BIOPROCESSING STRATEGIES FOR THE CULTIVATION OF …

BIOPROCESSING STRATEGIES FOR THE CULTIVATION

OF OLEAGINOUS YEASTS ON GLYCEROL

A thesis submitted to The University of Manchester for the degree of Doctor of Philosophy

in the Faculty of Science and Engineering

2016

ELENI KARAMEROU

School of Chemical Engineering and Analytical Science

The University of Manchester, UK

PhD thesis Bioprocessing strategies for the cultivation of oleaginous yeasts on glycerol

2

Table of Contents

3

Table of Contents

Abbreviations and nomenclature ................................................................................................. 7

Abstract .................................................................................................................................................... 9

Declaration ........................................................................................................................................... 10

Copyright statement .......................................................................................................................... 10

Acknowledgements ........................................................................................................................... 11

The author ............................................................................................................................................ 12

Chapter 1 ............................................................................................................................................... 13

1 Microbial oil as the basis for a sustainable society ........................................................ 14

1.1 Introduction ....................................................................................................................................... 14

1.2 Why do we need more oil and alternative oil sources? ..................................................... 15

1.3 Industrialisation possibilities of microbial oil ...................................................................... 17

1.4 Structure of the thesis .................................................................................................................... 18

Chapter 2 ............................................................................................................................................... 21

2 Microbial oil production from glycerol: a review of the literature .......................... 22

2.1 Introduction ....................................................................................................................................... 22

2.2 Microbial oil and its applications ............................................................................................... 22

2.3 Oleaginous microorganisms ......................................................................................................... 26

2.3.1 Oleaginous microalgae .......................................................................................................... 27

2.3.2 Oleaginous bacteria ................................................................................................................ 27

2.3.3 Oleaginous fungi....................................................................................................................... 28

2.4 The mechanism of lipid accumulation in yeasts .................................................................. 29

2.5 Factors affecting lipid accumulation in yeast species ........................................................ 32

2.5.1 Growth elements ...................................................................................................................... 32

2.5.2 Culture conditions ................................................................................................................... 34

2.6 The role of oxygen in the metabolism of oleaginous yeasts ............................................ 34

2.6.1 Oxygen transfer into the cell ............................................................................................... 35

2.6.2 The effect of oxygen in oleaginous yeasts ...................................................................... 36

2.7 Conversion of microbial oil into biodiesel .............................................................................. 37

2.8 Industrialisation of microbial oil ................................................................................................ 38

2.9 Process improvements for microbial oil production ......................................................... 41

2.9.1 Low-cost fermentation substrates .................................................................................... 41

2.9.2 Developing cultivation modes for efficient microbial oil yield .............................. 47

2.10 Robust cultivation conditions ................................................................................................. 51

2.11 By-products from oil production ........................................................................................... 52

2.12 Concluding remarks .................................................................................................................... 52

Chapter 3 ............................................................................................................................................... 55

3 Research objectives and experimental programme ...................................................... 56

3.1 Key objectives of the project ........................................................................................................ 56

3.2 Experimental programme ............................................................................................................. 57

3.2.1 Strain selection ......................................................................................................................... 58

3.2.2 Growth of oleaginous yeast Rh. glutinis on glycerol ................................................... 58

3.2.3 Model development ................................................................................................................ 58

3.2.4 Bioreactor cultivations for mode of operation development ................................. 59

Chapter 4 ............................................................................................................................................... 61

4 Materials and methods ............................................................................................................. 62

4.1 Introduction ....................................................................................................................................... 62


4

4.2 Microorganism ...................................................................................................................................62

4.3 Inoculum preparation .....................................................................................................................62

4.4 Analytical methods ...........................................................................................................................63

4.4.1 Cellular growth ..........................................................................................................................63

4.4.2 HPLC analysis .............................................................................................................................63

4.4.3 Total Nitrogen ............................................................................................................................63

4.4.4 Oil content (by extraction) ....................................................................................................64

4.4.5 Microscopic observation of cells ........................................................................................64

4.4.6 Calculation of cultivation parameters: Carbon to nitrogen ratio ...........................65

Chapter 5 ................................................................................................................................................ 67

5 Selecting a yeast based on potential growth and oil accumulation on glycerol ... 68

5.1 Introduction ........................................................................................................................................68

5.2 Theoretical Background .................................................................................................................68

5.2.1 Oleaginous yeasts .....................................................................................................................68

5.2.2 Observing intracellular lipids: Sudan Black B staining ..............................................70

5.3 Experimental design ........................................................................................................................70

5.4 Methodology .......................................................................................................................................71

5.4.1 Strain maintenance ..................................................................................................................71

5.4.2 Cultivation of yeasts on both glucose and glycerol .....................................................71

5.4.3 Sudan Black B staining experiments .................................................................................71

5.4.4 Cultivation of yeasts on glycerol .........................................................................................72

5.4.5 Analytical methods ..................................................................................................................72

5.5 Results ...................................................................................................................................................74

5.5.1 Morphology observation study ...........................................................................................74

5.5.2 Cultivation of oleaginous yeasts on both glucose and glycerol ..............................75

5.5.3 Observation of lipids using Sudan Black staining ........................................................77

5.5.4 Evaluation of growth on different concentrations of glycerol in batch mode ..78

5.6 Conclusions ..........................................................................................................................................83

Chapter 6 ................................................................................................................................................ 85

6 Growth aspects of Rhodotorula glutinis on glycerol based media ............................. 86

6.1 Introduction ........................................................................................................................................86

6.2 Theoretical background..................................................................................................................86

6.3 Methodology .......................................................................................................................................88

6.3.1 Experiments for defining growth conditions ................................................................88

6.3.2 Growth and lipid production experiments .....................................................................89

6.3.3 Kinetics of microbial oil production ..................................................................................90

6.4 Results ...................................................................................................................................................92

6.4.1 Experiments for defining growth conditions ................................................................92

6.4.2 Growth and lipid production experiments .....................................................................96

6.4.3 Kinetics of microbial oil production ............................................................................... 100

6.4.4 Growth on synthetic crude glycerol ............................................................................... 106

6.5 Conclusions ....................................................................................................................................... 112

Chapter 7 ............................................................................................................................................. 113

7 A biorefinery approach to microbial oil production from glycerol by Rhodotorula glutinis ................................................................................................................................................. 114

7.1 Introduction ..................................................................................................................................... 114

7.2 Theoretical background............................................................................................................... 114

7.2.1 Calculation of the oxygen uptake rate ........................................................................... 114

Table of Contents

5

7.3 Methodology ..................................................................................................................................... 116

7.3.1 Microorganism and media ................................................................................................. 116

7.3.2 Shake-flask experiments ..................................................................................................... 116

7.3.3 Batch bioreactor experiments .......................................................................................... 117

7.3.4 Fed-batch bioreactor experiments ................................................................................. 118

7.3.5 Specific growth rate .............................................................................................................. 119

7.3.6 Specific substrate uptake rate .......................................................................................... 119

7.3.7 Analytical methods ............................................................................................................... 119

7.4 Results and discussion ................................................................................................................. 119

7.4.1 Effect of glycerol concentration on growth and oil accumulation ...................... 119

7.4.2 Effect of initial nitrogen concentration on the specific growth rate and oil accumulation............................................................................................................................................. 123

7.4.3 Effect of air on growth and oil production ................................................................... 125

7.4.4 The effect of feeding strategies on growth, oil yield and glycerol consumption of Rh. glutinis ............................................................................................................................................. 127

7.4.5 Oxygen consumption during fermentation in 2-L bioreactor .............................. 131

7.4.6 Lipid accumulation over time and the ability of Rh. glutinis to grow and produce oil under acidic conditions ................................................................................................ 132

7.5 Conclusions ....................................................................................................................................... 133

Supplementary information for Chapter 7 ........................................................................................ 135

Chapter 8 ............................................................................................................................................. 143

8 Developing an unstructured model to describe batch cultivations of Rhodotorula glutinis .................................................................................................................................................. 144

8.1 Introduction ..................................................................................................................................... 144

8.2 Theoretical background ............................................................................................................... 144

8.3 Methodology ..................................................................................................................................... 145

8.3.1 Model development .............................................................................................................. 145

8.3.2 Parameter optimisation ...................................................................................................... 148

8.4 Results and discussion ................................................................................................................. 149

8.4.1 Specific growth rate estimation ....................................................................................... 149

8.4.2 Fitting the flask experiments ............................................................................................ 150

8.4.3 Model validation .................................................................................................................... 151

8.4.4 Testing a different expression for the specific lipid production rate ................ 153

8.4.5 Predicting shake-flask and bioreactor performance ............................................... 155

8.5 Conclusions ....................................................................................................................................... 157

Chapter 9 ............................................................................................................................................. 159

9 Evaluating feeding strategies for microbial oil production from glycerol by Rhodotorula glutinis ........................................................................................................................ 160

9.1 Introduction ..................................................................................................................................... 160

9.2 Materials and methods ................................................................................................................. 160

9.2.1 Fed-batch bioreactor experiments ................................................................................. 160

9.2.2 Oxygen uptake rate ............................................................................................................... 163

9.2.3 Analytical methods ............................................................................................................... 163

9.3 Results and Discussion ................................................................................................................. 163

9.3.1 Kinetic profiles of Rhodotorula glutinis using different feeding methods ....... 164

9.3.2 Effect of feeding style on the oxygen uptake rate ..................................................... 167

9.3.3 Effect of cumulative glycerol on growth and lipid production ............................ 168

9.3.4 Influence of the glycerol feeding rate on biomass yield from glycerol ............. 169


6

9.3.5 By-product formation and glycerol feeding rate ....................................................... 171

9.3.6 Comparison of fed-batch modes ...................................................................................... 172

9.4 Concluding remarks ...................................................................................................................... 176

Supplementary information for Chapter 9 ........................................................................................ 177

Chapter 10 .......................................................................................................................................... 183

10 Conclusions and recommendations .............................................................................. 184

10.1 Introduction ................................................................................................................................. 184

10.2 Discussion and concluding remarks ................................................................................... 184

10.3 Recommendations for future work .................................................................................... 187

References .......................................................................................................................................... 192

APPENDIX 1 ........................................................................................................................................ 202

APPENDIX 2 ........................................................................................................................................ 203

Word count: 49,850

Abbreviations and nomenclature

7

Abbreviations and nomenclature

ACCC Agricultural Culture Collection of China

ARA Arachidonic acid

AS Ammonium sulphate

C Concentration (g/L)

C/N Carbon to nitrogen ratio

CA Citric acid

CFB1 Fed-batch experiment with continuous feeding of glycerol at a rate close to the

glycerol uptake rate

CFB2 Fed-batch experiment with continuous feeding of glycerol at a rate twice as high

as the glycerol uptake rate

CFB3 Fed-batch experiment with continuous feeding of glycerol at a rate between that

of CFB1 and CFB2

CICC China Centre for Industrial Culture Collection

COD Chemical Oxygen Demand

DCW Dry Cell Weight (g/L)

DHA Docosahexaenoic acid

DO Dissolved oxygen (mg/L)

DOT Dissolved Oxygen Tension (%)

FB1 Fed-batch cultivation 1

FB2 Fed-batch cultivation 2

g gram (s)

GLA Gamma-linoleic acid

Glu Glucose

Gly Glycerol

kLa Mass transfer coefficient (h-1)

L Litre

nm nanometres

OD Optical density

PFB Pulsed fed-batch experiment

qi Specific nutrient (i) uptake rate (g/g/h)

R. Rhodosporidium

Rh. Rhodotorula


8

v/v Volume per volume

vvm Volume of gas flow per volume of vessel per minute

X Cell concentration (biomass ) (g/L)

Xf Lipid-free cell concentration (g/L)

YE Yeast extract

YL/Glu Yield of oil on glucose (g/g)

YX/Glu Yield of biomass on glucose (g/g)

YL/Gly Yield of oil on glycerol (g/g)

YX/Gly Yield of biomass on glycerol (g/g)

YX/N Yield of biomass on nitrogen (g/g)

YPD Yeast extract, peptone, glucose media

YPGly Yeast extract, peptone, glycerol media

w/w Weight per weight

Abstract The University of Manchester

9

BIOPROCESSING STRATEGIES FOR THE CULTIVATION OF OLEAGINOUS

YEASTS ON GLYCEROL

Abstract

Over recent years microbial oil has attracted much attention due to its potential to replace traditional oil sources in the production of biofuels and nutraceuticals. Its advantages arise from its independence of the food supply chain and its ease of production compared to conventional plant oils. Also, as concerns for the environment grow, microbially-synthesized oil emerges as potential competitor for the sustainable production of biodiesel. However, the high cost of its production currently hinders its large scale application. The bottlenecks to industrial microbial oil production are the cost of substrate and cultivation. Current research is focusing on process improvements to make microbial oil more competitive and worthwhile to produce. Several types of microorganisms have been explored so far and waste substrates have been utilised as cheap feedstocks. The overall cost is affected by the fermentation stage, therefore it is imperative to design cultivations with little operating requirements and high yields. Consequently, the present thesis aims to contribute to the field by developing and investigating a simple process for oleaginous yeast cultivation, focusing mainly on enhancing the yields during the bioreactor stage. Oleaginous yeasts were screened for their ability to grow on glycerol and the most promising strain was selected for further research. Then, the necessary conditions for its growth and oil accumulation were defined. Shake-flask cultivations showed that the specific growth rate and glycerol consumption of Rh. glutinis were higher at lower glycerol concentrations (≤40 g/L), while higher C/N elemental ratios enhanced oil content. Experimental data were used to construct an unstructured kinetic model to describe and predict the system’s behaviour. The Monod-based model took into account double substrate growth dependence and substrate inhibition. Following that, bioreactor cultivations extended the range of parameters studied, to include the influence of aeration rate and oxygen supply on cellular growth and microbial oil production. Cultivations at different air flow rates were performed in a 2 L bioreactor and showed that a low aeration rate of 0.5 L/min gave the best glycerol and nitrogen uptake rates, resulting in a concentration of biomass of 5.3 g/L with oil content of 33% under simple batch operation. This was improved by 68% to 16.8 g/L (cellular biomass) with similar oil content (34%) by applying a fed-batch strategy. Finally, different glycerol feeding schemes were evaluated in terms of their effect on oil accumulation. The concept of targeting first a cell proliferation stage, limited by the availability of nitrogen, followed by a lipid accumulation stage, fuelled by glycerol was tested. Continual feeding and pulsed feedings, delivering the same total amount of nitrogen (and glycerol), resulted in similar elevated values of both cellular biomass (~25 g/L) and oil content (~40%). Addition of glycerol at higher rates but giving the same total amount of nitrogen led to a further increase in oil content to 53%, resulting in an overall oil yield of more than 16 g/L (the highest achieved throughout the project). With comparable yields to those reported in the literature but achieved with a much poorer medium, there is every reason to be optimistic that microbial oil production from glycerol could be commercially viable in the future.

Eleni Karamerou PhD thesis September 2016


10

Declaration

No portion of this work referred to in the thesis has been submitted in support of an

application for another degree or qualification of this or any other university or other institute

of learning.

Copyright statement

i. The author of this thesis (including any appendices and/or schedules to this thesis) owns

certain copyright or related rights in it (the ‘Copyright’) and she has given The University

of Manchester certain rights to use such Copyright, including for administrative

purposes.

ii. Copies of this thesis, either in full or in extracts and whether in hard or electronic copy,

may be made only in accordance with the Copyright, Designs and Patents Act 1988 (as

amended) and regulations issued under it or, where appropriate, in accordance with

licensing agreements which the University has from time to time. This page must form

part of any such copies made.

iii. The ownership of certain Copyright, patents, designs, trade marks and other intellectual

property (the ‘Intellectual Property’) and any reproductions of copyright works in the

thesis, for example graphs and tables (‘Reproductions’), which may be described in this

thesis, may not be owned by the author and may be owned by third parties. Such

Intellectual Property and Reproductions cannot and must not be made available for use

without the prior written permission of the owner(s) of the relevant Intellectual Property

and/or Reproductions.

iv. Further information on the conditions under which disclosure, publication and

commercialisation of this thesis, the Copyright and any Intellectual Property University

IP Policy (see http://documents.manchester.ac.uk/display.aspx?DocID=24420), in any

relevant Thesis restriction declarations deposited in the University Library, The

University Library’s regulations (see

http://library.manchester.ac.uk/about/regulations/) and in The University’s policy on

Presentation of Theses.

http://documents.manchester.ac.uk/display.aspx?DocID=24420

http://library.manchester.ac.uk/about/regulations/)

Acknowledgements

11

Acknowledgements

First, I would like to thank my supervisors, Professor Colin Webb and Professor

Constantinos Theodoropoulos, for their unlimited help and undiminished interest in my

research effort. Especially, I thank Professor Webb for giving me freedom on my research as well

as for his tolerance and encouragement in every idea or research initiative, which aimed at

making me a responsible and independent researcher. Also, for his support and advice on

everyday life issues.

In addition, I thank the University of Manchester President’s Doctoral Scholar Award for

the financial support to pursue my doctoral studies.

I should not omit thanking my colleague and friend, Musaalbakri Abdul Manan, for

introducing me to microbiological techniques at the beginning of my project. Furthermore, I

would like to thank Dr Saul Alonso Tuero for his valuable comments and advice related to this

work and for sharing with me his expertise in experimental matters.

Many thanks to the MSc students, Simin Zhang and Ioannis Efthymiopoulos for their

contribution to the screening study and crude glycerol, respectively.

Furthermore, I would like to thank my friends Chen-Wei, Sara, Candice, Gonzalo and

Liliana for their friendship and the nice atmosphere they created.

I also appreciate the help from the technical personnel of the School.

Finally, I thank my parents for their love and moral support during these years and I

dedicate the present thesis to them.


12

The author

Eleni Karamerou graduated from the University of Patras (Greece) in 2011 with a 5 years

Diploma in Chemical Engineering with grade ‘Excellent’. The programme involved one year

research experience with the submission of a Diploma thesis that dealt with photocatalytic

reforming of glycerol conversion to hydrogen. During her studies she received funding from the

State Scholarships Foundation and in January 2012 was awarded First place in the rank of

Chemical Engineering graduates by the Technical Chamber of Greece.

In September 2012 Eleni joined the School of Chemical Engineering and Analytical

Science at the University of Manchester, sponsored by a University of Manchester President’s

Doctoral Scholar Award to pursue doctoral studies. The research, supervised by Professor Colin

Webb, involved bioprocessing aspects of cultivating yeasts on glycerol for microbial oil

production, as biodiesel feedstock.

The work reported in this thesis has been presented at national and international

conferences including the 23rd European Biomass Conference and Exhibition 2015 (Austria), the

3rd European Congress of Applied Biotechnology in 2015 (France), the International Bioenergy

Conference in 2014 (UK) and ChemengDayUK 2014 and 2015.

Publications arising from this research

1. Karamerou E., Theodoropoulos C., Webb C. (2016). “Evaluating feeding strategies for

microbial oil production from glycerol by Rhodotorula glutinis”, Engineering in Life

Sciences (Article in Press), DOI: 10.1002/elsc.201600073

2. Karamerou E., Theodoropoulos C., Webb C. (2016). “A biorefinery approach to microbial

oil production from glycerol by Rhodotorula glutinis”, Biomass and Bioenergy (89):113-

122, http://dx.doi.org/10.1016/j.biombioe.2016.01.007

3. Karamerou E., Theodoropoulos C., Webb C. (2015). “Yeast microbial oil from biodiesel

waste glycerol: a promising tool for biorefinery enhancement”, Proceedings of the 23rd

European Biomass Conference and Exhibition in Vienna, 2015. DOI:

10.71/23rdEUBCE2015-3CO.3.4

Manuscripts under preparation

1. Karamerou E. Webb C., State of the art in cultivation modes for microbial oil production

(review article).

2. Karamerou E., Zhang S., Theodoropoulos C., Webb C. A comparative study on oleaginous

yeasts grown on glycerol (research article).

http://dx.doi.org/10.1016/j.biombioe.2016.01.007

13

Chapter 1

Microbial oil as the basis for a sustainable society


14

1 Microbial oil as the basis for a sustainable society

1.1 Introduction

Due to the rise in prosperity levels in conjunction with the increased population, the

energy demand for living and transportation purposes, has grown significantly over recent

decades. For instance, according to the US Energy Information Administration (EIA) the world

energy consumption increased by 51% between 1990 and 2012 (Figure 1). Currently, the energy

required for most of the world’s activities is derived from fossil fuels, in particular oil. Human

life is strongly dependent on oil, since the majority of the materials we use daily are oil derived

(petrochemicals). Petroleum is the main feedstock for products such as fuels (gasoline, naphtha),

lubricants, paints, coatings, detergents, plastics, greasing agents, industrial gases (CO2, H2, Ar, N2)

and solvents for extraction processes, such as in food and pharmaceutical manufacturing.

However, extensive use of fossil fuels has released significant amounts of greenhouse gases

(GHG) with direct impact on the environment and consequently on human welfare, with the

energy industries and transportation sectors being the major causes for GHG emissions (Mata et

al., 2010).

Because of the imminent scarcity of petroleum reserves, the environmental concerns

related to the use of fossil fuels and the price instability of petroleum, alternative sources of

energy have been investigated and are already in use (Almeida et al., 2012). Taking actions to

reduce further the emissions related to the transportation sector ranks as an important factor in

Figure 1: Progress of world primary energy consumption over the period 1990-2012 (Data adapted from the following source: https://www.eia.gov/cfapps/ipdbproject/IEDIndex3.cfm?tid=44&pid=44&aid=2, accessed March 2016)

https://www.eia.gov/cfapps/ipdbproject/IEDIndex3.cfm?tid=44&pid=44&aid=2

Chapter 1 Microbial oil as the basis for a sustainable society

15

reducing further the environmental damage. Hence, biomass dependent biofuels, such as

bioethanol, biodiesel and biobutanol have emerged as technically feasible and renewable

options with lower emissions and long-term stability. GHG emissions from the transportation

sector have declined recently while it is expected that the application of renewable fuel sources

will further reduce the overall emissions by 2030 (Figure 2A). While still only a small portion of

total energy supply, biofuels represent a substantial and growing portion of the renewable

energy supply (Figure 2B).

The biomass used for biofuels production is usually plant material, such as oil-based,

sugar-based or starch-based crops. With the use of thermochemical or biological processes these

are converted into valuable products in a new type of refinery, the so-called biorefinery. Such

biorefineries are going to play a major role in a sustainable society.

1.2 Why do we need more oil and alternative oil sources?

Biodiesel is an environmentally friendly, renewable fuel, which is the bio-based analogue

of diesel. Supported by governmental subsidies, it is currently dominating the European biofuel

market (Bozbas, 2008). The most common feedstocks for biodiesel production are food grade

oils, extracted from edible plants such as rapeseed, soybean, sunflower or palm. The use of edible

plant oils for biofuel production raises concerns regarding their necessity as food ingredients.

Biodiesel is blamed for the increased prices of edible plants which affect the economics of both

food and biofuel production. Moreover, the current scarcity of these oil sources and the inability

(A) (B)

Figure 2: (A) Greenhouse gas emissions from the transportation sector (including aviation) for the period 1990-2013. Further reduction is expected by 2030. (Data adapted from the following source: http://www.eea.europa.eu/data-and-maps, accessed on March 2016 with search on Transport topic) (B) Global share of fuel sources in 2013, 1 data might not add up because of rounding. (Source:https://www.iea.org/publications/freepublications/publication/RENTEXT2015_PARTIIExcerpt.pdf , accessed March 2016).

http://www.eea.europa.eu/data-and-maps

https://www.iea.org/publications/freepublications/publication/RENTEXT2015_PARTIIExcerpt.pdf

https://www.iea.org/publications/freepublications/publication/RENTEXT2015_PARTIIExcerpt.pdf


16

to meet the increasing demand (currently 3.5 Mt/y) for biodiesel, create the need for oil from

non-edible sources (Pinzi et al., 2014). Alternative sources to edible oil such as waste oils, non-

edible plant oils and microbial oils have emerged as potential feedstocks and are summarised in

Table 1.

Table 1: Non-edible oil sources as potential biodiesel feedstocks (Pinzi et al., 2014; Adewale et al., 2015; Kumar and Sharma, 2015)

Oil source Advantages Disadvantages

Non-edible plants (Jatropha, Pongamia, Camelina)

Ability to grow on marginal lands Non-competition with food

Disruption of ecology Competition with agriculture and use of water Large amount of residues after extraction

Frying/waste oils (greases)

Waste material that needs disposal No conflict with land and food chain

High content of free fatty acids due to prolonged heating generates soap and low yield of biodiesel

Animal fats (poultry fat, tallow, lard)

High degree of saturation Oxidation stability of resulted fuel

High content of free fatty acids Limited by the requirements of livestock farming

Microalgal oil Saline water, CO2 capture, No competition with agriculture, faster growth compared to plants

High infrastructure costs Low growth rates compared to other microorganisms Likelihood of contamination

Yeast oil High growth rates Substrate versatility Conventional cultivation systems Suitable composition for biodiesel

High fermentation cost Non-competitive to plant oils Under development stage

Animal fats are generally the by-product of edible meat production. They are low cost,

easily harvested during butchering and are currently used in soap manufacturing. But animals

are slow-growing and large numbers are needed for a substantial amount of oil. As a result, they

cannot meet the forecasted requirements in terms of biodiesel production needs. Waste oils have

potential as biofuel feedstock but the prolonged heating they may have undergone can result in

some unwanted characteristics to the resulting biofuel. Non-edible plants have an indirect

competition with food, which stems from the occupation of land required for their cultivation.

Microbial oils, the so-called ‘single cell oils’, are produced by oil accumulating

microorganisms (mainly yeasts and microalgae) under nutrient-limited culture conditions and

have applications in production of cocoa butter substitutes, polyunsaturated fatty acids

(arachidonic, γ-linoleic) and have lately attracted attention as potential biodiesel feedstocks


17

(Huang et al., 2013). Microbial oil has similar composition to plant oils, in the sense that it is rich

in fatty acids such as C18:1, C18:2, C18:3 and C16:0 (Ratledge 2001). Furthermore, oleaginous

microorganisms have the edge over plants since they are not affected by climate change and they

achieve high oil yields in short periods (Louhasakul and Cheirsilp, 2013). It may even be the case

that petroleum is effectively fossilised microbial oil. Although there is no proof for the synthesis

of petroleum, hydrocarbon synthesizing microbes and the oil accumulating capacity of marine

organisms indicate that microorganisms could be responsible for its formation. Therefore,

utilising microorganisms for alternative oil production is a way of taking into our own hands,

Nature’s process.

1.3 Industrialisation possibilities of microbial oil

Although microbial oil production has great potential, currently its use as biodiesel

feedstock is not feasible due to the high processing costs, which have constrained its commercial

viability. The fermentation stage is the main factor affecting the cost due to the necessity for large

bioreactors, oxygen supply, pH control and monoculture maintenance. Therefore, advancements

such as higher substrate conversion yields, robust cultivations with little external control

requirements, resistance to contamination and stable production should be targeted.

Among the process improvements that are required to make microbial oil production a

commercial reality, the use of readily available and low-cost fermentation substrates plays a

major role in upstream processing. The high substrate versatility of yeast species allows the use

of inexpensive or negative-cost materials such as wastes and agro industrial residues as

potential feedstocks. One such material is the chemical by-product of biodiesel, glycerol. Large

amounts of crude glycerol, a mixture of glycerol, methanol and other impurities, are generated

during the transesterification of triacylglycerols to biodiesel (Figure 3). Stoichiometrically,

glycerol amounts to approximately 10% w/w of produced biodiesel.

The recent increase in biodiesel production is directly responsible for the corresponding

increase in crude glycerol, which has caused the prices of glycerol to drop. In view of that,

utilisation of this glycerol in oleaginous yeast fermentations presents an excellent opportunity

for improving the overall economics of microbial oil production, by minimising the contribution

of the raw materials to the overall cost. However, considerable research and development is

needed in order to design efficient processes. Major challenges behind the use of glycerol as a

carbon source are the conversion efficiency and the presence of inhibitors. To tackle these, it is

necessary to develop efficient fermentation strategies adjusted to the characteristics of an


18

oleaginous culture system with the aid of biochemical engineering tools to develop cultivation

modes for higher productivities.

The purpose of the present thesis is to report research investigating the possibility of

taking advantage of the availability of glycerol and utilising it as inexpensive carbon source for

yeast oil production. With the knowledge obtained from studies on the effect of nutrients, C/N

ratio and cultivation conditions on the growth and oil productivity, different fermentation

approaches have been evaluated in order to provide a better understanding of the role of glycerol

feeding mode on bioprocess parameters. It is hoped that this study will pave the way to

developing practical operating strategies to achieve efficient utilisation of the surplus glycerol in

the production of microbial oil.

1.4 Structure of the thesis

Following this introductory chapter, an overview of the literature related to microbial oil

production and process development is presented in Chapter 2. Following that, the objectives

(Chapter 3) of the research are presented along with the experimental programme required to

achieve these and overall goals. In Chapter 4 the main materials and methods are stated. The

results chapters then follow, starting with Chapter 5, where a screening study on several

oleaginous yeasts led to a selected species for further research, Chapter 6 presents the main

aspects of growing this yeast. Chapter 7 contains a study of the cultivation conditions for lipid

accumulation for this strain and considerations for scaling up the flask cultivations to bioreactor.

In Chapter 8, a kinetic model describing the system is presented. Following this, Chapter 9,

Figure 3: Biodiesel production process.


19

reports an investigation of glycerol feeding modes for efficient growth and oil production. Finally

conclusions are drawn in Chapter 10.


20

21

Chapter 2

Microbial oil production from glycerol: a review of

the literature


22

2 Microbial oil production from glycerol: a review of the literature

2.1 Introduction

In this Chapter, the main aspects of oil production are reviewed. After introducing

microbial oil and oil accumulating microorganisms, the conditions that lead to oil accumulation

are explained. Following this, focus is given to the current limitations of oleaginous systems in

in terms of industrialisation and the current state-of-the-art for these issues. These include

process improvements, which is also the scopus of the research presented in this thesis.

2.2 Microbial oil and its applications

Microbial oil can be considered to be any lipid material that is produced by

microorganisms and which is soluble only in organic solvents. Hence, it is also referred to as:

‘Microbial lipids’, or ‘Single Cell Oil’, with the latter equivalent to the term ‘single cell protein’

(SCP). Microbial lipids are intracellularly synthesized by microorganisms when they are subject

to imbalanced growth conditions and in certain microorganisms form discernible oil globules.

Figure 4 shows how these oil globules appear inside the cells of oleaginous yeasts.

Lipids are mainly composed of triacylglycerols (TAGs), which are three fatty acids

connected to a glycerol backbone (Figure 5). The fatty acids of oleaginous microorganisms

contain long chain carboxylic acids (ca. C16 to C18) and can be either saturated (with only single

bonds) or unsaturated (with one or more double bonds). Their main applications are supply of

Figure 4: Yeast cells (Rhodotorula glutinis) showing intracellular oil droplets. On the left untreated, on the right stained with Sudan Black B. Photos taken from the research reported in this thesis.

10μm

Chapter 2 Microbial oil production: literature review

23

polyunsaturated fatty acids (PUFAs), such as linoleic acid (GLA), arachidonic (ARA),

eicosapentaenoic (EPA) and docosahexaenoic acid (DHA), cocoa butter substitute and,

potentially, feedstock for biodiesel production (Huang et al., 2013).

The interest in Single Cell Oils has a long history starting from the beginning of the 20th

century. Research on the ability of microorganisms to accumulate oil originated in Germany

while investigating alternative oil supplies as well as food supplements to cope with the shortage

during World War I (Ratledge, 2001). This continued through World War II. By the 1950s the

basic understanding regarding oil-producing strains, the conditions for lipid accumulation as

well as the lipid composition had been established and research was expanded in Sweden,

Netherlands, UK, Canada and US (Lundin, 1950; Wynn and Ratledge, 2005; Sitepu et al., 2014).

Despite the potential of microbial oil, governmental subsidies towards agricultural development

brought forward the cultivation of oily plants, placing them ahead of their microbial

counterparts. At the same time the lack of large scale microbial facilities, the high price of glucose

as fermentation substrate and lack of fermentation development restricted industrialisation but

did not stop research on microbial oils.

Around the 1960s -1970s, it was acknowledged that microbial oils could serve as

precursors of important high valued oil-derived components for human consumption that were

either rare or impossible to obtain from the usual plants or fish. Microbial-derived PUFAs were

of superior quality with rich content of the desired fatty acid (Lundin, 1950). This opportunity

re-kindled interest in microbial oils and some signs of commercial production appeared. The

first production of fungal Gamma linoleic acid in UK was from the company of John and E Sturge

(Selby, North Yorkshire) in 1985 and many other fungal species were examined around that time

for their fatty acid production capabilities and suitability for human consumption (Ratledge,

2001; Ratledge, 2013; Wynn and Ratledge, 2005). Further large scale production of PUFAs,

(A) (B)

Figure 5: (A) General structural representation of a triacylglycerol. (B) Example of a triacylglycerol molecule, glycerol connected with, from top to bottom, palmitic, oleic and γ-linoleic acid (Designed based on: https://en.wikipedia.org/wiki/Triglyceride).

https://en.wikipedia.org/wiki/Triglyceride


24

continued with the production of ARA in Europe and Japan for infant nutrition and DHA from

Martek Biosciences. Research focused on ways of increasing the yields, the quality and the safety

of the products and processes (Wynn and Ratledge, 2005).

Another very important capability of oleaginous organisms, mainly yeasts, was the

synthesis of oils similar to cocoa butter (containing stearic, oleic and palmitic acid) fuelled during

the early 1980s by the high price of cocoa butter 0.8$/kg at that time (Papanikolaou and Aggelis,

2011b). Yeasts were engineered in order to obtain improved productivities and other potential

substrates were explored. However, such applications added to the final cost and the process

was considered less competitive to other ways of obtaining this fat.

Attention to microbial oil upscaling has been affected by the variability of prices of

conventional plant oils and agricultural development and funds. Commercial possibility of oils

for human nutrition depends on the use of cheaper carbon sources, further genetic engineering

approaches and fermentation design.

Over recent years, there has been a revival of interest in microbial oils as potential

feedstock for biodiesel production instead of vegetable oils. The generally high price of crops and

their importance for the food chain, limit biodiesel expansion. On the contrary, microorganisms

do not require arable land for their cultivation and can be ready for harvesting quickly as well as

having the suitable fatty acid content for biodiesel. Industrial use of microbial oil for biodiesel is

still in its infancy due to the main barrier that the high cost of the process imposes (Koutinas et

al., 2014). There are attempts at commercialisation of microbial oil in US with Solazyme

(http://solazyme.com/innovation/microalgae/, accessed on April 2016), while in Europe, the

company Neste Oil launched the first pilot plant for producing microbial based biodiesel in 2012.

Since it is evident that large scale oil production for biofuels will make sense only when the price

of oil is competitive to conventional oils or the process is characterised by added value products,

the current scenario in microbial oil research focuses on identification of high oil producing

species, conversion of low-cost carbon sources to oil and process optimisation in order to

improve the productivity, coupling the process with waste creating processes (Sitepu et al.,

2014; Huang et al., 2013; Liang 2016). Around the world, a continuously growing research

interest is focusing on developing improved and sustainable processes for microbial oil

production based on the above concerns. Figure 6 shows the increasing number of scientific

reports using the term ‘microbial oil’ over the last several decades.

http://solazyme.com/innovation/microalgae/


25

There has been a sharp increase in the number of relevant publications since 2010 on

microbial oil with almost 122 publications in 2014, while it is anticipated that it will continue

around that level. From these reports, 307 in total, 42% (128 reports) are referring to biodiesel

(search consisted of ‘Microbial oil’ or ‘Microbial lipids’ or ‘Single Cell Oil’ and ‘Biodiesel’ as the

search terms, information accessed in April 2016). This confirms that the biofuel application of

microbial oil is currently the main investigation. Leading countries in this research, according to

the data from Scopus by searching the terms ‘Microbial oil’, ‘Microbial lipids’ and ‘Single Cell Oil’

are China and United States, followed by United Kingdom, Germany and Greece (Figure 7).

Interestingly, the position of Greece is due almost entirely to the efforts of a single research

group, based in Athens Agricultural University. Moreover, in the UK, research on microbial oils

was initiated in the mid-1970s by Ratledge, with a focus on polyunsaturated fatty acids and

understanding the lipid biochemistry while nowadays the trend is towards biofuels from

oleaginous yeasts with the research groups of Webb and Chuck (Figure 7). These data suggest

that this research is worth investigating and there is much space for new findings and there are

many chances to move forward considering the high number of researchers dealing with

microbial oil.

Figure 6: Number of scientific reports with search on ‘Microbial Oil’ or ‘Microbial lipids’ or ‘Single Cell Oil’ based on data derived from Scopus in April 2016, by quoting the above terms. This number includes articles, book chapters, conference papers and review articles. The value for 2016 is estimated by linear extrapolation of the period 2012-2015.


26

2.3 Oleaginous microorganisms

All microorganisms are capable of synthesizing lipids with functional role (membranes

formation), however significant amounts of intracellular lipids are accumulated by only a few of

them, which are known as oleaginous. These microorganisms are able to accumulate large

proportions (more than 20% of their weight) of lipids (Meng et al., 2009). Oleaginous

microorganisms belong to various groups, including filamentous fungi, yeasts, microalgae and to

a lesser extent bacteria. The amounts and the compositions of accumulated lipids are case-

Figure 8: Leading countries in Microbial oil research. Data from Scopus by quoting ‘microbial oil’ or ‘microbial lipids’ or ‘single cell oil’ in April 2016.

Figure 7: Key authors from the highest in rank countries. Data from Scopus by quoting ‘microbial oil’ or ‘microbial lipids’ or ‘single cell oil’ in April 2016.


27

specific for each kind of microorganism. Also, with the progress in genetic engineering the oil

accumulating capacity of non-oleaginous microorganisms can be improved (Hegde et al., 2015).

Different oleaginous microorganisms and their lipid compositions are presented, in the

following sections.

2.3.1 Oleaginous microalgae

Microalgae have attracted much interest as lipid producers for various applications,

because they are able to produce large amounts of lipids, proteins and carbohydrates. Their

average lipid content is usually in the range 20-50% (Chisti, 2007). Autotrophic microalgae

utilise carbon dioxide as carbon source and sunlight as energy for oil accumulation thus

benefiting the environment. These microalgae require abundant sunlight, mild pH conditions

and a certain level of salinity in order to grow properly (Li et al., 2008). Well-known autotrophic

oleaginous microalgae include Chlorella sp., Nannochloropsis sp. and Botryococcus braunii

yielding up to 58% oil content (Mata et al., 2010). Microalgae can grow heterotrophically with a

carbon source for their growth and energy provision, in a conventional way and more efficiently

than their autotrophic counterparts with higher oil and biomass productivities (Chen et al.,

2011). Table 2 shows the yields of some oleaginous microalgae cultivated under heterotrophic

conditions. Microalgae can be cultivated in either bioreactors with light supply if needed or open

ponds (Chisti, 2007). In general, algal oil contains long chain polyunsaturated fatty acids with

medical and nutritional interest, omega-3, omega-6 fatty acids (Azocar et al., 2010; Bellou et al.,

2014) but suitable oils according to the biodiesel standards do exist (Huang et al., 2010).

However, the cost of algal biofuels remains high to-date.

2.3.2 Oleaginous bacteria

Bacteria accumulate lipids, but they are not to favourable oil producers like the other

microorganisms. Although bacteria proliferate at high rates and require only simple culture

modes, something that could be an advantage here, the majority of them cannot produce

significant amounts of lipids. In those that can, strains of the actinomycete group (Alvarez and

Steinbüchel, 2002), lipids are mainly in the form of lipoids. Bacteria are rather preferred for the

production of special lipids such as polyhydroxybutyrates (PHB), wax esters and free fatty acids.

Moreover, it can be difficult to recover bacterial lipids, because they are often collected in the

outer membrane and they don’t attract interest for biofuels production (Meng et al., 2009). A

promising option regarding biodiesel from bacteria is the use of genetically modified strains,


28

such as Escherichia Coli (Kalscheuer et al., 2006; Lu et al., 2008)). Some bacteria able to

accumulate lipids are listed in Table 2.

2.3.3 Oleaginous fungi

Fungi are divided into yeasts and filamentous fungi (moulds). The lipid yield and

composition are mainly dependent on the strain and the culture conditions.

