development of new cultivation strategies to …

182
DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO ENHANCE HETEROLOGOUS PROTEIN PRODUCTION IN PICHIA PASTORIS A thesis submitted to University College London for the degree of Doctor of Philosophy By Baolong Wang December 2018 The Advanced Centre for Biochemical Engineering Department of Biochemical Engineering University College London London WC1E 6BT

Upload: others

Post on 15-Apr-2022

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

DEVELOPMENT OF NEW

CULTIVATION STRATEGIES TO

ENHANCE HETEROLOGOUS

PROTEIN PRODUCTION IN PICHIA PASTORIS

A thesis submitted to University College London

for the degree of Doctor of Philosophy

By Baolong Wang

December 2018

The Advanced Centre for Biochemical Engineering

Department of Biochemical Engineering

University College London

London

WC1E 6BT

Page 2: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

2

Declaration

I, Baolong Wang, confirm that the work presented in this thesis is my own. Where information

has been derived from other sources, I confirm that this has been indicated in the thesis.

Signed:……………………………………………………………………….

Date:…………………………………………………………………………..

Page 3: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

3

In loving memory of my grandfather

and to my parents and grandmother

Page 4: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

4

Acknowledgement

First and foremost, I would like to thank my supervisor, Eli Keshavarz-Moore, for giving me

the opportunity to study under her supervision. Without her guidance, support and

encouragement, it would be impossible for me to move a step in my PhD study.

I also would like to thank my advisor, Darren Nesbeth, for his patience and constructive advices

in development of my project.

I would like to thank the people in the Research and Facilities Team, Dr. Michael Sulu, Dr.

Gareth Mannall and Dr. Brian O’Sulliva, for always being around to solve any problems of the

facilities and provide technical trainings.

I would like to thank Dr. Lourdes Velez Suberbie and Dr. Stefan Woodhouse for helping me

kick off the Pichia fermentation as well as for teaching me very detailed fermentation skills and

making me a qualified bioreactor operator.

I would like to thank all the postdoctoral research associates and PhD candidates in the

Vineyard office, Cheng, Steve, Shaleem, Ashikin, Haiyuan, Darryl, Neha, Folarin and Milena,

Vincent, Maximilian, Pedro, etc. for creating a funny and friendly environment to work in.

I would like to thank my good friends Chuanjie, Hongyu, Haoran, Yang, Yiling, Fuyao,

Zhongdi, Xinyu, Yi, Fang, Qingyu, etc. for accompanying me and driving away my loneliness

in the last four years.

Finally, I would like to thank my parents for giving me endless support and love since I was

born. It is not an easy path since the first day I went to school over twenty years ago. They have

always been behind me and given me everything they have. Without them, it would be

impossible for me to pursue a PhD degree.

Page 5: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

5

Abstract

Growth of the biopharmaceutical market and advances in host strain engineering have fuelled

the application of Pichia pastoris in recombinant protein production. Production of

recombinant protein in P. pastoris is commonly induced by continuous methanol feeding.

However, methanol induction challenges scale-up of cultivation due to high oxygen

requirement and substantial heat generation. Developing novel induction strategies to minimize

methanol consumption is desirable.

This thesis compared the standard methanol induction with a sorbitol/methanol mixed induction

strategy and developed the induction method at large scale by using oxygen transfer rates (OTR)

as a scale-down criterion. Influences of sorbitol/methanol mixed induction on product recovery

were also studied using scale-down methodology.

Compared to standard methanol induction, substituting 50% (C-mol/C-mol) methanol with

sorbitol improved cell viability from 92.8±0.3% to 97.7±0.1% but reduced product yield from

1.65±0.03 g•L-1 to 1.12±0.07 g•L-1. Oxygen uptake rate was reduced from 241.4±15.0 mmol•L-

1•h-1 to 145.5±4.8 mmol•L-1•h-1 by using the mixed induction. Proteomics study showed that

supernatant from mixed induction contained fewer host cell proteins (72 versus 96) and fewer

types of protease (1 versus 3).

OTR expected in the large scale bioreactors was used as scale-down criterion at one litre. By

measuring oxygen transfer coefficient (kLa) of the small bioreactor, a fermentation process with

OTR of 150 mmol•L-1•h-1 was defined. Standard methanol induction would cause oxygen

depletion and methanol accumulation in the medium. Residual methanol concentration

apparently influenced the cell growth and product expression. The biomass and product yield

reached 108.3 g•L-1 and 1.1 g•L-1 when the residual methanol concentration was below 5 g•L-

1, whereas they were reduced to 75.5 g•L-1 and 0.87 g•L-1 at a higher concentration (5~10 g•L-

1).

Decreasing methanol feeding rate avoided the oxygen depletion. However, the biomass and

product yield were reduced to 92.2 g•L-1 and 0.85 g•L-1. Partially replacing methanol with

sorbitol enhanced the biomass to 130 g•L-1 but the product yield was not enhanced.

An ultra scale-down approach enabled the prediction of cell culture dewatering in pilot and

industrial scale centrifuges. It was found that the cell culture from mixed induction had higher

centrifugal dewatering levels (84.5±3.3% versus 78.1±3.9%), which was likely to be attributed

to the decrease of cell diameter during induction.

Page 6: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

6

Impact statement

P. pastoris is becoming a popular host for production of heterologous proteins. Using methanol

as the inducer challenges P. pastoris cultivation especially at large scale. This thesis has

established a new induction strategy by using sorbitol as a co-substrate. The strategy has been

shown to improve product purity, process scalability and centrifugal dewatering in the

production of recombinant aprotinin.

In laboratorial research, findings of the project will provide guidance on how to select the

induction method of P. pastoris. The sorbitol/methanol mixed induction can be considered over

methanol induction when 1) the cell culture has low viability, 2) the products are easily

degraded by proteases, 3) a lot of host cell proteins are co-released with the product.

The key findings are also of benefit to the biopharmaceutical industry. Using the mixed

induction strategy will diminish the consumption of pure oxygen and thus reduce the capital

cost of manufacture. Cutting the usage of flammable methanol will make the manufacture safer

and easier to be managed. Since fewer host cell proteins are co-released, it is easier to reduce

the impurities to a suitable level for clinical use. In addition, cell culture from the mixed

induction has higher efficiency of centrifugal dewatering and thus product loss in product

recovery is minimized. Consequently, the manufacturing cost per unit of product will be

reduced.

Last but not least, scale-down methodology was used in product quantification, cell robustness

and centrifugal dewatering study in this project. It can be referred by other people either from

academic or industrial field. Using scale-down method will reduce the cost of materials, time,

capital and labours and improve efficiency of process development.

Overall, outputs of this project have real impact on different levels. The influence has been

expanded by showing the results at international conferences. Currently, two research

manuscripts are being prepared to be published in academic journals.

Page 7: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

7

Table of content

Declaration ................................................................................................................... 2

Acknowledgement ....................................................................................................... 4

Abstract ........................................................................................................................ 5

Impact statement ......................................................................................................... 6

Table of content ........................................................................................................... 7

List of figures ............................................................................................................. 11

List of tables ............................................................................................................... 14

Nomenclature ............................................................................................................. 15

Greek symbols ........................................................................................................... 17

Abbreviations and symbols ...................................................................................... 18

Chapter 1 Introduction ............................................................................................. 21

1.1 Trends in biopharmaceutical manufacturing ..................................................... 21

1.1.1 Overview of approved therapeutic proteins ............................................... 21

1.1.2 Recognised hosts for biopharmaceutical manufacturing ........................... 22

1.1.3 P. pastoris as an emerging host for biopharmaceutical ............................. 23

1.2 Introduction to P. pastoris expression system .................................................. 26

1.2.1 History of P. pastoris expression system ................................................... 26

1.2.2 Phenotype of P. pastoris strains ................................................................. 26

1.2.3 Promoters used in P. pastoris system ......................................................... 27

1.2.4 Cell engineering of P. pastoris strains ....................................................... 28

1.3 Fed-batch fermentation of P. pastoris ............................................................... 29

1.3.1 Cell culture medium of P. pastoris ............................................................ 29

1.3.2 Fed-batch fermentation of P. pastoris ........................................................ 29

1.3.3 Oxygen unlimited fed-batch fermentation ................................................. 31

1.3.4 Oxygen limited fed-batch fermentation ..................................................... 32

1.4 Mixed induction strategies of P. pastoris .......................................................... 33

Page 8: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

8

1.4.1 Metabolism of methanol in P. pastoris ...................................................... 33

1.4.2 Metabolism of mixed carbons in P. pastoris .............................................. 34

1.4.3 Glycerol/methanol mixed induction ........................................................... 35

1.4.4 Sorbitol/methanol mixed induction ............................................................ 36

1.5 Impurities of host cell protein in P. pastoris processing ................................... 40

1.5.1 Control of host cell proteins in biopharmaceutical .................................... 40

1.5.2 Analytical methods of host cell proteins .................................................... 41

1.5.3 Host cell proteins in P. pastoris culture ..................................................... 42

1.5.4 Product degradation in P. pastoris culture ................................................. 42

1.6 Product recovery from P. pastoris culture ........................................................ 43

1.6.1 Typical process of product recovery .......................................................... 43

1.6.2 Types of centrifuge for product recovery ................................................... 44

1.6.3 Depth filtration as an alternative to centrifuge ........................................... 46

1.6.4 Product recovery using single microfiltration ............................................ 46

1.7 Thesis objectives ............................................................................................... 48

Chapter 2 Materials and methods ........................................................................... 50

2.1 Materials ............................................................................................................ 50

2.2 Culture medium of P. pastoris .......................................................................... 50

2.2.1 Buffered complex medium ......................................................................... 50

2.2.2 Basal salts medium ..................................................................................... 50

2.3 P. pastoris strain and working cell bank ........................................................... 50

2.4 Cell culture methods .......................................................................................... 51

2.4.1 Shake flask culture ..................................................................................... 51

2.4.2 Microplate culture ...................................................................................... 51

2.4.3 Infors bioreactor culture ............................................................................. 52

2.5 Prediction of dewatering and clarification ........................................................ 53

2.6 Characterization of oxygen transfer coefficient ................................................ 53

2.7 Cell breakage method ........................................................................................ 54

2.8 Cell robustness study ......................................................................................... 54

2.9 Analytical methods ............................................................................................ 54

2.9.1 Aprotinin measurement .............................................................................. 54

Page 9: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

9

2.9.2 Cell density measurement .......................................................................... 56

2.9.3 Determination of methanol and sorbitol ..................................................... 56

2.9.4 DNA quantification .................................................................................... 56

2.9.5 Soluble protein quantification .................................................................... 56

2.9.6 NuPAGE Bis-Tris gel electrophoresis ....................................................... 57

2.9.7 Cell viability measurement ......................................................................... 57

2.9.8 Two-dimensional gel electrophoresis ......................................................... 57

2.9.9 Proteomics assay ........................................................................................ 59

2.9.10 Cell size measurement .............................................................................. 60

2.9.11 Rheology measurement ............................................................................ 60

2.9.12 AOX activity measurement ...................................................................... 60

Chapter 3 Impact of sorbitol/methanol mixed induction on cell growth and product

expression ................................................................................................................... 62

3.1 Introduction ....................................................................................................... 62

3.2 Theoretical considerations ................................................................................. 64

3.2.1 Specific growth rate ................................................................................... 64

3.2.2 Cell respiration ........................................................................................... 64

3.3 Results and discussion ....................................................................................... 65

3.3.1 Quantitative assay of aprotinin ................................................................... 65

3.3.2 Cell culture at small scale ........................................................................... 69

3.3.3 Cell culture at bioreactor scale ................................................................... 73

3.3.4 Impact of mixed induction on product yield and purity ............................. 79

3.4 Conclusion ......................................................................................................... 88

