Fermentation Process Development for the Production of Medium-chain-length Poly(3-hydroxyalkanoates) by Pseudomonas putida KT2440

by Zhiyong Sun

A thesis submitted to the Department of Chemical Engineering in conformity with the requirements for the degree of Doctor of Philosophy

Queens University Kingston, Ontario, Canada July, 2007 Copyright Zhiyong Sun, 2007

Abstract
Fed-batch fermentation processes were developed for efficient and automated lab scale production of saturated and unsaturated (functionalized) medium-chain-length poly(3-hydroxyalkanoates) (MCL-PHA) by Pseudomonas putida KT2440. Four automated glucose feeding strategies were first developed for the high-cell-density cultivation of P. putida KT2440. An overall biomass productivity of 4.3 g X l-1 h-1 (highest reported for P. putida KT2440) was obtained by continuous feeding based on cumulative glucose consumption estimation. An exponential feeding strategy proved to be an effective way of applying carbon-limitation and controlling the specific growth rate of the culture while also achieving a satisfactory cell concentration. Contrary to the common literature belief, nitrogen limitation proved to be unnecessary for MCL-PHA synthesis by P. putida KT2440. Instead, MCL-PHA production was achieved by single-stage, carbon-limited, exponential nonanoic acid feeding fed-batch fermentations. This approach resulted in an overall PHA productivity of 1.4 g PHA l-1 h-1 (highest reported for PHA production using nonanoic acid as substrate) at a specific growth rate of 0.25 h-1 and a high final PHA content of 75% (highest reported final MCL-PHA content) at a specific growth rate of 0.15 h-1. By co-feeding of nonanoic acid and glucose, the nonanoic acid to PHA yield was increased up to 30% without significantly altering the composition of the PHA or compromising the overall PHA productivity (1.4 g PHA l-1 h-1). This would lead to a substantial substrate cost reduction if MCL-PHA is produced at a commercial scale. Unsaturated (functionalized) MCL-PHA was also produced at a productivity of up to 1.1 g PHA l-1 h-1 (highest reported for unsaturated MCL-PHA production) and a final content of up to 56% by using nonanoic acid with 10-undecenoic acid as a co-substrate. The molar fraction of each monomeric component remained relatively constant through such fermentations. A linear relationship between the degree of unsaturation of the PHA product and the fraction of unsaturated carbon source in the substrate mixture was demonstrated. Therefore, such a co-feeding fed-batch process can be used to produce functionalized MCL-PHA with controlled compositions.
i

Statement of Co-authorship
The author wishes to acknowledge the contribution of Dr. Bruce A Ramsay, Dr. Juliana A Ramsay, and Dr. Martin Guay in the preparation of all manuscripts presented in this thesis. All experiments were performed by the author.

ii

Acknowledgements
I would first like to thank my parents. I rarely performed as well as I should have in those critical exams along my path of education, yet my parents have given me nothing but love and encouragement, which made a boy became a useful man. When I decided to pursue my graduate study abroad, they supported me without any hesitation. As their only child, I have barely spent two months with them during the past five years, and probably this will be even less in the near future. I know the only way I can make them happy is to work hard and live a fruitful life. I miss them so much. My thankfulness also goes to Xuan (Jade), my beloved wife, friend, and colleague. It is a blessing to have the opportunity to work with her in the same lab. Her love, care, and support made my life much brighter and my work much smoother. I cannot imagine what a different study experience abroad would be like without her. I also thank sincerely my supervisors and mentors, Dr. Bruce Ramsay, Dr. Juliana Ramsay, and Dr. Martin Guay. Starting a graduate study is one thing. Starting a new life by living in a different country and speaking a different language is another. Their guidance, encouragement, and friendship made the transition smooth and joyfull. From them, I learned advanced scientific knowledge, English, ways of being kind to people, and even Canadian ways of humor. I consider myself quite lucky to have supervisors that I can just walk in, sit down, and discuss my project with. I do not think there could be a better graduate study experience than what they have provided to me. My gratefulness to them is more than I can say. Last but not least, thank you to all my labmates, Yinghao, Paul, Thiru, Bozhi, and others. With all the equipment, I know I am working in an advanced lab and, with all you guys, I feel like I am living with a warm family.
iii

Table of Contents
Abstract Statement of Co-authorship Acknowledgements Table of Contents List of Tables List of Figures List of Abbreviations and Symbols CHAPTER 1 General Introduction 1.1 Background 1.2 Overall objective 1.3 Approaches 1.4 References CHAPTER 2 Literature Review Fermentation Process Development for the Production of Medium-chain-length Poly(3-hyroxyalkanoates) 2.1 Abstract 2.2 Introduction 2.3 Physiology and the kinetics of biosynthesis 2.3.1 General considerations 2.3.2 Carbon sources and their concentration 2.3.3 Growth rate and nutrient conditions 2.4 Process development 2.4.1 Continuous processes 2.4.2 Fed-batch processes 2.4.3 Production of MCL-PHAs containing functional groups 2.5 The effect of nutrient limitation on MCL-PHA accumulation 2.6 Economic considerations 2.7 Future commercial processes 2.8 References CHAPTER 3 Automated Feeding Strategies for High-cell-density Fed-batch Cultivation of Pseudomonas putida KT2440 3.1 Abstract 3.2 Introduction 3.3 Materials and methods 3.3.1 Analytical procedures
iv

i ii iii iv viii ix xii 1 1 4 4 8 10 10 11 11 13 13 15 18 19 20 21 26 31 32 35 36 43 43 44 44 47 47

3.3.2 Microorganism and growth Medium 47 3.3.3 Fermentation conditions 48 3.3.4 Substrate Feeding Strategies 49 3.4 Results 51 3.4.1 Growth yield study 51 3.4.2 Exponential feeding with a predetermined 53 3.4.3 Exponential feeding based on initial max estimation 54 3.4.4 Semi-continuous feeding based on the detection of substrate limitation (CPR-based pulse feeding) 56 3.4.5 Continuous feeding with real-time glucose consumption estimation based on CCP analysis 57 3.5 Discussion 58 3.5.1 Yield values 58 3.5.2 Exponential feeding strategies 60 3.5.3 CPR-based pulse feeding 62 3.5.4 Continuous feeding based on CCP 63 3.5.5 Comparison of methodologies 64 3.6 References 66 CHAPTER 4 69 Increasing the Yield of MCL-PHA from Nonanoic Acid by Co-feeding Glucose During the PHA Accumulation Stage in Two-stage Fed-batch Fermentations of Pseudomonas-putida KT2440 69 4.1 Abstract 70 4.2 Introduction 70 4.3 Materials and methods 72 4.4 Results and discussion 74 4.5 References 76 CHAPTER 5 78 Carbon-limited Fed-batch Production of Medium-chain-length Polyhydroxyalkanoates from Nonanoic Acid by Pseudomonas putida KT2440 78 5.1 Abstract 79 5.2 Introduction 79 5.3 Materials and methods 82 5.3.1 Microorganism and growth medium 82 5.3.2 Fermentation conditions 83 5.3.3 Substrate feeding and control methods 84 5.3.4 Analytical procedures 85 5.4 Results 86 -1 5.4.1 Single-stage, exponential feeding of nonanoic acid (target = 0.15 h ) 88
v

5.4.2 Single-stage, exponential feeding of nonanoic acid (target = 0.25 h-1) 89 5.4.3 Single-stage, exponential feeding of nonanoic acid (target = 0.25 h-1) followed by N-limited culture on nonanoic acid 90 5.4.4 Biomass and PHA yield 92 5.5 Discussion 93 5.5.1 Process design implications 93 5.5.2 Yield 95 5.5.3 Kinetics of MCL-PHA synthesis 96 5.6 References 98 CHAPTER 6 103 Fed-batch Production of Medium-chain-length Polyhydroxyalkanoates by Pseudomonas putida KT2440 using Nonanoic Acid and Glucose Co-feeding 103 6.1 Abstract 104 6.2 Introduction 104 6.3 Materials and methods 106 6.3.1 Microorganism and growth medium 106 6.3.2 Fermentation conditions 107 6.3.3 Substrate feeding and control methods 108 6.3.4 Analytical procedures 109 6.4 Results 110 6.4.1 Nonanoic acid and glucose co-feeding for MCL-PHA production 110 6.4.2 Monomeric compositions during co-substrate cultivation 113 6.4.3 Biomass and PHA Yield 114 6.5 Disucssion 118 6.5.1 High yield and PHA productivity by co-substrate carbon-limited fermentation 118 6.5.2 Yield potential of MCL-PHA synthesis 121 6.6 References 122 CHAPTER 7 125 Fed-batch Production of Unsaturated Medium-chain-length Polyhydroxyalkanoates with Controlled Composition by Pseudomonas putida KT2440 125 7.1 Abstract 126 7.2 Introduction 126 7.3 Materials and methods 129 7.3.1 Microorganism and growth medium 129 7.3.2 Fermentation conditions 129 7.3.3 Substrate feeding and control methods 130 7.3.4 Analytical procedures 131 7.4 Results 132
vi

7.4.1 Production of unsaturated MCL-PHAs by exponential feeding of nonanoic acid and 10-undecenoic acid 132 7.4.2 Monomeric compositions 135 7.5 Discussion 137 7.5.1 Process design for unsaturated MCL-PHAs production 137 7.5.2 Effect of specific growth rate and varied substrate composition on PHA synthesis 138 7.5.3 Unsaturated MCL-PHAs with controllable compositions 139 7.6 References 141 CHAPTER 8 Conclusions 143 8.1 Summary and contributions 143 8.1.1 High-cell-density cultivation of P. putida KT2440 143 8.1.2 Carbon-limited, single-stage MCL-PHA production 144 8.1.3 Co-substrate feeding strategy for functionalized MCL-PHA production 145 8.1.4 Yield enhancement and substrate cost reduction 145 8.2 Recommendations for future work 146

vii

List of Tables
Table 2-1 Summary of MCL-PHA (regular and functionalized) production processes. 29 Table 3-1 Comparison of the four feeding strategies and recommendations Table 6-1 Summary of fermentation results. Table 6-2 Summary of literature MCL-PHA production and yield Table 7-1 Fermentation results summary. Table 7-2 Molar fraction of all 3-OH-alkanoates detected in the MCL-PHA product from different fermentations. 65 117 117 134 135

viii

List of Figures
Figure 2-1 Chemical structure of polyhydroxyalkanoates (PHAs), which are classified as short-chain-length PHAs (SCL-PHAs) when the repeating units contain 4 or 5 carbons and medium-chain-length PHAs (MCL-PHAs) when the repeating units contain 6 or more carbons; R pendant group may also contains functional groups resulting in functionalized PHAs. 13 Figure 2-2 PHA content of P. putida strains using alkanes or alkanoates of varying length. P. putida GPo1 (formerly known as P. oleovorans GPo1) grown on n-alkanoates (Gross et al. 1989), P. putida GPo1 on n-alkanes (Lageveen et al. 1988), and P. sp. DSY-82 on n-alkanoates (Kang et al. 2001). 16 Figure 3-1 Yield of dry biomass from the principal growth substrates. Different symbols indicate data from different fermentations. The slope of the line indicates the calculated yield value. Only data indicated by closed and open circles were used to calculate yield from PO4 in (c). 52 Figure 3-2 Fed-batch P. putida cultivation using simple exponential feeding designed to achieve = 0.25 h-1. 53 Figure 3-3 Online cumulative CO2 production (CCP) data regressions according to Eq. 2 for max estimation. (a) Regression at 10 h cultivation time. (b) Regression at 12 h cultivation time. 55 Figure 3-4 Fed-batch P. putida cultivation using exponential feeding with online max estimation. 55 Figure 3-5 Fed-batch P. putida cultivation using CO2 production rate (CPR) based pulse feeding. 57

Figure 3-6 Fed-batch P. putida cultivation using continuous glucose feeding based on glucose consumption as estimated from cumulative CO2 production data. 58 Figure 4-1 PHA subunits produced during the nitrogen-limited stage of each fermentation. Nitrogen-limitation began at about the same time as the first data point shown in each figure. (a) G-only PHA synthesis. (b) NA-only PHA synthesis. (c) NA+G co-feeding PHA synthesis. The PHA subunits are 3-HHp, 3-hydroxyheptanoate, 3-HO, 3-hydroxyoctanoate, 3-HN, 3-hydroxynonanoate, 3-HD, 3-hydroxydecanoate. 75 Figure 4-2 Relationship between total carbon of nonanoic acid (NA) or glucose (G) consumed and total carbon accumulated into the corresponding PHA subunits. Closed squares are from G-only PHA synthesis, open squares
ix

from NA-only PHA synthesis, and half-closed squares from NA+G co-feeding.

75

Figure 5-1 Growth of P. putida KT2440 on glucose followed by N-limited growth on nonanoic acid (NA). The division between growth phase and N-limited phase is indicated by the dashed line. The nonanoic acid feeding rate (FNA) was increased from 0.47 g NA l-1 h-1 to 1.01 g NA l-1 h-1 at 27 h during the N-limited phase, as indicated by the dash-dot line. 87 Figure 5-2 Carbon-limited growth of P. putida KT2440 on nonanoic acid (NA) at = 0.15 h-1. 89

Figure 5-3 Effect of growth rate on PHA synthesis in P. putida KT2440 under nonanoic acid limitation. Closed and open symbols represent data from the = 0.15 h-1 and the = 0.25 h-1 fermentations, respectively. qPHA(Xr) is the specific PHA synthesis rate (g PHA g Xr-1 h-1). 91 Figure 5-4 Two-stage, fed-batch using nonanoic acid (NA) as the sole carbon source. Exponential feeding (= 0.25 h-1) was used during the growth phase with all other nutrients in excess. The division between the growth and N-limited phases is indicated by the dashed line. The nonanoic acid feed rate was increased from 1.31 g NA l-1 h-1 to 1.82 g NA l-1 h-1 at 41.5 h during the N-limited phase as indicated by the dash-dot line. 92 Figure 5-5 Yield of total biomass, PHA, and residual biomass (Xr) from nonanoic acid (NA). Closed and open symbols represent data from = 0.15 h-1 and the = 0.25 h-1 fermentations, respectively. The slopes of the curves indicate yields from nonanoic acid. 93 Figure 5-6 The specific rate of PHA synthesis (per gram of PHA) approaches a constant value as the concentration of PHA in the biomass increases. 98

Figure 6-1 Single-stage MCL-PHA production by P. putida KT2440 using nonanoic acid (NA) and glucose (G) co-feeding strategy with NA:G= 1:1 (w/w). Exponential feeding stopped and linear feeding (constant feed rate of 17 g l-1 h-1) began at 24.9 h. Only DO data around the occurance of oxygen limitation (25 h) are shown, before which DO was always maintained around 40% air saturation. 112 Figure 6-2 Single-stage MCL-PHA production by P. putida KT2440 using nonanoic acid (NA) and glucose (G) co-feeding strategy with NA:G= 1:1.5 (w/w). Exponential feeding was applied throughout the fermentation. 113 Figure 6-3 (a) 3-OH-nonanoate (C9) to 3-OH-heptanoate (C7) ratio from NA:G= 1:1.5 co-feeding fermentation and NA sole feeding fermentation. (b) PHA composition throughout the NA:G= 1:1.5 fermentation. C5, 3-OH-valerate;
x

C7, 3-OH-heptanoate; C9, 3-OH-nonanoate; C-even, total of mainly 3-OH-hexanoate, 3-OH-octanoate and 3-OH-decanoate. m is the slope of the corresponding regression line. 114 Figure 6-4 (a) Substrate to biomass yield. (b) nonanoic acid (NA) to PHA yield plots. m is the slope of the corresponding linear regression line. All linear regression curves were forced through origin. 116 Figure 6-5 The relationship between PHA content and the cumulative nonanoic acid to overall overall PHA yield ( YPHA / NA ). YPHA / NA is the cumulative PHA yield from nonanoic acid at a given sampling point. 122 Figure 7-1 Fermentation a, exponential feeding of nonanoic acid (NA) and 10-undecenoic acid (UDA=) (molar ratio 4.98:1) with desired = 0.15 h-1. The dissolved oxygen data are shown after 30 h, before when it was kept around 40% air saturation. Curve fitting (dashed line) was done by nonlinear regression function of Sigmaplot 9.01. 133 Figure 7-2 Biomass, PHA concentration and PHA content of fermentations b and c, with the same (0.25 h-1) and different NA to UDA= ratio (5.07:1 and 2.47:1, respectively).

134

Figure 7-3 (a) Linear relationship between total amount of saturated components and total amount of unsaturated components from three fermentations. (b) Relative monomeric composition of the MCL-PHA synthesized during fermentation b. (c) Relative monomeric composition of the MCL-PHA synthesized during fermentation c. The dash-dot line in (b) and (c) divides saturated and unsaturated components. C5, 3-OH-valerate, C7, 3-OH-heptanoate, C7:1, 3-OH-heptenoate, C9, 3-OH-nonanoate, C9:1, 3-OH-nonenoate, C11, 3-OH-undecanoate, C11:1, 3-OH-undecenoate. The small amount of even carbon number 3-OH-alkanoates is not included these plots. 136 Figure 7-4 Relationship between unsaturation of the MCL-PHA product and the substrate for fermentations at desired = 0.25h-1 (fermentations b and c). The r2 is 0.99 for the regression curve.

140

xi

List of Abbreviations and Symbols

3HB or C4 3HV or C5 3HHx or C6 3HHp or C7 3HO or C8 3HN or C9 3HD or C10 3HDD or C12 C14 C7:1 C9:1 C11:1 C12:1 C14:1 C10O HCDC pH-stat DO-stat PHA PHB PHB/V PHB/HHx PHO SCL MCL G NA UDA= DO D max X0 X 3-hydroxybutyrate 3-hydroxyvalerate 3-hydroxyhexanoate 3-hydroxyheptanoate 3-hydroxyoctanoate 3-hydroxynonanoate 3-hydroxydecanoate 3-hydroxydodecanoate 3-hydroxy-cis-5-tetradecanoate 3-hydroxyheptenoate 3-hydroxy-8-nonenoate 3-hydroxy-10-undecenate 3-hydroxy-cis-6-dodecenoate 3-hydroxy-cis-5-tetradecenoate 3-hydroxy-7-oxooctanoate High-cell-density cultivation or culture A substrate feeding strategy to maintain the pH of the culture at a certain value or within a certain range A substrate feeding strategy to maintain the dissolved oxygen of the culture at a certain value or within a certain range Polyhydroxyalkanoates or PHA concentration, g l-1 Poly(3-hydroxybutyrate) Poly(3-hydroxybutyrate-co-3-hydroxyvalerate) Poly(3-hydroxybutyrate-co-3-hydroxyhexanoate) Poly(3-hydroxyoctanoate) Short-chain-length Medium-chain-length Glucose Nonanoic acid 10-undecenoic acid Dissolved oxygen expressed as percentage of air saturation, % Dilution rate in continuous fermentation, h-1 Specific growth rate, h-1 Maximum specific growth rate, h-1 Initial biomass at the time of inoculation, g l-1 Biomass, g l-1
xii

Xr St qPHA(Xr) qPHA(PHA) F PHA productivity X productivity C N P Mg Fe C/N % PHA YX/G YX/NA YX/NH4 YX/PO4 YX/Mg YX/S YCO2/G YCO2/X YCO2/S YP/S YPHA/NA Y(C7+C9)/NA Y(C8+C10)/G CPR CCP CGC t rNA rG

Residual biomass (X minus PHA concentration), g l-1 Substrate amount required at time t, g Specific PHA synthesis rate based on Xr, dP/dt/Xr, g PHA g Xr-1 h-1 Specific PHA synthesis rate based on PHA, dP/dt/PHA, g PHA g PHA-1 h-1 Substrate (nonanoic acid or glucose) feed rate, g l-1 h-1 g PHA l-1 h-1 g X l-1 h-1 Carbon Nitrogen Phosphorus Meganesium Iron Carbon to nitrogen ratio PHA content expressed as percent of PHA among total biomass, % Yield of biomass from glucose, g g-1 Yield of biomass from nonanoic acid, g g-1 Yield of biomass from ammonium, g g-1 Yield of biomass from phosphate, g g-1 Yield of biomass from magnesium, g g-1 Yield of biomass from substrate, g g-1 Yield of CO2 from glucose, g g-1 Yield of CO2 from biomass, g g-1 Yield of CO2 from substrate, g g-1 Yield of PHA from substrate, g g-1 Yield of PHA from nonanoic acid, g g-1 Yield of C7 and C9 PHA components from nonanoic acid, g g-1 or mol mol-1 Yield of C8 and C10 PHA components from glucose, g g-1 or mol mol-1 CO2 production rate, g CO2 h-1 Cumulative CO2 production, g Cumulative glucose consumption, g Cultivation time, h Proportion of the mass of nonanoic acid among the total NA+G feed, % Proportion of the mass of glucose among the total NA+G feed, %

xiii

CHAPTER 1 General Introduction 1.1 Background

Plastics have been produced by the chemical industry since the 1930s. Their formability and durability make them among the most versatile materials ever developed and they have found their way into nearly every line of business. Despite their versatility and wide applicability, their drawbacks are becoming increasingly obvious and cannot be ignored. Most plastics are produced from petroleum, a non-renewable resource. Furthermore, after being disposed, most synthetic materials are very persistent in the environment. This causes serious problems in waste management and has a negative impact on the environment. Therefore, people have been trying for decades to develop new plastic materials that are biodegradable and derived from renewable resources. Biological polyesters such as poly(hydroxyalkanoates) (PHAs) and poly(lactic acid) may fulfill such requirements.

PHAs have drawn great attention as a group of biopolyesters with properties similar to petroleum-based thermoplastics and thermoelastomers. They are mainly synthesized by microorganisms as intracellular carbon and energy storage materials. They are commonly classified as short-chain-length PHA (SCL-PHA), which contains 4 or 5 carbons in their repeating units; and medium-chain-length PHA (MCL-PHA), which contains 6 or more
1

carbons in the repeating units. At present, approximately 150 different constituents occurring in PHAs as homopolyesters or copolyesters have been identified (Steinbchel and Ltke-Eversloh 2003). Generally speaking, the longer the chain length of PHA, the lower its crystallinity and melting temperature (van der Walle et al. 2001). Biosynthesized PHAs are commonly known as biodegradable and biocompatible. They are not only promising as environmental friendly materials, but also as biomaterials for medical applications (Zinn et al. 2001).

Poly(3-hydroxybutyrate) (PHB) was the first PHA discovered by Lemoigne in 1926. Studies on the physiology, biosynthesis, properties and applications of SCL-PHAs have been carried out intensively since the 1960s. Although PHB and poly(3-hydroxybutyrateco-3-hydroxyvalerate) (PHB/V) had been commercialized under the trademark Biopol in the 1980s, their initial mission to provide a biodegradable replacement for petrochemical plastics was hindered by the high production cost. With the further development of better production strains and fermentation techniques to lower the production cost, the increasing price of petroleum, and the general increasing desire for sustainable and environmentally benign plastics, interest in PHA has grown in recent years. Metabolix Inc. and Archer Daniels Midland Company (ADM) recently announced a joint venture to build their first commercial scale plant in Clinton, Iowa, USA. They are

expecting to produce MirelTM Natural Plastics, which are a series of SCL-PHAs (Peoples 2006), at an annual rate of 110 million pounds beginning in 2008.

In contrast to SCL-PHA, studies on MCL-PHA have taken place only within the last two decades since the discovery of poly(3-hydroxyoctanoate) (PHO) (de Smet et al. 1983). Production techniques are not as well established as for SCL-PHA. One challenge is the availability of suitable microorganisms. Although many genera of bacteria have the ability to synthesize significant amount of SCL-PHAs, only certain species of Pseudomonas are capable of accumulating MCL-PHAs. Among those, the physiology of MCL-PHA production seems to differ considerably from one strain to another, rendering the fermentation process to be strain-specific. The major process challenge lies in control of the carbon source concentration, as well as process automation. Most MCL-PHA fermentation processes reported so far involve a substantial level of instrumental complexity and of manual operation. Hence, they are not suitable for industrial development. The current high cost of suitable carbon substrates (mainly aliphatics) for MCL-PHAs synthesis present another challenge for industrial scale production. Certain substrates, such as oleic acid, albeit supporting substantial MCL-PHA synthesis, result in PHAs with a fixed but complex composition, and relatively undesirable mechanical properties.

It is well known that MCL-PHAs have wide diversity and are good candidates for chemical modification. Therefore, they are very promising in a wide range of potential applications, such as adhesives, paints, coatings (van der Walle et al. 2001), drug delivery (Pouton and Akhtar 1996), tissue engineering (Chen and Wu 2005) and other medical uses. It is now essential to develop highly efficient, repeatable and reliable MCL-PHA production processes in order to increase the availability of these materials for application-related research and commercial production.

1.2 Overall objective

The overall objective of this research is to develop automated, reliable, well-controlled, productive lab-scale fermentation processes for the production of MCL-PHAs from Pseudomonas putida KT2440 (ATCC 47054).

1.3 Approaches
The overall objective was achieved in five phases of study.

It was widely accepted in the literature that MCL-PHAs synthesis is stimulated by the limitation of certain major growth nutrients (such as nitrogen or phosphorus) and an excess supply of carbon source. Therefore, a two-stage fed-batch fermentation scheme is usually employed. During the first stage (growth stage), the microorganism is allowed to grow to a high cell concentration, which provides the basis for potential high intracellular
4

PHA synthesis. Then a major nutrient is limited but carbon supply is continued, such that the culture stops growth and begins PHA synthesis. Such a two-stage fed-batch scheme was initially adopted for our process development.

To make such a two-stage process more economically favorable, an inexpensive carbon source (glucose) was used for the growth stage, while a more expensive but also more efficient MCL-PHA-synthesis substrate (nonanoic acid) was used for most of the PHA-synthesis stage. Key aspects of a successful fed-batch fermentation process involve the selection of a suitable microorganism, the formulation of a balanced cultivation medium, and medium feeding strategies. Pseudomonas putida KT2440 was chosen and a balanced medium was formulated. Four substrate feeding strategies were developed to achieve high-cell-density culture. Each of these feeding strategies can be fully automated, and resulted in biomass productivities that were similar or higher than previous literature work with related strains of Pseudomonas. This phase of research is covered in Chapter 3, Automated feeding strategies for high-cell-density fed-batch cultivation of Pseudomonas putida KT2440 , published in Applied Microbiology and Biotechnology 71 (4): 423-431, 2006.

Following the high-cell-density growth phase of P. putida KT2440, the second-stage (PHA-synthesis stage) was investigated using nonanoic acid as the major carbon substrate, and also by imposing nitrogen limitation. According to the literature, nonanoic acid is the
5

most efficient carbon substrate for MCL-PHA synthesis in P. putida, and can be considered a renewable carbon source since it is produced from oleic or other carboxylic acids derived from temperate plants such as canola. However, no published fermentation process had used nonanoic acid as the carbon substrate, partly due to its toxicity to the culture. In my study of two-stage fed-batch fermentation, a nonanoic acid linear feeding strategy was developed based on the estimated substrate consumption rate, and resulted in a final PHA content of nearly 30%. A nonanoic acid and glucose co-feeding strategy was also established such that the yield of MCL-PHA from nonanoic acid was almost doubled by the co-feeding of glucose. This phase of research is covered in Chapter 4, Increasing the yield of MCL-PHA from nonanoic acid by co-feeding glucose during the PHA accumulation stage in two-stage fed-batch fermentations of Pseudomonas putida KT2440, published in the Journal of Biotechnology, March, 2007 online.

Although it was potentially possible to achieve a high final PHA content using the two-stage fed-batch process, the overall PHA productivity was low, due to prolonged fermentation time. A chemostat study on the optimal PHA synthesis conditions demonstrated that substantial PHA synthesis was possible in P. putida KT2440 when carbon-limited, while other growth nutrients (nitrogen or phosphorus) were in excess. Based on this finding, a single-stage exponential nonanoic acid feeding fed-batch process was developed. Substantial MCL-PHA synthesis occurred during the exponential growth
6

of P. putida KT2440. This process is by far the easiest and most automated approach for the production of MCL-PHAs, and resulted in the highest final MCL-PHA content (75%) reported in the literature. This phase of research is covered in Chapter 5, Carbon-limited fed-batch production of medium-chain-length polyhydroxyalkanoates from nonanoic acid by Pseudomonas putida KT2440, published in Applied Microbiology and Biotechnology, 74 (1): 69-77, 2007.

