinebetaine (or proline, which cannot be synthesised de novo either). Choline, a quaternary amine present in CSL (~0.5 % of the dry weight) can be converted into glycinebetaine by E. coli in the presence of a terminal electron acceptor like O2 (59). During anaerobic succinic acid production by AFP184 choline can thus not be used as an osmoprotectant. Under anaerobic conditions, or in media deprived of these osmoprotectants, E. coli will instead synthesise trehalose intracellular (60). Trehalose can accumulate to intracellular concentrations equal to approximately 20 % of the osmolar concentration of solutes in the cultivation medium (53). Carbon balances over the anaerobic phase can account for approximately 80-85 % of the carbon. Osmotically stressed AFP184 could hence synthesise trehalose during the anaerobic phase to substantial intracellular concentrations, which could explain the divergence between the achieved and theoretical yield and identify the unknown carbon fraction. We are currently conducting a study to characterise intracellular concentrations of metabolic products in AFP184 during succinic

13 acid production and also investigate the effects of addition of osmoprotectants such as glycinebetaine and proline to the medium. Conclusions In this study we have demonstrated that replacing the neutralising agent can provide substantial process improvements in the form of increased duration of high volumetric productivity and increased final titre. Compared to fermentations neutralised with NH4OH it was possible to achieve an almost 100% increase in final succinic acid concentration using

Na2CO3. It was also observed that the cell viability decreased during the anaerobic phase irrespective of the base used. The viable cell decrease is attributed to increased acid concentration coupled with possible cytoplasmic succinate accumulation, high medium osmolarity and long time exposure to a low energy generating environment and high maintenance demands during the course of acid production in the anaerobic phase. Further studies will be directed towards increasing the long term viability in the anaerobic phase by outlining intracellular metabolite concentrations and investigating the impact of supplementation of the medium with osmoprotectants.

References

1. T. Werpy and G. Petersen, Top value added chemicals from biomass, volume I. 2004. 2. M.J. Van der Werf, M.V. Guettler, M.K. Jain, and J.G. Zeikus, Environmental and physiological factors affecting the succinate product ratio during carbohydrate fermentation by Actinobacillus sp. 130Z. Archives of Microbiology, 1997. 167(6): p. 332-342. 3. M.V. Guettler, D. Rumler, and M.K. Jain, Actinobacillus succinogenes sp. nov., a novel succinic-acid-producing strain from the bovine rumen. International Journal of Systematic Bacteriology, 1999. 49: p. 207-216. 4. R. Chatterjee, C.S. Millard, K. Champion, D.P. Clark, and M.I. Donnelly, Mutation of the ptsG gene results in increased production of succinate in fermentation of glucose by Escherichia coli. Applied and Environmental Microbiology, 2001. 67(1): p. 148- 154. 5. P.C. Lee, S.Y. Lee, S.H. Hong, and H.N. Chang, Isolation and characterization of a new succinic acid-producing bacterium, Mannheimia succiniciproducens MBEL55E,