Filamentous fungi (moulds)

Fungi accumulate large amounts of oil, up 70% of their dry weight. Extensively studied

fungal species are Mucor circineloides, Mortierella alpina, Motrielella isabellina (Table 2). Fungal

oils are mainly composed by PUFAs and have been extensively investigated for that, they can

also be used as biodiesel feedstock (Vicente et al., 2009; Čertík et al., 2012). Filamentous fungi

can be cultivated under submerged conditions. However, their morphology allows them to

survive in solid state, a condition that mimics their natural habitat (Cheirsilp and Kitcha, 2015)

and as such, they can be used to decompose lignocellulosic materials and convert agricultural

residues to oil in a simple way.

Oleaginous Yeasts

Yeasts are fungi with mainly unicellular1 but in some cases mycelial form and they have

been used in several biotechnological applications, such as protein production. Their high lipid

content makes them attractive candidates for both human nutrition products and biofuel

production. Yeasts lipids are mainly in the form of triacylglycerols, with long chain fatty acids

(C16-C18) with oleic, stearic and palmitic as the predominant fatty acids (Sitepu et al., 2014).

Amongst several oleaginous yeast species, Rhodotorula glutinis, Rhodosporidium toruloides,

Yarrowia lipolytica and Cryptococcus curvatus have attracted much interest due to their lipid

capabilities and synthesis of other useful co-products (Chatzifragkou et al., 2011; Papanikolaou

et al., 2002b; Saenge et al., 2011a). Yeasts can be cultivated in conventional stirred bioreactors,

to high cell density at a high rate, they can consume a variety of carbon sources. Most of them

are obligate aerobes. From around 600 known yeast species to-date only a small portion has

been found to be oleaginous. According to a report from Ratledge and Wynn (2002), only 5% of

the known yeast species were classified as oleaginous. However research on oil accumulating

1 Their cellular size varies between 3-10 µ in width and 3-100 µ in length and their cell shape can be spherical, ovoid or elongated.


29

yeasts, known and newly isolated showed that between 3% and 10% could be considered to be

oleaginous (Sitepu et al., 2014) and considering the rapid development of microbial oil research

recently, the number of known oleaginous strains is expected to increase.

Table 2: Lipid yields of oleaginous microorganisms along with their carbon source and cultivation method.

Species Carbon source Oil content (% w/w)

Reference

Microalgae

Chlorella protothecoides Sugarcane bagasse hydrolysate 34 (Mu et al., 2015)

Chlorella protothecoides Crude glycerol 50 (Cerón-García et al., 2013)

Chlorella sorokiniana Glucose 38.7 (Zheng et al., 2013)

Scenedesmus sp. Glucose 43.4 (Ren et al., 2013)

Chlorella vulgaris Acetate 31 (Liang et al., 2009)

Bacteria

Rhodococcus rhodochrous Glucose 43 (Shields-Menard et al., 2015)

Rhodococcus opacus Glucose 38.4 (Kurosawa et al., 2010)

Rhodococcus opacus Glycerol-Glucose 51.2 (Kurosawa et al., 2015)

Filamentous fungi

Mortierella isabellina Xylose 64 (Gao et al., 2013)

Cuninghamella echinulata Molasses 32 (Chatzifragkou et al., 2010)

Mortierella isabellina Crude glycerol 53.2 (Fakas et al., 2009)

Aspergilus niger Crude glycerol 57 (Andre et al., 2010)

Mucor corcinelloides Glucose 44 (Carvalho et al., 2015)

Yeasts

Lipomyces starkeyi Crude glycerol 35.9 (Tchakouteu et al., 2015)

Yarowia lipolytica Crude glycerol 43 (Papanikolaou and Aggelis, 2002)

Rhodosporidium toruloides Glucose 48 (Uçkun Kiran et al., 2012)

Rhodotorula glutinis Crude glycerol 60.7 (Saenge et al., 2011b)

Cryptococcus curvatus Acetate 49.9 (Gong et al., 2015)

Trichosporon cutaneum Corncob hydrolysate 32.1 (Gao et al., 2014)

2.4 The mechanism of lipid accumulation in yeasts

Lipogenesis in yeasts takes place by two different mechanisms according to the

fermentation substrate. The first is referred to as ‘de novo’ lipid accumulation and takes place

when hydrophilic substrates (commercial sugars, lignocellulosic hydrolysates or glycerol) are

the carbon source and nitrogen or other growth-essential element is the limiting nutrient. ‘Ex

novo’ lipid accumulation is a non-growth associated process where hydrophobic substrates (oils

and fats) are the carbon source without nutrient depletion and with cell proliferation taking

place at the same time (Papanikolaou and Aggelis, 2011b). In this work ‘de novo’ oil production

is applied therefore only this case will be elaborated in this section.


30

‘De novo’ lipogenesis is induced by the depletion of an essential component for cell

proliferation, usually nitrogen in the presence of carbon excess. Lipids are formed in the cytosol

as energy storage for continuation of vital cellular functions in case of further nutrient deficiency

in the future (e.g. carbon extinction). In particular, cell proliferation is the dominant process

when carbon and nitrogen are both present in the medium. After the exhaustion of nitrogen, any

excess carbon is still taken up by the cells and converted into storage lipids rather than towards

building new cells (Figure 9). A combination of excess carbon and low nitrogen, a high C/N ratio,

is generally regarded as a prerequisite for lipid synthesis.

Triacylglycerol synthesis takes place inside the cytosol (Xu et al., 2013), however the step

that initiates its production takes place in the mitochondrion. In general, lipid accumulation is a

process resulting from the disruption of the metabolic cycles, during nitrogen deficiency, while

the cell tries to maintain vital functions. Details in these metabolic pathways and enzymatic

reactions can be found in other reports (Papanikolaou and Aggelis, 2011a; Ratledge and Wynn,

2002), which were consulted in order to provide here a basic understanding of lipid synthesis.

The regulatory steps for lipid synthesis are summarised in Figure 10.

Upon nitrogen depletion, cell proliferation ceases and synthesis of nucleic acids and

proteins stops. The cells continue to take up carbon source, which enters the cytosol as normally

and follows the EMP (Glycolysis) pathway in order to be converted to pyruvate. This enters the

mitochondrion, where it is enzymatically converted to oxaloacetate and acetyl-CoA by pyruvate

dehydrogenase (PD). Acetyl-CoA and oxaloacetate are converted to citric acid (CA) by citrate

synthetase (CS), within an irreversible reaction which is part of the Krebs cycle (or TCA cycle).

Figure 9: The critical role of nitrogen in lipid accumulation.


31

Citrate in its turn is isomerised to isocitrate which is then transformed to α-ketoglutarate by the

enzyme isocitrate dehydrogenase (ICDH). This is where the Krebs cycle is affected by the

external nitrogen deficiency and the conditions for TAG synthesis are created.

The lack of nitrogen input drives the cell to decompose adenosine monophosphate (AMP)

to inosine monophosphate (IMP) and NH4+, in order to provide itself with some nitrogen for

nucleic acid and protein biosynthesis. The function of the enzyme isocitrate dehydrogenase

(ICDH) is linked to the availability of AMP (Botham and Ratledge, 1979). Hence ketoglutaric acid

cannot be produced any more and the Krebs cycle cannot be completed. Accumulation of

isocitrate slows the reaction of its formation. Citrate then reaches a critical value inside the

mitochondrion and moves to the cytosol. In the cytosol, the enzyme ATP-citrate lyase (ATP-CL)

splits the citrate to oxaloacetate and acetyl-CoA. Regarding the oxaloacetate, it gives back

pyruvate after being converted to malate. The malic enzyme (ME) converts malate to pyruvate

with the conversion of NADP to NADPH. At the same time, acetyl-CoA, initiates a series of

reactions for the actual fatty acids biosynthesis after the generation of malonyl-CoA. These

reactions are catalysed by the fatty acid synthetases (FAS) along with the released NADPH from

the malic enzyme action. After that, esterification with glycerol, desaturation and elongation of

the carbon chain follow and all these constitute the pathway, which leads to the formation of

TAGs. Details of this pathway can be found in more detail elsewhere (Ratledge and Wynn, 2002;

Papanikolaou and Aggelis, 2011a).

The TAGS are organised in lipid bodies within the cytoplasm. The lipid bodies start their

formation from the endoplasmic reticulum and grow to a diameter of about 50 μm, maintaining

their shape and without coalescing due to the presence of proteins and polar lipids surrounding

the main TAG core (Ratledge and Wynn, 2002; Leber et al., 1994; Murphy and Vance, 1999). The

role of these proteins and lipids is to regulate the stored lipid and, when required, efficiently re-

utilise it.

It is acknowledged that the difference between oleaginous and non-oleaginous

microorganisms is the presence and function of the key enzymes ATP-citrate lyase (ATP-CL) and

the malic enzyme (ME). ATP-CL is not detected in most non-oleaginous species and the function

of malic enzyme affects the level of lipid production by generating the necessary NADPH (Sitepu

et al., 2014). If the malic enzyme does not work properly, because of inhibition or disabling, lipid

accumulation is slow.

These aspects are addressed by research on genetic engineering, either by improving the

lipid synthesis pathway of existing oleaginous yeasts or introducing relevant genes into non-

oleaginous yeasts (Sitepu et al., 2014). For example, overexpression of enzymes from the lipid


32

synthesis pathway of a Yarrowia lipolytica strain (Tai and Stephanopoulos, 2013), improved

significantly its lipid content. Nevertheless, despite the improvements made by genetic

engineering, process engineering is more important and beneficial for progress in microbial oil.

It is always worth seeking advances in process engineering, since they can be applicable to any

kind of strain.

2.5 Factors affecting lipid accumulation in yeast species

2.5.1 Growth elements

Many elements are related to cellular growth and their absence can induce lipid

accumulation. These include N, P, Zn, Fe, S or Mg, with nitrogen being the best case of nutrient

for studying this phenomenon, because it is directly related to cell proliferation (Ratledge, 2001).

Nitrogen is associated with the formation of macromolecules within the cell, such as nucleic acid

and proteins. These molecules are necessary for cell proliferation and are created during the

Figure 10: Biosynthesis of lipid accumulation (constructed based on data from (Wynn and Ratledge, 2005; Papanikolaou and Aggelis, 2011a). C-source, is a general representation of the carbon source e.g. glucose or glycerol; EMP pathway, is the Glycolysis pathway; TCA cycle, is the cycle for carboxylic acids synthesis; AMP, adenosine monophosphate; ICDH, isocitrate dehydrogenase; IMP, inosine monophosphate; PD, pyruvate dehydrogenase; Ac, Aconitase; MDc, malate dehydrogenase cytoplasmic; ACL, ATP-citrate lyase; ACC, Acetyl-CoA carboxylase; FAS, fatty acid synthetase; DAG, diacylglycerol; TAG, triacylglycerol; ME, malic enzyme; NADP, (cofactor) nicotinamide adenine dinucleotide phosphate; NADPH, the reduced form of NADP. The dashed blue circles are used to indicate the main enzymes responsible for TAG synthesis.


33

growth phase. Nitrogen limitation results in deceleration of that synthesis and consequently, a

carbon flux is directed to synthesis of TAGs (Beopoulos et al., 2009). However nitrogen exists in

organic and inorganic nitrogen sources and the use of them has different effect on the growth

and lipid yield. Whether organic or inorganic nitrogen source is better for oil accumulation

probably depends on the microorganism and the metabolic pathway that nitrogen follows for its

conversion to fat-free biomass. According to Li et al., (2008), inorganic nitrogen sources are good

for biomass generation, while organic ones are good for lipid production. However, Poli (2014)

achieved higher oil concentration with inorganic nitrogen than organic (Poli et al., 2014). In

another case, higher oil content was achieved with organic nitrogen source (Evans and Ratledge,

1984). Nevertheless, in a commercial system, inorganic nitrogen compounds would generally be

preferred because they are cheaper.

An important, and very widely used, parameter for design of fermentation media for lipid

accumulation is the C/N ratio, which indicates the relationship between the carbon and nitrogen

amounts, even though as a value (ratio) does not have direct impact on lipid accumulation

because it is absolute amounts of carbon and nitrogen that affect the process. A particular ratio

can be achieved by different combinations of carbon and nitrogen. However, it is a parameter

which indicates how much higher the carbon is than the nitrogen. Usually, high C/N ratios are

required for lipid accumulating conditions while low C/N are applied during the growth phase.

Sulphur is another element which is responsible for lipid accumulation. In general, it is

used for the construction of proteins (amino acids such as cysteine and methionine) and other

cellular compounds such as coenzyme A and biotin (Wu et al., 2011; Scott et al., 2007). Sulphur

limitation has been reported to induce lipid accumulation too, although it is not a popular target

for experimentation. It was found that regardless of an excess of nitrogen in the medium, sulphur

limitation induced oil production in Rhodosporidium toruloides Y4 resulting in a content of 57%

(Wu et al., 2011). Gill et al. (1977) examined sulphur and magnesium limitation in cultures of

Candida 107 and found that although these nutrient limitations induce lipid accumulation, their

use as limiting nutrients is much less effective than nitrogen limitation (Gill et al., 1977).

Phosphorus limitation also appears to induce oil accumulation even with nitrogen

present in the medium. However, for that to happen, very high C/P ratios are needed.

Rhodotorula glutinis accumulated 20% oil under C/P=637 (molar ratio), Rhodosporidium

toruloides 21% at C/P≤495 (molar ratio) and the oil content increased at ratios higher than 2,045

(Wu et al., 2010). Cell growth was observed to decrease at a later point than observed with usual

nitrogen limitation.


34

Phosphorus, nitrogen, iron and carbon limitation were applied separately in cultures of

Rhodotorula glutinis by Granger et al, (1993). Nitrogen depletion resulted in the highest lipid

yield (31.1%), while carbon limitation gave the lowest biomass (Granger et al., 1993). After that,

the second highest lipid content (23.35% w/w) was achieved by iron (Fe) limitation followed by

phosphorus (P) (18.2%). All these different nutrient limitations gave differences (small or large)

in oil composition as well. For example, phosphorus limitation results in the formation of more

neutral than polar lipids (Wu et al., 2010). Differences occur amongst strains and due to the type

of carbon source. Rhodotorula species, for example, when cultured on glucose displayed 24.9%

in C16 fatty acid, 32.2 % in C18:0 and 24.8% in C18:1, compared to 38.2, 28, 7.4 when cultured

on xylose and glycerol 16, 21.9, and 18 respectively (Xu et al., 2013).

It is clear from the above that the absence of any essential component for growth will

result in lipid accumulation if there is available carbon. Thus the carbon/element ratios

themselves are not actually the defining factors. It is also mentioned in the literature that other

culture parameters such as pH, temperature, aeration rate and concentration of elements also

influence lipid accumulation (Li et al., 2008).

2.5.2 Culture conditions

Temperature affects lipid accumulation but to a lesser extent than does nutrient

limitation. It can affect the composition of the lipids and the degree of saturation of the fatty acids

(Saxena et al., 1998). A temperature drop from 30oC to 25oC increased the yield of α-linoleic acid

from 7 mg/g to 9 mg/g (Granger et al., 1993). Yarrowia lipolytica was cultivated in a temperature

range of 24-33oC and the maximum lipid content of 44% (3.8 g/L) was obtained at 28oC

(Papanikolaou et al., 2002a).

The pH affects both oil yield and composition. A common pH range for yeasts for oil

accumulation research is 3-6.5. Some yeasts can tolerate acidic environments and this is an

advantage for them against other oleaginous microorganisms. Yeasts have been reported to

grow from pH 2 to 9, with pH 6 being optimum for the growth of Yarrowia lipolytica

(Papanikolaou et al., 2002a). Xue et al., (2006) achieved highest oil content at pH 5.5 after

examining pH values from 4.5 to 8.

2.6 The role of oxygen in the metabolism of oleaginous yeasts

Oxygen plays an important role in the growth of oleaginous microorganisms, especially

yeasts. The majority of the cultures have a high demand for air and consequently experimental


35

systems are normally provided with air. The following sections provide detail of the role of

oxygen in growth and lipid production.

2.6.1 Oxygen transfer into the cell

For aerobic microorganisms and cultivations, oxygen is an important substrate for

cellular growth, maintenance and production of metabolites (Garcia-Ochoa and Gomez, 2009).

The microbial cells are suspended in the culture broth, which is an aqueous environment,

therefore oxygen needs to be dissolved in the liquid phase in order to be utilised by the cells.

Oxygen transport from the gaseous to the liquid phase follows the steps below, which are

schematically represented as shown in Figure 11:

1) Transfer from the gaseous phase to the gas-liquid interphase

2) Movement within the gas-liquid interphase

3) Transfer through the liquid film surrounding the gas bubble

4) Oxygen reaches the bulk liquid area where it is moving through

5) Prior to entering the cell, oxygen passes across a stagnant liquid film surrounding

the outer part of the cellular wall

6) It moves through the liquid-cell interphase

7) It enters the cytosol in order to reach reaction sites

The microbial oxygen demand depends on the concentration and presence of other

nutrients, toxic end-products which accumulate into the broth and the supply of oxygen. Oxygen

solubility in water and especially nutrient broths is low, therefore it needs to be supplied

Figure 11: Oxygen transfer from the gas phase to the cell, adapted from (Garcia-Ochoa and Gomez, 2009)


36

continuously. Moreover, oxygen solubility decreases with increasing temperature and in most

cases the cultivation temperature is above 25°C. The dissolved oxygen concentration is affected

by the oxygen transfer rate (OTR) and the oxygen uptake rate (OUR) of the cells. The uptake of

oxygen is clearly limited by its availability, so OTR should be greater than OUR to ensure

maximum uptake. The driving force for OTR is the difference between the actual dissolved

oxygen concentration and its equilibrium value. The rate of change of dissolved oxygen

concentration in the liquid state during cultivation is given by Equation 1:

XqDODOakdt

dCOL

2)(OUROTR * (2-1)

The oxygen transfer from gas to liquid phase is affected by, and can be improved by

increasing the agitation rate. Increasing agitation breaks the incoming air bubbles into smaller

ones which have larger surface area to facilitate the oxygen transfer. It also increases gas hold-

up, residence time as well as improving the quality of mixing. OTR, OUR and qO2 are very

important when designing fermentations and scaling up. They can be calculated experimentally

as will be seen in Chapter 6.

2.6.2 The effect of oxygen in oleaginous yeasts

Oxygen affects both yeast growth and lipid production but its role does not lead to

conclusive results in the latter. In general, oleaginous yeasts are aerobic microorganisms and oil

production culture systems are supplied by air, in contrast to the case of alcohol production.

Usually, high dissolved oxygen enhances the growth rate. Most frequently, high dissolved oxygen

concentration is achieved by high air flow rate and high agitation rate. It was found that increasing

the headspace in flasks benefited lipid production, while increasing the agitation speed did not

(Poli et al., 2014). Similarly, Choi et al. (1982) observed increasing trends in both biomass and oil

production (Choi et al., 1982). The growth rate of Rhodotorula glutinis in an airlift bioreactor was

enhanced by increasing aeration rate up to 2.5 vvm while the oil content slightly decreased beyond

2 vvm (Yen and Liu, 2014). In another study with Rhodotorula glutinis (Pan et al., 1986), a

remarkably high cellular density was obtained with a supply of oxygen enriched air. However, the

oil content was quite low compared to the normal air supply but the high cellular concentration

resulted in a high overall oil yield.


37

Oxygen requirement during the lipogenesis phase is low, due to a lower generation of

metabolic heat, while some metabolic routes do not take place (Pan et al., 1986). Moreover, the

oxygen concentration has been reported to affect the degree of fatty acids saturation. Higher

oxygen availability may increase the degree of unsaturation and the oxygen limitation caused

either by limited supply or by higher biomass densities may result in a decrease in the proportion

of saturated fatty acids (Ratledge and Hall, 1977). Apparently the effects of oxygen on lipid

reduction differ amongst strains. There are though common practices to increase the oil yield.

These consist of using either controlled oxygen levels in the broth by modifying the agitation rate

-which is the most common practice- or skip oxygen control and leave the oxygen conditions

unregulated. In a recent study, Rakicka et al (2015), examined three different conditions:

unregulated oxygenation; regulated at 50% oxygen saturation; and high oxygenation. The

approach that gave the optimal results was the unregulated oxygen conditions, since the agitation

rates applied did not manage to avoid low oxygen concentrations.

2.7 Conversion of microbial oil into biodiesel

Triacylglycerols are intracellular products and in order to be accessible for conversion to

biodiesel they need to be released from the cell. Then they are converted to fatty acid methyl

esters (biodiesel). Yeasts cells are constructed by a thick cell wall, which makes their disruption

a difficult procedure, compared to other types of microorganisms. If the cell disruption is not

efficient, the oil yield is underestimated. Triacylglycerol conversion to biodiesel takes place in

two ways: conventional transesterification from extracted oil and direct transesterification

where oil recovery and transesterification happen in one step. Prior to the conversion, the

cellular mass is dried in order to deactivate the cellular enzymes.

A) Conventional transesterification of microbial lipids

The conventional method consists of three steps:

1. At first, the dried biomass undergoes mechanical disruption by cell homogenisers

or non-mechanical disruption using organic solvents, freezing, or application of

osmotic shock.

2. The lipids are separated from the cell debris with liquid-liquid extraction using

organic solvents such as, mixture of chloroform-methanol, hexane or other

solvents according to the fraction of oil required. Usually applied methods for oil

extractions follow procedures previously established for isolation of fatty


38

material from fish or meat samples (e.g. Soxhlet, Folch or Bligh and Dyer) adjusted

now to microbial cells. These methods are used with only very slight

modifications and probably the choice depends on the amount of sample

available.

3. The extracted lipids are then converted to biodiesel by reaction with a short-chain

alcohol as the reactant (usually methanol) and acid (e.g. HCl) or base (NaOH) as

the catalyst.

B) Direct transesterification of microbial oils

The direct method consists of simultaneous cell disruption, oil extraction and

transesterification. Acid or base methanolysis have been performed in yeasts and fungi. Many

parameters affect the final yield, such as reaction temperature and catalyst concentration. The

temperatures used are similar to those in the conventional transesterification but direct base

methanolysis is performed under milder conditions and for shorter time than acid based

methanolysis. The direct transesterifications are reported to be more efficient, with higher yields

(>90%) than the indirect transesterification but it is important to ensure cell lysis and therefore

the reaction time has great importance on the final yield (Thliveros et al., 2014). A techno-

economic study on both conventional and direct transesterification at industrial scale has shown

that the indirect process is cheaper than the direct (Koutinas et al., 2014). Noteworthy is a recent

process which aims to convert lipids to biodiesel inside the microbial cell (in situ). It refers to

microorganisms, engineered to synthesize fatty acid ethyl esters after the modification of

ethanol producing genes and expression of enzymes that enter the fatty acid pathway and

produce biodiesel (Shi et al., 2011).

2.8 Industrialisation of microbial oil

Biodiesel production from microbial oil is still in its infancy due to the high cost of the

process, mainly because of the fermentation cost. In order for the oil production to be viable at

large scale, its production cost should be lower than its selling price so that profits can be

derived. At the moment, the price of microbial oil is not competitive with that of conventional

oils, as can be seen in Table 3, and petroleum derived fuels are economically more attractive. In

2008 Ratledge reported that the price of yeast oil would be more than 3,000 $/t, which is cheaper

than algal oil (Ratledge and Cohen, 2008). More recent calculations estimated microbial oil cost

at 3,400 $/t for a production capacity 10,000 t per year from glucose at $400 per tonne of oil


39

produced (Koutinas et al., 2014). In another study, using glucose as carbon source the production

cost was $2,350/t (Braunwald et al., 2014).

Table 3: Prices of several oil feedstocks.

Oil Cost ($/kg) Reference

Nutritional oil (GLA, DHA, APA, EPA) 50-150 (Huang et al., 2013) Edible plant oils (peanut, olive, rapeseed, soybean)

1.5-2.8 (Huang et al., 2013)

Yeast oil 3.4a (Koutinas et al., 2014) a calculated value from the reference

In general, the process of producing yeast oils consists of the following steps: selection of

substrate, selection of oleaginous species according to the substrate, fermentation and

equipment design and downstream processing, as depicted in Figure 12. Amongst them, the raw

material (carbon source) and the fermentation play important roles in the process economics.

The carbon source determines the end use of oil and a high substrate to oil conversion is

necessary to result in efficient utilisation of the substrate, while the fermentation stage should

achieve high oil productivities and growth rates. The selection of carbon source, includes the cost

of its purchase, the transportation to the fermentation site and any pre-treatment that it may

require in order to be fully utilisable by the microorganisms. In a techno-economic study for

biodiesel production from yeast oil, the cost of glucose constituted 80 % of the raw materials

cost (Koutinas et al., 2014), which was about 44% of the overall cost. Use of lignocellulosic

materials, as presented in a recent review accounted for 40% of the total cost of the production

of 1 t of microbial oil , while the cost of microbial oil was 7,500 RMB (around $1,230/t) (Huang

et al., 2013).

The substrate has influence on the cost of microbial oil production, but the major cause

of the high prices is the fermentation stage. In the techno-economic study of Koutinas et al., the

fermentation cost accounted for about 80% of the installed equipment cost. First of all, to ensure

steady yield, it is necessary to maintain mono-cultures and the fermentation media should be

sterile prior to the inoculation of the desired species. Control of pH also contributes to the cost

and aeration is an energy consuming process. Moreover, smaller vessels are required for the

propagation of the seed cultures until the final fermentation, which would take place in

fermenters of 100,000 to 250,000 L, with each one needing sterilisation and probably pH control

and aeration (Ratledge and Cohen, 2008). The utilities of the bioreactor part are also higher than

other utilities of the process (Koutinas et al., 2007) and consequently a large vessel would be

accompanied by higher equipment and operation costs. It is important to provide suitable


40

cultivation conditions for the maximisation of the yield and high oil contents and productivities.

For efficient conversion to oil, a big portion of the cellular mass should be oil because less

residual defatted biomass minimises the waste processing after the fermentation. Finally,

regarding the oil recovery stage, the use of organic solvents and waste processing are expensive,

though the extraction step has quite standard contribution to the process, which is at about 13%

(Braunwald et al., 2014; Huang et al., 2013).

Over the years, there has been a trade-off between microbial oil and vegetable oil,

influenced by price variations and governmental policies. The use of plant oils clashes with food

needs, and prices fluctuate, though lately they show an increasing trend. However, it is necessary

to make commercial production of yeast oils attractive and economically viable as well as

sustainable in order to facilitate the transition from fossil to bio-based economy. The economic

feasibility of microbial oils depends on strain and process improvements. Biochemical

engineering approaches are able to contribute with the implementation of major improvements

to the fermentation as well as other parts of the process. For example, the microbial oil

production process can be coupled with another existing process or bioprocess thus improving

the economics. An example might be using the glycerol side stream from biodiesel as the carbon

source (Uçkun Kiran et al., 2013). Consequently the use of low cost materials and minimum pre-

treatment steps, along with robust and contamination-resistant yeast strains, would be

desirable. Regarding the fermentation part, achieving faster growth rates and higher cell

Figure 12: Design considerations for microbial oil production.


41

densities as well as fuller conversion of substrates are the main targets. More efficient extraction

methods can increase the product recovery and valorisation or recycling of fermentation wastes

may also reduce part of the raw materials cost. These aspects are discussed in more detail below.

2.9 Process improvements for microbial oil production

2.9.1 Low-cost fermentation substrates

A significant amount of research is focusing on the carbon source by investigating the

potential of utilising low or negative cost materials as fermentation substrates. The ability of

yeasts to utilise waste materials and tolerate impurities offers the possibility of utilising waste

materials from industrial side-streams. Materials that have been used so far include waste

streams from the food industry (e.g. starch and whey), industrial by products such as crude

glycerol and thin stillage, and wastewaters (such as sewage or olive mill wastewater).

Energy crops and lignocellulosic biomass

The most commonly used carbon and energy source is glucose. Glucose is easily

assimilated by almost all microorganisms and the biomass and oil yields are considerable.

Energy crops (cassava, sweet sorghum) and lignocellulosic biomass (straw, bagasse) have been

investigated as low-cost sources of sugars. They contain large amounts of fermentable

polysaccharides and they can be cultivated in marginal land (Xu et al., 2013). Moreover,

lignocellulosic biomass as a substrate does not compete with the food chain, is abundant in

nature, renewable and has already been used in ethanol production. It is composed of lignin,

hemicellulose and cellulose and requires pre-treatment and hydrolysis in order to convert

cellulose into assimilable sugars (e.g. glucose, xylose, cellobiose) (Leiva-Candia et al., 2014). This

is a drawback from a financial point of view that can add to the substrate related costs.

Hydrolysis methods are usually chemical, with the use of acids, or enzymatic. During acid

hydrolysis, several inhibitory by-products such as acetic acid and furfural, can be created.

Biochemical approaches are focusing on the improvement of lignocellulosic biomass

conversion to oil by investigating two key points: fuller and simultaneous utilisation of the

released sugars (pentoses and hexoses) and inhibitor tolerance. The ability of several yeasts was

investigated amongst others by Chen et al. (2009) and Sitepu et al. (2014) leading to

identification of less known and newly isolated yeast strains with satisfactory response to the

degradation compounds. Further to the presence of inhibitors, their amount and combination

can have different effects on the cultivation parameters. Minor amounts of furfural (1mM)


42

inhibited by 45% the growth of Rhodosporidium toruloides, though the presence of six inhibitors

in the broth did not affect the oil yield (Hu et al., 2009). The mechanism of utilisation of the

released sugars is of interest since not all yeasts are able to consume all sugars and even if they

can, they do not usually do it simultaneously. It has been demonstrated that the presence of both

hexoses and pentoses affects the consumption rate of an individual sugar (Yu et al., 2014). The

potential of simultaneous utilisation of glucose and xylose is of critical importance and the lipid

yields obtained so far are promising (Hu et al., 2011). It is therefore important to improve the

hydrolytic procedures so they can provide a higher yield of sugars and also develop suitable

strains for higher conversion of these carbon sources.

Wastewaters as fermentation substrates for microbial oil

Wastewaters and effluents can be used as substrates for microbial oil production since

they are considered as negative cost materials because of their high organic content and

capability of severe environmental pollution (Yang and Zheng, 2014; Yang et al., 2005)). These

wastes are rich in nutrients, they contain high COD (chemical oxygen demand), possibly large

amounts of salt and have low pH and therefore aerobic yeasts are suitable for their utilisation.

Several researchers have investigated the potential of utilising effluents from domestic use or

food processing units on yeast oil production. When such streams are utilised as raw material

for microbial oil production, they gain value and the high costs for their treatment and disposal

are skipped. For instance, untreated wastewater from ABE fermentation, containing remaining

sugars, acids and ethanol induced 14 % oil production in Trichosporon cutaneum (Chi et al.,

2011). In another approach brewery effluents were utilised by Rhodotorula glutinis for lipids and

carotenoids production (Schneider et al., 2013). Olive mill wastewaters were the substrate of

Trichosporon cutaneum without any supplementary nutrients (Yousuf et al., 2010). Municipal

wastewater has been examined on several oleaginous yeasts in order to find a suitable yeast

strain (Chi et al., 2011). Cultivation of oleaginous yeasts on such materials appears to be more

beneficial for growth than oil accumulation, although the yields are quite low (14-30%) due to

the nitrogen content and the fast depletion of carbon.

Food processing waste

By-products of the food industry are characterised by high nutritional value because of

their sugar and protein content and require treatment for safe disposal. Such wastes include

cheese whey (Vamvakaki et al., 2010), orange peel (Čertík et al., 2012), sweet potato starch (Li


43

et al., 2008) and sweet sorghum (Economou et al., 2011). Traditionally, such materials are used

for animal feed but they constitute a promising option as feedstock for microbial lipids since they

are available and have been successfully used for microbial oil production by oleaginous yeasts

and fungi (Huang et al., 2013). Sometimes though, the nitrogen content is much more than that

required for nitrogen limited conditions given a certain amount of substrate and might need

external supplement of carbon source.

Nitrogen sources from other processing streams

Rapeseed meal is the solid residue from rapeseed oil extraction during biodiesel

production. This residue is rich in nitrogen and although it is not an ideal animal feed and

fertiliser due to some toxic compounds it contains, it has been used in oleaginous yeast

fermentation following microbial treatment to release the nutrients (Uçkun Kiran et al., 2012).

Utilisation of rapeseed meal for microbial oil production can further reduce the cost by coupling

microbial oil production from biodiesel wastes in Europe. Tomato waste hydrolysate from

tomato processing units was used as supplementary nitrogen source in an oleaginous fungi

(Fakas et al., 2008)

Utilising biorefinery wastes for microbial oil production

Crude glycerol

The growth in biodiesel production over recent years has led to a surplus of crude

glycerol, which is unsuitable to be used directly in other applications and is treated as a high

strength polluting waste. This glycerol glut has resulted in a drop in the price of glycerol and

affects negatively the viability of biodiesel. Therefore, if crude glycerol could be used as low-cost

carbon source for microbial oil by oleaginous yeasts it would be of significant benefit to the

industry.

Background on glycerol

Glycerol (or 1,2,3-propanetriol, glycerine) is a clear, odourless and viscous alcohol with

the chemical formula C3H8O3. It has a molecular weight of 92.09 g/L, density 1.26 g/L (20°C) and

because of its three hydroxyl groups it is very soluble in water and other aqueous solutions

(Pagliaro and Rossi, 2010) . Glycerol is a structural component of lipids and can be derived by

hydrolysis of oils and fats, from saponification units and biodiesel and bioethanol production

(Tan et al., 2013). It is widely used in the cosmetic, pharmaceutical and food industry and other


44

chemicals by chemical or biochemical means (Wang et al., 2001; Johnson and Taconi, 2007;

Koutinas et al., 2008).

Currently the products that are produced chemically with catalytic conversions of

glycerol (reforming, oxidation, dehydration) are propylene glycol, propionic acid, propanol,

isopropanol, acrolein and hydrogen (Panagiotopoulou et al., 2013; Johnson and Taconi, 2007).

Catalytic conversions of glycerol require high temperatures and pressures. On the other hand,

biochemical processing of glycerol offers the same products and other biodegradable

compounds under benign transformation processes (Koutinas et al., 2007). Products that can be

obtained by microbial fermentation and were usually produced with petrochemical ways are:

1,3 propanediol, 2,3 butanediol, succinic acid, propionic acid, pigments, citric acid, lipids and

biopolymers and are shown in Figure 13 (Luo et al., 2016).

Glycerol is taken up by the cell in three ways: passive diffusion, facilitated diffusion by a

membrane protein or active uptake mechanism (da Silva et al., 2009; Lages et al., 1999). Then it

can be metabolised aerobically or anaerobically.

Characteristics of crude glycerol

When glycerol is released during the synthesis of biodiesel, it is contaminated by

methanol, salts, water and other materials as a result of the transesterification reaction and

biodiesel recovery efficiency. This mixture is generally referred to as crude glycerol. During the

production of biodiesel, fatty material reacts with methanol in the presence of an acid

Figure 13: Potential bio-products from glycerol, summarised from (Luo et al., 2016).


45

(hydrochloric, sulfuric) or base (potassium, sodium hydroxide) catalyst. The composition of

crude glycerol varies according to the oil feedstock used, the process followed (type of catalyst,

efficiency of reaction) and the treatment of the resulting mixture(Luo et al., 2016). For instance,

the glycerol fraction in crude glycerol can vary between 50-90 %. Crude glycerol derived from

soybean oil, for example, contained (w/w) 62% glycerol, 12.8% methanol, and 25.2% soap (Luo

et al., 2016). Various crude glycerol compositions, from different processes are presented in

Table 4.

Crude glycerol is a low-quality material with limited applications without purification

and the extent of impurities affects the range of applications in industry. Purification takes place

with distillation of this glycerol stream and is considered an expensive process (Ayoub and

Abdullah, 2012). The less pure the glycerol is, the more expensive its purification can be. The

continuously increasing biodiesel production has resulted in an oversupply of crude glycerol,

which has made its purification costly and not economically worthy for either producers or

clients (Dobson et al., 2012). The production is higher than the demand and so the crude glycerol

is often burnt or used as animal feed (Luo et al., 2016). This market glut results in even lower

prices of both crude and refined glycerol turning it from useful co-product to waste material

(Vlysidis et al., 2011). The drop in glycerol prices has direct impact on the economics of biodiesel,

according to a process model for estimation of biodiesel costs (Haas et al., 2006).

Table 4: Various compositions of crude glycerol from different transesterification processes.

Oil feedstock Glycerol (w/w)

Methanol (w/w)

Soap (w/w)

Water (w/w)

Salts (w/w)

Reference

Sunflower 30 50 13 2 2-3 (Luo et al., 2016) Jatropha 18-22 14.5 29 10 n.r. (Hiremath et al., 2011) Soybean 63 6.2 n.r. n.r. n.r. (Luo et al., 2016) Canola 56.5 28.3 15.3 n.r n.r (Luo et al., 2016) Rapeseed 81 <0.1 n.r. 10-12 5-6 (Chatzifragkou et al., 2011)

n.r.: non reported

Glycerol as a carbon source for microbial oil production

Utilisation of these glycerol-rich side streams in bioprocesses constitutes an outstanding

opportunity for integrated biorefinery processes achieving waste minimisation and utilisation

of all residues and co-products in other production lines. Additionally, single cell oil production

would benefit from a year-round supply of substrate allowing the plant to operate at full capacity

without need for special long-term storage for the raw materials. In this context, utilisation of

crude glycerol from biodiesel biorefinery enhances the biodiesel production chain and the


46

economical production of microbial oil if coupled with an existing process. Table 5, summarises

basic aspects of glycerol as fermentation substrate.

Previous works using pure and crude glycerol have demonstrated that this low-cost

material can be a suitable carbon source for microbial lipid production under appropriate

cultivation conditions (Meesters et al., 1996; Papanikolaou and Aggelis, 2002). Effective glycerol

utilisation is subject to the following considerations. To trigger microbial oil production is

necessary to apply carbon excess over nitrogen limitation. Therefore higher glycerol

concentrations are required to construct a high C/N ratio. Several studies have performed

screening of different glycerol amounts in batch mode to study the effect of low and high

concentrations and identify the optimum C/N ratio for their system. Glycerol may cause growth

inhibition due to osmotic stress on the cell membrane.

First, it is imperative to identify and develop species that are capable of assimilating

glycerol (thus crude glycerol) and several yeasts have been selected by taking into account their

ability to metabolise glycerol (Souza et al., 2014; Petrik et al., 2013). For example, Kitcha and

Cheirsilp (2011) focused their study of screening 889 newly isolated yeasts from wastes on their

ability to consume glycerol and selected two (Trichosporonoides spathulata and Kondamaea

ohmeri) who had high lipid content and good growth (Kitcha and Cheirsilp, 2011). Glycerol has

also been used alongside with sugars and has even been found to be superior (Yen et al., 2015b;

Easterling et al., 2009). The effects of glycerol concentration can be examined as well as the

differences between pure and crude glycerol. The effect of crude glycerol concentration on the

growth and lipid productivity of T. spathulata was then evaluated (Kitcha and Cheirsilp, 2013).

Growth and lipid inhibition was observed after 100 g/L glycerol. Many other studies have

examined the glycerol assimilating capacity of well-known or newly isolated yeasts. In another

approach (Liang et al., 2010), the growth and lipid production of the oleaginous yeast was first

evaluated on pure and crude glycerol and then the effect of crude glycerol concentrations

ranging from 20 to 80 g/L was examined. In this case glycerol inhibition appeared beyond 60

g/L. The critical concentration for inhibition was different between pure and crude glycerol with

that of crude being lower. The clear effect of glycerol might be strain dependent or even affected

by the cultivation conditions, however the inhibition can be avoided by manipulation of the

cultivation conditions such as fed-batch or continuous fermentation instead of batch. Therefore

developing fermentation modes is important in order to utilise effectively large amounts of

glycerol without compromising the yields.