Since the cost of carbon substrate(s) usually accounts for more than 30% of the total production cost for either SCL-PHA production (Choi and Lee 1997) or MCL-PHA (Hazenberg and Witholt 1997), it is highly desirable to reduce the cost of substrate and/or increase the PHA yield as much as possible. Therefore, a carbon-limited single-stage MCL-PHA fermentation process using nonanoic acid with glucose as co-substrate was developed. Compared to the nonanoic acid as sole substrate process, the co-substrate process resulted in a similar final PHA content and overall volumetric productivity, a similar PHA composition, and a nonanoic acid to PHA yield enhancement of 30%. Consequently, the total carbon substrate cost may be reduced by about 15%. This phase of research is covered in Chapter 6, Fed-batch production of medium-chain-length poly(3-hydroxyalkanoates) by Pseudomonas putida KT2440 using nonanoic acid and glucose co-feeding, to be submitted to Journal of Biotechnology in June, 2007.

The great potential of MCL-PHAs in many applications is largely due to the wide diversity of the possible structures. Besides saturated MCL-PHAs, functionalized MCL-PHAs (with functional groups in the side chains) can also be synthesized. Among possible functionalized MCL-PHAs, those with olefin groups (unsaturated MCL-PHAs) are considered the most useful functionalized MCL-PHAs, as many chemical modifications can be made to these polymers (Hazer and Steinbchel 2007). Based on the single-stage nonanoic acid exponential feeding process, nonanoic acid and 10-undecenoic acid co-feeding processes were developed and unsaturated MCL-PHA productivity as high as 1.1 g PHA l-1 h-1 (highest reported for unsaturated MCL-PHA production) was achieved. A linear relationship between the co-substrate mixture composition and the resulted MCL-PHA composition were confirmed, demonstrating that the production of functionalized MCL-PHA with controllable composition by fed-batch fermentation is possible. This phase of research is covered in Chapter 7, Fed-batch production of unsaturated medium-chain-length poly(3-hydroxyalkanoates) with controlled composition by Pseudomonas putida KT2440, to be submitted to Applied Microbiology and Biotechnology in June, 2007.

1.4 References
Chen GQ, Wu Q (2005) The application of polyhydroxyalkanoates as tissue engineering materials. Biomaterials 26:6565-6578

Choi JI and Lee SY (1997) Process analysis and economic evaluation for poly(3-hydroxybutyrate) production by fermentation. Bioprocess Eng 17:335-342 de Smet MJ, Eggink G, Witholt B, Kingma J, Wynberg H (1983) Characterization of intracellular inclusions formed by Pseudomonas oleovorans during growth on octane. J Bacteriol 154:870-878 Hazenberg W and Witholt B (1997) Efficient production of medium-chain-length poly(3-hydroxyalkanoates) from octane by Pseudomonas oleovorans: Economic considerations. Appl Microbiol Biotechnol 48:588-596 Hazer B, Steinbchel A (2007) Increased diversification of polyhydroxyalkanoates by modification reactions for industrial and medical applications. Appl Microbiol Biotechnol 74:1-12 Peoples OP (2006) PHA natural plastics: a disruptive technology for a sustainable future. International Symposium on Biological Polyesters 2006, Minneapolis/St. Paul, Minnesota, USA Pouton CW, Akhtar S (1996) Biosynthetic polyhydroxyalkanoates and their potential in drug delivery. Adv Drug Deliv Rev 18:133-162 Steinbchel A, Ltke-Eversloh T (2003) Metabolic engineering and pathway construction for biotechnological production of relevant polyhydroxyalkanoates in microorganisms. Biochem Eng J 16:81-96 van der Walle GAM, de Koning GJM, Weusthuis RA, Eggink G (2001) Properties, modifications and applications of biopolyesters. In Babel W and Steinbchel A edit, Adv Biochem Eng Biotechnol 71:263-291 Zinn M, Witholt B, Egli T (2001) Occurrence, synthesis and medical application of bacterial polyhydroxyalkanoate. Adv Drug Deliv Rev 53:5-21

CHAPTER 2 Literature Review

Fermentation Process Development for the Production of Medium-chain-length Poly(3-hydroxyalkanoates)

Zhiyong Sun, Juliana A. Ramsay, Martin Guay, Bruce A. Ramsay

Originally published June, 2007 in Applied Microbiology and Biotechnology, 75 (3): 475-485
10

2.1 Abstract
This paper presents a review of existing fermentation processes for the production of medium-chain-length poly(3-hydroxyalkanoates) (MCL-PHAs). These biodegradable polymers are usually produced most efficiently from structurally related carbon sources such as alkanes and alkanoic acids. Unlike alkanoic acids, alkanes exhibit little toxicity but their low aqueous solubility limits their use in high density culture. Alkanoic acids pose little mass transfer difficulty, but their toxicity requires that their concentration be well controlled. Using presently available technology, large-scale production of MCL-PHA from octane has been reported to cost from 5 to 10 US $ kg-1 with expenditures almost evenly divided between carbon source, fermentation process and the separation process. However, MCL-PHAs, even some with functional groups in their subunits, can also be produced from cheaper unrelated carbon sources, such as glucose. Metabolic engineering and other approaches should also allow increased PHA cellular content to be achieved. These approaches, as well as a better understanding of fermentation kinetics, will likely result in increased productivity and lower production costs.

2.2 Introduction
Poly(3-hydroxyalkanoates) (PHAs) have attracted extensive interest due to their biocompatibility and biodegradability. Since approximately 150 different PHA subunits
11

have been identified (Steinbchel and Ltke-Eversloh 2003), PHAs exhibit a wide variety of properties and thus may have many different applications. Short-chain-length PHAs (SCL-PHAs) such as poly(3-hydroxybutyrate) (PHB) have been studied in depth and have been produced on a commercial scale. Due to structural differences (Figure 2-1), the physical properties of medium-chain-length PHAs (MCL-PHAs), are generally quite different from PHB and other SCL-PHAs (Gagnon et al. 1992). PHAs are more expensive to produce than conventional plastics so they must find high value applications to be commercially viable. MCL-PHAs have shown promise as thermoelastomers for biomedical applications, such as drug delivery (Pouton and Akhtar 1996) and tissue engineering (Williams et al. 1999; Chen and Wu 2005). However, to date, the process development for the production of MCL-PHAs has been much less extensive than for SCL-PHAs. The consequent lack of availability has hampered application development. Efficient MCL-PHA synthesis occurs when using structural-related carbon sources, such as alkanes and aliphatic acids (Lageveen et al. 1988). These carbon sources are either poorly miscible with water and/or toxic to bacteria at relatively low concentrations. The major process challenge lies in automation of the control of these carbon source(s) concentration which generally govern the rates of growth and production. The highly reduced nature of these aliphatic carbon sources results in higher oxygen demand than carbohydrates and they are generally much more expensive than conventional

12

fermentation substrates such as carbohydrates. Increasing the efficiency of their use in PHA synthesis is critical to the overall economics.
R O R O R O

HO

O n

OH

Figure 2-1 Chemical structure of polyhydroxyalkanoates (PHAs), which are classified as short-chain-length PHAs (SCL-PHAs) when the repeating units contain 4 or 5 carbons and medium-chain-length PHAs (MCL-PHAs) when the repeating units contain 6 or more carbons; R pendant group may also contains functional groups resulting in functionalized PHAs.

In this paper, the effects of physiological conditions on biosynthesis kinetics of MCL-PHA-synthesizing bacteria and fermentation strategies for MCL-PHA production will be reviewed. Economic considerations will be briefly discussed.

2.3 Physiology and the kinetics of biosynthesis

2.3.1 General considerations
The occurrence of MCL-PHAs was first observed in Pseudomonas oleovorans GPo1 (ATCC 29347) (de Smet et al. 1983). Early screening studies showed that only certain species of fluorescent pseudomonads could synthesize MCL-PHAs (Huisman et al. 1989; Timm and Steinbchel 1990). More recently, 48 Pseudomonas species belonging to the rRNA homology group I were examined for their ability to synthesize PHA from octanoate (Diard et al. 2002). These organisms have been grouped into intrageneric clusters I and II (Yamamoto et al. 2000). The species in cluster II, as well as certain
13

strains of P. aeruginosa in cluster I synthesized MCL-PHAs with 3HO as the major component. Other cluster I organisms, including P. oleovorans strains, synthesized only PHB. It is important to note that one of the mostly frequently studied strains, P. oleovorans GPo1 (ATCC 29347) for the MCL-PHA synthesis studies, has been reclassified as P. putida GPo1 (ATCC 29347) (van Beilen et al. 2001; Diard et al. 2002). This strain is unique among MCL-PHA-synthesizing Pseudomonas, in that it can utilize alkanes such as octane for MCL-PHA synthesis due to its OCT plasmid (Huisman et al. 1989). Alkanoates such as octanoate are common MCL-PHA carbon sources for all MCL-PHA-synthesizing Pseudomonas.

Since the metabolism of MCL-PHA synthesis is quite different from that of SCL-PHA synthesis, early studies attempted to understand the physiological and kinetic fundamentals of MCL-PHA production. The most interesting aspects for process development are the effects of carbon sources and their concentration, specific growth rate (), and the ratio of carbon to other major nutrients on cell growth and MCL-PHA synthesis. Physiological conditions directly affect subunit composition, cellular PHA content (%), specific PHA synthesis rate based on residual biomass (qPHA(Xr), g PHA g Xr-1 h-1), and overall PHA productivity (g PHA l-1 h-1). Throughout this review, biomass (X) is used to indicate dry biomass while residual biomass (Xr) is defined as dry biomass minus PHA.
14

2.3.2 Carbon sources and their concentration

It has been clearly demonstrated that fatty acid -oxidation is involved in supplying the precursors for MCL-PHA synthesis from structurally related carbon sources (Eggink et al. 1992). MCL-PHA synthesis results in copolyesters containing C6 to C14 units but carbon sources, such as octane, octanoate, and nonanoate, are considered to be the most efficient substrates for MCL-PHA synthesis (Madison and Huisman 1999), also shown in Figure 2-2. When P. putida GPo1 was grown on octanoate and nonanoate under nitrogen limitation in a chemostat study (= 0.2 h-1), comparable PHA content, 42% and 31% were respectively obtained (Durner et al. 2001), but such results are not likely the maximum obtainable from these substrates since PHA content often depends on (Preusting et al. 1991; Ramsay et al. 1991). During the PHA accumulation phase of the batch cultivation of P. putida GPo1, the qPHA(Xr) using nonanoate was considerably higher than using octanoate (0.082 g PHA g Xr-1 h-1 vs. 0.036 g PHA g Xr-1 h-1) (Durner et al. 2001), suggesting that nonanoate is a more efficient carbon source for PHA synthesis. To produce MCL-PHAs from cheaper and more widely available resources, vegetable oils and animal fats are of interest as many Pseudomonas are able to utilize such carbon sources. Oleic acid, a common carboxylic acid in vegetable oils, has been used in the efficient production of MCL-PHAs by P. putida KT2442 (Huijberts and Eggink 1996; Lee et al. 2000). Other complex substrates, such as coconut oil and tallow (Solaiman et al. 1999; Thakor et al. 2005), co-products of soy-based biodiesel production (Ashby et al.
15

2004), and soy molasses (Solaiman et al. 2006) have been reported to support MCL-PHA synthesis by various Pseudomonas strains. Although some strains may accumulate as much as 40% PHA using such carbon sources, the yields are usually much lower than pure carboxylic acids. And since most of these studies were done in shake-flask culture or lab-scale batch fermentations, further work is required to evaluate their economic potential.

80 P. putida GPo1 on n-alkanoate P. putida GPo1 on n-alkane P. sp. DSY-82 on n-alkanoate 60

% PHA

40

20

0 5 6 7 8 9 10 11 12 13

Number of carbons

Figure 2-2 PHA content of P. putida strains using alkanes or alkanoates of varying length. P. putida GPo1 (formerly known as P. oleovorans GPo1) grown on n-alkanoates (Gross et al. 1989), P. putida GPo1 on n-alkanes (Lageveen et al. 1988), and P. sp. DSY-82 on n-alkanoates (Kang et al. 2001).

16

Structurally unrelated carbon sources, such as gluconate, can also be used for PHA synthesis by most MCL-PHA-synthesizing Pseudomonas (Huijberts et al. 1992; Timm and Steinbchel 1990), but this is generally less efficient than structurally related substrates (Madison and Huisman 1999). However, P. putida IPT046 (Table 2-1) has demonstrated very high MCL-PHA-synthesizing ability from carbohydrates (Sanchez et al. 2003) and has been evaluated in lab scale fermentations (Diniz et al. 2004).

The toxic nature of some carbon sources impedes process development. Hydrocarbons such as octane was often used as an organic second phase in the cultivation of P. putida GPo1 and exhibited little apparent toxicity (Preusting et al. 1993b), but growth at high cell concentration may be mass transfer limited (Hazenberg and Witholt 1997). Octane is also flammable. In contrast, alkanoate salts are miscible in aqueous media but exhibit toxicity. Growth of P. putida GPo1 was inhibited by 4.65 g l-1 octanoate (Ramsay et al. 1991) while P. putida KT2440 (ATCC 47054) was inhibited by 3-4 g l-1 nonanoic acid (Sun et al. 2007/Chapter 5). Vegetable oils are generally less inhibitory as are longer chain fatty acids. Oleic acid was good for growth of P. putida KT2442 if maintained below 15 g l-1 (Lee et al. 2000). Glucose did not inhibit the growth of P. putida if kept below 40 g l-1 (Hofer et al. 2002; Kim et al. 1996). The inhibitory effects of these carbon sources commonly used in MCL-PHA synthesis pose considerable difficulty in the development of high-cell-density production processes.
17

2.3.3 Growth rate and nutrient conditions

Chemostat studies of P. putida GPo1 on octane (Preusting et al. 1991) and octanoate (Durner et al. 2000; Ramsay et al. 1991), and P. putida KT2442 on oleic acid (Huijberts and Eggink 1996) have shown that cellular PHA content decreases with increasing dilution rate (equals to at steady state) with the highest content obtained at the lowest .

The effects of nutrient conditions on the synthesis of MCL-PHAs by various strains differed considerably. Although MCL-PHA synthesis was significant during nutrient (e.g. N or P) limitation with excess carbon, as shown by many batch and chemostat experiments (Hazenberg and Witholt 1997; Kim et al. 1997; Lageveen et al. 1988), this is not a necessity to obtain MCL-PHA for some strains using certain carbon sources. P. putida GPo1 synthesized PHA during batch exponential growth at a rate even higher than biomass formation; and synthesized 19% PHA under carbon limited chemostat conditions when grown on nonanoate but not octanoate (Durner et al. 2001). P. putida KT2440 synthesized up to 75% PHA during high-cell-density carbon-limited exponential growth (Sun et al. 2007/Chapter 5). Its rifampicin-resistant derivative P. putida KT2442 also demonstrated significant PHA-synthesizing ability during batch exponential phase on octanoate (Huisman et al. 1992) and has been cited as not requiring nutrient limitation for MCL-PHA synthesis (Kessler et al, 2001).

18

In a chemostat study of P. putida GPo1 growing on octanoate at = 0.2 h-1, the PHA content increased linearly from 2% to 42% as the carbon to nitrogen (C/N) ratio increased while both C and N were limited, but was constant at 40% as C/N ratio was further increased (the N-limited zone) (Durner et al. 2001). In another study with the same strain, constant PHA content (15%) was also observed at an increasing C/N ratio under N limitation (Ramsay et al. 1991). In contrast, chemostat study of P. resinovorans ATCC 14235 on octanoate (Ramsay et al. 1992) and P. putida KT2442 on oleic acid (Huijberts and Eggink 1996), showed that the highest PHA content was obtained via nitrogen limitation (17 mol C mol N -1 at 0.25 h-1 and 20 mol C mol N -1 at 0.1 h-1, respectively). Therefore, it cannot be generally stated that MCL-PHA synthesis is stimulated by non-carbon nutrient limitation. Based on the results reported, it is necessary to assess the optimal PHA-synthesis conditions when any new Pseudomonas strain, carbon source, or nutrient composition is to be employed.

2.4 Process development

To date, P. putida GPo1, P. putida KT2440 and KT2442, P. putida BM01 and P. putida IPT046 have been employed for fermentation process development using octane, octanoate, oleic acid, or carbohydrates (Table 2-1). Although efforts have been made to express MCL-PHA in recombinant Escherichia coli (Langenbach et al. 1997; Qi et al. 1998; Ren et al. 2000) or to improve PHA synthesis with recombinant Pseudomonas
19

(Kraak et al. 1997), few fermentation processes have been based on recombinant strains (Prieto et al. 1999). Due to the common belief that PHA synthesis is stimulated by nutrient limitation, almost all MCL-PHA production processes have incorporated an N or P limitation.

2.4.1 Continuous processes

Continuous fermentation is not only an appropriate method to study microbial physiology, but also a good way to achieve high productivity. Studies have been carried out to find suitable strategies for PHA production using this mode of operation. High-cell-density, continuous P. putida GPo1 fermentation processes using octane have been developed to high productivity levels by Witholt and coworkers. In their early study, P. putida GPo1 was cultivated at a cell concentration of 11.6 g l-1 containing 25% PHA with 12.5% (v/v) octane as the organic phase and ammonium limited conditions in a chemostat operated at D= 0.2 h-1 (Preusting et al. 1993a). The yields of biomass from major nutrients (P, Mg, Fe, N) were determined to allow medium composition optimization. This led to PHA productivity of 0.56 g PHA l-1 h-1. This process was operated continuously for 30 days with no significant degeneration of the culture. By further improving the medium composition and mixing capability of the bioreactor, enhanced PHA productivity (0.76 g PHA l-1 h-1) was achieved (Hazenberg and Witholt 1997). As previously mentioned, PHA content decreased with increased (or D in steady
20

state of continuous fermentations). Thus, a compromise between PHA content, concentration and productivity is required in a single-stage continuous process if a reasonable percent of PHA is to be attained in the biomass. To overcome this limitation, a two-stage continuous process was developed (Jung et al. 2001). Dilution rates to achieve maximum overall volumetric PHA productivity and content were determined to be 0.21 h-1 for cell growth (1st stage) and 0.16 h-1 for N-limited PHA-accumulation (2nd stage), using transient experiments. Under these conditions, a high-cell-density continuous fermentation was carried out, resulting in 18 g l-1 cell concentration containing 63% PHA in the second stage, and an overall PHA productivity of 1.06 g PHA l-1 h-1.

High-cell-density continuous processes of P. putida KT2442 using oleic acid have also been developed (Huijberts and Eggink 1996). Fed-batch operation was used to obtain a high cell concentration before switching to continuous mode. A dilution rate of 0.1 h-1 and a C/N ratio of 20 mol mol-1 was used to achieve maximum PHA productivity. Although the productivity was 0.67 g PHA l-1 h-1, comparable to that of P. putida GPo1 process (Hazenberg and Witholt 1997; Jung et al. 2001), the PHA content was only 23%.

2.4.2 Fed-batch processes

Fed-batch cultivation is often employed to achieve high product concentrations. Many different feeding control techniques have been reported (Lee et al, 1999). Fed-batch strategies have been applied to high-cell-density cultivation of P. putida GPo1 in a
21

two-phase system, using octane. Since octane does not appear to inhibit culture growth, ammonium limited feeding strategies were applied. In one study (Preusting et al. 1993b), ammonium was fed at a constant rate of 0.23 g h-1 for 38 h which led to a final biomass of 37.1 g l-1 containing 33% PHA (0.25 g PHA l-1 h-1). A high oxygen transfer rate was generally needed to obtain high cell density. Using a reactor with better oxygen-transfer capacity and optimized medium composition, P. putida GPo1 was grown up to 90 g l-1 using linear ammonium feeding and 112 g l-1 using exponential ammonium feeding (Kellerhals et al. 1999a). However, a maximum PHA content of only 25% was obtained in the middle of the process and decreased significantly until the end of the fermentation. Based on information from previous chemostat work (Hazenberg and Witholt 1997), it was possible that an octane mass transfer limitation hindered PHA accumulation. Therefore such two-phase systems might not allow efficient MCL-PHA production in high cell density fed-batch mode.

Alkanoates have also been used in MCL-PHA processes; but controlling their concentrations at a non-toxic level is a major challenge in achieving high-cell-density. Octanoate has been used as an alternative to octane in fed-batch processes for P. putida GPo1. In one such study, ammonium octanoate was first fed at an exponential rate resulting in a of 0.19 h-1, then adjusted to a constant rate to prevent oxygen limitation (Dufresne and Samain 1998). The pH was regulated by addition of octanoic acid which
22

also served as an additional carbon source. A final biomass of 47 g l-1 was obtained containing 55% PHA. Similarly, pH-stat feeding of a mixture of octanoic acid and ammonium nitrate was applied by Kim (2002). The C/N ratio of 20 g octanoic acid g-1 ammonium nitrate resulted in 75% PHA but a lower overall productivity (0.63 g PHA l-1 h-1) compared to C/N of 10 g g-1 that result in 1 g PHA l-1 h-1, mainly due to a prolonged fermentation time.

Other P. putida strains have been employed in fed-batch production of MCL-PHA. The P. putida BM01, isolated by Kims group, can grow on glucose up to 100 g l-1 cell concentration, but synthesized little PHA (Kim et al. 1996). Therefore a two-stage fed-batch process using glucose as growth substrate and octanoic acid as PHA-synthesis substrate was developed so that the overall yield of PHA from octanoic acid, which is much more expensive than glucose, could be improved (Kim et al. 1997). It was also found that a slight supplement of octanoic acid while using glucose as the main carbon source during the growth stage could shorten the adaptation period of the culture for the PHA-accumulating stage. Using such a two-stage process, biomass (55 g l-1) containing 65.5% of PHA was obtained and the overall yield of PHA from octanoate was 0.4 g g-1. It was not clear, however, how the octanoate concentration was maintained at an appropriate level during the feeding process.

23

Growth of P. putida KT2442 on oleic acid was also investigated (Weusthuis et al. 1997). One approach employed a DO-stat strategy in an oleic acid fed-batch process. Pulses of oleic acid below the inhibitory concentration were fed whenever an increase in dissolved oxygen indicated a depletion of carbon source. This feeding strategy produced 92 g l-1 of biomass containing 47% PHA (1.6 g PHA l-1 h-1). In another study, using the same strain and carbon source, phosphorus limitation was applied instead of nitrogen limitation (Lee et al. 2000). By extending the process using pH-stat feeding strategy following DO-stat feeding, the final biomass, PHA concentration and productivity was improved to 141 g l-1, 51% PHA and 1.9 g PHA l-1 h-1, respectively. Similar processes have been reported using a 30 l pilot-scale bioreactor (Kellerhals et al. 2000), but it seemed that nitrogen limitation hampered the increase in PHA content in the biomass and hence the overall productivity.

An alternative to the pH-stat or DO-stat feeding strategy is to control the feed directly by determining the actual carbon source concentration using rapid off-line measurement. Such off-line measurement equipment is now commonly used in glucose fermentations (Kim et al. 1996) but rare when alkanoates are the carbon source. A closed-loop off-line gas chromatography measurement system was developed to directly measure and control the medium concentration of octanoate within a selected range for the cultivation of P. putida KT2442 (Kellerhals et al. 1999b). The octanoate concentration was very well
24

maintained below 25 mM, but the PHA content declined from 50% during the middle of the fermentation to 34% at the end of the fermentation, and resulted in only 17.4 g l-1 of PHA. Despite the fact that such a system is excellent for the direct monitoring and control of toxic substrates, its complexity renders it less likely to be used commercially if, simpler, albeit indirect, feeding strategies are available.

In our recent study (Sun et al. 2007/Chapter 5), simple yet efficient exponential feeding strategies were applied to cultivate P. putida KT2440 using nonanoic acid. This approach was possible because MCL-PHAs synthesis occurs significantly along with cell growth, and no nutrient-limitation conditions were required. PHA content of 75% was obtained using a lower controlled specific growth rate (0.15 h-1) while higher overall productivity was achieved at higher controlled specific growth rate (0.25 h-1). This process was subsequently repeated in a 150 l bioreactor with similar success (unpublished data). From the reported results it seems that oxygen limitation was the only reason hampering the further increase of cell concentration and PHA content. Although most MCL-PHA-synthesizing strains can accumulate PHA from unrelated carbon sources (Huijberts et al. 1992; Timm and Steinbchel 1990), the achievable PHA content is usually unacceptably low. Many such strains also accumulate large amounts of undesirable polysaccharides if grown on carbohydrates. The exception is P. putida IPT046 (Sanchez et al. 2003) for which fermentation processes have been developed
25

(Diniz et al. 2004). Glucose and fructose pulse feeding, constant rate feeding and exponential feeding strategies with no nutrient limitation were all tested, but resulted in little difference in the final achievable cell concentration (about 30 g l-1). Following an exponentially-fed growth phase, ammonium limited or phosphate limited PHA-accumulation phases were investigated and resulted in a final biomass 40 g l-1 containing 21% PHA or 50 g l-1 containing 63% PHA, respectively. Such a process has industrial appeal as the control strategies are relatively simple, and more importantly, the carbon sources are cheaper and more widely available than alkanes and alkanoates.

2.4.3 Production of MCL-PHAs containing functional groups

In addition to MCL-PHAs containing solely 3-hydroxyalkanoates subunits, MCL-PHAs containing various functional groups, such as olefins (Fritzsche et al. 1990), branched alkyls (Hazer et al. 1994), halogens (Kim et al. 1996), phenyl (Hazer et al. 1996), and esters (Scholz et al. 1994) can often be incorporated when grown on suitable carbon substrates with such groups. However, the ability to utilize these carbon sources may differ considerably between strains (Kim et al. 2000). Few of these polymers have been produced at even gram quantities due to the toxicity of the carbon substrates, inefficient accumulation by the cultures employed, and unavailability of such substrates as many of them require dedicated chemical synthesis.

26

Among such functionalized PHAs, the synthesis of PHAs containing unsaturated subunits draws the most interest. Medium chain length alkenes and alkenoic acids, which are more easily available than many other functional substrates, have been used in such studies. When using octene or nonene as sole carbon source for P. oleovorans (Lageveen et al. 1988), the resulting PHA contained up to 55% of unsaturated subunits, probably due to the -oxidation at either end of the alkene. The biomass and PHA yields were lower from alkenes than from alkanes (Kim et al. 1995; Lageveen et al. 1988). To overcome such inefficiency but still obtain an unsaturated fraction in synthesized PHA, 10-undecenoic acid was mixed with octanoic or nonanoic acid as co-substrate (Kim et al. 1995), and the fraction of unsaturated units in the resulting polymers increased in proportion to the fraction of undecenoic acid in the mixture. This kind of co-substrate cultivation not only leads to better growth, but also allows control of the fraction of unsaturated subunits. Using the closed-loop gas chromatography system, sodium-octanoate and sodium-10-undecenoate were co-fed via a PID feedback loop in a fed-batch fermentation (Kellerhals et al. 1999b). Although this system was effective in maintaining the carbon compounds below their toxic level, the final cell concentration and MCL-PHA content were only 30 g l-1 and 33%, respectively, probably due to a limited nitrogen supply and insufficient fermentation time. In Hartmanns study, the variation in polymer compositions from batch and chemostat fermentations co-fed with octanoic and 10-undecenoic acids were compared (Hartmann et al. 2006). It was
27

concluded that chemostat cultivation is the most suitable method of obtaining functionalized MCL-PHAs with defined and consistent monomer compositions. Besides unsaturated MCL-PHAs, up to 2 g l-1 PHAs containing oxo side chains from octane and 2-octanone mixtures, and PHAs containing acetoxy side chain from octane and n-octylacetate mixture, have been produced using a two-phase continuous culturing technique described previously (Jung et al. 2000). As demonstrated in these studies, processes for producing functionalized PHAs can be very similar to that of regular MCL-PHAs. Such processes normally involve at least two carbon sources, one being a standard medium-chain-length carbon source and the other containing the functional group desired. It should be mentioned, however, that unsaturated subunits of MCL-PHAs can also be produced from unrelated carbon sources, such as glucose (Sanchez et al. 2003). This approach could potentially avoid the necessity of using expensive and inhibitory co-substrates for functionalized MCL-PHA production.