Chapter 4 Development of induction strategies in a bioreactor with limited oxygen

transfer rate ............................................................................................................... 89

4.1 Introduction ....................................................................................................... 89

4.2 Theoretical considerations ................................................................................. 91

4.2.1 Determining OTR of the bioreactor ........................................................... 91

4.2.2 Determining feeding rates of the induction solutions ................................ 92

4.3 Results and discussion ....................................................................................... 92

4.3.1 Measurement of oxygen transfer coefficient .............................................. 92

4.3.2 Induction in an oxygen limited condition .................................................. 95

Page 10: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

10

4.3.3 Induction in an oxygen unlimited condition ............................................ 101

4.3.4 Measurement of AOX enzyme activity .................................................... 108

4.4 Conclusion ....................................................................................................... 112

Chapter 5 Prediction of cell robustness and centrifugal dewatering using scale-down

methodology ............................................................................................................. 113

5.1 Introduction ..................................................................................................... 113

5.2 Theoretical considerations ............................................................................... 115

5.2.1 Mimic of shear in large scale centrifuges ................................................. 115

5.2.2 Scale down of large scale centrifuges ...................................................... 115

5.2.3 Calculation of clarification and dewatering ............................................. 116

5.3 Results and discussion ..................................................................................... 117

5.3.1 Determination of biomass and product .................................................... 117

5.3.2 Prediction of cell robustness to shear ....................................................... 120

5.3.3 Characterization of cell culture properties ............................................... 122

5.3.4 Prediction of centrifugal dewatering ........................................................ 126

5.4 Conclusion ....................................................................................................... 132

Chapter 6 Conclusions and future work ............................................................... 133

6.1 Conclusions ..................................................................................................... 133

6.2 Future work ..................................................................................................... 136

Chapter 7 References .............................................................................................. 139

Chapter 8 Appendix ................................................................................................ 158

Page 11: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

11

List of figures

Figure 1.1 (A) Number of therapeutic proteins approved by FDA from 2011 to 2016. (B)

Classes of therapeutic proteins approved by FDA from 2011 to 2016 (Lagassé et al. 2017).

..................................................................................................................................... 22

Figure 1.2 Diagram of carbon feeding (A) and cell growth (B) in P. pastoris fed-batch

fermentation... ............................................................................................................. 30

Figure 1.3 Metabolic pathways of methanol and sorbitol when P. pastoris was induced by pure

methanol or methanol/sorbitol mixture (Gao et al. 2012). .......................................... 34

Figure 1.4 Operating mode of tubular bowl (A), disc type (B) and decanter scroll (C) centrifuge.

Adapted from (Dévay 2013; Tarleton and Wakeman 2016). ...................................... 45

Figure 3.1 Effect of trypsin concentration on reaction rate of the enzymatic assay.. . 66

Figure 3.2 Inhibitory effect of BMMY and BSM mediums on the trypsin catalyzed reaction..

..................................................................................................................................... 67

Figure 3.3 Standard curve between inhibition rate and aprotinin concentration.. ...... 68

Figure 3.4 Study of cell growth on different sorbitol/methanol mixtures.. ................. 70

Figure 3.5 Growth and lysis of P. pastoris cells cultured on medium containing different

carbons.. ...................................................................................................................... 72

Figure 3.6 Comparison of methanol and sorbitol/methanol mixed induction strategies at

bioreactor scale.. .......................................................................................................... 74

Figure 3.7 Comparison of the OUR (A) and CER (B) of methanol and sorbitol/methanol (1:1,

C-mol/C-mol) mixture induced fermentations.. .......................................................... 76

Figure 3.8 Concentration of residual carbons during methanol and mixed induction..77

Figure 3.9 Comparison of the cell viabilities in methanol and sorbitol/methanol (1:1, C-mol/C-

mol) mixed induction strategies.. ................................................................................ 78

Figure 3.10 Comparison of product yields of methanol and sorbitol/methanol (1:1, C-mol/C-

mol) mixture induced fermentations.. ......................................................................... 80

Figure 3.11 Analysis of soluble proteins in supernatant using 2D protein gel.. ......... 83

Figure 3.12 Localization of the HCPs identified from methanol and sorbitol/methanol (1:1, C-

mol/C-mol) mixed induction strategies.. ..................................................................... 85

Page 12: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

12

Figure 3.13 Distribution of the molecular weights (A) and isoelectric points (B) of HCPs

identified from methanol and sorbitol/methanol (1:1, C-mol/C-mol) mixed induction strategies.

..................................................................................................................................... 87

Figure 4.1 kLa measurement using dynamic method.. ................................................ 94

Figure 4.2 Fermentation profile (A) and growth curve (B) of the OLFB fermentation..96

Figure 4.3 Residual methanol concentration in the OLFB fermentations. .................. 97

Figure 4.4 Change of the cell viability in the OLFB fermentations. ........................... 98

Figure 4.5 Product yields in the OLFB fermentations.. ............................................ 100

Figure 4.6 Fermentation profile (A) and growth curve (B) of P. pastoris in the oxygen unlimited

condition.. .................................................................................................................. 103

Figure 4.7 Residual methanol concentration in the oxygen unlimited condition.. .... 104

Figure 4.8 Cell viabilities in the oxygen unlimited condition.. ................................. 105

Figure 4.9 Product yields in the oxygen unlimited condition.. ................................. 107

Figure 4.10 Activity of AOX enzyme at different concentrations of total protein.. . 110

Figure 4.11 Comparing the activities of AOX enzyme from different induction strategies.. 111

Figure 5.1 Cell growth, viability and product expression of the cell culture used in dewatering

study.. ........................................................................................................................ 119

Figure 5.2 Predicting the cell robustness to shear stress in feeding zoon of the large scale

centrifuges.. ............................................................................................................... 121

Figure 5.3 Cumulative cell size distributions in methanol and sorbitol/methanol (1:1 C-mol/C-

mol) mixed induction.. .............................................................................................. 123

Figure 5.4 Comparing the viscosities of cell cultures in methanol and sorbitol/methanol (1:1 C-

mol/C-mol) mixed induction. .................................................................................... 125

Figure 5.5 Dewatering levels of the cell cultures induced by methanol and sorbitol/methanol

(1:1 C-mol/C-mol) mixture as predicted by the scale-down model of CSA-1 and BTPX305 disc

stack centrifuges.. ...................................................................................................... 128

Figure 5.6 Clarification levels of the cell cultures induced by methanol and sorbitol/methanol

(1:1 C-mol/C-mol) mixture as predicted by a scale down model of CSA-1 and BTPX305 disc

stack centrifuges.. ...................................................................................................... 129

Page 13: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

13

Figure 5.7 Cumulative size distribution and dewatering of cell cultures in fermentations

performed in an OTR-limited bioreactor.. ................................................................ 131

Figure 8.1 Detection of glycerol, methanol and sorbitol using UltiMate 3000 HPLC with

Aminex HPX-87h column.. ....................................................................................... 158

Figure 8.2 Correlation between peak area and concentration of methanol (A) and sorbitol (B)..

................................................................................................................................... 159

Figure 8.3 Viscosities of methanol, 0.75 g•ml-1 sorbitol solution and sorbitol/methanol (1:1, C-

mol/C-mol) mixture at the shear rate of 800s-1. ........................................................ 160

Figure 8.4 The impact of biomass concentration and treating time on specific protein release.

................................................................................................................................... 160

Figure 8.5 Correlation between DCW and WCW of P. pastoris .............................. 161

Page 14: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

14

List of tables

Table 1.1 Comparing the features CHO, P. pastoris and E. coli protein expression system.

..................................................................................................................................... 25

Table 1.2 Comparison of the P. pastoris phenotypes. ................................................ 27

Table 1.3 A summary of the literatures applying sorbitol as a co-substrate in P. pastoris

induction. ..................................................................................................................... 39

Table 2.1 Components of the reaction mixture in aprotinin quantification. ............... 55

Table 2.2 Voltage and time used in the IPG electro focusing. .................................... 59

Table 2.3 Components of the reaction mixtures in quantification of AOX enzyme. .. 61

Table 3.1 Number of host cell proteins, peptides, proteases and stress related proteins identified

from the methanol and sorbitol/methanol (1:1, C-mol/C-mol) mixed induction strategies. 85

Table 4.1 A summary of the induction strategies used in the bioreactor with an OTR of 150

mmol•L−1•h−1 ............................................................................................................... 90

Table 4.2 Feeding rates of different induction solutions ........................................... 102

Table 5.1 Dimensions of centrifuges used in the dewatering study. ......................... 126

Table 6.1 A summary of the fermentation and centrifugal dewatering investigated in this thesis.

................................................................................................................................... 135

Table 8.1 A summary of names, accession number, molecular weight, isoelectric point,

localization and function of the HCPs identified from the cell cultures induced by methanol or

sorbitol/methanol (1:1, C-mol/C-mol) mixture. ........................................................ 162

Page 15: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

15

Nomenclature

Symbol Description Units

ATM Standard atmosphere kPa

C Correlation factor -

DCW Dry cell weight g

DCWf Dry cell weight after filtration g

dwr DCWf by WCWf -

FO2 Volumetric fraction of oxygen -

FN2 Volumetric fraction of nitrogen -

FCO2 Volumetric fraction of carbon dioxide -

kLa Oxygen transfer coefficient h-1

m/v Mass by volume g•L-1

n Disc numbers of centrifuge -

N Rotation speed s-1

pH Power of hydrogen -

qp Specific productivity mg•gDCW-1•h-1

Q Liquid flow rate L•h-1

Qg Gas flow rate L•h-1

rmet Carbon fraction of methanol %

R Gas constant J•mol−1•K−1

r Radius of centrifuge rotor m

Page 16: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

16

Temp Temperature �

V Volume L

v/v Volume by volume -

WCW Wet cell weight g

WCWf Wet cell weight after filtration g

ΔH Enthalpy of combustion kJ•mol−1

Page 17: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

17

Greek symbols

ε Energy dissipation rate W•Kg-1

θ half disc angle of centrifuge -

μ specific growth rate h-1

Σ settling area of centrifuge m2

ω angular velocity of centrifuge -

Page 18: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

18

Abbreviations and symbols

ABTS 2,2'-Azino-bis (3-ethylbenzthiazoline-6-sulfonic acid)

AOX Alcohol oxidase

ATP Adenosine triphosphate

ATPS Aqueous two-phase system

BAPNA Nα-Benzoyl-DL-Arginine-p-Nitroanilide

BCA Bicinchoninic acid

BMGY/BMMY Buffered glycerol/methanol complex medium

BPTI Bovine pancreatic trypsin inhibitor

BSA Bovine serum albumin

BSM Basal salts medium

CFD Computational fluid dynamics

CHO Chinese hamster ovary

CHAPS 3-[(3-Cholamidopropyl) dimethylammonio]-1-propanesulfonate hydrate

CNBr Cyanogen bromide

CQA Critical quality attribute

DCW Dry cell weight

DE-MF Dead end microfiltration

DO Dissolved oxygen

DTT Dithiothreitol

EBA Expanded bed absorption

Page 19: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

19

ELISA Enzyme-linked immunosorbent assay

EPX1 Extracellular protein X1

ER Endoplasmic reticulum

FDA Food and Drug Administration

HCHO Formaldehyde

HCl Hydrogen chloride

HCPs Host cell proteins

HPLC High-performance liquid chromatography

HRP Horseradish peroxidase

IPG Immobilized pH gradient

LC-MS/MS Liquid chromatography tandem-mass spectrometry

LDS Lithium dodecyl sulfate

MLFB Methanol limited fed-batch

MOPS 3-(N-morpholino) propanesulfonic acid)

MW Molecular weight

NaCl Sodium chloride

NAD+ Nicotinamide adenine dinucleotide

NADH Reduced NAD+

OD Optical density

OLFB Oxygen limited fed-batch

pAOX Promoter of alcohol oxidase

Page 20: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

20

pGAP Promoter of glyceraldehyde-3-phosphate dehydrogenase gene

pI Isoelectric point

PTMs Post–translational modifications

PTM1 Pichia trace metals

RCT Research Corporation Technologies

rHuEPO Recombinant human erythropoietin

rIFN-α Recombinant Interferon α

scFv Single chain antibody fragment

SCP Single cell protein

TCA Tricarboxylic acid

TE Tris-EDTA

TFF-MF Tangential flow microfiltration

USD Ultra scale-down

WCW Wet cell density

YNB Yeast nitrogen base

YPD Yeast extract peptone dextrose

Page 21: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

21

Chapter 1 Introduction

1.1 Trends in biopharmaceutical manufacturing

1.1.1 Overview of approved therapeutic proteins

The recombinant protein has become a major class of medicine since the idea of recombinant

DNA was proposed. In the early 1970s, scientists from Genentech successfully synthesized

recombinant human insulin in E. coli and demonstrated the feasibility of recombinant protein

production using recombinant DNA technology (Goldner, 1972). From then, nearly 380

recombinant therapeutic proteins have been developed and approved to treat various clinical

indications and another 1300 recombinant products were undergoing pre-clinical or clinical

studies by 2017 (Usmani et al., 2017). Therapeutic area of these products covers metabolic

disorders, haematological disorders and oncology, etc. As the incidence of cancer soared in

recent years, the number of therapeutic proteins with indications for oncology or hepatology is

expanding. Of the 62 therapeutic proteins approved by US. Food and Drug Administration

(FDA) from 2011 to 2017, 55% of these products are indicated for oncology or hepatology

(Lagassé et al., 2017).