47

Table 5: Glycerol as a fermentation substrate

Advantages Disadvantages Points to improve

Cheap Presence of inhibitory compounds in crude glycerol

Identification of species that consume glycerol

Year-round availability

Not all microorganisms consume it

Adaptation of oleaginous species

Oxidative or reductive metabolism

Growth inhibition Identification of inhibitory concentration and performance of fed-batch or continuous fermentations for moderate supply of glycerol

2.9.2 Developing cultivation modes for efficient microbial oil yield

In addition to the carbon source, the cultivation conditions and mode have a big effect on

the oil yield. Therefore a proper midstream design is important to achieve effective utilisation of

the raw materials while maximising the yields (Pinzi et al., 2014; Anschau et al., 2014; Kitcha and

Cheirsilp, 2013; Thiru et al., 2011). Such a design should take into consideration the lipogenesis

conditions. In particular, cell proliferation is the main process when carbon and nitrogen are

available in the medium. Upon the exhaustion of nitrogen, carbon is still taken up by the cells but

is used in lipid synthesis rather than cell proliferation. Therefore a high C/N ratio, a combination

of excess carbon and low nitrogen would be a prerequisite for lipid synthesis. However this is

likely to cause growth and oil inhibition due to the high initial concentration of carbon source in

a system operating in batch mode. Several cultivation modes, used for oleaginous yeasts, are

summarised in Table 6.

In principle, batch fermentation is a closed system, where all the reaction components

are supplied at the beginning and the reaction takes place without external intervention (Bisen,

2013). Batch mode is the basic and most studied method for oil production and can be performed

either in flasks with uncontrolled culture conditions (pH, aeration) or bioreactors, where these

culture parameters can be controlled. The initial high C/N ratio cannot be controlled after the

initiation of the culture. The yeasts will continue growing until the exhaustion of nitrogen so the

residual C/N continuously increases until carbon gets consumed. The high initial C/N requires

large amounts of carbon source added at the beginning and this high substrate concentration

can sometimes lead to inhibition of growth, stressing conditions and discontinuity on cell and oil

yields (Christophe et al., 2012). Moreover due to the small amount of nitrogen provided, the

yields in cellular mass can be low.

Fed-batch cultivation has long been considered as an effective mode for promoting

cellular yields. With fed-batch fermentation, substrate inhibition can be avoided and a large

amount of carbon is eventually fed by the end of the fermentation. In this way higher cell


48

densities and oil concentrations can be achieved (Anschau et al., 2014). Fed-batch cultivation of

Candida freuschussii resulted in a 6-fold increase in the oil concentration while the oil

productivity was twice as high as that of batch fermentation (Raimondi et al., 2014). Clearly, fed-

batch fermentation has advantage over the simple batch mode. Nevertheless, a specific amount

of nitrogen must be provided to support cell density and make the oil yield meaningful and worth

applying at large scale. Hence, fermentations consisting of two stages, where at first biomass is

produced and then oil accumulation takes place provide a potential solution. A number of

cultivation modes have been developed based on this two-stage approach and some of them are

shown in Table 6.

Two stage batch cultivations

It is worth reviewing the case of two-stage cultivation by combining separate batch

cultivations, using cells from the first batch as concentrated inoculum for another batch

(Slininger et al., 2016; Lin et al., 2014). In this case, growth promoting media are supplied in the

first cultivation. For instance, Rhodosporidium toruloides AS 2.1389 was cultivated in glucose-

containing media with low C/N ratio (C/N=15) and after 4 days the cells were re-suspended in

acetic acid-containing media under the C/N of 200 (Huang et al., 2016). In this way a preferred

carbon source such as glucose can be used for biomass production and a possibly inhibitory

carbon source, such as acetic acid can be used as substrate when a certain level of biomass has

been achieved. However, in that study, they noticed a lag-phase after the inoculation of the cells

to the acetic acid media, which limited somehow the final productivity. In a very similar

approach, at bioreactor level, cells of Lipomyces starkeyi AS 2.1560 cultivated on glucose at a 15-

L bioreactor with glucose based media were re-suspended in a 7-L bioreactor containing only

glucose as the sole nutrient (Lin et al., 2011). This method resulted in an increase of 83.5 g/L and

64.7 g/L in cellular mass and lipid concentration respectively. However, these techniques

require two vessels at least for the making of the two stages. Such a case is not easily applicable

in large scale because it requires unloading of the previous vessel and loading of the second if

two different vessels are used, where in such a case many bioreactors are required. If the same

bioreactor is used throughout the process, cleaning of the first vessel is needed in order to be

filled with the new media. It would need some time where no cultivation is performed and

includes the risk of contamination as well as being potentially uneconomic in process terms.


49

Two stage fed-batch cultivations

A two-stage design can be performed with fed-batch or continuous fermentations, where

the two stages can be modified easily by altering the nutrient supply. Previous studies have

shown that addition of just the carbon source during the final feedings of a fed-batch cultivation

are able to boost the lipid yield. Fontanille et al. (2012) employed volatile fatty acids (VFAs) as

carbon source for the second stage while glycerol or glucose were supplied in the first stage. In

order to maintain a constant C/N ratio throughout the cultivation ammonium sulphate was

supplied at a basal level. The principle behind this approach was the bioconversion of VFAs when

enough biomass had been obtained (Fontanille et al., 2012). In another approach glycerol

additions took place after a 30-h batch cultivation, to reinstate glycerol concentration at 40 g/L

when previously added glycerol was consumed in order to boost lipid synthesis (Raimondi et al.,

2014). Many other strategies followed the approach of pulse-feeding carbon source (glucose or

glycerol) only when its concentration had fallen below a critical value to maintain it above a

certain level without nitrogen supply (Thiru et al., 2011; Zhang et al., 2011; Zhao et al., 2010a).

A comparison was made between one-stage and two-stage cultivation of

Trichosporonoides spathulata on crude glycerol (Kitcha and Cheirsilp, 2013). After a period of

batch cultivation, crude glycerol and ammonium sulphate were pulse fed (one-stage), enhancing

biomass production, while glycerol-only pulses (two-stage) gave higher lipid content and

slightly lower cell concentration. A combination of these two approaches could be very

advantageous by feeding rich medium prior to the carbon source supply, increasing in this way

the cell density to obtain higher oil yields. Such an approach was applied in a cultivation of

Rhodotorula glutinis on corncob hydrolysate (Liu et al., 2015). The two-stage fermentation, with

pulses of detoxified hydrolysate along with nitrogen, followed by carbon source alone, in this

case undetoxified hydrolysate, resulted in an increase of 21% in the oil content and despite the

slightly lower biomass concentration, higher lipid concentration was obtained compared to that

of one-stage cultivation with carbon and nitrogen supply.

Fed-batch fermentations with continuous nutrient supply

Some studies mention the importance of the substrate feeding mode on the

productivities. An appropriate substrate feeding makes the conversion of the carbon source to

lipids more efficient. Nevertheless, the number of studies investigating different substrate

feeding rates in a fed-batch cultivation are to-date limited. The effect of pulsed and continuous

glucose feedings on the growth and lipid productivity of Rhodosporidium toruloides Y4 were


50

examined (Zhao et al., 2010a). The application of continuous glucose feeding, which maintained

the glucose in the broth at low levels (5 g/L) improved, by 43% and 5%, the DCW and lipid

content compared to glucose pulses. Yen et al. (2015a) examined the effect of pulsed, constant

and exponential feeding of glycerol during fed-batch cultivation of Rhodotorula glutinis. Although

the exponential feeding gave the highest growth rate, DCW, lipid content and productivities were

higher in the cases of pulsed and constant feedings (Yen et al., 2015a). Continuous supply of

nutrients avoids inhibition caused by a large amount of substrate and cells are always provided

with medium. This is in contrast to a pulsed feeding where the nutrients might be completely

consumed before the next injection, resulting in lower yields (Raimondi et al., 2014).

Repeated batch and fed-batch cultures

Another interesting fermentation mode, which targets increasing cell concentration to

achieve higher substrate conversion and lipid yield, is the repeated batch cultivation. In this

approach, cells from one cycle are used as inoculum for the next cycle in the same bioreactor

vessel, to provide a very large inoculum and reduce downtime (for cleaning etc.). Lag phase is

reduced thanks to the concentrated cell suspension and the removal of the majority of

potentially toxic metabolic products from the previous cycle. These cultivations result in very

high cell densities and high lipid contents (Huang et al., 2016). However, they take longer (~238

h) and the prolonged time can result in a decrease in the cellular activity, as has been observed

in the case of Rhodosporidium toruloides Y4 on glucose (Zhao et al., 2010a).

Continuous fermentations

With continuous fermentation it is possible to maintain a steady substrate concentration

for a period of time and also still perform a two stage scheme. The advantage of the continuous

mode is that when both carbon and nitrogen source are supplied, a constant C/N ratio can be

maintained during steady state (Christophe et al., 2012). The main factor affecting the

productivities in a continuous system is the dilution rate, D, which is the inverse of the residence

time. Papanikolaou et al. (2002), tested dilution rates from 0.02 to 0.13 h-1 and detected higher

glycerol amounts in the broth at higher dilution rates, while lipid synthesis was favoured at low

dilution rates. Findings so far have shown that high dilution ratios favour cellular growth but not

oil accumulation and lipid-free cellular mass is benefited (Shen et al., 2013a; Anschau et al.,

2014). Another study showed that at a low C/N ratio of 10, the lipid and cellular yields are mainly


51

steady over the range of D from 0.01 to 0.22 h-1, while at a C/N of 50, the yields were reduced

when D was greater than 0.06 h-1 (Béligon et al., 2016).

Synergistic co-cultivation of oleaginous microorganisms

Recently, co-cultures of yeast and microalgae were applied at laboratory scale in order to

benefit from the potentially synergistic characteristics of the two systems. The yeast growth

results in CO2 release, which the microalgae use and produce O2 for the yeast. The co-culture is

not an easy case because there are many factors to be taken into account, such as the pH of the

culture, which can be acidified by the yeast growth, the different optimal temperatures for yeast

and algal growth and the matching of the growth rates (microalgae grow slower than yeast) and

there might be an imbalance on the production and utilisation of gases. By engineering these

parameters properly, suitable bioreactor design can be achieved, such as including membranes

permeable to gases or nutrients. One cultivation approach consisted of co-culture of the yeast

Rhodotorula glutinis and the microalga Chlorella vulgaris in a double system bubble column

photobioreactor (Zhang et al., 2014a).

Bioreactor configurations

Apart from the different types of fermentation, efforts to reduce the cost for scale up

include the use of particular bioreactors, such as airlift systems, which are said to have lower

operating cost than conventional stirred reactors (Yen and Liu, 2014) or open non-sterile ponds

(Santamauro et al., 2014). However, the most commonly utilised for oleaginous yeast cultivation

is the conventional stirred tank bioreactor.

2.10 Robust cultivation conditions

At large scale pH control increases the production cost. Use of microorganisms, especially

yeasts that can tolerate pH changes are desirable (Sitepu et al., 2014). Moreover, alkali addition

would inevitably result in accumulation inside the broth and complicate the final separation

processes. The reduction in pH can make the broth unfriendly to external microorganisms, as

was shown by a study, where the oleaginous yeast Metschnikowia pulcherrima was cultivated in

an open tank and the low pH along with the antimicrobial compounds it produced, ensured a

reasonable level of monoculture (Santamauro et al., 2014). If yeasts that do not need pH control

are utilised, these strains would be robust enough to require little external supply, resulting in a

simple fermentation process easier to scale-up.


52

2.11 By-products from oil production

Other metabolic products than oil consume part of the carbon source for their formation.

Some are produced by default and some because of the operating conditions, so these pathways

could be modified by either modification of the operating conditions or metabolic engineering in

order to avoid their formation. One of them, CO2, is a standard respiration product. The next, is

the fat-free biomass carbon content, which is not actually a by-product but is composed of

carbohydrates and proteins. Some yeasts can synthesise citric acid (Yarrowia lipolytica) from the

TCA cycle. A very common case is the production of pigments by various yeasts which are in

general valuable materials and their formation conditions are similar to those for oil

accumulation (e.g. nitrogen limitation). There is also possibility of cells to produce volatile

organic acids under low dissolved oxygen conditions, a fact that can be seen through pH decrease

(Zhang et al., 2014b; Beopoulos and Nicaud, 2012). In some cases the by-products are of value

and if they cannot be avoided they can further valorise the process through co-production of

value added products. Since oil recovery requires cell disruption, in all cases defatted biomass is

a waste that needs disposal, however it can serve as a nutrient source (e.g. fermentation

feedstock or animal feed). Researchers have explored the possibilities of feeding de-oiled

biomass back to another culture of oleaginous yeasts and this has brought up interesting results

(Thiru et al., 2011; Yang et al., 2015).

2.12 Concluding remarks

Several aspects of microbial oil production and its scalability potential were reviewed in

this Chapter. Although it is a potentially very useful product, its expansion and industrial

application are hindered by the high cost of its production. This can be attributed to three main

points: microorganism efficiency, substrate cost and productivities. The choice of substrate is

the most flexible parameter, since it can be selected according to the resources of a specific area.

However the bioreaction should ensure that substrate is not wasted but is utilised efficiently.

Further investigation is required to select a suitably robust microorganism, define the culture

conditions and develop the operational mode based on research. For this reason, the present

research has been focused on operational aspects of oleaginous yeast cultivation by developing

and exploring different fermentation modes.

Table 6: Cultivation modes using oleaginous yeast for microbial oil production.

Cultivation mode Principle Vessel Yeast C-source DCW (g/L)

Oil content (% w/w)

Oil (g/L)

Productivity (g/L/h)

Reference

One-stage

Batch Supply of nutrient medium at the beginning

Shake-flask Yarrowia lipolytica ACA-YC 5033

Crude glycerol 6.5 31 2 0.005 (Andre et al., 2009)

Batch Supply of nutrient medium at the beginning

2-L bioreactor Candida freyschussii ATCC 18737

Pure glycerol 15.6 32 4.7 0.16 (Raimondi

et al., 2014)

Fed-batch Pulses of carbon and nitrogen source

5-L bioreactor Trichosporonoides spathulata

Crude glycerol 17.3 41.9 7.25 0.05 (Kitcha and

Cheirsilp 2013)

Fed-batch Pulses of carbon and nitrogen source (constant C/N ratio)

2-L bioreactor Rhodotorula glutinis TISTR 5159

Crude glycerol 10.05 60.7 6.1 0.116 (Saenge et al., 2011)

Continuous Steady dilution rate 0.02 3-L bioreactor Rhodosporidium. toruloides AS 2.1389

Glucose 8.67 61.8 5.36 - (Shen et al.,

2013)

Two-stage

Batch Resuspension to the new vessel

Shake-flask Rhodosporidium toruloides AS 2.1389

1st stage: Glucose 2nd stage: Acetic acid

6.75 50.1 3.38 0.01 (Huang et al., 2016)

Fed-batch Addition of carbon source after 2 days of fermentation

Shake-flask Rhodosporidium toruloides Y2

1st stage: Bioethanol wastewater (sugars) 2nd stage: Glucose

85.5 35.2 30.1 0.14 (Zhou et al.,

2013)

Fed-batch C/N 30 at 1st stage No N at 2nd stage

2-L bioreactor Cryptococcus curvatus ATCC 20509

Crude glycerol 32.9 52 17.1 0.06 (Liang et al.,

2010)

Fed-batch No N in 2nd stage 5-L bioreactor Rhodotorula glutinis CGMCC 2.703

Corncob hydrolysate (sugars)

70.8 47.2 33.5 0.17 (Liu et al.,

2015)

Step-wise fed-batch

Change of vessel Growth medium at 1st stage, carbon source at 2nd stage

1st stage:15-L bioreactor

2nd stage 7-L bioreactor

Lipomyces starkeyi AS 2.1560

Glucose at both stages 104.6 64.9 67.9 1.6 (Lin et al.,

2011)

Continuous Higher carbon concentration at 2nd stage

5-L bioreactor Yarrowia lipolytica JMY 4086

Glycerol 59.8 40 24.2 0.43 (Rakicka et al., 2015)


54

55

Chapter 3

Research objectives and experimental

programme


56

3 Research objectives and experimental programme

3.1 Key objectives of the project

From the previous chapter, it is evident that microbial oil could be a very promising

alternative oil feedstock and could be produced at large scale given improvements and

optimisation of the process, particularly the upstream and midstream steps. The bottlenecks to

industrial microbial oil production are the substrate cost and the cultivation stage. The present

thesis therefore aims to contribute to the field of microbial oil production by developing and

investigating a simple process for oleaginous yeast cultivation, focusing mainly on enhancing the

yields during the bioreactor stage. Moreover, there is potential for enhancing the biodiesel

industry by coupling this process with biodiesel waste processing. By utilising glycerol as the

carbon source within an integrated biorefinery, where all participating materials are considered

valuable, it is possible to create a closed loop (Figure 14). Advanced and novel uses of glycerol

as a biological feedstock within a biorefinery approach can stimulate the biodiesel economy and

environmental sustainability towards a zero-waste manufacturing society.

To tackle the glycerol surplus and benefit from its current low price and worldwide year-

round availability, glycerol was selected as the sole carbon source for microbial production of oil

in this work. In the previous chapter, it was evident that yeasts are robust microorganisms, easy

to handle, as well as able to tolerate impurities in their substrates and with high lipid capabilities.

Therefore, oleaginous yeasts were chosen as the class of microorganism for the bioconversion

of glycerol to oil. Within this group, there are many potentially suitable strains. A preliminary

objective of this project was, therefore, to examine various yeast strains for their suitability and

select one for further study.

Having made these selections, it was necessary to focus on improving the midstream

stage (cultivation). Constraints for microbial oil scale-up lay in the fermentation stage. Although

new species identification, and implementation of cheap substrates are, of course, promising

steps towards this, the optimisation of existing bioprocesses is the core of the work required to

enhance the viability of microbial oil production. Good understanding of the phenomena taking

place is necessary to build a process with better glycerol uptake, minimisation of the residual

amounts and higher biomass yields to make the productivity meaningful. It is more important to

target high cell densities before oil contents because the greater the number of cells, the better

the oil yield. In this research, the main goal was the creation of a realistic design of a simple but

robust process with minimal control requirements, using oleaginous yeast. In order to achieve

this goal, it was necessary to perform experiments and develop a macroscopic model, after

Chapter 3 Research objectives and experimental programme

57

obtaining the initial experimental data. The model can then be used to describe and predict the

system for general use and scale-up. Exploring the system experimentally and developing a

suitable model were, therefore, key objectives of the research.

Several fed-batch approaches to the production of microbial oil are possible. Another

objective of the research was, therefore, to investigate various modes of operation of the

bioreaction step.

3.2 Experimental programme

Based on the information gathered from the literature and the experience gained through

analysing and evaluating this, an experimental programme was designed. If successful, this

experimental programme would enable the achievement of the various objectives described

above. Key stages in the experimental programme were as follows:

1. Obtain some oleaginous yeast strains, grow them on glycerol and select a suitable

one based on its growth, oil production and handling requirements.

2. Define the main cultivation parameters for this specific yeast strain and obtain

experimental data for the model construction.

Figure 14: Microbial oil production coupled with biodiesel production within an integrated biorefinery (Karamerou et al., 2016a).


58

3. Perform batch cultivations to investigate the effect of carbon and nitrogen source

concentration and once these are set, investigate the role of air supply in the

biomass and oil yields at bioreactor level.

4. Explore the possibility of enhancing the batch cultivation by performing fed-batch

mode and evaluate the difference between a two-stage and a single stage fed-

batch fermentation.

5. Investigate under fed-batch mode the role of feeding on the bioconversion yields.

3.2.1 Strain selection

The first part of the study consisted of evaluating the growth of several oleaginous yeasts

on glycerol. The aim of that stage was to identify an oleaginous yeast strain able to metabolise

glycerol to a satisfactory level. For that reason, the strains, which were initially cultivated on

glucose, were examined first under cultivation on both glucose and glycerol and then under

nitrogen limited conditions with glycerol as the sole carbon source. Rh. glutinis was the

oleaginous strain selected for further research. The fact that this yeast is also a well-known

carotenoid producer and a lot of literature is available on that, contributed to its selection as the

oleaginous yeast for further study. The specific experimental methods related to this research

are given in detail in Chapter 5.

3.2.2 Growth of oleaginous yeast Rh. glutinis on glycerol

The next two stages of the experimental programme involved growing the chosen strain.

After selecting a suitable yeast strain, it was necessary to identify which conditions were best for

its growth under nitrogen rich conditions. For this purpose, nitrogen sources were tested, initial

pH and flask level cultivations with air supply were performed. Then, the study was extended to

nitrogen limited batches to identify the effect of the initial glycerol, the initial nitrogen

concentration and different aeration conditions. The findings from these experiments would

inform scale up to bioreactor level. Specific experimental methods related to this research are

given in Chapter 6 and 7.

3.2.3 Model development

To make best use of the experimental results obtained from the above studies, an

unstructured model was constructed based on modified Monod kinetics. This was tested against

Chapter 3 Research objectives and experimental programme

59

experimental data from flasks and then used to optimise the kinetic parameters via stochastic

and deterministic optimisation. Results of the modelling work can be found in Chapter 8.

3.2.4 Bioreactor cultivations for mode of operation development

In order to understand the effect of oxygen on the oleaginous yeast cultivation, batch

fermentations under different aeration rates were performed. Then, the optimum aeration rate

would be used as a fixed parameter for evaluation of feeding modes using fed-batch cultivation

experiments. These aimed at defining the role of feeding modes on the cellular growth.

Details of the materials and methods used in for the above experiments are detailed in

the following chapter.


60

61

Chapter 4

Materials and methods


62

4 Materials and methods

4.1 Introduction

The main materials and methods used in this research are presented in this section. Due

to the different type of work presented in each chapter, specific equipment, methods and media

are given within the chapter in which the relevant results are presented.

4.2 Microorganism

The oleaginous yeast used Rhodotorula glutinis CICC 31596 (Figure 15), obtained from

the Centre for Industrial Culture Collection (China) was maintained at 4ºC in YPD stock vials. The

strain was subsequently cultivated on YP agar plates (10 g/L Peptone, 10 g/L yeast extract and

15 g/L agar) every month and maintained at 4°C. Other oleaginous yeasts that were also used in

this research as part of the strain selection study are described in Chapter 5.

4.3 Inoculum preparation

Prior to inoculation of a seed culture the strain was incubated for 4 days in a Petri dish

with YPGly media at 30°C. Prior to the inoculation of the seed culture a loopful of Rh. glutinis

from that Petri dish was transferred to a 500 mL Erlenmeyer flask filled with 100 mL medium

with composition 20 g/L glycerol and 10 g/L yeast extract at pH 5.5. The flask was incubated at

30°C at 200 rpm. From that seed culture a 10% v/v inoculum was transferred to the production

media unless otherwise stated.

Figure 15: Colonies from Petri dishes: Rh. glutinis 31596 (photo taken from above with the lid closed) (left), Cells of Rh. glutinis 31596 under light microscope with 100x objective lens (right), bar represents 10 μm.

10 µm

Chapter 4 Materials and methods

63

4.4 Analytical methods

All assays were analysed in triplicate with SD less than 10% unless otherwise stated.

4.4.1 Cellular growth

Cellular growth was monitored using Optical Density (OD) in a Shimadzu

Spectrophotometer (Shimadzu, UV mini-1240, Japan). For the dry cell mass determination 2 mL

of sample were centrifuged at 13,000 rpm for 5 min. The resulting supernatant was removed

and the cell pellet was washed with distilled water and dried for 24 h at 60°C in pre-dried

aluminium weighing disks (WVR, UK).

(L) VolumeSample

(g)disk driedofWeight(g) cellswithdiskdriedofWeight(g/L)DCW

(4-1)

Optical Density and Dry Cell Weight have been correlated with a calibration curve which

allows conversion of one measurement to the other. The Calibration equation is:

2466062572 .DCW. OD , R2=0.998 (4-2)

4.4.2 HPLC analysis

Glycerol concentration and the extracellular products were determined using HPLC

(Dionex Ultimate 3000, USA) equipped with a Refractive Index detector (Refractomax 521) and

an AMINEX HPX-87H column (BIORAD, USA). As mobile phase an aqueous solution of 5

mMH2SO4 (Sigma-Aldrich, 98.0%) was used at a flow rate of 0.6 mL/min. For this analysis, cell-

free samples were diluted appropriately and filtered through 0.45 µm cellulose filters (Millipore,

UK).

4.4.3 Total Nitrogen

Total nitrogen (TN) in the broth was estimated using a Total Organic Carbon analyser

(TOC-V series, Shimadzu) coupled with a TN detection unit. Zero grade air was used as carrier

gas at a flow rate of 150 mL/min to transfer the sample to a combustion chamber where N was

catalytically converted to NO. Then, a chemiluminescence detector was used to measure the NO

concentration, which was converted to mg/L of TN with the use of a calibration curve.


64

4.4.4 Oil content (by extraction)

Total cellular lipids were extracted chemically by following the Soxhlet extraction

method, previously set up in our laboratory (Thliveros et al., 2014), in a Soxtec-HT6 System

(Hӧgӓnӓs, Sweden). Briefly, culture broth was centrifuged at 9,500 rpm for 10 min and washed

with distilled water to obtain a medium-free cell pellet. This pellet was dried for 24 h at 60°C and

stored in tightly sealed containers until the analysis. The cell disruption was performed with

manual grinding using mortar and pestle with intermittent supply of liquid nitrogen to facilitate

the disruption. The procedure is schematically represented in Figure 16. The accuracy of

disruption was tested using an optical microscope (Figure 17). About 0.4 g of crushed dried cells

were added to cellulose extraction thimbles (60x26 mm, GE Life Sciences) and were loaded into

the Soxtec equipment. The extraction was carried out by boiling the dried cell mass for 2 h at

160°C in a suitable solvent mixture (Methanol:Chloroform 1:2 v:v, Fisherbrand, 99.5% pure and

99.8 % pure respectively) in pre-dried and pre-weighted aluminium cups. Then a 20-min rinsing

step followed to remove any remaining oil from the thimble. After the rinsing period, the solvent

was collected in the condenser of the Soxtec and the pure oil was retained in the metallic cups,

as shown at the last image within Figure 16. The oil-containing cups were dried overnight at

60°C and the oil was determined gravimetrically as follows:

%(g) Biomass Powder

(g) )extraction before Cup-oil with (Cup Lipids Total(%) content Lipid 100 (4-3)

(g/L) Weight Cell Drycontent Lipid(g/L) ionconcentrat Lipid (4-4)

4.4.5 Microscopic observation of cells

For cell morphology observation and check culture purity during the experiments,

samples were regularly checked in a light microscope, Olympus BH2. The big size of yeast cell

also allowed lipids to be observed and the process of oil accumulation to be monitored during

the cultivation.

Chapter 4 Materials and methods

65

Figure 16: Schematic representation of the oil extraction procedure used in this study.

Figure 17: Cell debris after crushing, 100x objective lens.

4.4.6 Calculation of cultivation parameters: Carbon to nitrogen ratio

The carbon-to-nitrogen ratio (C/N ratio) was calculated throughout this thesis by taking

into account the carbon content of glycerol (0.391 g/g)and the nitrogen content of yeast extract

(0.1 g/g):

Liquid nitrogen treatment

Crushed cells

Soxtec unit

Oil

Dried cells


66

(g/mol)AW

0.1(g/L)ionconcentratNitrogen

(g/mol)AW

0.391(g/L)ionconcentratGlycerol

l)C/N(mol/mo

N

C

(4-5)

Where AWC (g/mol) is the atomic weight of carbon (12 g/mol) and AWN is the atomic

weigh of nitrogen (14 g/L) respectively. The carbon content of yeast extract was not affecting the

C/N ratio (error <3%) and therefore it was ignored.

67

Chapter 5

A comparative study on yeasts for potential

growth and oil accumulation on glycerol as the

sole carbon source


68

5 Selecting a yeast based on potential growth and oil accumulation on glycerol

5.1 Introduction

After selecting glycerol as the carbon source for oil production, suitable microbial

factories are required for its efficient conversion to biomass and oil. In Chapter 2, strain selection

was stated as one of the ways of improving the economics of microbial oil production by

broadening the range of known microbes and exploiting their capabilities on common or unusual

carbon sources. The present chapter describes the screening of wild-type oleaginous yeasts for

their ability to convert glycerol to cellular mass and oil. Primary screening on nitrogen-rich

media and both glucose and glycerol allowed a first assessment of growth on glycerol. Then,

growth and oil yield were assessed under nitrogen limited submerged fermentation before the

selection of the best lipid-producing yeast strain for further research.

5.2 Theoretical Background

5.2.1 Oleaginous yeasts

Some yeasts are capable of accumulating high quantities of lipids. Moreover, yeasts

generally have been a preferred tool for biotechnological applications. Several yeast strains

along with their oil contents are summarised in Chapter 2 (Table 2). The oleaginous yeasts,

reported as glucose-consuming microorganisms previously, which were used in this study are

presented below.

1. Lipomyces starkeyi CICC 21390

2. Lipomyces kononenkoae CICC 1714

3. Trichosporon fermentans ACCC 21148

4. Trichosporon cutaneum ACCC 20119

5. Trichosporon fermentans ACCC 20243

6. Candida sp. (isolated from kitchen grease, kindly donated by Dr Zhuo Liang, Lab

of Microbiology and Enzyme Engineering Oil Crops Research Institute, CAAS,

China)

7. Rhodotorula glutinis CICC 31596

L. starkeyi, is an oleaginous aerobic yeast belonging to the Phylum of Ascomycota, and

family of Lipomycetaceae originally found in soil (Starkey, 1946). Several biochemical studies

Chapter 5 Selecting an oleaginous yeast

69

have investigated oil accumulation on strains of L. starkeyi, which are able to assimilate

commercial sugars such as glucose and lignocellulose-derived glucose (Anschau et al., 2014),

xylose (Gong et al., 2012) or both (Zhao et al., 2008) and accumulates high quantities of lipids up

to 72% of their total weight using different cultivation modes and scales of operation (Ageitos et

al., 2011). References regarding growth on glycerol for this yeast are limited but Tchakouteu et

al. (2015) have reported some such work recently. This yeast has also been reported to tolerate

acidic conditions (Calvey et al., 2016; Gong et al., 2012).

L. kononenkoae is another species of the family of Lipomyces. Although this species has

been characterised as oleaginous, research has focused so far on its amylase production and

starch assimilation rather than oil production. In 1978, the lipid composition of two

L. kononenkoae strains was analysed (Hossack and Spencer-Martins, 1978) while a recent study

investigated improvement of linoleic acid production by genetic modification of the same strain

CICC 1714 as used for the work reported in this chapter (Wang et al., 2011).

The oleaginous yeast T. cutaneum belongs to the Phylum of Bacidiomycota, and is able to

proliferate and accumulate oil (ca. 36%) with similar composition to vegetable oils (Chen et al.,

2012). Corn based hydrolysate sugars have been common carbon sources for its cultivation

while it has been reported to utilise them simultaneously and generally tolerate inhibitors within

the plant matter (Hu et al., 2011).

T. fermentans is an oleaginous yeast from the phylum of Ascomycetes, also known as

Geotrichum fermentans. Its oil composition is rich in palmitic, oleic and linoleic acid rendering it

as a potential regarding biodiesel production (Huang et al., 2009). It has been the subject of

extensive research regarding assimilation of mainly rice straw and sweet potato vine

hydrolysates. Furthermore, supplementation of sugar-based medium with pure glycerol

resulted in a 42% oil content suggesting that glycerol is also a suitable carbon source (Shen et

al., 2013b).

Candida sp. also belong to the phylum of Ascomycetes and some are able to accumulate

more than 33% oil (Duarte and Maugeri, 2014; Dey and Maiti, 2013; Ayadi et al., 2016).

Oleaginous species of Candida include C. tropicalis, C. viswanathii and Candida sp. A less studied

species, C. freyschussii utilised efficiently, large amounts of glycerol in fed-batch fermentation for

microbial oil synthesis (Amaretti et al., 2012).

The oleaginous Rh. glutinis is a mesophilic yeast, which belongs to the phylum

Basidiomycota, class Microbotryomycetes and order Sporodiobolales, found in lakes, seasalt and

dairy products (Kot et al., 2016). Rh. glutinis has been mainly known for its ability to produce


70

pigments such as carotenoids. Besides that the yeast can assimilate a variety of carbon sources,

including glycerol (Saenge et al., 2011b).

It is important to know, in a screening study, that none of the microorganisms is

pathogenic. Except for immunocompromised individuals, for whom there are only scarce reports

of infections, none of these yeasts have been shown to be pathogenic.

5.2.2 Observing intracellular lipids: Sudan Black B staining

Sudan Black B (Figure 18) belongs to the group of Sudan dyes, which are aromatic azo

and diazo compounds with the CAS (chemical substances) index name 2,3-dihydro-2,2-

dimethyl-6-[2-[4-(2-phenyldiazenyl)-1-naphthalenyl]diazenyl]- and chemical formula C29H24N6,

in room temperature it is powder with brown colour (Sabnis et al., 2010). Sudan Black B has

been used in analysis of tissue ceroid and lipofucsin, in vitro tests in humans and blood cell

staining. It is insoluble in water but soluble in ethanol, acetone, benzene, toluene, xylene and

ethylene glycol. It is usually used as a saturated solution in 70% ethanol (Hartman 1940,

Zakerhamidi et al. 2014). Sudan stains are widely used as colorants in food, cosmetics and other

materials such as textiles, solvents, etc. (Fonovich, 2012; Zakerhamidi et al., 2014). They are

considered slightly hazardous in respect to skin contact and mucous membranes and although

they are not classified as carcinogenic in humans (Zakerhamidi et al., 2014) they have been

mentioned as potential carcinogens and a oral toxins (Sabnis et al., 2010). For that reason,

nowadays use in food has been prohibited but they are still found as contaminants in foods and

as ingredients in cosmetics (Fonovich, 2012).

Sudan Black B was introduced as a stain for lipids in 1934 (Hartman, 1940) and is a basic

dye, which matches with neutral lipids and phospholipids. It colours lipids dark blue to light blue

according to the degree of saturation of the fatty acids (Thakur et al., 1988).

5.3 Experimental design

Taking into account the fact that all the available strains were normally grown on glucose,

the first step in the screening experiment was to see if the presence of glycerol alongside glucose

would have any detrimental effect on growth. Therefore, both glucose and glycerol were

Figure 18: Chemical representation of Sudan Black B (Zakerhamidi et al., 2014).


71

included in the initial screening media. The yeasts that were unaffected by the presence of

glycerol were selected for further evaluation. For these, growth was monitored on a medium

containing glycerol as the sole carbon source. This provided preliminary data on growth and oil

accumulation, which were used to select one strain as the subject for all further research.

5.4 Methodology

5.4.1 Strain maintenance

All seven strains were initially stored and maintained at 4oC in YPD micro-vials, as

received. From these, several stock cultures were created for subsequent use in the experiments

after continuous sub cultivation every month, in glucose based YPD agar slants and Petri dishes.

The composition of the YPD medium was (per L): 10 g Yeast extract (Sigma-Aldrich), 20 g

Glucose (Fisher Scientific), 20 g Peptone (Oxoid), 15 g Agar (Sigma-Aldrich) at pH 6-6.5.

Experiments to investigate the effect of glycerol utilised a slightly modified version of this

medium (YPGly1). The composition of this medium was derived from (Uçkun Kiran, 2012) and

is the following (per L): 3 g Malt extract (Fluka), 10 g Yeast extract (Sigma-Aldrich), 10 g Peptone

(Oxoid), 10 g NaCl (Sigma-Aldrich), 50 g Glycerol (Sigma-Aldrich) at pH 5.5.

5.4.2 Cultivation of yeasts on both glucose and glycerol

The fermentation was carried out in 500 mL shake flasks. The working volume was 150

mL in all cases. For the inoculum preparation, the yeasts were left to grow for 24 h in a flask that

contained 20 g/L glucose and 10 g/L yeast extract at pH=6. After that, they were inoculated at a

10% v/v ratio into different flasks with yeast extract at a constant concentration of 10 g/L and

glucose and glycerol in various concentrations (Table 7) at pH=6 and cultured for 24 h. The total

carbon source concentration in each flask was 36 g/L and the proportions of glucose and glycerol

varied to comply with this value.

5.4.3 Sudan Black B staining experiments

The composition of the nitrogen-limited solid medium was: 8 g Yeast extract (Sigma-

Aldrich), 30 g Glycerol (Sigma-Aldrich) at pH 5.5. For the microscopic staining, the yeasts were

grown for 74 h on 40 g/L glycerol and 8 g/L yeast extract at 30°C.


72

Table 7: Concentrations of glucose and glycerol used in media for the observation of growth of yeasts on glycerol.

Flask Glucose (g/L) Glycerol (g/L)

1 36 0 2 32 4 3 28 8 4 24 12 5 20 16 6 16 20 7 12 24 8 8 28 9 4 32

10 0 36

5.4.4 Cultivation of yeasts on glycerol

Glycerol concentrations from 0 to 150 g/L were used with constant yeast extract

concentration at 8 g/L at pH 5.5. The working volume was 100 mL in all cases. The pre-culture

consisted of 20 g/L glycerol and 10 g/L yeast extract. The yeasts were cultured for 72 h at 30°C

and shaken at 200 rpm. Cells grown in media with C/N ratios above 29 mol/mol were harvested

for quantitative estimation of their oil content.

5.4.5 Analytical methods

Carbon source analysis

Samples from the fermentation broth were centrifuged in a microcentrifuge at 13000

rpm for 8 minutes. The supernatant was removed and stored in a refrigerator at -30°C in

Eppendorf tubes until analysis.

Residual glucose from cell-free fermentation samples was analysed using a GL6 Analyzer

(Analox) which measures the oxygen uptake for the enzymatic conversion of glucose to gluconic

acid using glucose oxidase (http://www.analox.com/analyte-application-area/glucose-

industry-and-biotechnology). The oxygen uptake is proportional to the amount of glucose in the

sample. The machine was calibrated every twenty samples, using a standard solution of 4.5 g/L

glucose.

22(GO) oxidase Glucose

2` OHacidgluconicOglucoseD

Residual glycerol was analysed in the same GL6 Analyser as the glucose analysis. The

principle applied was conversion of glycerol to dihydroxyacetone phosphate (DAP) and H2O2 by


73

enzymes according to the following reactions (http://www.analox.com/analyte-application-

area/glycerol-industry-biotechnology). For the analysis, a standard solution of 20g/L glycerol was

used and dilutions were made as necessary.

ADPphosphate3GlycerolATPGlycerol(GK) kinaseGlycerol

22(GPO) kinasephosphate-3-Glycerol

2Ophosphate-3-Glycerol OHDAP

ATP refers to adenosine triphosphate, ADP adenosine triphosphate and DAP to

dihydroxyacetone phosphate.

Visual observations

During the cultivation of the yeast strains, especially on solid media, careful visual

examination was carried out throughout. Characteristics, such as the progress of colony

formation and development, type of colonies, the rate at which they grew, their morphology,

colour and texture were observed and recorded. In addition, the ease of handling, e.g. extracting

colonies and spreading on plates, was also noted.

Staining with Sudan Black B

The staining method was applied according to Thakur et al., (1988). Briefly, a drop of cell

suspension was spread on a microscope slide and was fixed by flame. After a washing step with

distilled water, Sudan Black B solution (0.3 g in 100 mL of 70% ethanol solution) was left in

contact with the cells for 20 mins. The excess stain was washed first with a 50% ethanol solution

and then xylene was used as a further stain removal (Sudan Black B is also soluble in xylene).

The slide was blotted dry and a 1% Safranine solution was added onto the slide and left for 15

min. Finally the slide was washed with distilled water. When this method was modified the

Safranine step was omitted.

The staining of yeast colonies directly from solid media was a simple technique derived

from the above procedure and it involved pouring the solution of Sudan Black B over the

colonies, leaving it to stain for 30 min and then rinsing with ethanol.

http://www.analox.com/analyte-application-area/glycerol-industry-biotechnology

http://www.analox.com/analyte-application-area/glycerol-industry-biotechnology


74

5.5 Results

5.5.1 Morphology observation study

Cultivation on solid media and familiarisation with the yeast strains

The yeasts were transferred from the vials to glucose-based Petri dishes using the streak

plate technique and their growth and morphology were observed. Figure 19 shows the colonies

of the yeast strains. Almost all of them formed creamy-white colonies, except for T. fermentans

21148 which had dark beige colonies. For example, L. kononenkoae has white smooth round-

shaped colonies and the colonies of Rh. glutinis had irregular shape. T. fermentans 20243 formed

rough creamy colonies. Candida sp. on the other hand, formed indiscernible colonies. A basic

description of the colonies is given in the figure.

Figure 19: Colonies of (A) L. starkeyi CICC 21390, (B) L. kononenkoae CICC 1714, (C) T. fermentans ACCC 21148, (AD) T. cutaneum ACCC 20119, (E) T. fermentans ACCC 20243, (F) Candida sp., (G) Rh. glutinis CICC 31596.