28

Table 2-1 Summary of MCL-PHA (regular and functionalized) production processes.

Fermentation mode Continuous (D= 0.2 h-1) Continuous (D= 0.2 h-1) Two-stage continuous (D= 0.21 h-1/ 0.16 h-1) Continuous (D= 0.1 h-1) Fed-batch Fed-batch Fed-batch Microorganism P. putida GPo1 (ATCC 29347) P. putida GPo1 P. putida GPo1 Carbon source (s) Octane Octane Octane Cell concentration (g l-1) 11.6 12.4 10.50; 18 PHA concentration (g l-1) 2.9 3.7 4.04; 11.34 PHA content (%) 25 30 38.5; 63 PHA productivity (g PHA l-1 h-1) 0.56 0.74 1.06 PHA subunits (high to low) C8, C6 C8, C6 C8, C6 Reference (Preusting et al. 1993a) (Hazenberg and Witholt 1997) (Jung et al. 2001) (Huijberts and Eggink 1996) (Preusting et al. 1993b) (Kellerhals et al. 1999a) (Dufresne and Samain 1998) (Kim 2002) (Kim et al. 1997) (Weusthuis et al. 1997) (Lee et al. 2000) (Kellerhals et al. 1999b) (Kellerhals et al. 2000)

P. putida KT2442 P. putida GPo1 P. putida GPo1 P. putida GPo1

Oleic acid Octane Octane Ammonium octanoate/ octanoic acid Octanoic acid Octanoic acid/ glucose Oleic acid Oleic acid Octanoic acid Oleic acid Octanoic acid Vegetable fatty acid Nonanoic acid

30 37.1 112 47

6.9 12.1 11.2 25.9

23 33 10 55

0.69 0.25 0.19 0.53

C8, C10, C12, C14:1, C6, C12:1 C8, C6 C8, C6 C8, C6

Fed-batch Two-stage fed-batch Fed-batch Fed-batch Fed-batch Fed-batch

P. putida GPo1 P. putida BM01 P. putida KT2442 P. putida KT2442 P. putida KT2442 P. putida KT2442

63 55 55 92 141 51.5 89.8 51.5 73 70 56

39 41 36 43 72 17.4 18.0 18.0 25 52.5 37.4 29

62 75 65.5 47 51 34 20 35 34 75 67

1 0.63 0.90 1.60 1.90 0.40 0.56 0.41 0.56 1.11 1.44

C8, C6 C8, C6, C10 C8, C10, C12, C14:1, C6 C8, C10, C6, C12, C14, C8, C6, C10 C8, C10, C12, C14:1, C6 C8, C6, C10 C8, C10, C6, C12:1, C14:1 C9, C7 C9, C7

Fed-batch

P. putida KT2440 (ATCC 47054)

(Sun et al. 2007)

Fed-batch Continuous (D= 0.2 h-1) Fed-batch Continuous (D= 0.1-0.3 h-1) Fed-batch

P. putida IPT046 P. oleovorans (putida) POMC1 (recombinant) E. coli 193MC1 (recombinant) P. putida GPo1

Glucose/fru ctose Octanoic acid/ citric acid Octanoic acid/ glycerol Octanoate (50%)/ 10-undecen oate (50%) 2-octanone/ octane

50 1.50 3.55 1.11-1.44

31.5 0.65 0.43 0.13-0.47

63 43 12 12-37

0.75 0.016 0.011 0.04-0.07

C10, C8 C8, C6 C8, C10, C6 C8, C9:1, C11:1, C6, C7:1 C8, C10O, C6

(Diniz et al. 2004; Sanchez et al. 2003) (Prieto et al. 1999) (Prieto et al. 1999) (Hartmann et al. 2006) (Jung et al. 2000)

P. putida GPo1

9.5

2.52

26.5

0.063

C6, 3-hydroxyhexanoate; C7, 3-hydroxyheptanoate; C7:1, 3-hydroxy-6-heptenoate; C8, 3-hydroxyoctanoate; C9, 3-hydroxynonanoate; C9:1, 3-hydroxy-8-nonenoate; C10, 3-hydroxydecanoate; C10O, 3-hydroxy-7-oxooctanoate; C11:1, 3-hydroxy-10-undecenoate; C12, 3-hydroxydodecanoate; C12:1, 3-hydroxy-cis-6-dodecenoate; C14, 3-hydroxy-cis-5-tetradecanoate; C14:1, 3-hydroxy-cis-5-tetradecenoate; P. putida GPo1 was formerly known as P. oleovorans GPo1.

30

2.5 The effect of nutrient limitation on MCL-PHA accumulation

When comparing the MCL-PHA production processes described in this report and summarized in Table 2-1, it is necessary to consider the effects of nutrient limiting and nutrient depleted conditions. Processes involving complete nutrient depletion are usually two-stage fed-batch processes with PHA accumulation occurring mainly during the nutrient-depleted stage (Kim et al. 1997; Lee et al. 2000; Diniz et al. 2004). In other studies, all nutrients are provided continuously, while ensuring that a key nutrient (most often N) is the growth-limiting factor. This is achieved either by limiting nitrogen feeding (Dufresne and Samain 1998; Kellerhals et al. 1999a) or by adjusting the C to N ratio in the feed (Kim 2002). Therefore, although not directly pointed out in these reports, it can be assumed that PHA synthesis occurred in conjunction with growth. One explanation for the increased PHA content that often results from nutrient depletion or limitation is that the growth of biomass other than PHA (i.e. residual biomass) is limited while PHA synthesis continues unabated (Ramsay et al. 1991). This is a very different phenomenon than that which occurs in many SCL-PHA accumulating bacteria, such as Ralstonia eutropha, where it has been clearly demonstrated that the rate of PHA synthesis is greatly increased when nutrients such as N are in short supply (Oeding and Schlegel 1973). Simultaneous rapid growth and PHA accumulation has been recently demonstrated in our
31

studies (Sun et al. 2007/Chapter 5). The continuously increasing qPHA(Xr) (g PHA g Xr-1 h-1) in our results indicated that PHA was synthesized at a rate higher than the residual biomass synthesis rate, suggesting that such PHA synthesis is not strictly associated with cell growth. The biochemical and genetic mechanisms controlling such kinetics needs to be understood if simple yet highly efficient MCL-PHA production processes similar to those available for SCL-PHA are to be developed.

2.6 Economic considerations

Little has been published concerning the cost of production of MCL-PHA. However, the technology is very similar to SCL-PHA production whose economics have been well studied (Choi and Lee 1999). Before transfer of their technology to Monsanto, Zeneca sold Biopol (their family of P(HB-HV) copolymers) for about 10 US $ kg-1. It has been estimated that large-scale production of MCL-PHA from octane would cost from 5 (de Koning et al. 1997) to 10 (Hazenberg and Witholt 1997) US $ kg-1 with costs almost evenly split between the carbon source, the fermentation process and the separation process. MCL-PHA production costs considerations differ from those for SCL-PHA in several ways. The substrates used for MCL-PHA production are generally different and more expensive. The fermentation processes developed to date have much lower PHA productivity comparing to industrial SCL-PHA processes and would lead to higher capital and labour costs. The solubility of the polymers in common solvents such as acetone and
32

the granule density are also different from SCL-PHAs, which affect separation costs. For example, PHA is recovered from the biomass either by solvent extraction or by dissolution of the residual biomass. If the latter approach is used, a latex suspension of PHA granules results. SCL-PHA suspensions are much easier to centrifuge since their density is significantly greater than that of MCL-PHA granules. As was previously discussed, fats, oils or carboxylic acids are usually used as substrates since -oxidation generally results in much higher yields than -addition pathways. Use of fats or oils requires lipase production by the cells or addition of lipase to the fermentation medium. This renders process control difficult for two reasons. Unless there is an excess of lipase, it is difficult to predict how much carboxylic acid is released into the growth medium, and these acids are toxic at relatively low concentrations. In addition, the release of glycerol provides an additional substrate which further complicates the process. Alkanes are less toxic than carboxylic acids but require the appropriate oxidase lacking in MCL-PHA producers unless they possess the OCT plasmid. They are also flammable and their mass transfer into the aqueous phase may limit the growth rate. Thus carboxylic acids, especially those in the C6 to C10 range, although inhibitory to growth above a certain concentration, are preferred substrates for MCL-PHA production. Unfortunately, they are expensive and often not readily available on a large scale. Commercial octanoic acid is usually derived from coconut oil while nonanoic acid can be easily synthesized from oils such as canola, which are rich in linoleic acid. These feedstocks are renewable, which is
33

an important environmental consideration. They are also becoming increasingly in demand for other green products, especially biodiesel. Although this review deals with fermentation processes, separation accounts for up to 30% of the overall production cost. The fermentation process also greatly influences PHA separation cost since this cost is inversely related to the PHA content in the biomass. As with SCL-PHA, MCL-PHA may be separated from cells either by solvent extraction, dissolution of biomass other than PHA, or a combination of both of these approaches. Solvent extraction of MCL-PHA would be less expensive than for SCL-PHA since it is soluble in a wider range of solvents including ones such as acetone, which can be produced via fermentation of renewable materials. In fact, it is often a co-product of butanol which can be used as a replacement for gasoline in automobiles. Dissolution of non-PHA biomass could be more challenging than in SCL-PHA processes. The released PHA granules are typically separated by centrifugation but capital and energy costs increase greatly if the granule density approaches that of the aqueous phase. The density of SCL-PHA granules are nearly 1.2 times the density of water, while those of MCL-PHA are very similar to that of water and may even float, depending on the subunit composition. A variety of methodologies such as froth flotation (van Hee et al. 2006) have been investigated to avoid centrifugation. A cost analysis on a non-PHA biomass dissolution approach employing filtration concluded that the separation cost could be as low as 1 US $ kg-1 if the PHA content in the biomass from the fermentor exceeded 80%
34

(de Koning et al. 1997). In summary, production of MCL-PHA is likely to be somewhat more expensive than SCL-PHA production due to additional substrate and separation costs. However, it is thought that MCL-PHAs may be more suitable for high value applications such as in the biomedical industry (Pouton and Akhtar 1996; Williams et al. 1999; Chen and Wu 2005).

2.7 Future commercial processes

Many of the studies presented in this review were process studies as opposed to process optimizations. The amount of PHA accumulated as a % of dry biomass is much less than in optimized SCL-PHA processes but there is no reason why similar productivity cannot be achieved in MCL-PHA processes. Advances in metabolic and biochemical engineering are likely to produce significant improvements in the near future. Although MCL-PHA production would be much simpler and cheaper using highly productive processes based on glucose fermentation or recombinant plants, use of related carbon sources such as octanoate allows tailoring of the final product to fit commercial demands. Thus, it is likely that such demand will be initially fulfilled by fermentation processes based on mixtures of glucose and carbon sources such as octanoate, especially those in which most or all of the glucose is used to supply energy and produce residual biomass.

35

2.8 References
Ashby RD, Solaiman DKY, Foglia TA (2004) Bacterial poly(hydroxyalkanoate) polymer production from the biodiesel co-product stream. J Polym Environ 12:105-112 Chen GQ, Wu Q (2005) The application of polyhydroxyalkanoates as tissue engineering materials. Biomaterials 26:6565-6578 Choi J, Lee SY (1999) Factors affecting the economics of polyhydroxyalkanoate production by bacterial fermentation. Appl Microbiol Biotechnol 51: 13-21 de Koning GJM, Kellerhals M, van Meurs C, Witholt B (1997) A process for the recovery of poly(hydroxyalkanoates) from Pseudomonads .2. Process development and economic evaluation. Biopro Eng 17: 15-21 de Smet MJ, Eggink G, Witholt B, Kingma J, Wynberg H (1983) Characterization of intracellular inclusions formed by Pseudomonas oleovorans during growth on octane. J Bacteriol 154:870-878 Diard S, Carlier JP, Ageron E, Grimont PAD, Langlois V, Guerin P, Bouvet OMM (2002) Accumulation of poly(3-hydroxybutyrate) from octanoate, in different Pseudomonas belonging to the rRNA homology group I. Syst Appl Microbiol 25:183-188 Diniz SC, Taciro MK, Gomez JG, da Cruz Pradella JG (2004) High-cell-density cultivation of Pseudomonas putida IPT 046 and medium-chain-length polyhydroxyalkanoate production from sugarcane carbohydrates. Appl Biochem Biotechnol 119:51-70 Dufresne A, Samain E (1998) Preparation and characterization of a poly(beta-hydroxyoctanoate) latex produced by Pseudomonas oleovorans. Macromolecules 31:6426-6433 Durner R, Witholt B, Egli T (2000) Accumulation of poly[(R)-3-hydroxyalkanoates] in Pseudomonas oleovorans during growth with octanoate in continuous culture at different dilution rates. Appl Environ Microbiol 66:3408-3414 Durner R, Zinn M, Witholt B, Egli T (2001) Accumulation of poly[(R)-3-hydroxyalkanoates] in Pseudomonas oleovorans during growth in batch and chemostat culture with different carbon sources. Biotechnol Bioeng 72:278-288
36

Eggink G, Dewaard P, Huijberts GNM (1992) The role of fatty-acid biosynthesis and degradation in the supply of substrates for poly(3-hydroxyalkanoate) formation in Pseudomonas putida. FEMS Microbiol Rev 103:159-163 Fritzsche K, Lenz RW, Fuller RC (1990) Production of unsaturated polyesters by Pseudomonas oleovorans. Int J Biol Macromol 12:85-91 Gagnon KD, Lenz RW, Farris RJ, Fuller RC (1992) The mechanical-properties of a thermoplastic elastomer produced by the bacterium Pseudomonas oleovorans. Rubber Chemistry and Technology 65:761-777 Gross RA, Demello C, Lenz RW, Brandl H, Fuller RC (1989) Biosynthesis and characterization of poly (-hydroxyalkanoates) produced by Pseudomonas oleovorans. Macromolecules 22: 1106-1115 Hartmann R, Hany R, Pletscher E, Ritter A, Witholt B, Zinn M (2006) Tailor-made olefinic medium-chain-length poly[(R)-3-hydroxyalkanoates] by Pseudomonas putida GPo1: Batch versus chemostat production. Biotechnol Bioeng 93:737-746 Hazenberg W, Witholt B (1997) Efficient production of medium-chain-length poly(3-hydroxyalkanoates) from octane by Pseudomonas oleovorans: economic considerations. Appl Microbiol Biotechnol 48:588-596 Hazer B, Lenz RW, Fuller RC (1996) Bacterial production of poly-3-hydroxyalkanoates containing arylalkyl substituent groups. Polymer 37:5951-5957 Hazer B, Lenz RW, Fuller RC (1994) Biosynthesis of methyl-branched poly(beta-hydroxyalkanoate)s by Pseudomonas oleovorans. Macromolecules 27:45-49 Hofer H, Mandl T, Steiner W (2002) Acetopyruvate hydrolase production by Pseudomonas putida O1 - optimization of batch and fed-batch fermentations. Appl Microbiol Biotechnol 60:293-299 Huijberts GNM, Eggink G (1996) Production of poly(3-hydroxyalkanoates) by Pseudomonas putida KT2442 in continuous cultures. Appl Microbiol Biotechnol 46:233-239 Huijberts GNM, Eggink G, Dewaard P, Huisman GW, Witholt B (1992) Pseudomonas putida KT2442 cultivated on glucose accumulates poly(3-Hydroxyalkanoates) consisting of saturated and unsaturated monomers. Appl Environ Microbiol 58:536-544
37

Huisman GW, de Leeuw O, Eggink G, Witholt B (1989) Synthesis of poly-3-hydroxyalkanoates is a common feature of fluorescent pseudomonads. Appl Environ Microbiol 55:1949-1954 Huisman GW, Wonink E, de Koning G, Preusting H, Witholt B (1992) Synthesis of poly(3-hydroxyalkanoates) by mutant and recombinant Pseudomonas strains. Appl Microbiol Biotechnol 38:1-5 Jung K, Hany R, Rentsch D, Storni T, Egli T, Witholt B (2000) Characterization of new bacterial copolyesters containing 3-hydroxyoxoalkanoates and acetoxy-3-hydroxyalkanoates. Macromolecules 33:8571-8575 Jung K, Hazenberg W, Prieto M, Witholt B (2001) Two-stage continuous process development for the production of medium-chain-length poly(3-hydroxyalkanoates). Biotechnol Bioeng 72:19-24 Kang HO, Chung CW, Kim HW, Kim YB, Rhee YH (2001) Cometabolic biosynthesis of copolyesters consisting of 3-hydroxyvalerate and medium-chain-length 3-hydroxyalkanaotes by Pseudomonas sp. DSY-82. Antonie Van Leeuvenhoek 80:185-191 Kellerhals MB, Hazenberg W, Witholt B (1999a) High cell density fermentations of Pseudomonas oleovorans for the production of mcl-PHAs in two-liquid phase media. Enzyme Microb Technol 24:111-116 Kellerhals MB, Kessler B, Witholt B (1999b) Closed-loop control of bacterial high-cell-density fed-batch cultures: Production of mcl-PHAs by Pseudomonas putida KT2442 under single-substrate and cofeeding conditions. Biotechnol Bioeng 65:306-315 Kellerhals MB, Kessler B, Witholt B, Tchouboukov A, Brandl H (2000) Renewable long-chain fatty acids for production of biodegradable medium-chain-length polyhydroxyalkanoates (mcl-PHAs) at laboratory and pilot plant scales. Macromolecules 33:4690-4698 Kessler B, Weusthuis R, Witholt B, Eggink G (2001) Production of microbial polyesters: fermentation and downstream processes. Adv Biochem Eng Biotechnol 71: 159-182 Kim BS (2002) Production of medium chain length polyhydroxyalkanoates by fed-batch culture of Pseudomonas oleovorans. Biotechnol Lett 24:125-130

38

Kim DY, Kim YB, Rhee YH (2000) Evaluation of various carbon substrates for the biosynthesis of polyhydroxyalkanoates bearing functional groups by Pseudomonas putida. Int J Biol Macromol 28:23-29 Kim GJ, Lee IY, Choi DK, Yoon SC, Park YH (1996) High cell density cultivation of Pseudomonas putida BM01 using glucose. Journal of Microbiology and Biotechnology 6:221-224 Kim GJ, Lee IY, Yoon SC, Shin YC, Park YH (1997) Enhanced yield and a high production of medium-chain-length poly(3-hydroxyalkanoates) in a two-step fed-batch cultivation of Pseudomonas putida by combined use of glucose and octanoate. Enzyme Microb Technol 20:500-505 Kim O, Gross RA, Hammar WJ, Newmark RA (1996) Microbial synthesis of poly(beta-hydroxyalkanoates) containing fluorinated side-chain substituents. Macromolecules 29:4572-4581 Kim YB, Lenz RW, Fuller RC (1995) Poly-3-hydroxyalkanoates containing unsaturated repeating units produced by Pseudomonas oleovorans. Journal of Polymer Science Part A-Polymer Chemistry 33:1367-1374 Kraak MN, Smits THM, Kessler B, Witholt B (1997) Polymerase C1 levels and poly(R-3-hydroxyalkanoate) synthesis in wild-type and recombinant Pseudomonas strains. J Bacteriol 179:4985-4991 Lageveen RG, Huisman GW, Preusting H, Ketelaar P, Eggink G, Witholt B (1988) Formation of polyesters by Pseudomonas oleovorans - effect of substrates on formation and composition of poly-(R)-3-hydroxyalkanoates and poly-(R)-3-hydroxyalkenoates. Appl Environ Microbiol 54:2924-2932 Langenbach S, Rehm BH, Steinbchel A (1997) Functional expression of the PHA synthase gene phaC1 from Pseudomonas aeruginosa in Escherichia coli results in poly(3-hydroxyalkanoate) synthesis. FEMS Microbiol Lett 150:303-309 Lee J, Lee SY, Park S, Middelberg APJ (1999) Control of fed-batch fermentations. Biotechnol Adv 17: 29-48 Lee SY, Wong HH, Choi JI, Lee SH, Lee SC, Han CS (2000) Production of medium-chain-length polyhydroxyalkanoates by high-cell-density cultivation of Pseudomonas putida under phosphorus limitation. Biotechnol Bioeng 68:466-470

39

Madison LL, Huisman GW (1999) Metabolic engineering of poly(3-hydroxyalkanoates): From DNA to plastic. Microbiology and Molecular Biology Reviews 63:21-53 Oeding V and Schlegel HG (1973) -ketothiolase from Hydrogenomons eutropha H16 and its significance in the regulation of poly--hydroxybutyrate metabolism. Biochem J 134: 239-248 Pouton CW, Akhtar S (1996) Biosynthetic polyhydroxyalkanoates and their potential in drug delivery. Adv Drug Deliv Rev 18:133-162 Preusting H, Hazenberg W, Witholt B (1993a) Continuous production of poly(3-hydroxyalkanoates) by Pseudomonas oleovorans in a high-cell-density, 2-liquid-phase chemostat. Enzyme Microb Technol 15:311-316 Preusting H, Kingma J, Witholt B (1991) Physiology and polyester formation of Pseudomonas oleovorans in continuous 2-liquid-phase cultures. Enzyme Microb Technol 13:770-780 Preusting H, Vanhouten R, Hoefs A, Vanlangenberghe EK, Favrebulle O, Witholt B (1993b) High-cell-density cultivation of Pseudomonas oleovorans - growth and production of poly (3-hydroxyalkanoates) in 2-liquid phase batch and fed-batch systems. Biotechnol Bioeng 41:550-556 Prieto MA, Kellerhals MB, Bozzato GB, Radnovic D, Witholt B, Kessler B (1999) Engineering of stable recombinant bacteria for production of chiral medium-chain-length poly-3-hydroxyalkanoates. Appl Environ Microbiol 65:3265-3271 Qi Q, Steinbchel A, Rehm BH (1998) Metabolic routing towards polyhydroxyalkanoic acid synthesis in recombinant Escherichia coli (fadR): inhibition of fatty acid beta-oxidation by acrylic acid. FEMS Microbiol Lett 167:89-94 Ramsay BA, Saracovan I, Ramsay JA, Marchessault RH (1992) Effect of nitrogen limitation on long-side-chain poly-beta-hydroxyalkanoate synthesis by Pseudomonas resinovorans. Appl Environ Microbiol 58:744-746 Ramsay BA, Saracovan I, Ramsay JA, Marchessault RH (1991) Continuous production of long-side-chain poly-beta-hydroxyalkanoates by Pseudomonas oleovorans. Appl Environ Microbiol 57:625-629

40

Ren Q, Sierro N, Witholt B, Kessler B (2000) FabG, an NADPH-dependent 3-ketoacyl reductase of Pseudomonas aeruginosa, provides precursors for medium-chain-length poly-3-hydroxyalkanoate biosynthesis in Escherichia coli. J Bacteriol 182:2978-2981 Sanchez RJ, Schripsema J, da Silva LF, Taciro MK, Pradella JGC, Gomez JGC (2003) Medium-chain-length polyhydroxyalkanoic acids (PHA(mcl)) produced by Pseudomonas putida IPT 046 from renewable sources. European Polymer Journal 39:1385-1394 Scholz C, Fuller RC, Lenz RW (1994) Growth and Polymer Incorporation of Pseudomonas oleovorans on Alkyl Esters of Heptanoic Acid. Macromolecules 27:2886-2889 Solaiman DKY, Ashby RD, Foglia TA (1999) Medium-chain-length poly(beta-hydroxyalkanoate) synthesis from triacylglycerols by Pseudomonas saccharophila. Curr Microbiol 38:151-154 Solaiman DKY, Ashby RD, Hotchkiss AT, Foglia TA (2006) Biosynthesis of medium-chain-length poly(hydroxyalkanoates) from soy molasses. Biotechnol Lett 28:157-162 Steinbchel A, Ltke-Eversloh T (2003) Metabolic engineering and pathway construction for biotechnological production of relevant polyhydroxyalkanoates in microorganisms. Biochem Eng J 16:81-96 Sun Z, Ramsay J, Guay M, Ramsay B (2007) Carbon-limited fed-batch production of medium-chain-length polyhydroxyalkanoates from nonanoic acid by Pseudomonas putida KT2440. Appl Microbiol Biotechnol 74:69-77 Thakor N, Trivedi U, Patel KC (2005) Biosynthesis of medium chain length poly(3-hydroxyalkanoates) (mcl-PHAs) by Comamonas testosteroni during cultivation on vegetable oils. Bioresour Technol 96:1843-1850 Timm A, Steinbchel A (1990) Formation of polyesters consisting of medium-chain-length 3-hydroxyalkanoic acids from gluconate by Pseudomonas aeruginosa and other fluorescent pseudomonads. Appl Environ Microbiol 56:3360-3367 van Beilen JB, Panke S, Lucchini S, Franchini AG, Rothlisberger M, Witholt B (2001) Analysis of Pseudomonas putida alkane-degradation gene clusters and flanking insertion sequences: evolution and regulation of the alk genes. Microbiology 147: 1621-1630
41

van Hee P, Elumbaring ACMR, van der Lans RGJM, van der Wielen LAM (2006) Selective recovery of polyhydroxyalkanoate inclusion bodies from fermentation broth by dissolved-air flotation. J Colloid Interface Sci 297:595-606 Weusthuis RA, Huijberts GNM, Eggink G (1997) Production of mcl-poly(hydroxyalkanoates) (review). In: Eggink G, Steinbchel A, Poirer Y, Witholt B (eds) 1996 International Symposium on Bacterial Polyhydroxyalkanoates. NRC Research Press, Ottawa Williams SF, Martin DP, Horowitz DM, Peoples OP (1999) PHA applications: addressing the price performance issue: I. Tissue engineering. Int J Biol Macromol 25:111-121 Yamamoto S, Kasai H, Arnold DL, Jackson RW, Vivian A, Harayama S (2000) Phylogeny of the genus Pseudomonas: intrageneric structure reconstructed from the nucleotide sequences of gyrB and rpoD genes. Microbiology 146:2385-2394

42

CHAPTER 3

Automated Feeding Strategies for High-cell-density Fed-batch Cultivation of Pseudomonas putida KT2440
Zhiyong Sun, Juliana A. Ramsay, Martin Guay, Bruce A. Ramsay

Originally published July, 2006 in Applied Microbiology and Biotechnology, 71 (4): 423-431
43

3.1 Abstract
Four automatic substrate feeding strategies were developed and investigated in this study to obtain rapid, repeatable and reliable high cell density cultivation of Pseudomonas putida KT2440 from glucose. Growth yield data of the key nutrients, YX/Glucose, YX/NH4, YX/PO4, YX/Mg, and YCO2/Glucose, were determined to be 0.41, 5.44, 13.70, 236, and 0.65 g g-1, respectively. Although standard exponential feeding strategy worked well when the predetermined was set at 0.25 h-1, an exponential glucose feeding strategy with online max estimation resulted in a higher average biomass productivity (3.4 vs. 2.8 g X l-1 h-1). A CO2 production rate based pulse glucose feeding strategy also resulted in good overall productivity (3.0 g X l-1 h-1) and can be used as an alternative to pH-stat or DO-stat feeding. A cumulative CO2 production based continuous feed with real-time cumulative glucose consumption estimation resulted in much higher biomass productivity (4.3 g X l-1 h-1) and appears be an excellent and reliable approach to fully automating high cell density fed-batch cultivation of P. putida.

3.2 Introduction
Many industrial fermentation processes require rapid growth to a high cell density. Fed-batch processes are considered to be the most efficient way of achieving high cell density cultures (HCDC) (Lee et al. 1999; Riesenberg and Guthke 1999) but depend on

44

suitable substrate feeding strategies to control key nutrient concentrations (typically the carbon source). Ideally feeding should be based on direct measurement of the substrate whose concentration requires control. Unfortunately automatic analysis of key substrates usually involves a substantial time delay, expensive, dedicated analytical equipment and moderately difficult computer interfacing, but there are alternatives. For example, substrate feeding may be based on knowledge of a microorganism's predicted growth rate. In this approach a predetermined exponential feeding rate designed to obtain a growth rate close to max is typically used for the growth phase, while a linear or decaying linear feed rate is often used in a nongrowth-associated production phase (Riesenberg et al. 1991). Feeding may also be based on physiology as in pH-stat and DO-stat techniques (Lee et al. 2000), where feeding depends on acid production or oxygen utilization. Lee et al (1999) have written an extensive review on fed-batch fermentation control technology.