14 from bovine rumen. Applied Microbiology and Biotechnology, 2002. 58(5): p. 663- 668. 6. G.N. Vemuri, M.A. Eiteman, and E. Altman, Effects of growth mode and pyruvate carboxylase on succinic acid production by metabolically engineered strains of Escherichia coli. Applied and Environmental Microbiology, 2002. 68(4): p. 1715- 1727. 7. D.Y. Kim, S.C. Yim, P.C. Lee, W.G. Lee, S.Y. Lee, and H.N. Chang, Batch and continuous fermentation of succinic acid from wood hydrolysate by Mannheimia succiniciproducens MBEL55E. Enzyme and Microbial Technology, 2004. 35(6-7): p. 648-653. 8. H. Lin, G.N. Bennett, and K.Y. San, Fed-batch culture of a metabolically engineered Escherichia coli strain designed for high-level succinate production and yield under aerobic conditions. Biotechnology and Bioengineering, 2005. 90(6): p. 775-779. 9. H. Lin, G.N. Bennett, and K.Y. San, Genetic reconstruction of the aerobic central metabolism in Escherichia coli for the absolute aerobic production of succinate. Biotechnology and Bioengineering, 2005. 89(2): p. 148-156. 10. S. Okino, M. Inui, and H. Yukawa, Production of organic acids by Corynebacterium glutamicum under oxygen deprivation. Applied Microbiology and Biotechnology, 2005. 68(4): p. 475-480. 11. A.M. Sanchez, G.N. Bennett, and K.Y. San, Efficient succinic acid production from glucose through overexpression of pyruvate carboxylase in an Escherichia coli alcohol dehydrogenase and lactate dehydrogenase mutant. Biotechnology Progress, 2005. 21(2): p. 358-365. 12. S.J. Lee, H. Song, and S.Y. Lee, Genome-based metabolic engineering of Mannheimia succiniciproducens for succinic acid production. Applied and Environmental Microbiology, 2006. 72(3): p. 1939-1948. 13. M.I. Donnelly, C.S. Millard, D.P. Clark, M.J. Chen, and J.W. Rathke, A novel fermentation pathway in an Escherichia coli mutant producing succinic acid, acetic acid, and ethanol. Applied Biochemistry and Biotechnology, 1998. 70-2: p. 187-198. 14. A.M. Sanchez, G.N. Bennett, and K.Y. San, Novel pathway engineering design of the anaerobic central metabolic pathway in Escherichia coli to increase succinate yield and productivity. Metabolic Engineering, 2005. 7(3): p. 229-239. 15. M.I. Donnelly, C.Y. Sanville-Millard, and N.P. Nghiem, Method to produce succinic acid from raw hydrolysates. 2004. U.S. Patent 6,743,610

15 16. H. Lin, G.N. Bennett, and K.Y. San, Chemostat culture characterization of Escherichia coli mutant strains metabolically engineered for aerobic succinate production: A study of the modified metabolic network based on metabolite profile, enzyme activity, and gene expression profile. Metabolic Engineering, 2005. 7(5-6): p. 337-352. 17. H. Lin, G.N. Bennett, and K.Y. San, Effect of carbon sources differing in oxidation state and transport route on succinate production in metabolically engineered Escherichia coli. Journal of Industrial Microbiology & Biotechnology, 2005. 32(3): p. 87-93. 18. H. Lin, G.N. Bennett, and K.Y. San, Metabolic engineering of aerobic succinate production systems in Escherichia coli to improve process productivity and achieve the maximum theoretical succinate yield. Metabolic Engineering, 2005. 7(2): p. 116- 127. 19. H. Lin, K.Y. San, and G.N. Bennett, Effect of Sorghum vulgare phosphoenolpyruvate carboxylase and Lactococcus lactis pyruvate carboxylase coexpression on succinate production in mutant strains of Escherichia coli. Applied Microbiology and Biotechnology, 2005. 67(4): p. 515-523. 20. N.R. Yun, K.Y. San, and G.N. Bennett, Enhancement of lactate and succinate formation in adhE or pta-ackA mutants of NADH dehydrogenase-deficient Escherichia coli. Journal of Applied Microbiology, 2005. 99(6): p. 1404-1412. 21. G.N. Vemuri, M.A. Eiteman, and E. Altman, Succinate production in dual-phase Escherichia coli fermentations depends on the time of transition from aerobic to anaerobic conditions. Journal of Industrial Microbiology & Biotechnology, 2002. 28(6): p. 325-332. 22. C. Andersson, D. Hodge, K.A. Berglund, and U. Rova, Effect of Different Carbon Sources on the Production of Succinic Acid Using Metabolically Engineered Escherichia coli. Biotechnol. Prog., 2007. 23. J. Zaldivar and L.O. Ingram, Effect of organic acids on the growth and fermentation of ethanologenic Escherichia coli LY01. Biotechnology and Bioengineering, 1999. 66(4): p. 203-210. 24. D.D. Axe and J.E. Bailey, Transport of Lactate and Acetate through the Energized Cytoplasmic Membrane of Escherichia-Coli. Biotechnology and Bioengineering, 1995. 47(1): p. 8-19.