Since recombinant human insulin was first produced, classes of the therapeutic proteins are also

expanding to benefit more therapeutic areas. In the 1980s, most of the approved recombinant

products, such as recombinant insulin, hormones, and interleukin, had relatively simple

structures and were recombinant version of native proteins from humans (Sanchez-Garcia et al.,

2016). With the advances in recombinant protein technology, recombinant proteins with

molecular weight (MW) over 100 kDa and complex post–translational modifications (PTMs)

are occupying a larger proportion of approved products in recent years (Sanchez-Garcia et al.,

2016). Meanwhile, new versions of recombinant protein with better efficacy or bi-functions are

developed using domain fusion technology (Levin et al., 2015, Strohl, 2015). Statistically, 48%

of the approved therapeutic proteins during 2011-2016 are monoclonal antibodies, whereas

coagulation factor, recombinant enzyme and fusion protein are the next three largest classes

(Fig 1.1).

Page 22: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

22

Figure 1.1 (A) Number of therapeutic proteins approved by FDA from 2011 to 2016. (B)

Classes of therapeutic proteins approved by FDA from 2011 to 2016 (Lagassé et al. 2017).

As their classes and therapeutic areas expand, recombinant proteins are becoming the dominant

player in pharmaceutical market. Four of the top five best-selling drugs in 2017, which

contributed 48 billion dollars in total, were therapeutic proteins (Philippidis, 2017). According

to a market forecast, the biopharmaceutical market in which therapeutic proteins play a major

role will expand at an annual growth rate of 11% and reach 209 $ billion to 480 $ billion by

2020 (Ecker et al., 2015). Monoclonal antibodies have a larger market share than the other

classes of therapeutic proteins. It is expected over 70 monoclonal antibody drugs will be

approved by 2020 and the annual sales will reach 125 $ billion (Ecker et al., 2015). Growth of

the market is stressing the importance of effective expression systems for therapeutic protein

production.

1.1.2 Recognised hosts for biopharmaceutical manufacturing

Development of host systems has been accelerated by the growth of biopharmaceutical market

and the advances in cell physiology. Over 100 host systems have been approved for therapeutic

protein production by 2011 and the species cover microorganism, plant cells, insect cells,

mammalian cells and transgenetic animals (Rader, 2018). In all of the host systems, bacterial

Escherichia coli (E. coli) and a mammalian cell line, Chinese hamster ovary (CHO) cells, are

most prevalently used in industry nowadays (Thomas Purkarthofer, 2017).

Since it produced the first recombinant protein, E. coli has been recognized as a reliable and

cost-effective approach for therapeutic protein production (Johnson, 1983). E. coli has a

Page 23: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

23

relatively small genome sequence and the sequence is readily edited by various molecular

biology tools (Sambrook et al., 1989). Thus, construction of the cell line is easily manipulated.

E. coli has a relative low requirement on nutrients and it can be easily cultured in minimal

medium with low cost. Scale-up of E. coli cultivation is straightforward and the fermentation

conducted over 200,000 litres has been reported (Alford, 2008). Nowadays, E. coli dominates

the production of ~30% recombinant products on the market and these products cover

recombinant insulin, interferons, hormone and antibody fragments (Overton, 2014). The major

drawback of E. coli system is that E. coli has limited ability of protein folding and lacks

endoplasmic reticulum and Golgi apparatus to perform PTMs. Overexpression of recombinant

proteins in E. coli often results in inclusion body and thus refolding is required in downstream.

Besides, E. coli lacks the capability to perform PTMs, such as glycosylation, which are critical

for the activity of some products. Eukaryotic expression systems, such as mammalian CHO

cells, are demanded in that situation.

In the current market, nearly 70% of the approved therapeutic proteins are produced by CHO

cells. CHO system is becoming very popular due to the following advantages: 1) ability to

perform human-like PTMs, 2) fewer endogenous secretory proteins, 3) robust cell growth in

chemical defined medium without serum, 4) relatively low susceptibility to human viruses, 5)

well-established history of the cell line. In 1987, FDA approved the first therapeutic protein

produced by CHO cells, recombinant human tissue plasminogen activator (Spellman et al.,

1989). Since then, CHO cells have been extensively used to produce protein drugs especially

monoclonal antibodies. As described in a review, over 50% of the products approved by FDA

after 2000 are produced by CHO cells (Lagassé et al., 2017). Kim JY and co-workers stated

that CHO cells would be more popular as a tool to produce therapeutic proteins in the near

future (Kim et al., 2012). However, the drawbacks of CHO cell system cannot be ignored. For

instance, cell line development of CHO cell, which usually takes over six months to obtain the

final clonal for production, is time-consuming (Lai et al., 2013). CHO cell cultivation requires

more complex medium than E. coli which increases the cost of goods in manufacture.

Meanwhile, it was reported that some products produced by CHO system had heterogeneous

product profiles (Yang and Butler, 2000). These drawbacks of CHO system call for other

eukaryotic hosts and Pichia pastoris (P. pastoris) is emerging as one of the most competitive

ones.

1.1.3 P. pastoris as an emerging host for biopharmaceutical

In the last few decades, P. pastoris has been well established as an effective host for

heterologous protein production, either in laboratory research or biopharmaceutical industry.

Page 24: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

24

By 2005, over 500 heterologous proteins were successfully expressed in P. pastoris (Thor et

al., 2005). According to the statistic from Research Corporation Technologies (RCT), over 70

recombinant proteins produced by P. pastoris have been approved for human use or in late

stage of clinical trials (Ahmad et al., 2014).

As shown in Table.1, P. pastoris combines the advantages of CHO and E. coli systems. Compared to E. coli, P. pastoris has the eukaryotic protein processing machinery and is capable

to perform proper protein folding (Ciarkowska and Jakubowska, 2013). Moreover, P. pastoris

has the capability to secrete recombinant protein into supernatant. In comparison,

homogenization of whole cells is usually required in the processing of E. coli, which generates

numerous host cell proteins (Ciarkowska and Jakubowska, 2013). Besides, P. pastoris

advantages E. coli system by providing glycosylation which is required by some proteins to

maintain the correct structure and biological efficacy (Higgins, 2001).

In contrast to CHO cell, P. pastoris has a relatively low nutrient requirement and cell culture is

easily implemented with cheap minimal medium. Doubling time of P. pastoris is much shorter

than that of CHO cell (2h versus 15h), thus P. pastoris exhibits higher growth rate in

fermentation and offers a faster speed platform for recombinant protein production (Maccani et

al., 2014). Meanwhile, P. pastoris is more robust to hydrodynamic shear and cell philosophy is

less influenced by culture condition, whereas shear stress and heterogeneity of environment

often impose great hurdles in scale-up of CHO cell fermentation (Senger and Karim, 2003).

However, single P. pastoris cell has lower productivity than CHO cell (Maccani et al., 2014),

and thus high density fermentation is required to achieve high titre. Glycosylation formed by P. pastoris is rich in mannose and genetic engineering of glycosylation pathway is required to

produce human like PTMs (Kalidas et al., 2001).

Page 25: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

25

CHO P. pastoris E. coli References

Nutrient requirement High Low Low (Kunert and Reinhart, 2016)

Time of cell line

development 6-12 month ~two weeks 2-3 days

(Lai et al., 2013) (Maccani et al.,

2014)

Protein folding capability Yes Yes Very

limited (Thomas Purkarthofer, 2017)

Secretion capability 2-3 mg•g-1 DCW•h-1 <0.01 mg•g-1 DCW•h-1 None (Maccani et al., 2014)

Glycosylation pattern Human-like High mannose None (Kalidas et al., 2001)

Product yield 1-10 g•L-1 1-20 g•L-1 < 1 g•L-1 (Thomas Purkarthofer, 2017, Pybus et

al., 2014)

Cost of Goods High Low Low (Maccani et al., 2014)

Viral risk High Low Low (Thomas Purkarthofer, 2017)

Table 1.1 Comparing the features CHO, P. pastoris and E. coli protein expression system.

Page 26: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

26

1.2 Introduction to P. pastoris expression system

1.2.1 History of P. pastoris expression system

The cultivation of P. pastoris was initially studied by the Phillips Petroleum Company in the 1970s. The company attempted to use P. pastoris to produce single cell protein (SCP), a substitute for protein rich foods. The attempt was abandoned soon because the oil crisis increased the cost of methanol, a major substrate for P. pastoris cultivation. In the following years, Phillips Petroleum transformed P. pastoris into an efficient system for recombinant protein expression by collaborating with the Salk Institute

Biotechnology/Industrial Associates Inc. (SIBIA) (Macauley�Patrick et al., 2005).

Alcohol oxidase (AOX) gene was cloned from P. pastoris and its promoter was used to drive recombinant protein expression. Expression vectors, strains, and transformation protocols were also developed by SIBIA. Research Corporation Technologies (RCT) patented the P. pastoris expression system in 1993, while most of the commercial P. pastoris expression system kits are supplied by the Invitrogen Corporation (Carlsbad, CA, USA).

1.2.2 Phenotype of P. pastoris strains

In P. pastoris, AOX is the first enzyme in methanol metabolism pathway by which methanol is oxidised to formaldehyde (Harder and Veenhuis, 1989). Based on the genes encoding AOX enzyme, P. pastoris can be classified into three phenotypes, Mut+, Muts, and Mut- (Cregg et al., 1998). In the phenotype of Mut+, AOX enzyme is encoded by two genes, AOX1 and AOX2. AOX1 promoter is much stronger than that of AOX2. Therefore, nearly 85% of AOX enzyme is regulated by AOX1 gene while only 15% by AOX2 gene (Ellis et al., 1985). In phenotype of Muts, AOX1 gene is disrupted and all of the AOX enzyme is regulated by AOX2 gene. Therefore, methanol utilization is much slower in the phenotype of Muts than Mut+. In the phenotype of Mut-, both genes are deleted, and cells cannot use methanol as the carbon source.