Pseudomonas putida KT2440 and its rifampicin-resistant derivative, strain KT2442, are metabolically versatile and well-characterized (Timmis 2002). The KT2440 genome has been sequenced and the putative gene products have been identified (Nelson et al. 2002). These strains are being exploited in a wide variety of applications, such as catabolic pathway designs for pollutant biodegradation (Rojo et al. 1987), biological desulphurization of petroleum feedstocks and products (Galan et al. 2000), production of recombinant proteins (Thuesen et al. 2003), and synthesis of medium-chain-length
45

poly(3-hydroxyalkanoates) (MCL-PHA) (Huijberts et al. 1992). Some of these applications, such as recombinant protein and MCL-PHA production, are most economically performed in fed-batch mode. While standard methods are available for the fed-batch cultivation of Escherichia coli (Riesenberg 1991; Kim et al. 2004), these are only now being developed for P. putida.

Kim et al. (1996) obtained 100 g l-1 of P. putida by maintaining glucose concentration within the range of 5-20 g l-1 using a glucose analyzer. Cell concentrations greater than 100 g l-1 have been obtained using a combination of pH-stat and DO-stat feeding strategies (Lee et al. 2000), while 38 g l-1 of a recombinant P. putida strain was obtained using an exponential glucose feeding (Thuesen et al. 2003). However, the substrate feeding and control strategies employed in those studies either required very complex online analysis, significant manual manipulation or have not been optimized to achieve maximum biomass productivity.

Although product yields depend on the type of carbon and energy source utilized, glucose, often obtained by starch hydrolysis, is by far the most commonly used source of carbon and energy in the fermentation industry. Glucose does not inhibit growth of P. putida below a concentration of at least 40 g l-1 (Hofer et al. 2002; Kim et al. 1996), provides reasonable yield efficiency, and has excellent aqueous solubility. Glucose is also cheap and widely available. Therefore, in this study, we investigated four automated
46

glucose feeding strategies designed to obtain rapid, repeatable, and reliable high cell densities of P. putida KT2440. Detailed growth yield information was also obtained. It is hoped that this information can be used for the development of specific fermentation processes using P. putida KT2440 or strains and species with similar physiology.

3.3 Materials and methods

3.3.1 Analytical procedures
During fermentations, an Accu-Chek Advantage blood glucose meter (Roche Diagnostics, Canada) was sometimes used to monitor real time glucose concentration according to an established standard curve. The cell dry weight was measured after centrifugation of 5 ml of culture broth at 17000 x g for 15 min, followed by lyophilization. A colorimetric assay of glucose using 4-hydroxybenzoic hydrazide and NaOH as reagents was used to determine the glucose concentration in all samples (Lever 1972). Phosphate concentration was measured via reduction of phosphomolybdate to molybdene blue (Clesceri et al. 1999). Ammonium concentration was determined by the phenol-hypochlorite reaction (Weatherburn MW 1967).

3.3.2 Microorganism and growth Medium

P. putida KT2440 (ATCC 47054) was maintained in lyophiles and on nutrient agar plates at 4 oC. The shake flask mineral salts medium contained the following components per liter: 4.70 g (NH4)2SO4, 0.80 g MgSO47H2O, 12.00 g Na2HPO47H2O, 2.70 g KH2PO4,
47

9.00 g glucose and 1.00 g nutrient broth. The initial fermentation culture medium contained the following components per liter: 4.70 g (NH4)2SO4, 0.80 g MgSO47H2O, 18.00 g Na2HPO47H2O, 4.05 g KH2PO4, 1 to 15.00 g of glucose (according to the feeding strategy) and 8 ml of trace element solution. Each liter of trace element solution contained 10 g FeSO47H2O, 3 g CaCl22H2O, 2.2 g ZnSO47H2O, 0.5 g MnSO44H2O, 0.3 g of H3BO3, 0.2 g CoCl26H2O, 0.15 g Na2MoO42H2O, 0.02 g NiCl26H2O, 1.00 g CuSO45H2O. Nitrogen source was supplied by using 14% (w/v) NH4OH to control the pH. Each liter of feeding medium contained 600 g glucose and 10 g MgSO47H2O.

3.3.3 Fermentation conditions

Seed culture was incubated in 500-ml shake flasks with 100 ml mineral salts medium at 301 oC and 200 rpm for 13 h before inoculation into the bioreactor. Fed-batch fermentations were all carried out at 311 oC with 0.8 l of initial working volume in a Multigen F-2000 2 l stirred-tank bioreactor (New Brunswick Scientific, Edison, N.J.). Data acquisition (dissolved oxygen, CO2 outlet gas concentration, NH4 and feed medium addition, and pH) and control was conducted with an IBM-compatible PC using LabVIEW 6.1 (National Instruments) software. pH was controlled between 6.80 and 6.90 by automatic addition of 14% (w/v) NH4OH with an Ingold sterilizable probe and a Bioengineering AG (Wald, Switzerland) controller. The dissolved oxygen was measured with an Ingold polarographic probe (Westech Industrial Ltd, Calgary, AB, Canada) and
48

maintained at or above 40% saturation (except where indicated), first, by adjusting the agitation speed up to 900 rpm and, thereafter, by controlling air and pure oxygen flow via mass flow controllers at a total gas flow rate of 1.5 l min-1. Exit gas (CO2) content was measured with an infrared monitor (Guardian Plus, Topac Inc. Hingham, MA, USA). Feed medium was continuously weighed on an Ohaus (Pine Brook, NJ, USA) GT 8000 balance. Feeding was conducted with a Cole-Parmer Ltd. (Vernon Hills, Il, USA) peristaltic pump automatically controlled with LabVIEW based on the weight of the feed.

3.3.4 Substrate Feeding Strategies

3.3.4.1 Exponential feeding with a predetermined With 1 g l-1 of initial glucose, substrate feeding was started at the beginning of the fermentation according to feed Eq.1 such that the culture grew at a predetermined specific growth rate.

St =

Xt X = 0 et YX / S YX / S

(1)

where St is the total amount of substrate required to produce a certain amount of biomass, Xt, at cultivation time t; YX/S is the yield of biomass from glucose, obtained previously and assumed to be constant; X0 is the initial biomass, obtained by measuring the optical density650 of the shake-flask inoculum culture before inoculation; and is the desired specific growth rate (0.25 h-1 was used in this study).
49

3.3.4.2 Exponential feeding based on initial maximum specific growth rate (max) estimation Fermentations were started in batch mode with 10 g l-1 of initial glucose. CO2 content in the exit gas flow was automatically measured once per minute, converted to CO2 production rate (CPR; g h-1), then integrated into cumulative CO2 production (CCP; g) with respect to time. Assuming the CO2 production yield from biomass (YCO2/X) is constant during the growth phase, then:

CCPt = YCO2 / X X t = YCO2 / X X 0 e t = a e t

(2)

where a is the product of YCO2/X and X0. After 10 hours of cultivation time, the CCP data were regressed nonlinearly using Sigmaplot 9.0.1 (Systat Software Inc. Point Richmond, CA, USA) to obtain the parameters a and for the above equation. The value obtained was considered to be the max that this culture could reach under these growth conditions. Once the model parameters were obtained, the substrate was fed according a modified form of equation (1).

St =

Xt X a a = 0 e t = e max t = e max t YX / S YX / S YCO2 / X YX / S YCO2 / S

(3)

50

3.3.4.3 Semi-continuous feeding based on the detection of substrate limitation Fermentations began with 15 g l-1 of glucose. When CPR dropped by 10% of its previous maximum (indicating a substrate limitation), sufficient glucose to raise the concentration by 10 g l-1 was fed automatically. 3.3.4.4 Continuous feeding based on cumulative glucose consumption (CGC) estimation Fermentations began with 10 g l-1 of glucose. Glucose was fed continuously as a function of cumulative CO2 production (CCP) and YCO2/G (Eq. 4). The cumulative substrate addition at time t (St) was kept equal to the estimated CGCt, hence, maintaining the estimated glucose concentration in the reactor at 10 g l-1. St = CGCt = CCPt YCO2 / S (4)

3.4 Results
3.4.1 Growth yield study
To obtain the information necessary to compose balanced growth media, data for the consumption of key nutrients from several nutrient-unlimited fed-batch fermentations were collected (Figure 3-1). Growth yields of biomass from glucose (YX/G), ammonia (YX/NH4), phosphate (YX/PO4), and the production of CO2 from glucose (YCO2/G) were determined as 0.41 g g-1 (r2= 0.98), 5.44 g g-1 (r2= 0.99), 13.7 g g-1 (r2= 0.95) (open and closed circle data only), and 0.65 g g-1 (r2= 0.99), respectively, using linear regression. In
51

addition, YX/Mg was determined as 236 g g-1 by imposing periodic magnesium limitation followed by pulse additions of magnesium sulfate to verify that magnesium was the growth-limiting nutrient.

70

100

60 80 50

Biomass (g)

Biomass (g)
0 20 40 60 80 100 120 140 160

40

60

30

40

20 20 10

0 0 2 4 6 8 10 12 14 16

Glucose consumption (g)

NH4 consumption (g)

(a) Biomass vs. glucose consumption

100 100

(b) Biomass vs. NH4 consumption

80

Cumulative CO2 production (g)

0 1 2 3 4 5 6 7

80

Biomass (g)

60

60

40

40

20

20

0 0 20 40 60 80 100 120 140

PO4 consumption (g)

Glucose consumption (g)

(c) Biomass vs. PO4 consumption

(d) CO2 vs. glucose consumption

Figure 3-1 Yield of dry biomass from the principal growth substrates. Different symbols indicate data from different fermentations. The slope of the line indicates the calculated yield value. Only data indicated by closed and open circles were used to calculate yield from PO4 in (c).

52

3.4.2 Exponential feeding with a predetermined

Exponential feeding with a predetermined was examined first, since it is the simplest and most commonly employed method of HCDC (Riesenberg and Guthke 1999). To avoid over-feeding, the value in the feeding Eq. 1 was set at the conservative value of 0.25 h-1. CPR (g h-1), an indicator of culture growth rate, exhibited an unlimited exponential increase during first seven hours (= 0.56 h-1), and then a steep decline (Figure 3-2). After this point, the CPR trajectory followed the glucose feeding curve (Figure 3-2), indicating a substrate-limited growth. By 22 h, the dissolved oxygen could no longer be maintained at 40% of saturation as oxygen demand surpassed the bioreactor's mass transfer capabilities. During the portion of the growth phase when growth was substrate-limited, glucose concentration was below 0.5 g l-1 and the total amount of dry biomass produced was 76.4 g (61.6 g l-1), with an average productivity of 2.8 g X l-1 h-1.
5 80 Biomass Glucose concentration CPR Glucose added 400 350 40 300 250 40 200 150 20 1 50 0 0 0 5 10 15 20 25 0 0 100 10 50

Glucose concentration (g l )

-1

4 60

20

Culture time (h)

Figure 3-2 Fed-batch P. putida cultivation using simple exponential feeding designed to achieve = -1 0.25 h .
53

CPR (g h-1)

Glucose added

Biomass (g)

30

3.4.3 Exponential feeding based on initial max estimation

The previous exponential feeding strategy was based on a conservative choice of for the feeding equation. An online estimation of the actual would allow for more rapid feeding and thus a more productive fermentation. This fermentation again began in batch mode. Data analysis 10 h after inoculation determined parameters a and max of Eq. 2 to be 0.15 g and 0.38 h-1, respectively (Figure 3-3a). These values, along with the YCO2/S for glucose of 0.65 g g-1, were then used in exponential feeding Eq. 3 to begin feeding. Since glucose feeding was based on St in Eq. 3, additional glucose was immediately automatically added in a step-wise fashion to compensate for the estimated use of glucose up to that point. This was followed by exponential addition to maintain the estimated St. At 12 h, the overall a and max values were redetermined as 0.09 g and 0.45 h-1, respectively (Figure 3-3b). The feeding equation was updated accordingly, resulting in more glucose being automatically added in a stepwise fashion to reach the updated St value. Again this was followed by exponential addition to maintain St as it increased with time. Biomass accumulated exponentially up to 64.6 g (54.4 g l-1) at 16 h, at which time the dissolved oxygen concentration could no longer be maintained at or above 40% saturation due to excessive O2 demand (Figure 3-4). This average biomass productivity throughout this fermentation was 3.40 g X l-1 h-1.

54

8
CCP Regression curve

20
CCP Regression curve

15

CCP (g)

CCP=0.15*EXP(0.38*t)

CCP (g)

10

CCP=0.09*EXP(0.45*t)

(a)
0 0 2 4 6 8 10 12 0 0 2 4 6 8 10

(b)
12 14

Culture time (h)

Culture time (h)

Figure 3-3 Online cumulative CO2 production (CCP) data regressions according to Eq. 2 for max estimation. (a) Regression at 10 h cultivation time. (b) Regression at 12 h cultivation time.

Biomass (g) and cumulative CO2 production (g)

200 180

200 180 160 140 120 100 80 60 40 20 0 0 2 4 6 8 10 12 14 16 18 Biomass Glucose added Glucose consumed CCP Glucose concentration

200 180 160 140 120 100 80 60 40 20 0

Glucose concentration (g l )

-1

160 140 120 100 80 60 40 20 0

Culture time (h)

Figure 3-4 Fed-batch P. putida cultivation using exponential feeding with online max estimation.

55

Glucose added (g) and glucose consumed (g)

3.4.4 Semi-continuous feeding based on the detection of substrate limitation (CPR-based pulse feeding)
The preceeding feeding method is dependent on an accurate YCO2/G value. A method allowing the actual level of glucose to be detected should be more reliable. Limitation of growth by falling glucose levels will cause the CPR to decline. Addition of a quantity of glucose insufficient to inhibit growth as soon as glucose limitation is detected in the CPR data should allow the glucose concentration to be maintained near its optimal value throughout the majority of the fermentation. In this approach, 10 g l-1 of glucose was fed automatically each time the CPR declined by 10% of its previous maximum value. Liquid samples were taken during the period of CPR decline. As expected, the glucose concentration during each CPR decline was very close to 0, which indicated that overall, the glucose concentration was always controlled at or below 10 g l-1 (Figure 3-5). Oxygen transfer became problematic at the 10 h point since increasingly large oxygen demands occurred after each addition of glucose. 45.9 g (30.0 g l-1) of biomass had been produced by that time giving an overall biomass productivity of 3.0 g X l-1 h-1.

56

14

50 Biomass Glucose concentration CPR Glucose added

120

40 35 30

12

40 10

80

Biomass (g)

30 60 20 40

25 20 15 10

4 10 2 20

5 0

0 0 2 4 6 8 10 12

Culture time (h)

Figure 3-5 Fed-batch P. putida cultivation using CO2 production rate (CPR) based pulse feeding.

3.4.5 Continuous feeding with real-time glucose consumption estimation based on CCP analysis
The three previous glucose feeding strategies all allow glucose to fall to growth-limiting levels for at least part of the fermentation. Besides being at least moderately sub-optimal in terms of biomass productivity, allowing glucose to limit growth may trigger undesirable physiological responses. If glucose limitation is to be avoided and automatic measurement of glucose concentration is impractical, accurate knowledge of YCO2/S values may allow substrates to be controlled based on CPR data. To assess whether this could be achieved, glucose was fed in a continuous manner according to feeding Eq. 4. Using this approach, the glucose concentration was maintained between 5 and 19 g l-1 (Figure 3-6) which is commonly considered to be neither growth-limiting nor
57

CO2 production rate (CPR, g h )

Glucose concentration (g l-1)

Glucose added (g)

-1

100

growth-inhibiting. By 12.5 h of fermentation the dissolved oxygen concentration had dropped to near 0 % so the fermentation was abandoned. By that time 63.5 g (54.2 g l-1) of biomass had been produced, giving an overall biomass productivity of 4.3 g X l-1 h-1.
Cumulative CO2 production (g) and Glucose added (g)

50

70 Biomass Glucose concentration CCP Glucose added

200 180 160 140 120 100

60

Glucose concentration (g l )

-1

40 50 30

Biomass (g)

40

20

30

80 60 40

20 10 10

20 0 0 0 2 4 6 8 10 12 14 0

Culture time (h)

Figure 3-6 Fed-batch P. putida cultivation using continuous glucose feeding based on glucose consumption as estimated from cumulative CO2 production data.

3.5 Discussion
3.5.1 Yield values
The yield values obtained are similar to literature data for other P. putida strains. P. putida IPT 046 was found to have a dry biomass yield from a mixture of glucose and fructose of 0.40 g g-1 (Diniz et al. 2004), while we found a dry biomass yield from glucose of 0.41 g g-1 for KT 2440. The YX/NH4 for P. putida KT 2440 (5.44 g g-1) was also
58

very similar to that of IPT 046 (5.50 g g-1). These differences appear to be within the range of experimental error and are thus essentially identical.

An analysis of data for P. putida KT2442 grown on oleic acid showed a YX/PO4 of less than or equal to 16 g g-1 (Lee et al. 2000) compared to 13.7 g g-1 for KT2440 on glucose. As stated previously, some phosphate data was excluded from our calculations. It is believed that significant precipitation of phosphate occurred in fermentations from which the excluded data was collected. The formation of insoluble phosphate complexes with other minerals should be considered during medium formulation and preparation, especially since phosphate limitation is believed to lead to the accumulation of poly(3-hydroxyalknoates) in P. putida (Lee et al. 2000).

Further analysis of data for P. putida KT2442 grown on oleic acid indicated a YX/Mg of less than or equal to 265 g g-1 (Lee et al. 2000), compared to 236 g g-1 for KT2440 on glucose. These values are lower than those typically found in bacterial cultures (Pirt, 1975). MgSO4 is generally added in small batches during high density fermentations or fed into the reactor with glucose as it too has a tendency to precipitate, possibly in complexes with phosphate.

CO2 production was found to be directly associated with glucose consumption in this strict aerobe (Figure 3-1d). Since the exit gas flowrate and the CO2 content can be easily

59

and accurately measured online, the CPR and consequently, the cumulative CCP, was used in various manners during the development of automatic substrate feeding strategies.

3.5.2 Exponential feeding strategies

Exponential feeding is a simple yet efficient automatic substrate feeding strategy that has been applied extensively in fed-batch high cell density cultivation of E. coli and other microorganisms to achieve desired specific growth rate in order to prevent accumulation of inhibitory products or substrates (Riesenberg et al. 1991; Tada et al. 2000; Yoon et al. 1994). It has been applied to HCDC of P. putida for intercellular protein (Thuesen et al. 2003) and MCL-PHA (Diniz et al. 2004) production. In our study, we began to evaluate this feeding strategy using a of 0.25 h-1 in Eq. 1. There was little risk of glucose overfeeding as the max of this species is about 0.67 h-1 when grown on glucose in a mineral salts medium (Diniz et al. 2004; Givskov et al. 1994). The glucose concentration limited growth rate to the desired value throughout the fermentation. Nonlinear regression of the CCP data with respect to time during the bulk of the growth phase (7.5 h to 21 h) according to a two parameter exponential growth model, y= aexp(bt) showed the damping parameter b to be 0.248 h-1, which was essentially the target value of (0.25 h-1). This demonstrated that CCP data can be used to fit the growth model. In addition, since CCP is the integral of CPR with respect to time, it is naturally filtered of noise.

60

Therefore CCP data are a good choice for off-line determination when biomass data are poor or, more importantly and practically, for online estimation.

Simple exponential feeding requires only basic equipment, interfacing and programming, and is commonly practiced. However, conservative feeding rates are usually chosen since may vary from batch to batch or during a particular fermentation due to the variations in the inoculum and physiological conditions. This results in sub-optimal biomass productivity. Online adjustment of feeding to the actual max would be preferable. As had been shown in our exponential feeding study (Figure 3-2) and also observed in other exponential feeding studies (Riesenberg et al. 1991), the culture usually exhibits a period of rapid growth early in the fermentation since all substrates are in excess. This may be considered as the max for the conditions in that portion of that particular fermentation. Sub-optimal feeding results in a rapid drop in CPR as excess carbon source is exhausted after the period of rapid growth. Using the CCP data collected online, we can determine the max before carbon source limitation, allowing us to set an appropriate glucose feeding rate and thus avoid limitation. This process was used to produce the data in Figure 3-3. As expected, "max" did vary, but the updated feeding model resulted in a very good match between the glucose addition and the actual glucose consumption during cultivation time 13 to 15 h (Figure 3-4). After that, subsequent glucose analysis showed a deviation of the actual glucose consumption from the glucose
61

feeding trajectory, which indicated that further actual max updating was required. However, using exponential feeding with online max estimation two estimations), the biomass productivity (3.40 g X l-1 h-1) and specific growth rate were enhanced and were comparable or higher than calculated from or reported in the literature data for other P. putida strains (Diniz et al. 2004; Thuesen et al. 2003). In order to accurately calculate the variation of and adjust the feeding rate accordingly, we are trying to develop a recursive least square algorithm to estimate in a real time. More frequent updating of estimation using this algorithm and consequently more frequent adjustment in the feeding rate should allow for even greater productivity.

3.5.3 CPR-based pulse feeding

DO-stat and pH-stat feeding methodologies have been developed to prevent the accumulation of substrate to inhibitory levels. These are based on the fact that DO increases rapidly when a key substrate is depleted (Cutayar and Poillon 1989) and that pH may also increase due to the excretion of ammonium ions (Suzuki et al. 1990). Both of these feeding strategies share the similar indirect feedback control scheme of keeping the substrate concentration between 0 and a noninhibitory level by pulse feeding, and are often used in fed-batch cultivation, including fermentations of Pseudomonas (Kim 2002; Lee et al. 2000). CPR, which can be obtained in real time almost as easily and rapidly as DO or pH, can also be used as the indication of substrate limitation. Although it was
62

previously claimed that such pulse feeding requires a full-time operator (Diniz et al. 2004), a control algorithm was written to add glucose automatically at every significant drop in CPR. Using this methodology, we were able to control the glucose concentration in the medium between 0 and 10 g l-1, leading to a biomass productivity of 3.0 g X l-1 h-1. This CPR-based feeding strategy can be an alternative method to DO-stat and pH-stat feeding strategies during the fed-batch cultivation of microorganisms, such as P. putida, especially when DO-stat or pH-stat feeding is not feasible, such as when pH is not significantly affected by substrate depletion (Kim et al. 1996).

3.5.4 Continuous feeding based on CCP

A simple continuous feeding strategy was developed based on real-time glucose consumption estimation. Without a complicated control algorithm and programming requirements, only real-time CCP data and an accurate YCO2/S are needed to control the substrate feeding according to Eq. 4. YCO2/S was already found to be highly stable during both glucose limitation and excess (Figure 3-1d). The regression of the CCP data with respect to time indicated that the max of the culture during the growth phase was 0.48 h-1, which was slightly higher than the max value obtained using the exponential feeding with updated strategy, and resulted in a high overall biomass productivity at 4.3 g X l-1 h-1. As can be seen from the increase of glucose concentration from 10 to 12 h of this fermentation, glucose consumption was overestimated, indicating either error in a
63

measurement or that the YCO2/S during that growth period must have varied to below the assumed value (0.65 g g-1). However, glucose was always maintained within a concentration range of 5-20 g l-1 in four fermentations using the same YCO2/S setting (data not shown). Thus the reliability and performance using this simple feeding strategy is satisfactory. Perfectly accurate CCP determinations could result in even better control.

3.5.5 Comparison of methodologies

In this study, we evaluated traditional exponential feeding strategy, and developed and investigated three other substrate feeding strategies for high cell density fed-batch cultivation of P. putida KT2440. In each case productivity was limited by the glucose feeding methodology and, at the higher oxygen demands found late in the fermentations, by the oxygen transfer capacity of the bioreactor. Although continuous feeding based on real-time CGC estimation produced the best results in terms of overall biomass productivity (Table 3-1), each of these approaches might be suitable for a particular set of conditions. Choice of feeding strategy would depend on availability of equipment and expertise as well as the microorganism. For example, some strategies require more sophisticated programming expertise (Table 3-1, automation programming). CPR based pulse feeding demands a tracking and comparison algorithm whose design is fairly challenging. Exponential feeding with max estimation requires automation of regression analysis which may be even more difficult to achieve. Continuous feeding with real-time
64

glucose consumption estimation is probably the most reliable methodology in terms of achieving a high biomass density since it can compensate for disturbances in the growth conditions (temperature, pH etc) that may affect (Table 3-1, reliability). However, this adaptability may make it less repeatable than simple exponential feeding at a conservative (Table 3-1, repeatability). In addition, most P. putida strains carry plasmids and copy number may be of importance depending on the application. Unless selective medium is used, none of the aforementioned growth strategies may be optimal, since plasmid replication is often independent of . The feeding strategies relying on YCO2/S would obviously not be suitable for organisms whose YCO2/S varies depending on physiological conditions (such as the facultative anaerobe E. coli). Nevertheless, all of these strategies offer the possibility of complete automation, allowing productive high density fed-batch cultivation of bacteria with physiologies similar to that of P. putida KT 2440.
Table 3-1 Comparison of the four feeding strategies and recommendations
Evaluation criteria Feeding strategies investigated Exponential with predetermined Exponential with max estimation CPR based pulse feed Continuous feed based on real-time CGC estimation Overall biomass productivity (g X l-1 h-1) Fair Good Automation programming Easy Difficult Reliability Good Fair Repeatability Good Fair Recommendations & further improvements Good for strictly controlled application More frequent online estimation required for reliable results Good alternative to pH-stat or DO-stat feeding strategies Good for maximizing biomass productivity and production

Fair

Fair

Fair

Fair

Best

Easy

Best

Fair

65

Acknowledgement This project was supported by the Natural Science and Engineering Research Council of Canada (NSERC).