Forming bubbles Smooth shiny circular shape Light brown

Dispersed indiscernible Discernible rough Dispersed rough

Circular wavy


75

Cellular morphology

All the yeasts were observed to be unicellular (Figure 20), but the Candida sp. also formed

mycelia. Various shapes of cells (oval to round with some elongate cells) were obtained.

Generally speaking, yeast cell shape varies with the age of cells and the medium and cellular

shape reflects the macroscopic morphology of colonies.

5.5.2 Cultivation of oleaginous yeasts on both glucose and glycerol

A series of experiments were performed in order to introduce glycerol into the nutrient

medium to show whether the yeasts would be affected by the presence of glycerol. The total

carbon source was maintained at a total concentration of 36 g/L and the partial concentrations

varied to match this value. With increasing glycerol proportion, the glucose concentration was

decreasing. In order to get quick results, a single measurement of growth was made at 24 h after

inoculation. It was expected that by increasing glycerol fraction, growth at 24 h would be less,

either because of reduced growth rate or extended lag phase. The results from this experiment

are shown in Figure 21.

Figure 20: (A) L. starkeyi, (B) L. kononenkoae, (C) T. fermentans 21148, (D) T. cutaneum, (E) T. fermentans 20243, (F) Candida sp. (G) Rh. glutinis. Photos taken with the 100x objective lens.


76

Three different trends were observed on the final biomass. Rh. glutinis showed an almost

linear decrease in cellular concentration with increasing glycerol content. L. starkeyi was

similarly affected by the presence of glycerol although it obtained lower cell concentration.

T. fermentans ACCC 20243 showed better levels of growth on mixtures of glycerol and glucose,

than on solely glucose or glycerol. Candida sp. also showed slightly higher cell concentration on

mixtures but the increase was less obvious. The cellular concentration for all the other strains

appeared relatively constant for the whole range of mixtures, indicating that there is no

inhibitory effect of glycerol. In similar studies, with more than one substrate, yeasts have been

shown to utilise one carbon source first and then a second, while in very limited cases they utilise

them concurrently (Zhao et al., 2008). The yeast T. cutaneum had the lowest cellular

concentration, indicating that there might have been a significant lag phase and that a longer

cultivation time was required even in the case of the glucose-only medium. Although growth was

the only criterion used for distinguishing the available strains in this experiment, four were

selected for further study based on the results obtained. Although inhibition due to higher

glycerol proportion was seen for Rh. glutinis, this yeast showed rapid growth during the first

24 h and was able to grow even in a medium only glycerol and yeast extract. Its DCW

concentration was the highest of all up to a 40% glycerol fraction, followed by Candida sp. The

T. fermentans strain 20243 and the L. starkeyi were not selected for the next stage because they

were not easy to cultivate in the first instance compared to the ease of growth of the slower

growing L. kononenkoae. The 36 g/L glycerol experiment suggests that pure glycerol was an

adequate carbon source and no further adaptation was needed for the four strains; Rh.glutinis,

Candida sp., T. fermentans (ACCC 21148) and L. kononenkoae and these were therefore selected

for further screening.

Figure 21: Cellular concentration of various yeast strains at 24 h of incubation when cultivated on different proportions of glycerol in glucose-glycerol media. In each case glycerol is represented as a fraction of the 36 g/L total carbon source with glucose being the remainder.


77

5.5.3 Observation of lipids using Sudan Black staining

Following the initial screening experiments, the four selected strains were cultivated on

nitrogen-limited solid media with glycerol as the only carbon source. Results are shown in Figure

22. The nitrogen-limited conditions resulted in slower growth compared to the full nutrients

solid medium used previously. However, signs of growth appeared by 48 h. Following this, Sudan

Black staining was performed directly in the Petri dishes in order to estimate the oleaginousness

of the strains after 72 h. Figure 23, shows that the stain was retained by most of the colonies.

This should be considered as an indicator of their ability to accumulate oil. Moreover, in two of

the strains (Candida sp. and Rh. glutinis) the colour was particularly intense, suggesting excellent

oil accumulation.

Figure 22: Oleaginous yeasts after 72 h of growth on nitrogen-limited solid media. (A) L. kononenkoae, (B) T. fermentans (C) Candida sp., (D) Rh. glutinis.

Figure 23: Yeast colonies stained with Sudan Black B. From left to right: L. kononenkoae, T. fermentans 21148, Candida sp., Rh. glutinis.


78

Using the same medium as above, the strains were cultivated in shake-flasks and samples

were withdrawn for microscopic examination. The method reported by Thakur et al., (1988) was

used to stain samples taken at 72 h. As can be seen in Figure 24, the additional Safranine step in

the staining method did not particularly improve the contrast. Much effort is needed to

distinguish the light grey oil globules within individual cells.

It was assumed that further method optimisation was needed, adjusted to each strain in

order to improve the visibility of the oil inside the cells. Although Sudan Black B staining is a

common preliminary tool for oleaginous yeasts screening (Kitcha and Cheirsilp, 2011; Dai et al.,

2007), some works have applied fluorescent staining for lipid observation (Sitepu et al., 2012).

This method is more powerful since the sensitivity of the microscope will detect better the oil

droplets.

5.5.4 Evaluation of growth on different concentrations of glycerol in

batch mode

The four preferred yeasts were cultivated on glycerol under nitrogen limitation with

constant nitrogen but variable initial glycerol concentrations, and used for quantitative oil

estimation. The time-course profiles of the growth of the oleaginous yeasts are shown in Figure

25. Graph A, shows the growth profiles of Rh. glutinis. Highest growth was obtained for lower

Figure 24: Stained cells of (A) Rh. glutinis, (B) L. kononenkoae, (C) Candida sp., (D) T. fermentans. Bar represents 10 μm. Photos were taken with the 100x magnification lens.


79

glycerol concentrations (20-45 g/L) and the decrease in growth as glycerol concentration

increased, was more obvious for the concentrations 90, 100 and 150 g/L. Even at 24 h of

cultivation there were minor differences in the growth levels but significant differences

appeared at 30 h of cultivation. Glycerol loading clearly affects the growth rate negatively,

suggesting glycerol inhibition. Similar behaviour has been reported for C. freyschussii which was

inhibited at concentrations above 40 g/L pure glycerol (Raimondi et al., 2014). In another study

Rh. glutinis tolerated up to 100 g/L initial pure glycerol, while beyond that a drop was seen in

the cell concentration (Souza et al., 2014).

Results for the Candida sp. (Figure 25B) showed that at 20 g/L and 50 g/L glycerol,

cellular concentration reached 22 and 24 g/L respectively. In contrast to Rh. glutinis, this yeast

was not affected by glycerol inhibition up to 100 g/L glycerol. As expected, the 5 g/L did not

provide enough growth but the 150 g/L experiment also gave very little growth.

Lower cell densities were obtained by L. kononenkoae (Figure 25C). Interestingly,

however, this strain reached similar levels of growth for all glycerol concentrations above 20

g/L, suggesting that there is no glycerol inhibition. Unfortunately, there is no literature

information to corroborate this finding but a recent study did report high levels of growth

(47.7g/L) for the strain L. kononenkoae NRRL Y-7042 growing on glycerol (Slininger et al., 2016)

Very low growth was observed for T. fermentans strain across all concentrations of

glycerol, suggesting that it can barely consume glycerol. This strain had initially been selected as

one of the four preferred strains because it showed a flat profile for DCW at 24 h on mixtures of

glucose and glycerol (Figure 21). However, it could be that for all of the flasks in that experiment,

the yeast was surviving on the glucose and/or carbon available from yeast extract and was not

consuming any of the glycerol.

To demonstrate the above findings more clearly, data for just the 72 h samples against

initial glycerol concentration are inset in the graphs in Figure 25.

Comparison of the oil yields of the four oleaginous yeasts

Figure 26 depicts the oil content and oil titre of the four oleaginous yeasts. Candida sp.

accumulated the highest oil content, which in combination with its biomass resulted in high total

oil titres. C. freyschussi has been reported to accumulate 26.4% oil and produce 11.9 g/L biomass

(Amaretti et al., 2012). In the present study, Candida sp. and Rh. glutinis had comparable cell

densities. However, the higher oil content meant that the Candida sp. reached a higher overall oil

concentration. These results are consistent with the qualitative plate staining, which showed

darker colonies for the Candida sp. and Rh. glutinis. There is much scope for optimisation of oil


80

production of Rh. glutinis since higher yields have been reported elsewhere (Wang and Ren,

2014). L. kononenkoae did not accumulate large amounts of oil and the yields obtained are below

the definition threshold for ‘oleaginous’.

The cell pellet of centrifuged cells belonging to Candida sp. was soft, which also could be

related to the high oil content of it. Regarding Figure 26A, the oil content of Candida sp and

Rh. glutinis can be correlated with a 2nd order polynomial line, while the two other yeasts are

characterised by a linear trend with very low slope. Similar trends were reflected in the oil titre

graphs (Figure 26B). The polynomial trend is a common phenomenon in other oleaginous yeasts.

Most often substrate inhibition causes the oil yield as well as the growth to cease beyond a

relatively high substrate concentration. Other times, a threshold value in oil content is achieved

and then other metabolites are produced (Makri et al., 2010). Similar trends were obtained with

R. toruloides cultivated in batch fermentations on pure and crude glycerol (Uçkun Kiran et al.,

2013).

Figure 25: Kinetic growth profiles of (A) Rh. glutinis, (B) Candida sp., (C) L. kononenkoae, (D) T. fermentans, cultivated on glycerol based media under nitrogen limitation. Inset are cellular concentrations of each strain at 72 h of fermentation.

(A) (B)

(C) (D)


81

In general, there is a trade-off relationship between lipid synthesis and cell proliferation.

In some cases other metabolites drive away carbon from lipid sythesis. To evaluate further the

lipid accumulation of the selected oleaginous yeast strains, the yields of biomass and oil on

glycerol is shown in Figure 27. Candida sp. had again the highest conversion yield (YL/Gly) which

was for all situations higher than the other strains. In terms of total biomass yield, the Rh. glutinis

showed the highest yield but only at low glycerol concentrations. Generally, trends were similar

for all but the T. fermentans strain.

(A)

(B)

Figure 26: Oil yields of four oleaginous yeasts obtained after cultivation on glycerol based media (A) Oil content (% w/w), (B) Oil concentration (g/L), SD<10%.


82

Figure 27: (A) Oil yield on glycerol (YL/Gly) and (B) Biomass yield on glycerol (YX/Gly) for Candida sp., Rh. glutinis, L. kononenkoae and T. fermentans.

(A)

(B)


83

5.6 Conclusions

Amongst seven oleaginous yeasts previously grown on glucose, four were able to grow

on, or at least tolerate, glycerol. These four strains were further screened at different initial

glycerol concentrations. The Candida sp. had the highest biomass, oil content and yield of oil on

glycerol, followed by Rh. glutinis. L. kononenkoae did not accumulate much oil while

T. fermentans was the least efficient of all. In general, Rh. glutinis and Candida sp. were shown to

be the more effective in terms of growth and oil accumulation. Although the Candida sp

accumulated higher amounts of oil and a higher oil titre, while Rh. glutinis showed inhibition of

growth at higher glycerol concentrations, the latter was chosen for further studies. This was

because the Candida sp. was not fully identified and therefore could not be compared directly

with equivalents in the literature. Moreover, it was difficult to handle due to its mycelial form

and harvesting cells from samples was unpredictable and unreliable. Rh. glutinis, on the other

hand, is well-documented in the literature and is the subject of current research elsewhere. It is

also very well-behaved in the lab always growing in unicellular fashion. The following chapters

report growth aspects of Rh. glutinis and present a modelling approach to the oleaginous system

based on the experiments performed.


84

85

Chapter 6

Growth aspects of Rhodotorula glutinis on

glycerol based media


86

6 Growth aspects of Rhodotorula glutinis on glycerol based media

6.1 Introduction

Following the screening study of oleaginous yeasts in the previous chapter, the selected

strain Rh. glutinis CICC 31596 was examined further. This chapter contains evaluation of the

growth rate, final biomass and lipid production on different types of medium, inoculum sizes and

aerated cultivations. Then, the presence and concentration of methanol and sodium chloride,

common growth inhibitors found in crude glycerol were investigated, followed by a

fermentation on synthetic crude glycerol. Consequently, the chapter is structured in four parts,

growth defining experiments, conditions for lipid accumulation, dynamic lipid production and

cultivation of Rh. glutinis on crude glycerol.

6.2 Theoretical background

The oleaginous yeast Rhodotorula glutinis

Rhodotorula glutinis (formerly known as Cryptococcus glutinis or Rhodotorula terrea)

belongs to the super kingdom of Eukaryota, kingdom of Fungi, phylum of Basidiomycota, genus

Rhodotorula (http://www.brenda-enzymes.org, date 31/3/13). It can be found in soil, lakes,

marine and polluted water, plant leaves and dairy products. It has been used for cheese

production (Gabier et al., 2005) and in the production of carotenoid pigments (Yen et al., 2015a).

Its cellular shape is ovoid to ellipsoidal or elongate, with some strains forming pseudohyphae

(http://www.brenda-enzymes.org). Reproduction occurs by budding, while some strains form

asexual teliospores. It can consume a wide range of carbon sources, including sugars such as

glucose, sucrose, and xylose and alcohols such as ethanol. It has also been reported to grow

successfully on glycerol (Louhasakul and Cheirsilp, 2013). Common growth conditions are:

T=25-30oC or 37oC and pH=5-6.5 (Louhasakul and Cheirsilp, 2013). Ethanol production in this

species has not been reported. Rh. glutinis has been reported as an oleaginous yeast with

satisfactory yields and attracts interest because it produces carotenoids along with the lipids.

Table 8 summarises biomass and oil yields of some Rh. glutinis strains.

http://www.brenda-enzymes.org/

Chapter 6 Growth of Rh. glutinis on glycerol

87

Table 8: Biomass and oil yields of Rh. glutinis Strain Carbon source DCW

(g/L) Oil content (%, w/w)

Oil (g/L)

Carotenoid (mg/L)

Reference

CGMCC 2258 Glucose 86.2 26.7 11.5 4.2 (Zhang et al., 2014b)

CICC 31643 Glucose n.r 44.5 7.93 n.r. (Wang and Ren, 2014)

ATCC 204091 Distillery effluent 6.06 27.02 1.86 n.r. (Gonzalez-Garcia et al., 2013)

CCT 20-2-26 Crude Glycerol 17.99 20 n.r. 42.24 (Petrik et al., 2013)

Non specified P. euramenicana 18.1 34.16 n.r. n.r. (Xu et al., 2013)

ATCC 204091 Levoglucosan 6.8 42.2 2.7 n.r. (Lian et al., 2013)

n.r.: non reported

In addition to the data provided for Rh. glutinis and substrate information, different

culture conditions have been investigated recently in order to improve the oil yield. Some of

them refer to mixed cultures of yeast and algae and show better results than single cultures. In

particular, in this kind of fermentation, algae consume carbon dioxide and produce oxygen which

is then consumed from the yeast, which produces carbon dioxide, to be taken up by the algae.

However, this approach is not straightforward since algae and yeast growth conditions have to

be carefully chosen (growth rates, optimum pH) and taken into account. A co-culture of

Rh. glutinis and Sp. platensis gave higher yeast biomass concentration (3.67 compared to 1.7 g/L

for single culture) and lipids (0.47 g/L compared to 0.14 g/L of single culture). Rh. glutinis and

C. vulgaris achieved increased biomass of 2.2 g/L and 1.7 g/L lipids compared to the 1.4 g/L for

the single culture (Santos and Reis, 2014).

Oil accumulation has been linked to the energy changes in the yeast cell (Andlid et al.,

1995). The lipid production of Rhodotorula glutinis was studied by microcalorimetry and the

energy content of the yeast was monitored throughout the fermentation and correlated with the

lipid accumulating phase. In particular, the heat production rate was monitored in situ by

connecting the reactor and creating a flow of broth to a conductive calorimeter. The results

showed that there is an increase in heat production, which reaches maximum at the start of oil

accumulation when nitrogen is exhausted, a finding verified by nitrogen analysis. After the start

of the lipid accumulating phase heat production is steady and decreases when consumption of

lipids occurs as the carbon finishes. The enthalpy of the yeast reached 30.6 kJ/g during the oil

accumulation. It is interesting that this study was conducted in order to characterize this yeast

as suitable aqua feed.


88

6.3 Methodology

6.3.1 Experiments for defining growth conditions

Investigation of initial pH in batch cultivations

In order to define how initial pH affects the growth of the yeast, three different pH

conditions were tested: pH 5, 5.5 and 6. The preculture consisted of 20 g/L glycerol and 10 g/L

yeast extract and the production media had the same composition. The pH was adjusted to 5, 5.5

and 6 with the addition of 1M HCl or 1M NaOH. The inoculum size was 10% v/v and the working

volume 100 mL in all cases. The flasks were incubated at 30oC and 200 rpm.

Growth on different nitrogen sources (organic- inorganic)

A study was performed in order to evaluate growth on some common nitrogen sources,

namely ammonium sulphate, urea and glutamic acid. A control medium without nitrogen source

was prepared in order to act as a reference point for the amount of yeast extract that would be

transferred during inoculation from the seed culture. All flasks contained the same amount of

nitrogen (as calculated from Equation 4-5). The same amount of glycerol was also present in

each flask (30 g/L). The culture volume was 100 mL and the incubation took place at 30oC, 200

rpm and pH 5.5. The experiment was conducted in duplicate.

Compound of Weight Molecular

Nitrogen of Weight Atomicatoms N of NumbercontentN

(4-5)

The nitrogen content of glutamic acid is 0.095, urea 0.466, and ammonium sulphate

0.212. According to this, in order to have a nitrogen concentration of 1.06 g/L in all the cases,

11.16 g/L glutamic acid, 2.27 g/L urea or 5 g/L ammonium sulphate were used in the respective

media.

The role of yeast extract in the growth medium

The purpose of the experiment was to check the importance of other nutrients in yeast

extract by providing nitrogen only through ammonium sulphate. The experiment was built in

three stages: first a seed culture for 24h with 20 g/L glycerol and 7 g/L yeast extract, then a

second seed culture for 48 h with composition 20 g/L glycerol and 10 g/L ammonium sulphate

(in order to increase the nitrogen content of the media), while at 48 h, inoculum was transferred


89

to the third stage flask. The third stage flask had the same composition as the second stage one.

This final fermentation was performed in duplicate. The volume of the first two stages was 100

mL while the third stage flask had working volume of 250 mL. In all the cases the pH was 5.5, the

incubation temperature was 30oC and the agitation was 200 rpm.

6.3.2 Growth and lipid production experiments

Defined versus semi-defined medium

A study to compare defined and semi-defined media on the growth and lipid

accumulation of Rh. glutinis was carried out to show whether providing each and every nutrient

had beneficial effect compared to yeast extract as the only nutrient source. The inoculum

propagation followed the procedure stated in 4.3. Prior to inoculation the cell suspension that

corresponded to 10% v/v inoculum was centrifuged to be free of nutrients and then was added

to fermentation media with the following compositions: 1) 30 g/L glycerol and 3.5 g/L yeast

extract, 2) 30 g/L glycerol and 1.5 g/L ammonium sulphate plus trace element solution (adapted

from Salakkam 2012) and 3) 30 g/L glycerol and 1.5 g/L ammonium sulphate. The trace element

solution was composed of: 10 g/L FeSO4·7H2O, 2.25 g/L ZnSO4·7H2O, 1 g/L CuSO4·5H2O, 0.5 g/L

MnSO4·H2O, 2 g/l CaCl2·2H2O, 0.106 g/L (NH4)6Mo7O24·4H2O, 0.22 g/L H3BO4, 0.2 g/L CoCl2·6H2O,

0.02 g/L NiCl2·6H2O and 10ml/L HCl (37%) and was added to medium 2 at a concentration of 1

mL/L. The nitrogen concentration in all flasks was adjusted to ~0.3 g/L.

Inoculum size effect

Cell suspensions, grown as described in 4.3, corresponding to inoculum sizes of 8, 10, 15

and 20% v/v of the final, inoculated, volume were collected in sterile centrifuge tubes,

centrifuged to separate the cells from the initial medium, re-suspended an equal amount of

sterile water. These were then added to the production medium, which had the following

composition: 30 g/L glycerol and 2 g/L yeast extract at pH 5.5 and working volume of 100 mL.

Small Aerobic Reactors (SARs): Nitrogen-limited cultures and air supply

In order to see whether aeration is beneficial for growth, air was supplied to the

fermentation. For this experiment the fermentation vessels consisted of 250 mL Duran bottles

capped with rubber lids. Each lid had three ports, one for sampling, one for the air inlet and one

for the vent. Polypropylene tubing was used for all the connections. The air (compressed air,


90

BOC) was supplied to the vessels at a flow rate of 0.2 mL/min through a metallic distributor and

filter sterilised through of 0.2 µm autoclavable filters (Figure 28).

The inoculum propagation was the same as previously (section 4.3). The final medium

contained 2 g/L yeast extract and the glycerol concentration was adjusted in such a way as to

provide initial C/N ratios of 80, 100, and 120 (34, 43 and 52 g/L glycerol respectively). The

inoculum size was 10% v/v and the working volume 100 mL. A non-aerated culture (Duran

bottle closed with the lid, Figure 28B, on the left) with C/N 100 was prepared as a control for the

aerated culture with C/N ratio of 100. The fermentation took place at 30oC and with just 140 rpm

because higher agitation was unnecessary due to the supply of air. The experiment was

performed in duplicate.

6.3.3 Kinetics of microbial oil production

In order to obtain data during the cultivation, samples for substrate, growth and product

were withdrawn regularly. One experiment, using glucose as carbon source, served as a control

for experiments with glycerol. For the inoculum preparation of that experiment, glucose was

used as a carbon source at 20 g/L and the composition of the production medium was: glucose

40g/L, yeast extract 10 g/L at pH=5.5. For the experiments with glycerol, the preculture had the

usual composition (section 4.3). The composition of the production medium was: glycerol at 30,

40 and 80 g/L, constant yeast extract at 2 g/L and pH=5.5. The working volume was 100 mL in

Figure 28: (A) Experimental set up of SARs (Small Aerobic Reactors), (B) Schematic representation of one SAR vessel. The control bottle is not supplied with air. The fermentation took place in an incubator room set at 30°C.

(A) (B)

Control

Vent

Sampling port


91

all cases. Cell harvesting was performed using the sacrificial culture methodology, where at each

sampling time, a whole flask was used to allow enough sample for oil extraction.

Staining the lipids with Sudan Black B

The method developed for the estimation of intracellular oil content was based on the

spectrophotometric measurement of the absorbance of the stained oil droplets under UV light

(Thakur et al., 1989). For the accurate performance of the procedure some steps of the original

method developed by Thakur were modified. First, 50 mL of culture were centrifuged at 9500

rpm for 10 minutes. Then, the cell pellet were washed with distilled water to completely remove

all the remaining nutrients in the pellet. After being centrifuged (9500 rpm for 10 minutes), the

cell pellet was re-suspended in 50 mL of distilled water. Next, 5 mL of this water suspension of

cells were added to one glass test tube which could be used as the test sample. Likewise, 5 mL of

the water suspension were transferred to another test tube as a blank. In this method unstained

cells were used as the reference for the measurement of absorbance so, any absorbance due to

the growth (number of cells) was eliminated and would be the reference for the oil estimation.

In order to proceed with the hydrolysis of the cells, 5 mL of HCl (37%) were added to each tube.

The adding of HCl avoids possible interference of the dye with the nuclear chromatin (Thakur et

al., 1989). Next, 0.4 mL of Sudan Black Solution (0.3 g in 100 mL of 70% ethanol) was added to

the test sample while 0.4 mL of 70% ethanol solution was added to the blank. Vigorous mixing

is required at this stage, otherwise two phases are visible after the addition of 0.4 mL of the

ethanol solution.

Once the mixing takes place, the tubes were placed in a boiling water bath for 60 s, with

intermittent shaking. After that, they were placed for 3-5 min in a mixture of ice and water in

order to stop the staining reaction immediately, leaving next the test tubes to cool down at room

temperature for 19 min. When ready, the stained suspensions were centrifuged at 9500 rpm for

10 min. The resulting pellet was washed up to 3 times with 10.4 mL of 50 % ethanol solution

(equal to the initial volume of liquid before the boiling step: 5 mL cell suspension and 5 mL HCl

and 0.4 mL ethanol of Sudan Black B solution) to remove the excess of staining dye. The blank

solution received the same treatment, even if it was not stained. Finally, the cells were re-

suspended in 50 mL of water and the absorbance was measured at 580 nm. Simultaneously,

samples were observed under the microscope in order to visualize the stained oil droplets.

A modification to the above staining procedure was introduced in this study, which

consisted of applying longer boiling times (30, 60 and 90 s) than those suggested in the original

method.


92

Growth on synthetic crude glycerol

To simulate the effects of contaminants in crude glycerol, synthetic media were prepared

with constant concentrations of glycerol (50 g/L) and yeast extract (10 g/L) but increasing

amounts of methanol (0 to 120 g/L) and/or sodium chloride (0-100 g/L). The pH was adjusted

to 5.5. In all cases, a 10 % v/v inoculum was transferred to the production media. The flasks were

incubated at 30°C and 200 rpm. For studies on the growth rate dependence of methanol, sodium

chloride, and combinations of both, flasks were set up as detailed in Table 9.

Table 9: Experimental details of the methanol and sodium chloride content of synthetic crude glycerol solutions used to simulate the effects of contaminants in real crude glycerol.

Medium composition

CH3OH (g/L) NaCl (g/L)

Experiment A

0

5

10

25

40

60

80

100

120

Experiment B

0

5

20

50

70

100

Experiment C

10 0

10 6

25 10

40 10

60 20

80 20

6.4 Results

6.4.1 Experiments for defining growth conditions

Investigation of initial pH in batch cultivations

Yeasts are able to survive in a wide range of pH (from 2 to 9) and even tolerate highly

acidic environments (Santamauro et al., 2014). Solid media of pH=5, 5.5 and 6 were prepared

and inoculated by loop from a Petri dish. Previous Petri dish cultivations suggested that the


93

plates with pH=5.5 had fastest growth (data not shown). To test if this was the same in liquid

media, flasks with pH 5, 5.5 and 6 were then incubated for 72 h and growth was monitored

throughout the cultivation. The highest growth was again achieved with pH=5.5 although the

differences were not large (Figure 29). These results are in agreement with published

observations about the optimum pH for growth of the yeast Rh. glutinis (Xue et al., 2006).

Therefore, it was decided that pH=5.5 would be applied in all following studies. Contamination

from bacteria is always a risk in fermentations and since the use of bacterial antibiotics has been

reported to harden the yeast cell wall in Rh. glutinis (Smith and Marchant, 1969), a more acidic

pH is desired.

Growth on different nitrogen sources (organic-inorganic)

The nature of the nitrogen source has been reported to influence growth and lipid

accumulation, but it is rather a microorganism-specific feature. In order to evaluate the growth

performance of the yeast on different nitrogen sources, media containing either inorganic or

organic nitrogen compounds, were prepared. The nitrogen sources were ammonium sulphate,

urea and glutamic acid, and the total nitrogen content of all the media was maintained the same.

Residual yeast extract from the seed culture was allowed to enter the fermentation medium

during inoculation. Should all of the yeast extract from the seed culture remain unconsumed, the

concentration of it in the preculture medium would be 10 g/L. With inoculum volume 10% v/v

and by taking into account the dilution effect, a maximum of 1 g/L of yeast extract would be

transferred to the production media. In most cases, however, the residual yeast extract would be

zero or very close to zero. To test the effect of the presence of residual yeast extract from the

inoculum culture, a control flask with no additional nitrogen source was included in the

Figure 29: Growth profiles of Rh. glutinis on different initial pH conditions.


94

experiment. Figure 30A shows the growth curves of Rh. glutinis when cultivated on the various

nitrogen sources and Figure 30B the yield of biomass on nitrogen.

Ammonium sulphate gave the highest growth compared to the other nitrogen sources

while urea reached only levels comparable to the control (yeast extract from the seed culture).

In particular, the average specific growth rate for the first 24 h was 0.113 h-1 for the ammonium

sulphate culture, 0.112, 0.098 and 0.095 h-1 for the glutamic acid, urea and control cultures

respectively. In this case, inorganic nitrogen appeared to be good for growth. On the other hand

urea had the lowest growth rate. Consequently, ammonium sulphate and yeast extract were

considered to be suitable nitrogen sources for further experiments. Similarly, in another study

(Gao et al., 2013), ammonium sulphate gave the highest growth after yeast extract. Contradictory

results were obtained in a batch fermentation of Rh. toruloides where urea resulted in higher

biomass and oil content than ammonium sulphate while different nitrogen sources were found

to have little effect on the final biomass (Evans and Ratledge, 1984).

The role of yeast extract in the growth medium

In the previous section the effect of nitrogen sources was investigated with ammonium

sulphate shown to be the best inorganic nitrogen source for the growth of the yeast. However, in

(A)

(B)

Figure 30: (A) Time-course profiles of cell concentration of Rh. glutinis grown on different nitrogen sources, (B) Yield of biomass on nitrogen (YX/N) for three nitrogen sources tested. AS: ammonium sulphate, GA: glutamic acid, U: urea and B: control.


95

these experiments yeast extract was also present, though often in very small quantities, having

been transferred with the inoculum. Since yeast extract contains many nutrients, a different

approach was applied to see how cells perform in a medium, where yeast extract was absent

throughout and only glycerol and ammonium sulphate were supplied. When cells were

transferred directly in such a medium, no growth was observed since only C, N, H and O were

not enough to support cellular functions. When other nutrients (malt extract) were added, a

minor level of turbidity was seen. Even after some days without growth, addition of yeast extract

in a fed-batch process stimulated growth and showed that the cells were still alive. Then it would

be worth seeing what happens if the medium started from a more complete form to a simpler

one. For that purpose, a simple experiment was designed to investigate the effect of carry-over

of nutrients from an initial medium containing yeast extract to the fermentation stages. It was

anticipated that the initial presence of yeast extract could provide the cells with much energy

and supplies so they can go through the dilution stage, with only carbon and nitrogen source.

The culture lasted for 168 h and the growth monitoring, that took place every day is shown

below. The experiment was performed in duplicate and Figure 31 shows the growth curve of

each replicate.

Growth was not supported in non-yeast extract media and the low growth rate is obvious

from Figure 31. The cells grew in the second stage on possibly available yeast extract. The very

little growth during the third stage might have been due to possible budding cells during the

inoculation to the second stage and possible nutrients transferred from the seed culture (1st

stage). The final DCW was half of the usual obtained in other experiments. Consequently, the

yeast was in need of other nutrients such as phosphorus, sulphur, vitamins that are usually

present in yeast extract.

A

Figure 31: Growth of Rh. glutinis during cultivation in media containing glycerol and yeast extract (area A) followed by cultivation in glycerol and ammonium sulphate (area B).

B


96

Finally, after 72 h, budding cells were observed under the microscope but the nutrients

were insufficient to provide enough stimulus. This led to another experiment where the

nutrients of yeast extract were replaced by mineral medium coupled with a study on oil

production.

6.4.2 Growth and lipid production experiments

Defined versus semi-defined medium

The effect of defined and semi-defined media on growth and lipid accumulation was

evaluated by growing the yeast in three different media: glycerol and ammonium sulphate (G-AS

experiment); glycerol and ammonium sulphate supplemented with other defined nutrients (G-

MIN experiment) and; glycerol with yeast extract (G-YE experiment). Although the mineral

medium contained all the elements that could be necessary, the yeast extract medium was shown

to be the most effective in terms of growth with the final cell concentration being 3.41 g/L. This

suggests that the nutrients from yeast extract are more easily assimilated by Rh. glutinis. Similar

DCW was obtained in the mineral medium culture (3.13 g/L) but the growth rate was slightly

lower than the culture with yeast extract. As expected the medium containing only ammonium

sulphate and glycerol supported the lowest growth.

The G-AS grown cells accumulated the highest oil content, 48% (Figure 32A). It seems

that the cells did not have all the necessary nutrients to proliferate and they directed glycerol to

lipid synthesis. Also, while previously with only glycerol and ammonium sulphate, the growth

was restricted, this time the cells had been separated from the inoculum medium so no nutrients

were transferred to the culture flask. At the time of inoculation there were many budding cells

in the inoculum so they probably provided the stimulus for higher cell densities even in this poor

medium. On the other hand, the medium (G-MIN) didn’t promote oil production significantly,

reaching 23.26% w/w oil content which is similar to the yield obtained with the yeast extract

medium G-YE (26.38 % w/w).

The G-MIN medium was rich in nutrients and this might not promote lipid accumulation.

As can be seen from Figure 33, the cells in G-MIN medium formed irregular shape, different to

the cells grown in G-YE and G-AS or other previously observed morphologies. Probably, the

amount of some salts was more than required, leading to imbalanced cell membranes, but it is

not straightforward to design an appropriate medium. Common fermentation media for

oleaginous yeasts are somehow simpler than the above mineral medium. For example, Y.

lipolytica was cultivated in 0.75 g/L yeast extract, 0.4 g/L KH2PO4, 1.5 g/LMgSO4·7H2O and 0.1


97

g/L (NH4)2SO4 (Sestric et al., 2014), while Rh. glutinis was cultivated in (NH4)2SO4, 6 g/L KH2PO4,

2 g/L MgSO4, 0.1 g/L CaCl2 and 0.12 g/L FeCl3 (Liu et al., 2015). However, Uckun Kiran (2012)

utilised an undefined medium containing malt extract, yeast extract and peptone along with

glycerol.

Despite the high oil content, the G-AS grown cells did not give sufficient DCW (Table 10).

Moreover, no process would be possible on such a poor medium and hence no further

consideration was given to the G-AS medium. Furthermore, considering the cellular morphology

of the G-MIN medium, the G-YE culture seemed to be the better one for Rh. glutinis (Figure 33).

The mineral medium (G-MIN) resulted in slightly better yield than the G-YE medium. It can be

concluded that 15-16% of the glycerol was converted to lipids. The overall oil concentrations

were 1.04 g/L for the G-AS, 0.73 g/L for the G-MIN medium and 0.90 g/L for the G-YE medium.

These results suggest that a simple medium with only yeast extract as the nitrogen and nutrient

source was sufficient for the cultivation, and was therefore used for all future experiments.

(A)

(B)

Figure 32: (A) Growth profiles of Rh. glutinis on G-AS (glycerol/ammonium sulphate), G-min (glycerol/mineral) and G-YE (glycerol/yeast extract) media. (B) Oil content of Rh. glutinis for the different types of media. G-AS: glycerol and ammonium sulphate medium, G-MIN: glycerol and mineral medium, G-YE: glycerol and yeast extract medium.


98

Table 10: Summary of experimental values obtained during the cultivation of Rh. glutinis on defined and semi-defined media

Medium DCW (g/L) Oil (g/L) YL/X (g/g) YL/Gly (g/g) YX/Gly (g/g)

G-AS 2.15±0.2 1.04±0.52 0.48 0.32 0.28 G-MIN 3.13±0.13 0.73±0.34 0.23 0.17 0.45 G-YE 3.41±0.04 0.90±0.10 0.26 0.15 0.40

Inoculum size effect

To examine the effect of inoculum size on oil accumulation, different inoculum sizes were

used: 8%, 10%, 12%, 15% and 20% v/v. The amount of inoculum has been reported to affect the

culture duration since more inoculum results in reduced lag phase. The final growth, oil content

and oil concentrations are shown in Figure 34.

The amount of biomass obtained was very similar for all the different inocula. The oil

content fluctuated in a similar way to the total biomass. The oil content showed bigger

differences with higher values obtained for inoculum sizes from 8% to 12% w/w. As can be seen

in Figure34 the 8% inoculum size resulted in the higher oil content, biomass and titre. Larger

Figure 33: Micrographs of Rh. glutinis from (A) G-MIN media (B) G-AS media (C) G-YE media samples.

Figure 34: Final biomass, oil concentration and oil content obtained for different inoculum sizes of Rh. glutinis grown on glycerol and yeast extract for 72 h.


99

inocula should result in larger final biomass and oil titre despite the oil content. However, here

the very similar biomass values did not allow for significant final differences in oil titre. Three

times as much inoculum resulted in a clear improvement in biomass and oil production for L.

starkeyi at bioreactor scale (Anschau et al., 2014). Admittedly the inoculum sizes used here were

closer to each other and that could be the reason for the lack of clear differences between them.

Since the inoculum experiment was inconclusive the standard 10% v/v inoculum was retained

for all further experiments reported in this thesis. Figure 35, depicts the glycerol profile. The

consumed glycerol slightly increased with increasing inoculum size but the values were very

close.

Small Aerobic Reactors (SARs): Nitrogen-limited cultures and air supply

As a preliminary step at flask scale the yeast was provided with air in order to see

whether air supply is beneficial for the system. As can be seen from Figure 36, aeration improved

the growth rate for the first 24 h compared to previous experiments (e.g. in the comparison

between defined and semi-defined media less DCW was obtained both at 24 and 74 h). The C/N

ratio of 80 resulted in higher cell density, indicating again that lower glycerol concentration is

not growth inhibitory.

The provision of air improved the oil content, resulting in higher oil titres for all

approaches. Higher C/N ratios promoted lipid accumulation, while the highest oil content was

obtained for the C/N 100 experiment. The air supply may enhance growth and/or oil

accumulation, so the increase in oil content can be the result of the presence of air. A non-aerated

culture with C/N=100 was serving as a control for the aerated culture with C/N=100. The lack

of growth in the control culture indicated that the geometry also plays an important role in the

Figure 35: Amount of glycerol consumed by Rh. glutinis in batch cultivations inoculated by different inoculum sizes.


100

aeration and proliferation of the culture. The shape of the Duran bottle was not suitable for non-

aerated culture while the shake-flask manages to distribute better the oxygen in the broth.

6.4.3 Kinetics of microbial oil production

To study the kinetics of oil production over time, samples for oil extraction were

withdrawn at regular intervals throughout the fermentation. First, a control experiment with

glucose was performed. The growth curve, glucose consumption and oil titre are shown in Figure

37. The dry cell weight did not exceed 4 g/L by 72 h while the growth rate ceased after 48 h.

Glucose consumption was sharp during the first 24 h and then tailed off, failing to achieve total

consumption by the end of the fermentation. This indicated that there was an excess of glucose

and the system required more time for fuller conversion. Low oil content was obtained,

corresponding to 13-18 % w/w of the dry biomass. Oil accumulation started after 48 h of

cultivation.

To further evaluate the kinetics of biomass and oil production, the yields of biomass and

oil on glucose were determined graphically. For that, the biomass and oil produced were plotted

against the glucose consumed (Figure 38). The YX/Glu (0.246 g/g) and YL/Glu (0.06 g/g) suggest that

the majority of glucose was directed to biomass synthesis rather than lipid production.

The dynamic oil production was then studied with glycerol as the sole carbon source at

three different concentrations. Figure 39, shows the time-course profiles of the biomass, oil and

substrate. Interestingly, glycerol was consumed at almost the same rate in all of the experiments

despite the different initial loads. However, during the first 24 h the 40 g/L glycerol experiment

gave the highest uptake rate (Figure 40A). As expected, 80 g/L glycerol were not fully consumed

by the end of the experiment at 168 h. No significant differences were noted in biomass

(A) (B)

Figure 36: (A) Time-course profiles of cellular concentration (B) Final values of cell and oil concentration and oil content at 72 h of cultivation.


101

production. However, by comparing the biomass at 72 h, the 40 g/L experiment had higher

growth and reached stationary phase after that. The oil concentration was similar between all

fermentations, however greater increase was observed for the lower glycerol concentrations,

while the system with the highest concentration required more time to reach nearly 2 g/L.

Nitrogen consumption was also similar, but in the case of 80 g/L initial glycerol, the consumption

was slower for the first few hours. Finally the pH of the system (Figure 40B) confirmed that the

yeast could survive highly acidic conditions.

Figure 37: (A) Cell concentration (B) Glucose (SD<5%) and (C) oil content and titre profiles and (D) specific glucose uptake rate during the cultivation of Rh. glutinis on glucose based media.

(A)

(B)

(C)


102

Figure 38: Graphical estimation of the (A) biomass yield on glucose (YL/Glu) and (B) and oil yield on glucose (YL/Glu) of Rh. glutinis cultivated in glucose based media.

(A) (B)

(A) (B)

(C) (D)

Figure 39: (A) Glycerol consumption, (B) Growth (C) Oil production and (D) nitrogen (TN) consumption (SD<3%) profiles during the cultivation of Rh. glutinis on different glycerol concentrations. In the experiment with 40 g/L initial glycerol, no TN was analysed.