Among the three phenotypes, Mut+ is most extensively used as a host of recombinant protein production. Mut+ cell has high cell growth rate and methanol consumption rate in a methanol sufficient condition, reaching 0.08 h-1 and 0.0682 g•gDCW−1•h−1 (Zhang et al., 2000), respectively. High oxygen demand is required in Mut+ P. pastoris fermentation especially at industrial scale since methanol is a high degree reductant (Schilling et al., 2001, Jenzsch et al., 2004). Muts is has slower methanol consumption

Page 27: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

27

rate and lower oxygen demand which makes process control easier at large scale cultivation (Cos et al., 2005). Muts has slower growth rate compared to Mut+ (0.015 h-

1 vs 0.043 h-1) which may delay the production process. But its productivity on methanol was even higher (18.2 vs 9.3 mgproduct•gmethanol−1) (Pla et al., 2006). Scarce reports could be found where Mut- was used as the host of recombinant protein production.

Table 1.2 Comparison of the P. pastoris phenotypes.

1.2.3 Promoters used in P. pastoris system

The promoters used in P. pastoris expression system can be classified into two categories: inducible promoters and constitutive promoters, represented by the promoter of AOX1 (pAOX1) and promoter of glyceraldehyde-3-phosphate dehydrogenase gene (pGAP), respectively.

pAOX1 is the most extensively used promoter and is one of the major driving forces that make P. pastoris expression system very popular (Potvin et al., 2012). As mentioned above, AOX1 gene encodes the majority of AOX enzyme. Due to the low affinity between AOX enzyme with oxygen, pAOX1 activity is up-regulated to produce more AOX enzyme when P. pastoris is cultured on solo methanol (Cregg et al., 2000). Amount of AOX can reach up to 30% of total protein. Therefore, pAOX1 can effectively prompt transcription of foreign genes when P. pastoris is cultured with methanol. Besides, inducible pAOX1 provides tight control on protein expression. Transcription of foreign gene is repressed during cell growth on glycerol or glucose, and production is switched on by changing carbon to methanol. It is especially helpful for process control when products are toxic to cells (Liu et al., 2005).

Methanol induction limits the usage of pAOX1 especially at industrial scale (Liu et al., 2005). Methanol is a high degree reductant and the oxidation of one mole methanol consumes 0.8~1.1 mole oxygen (Cunha et al., 2004, Jahic et al., 2003). Therefore,

Phenotype Encoding gene Promoter Inducer

Mut+ AOX1 AOX2 pAOX1 Methanol

Muts AOX2 pAOX2 Methanol

Mut- NA - -

Page 28: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

28

oxygen transfer and heat removal often constrain the feeding rate of methanol and cell density in large scale bioreactors (Potgieter et al., 2010b). Meanwhile, methanol is a flammable and hazardous liquid, and storing a large volume of methanol is a major concern to safety. Therefore, other inducible promoters, such as pAOX2, are also used as an alternative to diminish methanol consumption. Although the activity of pAOX2 is much weaker than that of pAOX1, successful recombinant protein expression controlled by pAOX2 has been reported (Mochizuki et al., 2001).

Using constitutive promoters, such as pGAP, can eliminate consumption of methanol. GAP is an enzyme in glycolysis and its promoter has been used to drive recombinant protein expression in P. pastoris without using methanol (Chen et al., 2007, Rupa et al., 2007). Protein expression controlled by pGAP was found to be affected by the carbon source. In a strain expressing recombinant β-lactamase, cells grew on glucose had higher product yield than that grew on glycerol (Waterham et al., 1997). Other constitutive promoters, such as promoter of translation elongation factor 1 (pTEF1), promotor of high affinity glucose transporter (pHGT1), were also reported (Prielhofer et al., 2013, Ahn et al., 2007).

1.2.4 Cell engineering of P. pastoris strains

Cell engineering, mainly glyco-engineering, has been used to improve recombinant protein production in P. pastoris. Glycosylation formed by wide type P. pastoris is rich in mannose. The glycan pattern has high immunogenicity to human and also reduces the in vivo activity of monoclonal antibodies (Bretthauer and Castellino, 1999). Therefore, genetic engineering has been used to reform the glycosylation pathway of P. pastoris to produce human-like glycoproteins. Manipulation of the genetic engineering mainly includes knocking out the enzymes introducing hypermannosylation, eliminating the enzymes for fungal glycosylation and introducing the enzymes for sialylation (Zha, 2012).

The monoclonal antibodies were successfully produced by the glyco-engineered P. pastoris strain with a product yield over 1 g•L-1 (Potgieter et al., 2009). The antibody produced by the glyco-engineered P. pastoris strain had more uniform glycan pattern compared to that produced by mammalian cells. Product yield of the antibody was enhanced to 1.6 g•L-1 by optimizing cultivation condition of the P. pastoris strain (Ye et al., 2011). In another study, specific productivity of antibody was found to be apparently affected by the specific growth rate of the strain (Potgieter et al., 2010b).

Page 29: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

29

Product yield of the antibody could be further enhanced by studying its expression kinetics and maintain cell growth at the optimal rate for product formation.

1.3 Fed-batch fermentation of P. pastoris

1.3.1 Cell culture medium of P. pastoris

In most of the publications regarding P. pastoris cultivation, yeast extract peptone dextrose medium (YPD), buffered glycerol/methanol complex medium (BMGY/BMMY) or basal salts medium (BSM) are used. The YPD medium and BMGY/BMMY medium contain all the amino acids necessary for yeast growth and provide easily utilized nutrients for rapid cell growth (Hanko and Rohrer, 2004). However, complex components in both mediums, such as yeast extract and peptone, not only increase capital cost but cause batch to batch variations, which is not preferred by regulatory agencies (Van der Valk et al., 2010). Therefore, YPD and BMGY/BMMY mediums are mainly used in laboratory research, whereas BSM medium is more popular at industrial scale fermentation.

Formulation of the BSM medium was shown in the Section 2.2.2, Chapter 2. In the medium, the nitrogen source is provided by adding ammonium hydroxide during cultivation and carbon sources are from the feeding of glycerol, methanol or sorbitol. Trace elements are normally complemented to the medium before inoculation to improve cells growth and product synthesis (Hélène et al., 2001). BSM medium is cheaper and has more consistent quality compared to YPD and BMGY/BMMY medium. However, it was also reported that cells cultured on BSM medium had to synthesize all metabolic intermediates, which reduced cell growth rate (Matthews et al., 2018). The most critical amino acids for cell growth was found out by screening the individual amino acids in complex medium. By supplementing these critical amino acids in BSM medium, maximum growth rate, biomass and product yield were enhanced.

1.3.2 Fed-batch fermentation of P. pastoris

Currently, fed-batch fermentation is the most applicable technique of P. pastoris cultivation in bioreactors. It is defined in bioprocess that one or more nutrients are fed into bioreactor during fermentation, and biomass or product are not harvested until the end of fermentation (Yamanè and Shimizu, 1984). Fed-batch fermentation is preferred

Page 30: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

30

over batch fermentation when cell growth is limited due to the lack of some nutrients. It is thought that fed-batch fermentation is easier to be manipulated and consumes less medium than continuous culture (Lim and Shin, 2013). Therefore, fed-batch fermentation is used in most reports about P. pastoris culture.

Figure 1.2 Diagram of carbon feeding (A) and cell growth (B) in P. pastoris fed-batch fermentation. The fed-batch protocol was recommended by Invitrogen (Invitrogen 2002a). Induction is indicated by the symbol of arrow where glycerol feeding is switched to methanol feeding.

Conventional fed-batch fermentation of P. pastoris contains three phases: a batch phase for cell growth on glycerol, a fed-batch phase for further cell growth on glycerol and a fed-batch phase for product synthesis using methanol. In batch phase, cells are cultured in medium with glycerol. A maximum specific growth rate (μmax) of 0.06 h-1 has been reported after cells adapted to the new environment (Cos et al., 2005). 40 g•L-1 glycerol is recommended to be used and further higher glycerol concentration may inhibit cell growth rate (Cos et al., 2006). After batch glycerol is consumed, fed-batch phase is started by feeding glycerol at a constant or decreasing rate. The specific growth rate in glycerol fed-batch phase is maintained to be lower than that in batch phase to avoid accumulation of toxic by-products. Finally, product synthesis is induced by feeding methanol. In the first few hours of induction, methanol feeding rate is increased stepwise to help cells adapt to methanol metabolism. Then it is maintained at constant until the end of fermentation. Feeding rate of methanol varied depending on strains in previous reports (Stratton et al., 1998, Tolner et al., 2006, Invitrogen, 2002a). Due to

Page 31: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

31

the decreased amount of methanol available per biomass, specific growth rate declines over time during constant feeding of methanol. When methanol is fed at the rate recommended by Invitrogen, μmax is as high as 0.05 h-1 in the beginning and then decrease to < 0.01 h-1 (Invitrogen, 2002a). Exponential feeding can be used to maintain cell growth at a constant rate which is optimal product formation (Potgieter et al., 2010b, Schenk et al., 2008, Heyland et al., 2011). While exponential feeding improves product yield, it requires more sophisticated control of feeding in fermentation, and it consumes much time to find the best cell growth rate.

1.3.3 Oxygen unlimited fed-batch fermentation

Oxygen unlimited fed-batch fermentation (OULFB) or methanol limited fed-batch fermentation (MLFB) is a control strategy in which methanol feeding rate is constrained to maintain dissolved oxygen (DO) at an optimal level, normally 20% or 30% of saturation (Singh et al., 2008, Invitrogen, 2002a). OULFB is the most prevailing control strategy in P. pastoris fermentation and it has been scaled up to demonstrated and industrial scales (Liu et al., 2016).

Dissolved oxygen was found to affect the expression of foreign proteins (Cregg et al., 2000, Lee et al., 2003a). Therefore, it is critical to develop optimal methanol feeding strategies and proper DO control in OULFB. Control of DO is commonly achieved through coupling DO values with agitation speed and oxygen fraction in aeration (Goodrick et al., 2001, D'anjou and Daugulis, 2001). Dissolved oxygen in the medium is detected by a sensor and agitation speed gradually raises when dissolved oxygen dropped below the set point. Pure oxygen is mixed into aeration once dissolved oxygen cannot be maintained by the agitation. Besides, Lim HK and co-workers proposed a novel strategy to control dissolved oxygen by coupling it with both methanol feeding rate and oxygen fraction in aeration (Lim et al., 2003). Dissolved oxygen was set at 40%~45% and methanol feeding rate was decreased when DO dropped below the setting value through a feedback control, and vice versa. Compared to the traditional control strategy, the novel DO control strategy improved cell growth and increased volumetric yield of recombinant Guamerin by 40%.

OULFB is not always reliable although it has been successfully implemented in many studies (Lauer et al., 2005). For instance, methanol accumulation inhibits cell growth and viability, and as a result, DO will increase shapely in the circumstance. Rising DO will reduce oxygen supply and cause further methanol accumulation. Moreover,

Page 32: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

32

complete exhaustion of methanol often happens in OULFB which is not desirable to maintain pAOX1 activity. It was reported that pAOX1 activity decreased to nearly zero at the end of OULFB fermentation.

1.3.4 Oxygen limited fed-batch fermentation

Although it is preferred to constrain methanol feeding and avoid oxygen limitation in P. pastoris induction, the fermentation with over-fed methanol and depleted oxygen, which is known as oxygen limited fed-batch (OLFB), were also successfully developed (Khatri and Hoffmann, 2006, Charoenrat et al., 2005, Barrigón et al., 2013). OLFB is a fermentation strategy where dissolved oxygen is depleted and methanol is accumulated in the medium.