3.6 References
Clesceri LS, Greenberg AE, Eaton AD (1999) Standard methods for the examination of water and wastewater 20th edition. American Puplic Health Association. Washington. DC. Cutayar JM, Poillon D (1989) High cell density culture of Escherichia coli in a fed batch system with dissolved-oxygen as substrate feed indicator. Biotechnol Lett 11:155-160 Diniz SC, Taciro MK, Gomez JG, da Cruz Pradella JG (2004) High-cell-density cultivation of Pseudomonas putida IPT 046 and medium-chain-length polyhydroxyalkanoate production from sugarcane carbohydrates. Appl Biochem Biotechnol 119:51-70 Galan B, Diaz E, Garcia JL (2000) Enhancing desulphurization by engineering a flavin reductase-encoding gene cassette in recombinant biocatalysts. Environ Microbiol 2:687-694 Givskov M, Eberl L, Moller S, Poulsen LK, Molin S (1994) Responses to nutrient starvation in Pseudomonas putida KT2442 - analysis of general cross-protection, cell-shape, and macromolecular content. J Bacteriol 176:7-14 Hofer H, Mandl T, Steiner W (2002) Acetopyruvate hydrolase production by Pseudomonas putida O1 - optimization of batch and fed-batch fermentations. Appl Microbiol Biotechnol 60:293-299 Huijberts GNM, Eggink G, Dewaard P, Huisman GW, Witholt B (1992) Pseudomonas putida KT2442 cultivated on glucose accumulates poly(3-Hydroxyalkanoates) consisting of saturated and unsaturated monomers. Appl Environ Microbiol 58:536-544 Kim BS (2002) Production of medium chain length polyhydroxyalkanoates by fed-batch culture of Pseudomonas oleovorans. Biotechnol Lett 24:125-130

66

Kim BS, Lee SC, Lee SY, Chang YK, Chang HN (2004) High cell density fed-batch cultivation of Escherichia coli using exponential feeding combined with pH-stat. Biopro Biosys Eng 26:147-150 Kim GJ, Lee IY, Choi DK, Yoon SC, Park YH (1996) High cell density cultivation of Pseudomonas putida BM01 using glucose. J Microbiol Biotechnol 6:221-224 Lee J, Lee SY, Park S, Middelberg APJ (1999) Control of fed-batch fermentations. Biotechnol Adv 17:29-48 Lee SY, Wong HH, Choi JI, Lee SH, Lee SC, Han CS (2000) Production of medium-chain-length polyhydroxyalkanoates by high-cell-density cultivation of Pseudomonas putida under phosphorus limitation. Biotechnol Bioeng 68:466-470 Lever M (1972) New reaction for colorimetric determination of carbohydrates. Anal Biochem 47:273-279 Nelson KE, Weinel C, Paulsen IT, Dodson RJ, Hilbert H, Martins dos Santos VA, Fouts DE, Gill SR, Pop M, Holmes M, Brinkac L, Beanan M, DeBoy RT, Daugherty S, Kolonay J, Madupu R, Nelson W, White O, Peterson J, Khouri H, Hance I, Chris Lee P, Holtzapple E, Scanlan D, Tran K, Moazzez A, Utterback T, Rizzo M, Lee K, Kosack D, Moestl D, Wedler H, Lauber J, Stjepandic D, Hoheisel J, Straetz M, Heim S, Kiewitz C, Eisen JA, Timmis KN, Dusterhoft A, Tummler B, Fraser CM (2002) Complete genome sequence and comparative analysis of the metabolically versatile Pseudomonas putida KT2440. Environ Microbiol 4:799-808 Pirt SJ (1975) Principles of microbe and cell cultivation. Blackwell Scientific Publications, Oxford UK Riesenberg D (1991) High-cell-density cultivation of Escherichia coli. Curr Opin Biotechnol 2:380-384 Riesenberg D, Guthke R (1999) High-cell-density cultivation of microorganisms. Appl Microbiol Biotechnol 51:422-430 Riesenberg D, Schulz V, Knorre WA, Pohl HD, Korz D, Sanders EA, Ross A, Deckwer WD (1991) High cell density cultivation of Escherichia coli at controlled specific growth rate. J Biotechnol 20:17-28 Rojo F, Pieper DH, Engesser KH, Knackmuss HJ, Timmis KN (1987) Assemblage of ortho cleavage route for simultaneous degradation of chloro- and methylaromatics. Science 238:1395-1398
67

Suzuki T, Yamane T, Shimizu S (1990) Phenomenological background and some preliminary trials of automated substrate supply in pH-stat modal fed-batch culture using a setpoint of high limit. J Ferment Bioeng 69:292-297 Tada K, Kishimoto M, Omasa T, Katakura Y, Suga KI (2000) L-lysine production by exponential feeding of L-threonine. J Biosci Bioeng 90:669-674 Thuesen MH, Norgaard A, Hansen AM, Caspersen MB, Christensen HE (2003) Expression of recombinant Pseudomonas stutzeri di-heme cytochrome c4 by high-cell-density fed-batch cultivation of Pseudomonas putida. Protein Expr Purif 27:175-181 Timmis KN (2002) Pseudomonas putida: a cosmopolitan opportunist par excellence. Environ Microbiol 4:779-781 Weatherburn MW (1967) Phenol-hypochlorite reaction for determination of ammonia. Anal Chem 39:971-974 Yoon SK, Kang WK, Park TH (1994) Fed-batch operation of recombinant Escherichia coli containing Trp promoter with controlled specific growth rate. Biotechnol Bioeng 43:995-999

68

CHAPTER 4

Increasing the Yield of MCL-PHA from Nonanoic Acid by Co-feeding Glucose During the PHA Accumulation Stage in Two-stage Fed-batch Fermentations of Pseudomonas-putida KT2440
Zhiyong Sun, Juliana A. Ramday, Martin Guay, Bruce A. Ramsay

Originally published online March 12th, 2007 in Journal of Biotechnology

69

4.1 Abstract
A method was developed to increase the yield of MCL-PHA from nonanoic acid in the PHA accumulation phase. Pseudomonas putida KT2440 was grown on glucose until ammonium-limitation was imposed. In the second (accumulation) stage, either glucose, nonanoic acid, or a mixture of these carbon and energy sources was supplied. Since the medium-chain-length poly(3-hydroxyalkanoate) (MCL-PHA) subunits produced are unique from each carbon source, their relative contribution to PHA yield could be calculated. Y(C7+C9)/NA was 0.254 mol mol-1 during PHA synthesis from nonanoic acid. Y(C8+C10)/G was only 0.057 mol mol-1 during PHA synthesis from glucose. When nonanoic acid and glucose were fed together, Y(C7+C9)/NA almost doubled to 0.450 mol mol-1 while Y(C8+C10)/G decreased to 0.011 mol mol-1. These results demonstrate that substantial savings can be obtained by feeding glucose with substrates that are good for PHA production but much more expensive than glucose.

4.2 Introduction
Pseudomonas putida can synthesize medium-chain-length PHAs (MCL-PHAs) from a variety of carbon sources (Huisman et al. 1989; Huijberts et al. 1992). Its metabolism has been well studied and depends on whether aliphatics or carbohydrates are supplied (Eggink et al. 1992). Consequently, MCL-PHA production processes have been

70

developed using either aliphatics (Kim et al. 1997; Kim 2002) or carbohydrates (Diniz et al. 2004) by employing two-stage, fed-batch fermentations with a nutrient-limited second stage for MCL-PHA accumulation. However, how a combination of these two types of carbon sources affect the efficiency of MCL-PHA synthesis is not known. Since aliphatics are generally much more expensive than common industrial carbohydrates, and since substrate costs make up a large proportion of the total production cost, the aim of this study was to better understand the relative contributions of these two types of carbon sources to MCL-PHA synthesis. This knowledge may aid in the development of more economical MCL-PHA fermentation processes.

Pseudomonas putida KT2440 was used because it is one of the most studied putida strains (Timmis 2002). To separate PHA synthesis from growth and because its a commonly employed approach to producing useful amounts of PHA, this study employed two-stage fed-batch fermentations and focused only on PHA synthesis during the nutrient-limited, non-growth (second) stage. The key to understand the contribution of each type of carbon source is to identify the corresponding PHA subunit yield from that carbon source. Nonanoic acid (NA) and glucose (G) were chosen as the representatives of aliphatics and carbohydrates, respectively, because it has been clearly shown that only odd carbon number monomers can be formed from nonanoic acid (Du and Yu 2002; Sun

71

et al. 2007/Chapter 5) and only even-numbered carbon subunits from glucose (Huijberts et al. 1992; Sanchez et al. 2003).

4.3 Materials and methods

Culture (P. putida KT2440, also known as ATCC 47054) conservation, media composition for inoculum and first stage (growth stage) of the fermentation, nutrients (glucose, nitrogen, and phosphorus) analyses as well as fermentation conditions were previously described (Sun et al. 2006/Chapter 3). PHA content (Sun et al. 2007/Chapter 5) and nonanoic acid concentration (Ramsay et al. 1991) were determined by gas chromatography.

During the growth stage of all three two-stage fed-batch fermentations glucose was fed based on cumulative glucose consumption estimated from CO2 data (Sun et al. 2006/Chapter 3). NH4OH was replaced by 2M KOH at an estimated cell concentration of 40 g l-1 such that NH4+ would be depleted at a cell concentration about 45-50 g l-1. Three fermentations were performed identically for the first stage, but carbon source(s) feeding strategy of the second stage (PHA accumulation stage) differed as following.

G-only PHA synthesis: only glucose was fed as the PHA synthesis substrate according to Eq. 1.

72

St PHAG = CGCt PHA =

CCPt PHA YCO2 / G PHA

(1)

where St -PHA-G is the amount of glucose fed, CGCt-PHA the estimated cumulative glucose consumption since PHA-time zero, CCPt-PHA is the cumulative CO2 production since PHA-time zero, which is the beginning of the PHA accumulation stage. YCO2/G-PHA is the yield of CO2 from glucose consumption during non-growth stage, which was obtained as 1.07 g g-1 from previous study.

NA-only PHA synthesis: only nonanoic acid was fed as the PHA substrate at linear rates according to Eq. 2, following a pulse addition of 3 g l-1 at the beginning of the second stage. St PHA NA = FPHA NA ( tPHA tPHA0 )

(2)

where St -PHA-NA is the amount of nonanoic acid fed, FPHA-NA is the linear feeding rate, which was 0.47 g NA l-1 h-1 during the first 13 h of the PHA stage then increased to 1.01 g NA l-1 h-1 till the end of the fermentation, and tPHA0 is the time point when a new feeding rate is applied.

NA+G co-feeding PHA synthesis: both carbon sources were fed simultaneously during the PHA accumulation stage. Nonanoic acid was fed at a linear rate of 0.47 g NA l-1 h-1

73

according to Eq. 2, following a pulse addition of 3 g l-1, and glucose was fed according to Eq. 1.

4.4 Results and discussion

The yield of CO2 from glucose during the growth phase of P. putida KT2440 has been shown to be about 0.63 g g-1 (Sun et al. 2006/Chapter 3). Although the yield during non-growth PHA accumulation using glucose is also constant (data not shown), it is much larger (1.07 g g-1). By feeding glucose according to the estimation of its consumption, the glucose concentration was maintained between 5 g l-1 and 20 g l-1, well below the inhibitory level but above a growth-limiting level during the growth phase of all fermentations. The feeding of nonanoic acid at the chosen linear rates resulted in nonanoic acid concentration below the 3 g l-1 level, above which growth inhibition occurs. (Sun et al. 2007/Chapter 5).

Only 3-hydroxyoctanoate (C8) and 3-hydroxydecanoate (C10) were found as significant subunits of the PHA synthesized during G-only fermentation (Figure 4-1a); while only 3-hydroxyheptanoate (C7) and 3-hydroxynonanoate (C9) formed during NA-only fermentation (Figure 4-1b). When the two carbon sources were fed together, all of these subunits were detected (Figure 4-1c). Based on these results and on the known synthetic pathways, it can be assumed that C8 and C10 components were solely derived

74

from glucose, while C7 and C9 were solely from nonanoic acid. Therefore, the relationships between carbon consumption and corresponding PHA components yield can be plotted (Figure 4-2).
(a) 1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 12 13 3-HO (C8) 3-HD (C10) (b) 10 9 8 7 6 5 4 3 2 1 0 19 10 3-HHp (C7) 3-HN (C9) 8 7 6 5 4 3 2 1 0 20 30 40 50 10 20 30 40 50 Culture time (h) Culture time (h) (c)

PHA subunits (g l-1)

14

15

16

17

18

Culture time (h)

Figure 4-1 PHA subunits produced during the nitrogen-limited stage of each fermentation. Nitrogen-limitation began at about the same time as the first data point shown in each figure. (a) G-only PHA synthesis. (b) NA-only PHA synthesis. (c) NA+G co-feeding PHA synthesis. The PHA subunits are 3-HHp, 3-hydroxyheptanoate, 3-HO, 3-hydroxyoctanoate, 3-HN, 3-hydroxynonanoate, 3-HD, 3-hydroxydecanoate.
Total carbon in (C8+C10) subunits (mmol) Total carbon of (C7+C9) subunits (mmol)

500 400 300 200 100 0 0 400 800 1200 1600 2000 Total nonanoic acid carbon consumed (mmol)

200 160 120 80 40 0 0

G-only NA+G co-feed NA-only

1000

2000

3000

4000

5000

Total glucose carbon consumed (mmol)

Figure 4-2 Relationship between total carbon of nonanoic acid (NA) or glucose (G) consumed and total carbon accumulated into the corresponding PHA subunits. Closed squares are from G-only PHA synthesis, open squares from NA-only PHA synthesis, and half-closed squares from NA+G co-feeding.

The slopes of the regression curves reflect the separate overall yields, which are Y(C7+C9)/NA (the C7 and C9 yield from nonanoic acid) and Y(C8+C10)/G (the C8 and C10 yield from glucose). Y(C7+C9)/NA was 0.254 mol mol-1 during NA-only PHA synthesis,
75

which is very close to Durners result (YC-PHA/C= 0.24 g g-1) when culturing P. oleovorans (now classified as a P. putida strain) on nonanoate (Durner et al. 2001). In contrast, Y(C8+C10)/G was only 0.057 mol mol-1 under G-only PHA synthesis. These yields demonstrate that nonanoic acid (or aliphatics in general) is a much more efficient carbon source than glucose (carbohydrates in general) for MCL-PHA synthesis. Furthermore, under NA+G co-feeding conditions, Y(C7+C9)/NA almost doubled to 0.450 mol mol-1 while Y(C8+C10)/G decreased to 0.011 mol mol-1. These data demonstrate that more nonanoic acid can be diverted to PHA synthesis simply by co-feeding glucose. Although glucose addition would also affect the subunit composition and thus the thermo-mechanical properties, glucose is presently much cheaper than nonanoic acid. Such a co-feeding strategy could significantly reduce the cost of production of certain types of MCL-PHA.

4.5 References
Diniz SC, Taciro MK, Gomez JG, da Cruz Pradella JG (2004) High-cell-density cultivation of Pseudomonas putida IPT 046 and medium-chain-length polyhydroxyalkanoate production from sugarcane carbohydrates. Appl Biochem Biotechnol 119:51-70 Du GC, Yu J (2002) Metabolic analysis on fatty acid utilization by Pseudomonas oleovorans: mcl-poly(3-hydroxyalkanoates) synthesis versus beta-oxidation. Process Biochemistry 38:325-332 Durner R, Zinn M, Witholt B, Egli T (2001) Accumulation of poly[(R)-3-hydroxyalkanoates] in Pseudomonas oleovorans during growth in batch and chemostat culture with different carbon sources. Biotechnol Bioeng 72:278-288

76

Eggink G, Dewaard P, Huijberts GNM (1992) The role of fatty-acid biosynthesis and degradation in the supply of substrates for poly(3-hydroxyalkanoate) formation in Pseudomonas putida. FEMS Microbiol Rev 103:159-163 Huijberts GNM, Eggink G, Dewaard P, Huisman GW, Witholt B (1992) Pseudomonas putida KT2442 cultivated on glucose accumulates poly(3-Hydroxyalkanoates) consisting of saturated and unsaturated monomers. Appl Environ Microbiol 58:536-544 Huisman GW, de Leeuw O, Eggink G, Witholt B (1989) Synthesis of poly-3-hydroxyalkanoates is a common feature of fluorescent pseudomonads. Appl Environ Microbiol 55:1949-1954 Kim BS (2002) Production of medium chain length polyhydroxyalkanoates by fed-batch culture of Pseudomonas oleovorans. Biotechnol Lett 24:125-130 Kim GJ, Lee IY, Yoon SC, Shin YC, Park YH (1997) Enhanced yield and a high production of medium-chain-length poly(3-hydroxyalkanoates) in a two-step fed-batch cultivation of Pseudomonas putida by combined use of glucose and octanoate. Enzyme Microb Technol 20:500-505 Ramsay BA, Saracovan I, Ramsay JA, Marchessault RH (1991) Continuous production of long-side-chain poly-beta-hydroxyalkanoates by Pseudomonas oleovorans. Appl Environ Microbiol 57:625-629 Sanchez RJ, Schripsema J, da Silva LF, Taciro MK, Pradella JGC, Gomez JGC (2003) Medium-chain-length polyhydroxyalkanoic acids (PHA(mcl)) produced by Pseudomonas putida IPT 046 from renewable sources. European Polymer Journal 39: 1385-1394 Sun Z, Ramsay JA, Guay M, Ramsay BA (2006) Automated feeding strategies for high-cell-density fed-batch cultivation of Pseudomonas putida KT2440. Appl Microbiol Biotechnol 71:423-431 Sun Z, Ramsay JA, Guay M, Ramsay BA (2007) Carbon-limited fed-batch production of medium-chain-length polyhydroxyalkanoates from nonanoic acid by Pseudomonas putida KT2440. Appl Microbiol Biotechnol 74:69-77 Timmis KN (2002) Pseudomonas putida: a cosmopolitan opportunist par excellence Environ Microbiol 4:779-781

77

CHAPTER 5

Carbon-limited Fed-batch Production of Medium-chain-length Polyhydroxyalkanoates from Nonanoic Acid by Pseudomonas putida KT2440
Zhiyong Sun, Juliana A. Ramsay, Martin Guay, Bruce A. Ramsay

Originally published February, 2007 in Applied Microbiology and Biotechnology, 74 (1): 69-77
78

5.1 Abstract
Pseudomonas putida KT2440 grew on glucose at a specific rate of 0.48 h-1 but accumulated almost no poly(3-hydroxyalkanoates) (PHA). Subsequent nitrogen-limitation on nonanoic acid resulted in the accumulation of only 27% medium-chain-length PHA (MCL-PHA). In contrast, exponential nonanoic acid-limited growth (= 0.15 h-1) produced 70 g l-1 biomass containing 75% PHA. At a higher exponential feed rate (= 0.25 h-1), the overall productivity was increased but less biomass (56 g l-1) was produced due to higher oxygen demand, and the biomass contained less PHA (67%). It was concluded that carbon-limited exponential feeding of nonanoic acid or related substrates to cultures of P. putida KT2440 is a simple and highly effective method of producing MCL-PHA. Nitrogen limitation is unnecessary.

5.2 Introduction
Poly(3-hydroxyalkanoates) (PHAs) have attracted extensive commercial interest due to their inherent biocompatibility and biodegradability (Zinn et al. 2001). In particular, medium-chain-length PHAs (MCL-PHAs), which contain 6 to 14 carbons in their repeating units, show great promise as thermoelastomers for biomedical applications, such as drug delivery (Pouton and Akhtar 1996) and tissue engineering (Williams et al. 1999; Chen and Wu 2005). Efficient production of MCL-PHAs is a prerequisite for

79

extensive study of these polyesters. Although it was observed that 8-carbon and 9-carbon alkanes or alkanoates result in the highest MCL-PHA accumulation (Brandl et al. 1988; Gross et al. 1989; Kang et al. 2001), and octanoic acid was used as the carbon source in pH-stat processes with nitrogen limitation to obtain 47 g l-1 biomass containing 55% PHA (Dufresne and Samain 1998) and 63 g l-1 containing 62% PHA (Kim 2002), almost no research into the development of production processes using nonanoic acid as the principle carbon source was reported. While octanoic acid is derived from coconut and palm kernal oils, nonanoic acid can be produced from oleic, linoleic and erucic and other carboxylic acids produced in temperate zone plants such as canola. For a variety of reasons, nonanoic acid has potential as a commodity feedstock for biorefinery processes in industrialized countries.

Nitrogen (N) or phosphorus (P) limitation stimulates rapid PHA synthesis in most of the well-studied SCL (short-chain-length)-PHA-synthesizing bacteria, such as Ralstonia eutropha. Therefore most SCL-PHA production processes employ a stage of rapid cell growth followed by a PHA accumulation stage, which is almost always N-limited or P-limited. Possibly due to an assumption that MCL-PHA production physiology is similar to that of most SCL-PHA accumulating bacteria, almost all publications dealing with MCL-PHA production have incorporated N or P limitation (Preusting et al. 1993; Huijberts and Eggink 1996; Dufresne and Samain 1998; Jung et al. 2001; Kim 2002).
80

Other than continuous plug-flow processes, which are impractical at a commercial scale, fed-batch fermentation is theoretically and practically the best method of achieving a high density of biomass containing the highest possible amount of PHA. High PHA content is critical to the reduction of PHA separation cost in commercial processes. Since the rate of substrate demand is difficult to measure or predict, design and control of a N-limited or P-limited accumulation phase in a fed-batch process is difficult when it involves the feeding of inhibitory substrates such as short-chain-length and medium-chain-length carboxylic acids. While a pH-stat approach is feasible, simple exponential feeding of substrate is more reliable and potentially more productive. However, exponential feeding is only possible when N or P limitation is not imposed.

The metabolic regulation of SCL-PHA has been well studied (Senior and Dawes 1971; Steinbchel and Ltke-Eversloh 2003), but less is known about the factors controlling MCL-PHA synthesis. While the rate of MCL-PHA production in P. resinovorans is greatly stimulated by N-limitation (Ramsay et al. 1992), this did not occur in P. oleovorans GPo1 ATCC 29347 (Ramsay et al. 1991) now recognized to be a strain of P. putida (Diard et al. 2002). Significant MCL-PHA accumulation during the exponential growth phase of P. putida KT2442 (Huisman et al. 1992), P. putida U (Carnicero et al. 1997) and P. putida GPo1 (Durner et al. 2001) has been reported, indicating little need for N or P limitation. We found that high MCL-PHA content
81

(>50%) occurred under carbon-limited conditions in a chemostat study of P. putida KT2440 using nonanoic acid as carbon source (unpublished data). P. putida strains are known to produce large amounts of MCL-PHA (Lee et al. 2000; Diniz et al. 2004), and strain KT2440 (ATCC 47054) has no plasmid load (Timmis 2002). Based on these observations, the aim of the present study was to compare a single-stage carbon-limited process with a two-stage process incorporating an N-limited stage for the production of MCL-PHAs by P. putida KT2440. Production from nonanoic acid and glucose was also compared.

5.3 Materials and methods

5.3.1 Microorganism and growth medium
P. putida KT2440 (ATCC 47054) was maintained in lyophiles and on nutrient agar plates at 4 C. For the fermentation using glucose as sole substrate during the growth phase, the inoculum medium, initial and feeding medium for the growth phase of the fermentation was described previously (Sun et al. 2006/Chapter 3). For all other fermentations the inoculum medium contained per liter: 4.70 g (NH4)2SO4, 0.80 g MgSO47H2O, 12.00 g Na2HPO47H2O, 2.70 g KH2PO4, 3.00 g nutrient broth. The initial culture medium contained per liter: 4.70 g (NH4)2SO4, 0.80 g MgSO47H2O, 18.00 g Na2HPO47H2O, 4.05 g KH2PO4, 1-3 g nonanoic acid and 10 ml trace element solution. Each liter of trace element solution contained 10 g FeSO47H2O, 3 g CaCl22H2O, 2.2 g ZnSO47H2O, 0.5 g
82

MnSO44H2O, 0.3 g H3BO3, 0.2 g CoCl26H2O, 0.15 g Na2MoO42H2O, 0.02 g NiCl26H2O and 1.00 g CuSO45H2O. Nitrogen was supplied using 14% (w/v) NH4OH to control the pH during growth and was replaced by 2 M KOH during N-limited stages. Nonanoic acid (96%, Sigma Aldrich) was filter-sterilized. To avoid precipitation, all the needed MgSO47H2O was not added to the initial medium. Additional MgSO47H2O was fed at a ratio of 0.033 g MgSO47H2O g-1 nonanoic acid addition assuming a YX/Mg of 240 g g-1 (Sun et al. 2006/Chapter 3) and a YX/NA of 0.80 g g-1.

5.3.2 Fermentation conditions

Inoculum culture was cultivated in 500 ml shake flasks (100 ml medium) at 301C and 200 rpm overnight. Fed-batch fermentations were all carried out at 311 C with a 3.5 l initial working volume in a Minifors 5 l stirred tank bioreactor (Infors HT, Bottmingen, Switzerland). Data acquisition (dissolved oxygen, outlet gas CO2, carbon source and base addition, and pH) and control was conducted with LabVIEW 6.1 (National Instruments). pH was controlled at 6.850.05. Dissolved oxygen was measured with an Ingold polarographic probe and maintained at or above 40% air saturation (except where indicated) by adjusting the agitation speed up to 1,200 rpm and by automatically adjusting the mixture of air and pure oxygen flow via mass flow controllers while maintaining total gas flow at 1 vvm. Exit gas CO2 content was measured with an infrared CO2 monitor

83

(Guardian Plus, Topac Inc. Hingham, MA, USA). Feeding of nonanoic acid with peristaltic pumps was automatically controlled based on the mass of the reservoirs.

5.3.3 Substrate feeding and control methods

5.3.3.1 Nutrient-unlimited growth on glucose followed by an N-limited accumulation phase on nonanoic acid (two-stage fed-batch) A continuous feeding strategy based on glucose estimation was applied to first obtain about 40 g l-1 of biomass, as previously described (Sun et al. 2006/Chapter 3). N-limitation was then imposed by replacing ammonium hydroxide with KOH. During this N-limited stage, a pulse of 3 g l-1 nonanoic acid was first added, followed by linear feeding at 0.47 g NA l-1 h-1 until 17 h, then 1.01 g NA l-1 h-1 until the end of the fermentation. 5.3.3.2 Single-stage, nonanoic acid-limited fed-batch The initial nonanoic acid concentration was 1 g l-1. Additional substrate was fed at the rate shown in Eq. 1.

St =

Xt X = 0 et YX / S YX / S

(1)

where St is total nonanoic acid required to produce biomass Xt at time t, YX/S is the yield of biomass from nonanoic acid (0.80 g g-1), X0 is the initial biomass, obtained by measuring the OD650nm of the inoculum culture and converted to cell concentration in the bioreactor, and is the desired specific growth rate.
84

5.3.3.3 Nonanoic acid-limited growth followed by an N-limited accumulation phase on nonanoic acid (two-stage fed-batch) With 3 g l-1 of initial nonanoic acid, exponential feeding was applied according to Eq. 1 with a target = 0.25 h-1. Shortly before reaching the maximum oxygen transfer capacity, nitrogen limitation was imposed by replacing ammonium hydroxide with KOH. During this N-limited stage, nonanoic acid was fed at a rate of 1.31 g NA l-1 h-1 until 41.5 h, then 1.82 g NA l-1 h-1 until the end of the fermentation.

5.3.4 Analytical procedures

Cell dry weight was determined after lyophilization of biomass obtained by centrifugation of 5 ml culture broth at 17,000 x g for 15 min and washing twice with 5 ml distilled water. The supernatant of the sample and the distilled water wash were saved for analysis of the nonanoic acid and key nutrients. Nonanoic acid concentration was determined by methylation followed by GC analysis with decanoic acid as internal standard (Ramsay et al. 1991). Phosphate was measured via reduction of phosphomolybdate to molybdene blue (Clesceri et al. 1999). Ammonium was determined by the phenol-hypochlorite method (Weatherburn 1967). PHA content and composition was determined by modification of the Braunegg et al. (1978) and Lageveen et al. (1988) methodologies. Lyophilized cell samples were suspended with 2 ml of methanol containing 15% (v/v) H2SO4 and 0.2% (w/v) benzoic acid (internal standard) and 2 ml of chloroform. Methanolysis was carried out in a 100 oC water bath for 3.5 h with periodic vigorous
85

mixing. After cooling to room temperature, 2 ml distilled water was added and vortexed for 1 min. The resulting emulsion was left overnight for phase separation. The organic phase was used for GC analysis. The PHA standard for this assay was prepared by repeated cycles of solvent extraction followed by precipitation in cold methanol. Its composition was determined by NMR. Residual biomass (Xr) was defined as the total cell concentration minus the PHA concentration. The specific PHA synthesis rate based on the amount of residual biomass (qPHA(Xr)) was defined as (dP/dt)/Xr (g PHA g of Xr -1 h-1). The specific PHA synthesis rate based on the amount of PHA (qPHA(PHA)) was defined as (dP/dt)/PHA (g PHA g PHA-1 h-1). They were calculated using the slope of the PHA concentration curve and the measured Xr and PHA concentration values, respectively, at that point.

5.4 Results
Growth on glucose followed by nonanoic acid linear feeding under N-limitation In SCL-PHA fermentations with bacteria such as Ralstonia eutropha or Burkholderia cepacia, glucose is typically fed to produce dense biomass (growth phase). N or P is then allowed to fall to a growth-limiting value which stimulates PHA synthesis. At the point of N or P limitation or slightly before, co-substrates such as propionic acid are fed to produce the desired copolymer composition during accumulation phase (Ramsay et al. 1990). High-cell-density cultivation of P. putida KT2440 on glucose had already been
86

established (Sun et al. 2006/Chapter 3). Based on this knowledge, 40 g l-1 of biomass was produced using glucose as the sole source of carbon and energy. Almost no PHA synthesis was detected in P. putida grown on glucose without N or P limitation (Figure 5-1). The specific growth rate throughout this phase was 0.48 h-1. Although some PHA was initially produced during the N-limited stage, the nonanoic acid feed rate of 0.47 g NA l-1 h-1 proved to be insufficient. Increasing the feed rate to 1.01 g NA l-1 h-1 enabled the cells to accumulate more PHA. The nonanoic acid concentration in the reactor remained below 2 g l-1 during this period. Eventually 26.8% PHA was accumulated in 46.1 g l-1 biomass giving a cumulative PHA productivity of 0.25 g NA l-1 h-1.

30

50 Biomass % PHA NH4+ NA qPHA (Xr)

10

0.06

40

NH4 and NA concentration (g l )

-1

25

0.05

PHA (%)

30

FNA=1.01 g l-1 h-1 FNA=0.47 g l h

-1 -1

15

0.03

20

10

0.02

10

0.01

0 0 10 20 30 40 50

0.00

Culture time (h)

Figure 5-1 Growth of P. putida KT2440 on glucose followed by N-limited growth on nonanoic acid (NA). The division between growth phase and N-limited phase is indicated by the dashed line. The nonanoic acid feeding rate (FNA) was increased from 0.47 g NA l-1 h-1 to 1.01 g NA l-1 h-1 at 27 h during the N-limited phase, as indicated by the dash-dot line.