103

The yield of biomass on glycerol was 17.8% for the 30 g/L glycerol experiment, 36.7%

for the 40 g/L glycerol and 34.9% for the 80 g/L initial glycerol (Figure 41A). On the other hand,

the conversion yield of glycerol to oil was much lower than the biomass yield but the 30 g/L

glycerol resulted in higher glycerol conversion (12.8%). Glycerol was consumed more than

glucose previously and the biomass obtained from it reached higher concentrations. The oil titre

was similar but slightly lower in the case of glycerol at 72 h. However prolonging the cultivation

increased substantially the oil yield. Finally, the improved yields, which were obtained with

glycerol suggest that the yeast had been successfully adapted to glycerol and that this is a

suitable carbon source for the fermentation.

Figure 40: (A) Maximum specific glycerol uptake rate by Rh. glutinis cultivated in different initial glycerol concentrations in batch mode (B) Profiles of pH during the cultivation of Rh. glutinis on different glycerol concentrations.

(A) (B)

Figure 41: Yields of (A) biomass and (B) oil on consumed glycerol.

(A) (B)


104

Improvement of staining procedure for qualitative oil observation

In the previous chapter, the staining method by Thakur (1988) failed to stain properly

the lipids. In particular, they had developed two staining methods, one for microscopic

observations and another for spectroscopic quantification of the oil content (Thakur et al., 1989).

In this section a modification of both methods is presented with application of the latter to

microscopic analysis as well. This analysis was performed using samples from the 40 g/L initial

glycerol fermentation.

Figure 42 shows the micrographs taken by applying the staining method (Thakur et al.,

1988) using two different microscopes. As can be seen, the Safranine obstructed the clear

observation of lipids, as reported in the previous chapter. However the droplets can be seen as

grey coloured globules within the yeast cells. Thakur et al., (1988) described the oil droplets as

light blue and linked this colour with the extent of saturation of the fatty acids, which changed

with time. Probably, the time of staining with Safranine could be reduced since it seems to have

over stained the whole sample and maybe affecting the observation of oil. In fact, it was simply

removed from the procedure (Figure 42C) and this improved the observation of lipid droplets.

Critical in the spectrophotometric method are the boiling (binding of the stain to the

lipids) and the washing steps (removal of excess stain). Initially the effectiveness of the washing

was evaluated using the supernatants of the centrifuged samples after the addition of distilled

water. According to this work, three washing steps are sufficient (Figure 43A). The absence of

stain in the supernatant should suggest that the reading would not be affected by remaining dye

in the sample. Figure 43B shows how the absorbance of the supernatant was reduced by the

washing treatment. Indeed, after three washing steps the absorbance was very low.

By following the method of Thakur et al, the absorbance of the sample took negative

values, suggesting that the absorbance of the sample was almost equal to the blank. This

indicated that the stain did not bind with the oil and had been removed at the washing steps. So,

the binding step was investigated by applying different boiling times and measurement of the

absorbance of each treated sample. Prolonging the boiling time increased the reading of the

sample by about four times. Sixty seconds did not have much difference from 90 s but extended

boiling can break the cell wall and decrease the accuracy of results. Micrographs were taken to

investigate that and are shown in Figure 44. Longer contact with the stain was more effective for

the oil staining. However, the integrity of the cells from the 90 s treatment was not as high as it

was with 60 s boiling as a result of the high temperature. This led to the selection of 60 s as the

optimal boiling time. By applying this to other samples, Figure 44C was obtained.


105

Figure 42: (A) pictures of Rh. glutinis taken by Leica microscope with exact application of the method of Thakur et al., (1988) (B) by an OLYMPUS microscope with the same method (C) Improved staining method pictured by a Leica microscope. All photos are taken using the 100x magnification lens.

Figure 43: (A) Absorbance of supernatant from cell suspensions after washes to remove excess stain (B) Absorbance of stained cell sample at 580 nm, treated at different boiling times.

(A) (B)


106

6.4.4 Growth on synthetic crude glycerol

Case study: is crude glycerol always inhibitory?

Methanol and salts, contained in crude glycerol, are common growth inhibitors for most

microorganisms and their presence in the fermentation medium is likely to have negative effects.

Many researchers have identified growth inhibition by crude glycerol (Liang et al., 2010) while

others have not seen significant inhibition (Papanikolaou and Aggelis, 2002). Although in the

present thesis focus is given to microbial oil production, rather than the raw material, the

process would be ultimately applicable to a biodiesel biorefinery where crude glycerol would be

utilised.

It is likely that crude glycerol could have negative effects on the fermentation but the very

concentrated and dark-coloured solution of crude glycerol would be diluted in order to achieve

a specific glycerol concentration in the target media. For a working volume of 1 L, the

components of crude glycerol might not affect the microorganism. Some simple calculations

based on different crude glycerol scenarios were performed by adjusting the volumes and

amounts that are used in this thesis in order to examine whether applying crude glycerol would

affect, and by how much, the process. Several cases of crude glycerol from the literature were

used for the calculations (Table 11).

Figure 44: (A) sample boiled for 30s, (B) sample boiled for 60 s, (C) sample boiled for 90s, (D) Sample boiled for 60 s after applying a modified spectrophotometric method.


107

Table 11: Crude glycerol compositions derived from literature

To identify the glycerol concentration in the stock solution and determine the required

volume to achieve the target concentration, Equation 4-6 was utilised:

Purity of glycerol · Density of glycerol (g/L)= Concentration (g/L) (4-6)

Stock 1 (Table 11):

The glycerol concentration is: LgL

mL

mL

g/

.819

1000261

100

65

If the desired concentration is 30 g/L then the stock crude glycerol solution should be

diluted 27 times. Then every other component would be diluted 27 times. The final methanol

and NaCl content would be: 1.4 g/L each.

Stock 2 (Table 11):

Similarly the stock glycerol concentration is 712 g/L and is diluted 24 times to give the

desired glycerol concentration. Therefore the methanol concentration would be 11.8 g/L and the

NaCl 6.38 g/L.

Stock 3 (Table 11):

The stock glycerol concentration is 1071 g/L and is diluted 36 times to give the desired

glycerol concentration. Therefore the methanol concentration would be 0.42 g/L and the NaCl

1.25 g/L.

According to the above calculations the inhibitors undergo significant dilution during

their addition into the culture media. Even if higher glycerol concentrations were desired and

the stock solution diluted less e.g. 10 times, then the highest methanol content (based on Table

11) would still only be 12.8 g/L. In the case of 28.3 % methanol (stock 2) the methanol amount

would be higher but that option could be potentially avoided due to the low glycerol content. If

Stock Glycerol (% w/v)

Methanol (% w/v)

Soap (% w/v)

Salt (%w/v)

Water Reference

1 65 4 - 4 27 (González-Pajuelo et al., 2005) 2 56.5 28.3 15.3 - - (Luo et al., 2016) 3 85 1.5 - 4.5 6 (Chatzifragkou et al., 2011)


108

there is no choice, higher dilution based on a different feeding plan should be considered (e.g

utilisation of fed-batch cultivation).

Fermentation of Rh. glutinis on ‘crude glycerol’: Dependence of the specific

growth rate on the presence of inhibitors

In conjunction with the above calculations it was considered useful to identify the limits

of the system when inhibitory compounds are present in the medium. For that reason different

methanol and sodium chloride amounts were added into fermentation media and the growth

curve and specific growth rate were determined. The amounts of methanol and sodium chloride

do not correspond to real situations (since they are in high amounts) but were applied in order

to determine the capacity of the system.

Individual and combined effect of impurities

The effect of methanol on growth

Methanol concentrations ranging from 0 to 120 g/L were added into the fermentation

media. The growth curves of Rh. glutinis are depicted in Figure 45, along with point values for

the end of the cultivations. There were no major differences between 0 and 40 g/L of methanol.

The inhibition was more obvious after 60 g/L methanol. All cultures had lag phases of about 4 h

and differences in growth appeared at around 5 h of the fermentation. The specific growth rate

for every concentration is shown in Figure 45C, where the inhibitory effect of high methanol

concentration is clear. The specific growth rate followed a downward trend, almost a straight

line up to 40 g/L, then steeper from 40 to 80 g/L methanol. Finally, there was no change in the

specific growth rate beyond 80 g/L methanol, where the high amount of methanol did not allow

the system to reach sufficient cell densities. This lower rate of growth resulted in significant drop

in the final DCW at 72 h. A similar study on the growth rate of Cupriavidus necator (Salakkam

and Webb, 2015), obtained a similar linear profile for increasing methanol up to 40 g/L. These

results demonstrated that Rh. glutinis was able to grow in the presence of methanol up to 40 g/L

without major inhibition. Specifically, for lower concentrations (<25 g/L) the specific growth

rate was not significantly affected and the final concentration reached similar levels to previous

experiments. However, inhibition was observed for higher concentrations of methanol while at

very high levels, the low growth rate did not allow for sufficient biomass.


109

The effect of salt and methanol on growth

In this experiment, the effect of salt (NaCl) on the growth of Rh. glutinis was examined. As

can be seen in Figure 46, NaCl did not cause major inhibition at concentrations up to 100 g/L.

The highest cellular concentration was obtained with the 50 and 70 g/L NaCl, while the 100 g/L

NaCl culture had the lowest growth rate with a final cell concentration being about 2/3 of the

highest (5.8 g/L). This suggests that the yeast could grow well in the presence of sodium chloride.

Similarly, increasing NaCl promoted biomass production for Rh. toruloides (Gao et al., 2016).

Presumably after 70 g/L osmotic stress was caused which resulted in a slower growth rate.

On the other hand when both NaCl and methanol were present in the media different

trends were seen at concentrations of NaCl, where growth had been unaffected when NaCl was

the only impurity (Figure 47). As with the cultures with just methanol, 60 g/L were inhibitory in

the presence of NaCl.

Conclusively, methanol is inhibitory for the system at very high concentrations. However,

considering the preliminary calculations the concentrations of methanol in the final media, it is

unlikely to be that high and crude glycerol is therefore not expected to be inhibitory if the

concentrations upon dilution are below the determined levels for the system (CH3OH<20 g/L,

NaCl<70 g/L).


110

(A) (B)

Figure 46: (A) Time-course profile of cellular concentration for different sodium chloride concentrations (B) DCW at 72 h of cultivation for different sodium chloride concentrations.

Figure 45: (A) Time-course profiles of growth of Rh. glutinis on different methanol concentrations during the whole fermentation, (B) Profiles of growth during the first 24h of fermentation on different methanol amounts, (C) Specific growth rate obtained for different methanol load, (D) DCW at 72 h of cultivation against increasing methanol concentrations.

(A)

(D)

(B)

(C)


111

Crude glycerol fermentation

Rh. glutinis was cultivated on synthetic crude glycerol. The growth rate was typical and

no negative effect was observed (Figure 48). In particular, biomass production was higher than

other times. Similarly to previous results with pure glycerol, the stationary phase started at

around 48 h. Using the same crude glycerol formula Uckun Kiran et al., (2013) observed growth

inhibition only beyond 60 g/L glycerol concentration. Rh. glutinis accumulated 15 % of its

cellular weight as oil which is quite promising. Further process optimisation is needed with pure

glycerol in order to establish a proper feeding process for crude glycerol utilisation.

Figure 47: (A) Dynamic cellular concentration for different methanol and sodium chloride contents. (B) Specific growth rate for the period 0-24 h. M stands for methanol and S for sodium chloride.

(A)

(B)


112

6.5 Conclusions

In this chapter, several cultivation conditions for Rh. glutinis were examined. The results

showed that the yeast can grow in a very simple medium despite the low pH. A fully inorganic

cultivation medium was not shown to be significantly superior to the semi-defined media with

only yeast extract as the nitrogen and nutrient source. Furthermore, aeration was beneficial for

growth. Oil is synthesized at a very low rate for the first48 h of the fermentation and accumulates

when the yeast is at stationary phase so harvesting should take place after 72 h. Regarding the

inhibitors in crude glycerol, dilution will take these impurities to levels where they are not

inhibitory for the growth of Rh. glutinis and a fermentation on synthetic crude glycerol

demonstrated this. These preliminary experiments helped to establish a simple cultivation

medium and conditions for the remaining Rh. glutinis research reported in this thesis.

Figure 48: Cellular concentration of Rh. glutinis and oil content and titre when cultivated on synthetic crude glycerol.

113

Chapter 7

A biorefinery approach to microbial oil

production from glycerol by Rhodotorula glutinis


114

7 A biorefinery approach to microbial oil production from glycerol by Rhodotorula glutinis

7.1 Introduction

The use of biodiesel-derived glycerol as a carbon source for microbial oil production is a

biorefinery engineering strategy aimed at reducing the glycerol surplus and making the

microbial oil process more cost-effective. In this chapter, glycerol was used as the sole carbon

source for the cultivation of the yeast Rh. glutinis using a very simple medium (glycerol plus yeast

extract) with no pH control in order to evaluate the influence of different nutrient-limiting

strategies. Cell growth and lipid formation are competitive processes but adequate cell density

levels are essential to achieve high oil yields. To this end, the present chapter reports the impact

of different feeding strategies on glycerol uptake by Rh. glutinis under fed-batch conditions,

where also significant increase in cell density was achieved. The results constitute part of a paper

published in the journal Biomass and Bioenergy (Karamerou et al., 2016). The theoretical

background and some supplementary information have been added to fit with the context and

format of the present thesis. The published article can be found in APPENDIX 1.


7.2.1 Calculation of the oxygen uptake rate

The dissolved oxygen in a well-mixed microbial system is given by Equation (2-1) in

Chapter 2. The Oxygen Uptake Rate (OUR) is an important parameter, which gives information

about the metabolic activity and status of the cells and can be determined experimentally

(Garcia-Ochoa et al., 2010). In most cases DO is high at the beginning of the fermentation, it

reaches low values during the exponential growth phase and increases towards the end of the

cultivation. OUR is high during fast growth but decreases when the requirements of the cells

become less.

A review by Garcia-Ochoa (2010) reports common methods for OUR determination: the

gas balancing method, the dynamic method, the yield coefficient method or by knowing the

oxygen transfer rate (OTR). The gas balancing method requires accurate oxygen analysers for

the incoming and unconsumed oxygen additionally to the oxygen probe, which is fitted to the

bioreactor. The yield coefficient method, makes sometimes unrealistic assumptions about the

dependence of the biomass yield on the carbon source and the last method for OTR

determination may not have taken into account the broth properties and the presence of

Chapter 7 A biorefinery approach to microbial oil production

115

microorganisms. Amongst these methods, the dynamic method comes as the most simple and

realistic method for determining the oxygen uptake rate within the actual culture and operating

conditions without special instrumentation and is based on the respiratory activity of the cells.

According to the dynamic method (Garcia-Ochoa and Gomez, 2009; Garcia-Ochoa et al.,

2010), the air supply to the bioreactor is interrupted momentarily (so as not to disturb the

experiment) and the decrease in DO values is attributed to the respiratory activity. When the air

is reintroduced to the bioreactor, the DO levels return to the initial values and this technique can

be applied many times throughout the cultivation process. The OUR is then calculated from the

slope of the decreasing values of Dissolved Oxygen concentration (DO), as shown in Figure 49

and expressed as Equation (7-1). From such a graph, the OUR, qO2 and kLa can be easily

calculated graphically (Alonso et al., 2012). In particular, the specific oxygen uptake rate (qO2,

mg/g/h) is calculated by dividing the OUR with the cell concentration (X, g/L) at each particular

time point, Equation (7-2). With the OUR known, the kLa can be calculated from the slope of by

plotting CO2 Equation (7-3). With known kLa, the oxygen transfer rate (OTR) can be calculated

from equation (7-4) and this can characterise the culture as oxygen-limited or not by comparing

it to the OUR. The OTR should be higher or equal to the OUR for not having oxygen limitation.

12

12

tt

DODO

t

DO tt

OUR (7-1)

XhgmgqO

OUR)//( 2

(7-2)

dt

dDOXq

akDODO O

L2

1* (7-3)

)((mg/L/h)OTR * DODOkL (7-4)

Where DO* (mg/L) is the oxygen saturation concentration in the liquid phase at

equilibrium to the gas phase.

The method assumes no oxygen exchange between the gas and liquid phase during the

interruption and no change in fluid dynamics of the broth. It also faces limitations at very low

levels of DO (<10% saturation) since the responses from the probe are inaccurate at these low

levels and there is not much oxygen available for uptake.


116

7.3 Methodology

7.3.1 Microorganism and media

Seed cultures of Rh. glutinis were prepared as follows: a 500 mL shake-flask was filled

with 100 mL medium, containing 20 g/L glycerol and 10 g/L yeast extract at pH initially adjusted

to 5.5. The inoculated yeast was left to proliferate for 24 h at 30°C in an orbital shaker at 200

rpm. From that seed culture, a 10% v/v inoculum was transferred to the glycerol and yeast

extract production media. Pure glycerol (≤99%), yeast extract and agar were purchased from

Sigma-Aldrich (Germany) while peptone was supplied by Oxoid (UK).

7.3.2 Shake-flask experiments

The shake-flask production media contained yeast extract and glycerol as nitrogen and

carbon source respectively. The mass fraction of nitrogen in yeast extract was confirmed by

measurement of total nitrogen to be 10% and this was taken into account for all the calculations.

For the specific growth rate estimation, Rh. glutinis was cultivated in different glycerol

concentrations at a constant nitrogen concentration of 0.8 g/L (supplied as 8 g/L yeast extract).

For the specific growth rate dependence on nitrogen, increasing amounts of ammonium sulphate

(purity 98.5%, SLS, UK) were used in combination with a constant glycerol concentration of 30

g/L. Since the main nitrogen source in the experiments (shake-flasks and bioreactor) was yeast

extract but here this could not be modified without increasing the amounts of other nutrients

Figure 49: Graphical estimation of oxygen uptake rate and volumetric mass transfer co-efficient. This figure is from DO data during the Dynamic method from this research.


117

present in it, ammonium sulphate was used to modify only the nitrogen concentration. For

general nutrients supply a basic amount of yeast extract (1.5 g/L) was present in all cultures at

the same concentration and ammonium sulphate was acting additionally to it in order to provide

the desired total nitrogen amount. A control flask with only yeast extract and glycerol provided

a baseline for the ammonium sulphate effect.

For the oil accumulation study, nitrogen limited media containing a constant amount of

0.2 g/L of N (2 g/L yeast extract), with different concentrations of glycerol from 30 to 80 g/L,

were used to give different initial C/N ratios. The agitation rate for all shake-flask experiments

was 200 rpm and the incubation temperature was 30°C. The initial pH before inoculation was

5.5.

7.3.3 Batch bioreactor experiments

For batch experiments, a 2 L bioreactor (Electrolab, UK) was used with 1L working

volume. The medium composition was 30 g/L glycerol, 2 g/L yeast extract and the pH was

adjusted to 5.5. For foam prevention 1 mL/L antifoam A (Sigma –Aldrich) was added to the

production medium prior to the pH adjustment. The incubation temperature was 30°C and the

agitation rate was 400 rpm. The pH was monitored but not controlled. Different aeration rates

(0, 0.5, 1 and 2 L/min) were applied. Dissolved Oxygen (DO) was measured with a polarographic

electrode (Broadley-James, UK) as Dissolved Oxygen Tension (DOT, percentage of air saturation)

and was also monitored but not controlled. Figure 50 shows the experimental set up of this group

of experiments.


118

7.3.4 Fed-batch bioreactor experiments

Fed-batch cultivations were conducted in the same 2 L bioreactor, by pulse-feeding at

regular intervals of 24 h, a 40 mL concentrated medium containing both glycerol (375 g/L) and

yeast extract (50 g/L) at final concentrations above 15 g/L and 2 g/L respectively, unless

otherwise stated. The working volume was 1 L. All starting and incubation conditions were the

same as those for the batch experiments mentioned previously and the aeration rate was 0.5

L/min. A Watson-Marlow 505U peristaltic pump (UK) was used to introduce the feeding medium

into the bioreactor. Figure 51 shows the experimental set up of this group of experiments.

Figure 50: Set up of a batch aerated fermentation of glycerol using Rh. glutinis. The inoculation bottle can be seen on the top of the fermenter stand. Photo taken from this work.

Figure 51: Set up of a pulsed fed-batch aerated fermentation of glycerol using Rh. glutinis. The additions include the pump for the rapid introduction of medium and two feeding bottles can be seen on the right and part of the air pump below the feeding pump, shown in detail at the left bottom edge. Photo taken from the research reported in this thesis.


119

7.3.5 Specific growth rate

The specific growth rate, μ (h-1) was calculated during the exponential growth phase as

the slope of the regression line between the logarithmic differences in cellular concentration

versus time according to the Equation (7-5).

0

0)/ln(

tt

XX

(7-5)

where X (g/L) and X0 (g/L) are the cellular concentrations at time t (h) and the beginning

of the exponential growth phase t0 (h), respectively.

7.3.6 Specific substrate uptake rate

The specific substrate uptake rate qS (g/g/h) was calculated according to Equation (7-6)

as the division of the substrate consumption by the cellular concentration:

Xt

SqS

(7-6)

Where S (substrate) is either glycerol or nitrogen (g/L), t (h) is the time and X (g/L) is the

cellular concentration.


Glycerol, nitrogen and oil analysis were performed as described in Chapter 4.

7.4 Results and discussion

7.4.1 Effect of glycerol concentration on growth and oil accumulation

To trigger lipid accumulation, the amount of carbon in the medium must be more than

that required for growth and maintenance of the yeast so that there is surplus for lipids

synthesis. However, glycerol has been reported to cause growth inhibition at high

concentrations due to osmotic stress on the cells (Rywińska et al., 2010; Raimondi et al., 2014).

To study the effect of initial glycerol concentration on the growth and oil yield and to identify a

suitable starting amount, Rh. glutinis was cultured on different initial glycerol concentrations,


120

ranging from 5 to 150 g/L. Figure 52A shows the specific growth rate of Rh. glutinis cultivated

on increasing initial concentrations of glycerol in batch mode in shake-flasks under nitrogen rich

conditions. The specific growth rate increased sharply with increasing glycerol up to 40 g/L (C/N

93 initial atom ratio), then reached its maximum value of 0.27 h-1 at 40 g/L glycerol but after that

decreased until 60 g/L. The specific growth rate eventually reached a plateau, meaning that

higher concentrations of glycerol had neither a positive nor a negative effect on growth. The

results indicate that lower glycerol concentrations were easily assimilated by the cells,

promoting growth. This is in agreement with the results of Raimondi et al., (2014), where

glycerol inhibition occurred above 40 g/L glycerol. Inhibition due to high concentration of

glycerol was observed from (Meesters et al., 1996; Liang et al., 2010) while evaluating the impact

of initial glycerol concentration on the growth of C. curvatus using pure glycerol. However,

growth was supressed above 64 g/L of glycerol. Liang et al., (2010) extended the study on the

same strain by applying different amounts of crude glycerol and reported a similar trend in

growth to that obtained when using lower concentrations of pure glycerol. Moreover, Uckun et

al., (2013) also reported same effect on R. toruloides, where growth suppression was obtained

above 70 g/L and 50 g/L of pure and crude glycerol, respectively. In contrast, Papanikolaou et

al., (2002) did not detect inhibition from crude glycerol on the growth of Y. lipolytica. The

different responses to initial glycerol concentration could be strain dependent (Papanikolaou

and Aggelis, 2002), but the point where inhibition occurs could be the result of the presence and

concentration of other nutrients in the medium that may alleviate the inhibitory effects due to

the dilution, which crude glycerol was subject to when added to the medium.

A similar profile to the specific growth rate was observed with the intracellular oil

content (Figure 52B,II) though under higher nitrogen limited conditions. These results are

summarised in Table 12. The maximum value was achieved at 60 g/L initial glycerol (ratio C/N

137) and dropped after that. From that, it can be deduced that higher glycerol concentrations

induce the accumulation of lipids as a result of supressing cellular growth, though the increase

of lipids cannot be unlimited since the decrease after 60 g/L means that further increase of initial

glycerol will not increase the lipid content. In fact here, the amount of residual glycerol followed

an increasing trend (Figure 52B, II), showing on the one hand that lower glycerol amounts were

used more effectively, resulting in adequate oil content and growth, while on the other hand

inhibition resulted in less glycerol conversion for higher initial amounts of it. Less glycerol

conversion was also observed in shake-flask cultures of R. toruloides on high initial crude

glycerol concentrations (180 g/L), however, the decrease in growth was gained in oil production

(54% w/w oil content compared to 40% at 120 g/L) (Tchakouteu et al., 2015). The maximum


121

achieved specific growth rate was 0.27 h-1 at 40 g/L glycerol, while the highest oil content at

60 g/L at a percentage of 29.8% w/w oil content. However, because lower glycerol

concentrations were used more effectively and did not cause inhibition of growth, they should

be preferred, for example 30 g/L, if one considers the amount of residual materials left after the

fermentation.

Table 12: Impact of initial glycerol concentration on the growth and lipid production of Rh. glutinis under nitrogen limited conditions. The initial TN concentration was 0.2 g/L (2 g/L of yeast extract) and the fermentation time 72 h in all cases.

Gly0a

(g/L) C:N

(atom ratio)

Glyconsb

(g/L) DCW (g/L)

Oil content (% w/w)

Oil concentration (g/L)

YX/Gly (g/g)

YL/Gly (g/g)

30 68 12.05 5.28 14.5 0.77 0.44 0.064 34 78 12.00 4.86 19.4 0.94 0.41 0.078 39 89 20.45 4.68 24.5 1.15 0.23 0.056 43 98 17.97 4.27 20.0 0.85 0.23 0.047 52 119 22.48 4.43 26.3 1.17 0.18 0.052 56 128 19.94 4.69 21.0 0.98 0.24 0.049 60 137 17.68 4.06 29.8 1.21 0.23 0.068 70 160 19.15 4.9 16.2 0.79 0.25 0.041 80 182 8.06 4.62 17.8 0.82 0.54 0.101

a Gly0 stands for the initial glycerol concentration b Glycons stands for the consumed glycerol

Figure 52A: The effect of initial glycerol concentration on the specific growth rate of Rh. glutinis CICC

31596, cultured in shake-flasks, on increasing initial glycerol and yeast extract at 8 g/L (diamonds),


122

Figure 52B: The effect of initial glycerol concentration on the oil content an glycerol consumption of Rh. glutinis CICC 31596, cultured in shake-flasks. (II) Oil content (%, w/w) at 72 h on increasing initial glycerol and yeast extract at 2 g/L, (C) Residual glycerol at 72 h on increasing initial glycerol and yeast extract at 2 g/L.


123

7.4.2 Effect of initial nitrogen concentration on the specific growth rate

and oil accumulation

Nitrogen exhaustion is the critical stimulus for transition between cell proliferation and

lipid synthesis as well as the necessary condition for adequate growth levels. To identify the

effect of initial nitrogen concentration on the growth and lipid yield, Rh. glutinis was cultured on

different initial nitrogen concentrations with constant glycerol at 30 g/L in shake-flasks for 72 h.

Details on the modification of the initial nitrogen concentration are summarized in Table 13.

Figure 53A shows the specific growth rate according to initial nitrogen concentrations. The

specific growth rate increased slightly by increasing nitrogen and then remained almost constant

at 0.2672 h-1. The growth kinetics related to nitrogen were described well by a Monod model

(Equation 7-7).

NK

N

N max (7-7)

Where μ is the specific growth rate (h-1), μmax the maximum specific growth rate equal to

0.2672 h-1, N is the total nitrogen concentration (g/L) and KN the half saturation constant at 0.05

g/L (obtained by data fitting in Excel).

Using Equation (7-7), the theoretical specific growth rate was calculated for several

initial nitrogen concentrations and the fit is shown in Figure 53A, confirming that increase in

nitrogen would not increase further the specific growth rate. The higher rates obtained for

TN≥0.4 g/L resulted in higher final cell concentration at 72 h, between 8 and 9 g/L, as shown in

Figure 53B). By plotting the maximum specific nitrogen consumption rates an increasing trend

was seen for concentrations up to 0.4 g/L TN (Figure 53C). However, no further increase of the

rate appeared for the two higher nitrogen concentrations (0.7 g/L TN and 0.95 g/L TN), which

is consistent with the specific growth rate results. The oil content of the cells was higher for

lower TN concentrations (26.99 % oil content and 28.7% oil content for 0.2 and 0.25 g/L of TN

respectively), while the opposite (lower oil content) happened for higher amounts than 0.4 g/L

TN (Figure 53D). Nitrogen should be exhausted for lipids to be synthesized and it happened for

the low concentrations of TN, while in the rest of the cultures there was still nitrogen available

(<0.27 g/L) to support further growth. The final cell concentration is in agreement with the

amount of consumed TN, with higher amount of nitrogen consumed resulting in higher cell


124

densities (Table 13). Lower nitrogen amount resulted in earlier initiation of the oil accumulating

phase, leading therefore to a larger amount of glycerol channelled to oil production. In the case

of 0.2 g/L of TN where almost all glycerol was consumed (27.79 g/L) when 28.7% oil content on

the cellular mass was produced (Table 13).

Table 13: Summary of results obtained from shake-flask batch cultivations of Rh. glutinis on different initial nitrogen concentrations. The initial glycerol concentration was 30 g/L and the fermentation time 72 h in all cases. N from YEa

(g/L) N from ASb

(g/L) DCW (g/L)

Oil content (% w/w)


Consumed glycerol (g/L)

Consumed nitrogen (g/L)

0.15 0 5.1 26.3 1.34 21.98 0.15 0.15 0.05 7.35 27.0 1.98 26.60 0.20 0.15 0.1 7.25 28.7 2.08 27.79 0.29 0.15 0.25 8.7 15.8 1.38 23.92 0.40 0.15 0.55 8.95 18.3 1.64 21.15 0.49 0.15 0.8 8.15 22.9 1.87 24.21 0.47 a YE stands for yeast extract b AS stands for ammonium sulphate

Figure 53: The effect of initial nitrogen concentration on the growth, oil content and nitrogen uptake rate of Rh. glutinis CICC 31596, cultured in shake-flasks for 72 h. TN 0.15 g/L corresponds to the control culture with only yeast extract. (A) Monod dependence of specific growth rate on initial total nitrogen concentration, (B) Cell concentration at 72 h, (C) Specific nitrogen uptake rate for increasing initial total nitrogen (TN) concentrations, (D) Yield of cellular growth on nitrogen (TN) (blue bars) and oil content of cells for different nitrogen amounts (grey bars).


125

7.4.3 Effect of air on growth and oil production

Like most oleaginous yeasts, Rh. glutinis is an aerobic microorganism (Pan et al., 1986;

Saenge et al., 2011b) and its growth and lipid production are dependent on aeration (Yen and

Zhang, 2011; Yen and Liu, 2014). Batch fermentations in a 2 L bioreactor were performed in

order to investigate further the effect of aeration rate. Aeration rates from 0 to 2 L/min were

performed using an initial glycerol concentration of 30 g/L. Compared to shake-flask cultivation

of Rh. glutinis at the same glycerol concentration, (μmax = 0.27 h-1) there was a clear increase in

the specific growth rate (0.31 h-1 and 0.37 h-1 in non-baffled and baffled bioreactor, respectively)

and the obtained lipid yield in the bioreactor, indicating that air supply plays an important role

in the yeast growth. Figure 54A shows that the aeration rate of 0.5 L/min gave the highest

specific growth rate of 0.31 h-1 while the absence of an air supply inhibited the growth of the

yeast and did not provide adequate cell concentration to allow for oil measurement. However,

higher aeration rates were not beneficial for growth or oil production. Use of baffles in the same

bioreactor raised the specific growth rate by about 20% and led to the highest final cell

concentration of 5.3 g/L (Figure 54B). Although the reasons why low aeration rates are better

are not really clear, one possibility could be the size of air bubbles in the bioreactor. Higher air

flow rates create larger air bubbles, which have a smaller specific surface area compared to the

smaller bubbles seen at low air flow rates. The mass transfer of oxygen is therefore likely to be

slower in the former case. Modification of agitation rates could be a solution to achieve higher

DO concentration. Figure 54C shows the oil content, cell and oil concentration for the different

aeration rates at 72 h.

The air flow rate of 0.5 L/min gave the highest oil content of 31.58% of the cellular mass

in the case of non-baffled bioreactor and 33% in the case of baffled bioreactor confirming the

beneficial effect of lower air flow rate. Moreover, for 0.5 L/min, the use of baffles also improved

the oil concentration (1.68 g/L) and lipids-free biomass (3.6 g/L), which are higher than the

other cases.

The specific consumption rates of glycerol and nitrogen for non-baffled bioreactor are

shown in Figure 55 and indicate that the rate of consumption of nutrients is also dependent on

the aeration rate. In particular, the specific consumption rate of both nutrients decreased linearly

from 0.5 L/min with increasing aeration rate, except for the case of 0 L/min air flow rate, where

the glycerol and nitrogen were not consumed. The specific glycerol and nitrogen consumption

rates were 33.98 g/g/h and 0.35 g/g/h respectively.


126

Figure 54: The effect of air flow rate on specific growth rate, final growth and oil mass fraction of Rh. glutinis CICC 31596. (A) Specific growth rate dependence on aeration rate in a 2 L non-baffled bioreactor, (B) Cell concentration at stationary growth phase, non-baffled 2 L bioreactor (Blue diamonds), baffled 2 L bioreactor (Red diamond), (C) Oil mass fraction of the cells along with the cell and oil concentration for baffled and non-baffled bioreactor for different aeration rates at 72 h of cultivation. Assays analysed in triplicate and SD less than 10%.


127

7.4.4 The effect of feeding strategies on growth, oil yield and glycerol

consumption of Rh. glutinis

Fed-batch cultivations were performed in order to tackle the case of glycerol inhibition,

by supplying moderate amounts of it and to increase the cell density with the aim of improving

the oil yield in terms of concentration. In the first approach (FB1), three feedings containing

glycerol and yeast extract in order to maintain the final concentrations at least 15 g/L and 2 g/L

(0.2 g/L TN) respectively took place, one every 24 h and cell harvesting for oil was done 48 h

later after the last feeding. In the second strategy (FB2), five glycerol and yeast extract feedings

with the same concentrations took place again one every 24 h, plus one feed containing only

glycerol at 750 g/L (final concentration 30 g/L) targeting oil accumulation.

As Figure 56C shows, moderate glycerol supply did not lead to any inhibition and all

added glycerol was totally consumed as well as the nitrogen (Figure 56D) in both cases. The

results of the fed-batch cultivations are summarised in Table 14. As can be seen in Figure 56C,

the low glycerol amount supplied each time (15 g/L) in the first strategy (FB1) was not enough

for improving the oil production. The fact that glycerol had been consumed at the end of the

cultivation, resulted in a slight decrease in oil content, which can be attributed to use of lipid for

cell maintenance, as it has been reported for other yeasts (Zhang et al., 2011). However, this

strategy did lead to higher biomass values (9.4 g/L) in comparison to batch cultures (5.3 g/L).

On the contrary, the implementation of one last pulse feeding of glycerol enabled the use of all

added glycerol for oil accumulation (30.6% oil content at 168 h and 34.62% at 192 h) in

comparison to lower oil contents achieved during the first fed-batch approach (27.5% oil content

of the dry cell weight at 168 h) and the batch cultivations. The lipid-free biomass was also higher

(A) (B)

Figure 55: (A) Maximum specific glycerol uptake rate for different aeration rates in a 2 L bioreactor, (B) Maximum specific nitrogen uptake rate for different aeration rates in a 2 L bioreactor.


128

in the second fed-batch cultivation (FB2, Figure 56B). Unfortunately due to foaming there was

considerable washout of cells from the fermenter so the two fermentations no longer followed

the same pattern in terms of cellular concentration.

Table 15 shows a comparison of different yeast cultivation approaches for microbial oil

production. The cell concentration (5.3 g/L) and oil content (33%) obtained in the present study

in batch mode were higher than those that obtained using Y. lypolitica, 4.68 g/L and 22.3%

respectively, cultured in pure glycerol along with yeast extract, ammonium sulphate and salts

(Makri et al., 2010). Despite the high oil content of 60% of the cellular mass of Rh. glutinis in

crude glycerol in fed-batch mode, the cell concentration it achieved was 10.5 g/L (Saenge et al.,

2011b), lower than the 16.8 g/L achieved in this study (Table 15). The oil yields on glycerol

(YL/Gly) are in most of the studies comparable and quite small (<0.11 g/g) except for Y. lipolytica

that displays yields around 0.35 g/g. Meesters et al., (1996) achieved a remarkable cell

concentration of 118 g/L using the yeast C. curvatus ATCC 20509, which led to an oil

concentration of 29.5 g/L with an oil content of only 25%. On the other hand, Y. lipolytica

produced 3.5 g/L of oil despite achieving 8.1 g/L of cells with a 43% oil content (Papanikolaou

and Aggelis, 2002). These two cases show that high cell density conditions are required in order

to obtain meaningful oil yields. Therefore, it is important to increase the cellular concentration

while obtaining sufficient oil contents. Likewise, the same strain of C. curvatus was found to

accumulate 44.6 % oil content of the cellular mass under fed-batch culture (Liang et al., 2010),

but the cell density was lower than those achieved by Meesters et al., (1996). As can be seen in

Table 15, the impact of increasing the cell density on the volumetric oil production (g of oil per

litre of bioreactor) was also found in the present study, suggesting that the stimulation of high

cell density conditions plays a key role in yielding high oil concentrations.

Considering the fact that Rh. glutinis, was cultivated on very simple nutrient medium,

compared to other yeasts, where additional nutrients were supplied along with the carbon and

nitrogen sources, the values obtained in this study are comparable to those obtained in other

cases. By using this simple medium and applying fed-batch cultivation the rise in cell

concentration was significant and the low pH did not affect the biomass production.


129

Table 14: Comparison of parameters obtained for the two fed-batch approaches.

Fed-batch1 Fed-batch2

Fermentation time (h) 168 192 Cell concentration (g/L) 9.4 16.8 Oil content (%, w/w) 27.5 34.62 Oil concentration ( g/L) 2.58 5.07 Productivity (g/L/d) 0.37 0.63 YL/Gly

a (g/g) 0.023 0.0307 Consumed glycerol (g/L) 116.33 166.62 Consumed yeast extract (g/L) 8 12 a L for lipids, Gly for glycerol

Figure 56: Time course profiles of cell growth oil production and nutrients consumption by Rh. glutinis for the two fed-batch approaches. (A) Kinetics of cell growth of Rh. glutinis. Black arrows represent feeds that were common for both cases, while dashed blue arrows represent the feeds that followed the three initial ones in the second fed-batch, (B) Oil content, dry cell weight and lipid-free biomass at 168 h of fermentation, (C) Glycerol consumption, (D) Total Nitrogen consumption.

Table 15: Comparison of microbial oil production approaches using yeasts cultivated on glycerol based media.

Yeast Cultivation medium C/N

(atom ratio)

Aeration rate

(L/L/min)

Cultivation mode

DCW (g/L)

Oil content (%, w/w)


YL/Gly

(g/g) Ref.

C.. curvatus ATCC 20509

Pure glycerol, yeast extract, salts

20-60 0.6 Fed-batch 118 25 29.5 0.11 (Meesters et

al., 1996)

Y. lypolitica ACA-DC 50109

Pure glycerol, yeast extract, salts, ammonium sulphate

82.7 0-3 Batch 4.68 22.3 1.34 0.08 (Makri et al.,

2010)

R.toruloides Y4 Crude glycerol, yeast extract, salts

60-90 - Fed-batch 12.1 50 6.1 0.12 (Uçkun Kiran et al., 2013)

Rh. glutinis TISTR 5159

Crude glycerol, ammonium sulphate

85 2 Fed-batch 10.5 60.7 6.1 0.058 (Saenge et al.,

2011b)

C. curvatus ATCC 20509

Crude glycerol, ammonium chloride, salts

35 0.8-1 Fed-batch 31.2 44.6 13.9 0.10 (Liang et al.,

2010)

Y. lipolytica LGAM S(7)1

Crude glycerol, yeast extract, ammonium sulfate, salts

- 1.8 Continuous 8.1 43 3.5 0.35 (Papanikolaou

and Aggelis, 2002)

L. starkeyi DSM 70296

Crude glycerol, yeast extract, peptone, salts

162 - Shake-flask 34.4 35.9 12.34 0.11 (Tchakouteu et al., 2015)

Rh. glutinis CICC 31596

Pure glycerol, yeast extract

68.4 0.5 Batch 5.3 33 1.68 0.038 This study



37-130 0.5 Fed-batch 9.4 27.5 2.58 0.023 This study



26-168.9 0.5 Fed-batch 16.8 34.62 5.07 0.0307 This study


131

7.4.5 Oxygen consumption during fermentation in 2-L bioreactor

A better insight on how oxygen was used per cell is given by the specific oxygen uptake

rate, shown in Figure 57. The high Dissolved Oxygen (DO) concentration at the beginning of the

fermentation in combination with the high oxygen transfer rates due to the low cell density

resulted in higher oxygen uptake rates per unit of biomass. During the cell growth phase, the DO

level decreased, resulting in less available oxygen per unit of biomass. Thus, the decrease in the

specific oxygen uptake rate was very sharp for the first 22 h, matching the offset of the stationary

phase and beginning of oil accumulation in the batch culture (Figure 57A). After that the rate

decreased while the cell concentration remained almost constant. For the fed-batch case (FB1),

in Figure 57B the oxygen uptake rate per unit of biomass (qO2) decreased with the production of

microbial cell mass in agreement to the first 24 h of the batch cultivation (Figure 57A). After that,

the qO2 increased at the beginning of the stationary growth phase, similarly to the batch case.