16 25. A.J. Roe, C. O'Byrne, D. McLaggan, and I.R. Booth, Inhibition of Escherichia coli growth by acetic acid: a problem with methionine biosynthesis and homocysteine toxicity. Microbiology-Sgm, 2002. 148: p. 2215-2222. 26. A.J. Roe, Perturbation of Anion Balance during Inhibition of Growth of Escherichia coli by Weak Acids. Journal of bacteriology, 1998. 180(4): p. 767. 27. B. Pieterse, R.J. Leer, F.H.J. Schuren, and M.J. van der Werf, Unravelling the multiple effects of lactic acid stress on Lactobacillus plantarum by transcription profiling. Microbiology-Sgm, 2005. 151: p. 3881-3894. 28. J.B. Russell and F. DiezGonzalez, The effects of fermentation acids on bacterial growth, in Advances in Microbial Physiology, Vol 39. 1998. p. 205-234. 29. T. Warnecke and R.T. Gill, Organic acid toxicity, tolerance, and production in Escherichia coli biorefining applications. Microbial Cell Factories, 2005. 4. 30. I.R. Booth, P. Cash, and C. O'Byrne, Sensing and adapting to acid stress. Antonie Van Leeuwenhoek International Journal of General and Molecular Microbiology, 2002. 81(1-4): p. 33-42. 31. R.K. Poole, Bacterial responses to pH - Summary, in Bacterial Response to Ph. 1999. p. 251-255. 32. I.R. Booth, Regulation of cytoplasmic pH in bacteria. Microbiological reviews, 1985. 49: p. 359-378. 33. J.W. Foster, Escherichia coli acid resistance: Tales of an amateur acidophile. Nature Reviews Microbiology, 2004. 2(11): p. 898-907. 34. M.P. Castanie-Cornet, T.A. Penfound, D. Smith, J.F. Elliott, and J.W. Foster, Control of acid resistance in Escherichia coli. Journal of Bacteriology, 1999. 181(11): p. 3525-3535. 35. H.T. Richard and J.W. Foster, Acid resistance in Escherichia coli, in Advances in Applied Microbiology, Vol 52. 2003. p. 167-186. 36. H. Richard and J.W. Foster, Escherichia coli glutamate- and arginine-dependent acid resistance systems increase internal pH and reverse transmembrane potential. Journal of Bacteriology, 2004. 186(18): p. 6032-6041. 37. J.S. Lin, M.P. Smith, K.C. Chapin, H.S. Baik, G.N. Bennett, and J.W. Foster, Mechanisms of acid resistance in enterohemorrhagic Escherichia coli. Applied and Environmental Microbiology, 1996. 62(9): p. 3094-3100. 38. P. Small, Acid and base resistance in Escherichia coli and Shigella flexneri: role of rpoS and growth pH. Journal of bacteriology, 1994. 176(6): p. 1729.