87

qPHA (Xr) (g g-1 h-1)

Biomass (g l )

20

0.04

-1

5.4.1 Single-stage, exponential feeding of nonanoic acid (target = 0.15 h-1)

To further assess whether N or P limitation is required for substantial MCL-PHA synthesis in P. putida, a nonanoic acid single-stage fed-batch process was conducted. To avoid overfeeding, the in the feeding equation (Eq. 1) was set at a conservative value of 0.15 h-1. The nonanoic acid concentration was maintained well below 1 g l-1 until 41 h (Figure 5-2). The exponential rates of consumption and growth indicate that nonanoic acid was the only limiting nutrient until the end of the fermentation. Control of pH with ammonium hydroxide allowed the ammonium concentration to be maintained between 1 and 2 g l-1 during the entire fermentation. Phosphate concentration was not controlled but it never decreased to a level that limited the growth rate. The total cell concentration increased up to 70.2 g l-1 and the measured , based on dry biomass samples, was the predicted value of 0.15 h-1. The inoculum culture (grown on nutrient broth) contained no measurable MCL-PHA but PHA content increased throughout the fermentation to a maximum of 75.4%, giving a cumulative PHA productivity of 1.11 g PHA l-1 h-1. Mainly hydroxynonanoate (C9) and hydroxyheptanoate (C7) PHA repeating units were detected. Also about 1 molar percent of a combination of hydroxyhexanoate (C6), hydroxyoctanoate (C8) and hydroxydecanoate (C10) was detected. The ratio of C9/C7 was about 2.4:1 (mol/mol) throughout the fermentation. Although agitation was maximal and 1 vvm of pure oxygen was supplied to the fermentor, after 41 h the dissolved oxygen concentration decreased to zero. Oxygen limited cell growth, and consequently caused
88

nonanoic acid to rapidly accumulate in the medium (to above 3 g l-1). This resulted in growth inhibition, cell lysis, and excessive foaming.

80 70 Biomass % PHA NH4+ PO4+ NA 50 40 30 20 10 0 0 10 20 30 40 50

8 7 6 5 4 3 2 1 0

60

Culture time (h)

Figure 5-2 Carbon-limited growth of P. putida KT2440 on nonanoic acid (NA) at = 0.15 h-1.

5.4.2 Single-stage, exponential feeding of nonanoic acid (target = 0.25 h-1)

To improve productivity and evaluate the effects of a higher growth rate on PHA accumulation, the single-stage carbon-limited fermentation was repeated but at a higher target (0.25 h-1) in the feeding equation (Eq.1). Key results were compared to those obtained at = 0.15 h-1 (Figure 5-3). Very similar patterns were observed in the two fermentations despite the difference in feed rate. The final total biomass, total PHA, and % PHA of the fermentation at = 0.25 h-1 were 56.0 g l-1, 37.4 g l-1, and 66.9%,
89

NH4 , PO 4 , and NA concentration (g l )

Biomass (g l ) and PHA (%)

-1

-1

respectively, which were lower than the final values at the lower (Figure 5-3a). However, the final cumulative PHA productivity (Figure 5-3b) increased to 1.44 g PHA l-1 h-1 at = 0.25 h-1 because the total cultivation time decreased significantly. The specific rate of PHA synthesis based on residual biomass (qPHA (Xr)) increased steadily at both feed rates up to 0.48 g g-1 h-1 at = 0.25 h-1 and 0.53 g g-1 h-1 at = 0.15 h-1.

5.4.3 Single-stage, exponential feeding of nonanoic acid (target = 0.25 h-1) followed by N-limited culture on nonanoic acid
Single-stage, exponential feeding of nonanoic acid resulted in high PHA productivity and content but perhaps N-limitation could further increase PHA synthesis. Therefore an experiment was conducted in which N-limitation was imposed after exponential feeding (target = 0.25 h-1) of nonanoic acid (Figure 5-4). Cell growth and PHA synthesis during the growth stage were very similar to the single-stage fed-batch where both were fed to achieve = 0.25 h-1 but nitrogen limitation was imposed at 27 h by replacing ammonium hydroxide with KOH. Linear feeding after this point allowed the nonanoic acid concentration to be maintained just below the 3 g l-1 level. However, although the total amount of biomass increased, there was little increase in the percentage of PHA in the biomass produced during the N-limited stage. The final PHA concentration was 52.4 g l-1, corresponding to an overall (i.e. cumulative) productivity of 0.91 g PHA l-1 h-1.

90

70

100 Biomass at =0.15 h-1 Biomass at =0.25 h-1 -1 PHA at =0.15 h -1 PHA at =0.25 h -1 %PHA at =0.15 h -1 %PHA at =0.25 h

Biomass and PHA concentration (g l-1)

60 50 40 30 20

80

60

40

20 10 0 2.0

(a)
Cumulative productivity at =0.15 h-1 Cumulative productivity at =0.25 h-1 qPHA (Xr) at =0.15 h-1

Cumulative PHA productivity (g l h )

.5

-1

-1

1.5

qPHA (Xr) at =0.25 h-1

.4

.3 1.0 .2 .5 .1

0.0 0 10 20 30 40

(b)
Culture time (h)

0.0

Figure 5-3 Effect of growth rate on PHA synthesis in P. putida KT2440 under nonanoic acid limitation. Closed and open symbols represent data from the = 0.15 h-1 and the = 0.25 h-1 -1 -1 fermentations, respectively. qPHA(Xr) is the specific PHA synthesis rate (g PHA g Xr h ).

91

q PHA (Xr) (g g h )

-1

-1

PHA (%)

100 Biomass %PHA NH4+ NA

8 7 6 5 4

Biomass (g l-1) and PHA (%)

80

60

40

3 2

20

FNA=1.31 g l-1 h-1

FNA=1.82 g l-1 h-1

1 0

0 0 10 20 30 40 50 60

Culture time (h)

Figure 5-4 Two-stage, fed-batch using nonanoic acid (NA) as the sole carbon source. Exponential feeding (= 0.25 h-1) was used during the growth phase with all other nutrients in excess. The division between the growth and N-limited phases is indicated by the dashed line. The nonanoic acid feed rate was increased from 1.31 g NA l-1 h-1 to 1.82 g NA l-1 h-1 at 41.5 h during the N-limited phase as indicated by the dash-dot line.

5.4.4 Biomass and PHA yield

Total biomass, PHA, and residual biomass (Xr) data from both single-stage (carbon-limited) fed-batch fermentations were plotted against total nonanoic acid consumption (Figure 5-5). There was a linear relationship between total biomass and nonanoic acid consumption demonstrating a yield of 0.83 g biomass g-1 nonanoic acid (r2= 0.995) throughout both fermentations. In contrast, the yield of PHA and residual biomass varied during the growth. The yield of PHA increased gradually throughout the growth phase as more carbon was directed to PHA synthesis and less towards residual
92

NH4+, and NA concentration (g l-1)

biomass. The overall PHA yield was about 0.60 g g-1 nonanoic acid while overall yield of residual biomass was 0.24 g g-1 nonanoic acid.

X at =0.15 h-1 200 X at =0.25 h-1 PHA at =0.15 h-1 PHA at =0.25 h-1 Xr at =0.15 h-1 150 Xr at =0.25 h-1

Biomass, PHA, and Xr (g)

100

50

0 0 50 100 150 200 250 300

Total NA consumption (g)

Figure 5-5 Yield of total biomass, PHA, and residual biomass (Xr) from nonanoic acid (NA). Closed and open symbols represent data from = 0.15 h-1 and the = 0.25 h-1 fermentations, respectively. The slopes of the curves indicate yields from nonanoic acid.

5.5 Discussion
5.5.1 Process design implications
Relatively little PHA was accumulated whether the N-limited PHA accumulation phase on nonanoic acid was preceded by growth on glucose or on nonanoic acid. Although better control of nonanoate concentration during N-limited production would increase PHA production, simple exponential feeding of nonanoic acid with no N or P limitation
93

was clearly more productive, based on results presented in this paper. Similar results are likely with octanoic acid, which is also carboxylic acid with one carbon less than nonanoic acid. In shake flasks, nonanoic acid concentrations higher than 3 g l-1 inhibit growth of P. putida KT2440. This has led process development researchers to work with less toxic substrates such as oleic acid. Although remarkable progress has been made in the production of PHA from oleic acid (Huijberts and Eggink 1996; Lee et al. 2000), its use limits the type of PHA that can be made and can lead to downstream processing difficulties. The standard approach to PHA production exemplified by the Biopol process (Byrom 1990) consists of dense biomass production followed by an accumulation phase. The Biopol process is relatively easy to control because the toxic co-substrate (propionic acid) used in the accumulation phase is typically mixed at low concentration with a high concentration of glucose. Use of a single toxic substrate such as nonanoic acid in an accumulation phase requires more sophisticated control. Fortunately a provoked accumulation phase (second phase in two-stage fermentation) does not appear to be required or even beneficial to MCL-PHA synthesis in P. putida KT2440. The Biomer process (Hnggi 1990) currently used for PHB production employs a single-stage fermentation with no accumulation phase. The organism used in the Biomer process, A. latus, exhibits a growth-associated PHA production pattern (Braunegg and Bogensberger 1985). Co-substrates may be fed to produce copolymers in such a single-stage process (Ramsay et al. 1990). Although the kinetics of PHA synthesis in P. putida KT2440 are
94

not directly linked to growth, we propose a similar single-stage process in which nonanoic acid or related substrates are fed exponentially. Co-substrates may be mixed with the major substrate to produce the desired polymer subunit composition.

Since a lower specific growth rate demands less oxygen transfer, a higher final PHA content (75%) was achieved than when was controlled at 0.25 h-1 (67%). However, the cumulative PHA productivity decreased from 1.44 g PHA l-1 h-1 to 1.11 g PHA l-1 h-1 when operating at a lower due to the longer culturing time. Thus there is a trade-off between the amount of PHA in the biomass and fermentation productivity. An optimal approach can be calculated based on economic considerations such as substrate and separation costs, as well as the oxygen transfer and mixing properties of the fermentor to be employed.

5.5.2 Yield
Yields of total biomass and PHA from nonanoic acid were 0.83 g g-1 and 0.60 g g-1, respectively (Figure 5-5). These are relatively high when compared to 0.40 g PHA g octanoate-1 under glucose co-feeding conditions (Kim et al. 1997), or 0.63 g PHA g octane-1 under optimized PHA accumulating conditions in the second stage of a continuous process (Hazenberg and Witholt 1997). The consistency of the yield values (Figure 5-5) and C9/C7 ratio of fermentations at different growth rates indicate that this is a highly repeatable process.
95

There appears to be a low maintenance energy requirement in this strain because the biomass yield values were essentially the same at both growth rates. Much more PHA is being synthesized during latter parts of the fermentation yet the yield remained the same (Figure 5-5). This indicates that the true yields for YX/S and YP/S are very similar. Since the % PHA, PHA concentration and qPHA(Xr) increased continuously at both growth rates until oxygen transfer limited production, higher final PHA content and productivity could be achieved in a bioreactor system with better oxygen transfer capacity. Extended cultivation using linear feeding of nonanoic acid may further increase PHA accumulation.

5.5.3 Kinetics of MCL-PHA synthesis

Although nutrient (mainly N and P) limitation was commonly applied to MCL-PHA production processes, some controversial results have been reported. Limitation of N or P was shown to be stimulatory to the synthesis of PHA in P. oleovorans ATCC 29347 growing on n-alkanes (Lageveen et al. 1988), P. resinovorans and P. putida BM01 on octanoate (Ramsay et al. 1992; Kim et al. 1997) and P. putida KT2442 on oleic acid (Huijberts and Eggink 1996; Lee et al. 2000). In contrast, it was also reported that nutrient limitation is unnecessary for the production of significant amounts of PHA in P. putida KT2442 (Huisman et al. 1992) , P. oleovorans ATCC 29347 (Ramsay et al. 1991; Durner et al. 2001 ) or P. putida U (Carnicero et al. 1997) on octanoate. In the present study, high PHA accumulation was achieved during nonanoic acid limited growth with no other
96

apparent nutrient limitation. The link between MCL-PHA accumulation and growth limitation by key nutrients such as nitrogen or phosphate seems to differ depending on the strain, carbon sources, the cultivation conditions, or possibly a combination of these factors.

There are many possible explanations for the production of substantial amounts of PHA during carbon-limited growth. Some deficiency in the TCA cycle, blockage in -oxidation, a detoxification mechanism or, as in the case of Azotobacter vinelandii UWD an ineffective NADH oxidase (Page and Knosp 1989) may divert substrate and energy toward PHA synthesis to cause this effect. Although it was not our purpose to study the mechanisms governing PHA production in this organism, data analysis suggests that unlike A. latus (Braunegg and Bogensberger 1985), the kinetics of MCL-PHA synthesis during the growth of P. putida KT2440 are not strictly linked to growth kinetics. Rather than remaining constant, the PHA content increased from close to zero in the inoculum to above 70%. The continually increasing qPHA (Xr) (Figure 5-3b) demonstrated that the growing cells were synthesizing PHA at a rate higher than the growth rate of the other biomass components (residual biomass). After a certain point in the fermentation process, the specific rate of PHA synthesis remains constant only if calculated based on the amount of PHA in the cells (Figure 5-6). This indicates that the rate of PHA synthesis is controlled by a property of the granules themselves, possibly related to phasin proteins
97

found on the surface of PHA granule. Much more kinetic, physiological, and biochemical study is required to understand the mechanisms of MCL-PHA synthesis during growth of P. putida KT2440 and other MCL-PHA accumulating bacteria.
1.2 qPHA (PHA) at =0.15 h-1 1.0 qPHA (PHA) at =0.25 h-1

qPHA (PHA) (g g-1 h-1)

0.8

0.6

0.4

0.2

0.0 0 10 20 30 40 50 60 70 80

% PHA
Figure 5-6 The specific rate of PHA synthesis (per gram of PHA) approaches a constant value as the concentration of PHA in the biomass increases.

Acknowledgement This research was supported by the Natural Science and Engineering Research Council of Canada and a grant from BIOCAP Canada.

5.6 References
Brandl H, Gross RA, Lenz RW, Fuller RC (1988) Pseudomonas oleovorans as a source of poly(beta-hydroxyalkanoates) for potential applications as biodegradable polyesters. Appl Environ Microbiol 54:1977-1982
98

Braunegg G, Bogensberger B (1985) About kinetics of growth and accumulation of poly-D(-)-3-hydroxybutric acid in Alcaligenes latus strains. Acta Biotechnol 5:339-345 Braunegg G, Sonnleitner B, Lafferty RM (1978) Rapid gas-chromatographic method for determination of poly-beta-hydroxybutyric acid in microbial biomass. Eur J Appl Microb Biotechnol 6:29-37 Byrom D (1990) Industrial production of copolymer from Alcaligenes eutrophus. In: Dawes EA (eds) Novel biodegradable microbial polymers, Kluwer Academic Publishers, Netherlands, pp 113-117. Carnicero D, FernandezValverde M, Canedo LM, Schleissner C, Luengo JM (1997) Octanoic acid uptake in Pseudomonas putida U. FEMS Microbiol Lett 149:51-58 Chen GQ, Wu Q (2005) The application of polyhydroxyalkanoates as tissue engineering materials. Biomaterials 26:6565-6578 Clesceri LS, Greenberg AE, Eaton AD (1999) Standard methods for the examination of water and wastewater 20th edition. American Public Health Association, Washington, DC. Diard S, Carlier JP, Ageron E, Grimont PAD, Langlois V, Guerin P, Bouvet OMM (2002) Accumulation of poly(3-hydroxybutyrate) from octanoate, in different Pseudomonas belonging to the rRNA homology group I. Syst Appl Microbiol 25:183-188 Diniz SC, Taciro MK, Gomez JG, da Cruz Pradella JG (2004) High-cell-density cultivation of Pseudomonas putida IPT 046 and medium-chain-length polyhydroxyalkanoate production from sugarcane carbohydrates. Appl Biochem Biotechnol 119:51-70 Dufresne A, Samain E (1998) Preparation and characterization of a poly(beta-hydroxyoctanoate) latex produced by Pseudomonas oleovorans. Macromolecules 31:6426-6433 Durner R, Zinn M, Witholt B, Egli T (2001) Accumulation of poly[(R)-3-hydroxyalkanoates] in Pseudomonas oleovorans during growth in batch and chemostat culture with different carbon sources. Biotechnol Bioeng 72:278-288

99

Gross RA, Demello C, Lenz RW, Brandl H, Fuller RC (1989) Biosynthesis and characterization of poly(beta-hydroxyalkanoates) produced by Pseudomonas oleovorans. Macromolecules 22:1106-1115 Hnggi, UJ (1990) Pilot scale production of PHA with A. latus In: Dawes EA (eds), Novel biodegradable microbial polymers, Kluwer Academic Publishers, Netherlands, pp 65-70 Hazenberg W, Witholt B (1997) Efficient production of medium-chain-length poly(3-hydroxyalkanoates) from octane by Pseudomonas oleovorans: economic considerations. Appl Microbiol Biotechnol 48:588-596 Huijberts GNM, Eggink G (1996) Production of poly(3-hydroxyalkanoates) by Pseudomonas putida KT2442 in continuous cultures. Appl Microbiol Biotechnol 46:233-239 Huisman GW, Wonink E, De Koning G, Preusting H, Witholt B (1992) Synthesis of poly(3-hydroxyalkanoates) by mutant and recombinant Pseudomonas strains. Appl Microbiol Biotechnol 38:1-5 Jung K, Hazenberg W, Prieto M, Witholt B (2001) Two-stage continuous process development for the production of medium-chain-length poly(3-hydroxyalkanoates). Biotechnol Bioeng 72:19-24 Kang HO, Chung CW, Kim HW, Kim YB, Rhee YH (2001) Cometabolic biosynthesis of copolyesters consisting of 3-hydroxyvalerate and medium-chain-length 3-hydroxyalkanoates by Pseudomonas sp. DSY-82. Antonie Van Leeuwenhoek 80:185-191 Kim BS (2002) Production of medium chain length polyhydroxyalkanoates by fed-batch culture of Pseudomonas oleovorans. Biotechnol Lett 24:125-130 Kim GJ, Lee IY, Yoon SC, Shin YC, Park YH (1997) Enhanced yield and a high production of medium-chain-length poly(3-hydroxyalkanoates) in a two-step fed-batch cultivation of Pseudomonas putida by combined use of glucose and octanoate. Enzyme Microb Technol 20:500-505 Lageveen RG, Huisman GW, Preusting H, Ketelaar P, Eggink G, Witholt B (1988) Formation of polyesters by Pseudomonas oleovorans - effect of substrates on formation and composition of poly-(R)-3-hydroxyalkanoates and poly-(R)-3-hydroxyalkenoates. Appl Environ Microbiol 54:2924-2932
100

Lee SY, Wong HH, Choi JI, Lee SH, Lee SC, Han CS (2000) Production of medium-chain-length polyhydroxyalkanoates by high-cell-density cultivation of Pseudomonas putida under phosphorus limitation. Biotechnol Bioeng 68:466-470 Page WJ, Knosp O (1989) Hyperproduction of poly-beta-hydroxybutyrate during exponential growth of Azotobacter vinelandii UWD. Appl Environ Microbiol 55:1334-1339 Pouton CW, Akhtar S (1996) Biosynthetic polyhydroxyalkanoates and their potential in drug delivery. Adv Drug Deliv Rev 18:133-162 Preusting H, Vanhouten R, Hoefs A, Vanlangenberghe EK, Favrebulle O, Witholt B (1993) High-cell-density cultivation of Pseudomonas oleovorans - growth and production of poly (3-hydroxyalkanoates) in 2-liquid phase batch and fed-batch systems. Biotechnol Bioeng 41:550-556 Ramsay BA, Ramsay JA, Lomaliza K, Chavarie C, Bataille P (1990) Production of poly-(beta-hydroxybutyric-co-beta-hydroxyvaleric) acid copolymers. Appl Environ Microbiol 56:2093-2098. Ramsay BA, Saracovan I, Ramsay JA, Marchessault RH (1991) Continuous production of long-side-chain poly-beta-hydroxyalkanoates by Pseudomonas oleovorans. Appl Environ Microbiol 57:625-629 Ramsay BA, Saracovan I, Ramsay JA, Marchessault RH (1992) Effect of nitrogen limitation on long-side-chain poly-beta-hydroxyalkanoate synthesis by Pseudomonas resinovorans. Appl Environ Microbiol 58:744-746 Senior PJ, Dawes EA (1971) Poly-beta-hydroxybutyrate biosynthesis and regulation of glucose metabolism in Azotobacter berjerinchii. Biochem J 125(1): 55-66 Steinbchel A, Ltke-Eversloh T (2003) Metabolic engineering and pathway construction for biotechnological production of relevant polyhydroxyalkanoates in microorganisms. Biochem Eng J 16:81-96 Sun Z, Ramsay JA, Guay M, Ramsay BA (2006) Automated feeding strategies for high-cell-density fed-batch cultivation of Pseudomonas putida KT2440. Appl Microbiol Biotechnol 71 (4): 423-431 Timmis KN (2002) Pseudomonas putida: a cosmopolitan opportunist par excellence. Environ Microbiol 4:779-781
101

Weatherburn MW (1967) Phenol-hypochlorite reaction for determination of ammonia. Anal Chem 39:971-974 Williams SF, Martin DP, Horowitz DM, Peoples OP (1999) PHA applications: addressing the price performance issue: I. Tissue engineering. Int J Biol Macromol 25:111-121 Zinn M, Witholt B, Egli T (2001) Occurrence, synthesis and medical application of bacterial polyhydroxyalkanoate. Adv Drug Deliv Rev 53:5-21

102

CHAPTER 6

Fed-batch Production of Medium-chain-length Polyhydroxyalkanoates by Pseudomonas putida KT2440 using Nonanoic Acid and Glucose Co-feeding
Zhiyong Sun, Juliana A. Ramsay, Martin Guay, Bruce A. Ramsay

Originally submitted June, 2007 to Journal of Biotechnology

103

6.1 Abstract
Medium-chain-length polyhydroxyalkanoates (MCL-PHA) were produced by carbon-limited, single-stage fed-batch fermentations of Pseudomonas putida KT2440 using a nonanoic acid (NA) and glucose (G) co-feeding strategy, in order to enhance the yield of PHA from nonanoic acid. With NA:G= 1:1 in the feed, an exponential (= 0.25 h-1) feeding strategy followed by linear substrate feeding was used to achieve a dry weight biomass of 70.7 g l-1 containing 56% PHA. This resulted in an overall PHA productivity of 1.44 g PHA l-1 h-1, equal to that obtained by nonanoic acid alone fermentation at = 0.25 h-1, while the final overall yield of nonanoic acid to PHA was increased by 25% (0.69 g PHA g NA-1 versus 0.55 g g-1). Further increase of the glucose fraction in the feed (NA:G= 1:1.5) slightly increased the yield (0.71 g PHA g NA-1) but decreased PHA content (48%) and productivity (1.16 g PHA l-1 h-1). The addition of glucose to the feed did not change the PHA composition, although the ratio of 3-OH-nonanoate to 3-OH-heptanoate was slightly increased, probably due to a decrease in the -oxidation of nonanoate.

6.2 Introduction
Polyhydroxyalkanoates (PHA) are biocompatible, biodegradable polyesters synthesized by many microorganisms. PHAs are categorized as short-chain-length (SCL-PHA)
104

containing 3 to 5 carbons in their repeating units, medium-chain-length (MCL-PHA) containing 6 or more monomeric carbons, and SCL-MCL, which contain both SCL and MCL repeating units. SCL-PHAs, such as poly(3-hydroxybutyrate-co-3-hydroxyvalerate) (PHB/V), and SCL- MCL-PHAs, such as poly(3-hydroxybutyrate-co-3-hydroxyhexanoate) (PHB/HHx) can now be produced on an industrial scale (Chen et al. 2001; Lee et al. 1999). However, the production of MCL-PHAs remains at the lab scale due to the nature of the carbon sources, microorganisms, and the degree of process automation, etc. (Sun et al. 2007b/Chapter 2). For the development of potential applications of MCL-PHAs, as well as future industrial development, efficient and economical production systems are required. MCL-PHA fermentation process development has been recently reviewed by Sun et al. (2007b/Chapter 2).

The economics of PHA production is determined by the volumetric PHA productivity, final content in the biomass, carbon substrate cost and PHA yield, as well as the degree of complexity and automation of the fermentation process itself. Substrate cost and the yield of PHA from substrate have a substantial impact on the overall cost of production, and this effect tends to increase with process scale (Choi and Lee, 1997). Although both aliphatic substrates and carbohydrates have been used in lab scale production of MCL-PHA by strains of P. putida (Diniz et al. 2004; Kim, 2002), aliphatic
105

substrates typically result in MCL-PHA with a wider variety of monomeric composition, which is dependent on the substrates fed (Hartmann et al. 2006). Since substrates, such as octanoate and nonanoate, are much more expensive than carbohydrates, efforts have been made to use glucose and octanoate consecutively in two-step, fed-batch fermentations to increase the overall PHA yield from octanoate (Kim et al. 1997). In our recent two-stage, fed-batch fermentation study, we investigated the relative contribution of glucose and nonanoic acid to cell maintenance and MCL-PHA synthesis during the nitrogen-limited, second stage (Sun et al. 2007c/Chapter 4). We found that the PHA yield from nonanoic acid was almost doubled when glucose was fed as a co-substrate.

In this study, which was based on a recently reported single-stage carbon-limited fed-batch MCL-PHA fermentation process (Sun et al. 2007a/Chapter 5), we investigated the possibility of using glucose as a co-substrate in a more cost-effective process for the production of MCL-PHA from nonanoic acid.

6.3 Materials and methods

6.3.1 Microorganism and growth medium
Pseudomonas putida KT2440 (ATCC 47054) was maintained on nutrient agar plates at 4 C prior to inoculation. The inoculum medium contained per liter: 4.70 g (NH4)2SO4, 0.80 g MgSO47H2O, 12.00 g Na2HPO47H2O, 2.70 g KH2PO4, 9 g glucose, 1 g nutrient broth. The initial fermentation medium contained per liter: 4.70 g (NH4)2SO4, 0.80 g
106

MgSO47H2O, 18.00 g Na2HPO47H2O, 4.05 g KH2PO4, 1 g carbon substrates in total and 10 ml trace element solution. Each liter of trace element solution contained 10 g FeSO47H2O, 3 g CaCl22H2O, 2.2 g ZnSO47H2O, 0.5 g MnSO44H2O, 0.3 g H3BO3, 0.2 g CoCl26H2O, 0.15 g Na2MoO42H2O, 0.02 g NiCl26H2O and 1.00 g CuSO45H2O. Nonanoic acid (98%, Spectrum Chemicals) was fed in its pure form. Glucose (99.5%, Sigma-Aldrich) solution of 640 g l-1 was fed separately as it is immiscible with nonanoic acid. Ammonium was supplied by using 14% (w/v) NH4OH solution for pH control, throughout the fermentation. To avoid initial precipitation of medium components, additional MgSO47H2O was mixed in the glucose feeding solution at 33 g l-1 based on YX/Mg of 240 g g-1 (Sun et al. 2006/Chapter 3).

6.3.2 Fermentation conditions

The inoculum was cultivated in two 500 ml shake flasks (150 ml medium each) at 301C and 200 rpm for about 12 h. Fed-batch fermentations were carried out at 281C with 3.0 l initial working volume in a Minifors 5 l stirred tank bioreactor (Infors HT, Bottmingen, Switzerland). Data acquisition (dissolved oxygen, outlet gas CO2 content, substrate and base addition, and pH) and control was conducted with LabVIEW6.1 (National Instruments). pH was controlled at 6.850.05. Dissolved oxygen was measured with an Ingold polarographic probe and maintained at or above 40% air saturation (except where indicated) by adjusting the agitation speed up to 1,200 rpm and by automatically adjusting
107

the mixture of air and pure oxygen flow via mass flow controllers while maintaining the total gas flow at 1 vvm. Exit gas CO2 (%) was measured with an infrared CO2 monitor (Guardian Plus, Topac Inc. Hingham, MA, USA) as direct indication of culture activity. Feeding of nonanoic acid and glucose with peristaltic pumps was automatically and separately controlled based on the mass of the reservoirs. Antifoam 204 (Sigma Aldrich) was injected manually through a sterile septum whenever necessary.