Rise in the value of the specific oxygen uptake rate was also observed after each feeding (FB1,

Figure 57B) indicating that growth was again stimulated after the addition of nutrients and the

cells were using the higher available oxygen conditions, which was a result of the deceleration

of growth before the substrate addition. Moreover, for the batch case, the lowest DOT level of

13% was observed during the exponential growth phase at 8 hours indicating that oxygen is

primarily used for growth purposes, while after entering stationary phase the DOT remained at

levels of 80%. This oxygen consumption evidence at stationary-lipid accumulating phase

indicates that oxygen participates in lipid formation as well.

(A) (B)

Figure 57: Time course profiles of specific oxygen uptake rate (diamonds) and cell concentration (squares) for 2 L bioreactor. (A) Batch cultivation (B) Fed-batch cultivation (FB1). The three arrows represent the feedings.


132

7.4.6 Lipid accumulation over time and the ability of Rh. glutinis to grow

and produce oil under acidic conditions

As mentioned earlier, when air is supplied, lipid droplets appear within the cells earlier

in the cultivation. Figure 58 shows the progress of lipid accumulation over time and the

competition with growth since no budding occurs during the lipid accumulating stage, while in

the minority of cells that continue to divide the mother cell has no oil droplet inside.

Interestingly, pH was not controlled since batch cultivations at shake-flask scale showed that

Rh. glutinis was able to grow under low pH values (pH<3). As Figure 58 shows, the low pH

encountered after 24 h did not affect negatively cell proliferation nor oil accumulation. Budding

cells occurred at 24 h and pH 3.06 while oil droplets developed at pH <3 accompanied by a minor

budding. Therefore, these results provide evidence that Rh. glutinis is a robust strain that is able

to endure acidic culture conditions. According to Sitepu et al., (2014), pH control contributes to

the overall oil production costs whereas pH-control free conditions not only do not have

detrimental effects on growth and oil yield, but also prevent bacterial contamination. In fact,

Saenge et al., (2011b) obtained very similar oil yields and glycerol consumption when culturing

Rh. glutinis under either pH-control free or pH-controlled conditions. Finally, the variation of

aeration had no effect on the cell morphology and did not limit growth neither prevented cells

from lipid accumulation. Stained cells are shown on Figure 59.


133

Figure 58: Phase contrast microscope pictures of Rh. glutinis at 1000x lens magnification, showing the time of cultivation along with the corresponding pH value. The yeast was cultivated in a 2 L bioreactor in batch mode with 2 L min-1 air flow rate. Bar represents 10 µm.

Figure 59: Sudan Black B stained cells of Rh. glutinis cultivated in shake-flask at phase contrast microscope at 100x magnification lens. Bar represents 10 µm.

7.5 Conclusions

This study showed that Rh. glutinis is able to consume glycerol and can adequately grow

and produce oil when cultivated on simple medium and pH-control free conditions. Furthermore

the application of a fed-batch strategy managed to increase substantially, cell growth and with

the right combination of feedings, targeting first growth and then oil accumulation, cell


134

concentration did not act at the expense of oil production. Finally, insights were given according

to the effect of air on the glycerol and nitrogen uptake, which showed certain trends according

to aeration, indicating that low aeration rates benefit the process variables. Further process

optimisation, such as glycerol feeding studies, can build on these yields and provide a scalable

process to produce microbial oil cost effectively.

Chapter 7 Supplementary information

135

Supplementary information for Chapter 7

Dissolved oxygen profiles

The Dissolved Oxygen Tension (DOT) profiles obtained during the batch experiments

under different oxygen supply conditions, reported in Chapter 6, are shown in Figure 60. The

same trend was observed under all of the conditions tested and regardless of the oxygen supply

strategy employed. A pronounced drop in DOT values was observed during 6-8 h which

coincided with the end of the exponential growth phase and the depletion of the N-source.

Oxygen transfer rate in 2-L bioreactor fermentations

The oxygen mass transfer coefficient, kLa, depends on the liquid properties and also on

the properties and nature of the medium employed. In addition, the cell proliferation and the

oxygen consumption by the yeasts may exert an effect on the kLa. The mass transfer also changes

according to the combination of working volume and aeration employed. Theoretically, the mass

transfer should be steady throughout a fermentation process. However, the properties of the

broth may change as the growth progresses and due to a reduction in the working volume after

the removal of samples. Such alterations in the system can be observed and monitored using the

dynamic gassing out method. Figure 61 shows the evolution of kLa and OTR throughout the

aforementioned batch and fed-batch fermentations. The obtained kLa fluctuated, showing an

increase at 24 h when the stationary phase starts. If the OTR is higher than the OUR, or OTR equal

Figure 60: Profiles of dissolved oxygen tension obtained in batch experiments performed under different oxygen supply conditions.


136

to OUR then the culture is not oxygen-limited. To this end, OTR is shown in the same plot (Figure

61). The theoretical OTR is using the value from the Henry equation (7.56 mg/L) as saturated

oxygen concentration, whereas the experimental OTR was calculated using the saturated value

from the intercept of the line with the Y axis used for the kLa determination. Interestingly, there

was a big difference between the two OTR profiles since the experimental saturated value was

closer to the dissolved oxygen concentration at each time. However, both OTR profiles were

higher than the OUR which suggested that the fermentation system was not oxygen-limited

under either batch or fed-batch cultivation mode.

Evolution of pH

Oleaginous yeasts have been reported to secrete organic acids into the medium, although

not much emphasis has been put on the nature of the acids nor even on the process implications

derived, except for the case of Y. lipolytica (André et al., 2009; Papanikolaou et al., 2002b).

Figure 61: Oxygen transfer parameters obtained during batch (A) and fed-batch 1 experiments (B).

(A)

(B)


137

Throughout the batch cultivations mentioned in this chapter, pH was progressively reduced over

time (Figure 62). In several studies Rh. glutinis was reported to acidify the broth by releasing

organic acids (Zhang et al., 2014a; Cheirsilp et al., 2011), which were identified as pyruvic, formic

and acetic acid (Xue et al., 2010). During the above bioreactor studies none of these acids were

detected apart from citric in non-significant amounts compared to the oil produced (Figure 63).

However, the main goal of the present cultivation experiments was to assess the effects of the

cultivation parameters on the cell growth and oil production rather than an evaluation of the

potential organic acids generated. The availability of carbon in the media and its uptake are

responsible for the metabolic generation and further secretion of citric acid into the media. Thus,

the yields of acid on carbon consumed (YCA/Gly) were almost similar (~0.11 g/g) indicating that

glycerol availability was the main factor for the synthesis of citric acid regardless of the oxygen

supply scheme employed. The specific CA productivity was calculated and showed that the

higher final DCW had the lowest citric acid production. Rakicka et al. (2015) noticed a trade-off

between citric acid and biomass formation of Y. lipolytica. In the case of fed-batch cultivations

some more citric acid was produced with increasing consumed glycerol, with 3.21 g/L of acid at

the end of the experiment FB1(YCA/Gly=0.026 g/g) and 7.51 g/L of acid (YCA/Gly=0.045 g/g) at the

end of the FB2 experiment (Figure 64).

Figure 62: Time-course profiles of pH found in the batch cultivations. In all cases pH reached a plateau value (pH~2.6) and then it remained constant over time.


138

Batch fermentation at two different aeration rates

DOT usually increased when the cells were in the stationary growth phase and according

to previous oxygen uptake data obtained in this research, the demand in oxygen was low towards

the end of the fermentation. Therefore, one experiment with two different aeration rates was

performed, in order to test if reduction of air flow rate would affect the biomass and oil

production. The starting conditions were the same as other batch fermentations, while the

aeration rate was changed from 0.5 to 0.2 L/min at 29 h of cultivation. Figure 65 shows the

dynamic profiles of DCW, glycerol and nitrogen as well as the oxygen transfer rate and uptake.

Figure 64: Time-course profile of citric acid production during fed-batch fermentations of Rh. glutinis.

Figure 63: (A) Profiles of citric acid (CA) production under different aeration rates in a 2 L batch bioreactor. (B) Specific citric acid productivity (qCA) obtained in batch fermentations under different aeration rates.

(A) (B)


139

As can be seen from Figure 65A, the change in aeration rate did not have a profound effect

on the biomass and substrates profiles. There were, however, some disruptions in the OTR

profiles, kLa and OUR (Figure 65B). In general kLa did not undergo major changes and was at

usual levels to other fermentations. Overall, the final values of DCW (4.86 g/L), oil content

(33.6%) and oil (1.63 g/L), indicated that changing the aeration rate to a lower value after the

cells have reached stationary phase did not affect significantly the final yields. However, even

though the oil content was similar, the biomass of 4.86 g/L was lower than that achieved with

constant 0.5 L/min and baffled bioreactor.

(A)

(B)

Figure 65: (A) Time-course profiles of Rh. glutinis grown on glycerol at two aeration rates. (B) Oxygen transfer parameters obtained during batch (A) and fed-batch 1 experiments (B).


140

Validation of the Optical Density-Dry Cell Weight correlation

Measurement of dry cell weight is the most direct technique for monitoring biomass

concentration (and hence growth) in oleaginous yeasts systems. However, it requires a

significant sample volume (e.g. 6 mL) and considerable time in order to obtain the results. Optical

density, on the other hand, is a simple and quick technique, though it is indirect and relies on the

uniformity of cells in the population. There are several ways of correlating these two

measurements, the one applied in the present thesis was based on sampling exponential phase

cells (24 h), measuring dry cell weight for part of the sample and measuring optical density for

serial dilutions of the remainder. Plotting measured OD against the DCW shows a linear

relationship. An example can be seen in Figure 66.

In order to test the validity of the OD-DCW correlation based on dilution as well as

examining the reproducibility of biomass measurements over a wide range of separate

experiments, the OD and DCW values obtained from batch and fed-batch cultivations were

plotted on the same graph (Figure 67). Although the aeration rates were not the same in all

fermentations, the relationship between OD and DCW indicated a clustered set of lines, while

absorbing the differences in biomass from one experiment to the other.

All the correlations show very good agreement for the slopes of the linear regression. By

calculating the mean value and standard deviation of the slopes in Figure 67, the SD between

them is less than 5%, while if the mean value (3.3166) is compared with the slope of the

calibration curve used throughout the present thesis, the SD is 16%.

Figure 66: Correlation of DCW and OD, based on serial dilutions of one sample.


141

Figure 67: Relationships between OD and DCW values obtained from samples during batch and fed-batch cultivations of Rh. glutinis. The ‘Main correlation’ dataset refers to the correlation curve shown in Figure 66.


142

143

Chapter 8

Developing an unstructured model to describe

batch cultivations of Rhodotorula glutinis


144

8 Developing an unstructured model to describe batch cultivations of Rhodotorula glutinis

8.1 Introduction

Modelling is a useful engineering tool for process design, because it can be used

predictively to reduce the experimental load. Although there is a plethora of experimental

findings on microbial oil production from different yeasts and substrates, the computational

approaches are limited. In the research presented in this thesis, an unstructured kinetic model

was developed to describe and predict the behaviour of the batch cultivation of Rh. glutinis. The

model takes into account glycerol inhibition and is based on a double substrate dependence of

the growth rate. The kinetic parameters of the model are estimated by fitting experimental data

presented in Chapter 6, minimising the difference between experimental and predicted values

and finally validation was performed by independent prediction of results from further

experiments. For this, flask experiments were used along with one bioreactor experiment from

Chapter 7. A review of existing models is presented in section 8.2, followed by a detailed

description of the model development in this work.


Oil accumulation is a process induced by nutrient limitation and growth interruption.

Only recently has there been a significant modelling effort on microbial oil kinetics using yeasts

(Meeuwse et al., 2011; Economou et al., 2011) but there is a lot of scope for further research,

since oleaginous systems are not straightforward to generalise. Initially, one study proposed a

model to predict microbial oil production cost using the C/N ratio and the oil yield and

productivity (Ykema et al., 1986). Their approach was based on a generalised stoichiometric

equation of the biochemical oil production reaction. Later works focused on the kinetics of the

process rather than the cost estimation. Economou et al., (2011) developed a batch model using

common Monod equations for microbial growth and applied it to the oleaginous fungus M.

isabellina on glucose. This approach assumed a well-mixed system, fungal growth dependence

on carbon and nitrogen, and carbon conversion to lipids and biomass. The model could predict

well batch fermentations with different C/N ratios.

Extensive study on different cultivation modes was performed by Meeuwse et al., (2011),

who modelled the cultivation of M. isabellina in submerged batch and continuous cultures using

glucose as carbon source. Their first model described different nutrient limitations including

Chapter 8 Developing a model for microbial oil production

145

carbon while their second work for the continuous mode investigated three different phases:

exponential growth, lipid accumulation and lipid turnover (Meeuwse et al., 2012). Even though

Economou et al.,. (2011) initially had utilised lipid degradation, they did not observe such

phenomenon taking place and removed this term from the model.

Very recently, another approach was developed by Beligon et al., (2016) with the yeast

C. curvatus. The authors developed a model which incorporates oxygen in the growth rate,

making it essentially a triple-substrate dependent model. After developing the model for fed-

batch culture, they used it predictively to optimise parameters for continuous cultivation and to

predict the growth of the oleaginous yeast on acetate.

The above mentioned studies are all good approaches to modelling of oil production.

However, none of them used glycerol as a carbon source and the assumptions applied are not

widely applicable to other systems due to the specificity of oil accumulation and the strain

dependent responses to different factors. In the present thesis, a set of kinetic parameters were

calculated to fit properly the state variables (glycerol, biomass and oil accumulation). These are

presented in Section 8.3 and further experimental results were used to validate the model

(Section 8.4).

8.3 Methodology

8.3.1 Model development

As reported previously, nitrogen is the limiting nutrient and its consumption under

carbon excess triggers lipid accumulation. Therefore, the growth of the yeast depends on both

carbon and nitrogen and both factors were taken into account for the development of the specific

growth rate expression, Equation (8-1). It was evident from Chapter 7 that high glycerol

concentrations inhibit growth since the specific growth rate was reduced at initial glycerol

concentrations beyond 40g/L. In order to accommodate this phenomenon effectively, the

Haldane equation (Andrews, 1968) was introduced to the carbon-dependent part of the growth

rate expression, Equation (8-2). The nitrogen screening experiments from Chapter 7, showed

clear Monod dependence of the growth rate on nitrogen, therefore the nitrogen contribution to

the growth rate did not include any inhibition term, as shown in Equation (8-3). A modified

Monod expression (Villadsen et al., 2011) was used to describe the specific growth rate. This

expression for the specific growth rate, Equation (8-4), is similar to that developed by Economou

et al., (2011), but they used the sum of the two rates, while here the rates are multiplied.


146

NGlyGlyN (8-1)

IGly

GlyGly

K

GlyKGly

Gly2max, (8-2)

NNN

KN

Nmax, (8-3)

Where μGlyN is the specific growth rate based on glycerol and nitrogen (h-1), μGly and μN are

the specific growth rates on glycerol and nitrogen respectively (h-1), μGly,max and μN,max are the

maximum specific growth rates on glycerol and nitrogen respectively, Gly is the glycerol

concentration (g/L), N is the nitrogen concentration (g/L), which is the limiting substrate, KGly

and KN are the half saturation constants (g/L) for glycerol and nitrogen respectively and KI is the

substrate inhibition constant for glycerol (g/L).

By substituting Equations (8-3) and (8-2) into Equation (8-1) the resulting expression

for the growth rate is:

N

IGly

GlyNGlyNKN

N

K

GlyKGly

Gly2max, (8-4)

Where µGlyN, max is the multiplication of the two individual maximum specific growth rates

(µGly,max · µN,max).

The rate of lipids accumulation can also be described by a Monod expression in relation

to the carbon source (Economou et al., 2011). Glycerol would not be inhibitory for lipid synthesis

because the oil yield increased with increasing amounts of glycerol. However, lipid accumulation

does not start if nitrogen is present in the medium and in most cases, it is not a growth-associated

process. To ensure that oil will be produced when a conditional term, (N0-N)/N0, was introduced

to the lipid specific formation rate, the specific lipid production rate becomes:


147

0

0

N

NN

KGly

Glyqq

LGlyLL

max, (8-5)

Where qL is the specific lipid synthesis rate (h-1), qL,max is the maximum specific lipid

synthesis rate (h-1) and N0 is the initial concentration of nitrogen (g/L).

Although this equation assisted in matching the experimental glycerol profile, the lipid

titre obtained negative values (data not shown) at the beginning of the fermentation when the

nitrogen concentration (N) was not very low. Then the term (N0-N)/N0 of Equation (8-5) was

replaced by the term k1/(k1+N) (Papanikolaou and Aggelis, 2011b) to satisfy the conditions for

lipid synthesis when nitrogen has been consumed or is very low. The new equation is:

Nk

k

KGly

Glyqq

LGly

LL

1

1max, (8-6)

Where qL,max is the maximum specific lipid production rate (h-1), KLGly is the half saturation

constant of lipids on glycerol (g/L) and k1 (g/L) is a parameter to ensure that while nitrogen is

available lipid accumulation is low. As with Economou et al., (2011), lipid degradation was not

observed, therefore such term was not included in the present model.

The kinetic model proposed here is based on the following assumptions:

The system is well-mixed.

Total cellular mass (biomass) consists of lipid-free biomass and intracellular oil.

Lipid-free biomass and oil are the only products derived from glycerol.

Lipid-free biomass is the only product derived from nitrogen.

The variables describing the biomass, glycerol, oil and nitrogen consumption were

constructed based on mass balances for carbon and nitrogen. The growth equation is described

by:

fGlyN Xdt

dX (8-7)

Where Xf (g/L) is the lipid-free biomass, t (h) is the time.

Lipid formation requires glycerol as input, which is taken into account within the qL as

shown in Equation (8-6). No lipid degradation has been taken into account in this equation.


148

fL Xqdt

dL (8-8)

Glycerol is converted to biomass and storage lipids so the expression for substrate consumption

is:

fGlyL

L

GlyX

GlyNX

Y

q

Ydt

dGly

f

//

(8-9)

Where YXf/Gly and YL/Gly are the yield of lipid-free biomass on glycerol (g/g) and yield of oil on

glycerol (g/g) respectively. Similarly, the nitrogen consumption is given by:

fNX

GlyNX

Ydt

dN

f /

(8-10)

Where YXf/N (g/g) is the yield of lipid-free biomass on nitrogen.

8.3.2 Parameter optimisation

Equations (8-7) to (8-10) constitute a system of four ordinary differential equations, with

four state variables (glycerol, biomass, oil and nitrogen) and ten kinetic parameters, which can

be summarised in the form of Equation (8-11). The parameter estimation was carried out

combining both stochastic and deterministic optimisation approaches to fit the kinetic model

with the experimental fermentation results. For the optimisation process, the parameter

determination used constraints (upper and lower limits) informed by literature values or

experience while an initial ‘guess’ of the state variables was made in order to calculate the

dynamic profile. A non-linear least squares method was used to minimise the objective function

that included the sums of the squared errors between the experimental and predicted values of

the state variables. In particular, a stochastic optimisation algorithm (Simulated Annealing) was

implemented in MATLAB R2014 to calculate a family of solutions around the global minimum.

This was then followed by a deterministic optimisation algorithm which used the ‘fmincon’

function in MATLAB to identify an exact optimum and estimate the final parameters. The system


149

of ODEs represented by Equation (8-11) was solved using the ‘ode45’ function in MATLAB, which

applies a 4th order Runge-Kutta integration method.

)),(()(

kktYfdt

tdY (8-11)

Where Y is a vector containing the five state variables (X-biomass, Gly-glycerol, L-oil, N-nitrogen)

and kk is a vector containing the parameters (μGly,Nmax, KGly, KN, KI, qLmax, KLGly, YX/Gly, YL/Gly, YX/N, k1).

8.4 Results and discussion

8.4.1 Specific growth rate estimation

For a fixed nitrogen concentration, the specific growth rate versus glycerol was described

by the Haldane equation, Equation (8-2). The parameters of this equation, were optimised

separately from the model in Excel, using the Solver to fit the experimental data by minimizing

the squared differences of the experimental and theoretical values. The standard deviation of the

fit was 0.002. Figure 68 depicts the experimental and the estimated specific growth rates on

glycerol. These results confirm that the Haldane equation can be utilised in relation to carbon in

the present model. The growth rate was affected by higher glycerol concentrations. A similar

trend to that of the experimental data for the specific growth rate dependence on glucose was

obtained by Economou et al., (2011).

Figure 68: Application of the Haldane expression to describe the values of the specific growth rate for Rh. glutinis on different initial glycerol concentrations, as reported in Chapter 7.


150

8.4.2 Fitting the flask experiments

The optimised parameters and the limits/constraints, used for this model, are included

in Table 16. These values are also compared with other literature values for the same parameters

in Table 16. However, the literature values are not based on a system with glycerol as the carbon

source. Apparently glucose is a more preferred substrate than glycerol, which could explain why

some parameters from the literature were higher than those of the present study. However,

others were very similar.

Table 16: Estimated model parameters for Equations (8-4) to (8-10) and comparison with literature results. The parameters are derived from optimisation following an initial guess, using upper and lower values as constraints for the optimisation.

Parameter Description Units Value Literature values

μGlyN,max Maximum specific growth rate h-1 0.2380 0.56a

qL Maximum specific rate of lipid production

h-1 0.05 0.06b

KGly Substrate half saturation constant

g/L 20 20c

KI Substrate inhibition constant g/L 45 20.98a

KN Half saturation constant on nitrogen

g/L 0.15 0.085a

KLGly Half saturation constant on glycerol for lipid formation

g/L 69.99 69.27a

YXf/Gly Yield of biomass on glycerol g/g 0.2 0.35a

YL/Gly Yield of oil on glycerol

g/g 0.1472 0.24a

YXf/N Yield of biomass on nitrogen g/g 38.6 3.93a

k1 Parameter for the onset of lipid accumulation

g/L 0.033 0.399a

a data from (Economou et al., 2011), b datum from (Aggelis and Sourdis, 1997) and c datum from (Fakas et al., 2009)

Figure 69 shows the model predicted time-course profiles for glycerol, cell concentration,

oil and nitrogen for a batch experiment with 30 g/L glycerol and 0.2 g/L nitrogen. To compare

easily the experimental and estimated data, each state variable was plotted separately. As can

be seen, the model fits reasonably well to the experimental data.


151

8.4.3 Model validation

The time-course profiles for the concentrations of the key variables for 40 and 80 g/L

initial glycerol are shown in Figures 70 and 71. The trend of the cellular growth was predicted

quite adequately for the first part of the growth curve (<100h), while in the case of 80 g/L the

exponential phase started earlier than expected. Cellular proliferation was taking place as long

as nitrogen was available in the broth. When nitrogen reached very low values, cell concentration

reached a plateau. Since the total nitrogen supplied was the same in all the experiments, the

differences in the growth rate and final biomass were also a result of the amount of carbon as

mentioned in previous chapters. The nitrogen consumption converged with the experimental

data after 50 h of cultivation, as was also seen in the experiment with 30 g/L initial glycerol.

The glycerol profile was estimated, based on mass balances for the glycerol conversion.

Carbon is theoretically converted to fat-free biomass, triacylglycerols (oil), carbon dioxide and

cell maintenance. Although carbon dioxide was not included in the model since it was not

measured experimentally, acids production and maintenance rate were initially included. Cell

maintenance was estimated to be very close to zero and because its absence did not affect the fit,

the maintenance term was omitted from the substrate equation in the model. Glycerol, was quite

Figure 69: Experimental (symbols) and estimated (lines) data for the batch cultivation of Rh. glutinis in 30 g/L initial glycerol concentration. Estimations were made using Equations (8-7) to (8-10). TN: total nitrogen.


152

adequately predicted until around 100 h (Figure 71), when the model suggested faster

consumption than was seen in the experimental data. The best fit was in the case of 80 g/L.

Nevertheless, the effect of inhibition is not strong at concentrations of that level compared to

glycerol above 100 g/L.

Regarding the lipid accumulation predictions, the parameter k1 introduced in Equation

(8-6), which ensured no lipid synthesis prior to nitrogen consumption, succeeded in describing

well the experimental lipid titre. The estimated specific lipid production rate (0.05 h-1) predicted

substantial lipid synthesis at 48-50 h and an increase later on, which was in agreement with the

experimental observations. Lipid synthesis was very low during the first hours of fermentation,

as reflected by the model estimation. Such a result would indicate that cell harvesting for lipid

measurement is meaningful when the cells are in stationary growth phase.

Figure 70: Experimental (symbols) and estimated (lines) data for the batch cultivation of Rh. glutinis in 40 g/L initial glycerol concentration. Estimations were made using Equations (8-7) to (8-10). TN: total nitrogen. In this experiment, no TN was analysed.


153

8.4.4 Testing a different expression for the specific lipid production rate

A minor modification of the conditional term for oil accumulation in Equation (8-6)

(k1/k1+N), was made to test how the nitrogen concentration affected the lipid formation rate and

consequently the fit to experimental data. This modification, given in Equation (8-12) takes into

account that nitrogen is not fully consumed but reaches low levels in relation to its initial

concentration. The corresponding fits are shown in Figures 72-74. As can be seen from the

figures, the general fit was similar to that of Equation (8-6), shown in Figures 70 and 71, but lipid

concentration was fitted better. This modification was only a test and was not used to alter the

model as presented in the previous sections. The kinetic parameters, which changed only slightly

are: µGlyN,max=0.2235 h-1, YL/Gly=0.0599 g/g, KN=0.153 g/L, KLGly=70 and k1=0.2598, while the rest

remained the same. The differences from Equation (8-6) were not significant, so to keep it

simple, Equation (8-6) was retained.

Figure 71: Experimental (symbols) and estimated (lines) data points for the batch cultivation of Rh. glutinis in 80 g/L initial glycerol concentration. Estimations were made using Equations (8-7) to (8-10). TN: total nitrogen.


154

0

1

1

N

Nk

k

KGly

Glyqq

LGlyLL

max, (8-12)

Figure 73: Experimental (symbols) and estimated (lines) data points for the batch cultivation of Rh. glutinis in 30 g/L initial glycerol concentration. Estimations were made using Equations (8-7) to (8-10) with implementation of Equation (8-12). TN: total nitrogen.

Figure 72: Experimental (symbols) and estimated (lines) data points for the batch cultivation of Rh. glutinis in 40 g/L initial glycerol concentration. Estimations were made using Equations (8-7) to (8-10) with implementation of Equation (8-12). TN: total nitrogen. In this experiment, no TN was analysed.


155

8.4.5 Predicting shake-flask and bioreactor performance

The model presented above was used to predict performance in further flask and

bioreactor experiments. For the validation of the shake-flask cultivations, dynamic data from the

study on the effect of nitrogen (Section 7.4.2) with 0.15, 0.25 and 0.4 g/ initial nitrogen were

used. Data from the bioreactor fermentation with 0.5 L/min aeration rate were used for the

model validation at bioreactor scale. However, data were not available for all variables. For

example, oil was extracted only in the final point for these flask experiments and only after 72 h

in the bioreactor and for that reason the total cellular mass is shown in Figures 74 and75.

As can be seen in Figure 75, the model predicted the behaviour of the flask cultivations

quite well without any modification or optimisation of parameters and the key variables follow

the same pattern (Figure 75). Glycerol consumption was predicted very well until 15 h of

fermentation, while a similar to the experimental data trend was observed for the lipid-free

biomass. The lipid-free biomass and the total biomass followed the same trend. As expected the

simulated values of the lipid-free biomass are lower than the experimental values of total cell

mass.

Figure 74: Experimental (symbols) and estimated (lines) data points for the batch cultivation of Rh. glutinis in 80 g/L initial glycerol concentration. Estimations were made using Equations (8-7) to (8-10) with implementation of Equation (8-12). TN: total nitrogen.


156

Regarding the bioreactor cultivation, the model failed to describe the improved growth

rate of the yeast, resulting in significant deviations in the first 24 h of the fermentation. However,

the bioreactor cultivations are a different case since the system was supplied with air. According

to Chapter 7, the air supply benefited the growth rate and final biomass and oil concentration.

(A)

(B)

(C)

Figure 75: Experimental (symbols, ‘Exp’) and simulated (lines, ‘Sim’) data points for the batch cultivation of Rh. glutinis on 30 g/L glycerol in shake-flasks. (A) 0.15 g/L TN, (B) 0.25 g/L TN, (C) 0.4 g/L TN. TN: total nitrogen, Xf: lipid-free biomass, X:total biomass.


157

Therefore, the inclusion of an oxygen term in the model would be desirable for these systems.

However, here a simpler modification was used. The optimised maximum specific growth rate,

obtained from flask experiments (0.2380 h-1) was replaced by the one reported in Chapter 7 for

the 0.5 L/min aerated bioreactor cultivation (0.6 h-1). No other changes were made to any of the

model parameters. Predictions using the model with this minor modification are shown in Figure

76.

8.5 Conclusions

In this chapter an unstructured kinetic model describing oil production from glycerol in

batch mode was developed. The model adequately described flask cultivations and was used to

predict other experimental conditions in flasks. Taking into account glycerol inhibition resulted

in better fit to glycerol consumption data. Similarly to other models, Monod kinetics was an

adequate assumption for the growth rate and product description. However, for bioreactor

fermentations where oxygenation leads to better growth, modification in the model parameters

and introduction of new terms into the equations would be desirable but would increase the

complexity of the model. Nevertheless, even without significant changes it worked well. In the

next chapter different feeding methods are investigated which could offer more experimental

data for further model development.

Figure 76: Simulation of a bioreactor fermentation of Rh. glutinis with 30 g/L initial glycerol and 0.2 g/L initial nitrogen. Sim: refers to the simulated results and Exp to the experimental data


158

159

Chapter 9

Evaluating feeding strategies for microbial oil

production from glycerol by Rhodotorula glutinis


160

9 Evaluating feeding strategies for microbial oil production from glycerol by Rhodotorula glutinis

9.1 Introduction

Following the successful fed-batch improvements reported in Chapter 7, different

glycerol feeding schemes (pulsed and continual feeding at rates equal to or higher than the

glycerol uptake rate) were evaluated under fed-batch cultivation of the oleaginous yeast

Rh. glutinis. The feeding strategy was an extension of the two-stage cultivation approach,

referred to in chapter 7. After a batch period of 24 h, a further growth stage took place (to

enhance cell production), followed by a lipid promoting stage with only glycerol supply (to

enhance oil production). These feeding strategies enabled the evaluation of the effect of feeding

rates on the glycerol uptake, cellular concentration and lipid productivity, in order to build up a

process for microbial oil production from glycerol. The results constitute part of a paper

published in the journal Engineering in Life Sciences (Karamerou et al., 2016b). Some

supplementary information has been added to fit with the context and format of the present

thesis. The published article can be found in APPENDIX 2.

9.2 Materials and methods

9.2.1 Fed-batch bioreactor experiments

Fed-batch cultivations were conducted in the 2-L bioreactor (Electrolab, UK) that was

used previously with the same starting conditions for all experiments: 30 g/L glycerol and 2 g/L

yeast extract, 1 mL/L antifoam A (Sigma –Aldrich) and pH 5.5. The incubation temperature was

30°C, the agitation rate was 400 rpm and the air flow rate was 0.5 L/min throughout the

cultivation. The used set up is depicted in Figures 77 and 78.

Feeding scheme

A two-stage culture scheme was applied in the fed-batch cultivations. After an initial

batch phase of 24 h, glycerol and yeast extract were fed to extend the ‘Growth’ stage from 24 to

96 h, while from 96 to 144 h the ‘Lipogenesis’ stage was promoted by feeding only glycerol. After

that, the culture was left without nutrient input until the final harvesting of biomass. The first

strategy (pulsed fed-batch, experiment PFB), consisted of three pulses of glycerol and yeast

extract, every 24 h to maintain the glycerol and nitrogen concentration in the broth above 30

Chapter 9 Feeding strategies for microbial oil production

161

g/L and 0.2 g/L, respectively. At 96 and 120 h two glycerol injections restored the glycerol

concentration to 30 g/L. The average glycerol uptake rate was calculated for the growth stage

0.8 g/L/h and for the oil production stage 1.14 g/L/h. Samples were withdrawn immediately

prior to the injection and within 5 min after the injection, during the PFB experiment. In the

second strategy (fed-batch with continuous feeding, experiment CFB1) each feeding, lasted for

24 h to supply by the end of the 24 h period the same amount of nutrients as the PFB did at the

beginning (Figure 79). Experiments PFB and CFB1 had the same stock media composition, which

is shown in Table 17. The glycerol supply in CFB1 was 0.83 g/L/h from 24 to 48 h, 1.04 g/L/h

from 48 to 96 h and 1.25 g/L/h from 96 to 144 h. In the third strategy (experiment CFB2) the

glycerol supply rate was twice as high as the average glycerol uptake rate of each stage in the

PFB cultivation. Therefore, a feeding rate of 2 mL/h (1.6 g/L/h glycerol) for the ‘Growth stage’

and another feeding rate of 2.85 mL/h (2.28 g/L/h glycerol) for the ‘Lipogenesis stage’ were

applied. The strategy for experiment CFB3 involved a constant medium feeding rate of 1.65

mL/min (1.32 g/L/h glycerol) from 24 h to 96 h, (glycerol and yeast extract) followed by glycerol

only, at the same rate, from 96 to 144 h. The stock solutions in the CFB2 and CFB3 schemes

contained 800 g/L glycerol in both growth and lipogenesis stages while the yeast extract was

41.5 g/L in CFB2 and 50.3 g/L in CFB3 to provide a constant, between all experiments, feeding

rate of 0.083 g/L/h during the growth stage. Antifoam was supplied when needed to prevent

foam formation. A LKB Perspex peristaltic pump was used to transfer the media into the

bioreactor. Information regarding feeding rates and stock media compositions are summarised

in Table 17.

Figure 77: Experimental set up for continuous fed-batch cultivations. Inputs consist of feeding bottle, antifoam bottle. The air pump and controller are on the right part of the photo, while on the left there is a stirrer for constant mixing of the feeding solution as well as a balance for measuring the feeding volume.


162

Figure 78: Schematic representation of the continuous fed-batch system shown in Figure 77.

Figure 79: (A) Glycerol feeding schemes presented as cumulative concentration during fed-batch cultivation of Rh. glutinis on glycerol-based media. (B) Total nitrogen (TN) feeding schemes presented as cumulative concentration during fed-batch cultivation of Rh. glutinis on glycerol-based media.


163

Table 17: Stock media composition and feeding rates employed in the different feeding schemes.

Feeding strategy

Feed stages Stock media composition Fed volume

(mL)

Glycerol feeding rate

(g/L/h)

Yeast extract feeding rate

(g/L/h) Glycerol N-source

PFB 24 h: Gly+YE 333 g/L 33 g/L YE 60 Pulse-fed Pulse-fed

48 h: Gly+YE 417 g/L 38 g/L YE 60 " "

72 h: Gly+YE 417 g/L 38 g/L YE 60 " "

96 h: Gly 500 g/L - 60 " "

120 h: Gly 500 g/L - 60 " "

CFB1 24-48 h: Gly+YE 333 g/L 33 g/L YE 60 0.83 0.083

48-72 h: Gly+YE 417 g/L 38 g/L YE 60 1.04 0.095

72-96 h: Gly+YE 417 g/L 38 g/L YE 60 1.04 0.095

96-120 h: Gly 500 g/L - 60 1.25 -

120-144 h: Gly 500 g/L - 60 1.25 -

CFB2 24-96 h: Gly+YE 800 g/L 41.5 g/L YE 144 1.6 0.083

96-144 h: Gly 800 g/L - 109 2.28 -

CFB3 24-96 h: Gly+YE 800 g/L 50.3 g/L YE 119 1.32 0.083

96-144 h: Gly 800 g/L - 80 1.32 -

Gly refers to glycerol, YE to yeast extract

-: No yeast extract supply

9.2.2 Oxygen uptake rate

The Oxygen Uptake Rate (OUR, mg/L/h) was calculated as shown in Chapter 7.


Glycerol, nitrogen and oil were analysed as described in Chapter 4. All assays were

carried out in triplicate and the results presented here are the average values. In all cases

SD<10%.

9.3 Results and Discussion

There are studies that highlight the importance of controlling the substrate feed in order

to make the microbial oil production process more efficient (Zhao et al., 2010a; Yen et al., 2015a).

By following a two-stage fed-batch approach, different glycerol feeding styles (pulse and

continuous) were evaluated in order to select an appropriate cultivation process for improved

lipid production.


164

9.3.1 Kinetic profiles of Rhodotorula glutinis using different feeding

methods

At first a pulsed fed-batch experiment (PFB) was carried out with three glycerol and yeast

extract feedings during the growth stage, followed by two glycerol-only pulses in the lipogenesis

stage. As can be seen from Figure 80A, the cells grew quickly between 24 and 96 h, then the

growth rate decreased when the lipogenesis stage began (phase III, Figure 80A). The final

cellular concentration was 23 g/L (IV, Figure 80A). The oil concentration increased from 8.25 at

128 h to 9.38 g/L to 168 h. Since the changes in cellular density were insignificant during phase

III and IV, the glycerol consumption was attributed to oil accumulation. All of the glycerol added

was eventually consumed within 24 h of each injection. Nevertheless, in a pulsed fed-batch

cultivation the substrate is supplied at once and the cells are left free to consume it at a rate

probably influenced by the driving force (local concentration) of each component of the medium.

In order to evaluate the effect of supplying the same amount of glycerol at a lower rate, a

fed-batch approach (CFB1) with continuous supply of the same stock medium as that used in

experiment PFB was performed. According to Figure 80B, the increase in cellular concentration

was not quite as sharp between 24 and 72 h. However, it reached a concentration of 19.6 g/L at

96 h and continued to increase at the beginning of lipid stage (III), reaching a final concentration

of 24.23 g/L (Figure 80B). The oil concentration increased from 7.42 at 128 h to 9.55 g/L at 186

h. The only difference was the longer time that cells in CFB1 took to reach the stationary phase.

As in PFB, all added glycerol was consumed, despite a slight accumulation from 24 h to 79 h

(Figure 80B), which was later eliminated during the lipogenesis stage.

Another experiment, CFB2, supplied glycerol continuously at rates twice as high as the

glycerol uptake rate of each stage (growth and lipogenesis). In this way the cells would have

available more glycerol than they apparently required. The cellular concentration increased

smoothly and more sharply than in CFB1, even during lipogenesis stage (III) to 30.63 g/L at

168 h accompanied by the accumulation of lipids (Figure 80C). The high glycerol supply rate

during the lipid accumulation stage led to a glycerol peak of 98 g/L at 144 h, a value that could

be inhibitory. This peak was a result of the accumulation of glycerol at a rate of 1.05 g/L/h,

confirming that the supply rate was surplus to that required. However, the fact that this peak

occurred during the lipid accumulation stage did not seem to have major detrimental impact on

the final yields but resulted in a residual glycerol concentration of 68 g/L at the end of

fermentation. Interestingly, the oil production was enhanced, reaching a concentration of

16.28 g/L by 168 h. In contrast to this study, in a continuous fed-batch cultivation of


165

C. freyschussii at a rate 3-fold higher than the glycerol uptake rate, the accumulation of glycerol

occurred earlier and no residual glycerol was detected in the broth (Raimondi et al., 2014).