OLFB was also reported to increase product yield. In one study, while biomass concentration was almost the same in OULFB and OLFB, volumetric product yield of recombinant β-glucosidase in OLFB was enhanced by 16% and percentage of product in total soluble proteins was improved by 64% (Charoenrat et al., 2005). Khatri NK and co-workers further investigated the impact of methanol concentration on product yield in OLFB fermentation (Khatri and Hoffmann, 2006). It was found that methanol concentration apparently affected the product yield of recombinant single chain antibody fragment (scFv). When methanol concentration increased from 0.3% to 3% (v/v), volumetric product yield of scFv increased from 60 mg•L-1 to 350 mg•L-1. However, it seems that methanol concentrations affected cell growth and product yield depending on cell strains. In another OLFB study, both cell growth and formation of

recombinant Rhizopus oryzae lipase were inhibited by 10 g�L-1 of methanol (Barrigón

et al., 2013). Compared to OULFB, both volumetric and specific productivities were enhanced in OLFB when methanol concentration was in the range of 3~10 g•L-1.

Methanol control strategy is critical in OLFB fermentation. Constant methanol concentration was achieved by a feedback control system (Khatri and Hoffmann, 2006). In the system, methanol sensor based on photoelectric plethysmograph was used to detect methanol concentration in the medium. Methanol addition was activated for a dosage time when methanol concentration dropped below the set point. When on-line monitoring of methanol is not available, methanol can be quantified using off-line methods, like high-performance liquid chromatography (HPLC), gas chromatography or enzymatic reaction-based method (Minning et al., 2001, Parpinello and Versari, 2000, Kučera and Sedláček, 2017).

Page 33: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

33

1.4 Mixed induction strategies of P. pastoris

1.4.1 Metabolism of methanol in P. pastoris

In the first step of methanol metabolism, methanol is oxidized to formaldehyde (HCHO) by AOX enzyme in peroxisome (Eq1.1). Half mole of methanol is consumed and no adenosine triphosphate (ATP) is produced in the initial oxidation. The further formaldehyde flux is divided into two pathways: formaldehyde dissimilatory (energizing) pathway and biomass synthesis pathway (Jahic et al., 2002, Gao et al., 2012). In the formaldehyde dissimilatory pathway, formaldehyde is oxidized to CO2 by NAD+ (nicotinamide adenine dinucleotide) and two mole NADH (reduced NAD+) is formed by oxidizing one mole HCHO (Eq1.2). NADH is further oxidized to NAD+ by O2 and ATP is also produced in the reaction (Eq1.3). In total, one mole formaldehyde consumes one mole O2 and produces two mole ATP in the formaldehyde dissimilatory pathway. In the biomass synthesis pathway, neither oxygen uptake nor ATP production is significant.

CH#OH + 0.5O) = HCHO + H)O Eq1.1

HCHO + 2NAD/ = 2NADH + CO) Eq1.2

O) + 2NADH + 2ADP + 2P1 = 2ATP + 2NAD/ + 2H)O Eq1.3

Flux distribution of formaldehyde varies in different culture conditions. In one study, 57% and 43% of carbon flux were divided into energizing and biomass synthesis pathway, respectively, when a P. pastoris strain producing recombinant Interferon α

(rIFN-α) was cultured at 30� (Gao et al., 2012). When temperature decreased to 20�,

flux distribution in energizing pathway decreased to 35% while distribution in biomass synthesis pathway increased to 65%. Volumetric product yield of IFN-α was significantly enhanced due to the increasing flux distribution in biomass synthesis pathway. Meanwhile, flux distribution also varied on specific growth rates of P.

pastoris cells (Jahic et al., 2002). At μ=0.033 h-1, about 33% of carbon flux was distributed to the biomass synthesis pathway, while it decreased to 20% when μ is maintained at 0.005 h-1. As a result, the oxygen uptake rate increased significantly at the slower growth rate. These studies indicate that flux distribution analysis is an effective approach to help determine culture conditions of P. pastoris.

Page 34: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

34

1.4.2 Metabolism of mixed carbons in P. pastoris

Glycerol and sorbitol are utilized by a different pathway from methanol. In P. pastoris, glycerol or sorbitol is initially converted to glyceraldehyde 3-phosphate (GAP) in glycolysis. GAP is further divided into biomass synthesis and tricarboxylic acid (TCA) cycle (Niu et al., 2013). In the TCA cycle, one mole carbon is oxidized by two mole NAD+ and produces two mole NADH and 2/3mol ATP (Eq1.4). Afterwards, two mole NADH is oxidized by one mole O2 and produces two mole ATP (Eq1.5). Metabolism of one mole carbon by the TCA cycle consumes one mole oxygen and produces 2.66mol ATP.

C + 2NAD/ + 23ADP +23P1 = 2NADH + CO) +

23ATP Eq1.4

O) + 2NADH + 2ADP + 2P1 = 2ATP + 2NAD/ + 2H)O Eq1.5

Figure 1.3 Metabolic pathways of methanol and sorbitol when P. pastoris was induced by pure methanol or methanol/sorbitol mixture (Gao et al. 2012).

Adding sorbitol or glycerol as a co-substrate causes shift of methanol flux distribution. In one study, a P. pastoris strain producing recombinant β-galactosidase was cultured on sorbitol/methanol mixture with different ratios in a transient continuous culture (Niu et al., 2013). It was found that methanol flux in energizing pathway decreased after the

Page 35: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

35

addition of sorbitol. Energy production was complemented by the increase of sorbitol flux in TCA cycle. Similar shift of methanol flux was observed when sorbitol/methanol mixed induction was used in the fed-batch fermentation (Gao et al., 2012). Methanol flux in energizing pathway reduced by half while flux in biomass synthesis enhanced by 63% compared to culture on sole methanol. A shift of methanol flux distribution was also observed in a strain producing recombinant human erythropoietin (rHuEPO), where the synthetic flux of rHuEPO was enhanced 4.4-fold after sorbitol addition (Çelik et al., 2010). The shift of methanol flux was likely to enhance product yield as reported by several studies (Wang et al., 2010, Çelik et al., 2009).

1.4.3 Glycerol/methanol mixed induction

As a high degree reductant, methanol has a high enthalpy of combustion (−727 kJ•C-mol−1). Considerable heat is generated when P. pastoris is grown on methanol. Glycerol has a lower enthalpy of combustion (−549.5 kJ•C-mol−1) than methanol and using glycerol as co-substrate reduces the heat generation in P. pastoris induction. Since heat removal is one of the major challenges at large scale, glycerol/methanol mixed induction has been developed to increase the scalability of cell culture (Jungo et al., 2007). The impact of mixed induction on cell growth, product expression and oxygen uptake were studied (Zhang et al., 2003, Berrios et al., 2017, Canales et al., 2015).

In the fermentation of a P. pastoris strain expressing recombinant CD40 ligand, glycerol/methanol (1:1, C-mol/C-mol) mixed induction increased cell growth and volumetric product yield by about one fold (McGrew et al., 1997). Product yield of recombinant β-glycoprotein I domain V was also significantly improved when glycerol was used as a co-substrate (Katakura et al., 1998). Transient continuous culture was used to determine the optimal ratio of glycerol and methanol (Jungo et al., 2007). It found that both volumetric and specific productivity kept stable when methanol fraction was kept between 60% and 100% (C-mol/C-mol), while the specific productivity declined when methanol fraction dropped below 60% (C-mol/C-mol). Methanol consumption was decreased by 40% without impairing the product yield.

The major drawback of glycerol/methanol mixed induction is that glycerol has an inhibitory effect on pAOX1. Glycerol addition may inhibit product expression during methanol induction. In one study, product expression was totally inhibited when glycerol feeding rate was higher than 6 mg•g WCW−1•h−1 (Hellwig et al., 2001). Although the product was expressed at a slower glycerol feeding, its volumetric product

Page 36: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

36

yield decreased by half compared to solo methanol induction. Due to the inhibitory effect of glycerol, more attempts are seeking to use other non-inhibitory carbons such as sorbitol in mixed induction of P. pastoris.

1.4.4 Sorbitol/methanol mixed induction

Enthalpy of combustion of sorbitol is relatively lower than that of methanol (504.3 kJ•C-mol−1 versus −727 kJ•C-mol−1). Besides, sorbitol does not inhibit the activity of pAOX1 at the concentration as high as 50 g•L−1 (Çelik et al., 2009). Therefore, it has been widely used as a co-substrate of P. pastoris to reduce the drawbacks of methanol induction, such as high oxygen consumption and heat removal.

Strategies of sorbitol/methanol mixed induction can be categorized into batch-wise sorbitol addition and constant sorbitol feeding. Celik E and co-workers showed that sorbitol did not repressed pAOX1 activity with a concentration lower than 50g•L−1

(Çelik et al., 2009). Therefore, sorbitol/methanol mixed induction is easily implemented by batch-wise sorbitol addition. It was observed that adding sorbitol accelerated cell growth and enhanced volumetric product yield by 1.8 folds after 18 h induction compared to sole methanol induction. Constant sorbitol feeding was also used in several studies. In a strain expressing recombinant porcine interferon- α, it was shown that both sorbitol co-feeding and low temperature improved product yield (Gao et al., 2012). Compared to standard methanol induction, product yield was enhanced 10 to 200 folds by using sorbitol/methanol mixed induction. Transient continuous culture was used to determine the optimal ratio of methanol and sorbitol. In a continuous culture where methanol fraction was varied from 0 to 100% (C-mol/C-mol), it was found that comparable volumetric product yield was obtained by maintaining methanol fraction in the range of 45%~100% (C-mol/C-mol) (Niu et al., 2013). When the sorbitol/methanol (45:55, C-mol/C-mol) mixed induction was applied to a fed-batch fermentation, same amount of product was obtained while oxygen uptake was significantly reduced. Besides, it was also reported that sorbitol co-feeding reduced cell mortality from 23.1% to 8.8% in a strain expressing recombinant interferon-α compared to methanol induction (Niu et al., 2013). As a result, product degradation by co-released proteases was significantly reduced.

Compared to glycerol/methanol mixed induction, the major disadvantage of sorbitol/methanol mixed induction is that P. pastoris has a lower sorbitol uptake rate (Valero, 2013), which delays cell growth during induction. Besides, product yield is

Page 37: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

37

influenced by sorbitol/methanol mixed induction depending on strains. The product yield was even reduced by the mixed induction in several reports (Woodhouse, 2016, Niu et al., 2013).

Page 38: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

38

Mixed induction method Product Impact References

Exponential feeding Rhyzopus oryzae lipase Reduced product yield at 30� (Berrios et al., 2017)

Linear feeding and lower

temperature (20�) Interferon-α

1) Enhanced product yield by 1.3 folds

2) Enhanced product activity by 2.1 folds (Gao et al., 2015)

Automatically regulated

methanol and sorbitol feeding

Porcine circovirus cap

protein

1) DO stayed at stable

2) Product yield was increased by 64% (Ding et al., 2014)

Linear feeding Thermomyces lanuginosus lipase

Increased product yield by 1.4 folds (Fang et al., 2014)

Batch-wise sorbitol addition β-glucosidase Enhanced product yield by 1.4 folds (Batra et al., 2014)

Transient continuous cultures β-galactosidase 1) Comparable product yield

2) O2 uptake was reduced by 30% (Niu et al., 2013)

Linear feeding and lower

temperature (26�) β-mannanases Increased product yield by 1.5 folds (Zhu et al., 2011)

Page 39: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

39

Linear constant feeding Interferon-α 1) Enhanced product yield by 1.85-fold

2) Decreased mortality from 23.1% to 8.8% (Wang et al., 2010)

Batch-wise sorbitol addition Growth hormone Maximum product yield at μ = 0.03 h−1 (Çalık et al., 2010)

Batch-wise sorbitol addition Erythropoietin

1) Accelerated cell growth

2) Enhanced product yield by 1.8 folds at

t=18 h

3) Reduced protease by 1.2 folds

4) Eliminated lactic acid accumulation

5) Decreased oxygen uptake by 2 folds

(Çelik et al., 2009)

Batch-wise sorbitol addition Rhizopus oryzae lipase Enhanced product yield by 1.7 folds (Ramón et al., 2007)

Table 1.3 A summary of the literatures applying sorbitol as a co-substrate in P. pastoris induction.