6.3.3 Substrate feeding and control methods

A total of 1 g l-1 of nonanoic acid (NA) and glucose (G) was added to the initial medium. The total carbon requirement was calculated according to the exponential growth equation (Eq. 1).

St =

X t X = 0 ( e t 1) YX / S YX / S

(1) (2)

Where YX / S = YX / NA rNA + YX / G rG

The amount of nonanoic acid and glucose required were calculated separately according to pre-defined ratio, and fed accordingly.

St NA = St rNA St G = St rG

(3) (4)

108

In the equations above, St (g) is the total carbon source required to produce biomass Xt (g) at cultivation time t (h); YX/S, YX/NA, and YX/G are the yield of biomass from total carbon, nonanoic acid, and glucose, respectively; X0 (g) is the initial biomass, obtained by measuring the optical density of the inoculum; (h-1) is the desired specific growth rate; St-NA (or St-G, g) and rNA (or rG, %) are the total nonanoic acid (or glucose) required at cultivation time t and the proportion of the mass of nonanoic acid (or glucose) in the feed, respectively.

Two fermentations with different nonanoic acid to glucose ratio were conducted in this study. During the NA:G= 1:1 fermentation, at 24.9 h when the estimated cell concentration was about 50 g l-1, the substrate feed rate (F, g l-1 h-1) was set at 17 g l-1 h-1 until the end of the fermentation (27.3 h). The NA:G= 1:1.5 fermentation was operated using the exponential feeding strategy (Eq. 1) until 26.45 h.

6.3.4 Analytical procedures

Cell dry weight was determined with lyophilized biomass, which was obtained by centrifugation of 5 ml culture broth at 10,000 x g for 15 min and washed with distilled water. The supernatant of the centrifuged broth was saved for analysis of the major nutrients, nonanoic acid (NA), and glucose (G). Phosphate, ammonium and glucose concentrations were determined as described previously (Sun et al. 2006/Chapter 3). NA concentration was determined by methylation reaction followed by GC analysis (Ramsay
109

et al. 1991), using benzoic acid as the internal standard. Samples for GC analysis of cellular PHA content and composition were prepared as described previously (Sun et al. 2007a/Chapter 5). A gas chromatography (CP-3800, Varian Inc.) with FID was used in all GC analysis. For PHA analysis the GC parameters were: injector temperature 250 C, detector temperature 275 C, 1 l injection, and a split ratio 10. The oven temperature profile was: 90 C, 0.5 min, 6 C min-1 to 96 C, 7 C min-1 to 131 C, 20 C min-1 to 181 C, 5 min. The MCL-PHA standard for the GC analysis was prepared as described by Jiang et al. (2006). The monomeric components were confirmed by GC-MS and composition determined by nuclear magnetic resonance (NMR). The SCL-PHA standard was Biopol, which was poly(3-hydroxybutyrate-co-3-hydroxyvalerate) (81 and 19 mol % of HB and HV, respectively), originally purchased from Imperial Chemical Industries (ICI). All analyses were done in duplicate with the average of the results presented in this paper.

6.4 Results
6.4.1 Nonanoic acid and glucose co-feeding for MCL-PHA production
As previously reported (Sun et al. 2007a/Chapter 5), P. putida KT2440 is able to synthesize substantial amount of MCL-PHA using nonanoic acid as the only carbon source during carbon-limited, exponential growth, without nitrogen nor phosphorus limitation. To lower the total substrate cost, nonanoic acid and glucose were used as
110

co-substrates first at a 1:1 (w/w) ratio ( rNA = rG = 50%). Since YX/NA= 0.83 g g-1 and YX/G= 0.41 g g-1 (Sun et al. 2007a/Chapter 5; Sun et al. 2006/Chapter 3), YX/S was calculated as 0.62 g g-1 for the substrate feeding equation (Eq 1). X0 was set at 0.30 g (0.10 g l-1) based on the amount of biomass measured in the inoculum. The desired specific growth rate was 0.25 h-1. Ammonium and phosphate concentration was analysed and shown always in excess during the fermentation (data not shown), similar to previous nonanoic acid fed-batch studies (Sun et al. 2007a/Chapter 5). Nonanoic acid and glucose concentrations confirm that growth was carbon limited throughout the fermentation (Figure 6-1). Biomass first increased exponentially to 51 g l-1 due to exponential feeding, then increased linearly up to 71 g l-1 due to linear feeding. The culture initially contained very little PHA since the inoculum was grown on glucose and nutrient broth. The PHA content increased to 56% at the end of cultivation. Very soon after the start of linear feeding, dissolved oxygen dropped to zero although 1 vvm of pure oxygen was supplied. Feeding was maintained for another two hours before excessive foaming terminated the fermentation. This fermentation resulted in 39 g l-1 PHA, which gave a volumetric productivity of 1.44 g PHA l-1 h-1, with total nonanoic acid and glucose consumption of 178 g and 177 g, respectively.

111

Cell and PHA concentration (g l ) and PHA content (%)

10 70 60 50 40 30 20 2 10 0 5 10 15 20 25 30 0 4 Biomass PHA % PHA NA G DO

200

NA and G concentration (g l-1)

150

100

50

Culture time (h)

Figure 6-1 Single-stage MCL-PHA production by P. putida KT2440 using nonanoic acid (NA) and glucose (G) co-feeding strategy with NA:G= 1:1 (w/w). Exponential feeding stopped and linear feeding (constant feed rate of 17 g l-1 h-1) began at 24.9 h. Only DO data around the occurance of oxygen limitation (25 h) are shown, before which DO was always maintained around 40% air saturation.

To further increase the yield of PHA from nonanoic acid, the ratio of nonanoic acid to glucose was decreased. At NA:G= 1:1.5 (w/w), the feeding equation values of rNA= 40% and rG= 60%, YX/S was lowered to 0.58 g g-1, again based on YX/NA= 0.83 g g-1 and YX/G= 0.41 g g-1. X0 was set as 0.29 g (0.097 g l-1) based on the biomass of the inoculum while the desired specific growth rate remained as 0.25 h-1. A slightly higher yield of PHA from nonanoic acid was achieved (Figure 6-4b), while the final biomass, PHA content, and PHA productivity were all lower than the NA:G= 1:1 fermentation (Figure 6-2). Key results of both fermentations are summarized in Table 6-1. Data for MCL-PHA
112

Dissolved Oxygen (% air saturation)

-1

production from nonanoic acid with no glucose are taken from a previous study, at a specific growth rate of 0.25 h-1 (Sun et al. 2007a/Chapter 5).
Cell and PHA concentration (g l-1) and PHA content (%)
10 70 60 50 40 30 20 2 10 0 5 10 15 20 25 30 0 4 Biomass PHA % PHA NA G 8

Culture time (h)

Figure 6-2 Single-stage MCL-PHA production by P. putida KT2440 using nonanoic acid (NA) and glucose (G) co-feeding strategy with NA:G= 1:1.5 (w/w). Exponential feeding was applied throughout the fermentation.

6.4.2 Monomeric compositions during co-substrate cultivation

Analysis of the NA:G= 1:1.5 fermentation showed that co-feeding this amount of glucose had no significant effect on the monomeric composition (Figure 6-3b). The major components were still 3-OH-nonanoate (about 65%), 3-OH-heptanoate (about 33%) and 3-OH-valerate (about 1.5%). There was only about 1% of 3-OH-octanoate, 3-OH-hexanoate and 3-OH-decanoate combined. A similar percentage of these C-even repeating units were detected when only nonanoic acid was fed (Sun et al. 2007a/Chapter
113

NA and G concentration (g l )

-1

5). Therefore these were not likely derived from glucose. The ratio of C9 to C7 repeating units was slightly increased by co-feeding glucose when compared to nonanoic acid alone (from 1.90 mol mol-1 to 2.10 mol mol-1) (Figure 6-3a).
180 160

(a)

C9 molar concentration (mmol l-1)

140 120

80

60 100 80 40 60 40 20 0 0 20 40 60 80 100
-1

NA:G=1:1.5 m=2.10 mol mol-1 NA only m=1.90 mol mol-1

20

0 10 15 20 25

C7 molar concentration (mmol l )

Culture time (h)

Figure 6-3 (a) 3-OH-nonanoate (C9) to 3-OH-heptanoate (C7) ratio from NA:G= 1:1.5 co-feeding fermentation and NA sole feeding fermentation. (b) PHA composition throughout the NA:G= 1:1.5 fermentation. C5, 3-OH-valerate; C7, 3-OH-heptanoate; C9, 3-OH-nonanoate; C-even, total of mainly 3-OH-hexanoate, 3-OH-octanoate and 3-OH-decanoate. m is the slope of the corresponding regression line.

6.4.3 Biomass and PHA Yield

When a simple exponential substrate feeding strategy, such as Eq. (1), is applied, it is assumed that the biomass yield (YX/S) is accurate and constant. For the NA:G= 1:1 and NA:G= 1:1.5 fermentations, the YX/C were calculated to be 0.62 g g-1 and 0.58 g g-1, respectively, based on previous research. After plotting the total biomass versus total carbon substrates consumption obtained from the two glucose-nonanoic acid co-feeding fermentations (Figure 6-4a), it is clear that the yield was fairly constant throughout the
114

Monomeric composition (mol %)

C5 C7 C9 C-even

(b)

100

fermentation. The slopes of the regression curves are very close to what was predicted based on YX/NA= 0.83 g g-1 and YX/G= 0.41 g g-1. Therefore, Eq. 2 proved to be a reasonable way to estimate the biomass yield from the carbon substrates under co-substrate conditions, when the yield from each single substrate has been predetermined.

Total PHA accumulation at each sampling point was also plotted against cumulative nonanoic acid consumption, using data from three fermentations (Figure 6-4b). Linear regression was applied to each series of data and forced through origin, and the slope of
average each regression curve was considered as the average nonanoic acid to PHA yield ( YPHA / NA

). Comparing the results from the NA only feeding fermentation and the NA:G= 1:1
average fermentation, YPHA / NA increased by 25% (0.53 g g-1 to 0.66 g g-1). Further increase in average glucose feeding portion (NA:G= 1:1.5) resulted in slightly higher YPHA / NA , which was

0.69 g g-1 and about 30% increase from using nonanoic acid alone.

115

250 NA:G=1:1 m=0.63 g g-1 NA:G=1:1.5 m=0.59 g g-1

200

150

80 60 40

100

50 20

(a)
0
0 0 0 0 0 0 50 0 0 0 10 20 30 40 10 15 20

(b)
0
25 0

Total carbon substrates consumption (g)

NA consumption (g)

Figure 6-4 (a) Substrate to biomass yield. (b) nonanoic acid (NA) to PHA yield plots. m is the slope of the corresponding linear regression line. All linear regression curves were forced through origin.

116

PHA accumulation (g)

NA only -1 m=0.53 g g NA:G=1:1 m=0.66 g g-1 NA:G=1:1.5 m=0.69 g g-1

140 120 100

Total biomass (g)

Table 6-1 Summary of fermentation results.

Fermentation Final X (g l-1) 56.0 70.7 64.2 Final PHA (g l-1) 37.4 39.4 30.8 Final PHA content (%) 67 56 48 PHA overall productivity (g l-1 h-1) 1.44 1.44 1.16 Total NA consumption (g) 231 178 137 Total G consumption (g) 0 177 217 Average YPHA/NA (g g-1) 0.53 0.66 0.69 Final Overall YPHA/NA (g g-1) 0.55 0.69 0.71 Reference

NA only NA:G=1:1 NA:G=1:1.5

Sun et al. 2007a This work This work

Table 6-2 Summary of literature MCL-PHA production and yield

Microorganism P. putida IFO14464 P. putida BM01 P. putida BM01 P. putida IPT046 P. putida GPo1 P. putida KT2442 P. putida KT2440 P. putida KT2440 Fermentation Two-stage batch Two-stage fed-batch Two-stage fed-batch Two-stage fed-batch Two-stage continuous Continuous Single-stage Single-stage Notes 2nd stage N limited Octanoate as sole substrate 1st stage glucose, 2nd stage octanoate, 2nd stage N limited 2nd stage P limited 2nd-stage N limited, estimated yield N limited, estimated yield C limited C limited, co-feeding X (g l-1) 4.8 41.7 54.8 50 18 30 56 71 Final PHA content (%) 18 60 65.5 63 63 23 67 56 PHA overall productivity (g l-1 h-1) 0.03 0.71 0.92 0.80 1.06 0.69 1.44 1.44 Major YPHA/C (g g-1) 0.17 0.28 0.40 0.19 0.63 0.15 0.55 0.69 Major substrate Octanoate Octanoate Octanoate Glucose Octane Oleic acid Nonanoic acid Nonanoic acid Supplementary substrate N/A N/A Glucose Fructose N/A N/A N/A Glucose Reference (HORI et al. 1994) (Kim et al. 1997) (Kim et al. 1997) (Diniz et al. 2004) (Hazenberg and Witholt 1997) (Huijberts and Eggink 1996) (Sun et al. 2007a) This work

117

6.5 Disucssion
6.5.1 High yield and PHA productivity by co-substrate carbon-limited fermentation
Based on computer simulation of industrial scale PHA production processes, substrate accounts for about 30% of the total MCL-PHA production cost if an MCL carboxylic acid such as octanoate is used (deKoning et al. 1997) and, at least in the case of SCL-PHA, this increases as the scale of production increases (Choi and Lee, 1997). Therefore, it is highly desirable to use a low cost carbon source, and/or increase the carbon substrate to PHA yield for more economical MCL-PHA production. The present study demonstrates that these objectives can be achieved by using glucose to supplement nonanoic acid feeding in single-stage carbon-limited fed-batch MCL-PHA fermentations. Assuming the product contains 65 molar percent of 3-OH-nanonoate and 35 percent of 3-OH-heptanoate, the maximum yield of MCL-PHA from nonanoic acid is 0.93 g g-1. Using nonanoic acid as the sole substrate, 56 g l-1 biomass containing 67% MCL-PHA could be obtained by the carbon-limited single-stage fed-batch fermentation (Sun et al.
average 2007a/Chapter 5). However, the YPHA / NA was only 0.53 g g-1 (Figure 6-4b), as a large

portion of the nonanoic acid was used for synthesis of residual biomass (total biomass minus PHA). As the NA:G= 1:1 fermentation results demonstrated (Figure 6-1), nonanoic
average acid and glucose could be consumed simultaneously by P. putida, and the YPHA / NA was

enhanced by 25% (0.53 to 0.66 g g-1) by the co-feeding of glucose. An important

118

difference between NA+G co-feeding during a nitrogen-limited second stage (the standard method of producing PHA) and co-feeding during exponential growth, is the effect of glucose on the PHA composition. While during the nitrogen-limited stage glucose yielded 3-OH-decanoate and 3-OH-octanoate accounting for 10 molar percent of the total PHA synthesized (Sun et al. 2007c/Chapter 4), it yielded an insignificant amount of C-even (i.e. glucose derived) repeating units during C-limited exponential fed batch. Similar results have been reported in other studies (Diniz et al. 2004; Kim et al. 1996; Sun et al. 2007a/Chapter 5), where little PHA could be synthesized during the growth phase of various strains of P. putida when glucose was the sole substrate. Therefore, when the second energy source (glucose) was supplied, the -oxidation pathway, which is the energy generating pathway when fatty acid was sole substrate, was inhibited, leading to more nonanoic acid being diverted to MCL-PHA synthesis. As a consequence, the C9 to C7 ratio during the co-feeding also slightly increased when compared to the ratio obtained when only NA is fed (Figure 6-3a).

When the nonanoic acid to glucose ratio was further decreased (NA:G= 1:1.5 fermentation), it was hoped that glucose would provide enough carbon source for all Xr production, leading to the theoretical maximum nonanoic acid to PHA conversion (0.93 g
average PHA g-1 NA). However, the YPHA / NA increased only slightly (0.69 g g-1, Figure 6-4b),

and only 47% PHA was accumulated at the end of the fermentation. Apparently, simply
119

supplying glucose as the second substrate cannot completely inhibit -oxidation and allow the diversion of all nonanoic acid to PHA. The decreased nonanoic acid to glucose ratio likely resulted in an insufficient supply of nonanoic acid for PHA synthesis, probably due to a constant flux of acetyl-CoA from -oxidation toward cell growth, and led to a lower final PHA content and lower final PHA productivity (Table 6-1). Thus, the NA:G= 1:1 co-substrate ratio resulted in better production of MCL-PHA with similar composition (in terms of C9 and C7 molar portion among the total) and productivity (1.44 g PHA l-1 h-1) when compared to fermentations using a only nonanoic acid (Sun et al.
average 2007a/Chapter 5), but with significant enhancement in YPHA / NA .

The purpose of linear (i.e. constant rate) feeding after biomass reached above 50 g l-1 during the NA:G= 1:1 fermentation was to avoid oxygen limitation and achieve higher biomass, hence the rate of oxygen consumption should be constant when the rate of biomass synthesis (dX/dt) was constant. Indeed, a slightly higher final biomass (71 g l-1 versus 64 g l-1) was achieved when the result of this fermentation is compared to those fermentations where only exponential feeding (NA feeding and NA:G= 1:1.5 feeding fermentations) was employed. However, in all cases, oxygen limitation eventually led to foaming resulting in the termination of the fermentation. Foaming seems to be less of a problem for MCL-PHA fermentations using oleic acid (Lee et al. 2000), probably due to the antifoaming properties of oleic acid.
120

6.5.2 Yield potential of MCL-PHA synthesis

MCL-PHA synthesis yield data from the literature are summarized in Table 6-2. Clearly, single-stage, carbon-limited co-substrate fermentation resulted in the highest substrate to
overall PHA yield. If YPHA / NA is defined as PHA synthesized per gram nonanoic acid consumed overall at a given point of the fermentation, the relation between YPHA / NA and PHA content

during fed-batch fermentations may be studied as in Figure 6-5. It is obvious that the
overall YPHA / NA increases with PHA content during a fermentation. To study the theoretical overall overall maximum YPHA / NA , the experimental YPHA / NA data in Figure 6-5 are extrapolated with

respect to the PHA content. For nonanoic acid, single substrate feeding, the maximum yield (0.93 g PHA g-1 NA) can be achieved only at 100% PHA content, while for NA+G co-feeding cultivation the maximum yield can theoretically be achieved at 70% PHA, which should be much easier to obtain. A similar relation between overall PHA yield and the PHA content was calculated by Hazenberg and Witholt (1997). The above analysis, as well as the simplicity of the process itself, indicates that this NA+G co-feeding process is the most practical way to achieve cost-effective MCL-PHA production.

Acknowledgement This project was supported by the Natural Science and Engineering Research Council of Canada (NSERC) and BIOCAP Canada.

121

1.0

.8

YieldoverallPHA/NA (g g-1)

.6

.4

.2

NA:G=1:1.5 NA:G=1:1 NA only

0.0 0 20 40 60 80 100

PHA content (%)

Figure 6-5 The relationship between PHA content and the cumulative nonanoic acid to PHA yield (
overall overall YPHA / NA ). YPHA / NA is the cumulative PHA yield from nonanoic acid at a given sampling point.

6.6 References
Chen GQ, Zhang G, Park SJ, Lee SY (2001) Industrial scale production of poly(3-hydroxybutyrate-co-3-hydroxyhexanoate). Appl Microbiol Biotechnol 57:50-55 Choi JI, Lee SY (1997) Process analysis and economic evaluation for poly(3-hydroxybutyrate) production by fermentation. Bioprocess Eng 17:335-342 deKoning GJM, Kellerhals M, vanMeurs C, Witholt B (1997) A process for the recovery of poly(hydroxyalkanoates) from pseudomonads. 2. process development and economic evaluation. Bioprocess Eng 17:15-21 Diniz SC, Taciro MK, Gomez JG, da Cruz Pradella JG (2004) High-cell-density cultivation of Pseudomonas putida IPT 046 and medium-chain-length polyhydroxyalkanoate production from sugarcane carbohydrates. Appl Biochem Biotechnol 119:51-70
122

Hartmann R, Hany R, Pletscher E, Ritter A, Witholt B, Zinn M (2006) Tailor-made olefinic medium-chain-length poly[(R)-3-hydroxyalkanoates] by Pseudomonas putida GPo1: Batch versus chemostat production. Biotechnol Bioeng 93:737-746 Hazenberg W, Witholt B (1997) Efficient production of medium-chain-length poly(3-hydroxyalkanoates) from octane by Pseudomonas oleovorans: Economic considerations. Appl Microbiol Biotechnol 48:588-596 Hori K, Soga K, Doi Y (1994) Effects of culture conditions on molecular-weights of poly(3-hydroxyalkanoates) produced by Pseudomonas putida from octanoate. Biotechnol Lett 16:709-714 Huijberts GNM, and Eggink G (1996) Production of poly(3-hydroxyalkanoates) by Pseudomonas putida KT2442 in continuous cultures. Appl Microbiol Biotechnol 46:233-239 Jiang X, Ramsay JA, Ramsay BA (2006) Acetone extraction of mcl-PHA from Pseudomonas putida KT2440. J Microbiol Methods 67:212-219 Kim BS (2002) Production of medium chain length polyhydroxyalkanoates by fed-batch culture of Pseudomonas oleovorans Biotechnol Lett 24:125-130 Kim GJ, Lee IY, Choi DK, Yoon SC, Park YH (1996) High cell density cultivation of Pseudomonas putida BM01 using glucose. Journal of Microbiology and Biotechnology 6:221-224 Kim GJ, Lee IY, Yoon SC, Shin YC, Park YH (1997) Enhanced yield and a high production of medium-chain-length poly(3-hydroxyalkanoates) in a two-step fed-batch cultivation of Pseudomonas putida by combined use of glucose and octanoate. Enzyme Microb Technol 20:500-505 Lee SY, Choi JI ,Wong HH (1999) Recent advances in polyhydroxyalkanoate production by bacterial fermentation: Mini-review. Int J Biol Macromol 25:31-36 Lee SY, Wong HH, Choi JI, Lee SH, Lee SC, Han CS (2000) Production of medium-chain-length polyhydroxyalkanoates by high-cell-density cultivation of Pseudomonas putida under phosphorus limitation. Biotechnol Bioeng 68:466-470 Ramsay BA, Saracovan I, Ramsay JA, Marchessault RH (1991) Continuous production of long-side-chain poly-beta-hydroxyalkanoates by Pseudomonas oleovorans. Appl Environ Microbiol 57:625-629
123

Sun Z, Ramsay JA, Guay M, Ramsay BA (2006) Automated feeding strategies for high-cell-density fed-batch cultivation of Pseudomonas putida KT2440. Appl Microbiol Biotechnol 71:423-431 Sun Z, Ramsay JA, Guay M, Ramsay BA (2007a) Carbon-limited fed-batch production of medium-chain-length polyhydroxyalkanoates from nonanoic acid by Pseudomonas putida KT2440. Appl Microbiol Biotechnol 74:69-77 Sun Z, Ramsay JA, Guay M, Ramsay BA (2007b) Fermentation process development for the production of medium-chain-length poly-3-hydroxyalkanoates. Appl Microbiol Biotechnol 75:475-485 Sun Z, Ramsay JA, Guay M, Ramsay BA (2007c) Increasing the yield of MCL-PHA from nonanoic acid by co-feeding glucuose during the PHA accumulation stage in two-stage fed-batch fermentations of Pseudomonas putida KT2440. J Bacteriol DOI:101016/jbiotec200702023

124

CHAPTER 7

Fed-batch Production of Unsaturated Medium-chain-length Polyhydroxyalkanoates with Controlled Composition by Pseudomonas putida KT2440
Zhiyong Sun, Juliana A. Ramsay, Martin Guay, Bruce A. Ramsay

Originally submitted June, 2007 to Applied Microbiology and Biotechnology

125

7.1 Abstract
Unsaturated medium-chain-length polyhydroxyalkanoates (MCL-PHA) were produced at a productivity range of 0.63-1.09 g PHA l-1 h-1 with final PHA content ranging from 42.6% to 55.8%, by single-stage, carbon-limited fed-batch fermentations of Pseudomonas putida KT2440, using exponential feeding of nonanoic acid (NA) and 10-undecenoic acid (UDA=) mixture. Different specific growth rate (0.14 h-1 vs. 0.23 h-1) with similar substrate composition (NA:UDA== 5:1) had little effect on the final achievable PHA content and relative composition. Co-feeding of the two carbon sources resulted in PHA content plateau (about 55% when NA:UDA== 5.07:1, and 42% when NA:UDA== 2.47:1), which was not found from previous nonanoic acid only fermentations. The relative molar fraction of all 3-OH-alkanoates of the PHA product was fairly constant throughout each fermentation. Linear relationship between the unsaturation in PHA product and the unsaturated carbon source fraction in the feed was demonstrated, which makes it possible to produce unsaturated MCL-PHAs with controlled polymeric compositions.

7.2 Introduction
Medium-chain-length poly(3-hydroxyalkanoates) (MCL-PHAs) have attracted extensive interest, not only because of their biocompatibility and biodegradability, but also their diversity. The large variability of chemical structure and material properties (Hazer and

126

Steinbchel 2007) make them candidates for many applications, such as drug delivery (Pouton and Akhtar 1996) and tissue engineering (Chen and Wu 2005). Most MCL-PHAs consist solely of repeating units of 3-hydroxyalkanoates, but MCL-PHAs containing functional groups in the side chain, often referred to as functionalized PHAs (Hazer and Steinbchel 2007), can also be synthesized. These functional groups, such as carbon-carbon double bonds (Fritzsche et al. 1990), triple bonds (Kim et al. 1998), halogens (Kim et al. 1996), and phenyl groups (Hazer et al. 1996), are usually incorporated by appropriate microorganisms grown on suitable carbon source(s) containing the corresponding functional group(s). To date, however, few such functionalized PHAs have been produced at even a gram scale possibly due to the toxicity of the functional carbon substrates, inefficiency in substrate utilization by the cultures employed, and/or unavailability of such substrates, as many of them require dedicated chemical synthesis. Among various functionalized PHAs, those containing olefinic groups in the side chains, i.e. unsaturated PHAs, are likely one of the most useful functional PHAs (Hazer and Steinbchel 2007) due to the possibility of chemical modifications on or via the unsaturated sites, such as chlorination (Arkin et al. 2000), cross-linking (Dufresne et al. 2001), carboxylation (Kurth et al. 2002) and epoxidation (Park et al. 1998). Unsaturated PHAs are typically soft, sticky materials. Since the melting and the glass transition temperature vary according to the fraction of unsaturated components (Kim et al. 1995), and since different applications have different
127

requirements, the degree of unsaturation needs to be finely controlled in order to obtain materials with desired properties.

Most studies on the synthesis of MCL-PHA employed strains of Pseudomonas (Sun et al. 2007b/Chapter 2). Edible oils (Ashby and Foglia 1998) and carbohydrates (Sanchez et al. 2003) have been shown to support the synthesis of MCL-PHAs with unsaturated fractions, but the degree of unsaturation is difficult to be controlled using those carbon sources. However, when mixtures of saturated and unsaturated aliphatic acids were used as the carbon sources, a clear relationship between fraction of the unsaturated carbon source in the medium and the unsaturated fraction in the PHA was demonstrated (Kim et al. 1995; Kellerhals et al. 1999; Hartmann et al. 2006). Unfortunately, the highest volumetric productivity of unsaturated PHAs reported was only 0.20 g PHA l-1 h-1 (Kellerhals et al. 1999, Table 7-1), which is much less than that of saturated MCL-PHAs (Sun et al. 2007b/Chapter 2).

We have recently reported efficient MCL-PHA production by single-stage, carbon-limited, fed-batch fermentation using a nonanoic acid exponential feeding strategy (Sun et al. 2007a/Chapter 5). It was clearly demonstrated that nitrogen limitation was not necessary for MCL-PHA synthesis in Pseudomonas putida KT2440. In the current study, such a fed-batch technique was applied to the production of unsaturated MCL-PHAs and

128

to control the monomeric composition. The effect of specific growth rate on the PHA composition was also investigated.