The last experiment, CFB3, examined further the effect of supplying glycerol at a constant

rate throughout the cultivation, lower than CFB2 to avoid overfeeding of glycerol. The rate of

1.32 g/L/h glycerol was between the lower and upper rates applied in CFB1 and CFB2. The final

cellular concentration reached 27.8 g/L at 168 h (Figure 80D), lower than that achieved in CFB2

and higher than that of CFB1, accompanied by complete consumption of glycerol and production

of 11.38 g/L oil. Similarly to CFB1, glycerol accumulated only during the growth stage (0.45

g/L/h), while during the lipid stage it remained almost steady (Figure 80D).

Glycerol, or in general, substrate accumulation is a common phenomenon in continuous

fed-batch cultures, when supply is higher than the uptake rate. However, subsequent cellular

growth increases the uptake rate and eventually the substrate gets consumed.

The nitrogen (yeast extract) amount in all the continuous fed-batch experiments was

controlled in such a way as to provide no more than the pulsed fed-batch fermentation over the

same period of time (growth stage) and with the same supply rate of nitrogen. As can be seen in

Figure 81, nitrogen was immediately consumed after each injection. Similar levels of nitrogen

were observed in all the continuous fed-batch fermentations as driven by the supply rate,

confirming that nitrogen supply rate did not play a major role in the growth of Rh. glutinis but

the biomass yield was rather a result of the supply and local concentration of carbon. Other

studies on nitrogen have shown that the amount of nitrogen is linked to growth and increasing

amounts affect beneficially the cellular concentration. Batch cultivations of Rh. glutinis at

increasing C/N ratios, with increasing nitrogen concentrations favoured growth, delayed the

lipid accumulation phase but did not affect the glucose consumption profiles (Braunwald et al.,

2013). In another study of Rh. glutinis with constant medium supply, it was shown that sudden

nitrogen limitation induced the lipid synthesis earlier (Cescut et al., 2014).


166

Figure 80: Time-course profiles of cell growth (■), oil concentration (●), citric acid-CA (○), glycerol (▲), and oxygen uptake rate (◊) during different fed-batch fermentations of Rh. glutinis on glycerol. (A) pulsed fed-batch cultivation PFB, (B) continuously fed-batch cultivation CFB1 (0.83 and 1.04 g/L/h from 24 to 96 h and 1.25 g/L/h from 96 to 144 h), (C) continuous fed-batch experiment CFB2 (1.6 g/L/h from 24 to 96 h and 2.28 g/L/h from 96 to 144 h), (D) continuous fed-batch experiment CFB3 (1.32 g/L/h from 24 to 144 h).

The area I corresponds to the batch phase, area II to the extended growth stage, area III to the lipogenesis stage, and area IV to the last phase of the fermentation (harvesting stage). Values represent the average of triplicate assays with SD lower than 10%.

(A) (B)

(C) (D)


167

9.3.2 Effect of feeding style on the oxygen uptake rate

The oxygen uptake rate (OUR) over time for each fed-batch cultivation is shown in Figure

80. The sharp increase in OUR from 0 to 8 h coincided with the exponential growth phase and

was followed by a drop at 24 h in all cases. At that time the nitrogen had been consumed, the

growth rate ceased and the oxygen requirements and uptake became less. With the onset of the

fed-batch phase, the cells started consuming the new nutrients and proliferating. This led to an

increase in OUR, which remained at high levels during the growth stage and most of the

lipogenesis stage. In the case of PFB, the OUR was almost constant during the growth stage while

the continuous fed-batch cultivations showed a clear increasing trend with the rising cell weight.

OUR is linked to the metabolic activity as a result of growth which was a result of the available

nutrients. High DO was used by Meesters et al., (1996) as a sign of growth deceleration after

exhaustion of nutrients in order to feed glycerol during fed-batch cultivation of C. curvatus. In

contrast to the pulse fed-batch cultivation, the medium introduced gradually into the bioreactor

in all the CFB experiments and that is the reason for the slower increase in OUR. The OUR

decreased during the lipid stage while the DO was increasing, since the cells were not consuming

much oxygen. However, the decrease in qO2 (Figure 82) meant less oxygen consumption per cell

due to the rise in the number of cells present. The highest qO2 values after the onset of the fed-

batch phase was in PFB, while in the other experiments the qO2 did not rise significantly.

Figure 81: Residual total nitrogen (TN) concentration during the PFB (●), CFB1 (○), CFB2 (◊) and CFB3 (■) fed-batch fermentations of Rh. glutinis. Values represent the average of triplicate assays with SD lower than 10%.


168

Moreover, in all fermentations except CFB2, the peak in qO2 appeared before 32 h while despite

the differences in feeding method, all cases reached the same level of qO2 ~5-6 mg/g from 96 h,

showing that the DCW above 19 g/L is the limiting factor for qO2.

9.3.3 Effect of cumulative glycerol on growth and lipid production

Feeding the same stock medium under two different modes (pulsed and continuous

feeding) did not result in major differences in the final DCW and oil yields of PFB and CFB1 as

can be seen in Figure 83A. The final DCW in CFB1 was 24.23 g/L while that of PFB was 23 g/L,

indicating that growth simply reflected the overall stoichiometry. Similarly the oil content was

around 40% in both cases. Different glycerol feeding rates however, add different total amounts

of glycerol causing differences in effect (Figure 83B). Feeding glycerol at higher supply rates led

to an upward trend in DCW, oil content and oil concentration. Interestingly, the surplus of

glycerol during the lipogenesis stage in CFB2 appeared to be extra-beneficial for oil production,

giving a 53% w/w oil content, rather than about 46% w/w that would be expected from a simple

extrapolation of CFB1 based on total glycerol provided. A similar effect has been reported for a

sequential fed-batch fermentation using Y. lipolytica on crude glycerol (Rakicka et al., 2015).

Figure 82: Evolution of specific oxygen uptake rate (qO2) during the PFB (●), CFB1 (○), CFB2 (◊) and CFB3 (■) fed-batch cultivations of Rh. glutinis on glycerol-based media. Area I corresponds to the batch phase, area II to the extended growth stage, area III to the lipogenesis stage, and area IV to the harvesting stage. Values represent the average of triplicate assays with SD lower than 10%.


169

9.3.4 Influence of the glycerol feeding rate on biomass yield from

glycerol

To evaluate the effect of feeding rate on Rh. glutinis growth after the initiation of feeding,

the yield of biomass (YX/Gly, g/g) was calculated as the slope of the regression line of a plot of DCW

against consumed glycerol (Figure 84A-D). In Figure 84A-D, growth and lipogenesis are

distinguished by two different slopes. These slopes were less steep for the lipogenesis stage due

to the near absence of cell growth. As can be seen in Figure 84E, the biomass yield on glycerol

decreased with increasing glycerol feeding rate. A different trend was seen during the

lipogenesis stage (Figure 84F) where increasing feeding rate led to higher YX/Gly values, indicating

Figure 83: (A) Comparison of the final DCW and oil content for the experiments PFB (pulsed feeding

of glycerol, black bars) and CFB1 (continuous feeding of glycerol, grey bars). (B) Final DCW (♦), oil content (bars) and oil concentration (■) according to the total glycerol added in each continual fermentation. Values represent the average of triplicates with SD<10%.


170

that higher local glycerol concentration is preferred for oil production over cell proliferation.

This behaviour is consistent with that reported for a two-stage cultivation of Y. lipolytica on

glucose (Fontanille et al., 2012), where high and low yields were found for the growth and

lipogenesis stages, respectively.

Figure 84: Graphical estimation of the yield of biomass on glycerol (YX/Gly) for the growth stage (●) and the lipogenesis stage (○) during the fed-batch experiments (A) PFB, (B) CFB1, (C) CFB2, (D) CFB3. (E) Biomass yield on glycerol (YX/Gly) as a function of the glycerol feeding rate applied during the growth stage. (F) YX/Gly according to the glycerol feeding rate applied during the lipogenesis stage.


171

9.3.5 By-product formation and glycerol feeding rate

Citric acid was detected in the broth mainly during the lipogenesis stage (Figure 80).

Although citric acid is a potentially valuable by-product it diverts carbon away from oil

production. Lipid production along with citric acid secretion has been described extensively for

the yeast Y. lipolytica (Moeller et al., 2007; André et al., 2009; Papanikolaou et al., 2002b). In most

of the cases citric acid is produced under conditions of nitrogen-limitation and carbon excess,

during the last part of the fermentations (Makri et al., 2010). Regarding Rh. glutinis, acidification

of the broth has been reported but without further details of the acids or their relationship with

oil or carbon source (Zhang et al., 2014a; Cheirsilp et al., 2011; Kitcha and Cheirsilp, 2013). The

citric acid titres and yields are summarized in Table 18. About 11 g/L of citric acid were

produced in the PFB experiment, which is very close to the levels of oil production (9.4 g/L).

However, only about half as much (5.3 g/L) was produced in CFB2, where the glycerol was fed

continuously but slowly. This suggests that when large amounts of glycerol are available

instantly, there is more chance that it will be converted to both oil and citric acid while the

continuous supply channelled the carbon to oil production. In chemostat cultures using crude

glycerol with Y. lipolytica, citric acid and oil production decreased with increasing dilution rate

from 0.03 to 0.13 h-1 (Papanikolaou and Aggelis, 2002). Conversely, Rakicka et al., (2015)

observed a competitive relationship between lipid and citric acid production. Low values of

biomass and oil were attributed to the high value of citric acid concentration. This might explain

the low citrate production in our CFB2, where biomass and oil were higher. Concerning the CFB3

experiment, the citric acid titre increased after the lipid titre had reached a high but constant

value. Although the glycerol supply rate was higher than in CFB1, it was consumed faster in CFB3

possibly due to this high acid production. It has been reported that in some cases oil

concentration reaches a threshold value after which citric acid is mainly synthesized

(Dobrowolski et al., 2016). Moreover, increasing the glycerol feeding rate decreased the yield of

citric acid from glycerol (YCA/Gly). In terms of yield, Figure 85 shows a clear trend towards lower

production as feeding rate is increased. Similar trends are shown by the specific productivity,

where again citric acid per cell was reduced by increasing flow rate while the pulsed supply of

glycerol again led to more acid secretion. On the contrary, constant yields were observed studies

on Y. lipolytica at increasing crude glycerol concentrations (André et al., 2009) and Rywińska et

al., (2010) using pulsed and continuous fed-batch cultivation of the same yeast.


172

9.3.6 Comparison of fed-batch modes

Among the fed-batch strategies studied here, the continual feeding of glycerol was seen

to be the most beneficial mode, yielding the highest biomass and lipid production as confirmed

in Table 18. However, the glycerol to oil conversion yields (YL/Gly) were very similar (6-9%) with

a slightly increasing trend towards the higher glycerol supply rate, while the pulsed glycerol feed

gave the lowest YL/Gly. Similar yields can be seen in fermentations included in Table 19. In the

continuous feeding mode, the cells are supplied constantly with nutrients and there is less

likelihood of them running out. Other studies have stated similar benefits of continual feeding

compared to pulsed feeding. For example, Raimondi et al., (2014a) obtained their best results

with continuous fed-batch fermentation while cultivating the yeast C. freyschussi on pure

glycerol. Improved productivities were also obtained by Anschau et al. (2014) during continuous

cultivation of L. starkeyi on hemicellulose hydrolysate. The advantage of a continuous supply of

substrate was also confirmed by Zhao et al., (2010a) who achieved 30% increase in DCW and

33% in oil concentration when using a continuous supply of glucose to maintain the broth

concentration at 5 g/L throughout the cultivation of R. toruloides Y4. R. toruloides DSM 4444

obtained high DCW (62.4 g/L) with 61% oil content in a fed-batch cultivation with continuous

supply of glucose, targeting a constant concentration in the broth (Tsakona et al., 2016). These

values including the lipid productivity and glucose conversion yield were better than pulsed

Figure 85: Citric acid yield on glycerol (YCA/Gly, circles) and specific volumetric productivity of citric acid (diamonds) according to the glycerol feeding rate applied during the lipogenesis stage. Data presented herein are the average values of triplicates with SD<10%


173

supply of glucose. Yen et al., (2015a) compared pulsed, constant and exponential feeding of crude

glycerol on Rh. glutinis. In contrast to exponential feeding, pulsed and constant feedings were

more efficient for growth and lipid production.

As well as continuous feeding being better than pulsed feeding, (see Chapter 7) it was

shown that two-stage feeding was better than single-stage. The results reported here confirm

these findings (all values in Table 18 are higher than those reported previously). Table 19

summarizes the various cultivation strategies for oleaginous yeasts on glycerol. For the PFB

experiment, results for both biomass and oil were higher than those obtained by Kitcha et al.,

(2013): (13.8 g/L biomass and oil 7.78 g/L) using a similar two stage fed-batch strategy with

Rh. glutinis. Comparable oil content to the CFB1 and CFB3 experiments (~40%) but higher DCW

(113 g/L) resulting in substantial lipid titre were acquired with the ‘red’ yeast R. toruloides

cultivated on sugars extracted from Jerusalem artichoke, with intermittent feeding (Zhao et al.,

2010b). The results for CFB2 were similar to those obtained for a two-stage cultivation of C.

curvatus on crude glycerol (32.9 g/L cellular concentration and 52% oil content) (Liang et al.,

2010). An equivalent DCW of 30.5 g/L was reached by C. freyschussii cultivated on glycerol in

pulsed fed-batch mode but the lower oil content (30%) led to lower oil titre than in the present

study (Raimondi et al., 2014). Higher values were, however, were obtained when auxiliary

nutrients were added alongside glycerol. Two-stage fed-batch cultivation of R. toruloides on

glycerol (Yang et al., 2014) resulted in similar values of oil content (40.3%) and oil concentration

(8.1 g/L) to the PFB and CFB1 experiments of the present work, as can be seen in Table 19. In

another two-stage approach using Rh. glutinis on sugars, oil content (47.2%) was comparable to

the present work, though a higher cell concentration of 70.8 g/L led to a higher overall oil

production (Liu et al., 2015). Supplementation of lignocellulose derived sugars with crude

glycerol improved the yields of Rh. glutinis compared to those with only sugars in a batch study

(Yen et al., 2015b). Fontanille et al., (2012) utilized volatile fatty acids as carbon source following

a growth stage on glycerol using Y. lipolytica. They obtained similar results of 31 g/L biomass

and 12.4 g/L oil concentration (oil content 40%), supporting the suggestion that two-stage

cultivations are beneficial for oil production systems and indicating that during the second stage

no inhibition occurs, which broadens the range of carbon sources and concentrations that can

be used. Another aspect of continuous fed-batch cultivation was investigated by Cescut et al.,

(2014) who investigated the effect of sudden and progressive nitrogen limitation using

Rh. glutinis. The biomass and oil yields as well as the glucose uptake rate were higher in the case

of sudden nitrogen limitation.


174

Table 18: Experimental yields of Rh. glutinis CICC 31596 during the fed-batch cultivations performed in the present work. Values represent the average of triplicates with SD<10%.

Feeding approach

DCW (g/L)

YX/Gly growth stage

YX/Gly lipid stage

Oil (g/L)

Oil content (%, w/w)

YL/Gly (g/g)

Citric acid (g/L)

YCA/Gly (g/g)

PFB 23.00±0.03 0.225 0.043 9.38±0.17 40.8±0.01 0.059 10.96±0.14 0.22 CFB1 24.23±0.74 0.308 0.058 9.55±0.04 39.4±0.16 0.060 10.56±0.09 0.21 CFB2 30.63±1.44 0.229 0.091 16.28±0.23 53.0±0.00 0.087 5.46±0.04 0.10

CFB3 28.00±0.9 0.287 0.079 11.38±0.11 41.5±0.28 0.061 15.61±0.06 0.03

X: Biomass, CA: citric acid

Table 19: Comparison of cultivation modes employed for single oil production from glycerol by oleaginous yeasts.

Yeast strain Carbon source Cultivation mode Cultivation scale DCW (g/L)

Oil content (% w/w)

Oil titre (g/L)

YL/Gly (g/g) Qoil (g/L/h) Reference

C. curvatus ATCC 20509 Pure glycerol Fed-batch: two-stage Bioreactor 118 25 29.5 0.11 0.59 (Meesters et al., 1996)

Rh. glutinis CICC 31596 Pure glycerol Fed-batch: two-stage Bioreactor 16.8 34.6 5.07 0.03 0.03 (Karamerou et al., 2016a)

R.. toruloides Y4 Pure glycerol Batch Bioreactor 35.3 46.0 16.2 0.26 0.14 (Uçkun Kiran et al., 2013)

Y. lipolytica MUCL 28849 Pure glycerola Fed-batch: two stage Bioreactor 40.9 38.4 15.73 0.17 0.33 (Fontanille et al., 2012)

R. toruloides Y4 Pure glycerol Fed-batch: two-stage Shake-flask 21.1 40.3 8.5 0.22 0.07 (Yang et al., 2014)

Rh. glutinis TISTR 5159 Crude glycerol Fed-batch Bioreactor 10.5 60.7 6.1 0.06 0.12 (Saenge et al., 2011b)

R. toruloides Y-27012 Crude glycerol Batch Bioreactor 30.1 34 10.2 0.07 0.28 (Tchakouteu et al., 2015)

Y. lipolytica LGAM S(7)1 Crude glycerol Continuous Bioreactor 8.1 43 3.5 0.09 0.11 (Papanikolaou and Aggelis,

2002)

C. freyschussii ATCC 18737 Pure glycerol Fed-batch (glycerol pulses) Bioreactor 30.5 30 9.1 0.08 0.03 (Raimondi et al., 2014)

C. freyschussii ATCC 18737 Pure glycerol Fed-batch (continuous feeding)c Bioreactor 82 34 28 0.07 1.8 (Raimondi et al., 2014)

Y. lipolytica JMY 4086 Crude glycerolb Fed-batch: two stage (continuous feeding)c

Bioreactor 49.1 46 22.6 0.08 0.31 (Rakicka et al., 2015)

Rh. glutinis BCRC 21418 Crude glycerol Fed-batch (continuous glycerol supply)

Bioreactor 44.8 62.1 27.82 - 0.45 (Yen et al., 2015a)

Rh. glutinis BCRC 21418 Crude glycerol Fed-batch (exponential glycerol supply)

Bioreactor 39.2 43.3 16.97 - 0.23 (Yen et al., 2015a)

Rh. glutinis CICC 31596 Pure glycerol Fed-batch: two stage (pulsed) Bioreactor 23 40.8 9.38 0.059 0.06 This study

Rh. glutinis CICC 31596 Pure glycerol Fed-batch: two stage (continuous feeding)d

Bioreactor 24.23 39.4 9.55 0.06 0.06 This study


Bioreactor 28 41.5 11.38 0.061 0.07 This study


Bioreactor 30.63 53 16.28 0.087 0.10 This study

a Pure glycerol was used only in the growth stage, b crude glycerol was fed only during the lipid stage, c glycerol and nutrients were fed throughout the fermentation, d glycerol and yeast extract in the first stage, glycerol only feeding in the second stage


176

9.4 Concluding remarks

This study used a two-stage fed-batch cultivation concept in order to examine the effect

of the glycerol feeding rate on cellular growth and lipid production. Results showed that fed-

batch cultivation with continuous feeding of glycerol is more efficient for both biomass and oil

production than cultivation with pulsed feeding. Continuous feeding kept the cellular

metabolism active, leading to high biomass and oil yields. However, provision of the same

amount of nutrients in different ways (pulsed or continuous supply) did not significantly affect

the final concentrations of biomass and lipids (and also citric acid) as these are defined more by

the stoichiometry than the mode of operation. In addition, increasing the supply rate of glycerol

had beneficial effects on the biomass production. Continuous glycerol supply at high rates

resulted in enhanced cell densities and oil content, leading to higher overall productivities.

Moreover, it was demonstrated that high glycerol levels were not inhibitory during the

lipogenesis stage, resulting in less citric acid formation by channelling the available surplus of

carbon source into oil production. In conclusion, a continuous feeding strategy with different

nutrient supply rates for each stage was an efficient cultivation mode for enhanced microbial oil

production while reducing the by-product formation.


177

Supplementary information for Chapter 9

Glycerol consumption during fed-batch fermentation

Figure 86 represents the cumulative amounts of glycerol introduced into the medium

upon the initiation of the cultivations reported above, in the main part of Chapter 9. From these

data along with the dynamic residual glycerol concentration in the broth, the efficiency of

glycerol consumption can be characterised though the comparative graphs of fed and residual

glycerol. As the figure suggests, the consumption was high in all cases, including that of CFB2,

where the glycerol feeding rate was much higher than the consumption rate. The different stages

can be seen in the figure and reflect the effect of feeding rate on the residual concentration. The

experiment CFB2 had the least citric acid production. Similarly the comparative graphs for total

nitrogen (TN) are following (Figure 86).

Figure 86: Comparative glycerol graphs for the fed-batch cultivations of Rh. glutinis reported in Chapter 9.


178

Oxygen concentration as an indicator of metabolic activity

Oxygen consumption is related to metabolic activity. When the cells do not require much

oxygen, it is not taken up at a high rate and the dissolved oxygen concentration remains high. At

the end of fermentation any oxygen consumption could be related to cell maintenance. The batch

cultivations, regardless of the aeration rate had the same profile of dissolved oxygen. After a

sharp drop during the exponential growth phase, the DOT increased slowly to values close to

saturation by the end of the cultivation. Figure 89, shows the DOT profiles during the fed-batch

cultivations reported in Chapter 9. In contrast to the profiles of batch cultivations shown

previously, the fed-batch cultivations had lower concentrations of dissolved oxygen for longer

periods of time, reached the initial level only during the last 144 h and in general followed a

totally different trend.

Figure 87: Comparative total nitrogen (TN) graphs for the fed-batch cultivations of Rh. glutinis reported in Chapter 9.


179

Microscopic observations during the PFB, CFB1, CFB2 and CFB3

experiments

Micrographs were taken regularly during the above fermentations to monitor the cells

status and the progress of oil accumulation (Figures 90-93). In all cases unicellular form was

observed and the yeast did not suffer from the acidic pH.

Figure 89: Dissolved oxygen tension for fed-batch experiments from Chapter 9.


180

Experiment PFB

Experiment CFB1

Figure 90: Photomicrograps of Rh. glutinis during the experiment PFB. A) t=0 h, pH=4.38, B) t=28 h pH=2.85, C) t=52 h pH=2.65, D) t=72 h pH=2.93, E) t=96 h pH=2.37, F) t=99.5 h pH=2.63, G) t=120 h pH=2.6, H) t=144 h pH=2.41.

Figure 91: Photomicrographs of Rh. glutinis during the experiment CFB1. A) t=24 h, pH=2.71, B) t=46.5 h pH=2.71, C) t=79.5 h pH=2.95, D) t=99 h pH=2.65, E) t=128 h pH=2.6.


181

Experiment CFB2

Experiment CFB3

Figure 92: Photomicrographs of Rh. glutinis during the experiment CFB2. A) t=24 h, pH=2.72, B) t=48 h pH=2.74, C) t=56.4 h pH=2.67, D) t=101.5 h pH=2.54, E) t=128 h pH=2.48, F) t=168 h pH=2.4.

Figure 93: Photomicrographs of Rh. glutinis during the experiment CFB3. A) t=24 h, pH=2.6, B) t=32 h pH=2.65, C) t=72 h pH=2.6, D) t=96 h pH=2.79, E) t=128 h pH=2.76, F) t=99.5 h pH=2.3


182

Reproducibility of results

The data used for the calculation of standard deviation are shown in Figure 90. For the

calculation of the SD data from the batch phase, four independent experiments were used. As can

be seen from the graphs there is good agreement between all plots. In all cases SD<10%. The

formula utilised for the calculation of the standard deviation (unbiased estimation) was:

1

)(

1

2

N

xx

SD

N

i

i

(S-1)

Where N is the sample size, xi is the value and x is the mean value.

Figure 94: Data utilised for calculating the standard deviation of experimental data presented in Chapter 9.

183

Chapter 10

Conclusions and recommendations


184

10 Conclusions and recommendations

10.1 Introduction

As discussed in the introduction to this thesis, biofuels are emerging as the bio-based

alternative to fossil fuels. Biodiesel, is currently dominating the European biofuel market but its

expansion is impeded by the high production costs which are mainly influenced by the price of

raw materials. Oleaginous microorganisms, are one of the non-edible oil sources which

constitute an attractive choice thanks to their versatility in terms of assimilable substrates, as

discussed in Chapter 2. Consequently, their cultivation can be coupled with processing of

biodiesel waste and by-products processing.

Microbial oil, as biodiesel feedstock, has not yet been adopted commercially. Currently,

research is focusing on optimising the process and defining new routes for reducing cost and

increasing productivity. As highlighted in Chapter 2, biochemical engineering tools can play a

major role in designing and optimising bioprocesses. The utilisation of a low-cost carbon source

such as glycerol is not enough, on its own, to sufficiently improve the economics of microbial oil

production. But with proper operational design, the mid-stream stage (fermentation) can

achieve high yields at maximum conversion.

The present thesis contributes to the field of bioprocessing by presenting a fermentation

process for cultivation of the oleaginous yeast Rh. glutinis on glycerol. Novel feeding modes were

explored and evaluated in order to provide better understanding of the relationship between

glycerol supply and biomass/oil production. A predictive kinetic model was also introduced. The

value of this work lies in the very simple medium utilised, while achieving comparable or better

yields and the insights into the use of oxygen for the yeast growth, an under-characterised

fermentation parameter in other studies. The main findings of the research presented in this

thesis are discussed in the following section, followed by ideas on how the research can be

continued and advanced.

10.2 Discussion and concluding remarks

The research began with a selection study on the performance and glycerol consumption

of seven oleaginous yeasts (Chapter 5). The selection was based on a preliminary screening of

both glucose and glycerol consumption which resulted in four strains being selected for the

second screening stage. These were cultivated on different initial glycerol concentrations under

nitrogen limitation. This showed that pure glycerol can be inhibitory at high concentrations. The

best oleaginous yeasts from this study were Rh. glutinis and a Candida sp. Although the highest

Chapter 10 Conclusions and recommendations

185

biomass and oil contents were obtained for the Candida sp., the strain was not fully identified

and, while its potential for high productivity was noted, it was not selected for further study. On

the contrary, Rh. glutinis was a well-documented species and was also easy to handle, compared

to the Candida sp. and so was chosen as the focus for the remainder of the research. The fact that

Rh. glutinis was able to grow on all the glycerol concentrations tested suggested that no

adaptation was necessary and a proper fermentation design for improved glycerol uptake could

be investigated.

Growth aspects of Rh. glutinis were investigated in Chapter 6. The optimum initial pH was

found to be 5.5, which is in agreement with literature for this yeast, while yeast extract was

shown to be a suitable nitrogen and nutrients source. This allowed fermentation media

consisting only of glycerol and yeast extract to be used, which was quite simple compared to

those used in other research. Using these conditions, the dynamic oil accumulation profile was

determined and it was shown that cell harvesting for oil extraction could take place from 72 h.

Rh. glutinis was also able to grow on glycerol containing methanol at concentrations up to 60

g/L, with growth being only slightly affected beyond 20 g/L of methanol. This indicates that

crude glycerol, a biodiesel by-product, could potentially be used directly.

The effect of initial glycerol and nitrogen concentrations on the growth and lipid

production was examined in flasks under nitrogen limitation in order to develop starting

conditions for bioreactor cultivations. Glycerol concentrations higher than 40 g/L inhibited

growth, though oil accumulation was seen to be proportional to the amount of glycerol. Nitrogen

was not growth-inhibitory and Monod kinetics described well the growth rate. With these

conditions set, bioreactor cultivations were performed at different aeration rates and the oxygen

effect on growth and oil production were evaluated. An aeration rate of 0.5 vvm was best for

growth (5.38 g/L) and oil content (33%) and led to the highest specific glycerol and nitrogen

consumption rates. The provision of air improved the growth rate and glycerol consumption

compared to the flask cultivations.

The chosen aeration rate was used as a fixed parameter for building an effective

cultivation strategy. The inhibitory nature of glycerol at high concentrations favoured fed-batch

operation and so various feeding strategies were explored. A two-stage fed-batch was found to

be much better than a single stage in terms of biomass production (16.8 compared to 9.4 g/L)

and oil content (34.6 compared to 27.5%).

A model taking into account double-substrate dependence of the growth rate (on carbon

and nitrogen) coupled with glycerol inhibition at high concentrations was constructed and was

shown to be suitable for describing and predicting the behaviour of the system. This worked


186

particularly well for shake-flask studies but for the bioreactor cultivations the model tended to

underestimate the growth rate. Further model development is needed to improve bioreactor

performance predictions and to include by-product formation and oxygen consumption.

In two-stage feeding trials involving different glycerol feeding rates, results suggested

that each stage has different substrate requirements. Pulsed or continuous supply of the same

nutrient amount by the end of a certain time period did not affect the final concentrations.

Increasing the glycerol feeding rate enhanced cell and oil production. The glycerol surplus during

the lipid accumulating stage did not have negative effect on the lipid synthesis but boosted the

oil content. This clearly indicated that when a certain level of growth has been achieved,

overfeeding of carbon is not inhibitory. Moreover, secretion of by-product (citric acid) was

limited at the highest glycerol feeding rate, where oil production was maximum.

Such a feeding scheme has not been reported so far and the improvement in biomass and

oil production was based only on operational parameters and cellular uptake rates on a very

simple cultivation medium without pH control. The yields achieved are higher than others

reported with a richer medium (Table 20). In cases where the oil content is higher than that

achieved in the present thesis, the biomass is lower than the 30.63 g/L, obtained in this research.

This work demonstrates that with little input but good bioprocessing design, significant

improvements can be achieved in overall product yields. This, potentially, paves the way for

improved bioprocesses, helping to drive the growth of integrated biorefineries.

Table 20: Oil contents in relation to the nutrient medium for several oleaginous yeasts cultivated with glycerol as the carbon source.

Yeast Medium Oil content (%)

Oil titre (g/L)

DCW (g/L)

Reference

Rh. glutinis Pure glycerol, ammonium sulphate

60.7 6.1 10.5 (Saenge et al., 2011b)

Rh. glutinis Pure glycerol, yeast extract 53 16.28 30.63 This work

R. toruloides Pure glycerol, yeast extract, salts

46 16.2 35.3 (Uçkun Kiran et al., 2013)

C. curvatus Crude glycerol, ammonium chloride, salts

44.6 13.9 31.2 (Liang et al., 2010)

Y. lipolytica Crude glycerol, yeast extract, salts

43 3.5 8.1 (Papanikolaou and Aggelis, 2002)

C. freyschussii Pure glycerol, yeast extract salts

30 9.1 30.5 (Raimondi et al., 2014)


187

10.3 Recommendations for future work

Producing microbial oil within the concept of an integrated biorefinery has potential to

be produced in large scale and diminish dependence on fossil sources. The work reported in the

present thesis followed this idea and results confirmed the potential for microbial oil production

to be coupled with biodiesel production through the utilisation of glycerol. However, there is

scope for further improvement by addressing some critical aspects.

Cultivation modes

Continuous feeding of glycerol was shown to be better for both growth and oil

production. To build on this, exponential feeding should also be investigated. As shown in

Chapter 8, glycerol supply at a rate twice as high as the calculated one was beneficial for oil

production. The cellular uptake rate changes throughout the cultivation due to the continued

production of cells. Very recently, Yen et al., (2015) reported exponential feeding of crude

glycerol during fed-batch culture of Rh. glutinis. Exponential feeding needs complex design and

carefully designed mathematical relationships for its successful performance. In contrast to that

study, it is possible that using the two-stage approach developed in this research (supply of both

glycerol and nitrogen, followed by glycerol-only feeding) would be advantageous since the

nutrient supply would follow the population growth and is thus worth investigating further.

In addition to investigating alternative feeding strategies, continuous operation should

also be explored. The lipids are intracellular products, hence maintaining a constant number of

cells could improve overall performance and the yield. This can be achieved through continuous

fermentation, along with cell recycling. In this way, acids and other extracellular metabolites are

removed while the cells are left to consume the nutrients and accumulate oil while the younger

yeast cells can still proliferate.

In both of the above operational strategies, more cells should mean more oil. Therefore,

further research to optimise cell production, for example through the use of oxygen enriched

aeration, should be investigated.

Integration with biorefinery

Preliminary studies showed that Rh. glutinis could grow in synthetic crude glycerol.

Therefore, using crude glycerol as carbon source would make the process more profitable.

Furthermore, solid by-products of biodiesel such as rapeseed meal could serve as nitrogen and


188

nutrient source. Rapeseed meal has been successfully used in the author’s research group after

fungal pre-treatment for the growth of another oleaginous yeast, R. toruloides.

Scale-up of the process

With the help of techno-economic evaluation, starting from the biodiesel feedstock (if

crops are used) up to the end material and possible recycle of the product oil, the process

economics could be significantly improved. The applicability to large scale can be assessed in

this way and moreover, areas for further cost reduction could be identified. It is, in any case

worth performing cultivations in larger scale vessels. Shifting the fermentation from bench-top

bioreactor to pilot scale vessel, step by step, will provide insights into the effect of increasing the

working volume and adjusting the cultivation parameters. For this, principles such as the

volumetric mass transfer co-efficient or stirrer speed and dimensions to allow proper scale-up

and prediction of the bioreactor performance, should be studied.

Metabolic engineering

Metabolic flux analysis can be used to identify metabolic reactions taking place and

indicate which pathways should be supressed or promoted to enhance glycerol conversion to

microbial lipids. In order to maximise yields by blocking pathways, which would otherwise lead

the carbon flux to by-products (e.g. citric acid) metabolic analysis by computational and

experimental means should be carried out.

Further investigation of the Candida sp.

In the beginning of this work seven oleaginous yeasts were screened regarding their

growth and oil accumulation on glycerol. One of them Candida sp. showed a remarkable oil

content of 60% when cultivated on glycerol but was not selected for further research due to its

mycelial form. Moreover, it was not fully characterised. Using the findings of this research and

based on the behaviour of the Candida sp. on glycerol, cultivation of this strain can be assessed

at bioreactor scale as part of the future work. Clearly it would also be necessary to identify fully

the strain in order to make direct comparisons with existing literature.


189

Summary

In conclusion, the research reported in this thesis sets the basis for constructing a

microbial oil production process by investigating operational aspects of the yeast cultivation.

Further improvements can be implemented to ensure the applicability of the process and make

it more economically attractive.


190

191

References


192

References

Adewale, P., Dumont, M.-J. & Ngadi, M. (2015). Recent trends of biodiesel production from animal fat wastes and associated production techniques. Renewable and Sustainable Energy Reviews, 45, 574-588.

Ageitos, J. M., Vallejo, J. A., Veiga-Crespo, P. & Villa, T. G. (2011). Oily yeasts as oleaginous cell factories. Applied Microbiology and Biotechnology, 90(4), 1219-1227.

Aggelis, G. & Sourdis, J. (1997). Prediction of lipid accumulation-degradation in oleaginous micro-organisms growing on vegetable oils. Antonie van Leeuwenhoek, 72(2), 159-165.

Almeida, J. R. M., Fávaro, L. C. L. & Quirino, B. F. (2012). Biodiesel biorefinery: Opportunities and challenges for microbial production of fuels and chemicals from glycerol waste. Biotechnology for Biofuels, 5.

Alonso, S., Rendueles, M. & Díaz, M. (2012). Role of dissolved oxygen availability on lactobionic acid production from whey by Pseudomonas taetrolens. Bioresource Technology, 109, 140-147.

Alvarez, H. & Steinbüchel, A. (2002). Triacylglycerols in prokaryotic microorganisms. Applied Microbiology and Biotechnology, 60(4), 367-376.

Amaretti, A., Raimondi, S., Leonardi, A. & Rossi, M. (2012). Candida freyschussii: An oleaginous yeast producing lipids from glycerol. Chemical Engineering Transactions, 27, 139-144.

Andlid, T., Larsson, C., Liljenberg, C., Marison, I. & Gustafsson, L. (1995). Enthalpy content as a function of lipid accumulation in Rhodotorula glutinis. Applied Microbiology and Biotechnology, 42(6), 818-825.

Andre, A., Diamantopoulou, P., Philippoussis, A., Sarris, D., Komaitis, M. & Papanikolaou, S. (2010). Biotechnological conversions of bio-diesel derived waste glycerol into added-value compounds by higher fungi: production of biomass, single cell oil and oxalic acid. Industrial Crops and Products, 31(2), 407-416.

Andrews, J. F. (1968). A mathematical model for the continuous culture of microorganisms utilizing inhibitory substrates. Biotechnology and Bioengineering, 10(6), 707-723.

André, A., Chatzifragkou, A., Diamantopoulou, P., Sarris, D., Philippoussis, A., Galiotou-Panayotou, M., Komaitis, M. & Papanikolaou, S. (2009). Biotechnological conversions of bio-dieselderived crude glycerol by Yarrowia lipolytica strains. Engineering in Life Sciences, 9(6), 468-478.

Anschau, A., Xavier, M. C. A., Hernalsteens, S. & Franco, T. T. (2014). Effect of feeding strategies on lipid production by Lipomyces starkeyi. Bioresource Technology, 157(0), 214-222.

Ayadi, I., Kamoun, O., Trigui-Lahiani, H., Hdiji, A., Gargouri, A., Belghith, H. & Guerfali, M. (2016). Single cell oil production from a newly isolated Candida viswanathii Y-E4 and agro-industrial by-products valorization. Journal of Industrial Microbiology and Biotechnology, 1-14.

Ayoub, M. & Abdullah, A. Z. (2012). Critical review on the current scenario and significance of crude glycerol resulting from biodiesel industry towards more sustainable renewable energy industry. Renewable and Sustainable Energy Reviews, 16(5), 2671-2686.

Azocar, L., Ciudad, G., Heipieper, H. J. & Navia, R. (2010). Biotechnological processes for biodiesel production using alternative oils. Applied Microbiology and Biotechnology, 88(3), 621-636.

Bellou, S., Baeshen, M. N., Elazzazy, A. M., Aggeli, D., Sayegh, F. & Aggelis, G. (2014). Microalgal lipids biochemistry and biotechnological perspectives. Biotechnology Advances, 32(8), 1476-1493.

Beopoulos, A., Cescut, J., Haddouche, R., Uribelarrea, J.-L., Molina-Jouve, C. & Nicaud, J.-M. (2009). Yarrowia lipolytica as a model for bio-oil production. Progress in Lipid Research, 48(6), 375-387.

Beopoulos, A. & Nicaud, J.-M. (2012). Yeast: A new oil producer? Oilseeds &fats Crops and Lipids, 19(1), 22-28.

Bisen, P. S., Sharma, A. (2013). Introduction to instrumentation in life sciences. USA, CRC Press Taylor and Francis group.

Botham, P. A. & Ratledge, C. (1979). A biochemical explanation for lipid accumulation in Candida 107 and other oleaginous micro-organisms. Journal of General Microbiology, 114(2), 361-75.

Bozbas, K. (2008). Biodiesel as an alternative motor fuel: Production and policies in the European Union. Renewable and Sustainable Energy Reviews, 12(2), 542-552.

References

193

Braunwald, T., French, W. T., Claupein, W. & Graeff-Hönninger, S. (2014). Economic Assessment of microbial Biodiesel Production Using Heterotrophic Yeasts. International Journal of Green Energy, 13, 274-282.

Braunwald, T., Schwemmlein, L., Graeff-Hönninger, S., French, W., Hernandez, R., Holmes, W. & Claupein, W. (2013). Effect of different C/N ratios on carotenoid and lipid production by Rhodotorula glutinis. Applied Microbiology and Biotechnology, 97(14), 6581-6588.

Béligon, V., Poughon, L., Christophe, G., Lebert, A., Larroche, C. & Fontanille, P. (2016). Validation of a predictive model for fed-batch and continuous lipids production processes from acetic acid using the oleaginous yeast Cryptococcus curvatus. Biochemical Engineering Journal, 111, 117-128.

Calvey, C. H., Su, Y.-K., Willis, L. B., McGee, M. & Jeffries, T. W. (2016). Nitrogen limitation, oxygen limitation, and lipid accumulation in Lipomyces starkeyi. Bioresource Technology, 200, 780-788.

Carvalho, A. K. F., Rivaldi, J. D., Barbosa, J. C. & de Castro, H. F. (2015). Biosynthesis, characterization and enzymatic transesterification of single cell oil of Mucor circinelloides – A sustainable pathway for biofuel production. Bioresource Technology, 181, 47-53.