Page 40: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

40

1.5 Impurities of host cell protein in P. pastoris processing

1.5.1 Control of host cell proteins in biopharmaceutical

Once a recombinant product is produced by host cells, it requires to remove the impurities such

as host cell proteins (HCPs), DNA and lipids. HCPs, a complex mixture of proteins, are one of

major of these impurities that must be strictly controlled. These proteins are encoded by gene-

modified host cells and produced together with recombinant products during cell culture. They

are usually related to normal functions of cells such as growth, proliferation, and apoptosis

(Wang et al., 2015).

HCPs are introduced into the product through different ways during cell culture and product

harvest. Some HCPs are co-secreted into the medium with products in cell culture. Cell lysis

also happens especially in late stage of cell culture. As a result, intracellular proteins will be

released into the medium. When shear sensitive cells, such as mammalian cells, are used as an

expression host, stir agitation and bubble breakage also cause cell lysis and HCPs release (Walls

et al., 2017). During harvest using centrifuges, shear stress of fluid also damages cells and

introduces intracellular proteins into the culture medium (Tait et al., 2012). When the

recombinant product is expressed intracellularly, total break of host cells is required to harvest

products, which results in the release of endogenous proteins.

HCPs level is one of the critical quality attributes (CQA) that is required to be strictly controlled

during biopharmaceutical manufacturing. HCPs are usually removed from products based on

their different properties such as affinity to resins, protein charge, and hydrophobicity

(Bracewell et al., 2015). However, it was also observed that some HCPs and recombinant

products have similar properties or strong interaction with each other and therefore these HCPs

may co-eluted with the products during purification (Levy et al., 2014, Bracewell et al., 2015).

If HCPs are not thoroughly cleared, they may cause potential safety risk even at extremely low

levels in the final product by stimulating immune responses of patients (Doneanu et al., 2012,

Huang et al., 2009). Besides, some HCPs have proteolytic activity and they will degrade the

recombinant product and affect its stability (Wang et al., 2015). According to the regulations

of FDA and European Commission, HCPs level in final product must be controlled less than

100 parts per million (Liu et al., 2012). In order to minimize the risk, it is necessary to reduce

the release of HCPs in cell culture and monitor the clearance of HCPs in downstream.

Page 41: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

41

1.5.2 Analytical methods of host cell proteins

Several methods have been developed to visualize, quantify and identify HCPs produced in

biological manufacture. This section will introduce the theory, advantages, and drawbacks of

the methods used in HCPs detection.

Protein electrophoresis is the most commonly used approach to separate and visualize HCPs.

HCPs are separated by a polyacrylamide gel based on their MW, PI or both. And HCP spots on

gel can be visualized by protein staining after the separation. Coomassie Blue, silver staining,

and Sypro Ruby staining are the most commonly used staining reagents (Olson et al., 2007).

0.05-0.1 μg protein per spot can be detected by coomassie blue while silver and Sypro Ruby

staining improve the detection limit to 1-5 ng/spot. Detecting HCPs using protein

electrophoresis is cost-effective and easily implemented. Spots of interest can be spliced from

gel and further analysed by mass spectrometry. The major drawback of this method is that low

abundant proteins are easily overlaid by overexpressed proteins such as the recombinant

product (Wang et al., 2015).

When it comes to quantitatively measure HCPs, enzyme-linked immunosorbent assay (ELISA)

is the standard method used in industry (Wang et al., 2015). In ELISA, anti-HCP antibodies are

firstly embedded on the bottom of microplate and afterwards, HCPs are added and bound. Anti-

HCP antibodies labelled by horseradish peroxidase (HRP) are added and free antibodies are

washed away. Substrate of HRP is finally added and causes a colour change. ELISA provides

a quantitative assay of HCPs with high throughout. However, the accuracy is highly impacted

by the HCPs used to generate anti-HCP antibodies (Wang et al., 2009). When new HCPs are

produced due to process change or fail, they will not be detected by ELISA assay. Besides, the

assay can only measure the total population of HCP, but some high immunogenic HCPs cannot

be separately determined.

Mass spectrometry (MS) based method makes identification of HCPs achievable (Doneanu et

al., 2012). HCPs are firstly separated and stained on protein gel, and each protein spot is spliced

and digested into peptides. Peptides are then separated and analysed by HPLC-MS. In addition,

proteomic method is also developed where peptides from digestion of HCPs mixture are

directly analysed by LC-MS/MS. HCPs can be identified by searching algorithm of peptides

against protein data bank. Although it requires expensive equipment, MS advantages protein

electrophoresis and ELISA by detecting low abundant and high immunogenic HCPs. HCPs

profile from cell culture to protein purification can also be tracked (Zhang et al., 2014).

Page 42: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

42

1.5.3 Host cell proteins in P. pastoris culture

HCPs profile is closely related to the expression system and manner that product is produced

(Wang et al., 2009). Compared to CHO and E. coli systems, P. pastoris is featured by secreted

expression of products with low HCPs impurities. P. pastoris has ~7900 genes totally whereas

the gene number is ~30000 in CHO cells (Gibbs et al., 2004, Mattanovich et al., 2009). As a

result, the composition of HCPs in P. pastoris culture is much simpler than that in CHO cell

culture. Meanwhile, recombinant proteins can be secreted into supernatant by P. pastoris cells

and downstream processing does not cause further release of intracellular HCPs. However,

product from E. coli system often form inclusion body and homogenization of whole E. coli in

product recovery releases bulk of intracellular HCPs.

Generally, HCPs produced by P. pastoris can be separated into native secretome and released

intracellular proteins. P. pastoris only secretes a few native proteins into cell culture

(Mattanovich et al., 2009). In a P. pastoris secretome study, it was found that only 20 secreted

proteins were detectable when P. pastoris was grown on glucose. Major of these proteins had

theoretical pI values ranging from 4.0 to 6.2 and MW values less than 70 kDa. Despite the

simple secretome, high concentration of HCPs may be produced due to the high cell density.

As it was estimated, around one gram per litre of HCPs was secreted into the culture medium

in a fed-batch fermentation with 100g•L−1 final dry biomass (Heiss et al., 2013). A 65 kDa

protein, named extracellular protein X 1 (EPX1), was identified as the major secreted HCPs.

Secretion of EPX1 was affected by the fermentation condition, such as temperature, medium

osmolality and substrates. The authors also demonstrated the feasibility of enhancing product

purity by deleting the gene encoding EPX1.

Besides, intracellular proteins also release into the culture medium and contaminate the product

during fermentation especially in late stage of fermentation. It has been shown that methanol

induction caused cell lysis and intracellular protein release (Cregg et al., 2000). In a proteomic

study, 110 proteins were identified from the culture medium after a P. pastoris strain expressing

recombinant vaccine was induced by methanol for 24 h (Huang et al., 2011). While the majority

of the identified HCPs were cell wall related proteins, some intracellular proteins such as

alcohol oxidase were also detected. Besides, it was also found that growing cells at a lower

temperature (25) reduced concentration of co-released HCPs.

1.5.4 Product degradation in P. pastoris culture

Recombinant protein production by P. pastoris is challenged by proteolytic degradation

(Kobayashi et al., 2000, van den Hazel et al., 1996). While high density fermentation improves

Page 43: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

43

expression of the desired product, it also increases the concentration of host cell proteases in

the culture medium. P. pastoris cells are stressed by starvation, toxic metabolites of methanol

and formation of products. Considerable proteases are released to culture medium due to high

cell concentration. Product degradation has been reported in several studies (Sinha et al., 2005,

Zhang et al., 2007). In the fermentation of a strain producing recombinant ovine interferon-τ,

proteases were undetectable during cell growth on glycerol but increased significantly after 48

h of methanol induction (Sinha et al., 2005). As a result, proteolytic degradation pattern of the

product was detected in the end of fermentation.

The protease can be divided into three catalogues: cytosolic proteases, vacuolar proteases and

proteases in secretory pathway (Jones, 1991, Zhang et al., 2007). Cytosolic proteases play a

pivotal in cell response to stress and are responsible for degradation of short life and detrimental

proteins (Hilt and Wolf, 1992, Zhang et al., 2007, Jones, 1991). Vacuolar proteases are mainly

involved in cellular proteolysis in yeast especially during cell starvation. Several different

vacuolar proteases have been identified including endoproteinases, carboxypeptidases,

aminopeptidases and dipeptidyl aminopeptidase B (Zhang et al., 2007). Vacuolar proteases

were often identified in culture medium and played a critical role in product degradation.

Finally, proteases in secretory pathway locate at Golgi complexes and plasmatic membrane. In

nature, they function in cutting signal peptide and maturation of protein precursors (Flores et

al., 1999).

Several strategies have been proposed to protect products from proteolytic degradation.

Protease-deficient strain was developed and it found to improve yield and product quality

(Cereghino and Cregg, 2000). Protease activity could be inhibited by running fermentation in

specific conditions. For instance, reducing the temperature from 30 to 23 was found to

improve cellular viability and prevent product degradation during induction (Jahic et al., 2003).

Besides, using sorbitol as a co-substrate reduced proteases release by decreasing cell mortality

during methanol induction (Wang et al., 2010). In addition, adding protease inhibitors in the

culture medium was also an effective way to protect the product (Shi et al., 2003).

1.6 Product recovery from P. pastoris culture

1.6.1 Typical process of product recovery

Product recovery from cell culture is the step after bioreactor production. In this step, cells, cell

debris and solid contaminant are removed and particle free liquid is generated for

chromatography steps (Sampath et al., 2014). A typical process of product recovery contains

Page 44: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

44

primary recovery, secondary clarification and virus filtration (Yavorsky et al., 2003). Bulk of

cells, residual cell or cell debris and viruses are clarified from liquid in each step, respectively.

Primary recovery, the first step of clarification, refers to the process of removing bulk of whole

cells and cell debris. Due to the high cell density, harvesting products from P. pastoris culture

is quite challenging. Centrifugation, filtration and depth filtration are the most commonly used

options in primary recovery of P. pastoris culture (Sampath et al., 2014). Besides, expanded

bed absorption (EBA) and aqueous two-phase system (ATPS) are also developed as alternative

options (Dong et al., 2012, Charoenrat et al., 2006).

It is notable that primary recovery removes most of but not all particles. For instance, only

particles larger than the pore will be removed when filtration is applied in primary recovery. In

secondary clarification, liquid obtained from primary recovery is filtered to remove residual

cells, cell debris and any remaining particles that are not removed in primary recovery.

Secondary clarification provides a protection to the downstream by removing any particles that

may foul chromatography resin. One or two depth filtrations are used in secondary clarification.

Different options can be combined in clarification based on cell type, cell density and sample

volume. Maximizing product recovery and clarification is the major driver in selection of these

options. Meanwhile, time cost, process reproducibility and consistency should be considered.

Three clarification strategies, centrifugation followed by depth filtration, centrifugation

followed by filter-aid enhanced depth filtration, and microfiltration were compared in one study

(Wang et al., 2006). It was found all the three approaches achieved high product recovery and

clarification. Processing time was minimized by using centrifugation, but the capital cost was

high and process scale-up was challenging. Using microfiltration simplified the process but

significantly increased the filtration time. Therefore, it is necessary to consider each aspect of

the clarification options.

1.6.2 Types of centrifuge for product recovery

Centrifugation is a process that uses centrifugal force to separate solid and liquid based on their

different densities. P. pastoris cells are relatively large entities and present high sedimentation

velocities. Therefore, centrifuge is most commonly used to recovery P. pastoris culture in

industry. Compared to other approaches, centrifugation is feasible to handle with a large volume

of samples at high speed. The major drawback is that scaling up centrifugation from lab to

industrial scale is difficult to be predicted. Besides, installation and maintenance of centrifuges

are high-cost.