17 39. S. Bearson, Acid stress responses in enterobacteria. FEMS microbiology letters, 1997. 147(2): p. 173-180. 40. B.O. Hersh, A glutamate-dependent acid resistance gene in Escherichia coli. Journal of bacteriology, 1996. 178(13): p. 3978. 41. R. Iyer, Arginine-Agmatine Antiporter in Extreme Acid Resistance in Escherichia coli. Journal of bacteriology, 2003. 185(22): p. 6556. 42. S. Gong, YjdE (AdiC) Is the Arginine: Agmatine Antiporter Essential for Arginine- Dependent Acid Resistance in Escherichia coli. Journal of bacteriology, 2003. 185(15): p. 4402. 43. J. Lin, I.S. Lee, J. Frey, J.L. Slonczewski, and J.W. Foster, Comparative analysis of extreme acid survival in Salmonella typhimurium, Shigella flexneri, and Escherichia coli. Journal of bacteriology, 1995. 177(14): p. 4097-104. 44. G. Naundorf and N.G. Aumen, Ammonia-Induced Cell-Envelope Injury in Escherichia-Coli and Enterobacter-Aerogenes. Canadian Journal of Microbiology, 1990. 36(8): p. 525-529. 45. J. Shiloach and R. Fass, Growing E-coli to high cell density - A historical perspective on method development. Biotechnology Advances, 2005. 23(5): p. 345-357. 46. D. Riesenberg, High-Cell-Density Cultivation of Escherichia-Coli. Current Opinion in Biotechnology, 1991. 2(3): p. 380-384. 47. E.E.F. Gale, Factors influencing the enzymatic activities of bacteria. Bacteriological reviews, 1943. 7(3): p. 139-73. 48. E.E.F. Gale and H.H.M. Epps, The effect of the pH of the medium during growth on the enzymic activities of bacteria (Escherichia coli and Micrococcus lysodeikticus) and the biological significance of the changes produced. The Biochemical journal, 1942. 36(7-9): p. 600-18. 49. A.F. Mendonca, T.L. Amoroso, and S.J. Knabel, Destruction of Gram-Negative Food- Borne Pathogens by High pH Involves Disruption of the Cytoplasmic Membrane. Applied and Environmental Microbiology, 1994. 60(11): p. 4009-4014. 50. G. Gottschalk, Bacterial metabolism. 2nd ed. 1986: Springer-Verlag New York Inc. 51. E.R. Kashket, Bioenergetics of Lactic-Acid Bacteria - Cytoplasmic Ph and Osmotolerance. Fems Microbiology Reviews, 1987. 46(3): p. 233-244. 52. C. Kirkpatrick, L.M. Maurer, N.E. Oyelakin, Y.N. Yoncheva, R. Maurer, and J.L. Slonczewski, Acetate and formate stress: Opposite responses in the proteome of Escherichia coli. Journal of Bacteriology, 2001. 183(21): p. 6466-6477.

18 53. L.N. Csonka, Physiological and Genetic Responses of Bacteria to Osmotic-Stress. Microbiological Reviews, 1989. 53(1): p. 121-147. 54. A. Ishida, Y. Kawatake, and N. Ono, Osmotic-Stress Conditioning for Induction of Acquired Osmotolerance in Escherichia-Coli. Journal of General and Applied Microbiology, 1994. 40(1): p. 35-42. 55. I. Poirier, P.A. Marechal, C. Evrard, and P. Gervais, Escherichia coli and Lactobacillus plantarum responses to osmotic stress. Applied Microbiology and Biotechnology, 1998. 50(6): p. 704-709. 56. D.S. Cayley, H.J. Guttman, and M.T. Record, Biophysical characterization of changes in amounts and activity of Escherichia coli cell and compartment water and turgor pressure in response to osmotic stress. Biophysical Journal, 2000. 78(4): p. 1748- 1764. 57. W.S. Epstein, Osmoregulation by potassium transport in Escherichia coli. FEMS microbiology reviews, 1986. 39: p. 73. 58. P.I. Larsen, L.K. Sydnes, B. Landfald, and A.R. Strøm, Osmoregulation in Escherichia coli by accumulation of organic osmolytes: betaines, glutamic acid, and trehalose. Archives of microbiology, 1987. 147(1): p. 1-7. 59. B. Landfald, Choline-glycine betaine pathway confers a high level of osmotic tolerance in Escherichia coli. Journal of bacteriology, 1986. 165(3): p. 849. 60. H.M. Giaever, O.B. Styrvold, I. Kaasen, and A.R. Strøm, Biochemical and genetic characterization of osmoregulatory trehalose synthesis in Escherichia coli. Journal of bacteriology, 1988. 170(6): p. 2841-9.

19