Cerón-García, M. C., Macías-Sánchez, M. D., Sánchez-Mirón, A., García-Camacho, F. & Molina-Grima, E. (2013). A process for biodiesel production involving the heterotrophic fermentation of Chlorella protothecoides with glycerol as the carbon source. Applied Energy, 103, 341-349.

Cescut, J., Fillaudeau, L., Molina-Jouve, C. & Uribelarrea, J.L. (2014). Carbon accumulation in Rhodotorula glutinis induced by nitrogen limitation. Biotechnology for Biofuels, 7(1), 1-11.

Chatzifragkou, A., Fakas, S., Galiotou-Panayotou, M., Komaitis, M., Aggelis, G. & Papanikolaou, S. (2010). Commercial sugars as substrates for lipid accumulation in Cunninghamella echinulata and Mortierella isabellina fungi. European Journal of Lipid Science and Technology, 112(9), 1048-1057.

Chatzifragkou, A., Makri, A., Belka, A., Bellou, S., Mavrou, M., Mastoridou, M., Mystrioti, P., Onjaro, G., Aggelis, G. & Papanikolaou, S. (2011). Biotechnological conversions of biodiesel derived waste glycerol by yeast and fungal species. Energy, 36(2), 1097-1108.

Cheirsilp, B. & Kitcha, S. (2015). Solid state fermentation by cellulolytic oleaginous fungi for direct conversion of lignocellulosic biomass into lipids: Fed-batch and repeated-batch fermentations. Industrial Crops and Products, 66, 73-80.

Cheirsilp, B., Suwannarat, W. & Niyomdecha, R. (2011). Mixed culture of oleaginous yeast Rhodotorula glutinis and microalga Chlorella vulgaris for lipid production from industrial wastes and its use as biodiesel feedstock. New Biotechnology, 28(4), 362-368.

Chen, C.-Y., Yeh, K.-L., Aisyah, R., Lee, D.-J. & Chang, J. S. (2011). Cultivation, photobioreactor design and harvesting of microalgae for biodiesel production: A critical review. Bioresource Technology, 102(1), 71-81.

Chen, X. F., Huang, C., Xiong, L., Chen, X. D. & Ma, L. L. (2012). Microbial oil production from corncob acid hydrolysate by Trichosporon cutaneum. Biotechnology Letters, 34(6), 1025-8.

Chi, Z., Zheng, Y., Jiang, A. & Chen, S. (2011). Lipid production by culturing oleaginous yeast and algae with food waste and municipal wastewater in an integrated process. Applied Biochemistry and Biotechnology, 165(2), 442-453.

Chisti, Y. (2007). Biodiesel from microalgae. Biotechnology Advances, 25(3), 294-306. Choi, S. Y., Ryu, D. D. Y. & Rhee, J. S. (1982). Production of microbial lipid: Effects of growth rate and

oxygen on lipid synthesis and fatty acid composition of Rhodotorula gracilis. Biotechnology and Bioengineering, 24(5), 1165-1172.

Christophe, G., Kumar, V., Nouaille, R., Gaudet, G., Fontanille, P., Pandey, A., Soccol, C. R. & Larroche, C. (2012). Recent developments in microbial oils production: a possible alternative to vegetable oils for biodiesel without competition with human food? Brazilian Archives of Biology and Technology, 55, 29-46.

da Silva, G. P., Mack, M. & Contiero, J. (2009). Glycerol: A promising and abundant carbon source for industrial microbiology. Biotechnology Advances, 27(1), 30-39.


194

Dai, C. C., Tao, J., Xie, F., Dai, Y. J. & Zhao, M. (2007). Biodiesel generation from oleaginous yeast Rhodotorula glutinis with xylose assimilating capacity. African Journal of Biotechnology, 6(18), 2130-2134.

Dey, P. & Maiti, M. K. (2013). Molecular characterization of a novel isolate of Candida tropicalis for enhanced lipid production. Journal of Applied Microbiology, 114(5), 1357-1368.

Dobrowolski, A., Mituła, P., Rymowicz, W. & Mirończuk, A. M. (2016). Efficient conversion of crude glycerol from various industrial wastes into single cell oil by yeast Yarrowia lipolytica. Bioresource Technology, 207, 237-243.

Dobson, R., Gray, V. & Rumbold, K. (2012). Microbial utilization of crude glycerol for the production of value-added products. Journal of Industrial Microbiology and Biotechnology, 39(2), 217-226.

Duarte, S. H. & Maugeri, F. (2014). Prediction of quality properties for biodiesel production by oleaginous yeast cultivated in pure and raw glycerol. Chemical Engineering Transactions.

Easterling, E. R., French, W. T., Hernandez, R. & Licha, M. (2009). The effect of glycerol as a sole and secondary substrate on the growth and fatty acid composition of Rhodotorula glutinis. Bioresource Technology, 100(1), 356-61.

Economou, C. N., Aggelis, G., Pavlou, S. & Vayenas, D. V. (2011). Modeling of single-cell oil production under nitrogen-limited and substrate inhibition conditions. Biotechnology and Bioengineering, 108(5), 1049-1055.

Evans, C. T. & Ratledge, C. (1984). Effect of nitrogen source on lipid accumulation in oleaginous yeasts. Journal of General Microbiology, 130, 1693-1704.

Fakas, S., Papanikolaou, S., Batsos, A., Galiotou-Panayotou, M., Mallouchos, A. & Aggelis, G. (2009). Evaluating renewable carbon sources as substrates for single cell oil production by Cunninghamella echinulata and Mortierella isabellina. Biomass and Bioenergy, 33(4), 573-580.

Fakas, S., Papanikolaou, S., Galiotou-Panayotou, M., Komaitis, M. & Aggelis, G. (2008). Organic nitrogen of tomato waste hydrolysate enhances glucose uptake and lipid accumulation in Cunninghamella echinulata. Journal of Applied Microbiology, 105(4), 1062-70.

Fonovich, T. M. (2012). Sudan dyes: are they dangerous for human health? Drug and Chemical Toxicology, 36(3), 343-352.

Fontanille, P., Kumar, V., Christophe, G., Nouaille, R. & Larroche, C. (2012). Bioconversion of volatile fatty acids into lipids by the oleaginous yeast Yarrowia lipolytica. Bioresource Technology, 114, 443-449.

Gabier, A. C., Gourdon, P., Reitz, J., Leveau, J. Y. & Bouix, M. (2005). Intracellular physiological events of yeast Rhodotorula glutinis during storage at +4 degrees C. International Journal of Food Microbiology, 105(2), 97-109.

Gao, D., Zeng, J., Zheng, Y., Yu, X. & Chen, S. (2013). Microbial lipid production from xylose by Mortierella isabellina. Bioresource Technology, 133, 315-321.

Gao, Q., Cui, Z., Zhang, J. & Bao, J. (2014). Lipid fermentation of corncob residues hydrolysate by oleaginous yeast Trichosporon cutaneum. Bioresource Technology, 152, 552-556.

Gao, Z., Ma, Y., Wang, Q., Zhang, M., Wang, J. & Liu, Y. (2016). Effect of crude glycerol impurities on lipid preparation by Rhodosporidium toruloides yeast 32489. Bioresource Technology, 218, 373-379.

Garcia-Ochoa, F. & Gomez, E. (2009). Bioreactor scale-up and oxygen transfer rate in microbial processes: An overview. Biotechnology Advances, 27(2), 153-176.

Garcia-Ochoa, F., Gomez, E., Santos, V. E. & Merchuk, J. C. (2010). Oxygen uptake rate in microbial processes: An overview. Biochemical Engineering Journal, 49(3), 289-307.

Gill, C. O., Hall, M. J. & Ratledge, C. (1977). Lipid accumulation in an oleaginous yeast (Candida 107) growing on glucose in single-stage continuous culture. Applied and Environmental Microbiology, 33(2), 231-239.

Gong, Z., Shen, H., Zhou, W., Wang, Y., Yang, X. & Zhao, Z. K. (2015). Efficient conversion of acetate into lipids by the oleaginous yeast Cryptococcus curvatus. Biotechnology for Biofuels, 8(189).

References

195

Gong, Z., Wang, Q., Shen, H., Hu, C., Jin, G. & Zhao, Z. K. (2012). Co-fermentation of cellobiose and xylose by Lipomyces starkeyi for lipid production. Bioresource Technology, 117, 20-24.

Gonzalez-Garcia, Y., Hernandez, R., Zhang, G., Escalante, F. M. E., Holmes, W. & French, W. T. (2013). Lipids accumulation in Rhodotorula glutinis and Cryptococcus curvatus growing on distillery wastewater as culture medium. Environmental Progress and Sustainable Energy, 32(1), 69-74.

González-Pajuelo, M., Meynial-Salles, I., Mendes, F., Andrade, J. C., Vasconcelos, I. & Soucaille, P. (2005). Metabolic engineering of Clostridium acetobutylicum for the industrial production of 1,3-propanediol from glycerol. Metabolic Engineering, 7(5–6), 329-336.

Granger, L. M., Perlot, P., Goma, G. & Pareilleux, A. (1993). Effect of various nutrient limitations on fatty acid production by Rhodotorula glutinis. Applied Microbiology and Biotechnology, 38(6), 784-789.

Haas, M. J., McAloon, A. J., Yee, W. C. & Foglia, T. A. (2006). A process model to estimate biodiesel production costs. Bioresource Technology, 97(4), 671-678.

Hartman, T. L. (1940). The Use of Sudan Black B as a Bacterial Fat Stain. Biotechnic & Histochemistry, 15(1), 23-28.

Hegde, K., Chandra, N., Sarma, S. J., Brar, S. K. & Veeranki, V. D. (2015). Genetic Engineering Strategies for Enhanced Biodiesel Production. Molecular Biotechnology, 57(7), 606-624.

Hiremath, A., Kannabiran, M. & Rangaswamy, V. (2011). 1,3-Propanediol production from crude glycerol from jatropha biodiesel process. New Biotechnology, 28(1), 19-23.

Hossack, J. A. & Spencer-Martins, I. (1978). Lipid composition of the yeast. European journal of applied microbiology and biotechnology, 5(4), 273-278.

Hu, C., Wu, S., Wang, Q., Jin, G., Shen, H. & Zhao, Z. K. (2011). Simultaneous utilization of glucose and xylose for lipid production by Trichosporon cutaneum. Biotechnology for Biofuels, 4(1), 1-8.

Hu, C., Zhao, X., Zhao, J., Wu, S. & Zhao, Z. K. (2009). Effects of biomass hydrolysis by-products on oleaginous yeast Rhodosporidium toruloides. Bioresource Technology, 100(20), 4843-4847.

Huang, C., Chen, X. F., Xiong, L., Chen, X. D., Ma, L. L. & Chen, Y. (2013). Single cell oil production from low-cost substrates: The possibility and potential of its industrialization. Biotechnology Advances, 31(2), 129-139.

Huang, C., Zong, M.-h., Wu, H. & Liu, Q.-p. (2009). Microbial oil production from rice straw hydrolysate by Trichosporon fermentans. Bioresource Technology, 100(19), 4535-4538.

Huang, G., Chen, F., Wei, D., Zhang, X. & Chen, G. (2010). Biodiesel production by microalgal biotechnology. Applied Energy, 87(1), 38-46.

Huang, X.-F., Liu, J.-N., Lu, L.-J., Peng, K.-M., Yang, G.-X. & Liu, J. (2016). Culture strategies for lipid production using acetic acid as sole carbon source by Rhodosporidium toruloides. Bioresource Technology, 206, 141-149.

Johnson, D. T. & Taconi, K. A. (2007). The glycerin glut: Options for the value-added conversion of crude glycerol resulting from biodiesel production. Environmental Progress, 26(4), 338-348.

Kalscheuer, R., Stölting, T. & Steinbüchel, A. (2006). Microdiesel: Escherichia coli engineered for fuel production. Microbiology, 152(9), 2529-2536.

Karamerou, E. E., Theodoropoulos, C. & Webb, C. (2016a). A biorefinery approach to microbial oil production from glycerol by Rhodotorula glutinis. Biomass and Bioenergy, 89, 113-122.

Karamerou, E. E., Theodoropoulos, C. & Webb, C. (2016b). Evaluating feeding strategies for microbial oil production from glycerol by Rhodotorula glutinis. Engineering in Life Sciences, DOI: 10.1002/elsc.201600073 (In press).

Kitcha, S. & Cheirsilp, B. (2011). Screening of Oleaginous Yeasts and Optimization for Lipid Production Using Crude Glycerol as a Carbon Source. Energy Procedia, 9, 274-282.

Kitcha, S. & Cheirsilp, B. (2013). Enhancing Lipid Production from Crude Glycerol by Newly Isolated Oleaginous Yeasts: Strain Selection, Process Optimization, and Fed-Batch Strategy. Bioenergy Research, 6(1), 300-310.

Kot, A. M., Błażejak, S., Kurcz, A., Gientka, I. & Kieliszek, M. (2016). Rhodotorula glutinis—potential source of lipids, carotenoids, and enzymes for use in industries. Applied Microbiology and Biotechnology, 1-15.


196

Koutinas, A. A., Chatzifragkou, A., Kopsahelis, N., Papanikolaou, S. & Kookos, I. K. (2014). Design and techno-economic evaluation of microbial oil production as a renewable resource for biodiesel and oleochemical production. Fuel, 116, 566-577.

Koutinas, A. A., Du, C., Wang, R. H. & Webb, C. (2008). Production of Chemicals from Biomass. Introduction to Chemicals from Biomass. Chapter 4, pp. 77-101. John Wiley & Sons, Ltd.

Koutinas, A. A., Wang, R.-H. & Webb, C. (2007). The biochemurgist –Bioconversion of agricultural raw materials for chemical production. Biofuels, Bioproducts and Biorefining, 1(1), 24-38.

Kumar, M. & Sharma, M. P. (2015). Assessment of potential of oils for biodiesel production. Renewable and Sustainable Energy Reviews, 44, 814-823.

Kurosawa, K., Boccazzi, P., de Almeida, N. M. & Sinskey, A. J. (2010). High-cell-density batch fermentation of Rhodococcus opacus PD630 using a high glucose concentration for triacylglycerol production. Journal of Biotechnology, 147(3–4), 212-218.

Kurosawa, K., Radek, A., Plassmeier, J. K. & Sinskey, A. J. (2015). Improved glycerol utilization by a triacylglycerol-producing Rhodococcus opacus strain for renewable fuels. Biotechnology for Biofuels, 8(1).

Lages, F., Silva-Graca, M. & Lucas, C. (1999). Active glycerol uptake is a mechanism underlying halotolerance in yeasts: a study of 42 species. Microbiology, 145, 2577-85.

Leber, R., Zinser, E., Paltauf, F., Daum, G. & Zellnig, G. (1994). Characterization of lipid particles of the yeast, Saccharomyces cerevisiae. Yeast, 10(11), 1421-1428.

Leiva-Candia, D. E., Pinzi, S., Redel-Macías, M. D., Koutinas, A., Webb, C. & Dorado, M. P. (2014). The potential for agro-industrial waste utilization using oleaginous yeast for the production of biodiesel. Fuel, 123, 33-42.

Li, Q., Du, W. & Liu, D. (2008). Perspectives of microbial oils for biodiesel production. Applied Microbiology and Biotechnology, 80(5), 749-756.

Lian, J., Garcia-Perez, M. & Chen, S. (2013). Fermentation of levoglucosan with oleaginous yeasts for lipid production. Bioresource Technology, 133, 183-189.

Liang, Y. (2016). Update on research and development of microbial oils. In: Commercialising biobased products: Opportunities, Challenges, Benefits and Risks, Seth W Snyder (ed.), RSC Green Chemistry, 166-189.

Liang, Y., Cui, Y., Trushenski, J. & Blackburn, J. W. (2010). Converting crude glycerol derived from yellow grease to lipids through yeast fermentation. Bioresource Technology, 101(19), 7581-7586.

Liang, Y., Sarkany, N. & Cui, Y. (2009). Biomass and lipid productivities of Chlorella vulgaris under autotrophic, heterotrophic and mixotrophic growth conditions. Biotechnology Letters, 31(7), 1043-1049.

Lin, J., Li, S., Sun, M., Zhang, C., Yang, W., Zhang, Z., Li, X. & Li, S. (2014). Microbial lipid production by oleaginous yeast in d-xylose solution using a two-stage culture mode. RSC Advances, 4(66), 34944-34949.

Lin, J., Shen, H., Tan, H., Zhao, X., Wu, S., Hu, C. & Zhao, Z. K. (2011). Lipid production by Lipomyces starkeyi cells in glucose solution without auxiliary nutrients. Journal of Biotechnology, 152(4), 184-188.

Liu, Y., Wang, Y., Liu, H. & Zhang, J. a. (2015). Enhanced lipid production with undetoxified corncob hydrolysate by Rhodotorula glutinis using a high cell density culture strategy. Bioresource Technology, 180, 32-39.

Louhasakul, Y. & Cheirsilp, B. (2013). Industrial Waste Utilization for Low-Cost Production of Raw Material Oil Through Microbial Fermentation. Applied Biochemistry and Biotechnology, 169(1), 110-122.

Lu, X., Vora, H. & Khosla, C. (2008). Overproduction of free fatty acids in E. coli: Implications for biodiesel production. Metabolic Engineering, 10(6), 333-339.

Lundin, H. (1950). Fat synthesis by miroorganisms and its possible applications in industry. Journal of the Institute of Brewing, 56(1), 17-28.

Luo, X., Ge, X., Cui, S. & Li, Y. (2016). Value-added processing of crude glycerol into chemicals and polymers. Bioresource Technology, 215, 144-154.

References

197

Makri, A., Fakas, S. & Aggelis, G. (2010). Metabolic activities of biotechnological interest in Yarrowia lipolytica grown on glycerol in repeated batch cultures. Bioresource Technology, 101(7), 2351-2358.

Mata, T. M., Martins, A. A. & Caetano, N. S. (2010). Microalgae for biodiesel production and other applications: A review. Renewable and Sustainable Energy Reviews, 14(1), 217-232.

Meesters, P. A. E. P., Huijberts, G. N. M. & Eggink, G. (1996). High-cell-density cultivation of the lipid accumulating yeast Cryptococcus curvatus using glycerol as a carbon source. Applied Microbiology and Biotechnology, 45(5), 575-579.

Meeuwse, P., Akbari, P., Tramper, J. & Rinzema, A. (2012). Modeling growth, lipid accumulation and lipid turnover in submerged batch cultures of Umbelopsis isabellina. Bioprocess and Biosystems Engineering, 35(4), 591-603.

Meeuwse, P., Tramper, J. & Rinzema, A. (2011). Modeling lipid accumulation in oleaginous fungi in chemostat cultures. II: Validation of the chemostat model using yeast culture data from literature. Bioprocess and Biosystems Engineering, 34(8), 951-961.

Meng, X., Yang, J. M., Xu, X., Zhang, L., Nie, Q. J. & Xian, M. (2009). Biodiesel production from oleaginous microorganisms. Renewable Energy, 34(1), 1-5.

Moeller, L., Strehlitz, B., Aurich, A., Zehnsdorf, A. & Bley, T. (2007). Optimization of Citric Acid Production from Glucose by Yarrowia lipolytica. Engineering in Life Sciences, 7(5), 504-511.

Mu, J., Li, S., Chen, D., Xu, H., Han, F., Feng, B. & Li, Y. (2015). Enhanced biomass and oil production from sugarcane bagasse hydrolysate (SBH) by heterotrophic oleaginous microalga Chlorella protothecoides. Bioresource Technology, 185, 99-105.

Murphy, D. J. & Vance, J. (1999). Mechanisms of lipid-body formation. Trends in Biochemical Sciences, 24(3), 109-115.

Pagliaro, M. & Rossi, M. (2010). Chapter 1 Glycerol: Properties and Production. The Future of Glycerol (2). The Royal Society of Chemistry.

Pan, J., Kwak, M. & Rhee, J. (1986). High density cell culture of Rhodotorula glutinis using oxygen-enriched air. Biotechnology Letters, 8(10), 715-718.

Panagiotopoulou, P., Karamerou, E. E. & Kondarides, D. I. (2013). Kinetics and mechanism of glycerol photo-oxidation and photo-reforming reactions in aqueous TiO2 and Pt/TiO2 suspensions. Catalysis Today, 209, 91-98.

Papanikolaou, S. & Aggelis, G. (2002). Lipid production by Yarrowia lipolytica growing on industrial glycerol in a single-stage continuous culture. Bioresource Technology, 82(1), 43-49.

Papanikolaou, S. & Aggelis, G. (2011a). Lipids of oleaginous yeasts. Part I: Biochemistry of single cell oil production. European Journal of Lipid Science and Technology, 113(8), 1031-1051.

Papanikolaou, S. & Aggelis, G. (2011b). Lipids of oleaginous yeasts. Part II: Technology and potential applications. European Journal of Lipid Science and Technology, 113(8), 1052-1073.

Papanikolaou, S., Chevalot, I., Komaitis, M., Marc, I. & Aggelis, G. (2002a). Single cell oil production by Yarrowia lipolytica growing on an industrial derivative of animal fat in batch cultures. Applied Microbiology and Biotechnology, 58(3), 308-312.

Papanikolaou, S., Muniglia, L., Chevalot, I., Aggelis, G. & Marc, I. (2002b). Yarrowia lipolytica as a potential producer of citric acid from raw glycerol. J Appl Microbiol, 92(4), 737-44.

Petrik, S., Marova, I., Haronikova, A., Kostovova, I. & Breierova, E. (2013). Production of biomass, carotenoid and other lipid metabolites by several red yeast strains cultivated on waste glycerol from biofuel production - A comparative screening study. Annals of Microbiology, 63(4), 1537-1551.

Pinzi, S., Leiva, D., López-García, I., Redel-Macías, M. D. & Dorado, M. P. (2014). Latest trends in feedstocks for biodiesel production. Biofuels, Bioproducts and Biorefining, 8(1), 126-143.

Poli, J. S., da Silva, M. A. N., Siqueira, E. P., Pasa, V. M. D., Rosa, C. A. & Valente, P. (2014). Microbial lipid produced by Yarrowia lipolytica QU21 using industrial waste: A potential feedstock for biodiesel production. Bioresource Technology, 161, 320-326.

Raimondi, S., Rossi, M., Leonardi, A., Bianchi, M., Rinaldi, T. & Amaretti, A. (2014). Getting lipids from glycerol: new perspectives on biotechnological exploitation of Candida freyschussii. Microbial Cell Factories, 13(1), 83.


198

Rakicka, M., Lazar, Z., Dulermo, T., Fickers, P. & Nicaud, J. M. (2015). Lipid production by the oleaginous yeast Yarrowia lipolytica using industrial by-products under different culture conditions. Biotechnology for Biofuels, 8(1).

Ratledge, C. (2001). Microbial Lipids. Biotechnology Set. Wiley-VCH Verlag GmbH. Ratledge, C. (2013). Microbial oils: An introductory overview of current status and future prospects.

OCL - Oilseeds and fats, crops and lipids, 20(6). Ratledge, C. & Cohen, Z. (2008). Microbial and algal oils: Do they have a future for biodiesel or as

commodity oils? Lipid Technology, 20(7), 155-160. Ratledge, C. & Hall, M. J. (1977). Oxygen demand by lipid-accumulating yeasts in continuous culture.

Applied and Environmental Microbiology, 34(2), 230-231. Ratledge, C. & Wynn, J. P. (2002). The biochemistry and molecular biology of lipid accumulation in

oleaginous microorganisms. Adv Appl Microbiol, 51, 1-51. Ren, H.-Y., Liu, B.-F., Ma, C., Zhao, L. & Ren, N.-Q. (2013). A new lipid-rich microalga Scenedesmus sp.

strain R-16 isolated using Nile red staining: effects of carbon and nitrogen sources and initial pH on the biomass and lipid production. Biotechnology for Biofuels, 6(1), 1-10.

Rywińska, A., Rymowicz, W. & Marcinkiewicz, M. (2010). Valorization of raw glycerol for citric acid production by Yarrowia lipolytica yeast. Electronic Journal of Biotechnology; Vol 13, No 4.

Sabnis, R. & W. (2010). Handbook of Biological Dyes and Stains: John Wiley and Sons. Saenge, C., Cheirsilp, B., Suksaroge, T. T. & Bourtoom, T. (2011a). Efficient concomitant production of

lipids and carotenoids by oleaginous red yeast Rhodotorula glutinis cultured in palm oil mill effluent and application of lipids for biodiesel production. Biotechnology and Bioprocess Engineering, 16(1), 23-33.

Saenge, C., Cheirsilp, B., Suksaroge, T. T. & Bourtoom, T. (2011b). Potential use of oleaginous red yeast Rhodotorula glutinis for the bioconversion of crude glycerol from biodiesel plant to lipids and carotenoids. Process Biochemistry, 46(1), 210-218.

Salakkam, A. & Webb, C. (2015). The inhibition effect of methanol, as a component of crude glycerol, on the growth rate of Cupriavidus necator and other micro-organisms. Biochemical Engineering Journal, 98, 84-90.

Santamauro, F., Whiffin, F., Scott, R. & Chuck, C. (2014). Low-cost lipid production by an oleaginous yeast cultured in non-sterile conditions using model waste resources. Biotechnology for Biofuels, 7(1), 34.

Santos, C. A. & Reis, A. (2014). Microalgal symbiosis in biotechnology. Applied Microbiology and Biotechnology.

Saxena, V., Sharma, C. D., Bhagat, S. D., Saini, V. S. & Adhikari, D. K. (1998). Lipid and fatty acid biosynthesis by Rhodotorula minuta. JAOCS, Journal of the American Oil Chemists' Society, 75(4), 501-505.

Schneider, T., Graeff-Hönninger, S., French, W. T., Hernandez, R., Merkt, N., Claupein, W., Hetrick, M. & Pham, P. (2013). Lipid and carotenoid production by oleaginous red yeast Rhodotorula glutinis cultivated on brewery effluents. Energy, 61, 34-43.

Scott, C., Hilton, M. E., Coppin, C. W., Russell, R. J., Oakeshott, J. G. & Sutherland, T. D. (2007). A global response to sulfur starvation in Pseudomonas putida and its relationship to the expression of low-sulfur-content proteins. FEMS Microbiology Letters, 267(2), 184-193.

Sestric, R., Munch, G., Cicek, N., Sparling, R. & Levin, D. B. (2014). Growth and neutral lipid synthesis by Yarrowia lipolytica on various carbon substrates under nutrient-sufficient and nutrient-limited conditions. Bioresource Technology, 164(0), 41-46.

Shen, H., Gong, Z., Yang, X., Jin, G., Bai, F. & Zhao, Z. K. (2013a). Kinetics of continuous cultivation of the oleaginous yeast Rhodosporidium toruloides. Journal of Biotechnology, 168(1), 85-89.

Shen, Q., Lin, H., Zhan, J., Wang, Q. & Zhao, Y. (2013b). Sweetpotato vines hydrolysate induces glycerol to be an effective substrate for lipid production of Trichosporon fermentans. Bioresource Technology, 136, 725-729.

Shi, S., Valle-Rodriguez, J. O., Siewers, V. & Nielsen, J. (2011). Prospects for microbial biodiesel production. Biotechnol J, 6(3), 277-85.

References

199

Shields-Menard, S. A., Amirsadeghi, M., Sukhbaatar, B., Revellame, E., Hernandez, R., Donaldson, J. R. & French, W. T. (2015). Lipid accumulation by Rhodococcus rhodochrous grown on glucose. Journal of Industrial Microbiology & Biotechnology, 42(5), 693-699.

Sitepu, I. R., Garay, L. A., Sestric, R., Levin, D., Block, D. E., German, J. B. & Boundy-Mills, K. L. (2014). Oleaginous yeasts for biodiesel: Current and future trends in biology and production. Biotechnology Advances, 32(7), 1336-1360.

Sitepu, I. R., Ignatia, L., Franz, A. K., Wong, D. M., Faulina, S. A., Tsui, M., Kanti, A. & Boundy-Mills, K. (2012). An improved high-throughput Nile red fluorescence assay for estimating intracellular lipids in a variety of yeast species. Journal of Microbiological Methods, 91(2), 321-328.

Slininger, P. J., Dien, B. S., Kurtzman, C. P., Moser, B. R., Bakota, E. L., Thompson, S. R., O'Bryan, P. J., Cotta, M. A., Balan, V., Jin, M., Sousa, L. d. C. & Dale, B. E. (2016). Comparative lipid production by oleaginous yeasts in hydrolyzates of lignocellulosic biomass and process strategy for high titers. Biotechnology and Bioengineering.

Smith, D. G. & Marchant, R. (1969). Unbalanced cell-wall synthesis in chloramphenicol-grown Rhodotorula glutinis. Antonie van Leeuwenhoek, 35(1), 113-119.

Souza, K. S. T., Schwan, R. F. & Dias, D. R. (2014). Lipid and citric acid production by wild yeasts grown in glycerol. Journal of Microbiology and Biotechnology, 24(4), 497-506.

Starkey, R. L. (1946). Lipid Production by a Soil Yeast. Journal of Bacteriology, 51(1), 33-50. Tai, M. & Stephanopoulos, G. (2013). Engineering the push and pull of lipid biosynthesis in oleaginous

yeast Yarrowia lipolytica for biofuel production. Metabolic Engineering, 15, 1-9. Tan, H. W., Abdul Aziz, A. R. & Aroua, M. K. (2013). Glycerol production and its applications as a raw

material: A review. Renewable and Sustainable Energy Reviews, 27(0), 118-127. Tchakouteu, S. S., Kalantzi, O., Gardeli, C., Koutinas, A. A., Aggelis, G. & Papanikolaou, S. (2015). Lipid

production by yeasts growing on biodiesel-derived crude glycerol: strain selection and impact of substrate concentration on the fermentation efficiency. Journal of Applied Microbiology, 118(4), 911-27.

Thakur, M. S., Prapulla, S. G. & Karanth, N. G. (1988). Microscopic observation of Sudan Black B staining to monitor lipid production by microbes. Journal of Chemical Technology & Biotechnology, 42(2), 129-134.

Thakur, M. S., Prapulla, S. G. & Karanth, N. G. (1989). Estimation of intracellular lipids by the measurement of absorbance of yeast cells stained with Sudan Black B. Enzyme and Microbial Technology, 11(4), 252-254.

Thiru, M., Sankh, S. & Rangaswamy, V. (2011). Process for biodiesel production from Cryptococcus curvatus. Bioresource Technology, 102(22), 10436-10440.

Thliveros, P., Uçkun Kiran, E. & Webb, C. (2014). Microbial biodiesel production by direct methanolysis of oleaginous biomass. Bioresource Technology, 157(0), 181-187.

Tsakona, S., Skiadaresis, A. G., Kopsahelis, N., Chatzifragkou, A., Papanikolaou, S., Kookos, I. K. & Koutinas, A. A. (2016). Valorisation of side streams from wheat milling and confectionery industries for consolidated production and extraction of microbial lipids. Food Chemistry, 198, 85-92.

Uçkun Kiran, E. (2012). Developing value-added products from agricultural and biorefinery by-products by using biological pretreatment strategies. PhD thesis, Middle East Technical University.

Uçkun Kiran, E., Salakkam, A., Trzcinski, A. P., Bakir, U. & Webb, C. (2012). Enhancing the value of nitrogen from rapeseed meal for microbial oil production. Enzyme and Microbial Technology, 50(6-7), 337-342.

Uçkun Kiran, E., Trzcinski, A. & Webb, C. (2013). Microbial oil produced from biodiesel by-products could enhance overall production. Bioresource Technology, 129, 650-654.

Vamvakaki, A.-N., Kandarakis, I., Kaminarides, S., Komaitis, M. & Papanikolaou, S. (2010). Cheese whey as a renewable substrate for microbial lipid and biomass production by Zygomycetes. Engineering in Life Sciences, 10(4), 348-360.


200

Vicente, G., Bautista, L. F., Rodríguez, R., Gutiérrez, F. J., Sádaba, I., Ruiz-Vázquez, R. M., Torres-Martínez, S. & Garre, V. (2009). Biodiesel production from biomass of an oleaginous fungus. Biochemical Engineering Journal, 48(1), 22-27.

Villadsen, J., Nielsen, J. & Lidén, G. (2011). Bioreaction Engineering Principles. Boston, MA: Springer US: Boston, MA.

Vlysidis, A., Binns, M., Webb, C. & Theodoropoulos, C. (2011). Glycerol utilisation for the production of chemicals: Conversion to succinic acid, a combined experimental and computational study. Biochemical Engineering Journal, 58-59(1), 1-11.

Wang, P., Wan, X., Zhang, Y. & Jiang, M. (2011). Production of γ-linolenic acid using a novel heterologous expression system in the oleaginous yeast Lipomyces kononenkoae. Biotechnology Letters, 33(10), 1993-1998.

Wang, X. & Ren, H. (2014). Microbial oil production by Rhodotorula glutinis CICC 31643 using sugar cane molasses. Journal of Renewable and Sustainable Energy, 6(1).

Wang, Z., Zhuge, J., Fang, H. & Prior, B. A. (2001). Glycerol production by microbial fermentation: A review. Biotechnology Advances, 19(3), 201-223.

Wu, S., Hu, C., Jin, G., Zhao, X. & Zhao, Z. K. (2010). Phosphate-limitation mediated lipid production by Rhodosporidium toruloides. Bioresource Technology, 101(15), 6124-6129.

Wu, S., Zhao, X., Shen, H., Wang, Q. & Zhao, Z. K. (2011). Microbial lipid production by Rhodosporidium toruloides under sulfate-limited conditions. Bioresource Technology, 102(2), 1803-1807.

Wynn, J. P. & Ratledge, C. (2005). Oils from Microorganisms. Bailey's Industrial Oil and Fat Products. John Wiley & Sons, Inc.

Xu, J., Du, W., Zhao, X., Zhang, G. & Liu, D. (2013). Microbial oil production from various carbon sources and its use for biodiesel preparation. Biofuels, Bioproducts and Biorefining, 7(1), 65-77.

Xue, F., Miao, J., Zhang, X. & Tan, T. (2010). A new strategy for lipid production by mix cultivation of Spirulina platensis and Rhodotorula glutinis. Applied Biochemistry and Biotechnology, 160(2), 498-503.

Xue, F., Zhang, X., Luo, H. & Tan, T. (2006). A new method for preparing raw material for biodiesel production. Process Biochemistry, 41(7), 1699-1702.

Yang, M. & Zheng, S. (2014). Pollutant removal-oriented yeast biomass production from high-organic-strength industrial wastewater: A review. Biomass and Bioenergy, 64, 356-362.

Yang, Q., Yang, M., Zhang, S. & Lv, W. (2005). Treatment of wastewater from a monosodium glutamate manufacturing plant using successive yeast and activated sludge systems. Process Biochemistry, 40(7), 2483-2488.

Yang, X., Jin, G., Gong, Z., Shen, H., Bai, F. & Zhao, Z. K. (2014). Recycling biodiesel-derived glycerol by the oleaginous yeast Rhodosporidium toruloides Y4 through the two-stage lipid production process. Biochemical Engineering Journal, (0).

Yang, X., Jin, G., Gong, Z., Shen, H., Bai, F. & Zhao, Z. K. (2015). Recycling microbial lipid production wastes to cultivate oleaginous yeasts. Bioresource Technology, 175, 91-96.

Yen, H.-W., Liu, Y. X. & Chang, J.-S. (2015a). The effects of feeding criteria on the growth of oleaginous yeast—Rhodotorula glutinis in a pilot-scale airlift bioreactor. Journal of the Taiwan Institute of Chemical Engineers, 49, 67-71.

Yen, H.-W. & Zhang, Z. (2011). Effects of dissolved oxygen level on cell growth and total lipid accumulation in the cultivation of Rhodotorula glutinis. Journal of Bioscience and Bioengineering, 112(1), 71-74.

Yen, H. W., Chang, J. T. & Chang, J. S. (2015b). The growth of oleaginous Rhodotorula glutinis in an internal-loop airlift bioreactor by using mixture substrates of rice straw hydrolysate and crude glycerol. Biomass and Bioenergy, 80, 38-43.

Yen, H. W. & Liu, Y. X. (2014). Application of airlift bioreactor for the cultivation of aerobic oleaginous yeast Rhodotorula glutinis with different aeration rates. Journal of Bioscience and Bioengineering.

Ykema, A., Verbree, E. C., Van Verseveld, H. W. & Smit, H. (1986). Mathematical modelling of lipid production by oleaginous yeasts in continuous cultures. Antonie van Leeuwenhoek, International Journal of General and Molecular Microbiology, 52(6), 491-506.

References

201

Yousuf, A., Sannino, F., Addorisio, V. & Pirozzi, D. (2010). Microbial Conversion of Olive Oil Mill Wastewaters into Lipids Suitable for Biodiesel Production. Journal of Agricultural and Food Chemistry, 58(15), 8630-8635.

Yu, X., Zheng, Y., Xiong, X. & Chen, S. (2014). Co-utilization of glucose, xylose and cellobiose by the oleaginous yeast Cryptococcus curvatus. Biomass and Bioenergy, 71, 340-349.

Zakerhamidi, M. S., Golghasemi Sorkhabi, S. & Shamkhali, A. N. (2014). Polar and low polar solvents media effect on dipole moments of some diazo Sudan dyes. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 127(0), 340-348.

Zhang, J., Fang, X., Zhu, X. L., Li, Y., Xu, H. P., Zhao, B. F., Chen, L. & Zhang, X. D. (2011). Microbial lipid production by the oleaginous yeast Cryptococcus curvatus O3 grown in fed-batch culture. Biomass and Bioenergy, 35(5), 1906-1911.

Zhang, Z., Ji, H., Gong, G., Zhang, X. & Tan, T. (2014a). Synergistic effects of oleaginous yeast Rhodotorula glutinis and microalga Chlorella vulgaris for enhancement of biomass and lipid yields. Bioresource Technology, 164, 93-99.

Zhang, Z., Zhang, X. & Tan, T. (2014b). Lipid and carotenoid production by Rhodotorula glutinis under irradiation/high-temperature and dark/low-temperature cultivation. Bioresource Technology, 157, 149-153.

Zhao, X., Hu, C., Wu, S., Shen, H. & Zhao, Z. K. (2010a). Lipid production by Rhodosporidium toruloides Y4 using different substrate feeding strategies. Journal of Industrial Microbiology & Biotechnology, 38(5), 627-632.

Zhao, X., Kong, X., Hua, Y., Feng, B. & Zhao, Z. (2008). Medium optimization for lipid production through co‐fermentation of glucose and xylose by the oleaginous yeast Lipomyces starkeyi. European Journal of Lipid Science and Technology, 110(5), 405-412.

Zhao, X., Wu, S., Hu, C., Wang, Q., Hua, Y. & Zhao, Z. K. (2010b). Lipid production from Jerusalem artichoke by Rhodosporidium toruloides Y4. Journal of Industrial Microbiology & Biotechnology, 37(6), 581-585.

Zheng, Y., Li, T., Yu, X., Bates, P. D., Dong, T. & Chen, S. (2013). High-density fed-batch culture of a thermotolerant microalga Chlorella sorokiniana for biofuel production. Applied Energy, 108, 281-287.

Zhou, W., Wang, W., Li, Y. & Zhang, Y. (2013). Lipid production by Rhodosporidium toruloides Y2 in bioethanol wastewater and evaluation of biomass energetic yield. Bioresource Technology, 127, 435-440.

Čertík, M., Adamechová, Z. & Laoteng, K. (2012). Microbial production of γ-linolenic acid: Submerged versus solid-state fermentations. Food Science and Biotechnology, 21(4), 921-926.


202

APPENDIX 1

The appended file, “EK paper 1.pdf” contains a full reprint copy of the following publication

arising from the research presented in this thesis.

Karamerou E., Theodoropoulos C., Webb C. (2016). “A biorefinery approach to microbial oil

production from glycerol by Rhodotorula glutinis”, Biomass and Bioenergy (89):113-122,



Appendix

203

APPENDIX 2

The appended file, “EK paper 2.pdf” contains a full preprint copy of the following publication

arising from the research presented in this thesis.

Karamerou E., Theodoropoulos C., Webb C. (2016). “Evaluating feeding strategies for microbial

oil production from glycerol by Rhodotorula glutinis”, Engineering in Life Sciences (Article in

Press), DOI: 10.1002/elsc.201600073

http://onlinelibrary.wiley.com/doi/10.1002/elsc.201600073/full

Documents

BIOPROCESSING STRATEGIES FOR THE CULTIVATION OF …