Page 45: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

45

Figure 1.4 Operating mode of tubular bowl (A), disc type (B) and decanter scroll (C) centrifuge.

Adapted from (Dévay 2013; Tarleton and Wakeman 2016).

In industry, tubular bowl, disc stack and decanter scroll centrifuge are widely used (Fig 1.4).

Tubular bowl centrifuge has a rapidly rotating bowl that provides a sedimenting acceleration.

Feeding stream enters the bottom of the bowl and the solid is separated out by the centrifugal

force. Disc stack centrifuge is featured by a vertical stack of thin discs. Number of discs ranges

from 50 to 150 depending on dimension of the centrifuge. Feeding stream flows towards axis

of the bowl and fills the spaces between discs. Bowl rotation settles solid to underside of each

bowl, while clarified liquid stays in centre of the bowl and is discharged from top or bottom.

Decanter scroll centrifuge has a rotating bowl and conveyor scroll inside the bowl. Feeding

stream is introduced into the bowl through a pipe. Bowl rotation settles solids to wall of the

bowl and leaves clarified liquid in centre of the bowl. Conveyor scroll rotates in the same

direction at a slower speed and scroll the sediment to the end of the bowl.

Each type of centrifuge has advantages and drawbacks. Tubular bowl centrifuge has a higher

separation force than disc type or decanter scroll centrifuge. However, it has a much lower solid

capacity and solid can only be removed by stopping the centrifuge to flush solid out. Disc stack

Page 46: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

46

centrifuge is able to generate liquid with higher clarification, but its installation is more

expensive. Decanter scroll centrifuge advantages over tubular bowl and disc stack centrifuges

by processing fluids with solid concentrations as high as 60%-80%. The major drawback is that

it can only provide an acceleration up to 4000 g.

1.6.3 Depth filtration as an alternative to centrifuge

Depth filtration is a process that particles are separated from liquid by retaining particles with

filter medium. Depth filters have an open-pore structure and structure inside filter is tortuous

and channel-like (Obrien et al., 2012). Pore size of depth filtration varies from 0.1 μm to over

10 μm to retain different cell types. When cell culture flow through a depth filter, cells are

retained in the channels and liquid flows throughout filters. In contrast, cells are only retained

on surface of the filter in microfiltration. Depth filters used in biopharmaceutical are normally

made of cellulose fiber. Each filter is a flat sheet and many filters are assembled together to

form a multistack column (Obrien et al., 2012).

Depth filtration can retain mass particles before being fouled (Shukla and Kandula, 2008), and

thus it can replace centrifuge as an option of primary recovery. In a case study, the centrifuge

was replaced by a disposable depth-filter system in clarification of CHO cell culture (Obrien et

al., 2012). Compared to centrifuge, depth-filter was easier to set up and had relatively low

capital cost. Besides, it eliminated the needs of equipment cleaning and cleaning validation.

Depth filtration was also found an effective option to clarify P. pastoris cell culture (Chandler

and Zydney, 2005). Filters with smaller size were found to provide higher clarification at low

capacity and vice versa.

Depth filtration can be scaled up by keeping a constant flux rate (flow per unit membrane area),

which is much easier than scale-up of centrifuge. Linear scaling of depth filtration from lab

scale to 10, 000 L has been reported (Deakin, 2012). Solid amount that can be retained in a

depth-filter is limited. Depth filtration is not suitable in primary harvest of high density cell

culture (Yavorsky et al., 2003). In that case, centrifugation followed by depth filtration or

microfiltration is more effective.

1.6.4 Product recovery using single microfiltration

Filtration, specifically microfiltration, is a process that cell culture flows through a specific

pore-sized membrane and cells are retained on surface of the membrane (Cheryan, 1998).

Compared to centrifugation, microfiltration provides higher efficiency of solid removal and

over 99.5% of solids can be removed in one step (Yavorsky et al., 2003). Therefore, one step

Page 47: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

47

of microfiltration can be robust enough to provide clarified liquid for purification.

Microfiltration removes all partials upper the range of pore size through size exclusion, and

thus it is more resistant to variabilities in feeding stream (Yavorsky et al., 2003).

Dead end (DE-MF) and tangential flow microfiltration (TFF-MF) are two main types of

microfiltration. In DE-MF, feeding fluid is vertical to membrane surface and liquid is pushed

through the membrane (Igunnu and Chen, 2012). Solid tends to build upon membrane surface

of DE-MF and the increasing layer thickness will decline the flux. Thus, stopping filtration to

clean or change the membrane is generally required in DE-MF, which limits its application in

industry (Li and Li, 2015). TFF-MF is faster and more efficient than DE-MF and is more

preferable in industrial use. In TFF-MF, most of the liquid travels tangentially across the

membrane and keeps washing surface of the membrane, thus there is a lower tendency to build

up cake upon membrane surface (Igunnu and Chen, 2012). A single TFF-MF was found robust

to deliver high product recovery and clarification (Wang et al., 2006). It was also combined

with centrifugation to further avoid membrane fouling (Sampath et al., 2014).

TFF-DF was shown effective to harvest product from P. pastoris culture with over 30% (v/v)

cells (Wang et al., 2006, Polez et al., 2016). Comparable clarification was achieved by TFF-DF

and centrifugation followed by depth filtration.

Page 48: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

48

1.7 Thesis objectives

Pichia pastoris (P. pastoris) is becoming a popular host for the manufacture of a wide range of

products including recombinant proteins, enzymes and vaccines. Whist high density P. pastoris

fermentation is currently in used, the bioprocessing challenges remain significant especially at

large scale. Methanol induction makes process scale-up difficult due to high oxygen

requirement and substantial heat generation. The productivity of the cell is also low and product

degradation caused by protease release is likely.

In a previous study, a sorbitol/methanol mixed induction strategy was established for a P.

pastoris strain producing recombinant aprotinin. Compared to standard methanol induction, the

mixed induction strategy was shown to efficiently induce product expression and reduce

oxygen consumption and heat generation. However, its impact on product quality attributes and

early downstream processing still remains unclear. Besides, sorbitol/methanol mixed induction

was only studied at small scale cell culture in most of the reports and its performance at large

scale has not been studied.

In this thesis, a sorbitol/methanol mixed induction strategy was investigated to assess its impact

on both upstream and early downstream processing of high density P. pastoris culture. In

addition, oxygen transfer rates (OTR) expected in large scale fermentation was used as the

scale-down criteria at 1 litre to compare the performances of methanol and mixed induction

strategies.

Chapter 1: To overview the current hosts of biopharmaceutical production, and to introduce P. pastoris expression host, fermentation strategies, host cell protein impurities, and product

recovery options.

Chapter 2: To introduce the materials, cell strain, equipment, sample analytical techniques and

experimental methodologies used in this research.

Chapter 3: A sorbitol/methanol mixed induction strategy was applied to P. pastoris

fermentation. A quantification assay of the product was established and the impact of

sorbitol/methanol mixed induction on cell growth, volumetric product yield, cell viability and

HCPs profile were studied.

Chapter 4: In this chapter, sorbitol/methanol mixed induction was applied in a bioreactor that

has comparable OTR with large scale ones. Oxygen transfer coefficient of one litre bioreactor

was characterized. Methanol and sorbitol/methanol mixed induction strategies were compared.

Page 49: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

49

Chapter 5: To study the impact of sorbitol/methanol mixed induction on centrifugal dewatering.

Cell properties influencing centrifugal dewatering were studied and the dewatering efficiency

of cell cultures from methanol and sorbitol/methanol mixed induction strategies were compared

by using a scale-down model of disc stack centrifuge.

Chapter 6: To summarize the main conclusions and introduce possible future works.

Page 50: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

50

Chapter 2 Materials and methods

2.1 Materials

Methanol and sorbitol were purchased from VWR International Ltd (Lutterworth, UK). All the

other materials were purchased from Sigma-Aldrich Cooperation (Dorset, UK) unless

otherwise specified.

2.2 Culture medium of P. pastoris

2.2.1 Buffered complex medium

Buffered complex medium was prepared by dissolving 20 g peptone, 10 g yeast extract, 13.4 g

yeast nitrogen base (YNB), 1.15 g potassium hydrogen and 5.9 g potassium dihydrogen

phosphate in one litre Milli-Q water. The medium was filtered using Millipore® Stericup™

filtration system (Merck Millipore, Watford, UK) and stored at 4 for further use. Before use,

the medium was supplemented with glycerol or methanol to obtain BMGY and BMMY

medium.

2.2.2 Basal salts medium

Basal salts medium (BSM) recipe developed by Invitrogen™ Thermo Fisher Scientific

(Carlsbad, CA, US) was used. One litre of the medium consisted of 26.7 ml 85% (v/v)

phosphoric acid, 0.93 g calcium sulfate, 18.2 g potassium sulfate, 14.9 g magnesium

sulfate•7H2O, 4.1 g potassium hydroxide, and 40.0 g glycerol. Before sterilization in an

autoclave, the pH was adjusted to 4.0 using 15% (m/v) ammonium hydroxide. Each litre of the

medium was supplemented with 4.35 ml Pichia trace metals (PTM1) solution before use.

PTM1 solution was prepared by dissolving 6.0 g cupric sulfate•5H2O, 0.08 g sodium iodide,

3.0 g manganese sulfate•H2O, 0.2 g sodium molybdate•2H2O, 0.02 g boric acid, 0.5 g cobalt

chloride, 20.0 g zinc chloride, 65.0 g ferrous sulfate•7H2O, 0.2 g biotin and 5.0 ml sulfuric acid

in one litre Milli-Q water. PTM1 solution was sterilized using Millipore® Stericup™ filtration

system and stored at 4 for further use.

2.3 P. pastoris strain and working cell bank

A recombinant strain of P. pastoris Mut+ strain is kindly provided by Fujifilm Diosynth

Biotechnologies (Billingham, UK). The strain has been stably transfected with a plasmid having

Page 51: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

51

sequence of aprotinin and the recombinant protein is expressed extracellularly. Aprotinin, also

named as bovine pancreatic trypsin inhibitor (BPTI), consists of 58 amino acid residues and the

residues are arranged in a single chain that containing 3 disulfides. Aprotinin has a MW of 6512

Da and an isoelectric point of 10.5.

10 20 30 40 50

RPDFCLEPPY TGPCKARMIK YFYNIRSRSC EEFIYGGCEA KKNNFEAMED

CMRTCGGA

Seq.1 Amino acid sequence of aprotinin (Uniprot, N.A.).

To generate a working cell bank, frozen cells were thawed and cultured in one litre shake flask

with 150 ml BMGY. The flask was cultured at 30 and agitated at 250 RPM. Optical density

at 600 nm (OD600) of the cell culture was monitored by Ultraspec 500 pro spectrophotometer

(Amersham Bioscience Corp, Little Chalfont, UK) and when it reached 20, cell culture was put

into 15 ml Fisherbrand centrifuge tubes (Thermo Fisher Scientific, Cramlington, UK) and

centrifuged at 4000 RPM for 10 min using an Eppendorf 5810R benchtop centrifuge

(Eppendorf UK, Stevenage, UK). After the supernatant was discarded, cell pellet was re-

suspended using 50% (v/v) sterilized glycerol and aliquoted into 2 ml Eppendorf tubes for long

term storage at -70

2.4 Cell culture methods

2.4.1 Shake flask culture

Cells were grown on different carbons in shake flasks. 50 ml buffered complex medium was

added to 250 ml shake flask and was supplemented with 1% (v/v) glycerol, methanol, sorbitol

or methanol/sorbitol mixture (1:1 C-mol/C-mol). After flasks were inoculated to obtain starting

OD600 of 1.0, they were cultured at 30 and agitated at 250 RPM for 48 h. OD600 of cell culture

was measured every 12h by Ultraspec 500 pro spectrophotometer.