7.3 Materials and methods

7.3.1 Microorganism and growth medium
Pseudomonas putida KT2440 (ATCC 47054) was maintained in lyophiles and on nutrient agar plates at 4 C. For all fermentations, 9 g l-1 of glucose plus 1 g l-1 of nutrient broth was used as the carbon sources in the inoculum medium. The initial fermentation medium contained about 0.66 g l-1 of the mixture of nonanoic acid (NA, 98%, Spectrum) and 10-undecenoic acid (UDA=, 97%, Sigma Aldrich) with a desired molar ratio (specified in the results section). Such substrate mixture was prepared prior to sterilization. Other medium components are reported by Sun et al. (2007a/Chapter 5). Nitrogen was supplied by using a 14% (w/v) NH4OH solution for pH control. To avoid initial precipitation of medium components, additional MgSO47H2O was fed at a ratio of 0.033 g MgSO47H2O g-1 carbon mixture added, assuming a YX/Mg of 240 g g-1 (Sun et al. 2006/Chapter 3) and a YX/S of 0.80 g g-1.

7.3.2 Fermentation conditions

Inocula were cultivated in 500-ml shake flasks (150 ml medium) at 301C and 200 rpm for about 12 h. Fed-batch fermentations were carried out at 281C with 3.0 l initial working volume in a Minifors 5 l stirred tank bioreactor (Infors HT, Bottmingen,
129

Switzerland). Data acquisition, pH control, dissolved oxygen control, and exit gas CO2 measurement were previously described (Sun et al. 2007a/Chapter 5). Substrate feeding using peristaltic pumps was automatically controlled based on the mass of the reservoir containing the substrate mixture. Antifoam 204 (Sigma Aldrich) was added drop wise manually through a sterile septa using an injection needle whenever required.

7.3.3 Substrate feeding and control methods

Three carbon-limited fed-batch fermentations, each at a controlled specific growth rate (), were carried out with different molar ratios of nonanoic acid to undecenoic acid as follows: (a) NA:UDA== 4.98:1 (molar ratio) and = 0.15 h-1; (b) NA:UDA == 5.07:1 and = 0.25 h-1; (c) NA:UDA== 2.47:1 and = 0.25 h-1. The initial substrate concentration was about 0.66 g l-1. To achieve the desired specific growth rate, substrate was then fed continuously to the culture according to the feed Eq. (1):

St =

X t X = 0 ( et 1) YX / C YX / C

(1)

Where St (g) is the total substrate required to produce biomass Xt (g) at cultivation time t (h); YX/C is the yield of biomass from substrate, assumed to be constant at 0.80 g g-1 based on a previous study (Sun et al. 2007a/Chapter 5); X0 (g) is the initial biomass, obtained by measuring the optical density of the inoculum; and (h-1) is the desired specific growth rate.
130

7.3.4 Analytical procedures

Cell dry weight of the lyophilized biomass was determined after centrifugation of 5 ml culture broth at 10,000 x g for 15 min and washing with distilled water. The supernatant of the centrifuged broth was saved for analysis of nonanoic acid (NA), undecenoic acid (UDA=), and other major nutrients. Phosphate and ammonium concentration were determined colorimetrically (Sun et al. 2007a/Chapter 5). Nonanoic acid and undecenoic acid concentration were determined by methylation followed by GC analysis (Ramsay et al. 1991), using benzoic acid as the internal standard. The procedures of methylation reaction for GC analysis of the cellular PHA content and composition were described in Sun et al. (2007a/Chapter 5). A gas chromatography (CP-3800, Varian Inc.) equipped with flame-ionization-detector was used in all GC analysis. The GC parameters were: injector temperature 250 C, detector temperature 275 C, 1 l injection, split ratio 10. The oven temperature profile was: 90 C, 0.5 min, 6 C min-1 to 96 C, 7 C min-1 to 131 C, 20 C min-1 to 181 C, 5 min. The MCL-PHA standard for the GC analysis was prepared as described by Jiang et al. (2006). The monomeric components were confirmed by GC-MS and composition determined by nuclear magnetic resonance (NMR). The SCL-PHA standard was Biopol, which was poly(3-hydroxybutyrate-co-3-hydroxyvalerate) (81 and 19 mol % of HB and HV, respectively), originally purchased from Imperial Chemical Industries (ICI). All analysis was done in duplicates with the average of the results presented in this paper.
131

7.4 Results
7.4.1 Production of unsaturated MCL-PHAs by exponential feeding of nonanoic acid and 10-undecenoic acid
A fed-batch fermentation (fermentation a) was conducted by exponentially feeding nonanoic acid (NA) and undecenoic acid (UDA=) at a molar ratio of 4.98:1. The desired was set at 0.15 h-1 in the feed equation (Eq. 1). As seen in Figure 7-1, the overall biomass increased exponentially, and the actual was determined to be 0.14 h-1 by nonlinear curve fitting (dashed line in Figure 7-1). Shortly after inoculation, the culture became carbon-limited, and remained carbon-limited throughout the balance of the fermentation until the 42 h point. At this time, cell growth became limited by oxygen, causing both nonanoic acid and undecenoic acid to accumulate rapidly in the fermentation broth. The other major nutrients, ammonium and phosphorus, were kept at concentrations very similar to those of a previous study (Sun et al. 2007a/Chapter 5). The inoculum culture (grown mainly on glucose) contained negligible PHA, but the PHA content increased to 55.6% of cell dry weight at the end of the fermentation, giving a cumulative unsaturated MCL-PHA productivity of 0.71 g PHA l-1 h-1 (Table 7-1). The PHA content achieved a plateau when the cell concentration reached above 20 g l-1.

132

Biomass (g l ), PHA (g l ) and PHA content (%)

60 Biomass % PHA PHA NA = UND Curve fit DO %

10

200

NA and UDA= concentration (g l-1)

50

150

40

30 4 20 2

100

50

10

0 0 10 20 30 40 50

Culture time (h)

Figure 7-1 Fermentation a, exponential feeding of nonanoic acid (NA) and 10-undecenoic acid (UDA=) (molar ratio 4.98:1) with desired = 0.15 h-1. The dissolved oxygen data are shown after 30 h, before when it was kept around 40% air saturation. Curve fitting (dashed line) was done by nonlinear regression function of Sigmaplot 9.01.

Fermentations with higher desired (0.25 h-1) were also conducted with different nonanoic acid to undecenoic acid ratio (fermentations b and c, with NA:UDA== 5.07:1 and 2.47:1, respectively) (Figure 7-2), to determine the effect of growth rate and substrate mixture on the PHA synthesis. The cell growth curves from both fermentations were very similar, indicating that the ratio of nonanoic acid and undecenoic acid had little effect on biomass yield. Using a similar nonanoic acid to undecenoic acid ratio (fermentation b), the PHA content also plateaued at around 55%, very similar to that of the = 0.14 h-1 fermentation, despite the much higher specific growth rate (0.23 h-1 vs. 0.14 h-1). The PHA overall productivity also increased to 1.09 g PHA l-1 h-1 due to the shortened
133

Dissolve oxygen (% air saturation)

-1

-1

fermentation time. However in fermentation c, the PHA content achieved the plateau only at around 42%. The final results of the three fermentations are summarized in Table 7-1, with some literature results as comparisons.
Biomass (g l-1), PHA (g l-1) and PHA content (%)
60 Biomass (5.07:1) Biomass (2.47:1) % PHA (5.07:1) % PHA (2.47:1) PHA (5.07:1) PHA (2.47:1)

50

40

30

20

10

0 5 10 15 20 25

Culture time (h)

Figure 7-2 Biomass, PHA concentration and PHA content of fermentations b and c, with the same (0.25 h-1) and different NA to UDA= ratio (5.07:1 and 2.47:1, respectively).

Table 7-1 Fermentation results summary.

Fermentation Actual NA:UDA= (molar ratio) 4.98:1 5.07:1 2.47:1 Varied 1:1 Actual (h-1) 0.14 0.23 0.24 Varied 0.1-0.3 Final X (g l-1) 54.10 48.12 33.58 30 1.11-1.44 Final PHA (g l-1) 30.07 26.86 14.31 10 0.13-0.47 Final PHA content (%) 55.59 55.82 42.61 33 12-37 PHA overall productivity (g l-1 h-1) 0.71 1.09 0.63 0.20 0.04-0.07 Reference

(a) (b) (c) Closed-loop fed-batch Continuous

This work This work This work Kellerhals et al. (1999) Hartmann et al. (2006)

134

7.4.2 Monomeric compositions

Throughout the fermentations, MCL-PHA containing various 3-OH-alkanoates with saturated or unsaturated side chains was synthesized. The relative molar fraction of saturated and unsaturated components remained fairly stable throughout the cultivation (Figure 7-3b, 3c). The approximate molar fractions of all detectable components are shown in Table 7-2. Despite the difference in nonanoic acid to undecenoic acid ratio in the feed or the actual cell growth rate, 3-OH-nonanoate was always the most abundant among the saturated components, while 3-OH-nonenoate was the most abundant among the unsaturated components. Interestingly, even carbon number 3-OH-alkanoates (3-OH-octanoate, 3-OH-decanoate, 3-OH-hexanoate, and 3-OH-butyrate) with a total of about 1 mol % were also detected, but only later in the fermentations.
Table 7-2 Molar fraction of all 3-OH-alkanoates detected in the MCL-PHA product from different fermentations.
Fermentation Actual NA:UDA= (molar ratio) 4.98:1 5.07:1 2.47:1
-1

C5

C7

C9

C11

C7:1

C9:1

C11:1

C-even

Saturated: unsaturated (molar ratio) 5.82 5.73 2.76

(a), = 0.14 h-1 (b), = 0.23 h (c), = 0.24 h

-1

1.5 1.6 3.0

29.5 30.0 25.0

53.0 52.5 44.5

0.5 0.5 0.6

1.0 1.0 2.2

7.2 7.2 13.3

6.3 6.3 10.8

1.0 0.9 0.6

C5, 3-OH-valerate, C7, 3-OH-heptanoate, C7:1, 3-OH-heptenoate, C9, 3-OH-nonanoate, C9:1, 3-OH-nonenoate, C11, 3-OH-undecanoate, C11:1, 3-OH-undecenoate, C-even, total of 3-OH-octanoate, 3-OH-decanoate, 3-OH-hexanoate, and 3-OH-butyrate (descending in relative fration).

Since the fractions of all components were relatively constant, the total molar amount of saturated repeating units versus the total molar of unsaturated ones from three fermentations were plotted in Figure 7-3a. The small amount of even carbon number
135

3-OH-alkanoates was not included in the saturated vs. unsaturated plots. The linear relationships demonstrate that the ratio of saturated to unsaturated components of the MCL-PHA remained constant throughout each fermentation (Table 7-2). They were always slightly higher than the corresponding ratios of nonanoic acid to undecenoic acid in the feed.
180
-1 4.98:1 at =0.14 h Slope=5.82 -1 5.07:1 at =0.23 h Slope=5.73 -1 2.47:1 at =0.24 h Slope=2.76

(b)

100% 80% 60% 40% 20% 0%

160

Total of saturated components (mmol)

140

120

100

80

(c)

100% 80%

60 C5 C7 C9 C11 C7:1 C9:1 C11:1 60% 40% 20% 0% 0 5 10 15 20 25 30 35 5 10 15 20 25

40

20

(a)
0

Total of unsaturated components (mmol)

Culture time (h)

Figure 7-3 (a) Linear relationship between total amount of saturated components and total amount of unsaturated components from three fermentations. (b) Relative monomeric composition of the MCL-PHA synthesized during fermentation b. (c) Relative monomeric composition of the MCL-PHA synthesized during fermentation c. The dash-dot line in (b) and (c) divides saturated and unsaturated components. C5, 3-OH-valerate, C7, 3-OH-heptanoate, C7:1, 3-OH-heptenoate, C9, 3-OH-nonanoate, C9:1, 3-OH-nonenoate, C11, 3-OH-undecanoate, C11:1, 3-OH-undecenoate. The small amount of even carbon number 3-OH-alkanoates is not included these plots.

136

PHA monomeric composition (mol%)

7.5 Discussion
7.5.1 Process design for unsaturated MCL-PHAs production
Although a highly promising material, unsaturated MCL-PHAs have not been produced at a scale comparable to saturated MCL-PHAs; and this could hamper the development of applications for these biopolyesters. In a closed-loop fed-batch process, unsaturated MCL-PHAs were produced up to 33% of the dry biomass (0.20 g PHAs l-1 h-1) by co-feeding of Na-octanoate and Na-undecenoate (Kellerhals et al. 1999) (Table 7-1). Since feeding of the two substrates was controlled individually, the unsaturated fraction varied throughout the cultivation, which is not ideal for the production of such MCL-PHAs with controlled properties. In another study (Hartmann et al. 2006), chemostat cultivation was found to be a suitable method to produce unsaturated MCL-PHAs with defined monomeric composition, while batch processes resulted in a slight change in the composition throughout the cultivation. However, the PHA productivity of such chemostat process was only within the 0.04-0.07 g PHA l-1 h-1 range.

Using a single stage, exponential feeding fed-batch fermentation process (Sun et al. 2007a/Chapter 5), we produced unsaturated MCL-PHAs at a productivity range of 0.63-1.09 g PHA l-1 h-1 with a final PHA content ranging from 42.6% to 55.8%, by feeding a mixture of saturated alkanoic acid (nonanoic acid), and unsaturated alkanoic acid (10-undecenoic acid). Both carbon sources are among the most common substrates
137

used to produce unsaturated MCL-PHAs (Sun et al. 2007b/Chapter 2), and are relatively inexpensive compared to other aliphatic carbon sources. Furthermore, these substrates mix very well with each other and are all miscible in the aqueous media, making the sterilization and feeding process much easier. The productivity and final PHA content are much higher than what has been reported in the literature, and the process does not involve many manual operations or expensive instrumentation. These features render this process relatively attractive for industrial application.

7.5.2 Effect of specific growth rate and varied substrate composition on PHA synthesis
Under very different specific growth rate (0.14 h-1 vs. 0.23 h-1) with similar substrate composition (NA:UDA= of 4.98:1 vs. 5.07:1), the final achievable PHA content and relative composition show little significant difference (Table 7-2).

However, the increasing fraction of unsaturated carbon substrate (undecenoic acid in this case) seems to have big impact on the final achievable PHA content. Reviewing the results of our previous report on saturated MCL-PHA production by nonanoic acid alone (Sun et al. 2007a/Chapter 5), the PHA content did not stop increasing before oxygen transfer limitation was encountered. It achieved 67% at the end of a = 0.25 h-1 fermentation and can potentially reach as high as 75% if there was sufficient oxygen transfer. However, during fermentation b of this study, when undecenoic acid was about

138

17% molar fraction in the substrate, a plateau of about 55% was attained before the end of the cultivation. When undecenoic acid fraction was further increased to 28% in fermentation c, the plateau occurred at only about 42%. This indicates that undecenoic acid is not a substrate as favorable as nonanoic acid to P. putida KT2440 for PHA synthesis, and maybe also inhibit cell growth. Similar observations were also reported by Kim et al. (1995).

7.5.3 Unsaturated MCL-PHAs with controllable compositions

As demonstrated in Figure 7-3, the ratio of saturated to unsaturated components of the PHAs stayed fairly constant throughout the fermentation, regardless of specific growth rate or substrate composition. It had been concluded (Hartmann et al. 2006) that batch cultivation was not ideal for achieving desired monomeric composition as a slight change over time was observed. This may have been due to unbalanced growth in batch cultivation, as the concentration of nutrients change with time. In fed-batch cultivation, such as what we presented in this study, a more balance growth and PHA synthesis is more likely to occur, which leads to a constant composition of the resulting PHAs.

When plotting the ratio of saturated to unsaturated components versus the ratio of saturated to unsaturated carbon source from two fermentations (b and c, Table 7-2), a straight line through the origin can be generated (r2= 0.99) (Figure 7-4). The slightly higher ratio of saturated components (compared to the ratio of saturated substrate in the
139

feed) indicates that nonanoic acid is slightly more favorable for PHA synthesis. This curve may be used to predict the monomeric composition during production of unsaturated MCL-PHAs. Since this ratio is very constant during the entire cultivation, unsaturated MCL-PHAs with a desired monomeric composition may be produced using P. putida KT2440 by adjusting the composition of the substrate according to Figure 7-4. The current process can also be considered as a model process, and likely be applied to produce other types of functionalized MCL-PHAs on a larger scale than those that have been previously reported.
6 5 4 3 2 1 0 0 1 2 3 4 5 6 Sat/Unsat ratio in the substrate
Figure 7-4 Relationship between unsaturation of the MCL-PHA product and the substrate for -1 2 fermentations at desired = 0.25h (fermentations b and c). The r is 0.99 for the regression curve.

Acknowledgement This project was supported by the Natural Science and Engineering Research Council of Canada (NSERC).

Sat/Unsat ratio in PHA

140

7.6 References
Arkin AH, Hazer B, Borcakli M (2000) Chlorination of poly(3-hydroxyalkanoates) containing unsaturated side chains. Macromolecules 33:3219-3223 Ashby RD, Foglia TA (1998) Poly(hydroxyalkanoate) biosynthesis from triglyceride substrates. Appl Microbiol Biotechnol 49:431-437 Chen GQ, Wu Q (2005) The application of polyhydroxyalkanoates as tissue engineering materials. Biomaterials 26:6565-6578 Dufresne A, Reche L, Marchessault RH, Lacroix M (2001) Gamma-ray crosslinking of poly (3-hydroxyoctanoate-co-undecenoate). Int J Biol Macromol 29:73-82 Fritzsche K, Lenz RW, Fuller RC (1990) Production of unsaturated polyesters by Pseudomonas oleovorans. Int J Biol Macromol 12:85-91 Hartmann R, Hany R, Pletscher E, Ritter A, Witholt B, Zinn M (2006) Tailor-made olefinic medium-chain-length poly[(R)-3-hydroxyalkanoates] by Pseudomonas putida GPo1: Batch versus chemostat production. Biotechnol Bioeng 93:737-746 Hazer B, Lenz RW, Fuller RC (1996) Bacterial production of poly-3-hydroxyalkanoates containing arylalkyl substituent groups. Polymer 37:5951-5957 Hazer B, Steinbchel A (2007) Increased diversification of polyhydroxyalkanoates by modification reactions for industrial and medical applications. Appl Microbiol Biotechnol 74:1-12 Jiang X, Ramsay JA, Ramsay BA (2006) Acetone extraction of mcl-PHA from Pseudomonas putida KT2440. J Microbiol Methods 67:212-219 Kellerhals MB, Kessler B, Witholt B (1999) Closed-loop control of bacterial high-cell-density fed-batch cultures: Production of mcl-PHAs by Pseudomonas putida KT2442 under single-substrate and cofeeding conditions. Biotechnol Bioeng 65:306-315 Kim DY, Kim YB, Rhee YH (1998) Bacterial poly(3-hydroxyalkanoates) bearing carbon-carbon triple bonds. Macromolecules 31:4760-4763 Kim O, Gross RA, Hammar WJ, Newmark RA (1996) Microbial synthesis of poly(beta-hydroxyalkanoates) containing fluorinated side-chain substituents. Macromolecules 29:4572-4581
141

Kim YB, Lenz RW, Fuller RC (1995) Poly-3-hydroxyalkanoates containing unsaturated repeating units produced by Pseudomonas oleovorans. J Polym Sci Pol Chem 33:1367-1374 Kurth N, Renard E, Brachet F, Robic D, Guerin P, Bourbouze R (2002) Poly(3-hydroxyoctanoate) containing pendant carboxylic groups for the preparation of nanoparticles aimed at drug transport and release. Polymer 43:1095-1101 Park WH, Lenz RW, Goodwin S (1998) Epoxidation of bacterial polyesters with unsaturated side chains. I. Production and epoxidation of polyesters from 10-undecenoic acid. Macromolecules 31:1480-1486 Pouton CW, Akhtar S (1996) Biosynthetic polyhydroxyalkanoates and their potential in drug delivery. Adv Drug Deliv Rev 18:133-162 Ramsay BA, Saracovan I, Ramsay JA, Marchessault RH (1991) Continuous production of long-side-chain poly-beta-hydroxyalkanoates by Pseudomonas oleovorans. Appl Environ Microbiol 57:625-629 Sanchez RJ, Schripsema J, da Silva LF, Taciro MK, Pradella JGC, Gomez JGC (2003) Medium-chain-length polyhydroxyalkanoic acids (PHA(mcl)) produced by Pseudomonas putida IPT 046 from renewable sources. Eur Polym J 39:1385-1394 Sun Z, Ramsay JA, Guay M, Ramsay BA (2007a) Carbon-limited fed-batch production of medium-chain-length polyhydroxyalkanoates from nonanoic acid by Pseudomonas putida KT2440. Appl Microbiol Biotechnol 74:69-77 Sun Z, Ramsay JA, Guay M, Ramsay BA (2007b) Fermentation process development for the production of medium-chain-length poly-3-hydroxyalkanoates. Appl Microbiol Biotechnol 75: 475-485 Sun Z, Ramsay JA, Guay M, Ramsay BA (2006) Automated feeding strategies for high-cell-density fed-batch cultivation of Pseudomonas putida KT2440. Appl Microbiol Biotechnol 71:423-431

142

CHAPTER 8 Conclusions 8.1 Summary and contributions

8.1.1 High-cell-density cultivation of P. putida KT2440
High-cell-density is often required to improve biomass and product formation during microbial fermentations. With industrial applications in mind, such fermentation processes should be automated, easily controlled, and have high biomass productivity. The fermentation processes developed for P. putida KT2440 cultivation using four different automated substrate feeding strategies (Chapter 3) fulfill all these requirements. The highest biomass productivity (4.3 g X l-1 h-1) was achieved by continuous feeding based on real-time glucose consumption estimation, since during such cultivation the substrate requirements of the culture can be continuously and fully satisfied. This substrate feeding strategy is ideal for the growth stage (first stage) of a conventional two-stage PHA production process in order to achieve high productivity with a short cultivation time (Chapter 4). The exponential substrate feeding strategy proved to be an efficient and effective method to achieve a desired specific growth rate (, must be lower than the max of the microorganism). This feeding strategy is ideal to achieve carbon-limited and controlled growth, so that the toxic effects of substrate(s) can be virtually eliminated and the effect of specific growth rate on product formation can be revealed (Chapter 5, 6, 7). The high-cell-density cultivation strategies can not only benefit
143

the cultivation of P. putida and the production of MCL-PHA, but may also be used as valuable reference for the production of other microorganisms and microbial products.

8.1.2 Carbon-limited, single-stage MCL-PHA production

Despite the fact that most literature studies of MCL-PHA production employed either nitrogen or phosphorus limitation, carbon-limited (N and P in excess), single-stage, exponential fed fermentations were found to be highly suitable for MCL-PHA production by P. putida KT2440 (Chapter 5). Partially due to its high toxicity, nonanoic acid has never been used for the production of MCL-PHA except in shake-flasks. In this study, the toxicity was virtually eliminated by carbon-limitation so that no complicated instrumentation was required for process control. Full automation can be easily achieved by this approach. A PHA content of 64% was obtained by the end of a = 0.25 h-1 fermentation, while 75% PHA (highest MCL-PHA content ever shown in a fermentation process) was achieved when the oxygen demand was lowered by lowering (0.15 h-1). Therefore, MCL-PHA synthesis and content increase is likely to be limited only by oxygen transfer when using nonanoic acid alone. Such a process has the potential of even higher overall PHA productivity and content when fermentation systems with better oxygen transfer capacity are available. For a given system, the choice of specific growth rate has to consider the trade-off between achievable final PHA content and final PHA productivity such that the overall PHA productivity is satisfactory while final PHA
144

content is high enough for efficient separation of PHA from biomass. Since P. putida KT2440 has a wide substrate range, similar aliphatic substrates, such octanoic acid, heptanoic acid, and decanoic acid, are also expected to be suitable for the exponential feeding process. Therefore MCL-PHAs with different monomeric compositions can be produced.

8.1.3 Co-substrate feeding strategy for functionalized MCL-PHA production

Functionalized MCL-PHAs are usually more difficult to produce due to a low yield of product from the substrate carrying the functional group, and the lack of availability of such substrates. Using nonanoic acid with 10-undecenoic acid as the co-substrate (Chapter 7), not only was unsaturated MCL-PHA be produced at a productivity (1.09 l-1 h-1), much higher than literature reports, but the composition of the PHA product was controlled by the controlling the substrate ratio. Such a strategy can be applied to the production of other types of functionalized MCL-PHAs with controlled compositions, with other combinations of major substrates (nonanoic acid or other MCL carboxylic acids) and co-substrates carrying functional groups.

8.1.4 Yield enhancement and substrate cost reduction

Carbohydrates are among the least expensive substrates for microbial fermentations. In the synthesis of MCL-PHA by P. putida KT2440, glucose yielded substantial biomass but insignificant PHA. To take advantage of the low cost of glucose and the high PHA yield
145

efficiency of nonanoic acid, co-feeding strategies were developed in a conventional two-stage MCL-PHA production process (Chapter 4) and a carbon-limited, single-stage MCL-PHA production process (Chapter 6). Both studies demonstrated that a substantial nonanoic acid to PHA yield enhancement can be achieved by using glucose as the co-substrate. In the two-stage production process (Chapter 4), the YPHA/NA was almost doubled while the composition of the PHA product changed only slightly comparing to feeding only nonanoic acid in the same process. In the single-stage production process (Chapter 6), the YPHA/NA was increased up to 30% with no significant change in the PHA composition and an overall PHA productivity (1.44 g PHA l-1 h-1) was achieved (Chapter 5). Since co-feeding of glucose does not alter the composition of the PHA product, such a co-feeding strategy can be applied when other aliphatic substrates are to be used for MCL-PHA synthesis in order to enhance the yield of those substrates and lower the total substrate cost. The more expensive the substrate is (10-undecnoic acid for example), the more benefit it will be to use carbohydrate as a co-substrate.

8.2 Recommendations for future work

To further optimize the carbon-limited, single stage MCL-PHA production process developed in this study, the mechanisms governing the MCL-PHA synthesis in P. putida KT2440 should be studied in depth. Based on the fermentation results presented in this study (Chapter 5), it is clear that the mechanisms are different from those of typical
146

SCL-PHA-synthesizing microorganisms. Two types of SCL-PHA synthesis are well known, non-growth associated and growth associated. Non-growth associated SCL-PHA-synthesizing microorganisms, such as Ralstonia eutropha, accumulate PHA only (or mostly) during a non-growth phase when one or more nutrients (e.g. nitrogen or phosphorus) becomes limiting while carbon source is in excess. Growth-associated SCL-PHA-synthesizing microorganisms, such as Alcaligenes latus, synthesize PHA during growth phase with a constant PHA content and specific PHA synthesis rate. However, P. putida KT2440 synthesizes MCL-PHA at a rate that is higher than the specific growth rate of the residual biomass, resulting in increasing PHA content throughout the growth phase. A deficiency in the TCA cycle, blockage in -oxidation, a substrate detoxicification mechanism, and properties of the PHA granules should all be investigated as possible mechanisms controlling such PHA synthesis.

The oxygen transfer capacity of the bioreactor was found to limit the achievable PHA productivity. Approaches to increase the oxygen transfer of a given fermentation system are therefore necessary, such as pressurized bioreactors or by employing an oxygen carrier compound such as dodecane. Alternatively, other modes of cultivation should be considered. Continuous fermentation running at high-cell-density may be a feasible way of achieving higher volumetric PHA productivity. For example, if a steady state can be reached at cell concentration of 20 g l-1 containing 50% PHA at a dilution rate of 0.2 h-1,
147

the volumetric productivity would be 2 g PHA l-1 h-1. Optimization of such a continuous process may result in even higher PHA content and productivity.

Unfortunately, continuous culture is rarely used for industrial scale production since contamination is virtually impossible to eliminate. The possibility of using a decaying exponential fed-batch process should, therefore, also be examined. In this approach the controlled specific growth rate is gradually lowered throughout the fermentation by carbon-limitation until the oxygen transfer capacity of the bioreactor is nearly attained. At this point the feed rate can be maintained at a constant value or gradually lowered (kLa is known to decrease with culture density). In theory, such a feeding regime should give the maximum productivity obtainable in fed-batch mode fermentation. A preliminary result (not presented in this thesis) demonstrated that the final achievable biomass can be as high as 90 g l-1, a near 30% increase comparing to the highest biomass concentration (70 g l-1) presented in this thesis.

148