2.4.2 Microplate culture

Prior to cell culture in Corning 24 microplate (Corning Incorporated, Deeside, UK), cells were

initially cultured in one 250 ml shake flask to obtain enough cells. When OD600 of the culture

Page 52: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

52

in shake flask reached 20, cells were harvested by centrifuging at 4000 RPM for 10 min using

an Eppendorf 5810R benchtop centrifuge. After the supernatant was discarded, cells were re-

suspended and diluted using buffered complex medium until OD600 reached 1.0. Each well was

pipetted with 2 ml of diluted cell culture and the plate was covered by a gas permeable sealing

film. The plates were cultured at 30 in an 80% (v/v) humidified shaker and agitation speeds

of 250 RPM or 400 RPM were used respectively. OD600 of the culture was measured by

Ultraspec 500 pro spectrophotometer.

2.4.3 Infors bioreactor culture

Fermentation was carried out in Multifors bioreactor (Infors UK Ltd., Reigate, UK) which has

four one-litre vessels. Before the pH probes were installed, they were calibrated using standard

calibration buffers, pH4 and pH7 (PCE Instruments, Southampton, UK), respectively. After the

calibration, each vessel was filled with 550 ml BSM medium and sterilized in an autoclave.

Vessels were put back on bioreactor’s platform and probes of pH and dissolved oxygen were

connected to the bioreactor for at least 6 h before use. When the probe reading stabilized, DO

was calibrated to 100% and pH was adjusted to 5.0 using 15% (m/v) ammonium hydroxide.

Finally, 2.5 ml sterilized PTM1 solution was added to each vessel before inoculation.

To prepare cells for inoculation, cells from the working cell bank were initially cultured in a

one litre flask containing 150 ml complex medium. After incubation in a shaker agitating at 200

RPM for 16 h, cells were sucked into 20 ml syringes and injected into the inoculation port of

the bioreactor to obtain an initial OD600 of 1.0. During the fermentation, pH was maintained at

5.0 using 15% (m/v) ammonium hydroxide. DO was kept at 30% by coupling with changing

agitation speed and oxygen proportion in the inlet. When DO dropped below 30%, the agitation

would increase from 300 RPM to a maximum of 1100 RPM. When the DO dropped below 30%

at the maximum agitation speed, oxygen proportion would increase from 21% to a maximum

of 80%.

Cells initially grew on glycerol in the medium after inoculation. When glycerol was consumed,

indicated by a DO spike, 50% (v/v) glycerol was fed at a rate of 18.0 ml•h-1•L-1 until the

culture’s OD600 reached ~300. After glycerol feeding was stopped, cells were starved for 1 hour

until the thorough consumption of residual glycerol. Induction was switched on after starvation

phase by using pure methanol or sorbitol/methanol mixture (Woodhouse, 2016). In order to

adapt cells to new carbons, methanol or sorbitol/methanol were fed at a rate of 3.6 ml•h-1•L-1

for one hour and the feeding rate was kept at 7.2 ml•h-1•L-1 for two hours afterward. Finally,

feeding rate was increased to 10.8 ml•h-1•L-1 and kept constant until the harvest. Cell culture

Page 53: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

53

was harvested through sampling ports and stored at 4 for further use. After probes and gas

lines were disconnected with the bioreactor, remaining cell culture in vessels was killed by

autoclave.

Samples were taken regularly through sampling ports during fermentation. Samples were

pipetted into 1.5 ml Eppendorf tubes and centrifuged at 4000 RPM for 10 min in an Eppendorf

5424R benchtop centrifuge (Eppendorf UK, Stevenage, UK). The supernatant was collected

and stored at -20 for further analysis.

2.5 Prediction of dewatering and clarification

Cell culture was diluted to a volumetric cell fraction of 30% (v/v) using Milli-Q water. Then it

was added in 2 ml Eppendorf tubes or 15ml centrifuge tubes and spun by Eppendorf 5810R

(Eppendorf UK, Stevenage, UK) with the fixed rotor FA 45-30-11and Beckman Coulter Avanti

J-E Centrifuge (Beckman Coulter United Kingdom, High Wycombe, UK) with fixed rotor of

JA-21, respectively. Dewatering and clarification in CSA-1 or BTPX-305 disc stack centrifuges

were predicted by centrifugation in 2 ml and 15 ml tubes. Speed and dimension of centrifuges

were shown in Table 5.2. OD600 of supernatant was measured after centrifugation and

clarification was calculated using Eq 5.5. After supernatant was discarded thoroughly, tubes

with cell pellets were weighed before and after being dried at 100 for 24 h and dewatering

levels were calculated using Eq 5.6.

2.6 Characterization of oxygen transfer coefficient

Oxygen transfer coefficient (kLa) of the Multifors vessels were measured using the dynamic

method (Garcia-Ochoa and Gomez, 2009). DO probe was calibrated between 0% and 100%

using nitrogen and air flow. 700 ml BSM medium was added into each vessel and temperature

was set at 30. Nitrogen flow was initially introduced into the vessels until dissolved oxygen

dropped to 0%. Oxygen was then sparged into the vessels at the rate of 1 L•min-1 and DO

reading was recorded every 5 sec until it reached 100%. kLa values of bioreactor were calculated

at agitation speeds of 300, 500, 800, 1100 RPM, respectively.

Page 54: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

54

2.7 Cell breakage method

For scale-down studies, Covaris E220 Focused-ultrasonicator (Covaris, Inc., Woburn, MA,

USA) was used to break yeast cells according to manufacturer’s instruction. Fresh cells were

obtained from vessels and centrifuged at 4000 RPM for 10 min using Eppendorf 5810R

benchtop centrifuge. The supernatant was discarded, and cell pellet was re-suspended and

washed twice with cold 200 mM phosphate buffer. Cells were diluted to the concentration of

20 g•WCW-1, and 1 ml cell culture was loaded in borosilicate glass vials and sonicated for 1200

s. Temperature, duty cycle and cycles per burst were set at 10, 20% and 1000, respectively

(Bláha et al., 2017).

2.8 Cell robustness study

Robustness of P. pastoris cells to mechanical stress was studied using an ultra scale-down (USD)

shear device according to manufacturer’s instruction (Boychyn et al., 2004, Hutchinson et al.,

2006). Briefly, fresh cells were harvested from vessels and diluted to volumetric cell fraction

of 30% (v/v) using Milli-Q water. 20 ml diluted sample was then injected into device’s chamber

using syringe. Once injection port was closed, the rotation was switched on and worked at

speeds of 167, 233, and 300 rps for 20 s. Samples were removed with a clean syringe after

rotation, and the device’s chamber was washed three times using chilled water. Soluble protein

concentration and cell viability in samples were immediately measured.

2.9 Analytical methods

2.9.1 Aprotinin measurement

Sigma-Aldrich provided quantification method of aprotinin in Enzymatic Assay of Aprotinin

(Sigma-Aldrich-a). Absorbance of the following reaction was measured consecutively and

aprotinin was determined by calculating inhibition rate of the reaction.

Nα − Benzoyl − DL − Arginine − p − Nitroanilide56789:;<⎯⎯⎯⎯> Eq 2.1

Nα − Benzoyl − DL − Arginine +p-Nitroaniline

Page 55: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

55

200 mM triethanolamine buffer was obtained by preparing 13.6 mg•ml-1 solution of

triethanolamine using Milli-Q water. After pH was adjusted to 7.4 at room temperature with 1

M KOH, the buffer was sterilized using 0.22 μm filter and stored at room temperature. 0.1%

(w/v) Nα-Benzoyl-DL-Arginine-p-Nitroanilide (BAPNA) solution was obtained by preparing

1.0 mg•ml-1 BAPNA solution in Milli-Q water. The solution was prepared immediately before

use. Trypsin enzyme solution was obtained by dissolving enzyme into 1 mM hydrogen chloride

(HCl). Standard aprotinin solution was prepared immediately before use by dissolving aprotinin

in 0.9% (m/v) sodium chloride (NaCl) solution. The following reagents were pipetted into

Corning 96 wells plate with flat bottom (Corning Incorporated, Deeside, UK).

After

the

mixture

was

mixed

well in a

Thermomix, 100 μl 0.1% (w/v) BAPNA solution was added and mixed by pipette. Absorbance

at 405 nm was

measured consecutively every two minutes using a Tecan Microplate reader (Tecan Group Ltd.,

Männedorf, Switzerland).

Inhibition % was calculated by:

Inhibition% = 100 •∆405nm/minUninhibited − ∆405nm/min Inhibited∆405nm/minUninhibited − ∆405nm/minBlank Eq 2.2

Where ∆405 nm/min is the increasing rate of mixture’s absorbance at 405 nm.

Uninhibited Inhibited

Blank (μl) Control (μl) Tested sample (μl)

200 nm Triethanolamine 160 160 160

1 mM HCl 20 - -

Trypsin solution - 20 20

0.85% (v/v) NaCl 20 20 -

Inhibitor - - 20

Table 2.1 Components of the reaction mixture in aprotinin quantification.

Page 56: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

56

2.9.2 Cell density measurement

OD600, wet cell density (WCW) and dry cell weight (DCW) were measured during fermentation.

Samples were diluted properly using Milli-Q water and loaded into 1 ml cuvette for absorbance

measurement by Ultraspec 500 pro spectrophotometer. For cell density measurement, weights

of blank 1.5ml Eppendorf tubes were measured before sample loading. 1ml samples were

pipetted into tubes and centrifuged at 4000 RPM for 10min using an Eppendorf 5424R benchtop

centrifuge. Tubes with wet cells were weighed again after supernatant was removed. Dry cell

weight was measured by weighing the remaining solids after wet cells were dried at 100 for

24 h.

2.9.3 Determination of methanol and sorbitol

Methanol and sorbitol concentrations were determined using UltiMate 3000 HPLC (Thermo

Fisher Scientific, Cramlington, UK) with Aminex HPX-87 column (Bio-rad Laboratories In.,

Watford, UK). The column was balanced with 0.5% (v/v) trifluoroacetic acid at constant flow

rate of 0.6 ml•h-1 before use. After 20 μl samples were loaded through the column for 30 min,

the output was detected by Thermo Scientific™ RefractoMax 520 Refractive Index Detector

(Thermo Fisher Scientific, Cramlington, UK) at 55 (Parpinello and Versari, 2000).

2.9.4 DNA quantification

DNA concentration in the supernatant was quantified using Quant-iT™ PicoGreen dsDNA

Reagent (Thermo Fisher Scientific, Cramlington, UK). Samples were diluted properly using

TE buffer (10 mM Tris-HCl, 1 mM EDTA, pH 7.5). 100 μl diluted samples and 100 μl diluted

Quant-iT™ PicoGreen® Reagent were pipetted into a 96 wells plate. After being incubated at

room temperature for 5 min, fluorescence in the reaction mixture was measured by a microplate

reader (excitation ~480 nm, emission ~520 nm). Standard Lambda DNA provided in the kit

was used to build the standard curve.

2.9.5 Soluble protein quantification

Soluble protein concentration was measured using bicinchoninic acid (BCA) assay. Pierce BCA

Protein Assay Kit and standard bovine serum albumin (BSA) were purchased from Thermo

Fisher Scientific. 0.13 mg•ml-1, 0.25 mg•ml-1, 0.50 mg•ml-1 and 1.0 mg•ml-1 of BSA standard

solution were prepared, and samples were diluted using Milli-Q water. In a 96 wells plate with

flat bottom, each well was firstly pipetted with 20 μl BSA standard protein or sample and then

Page 57: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

57

added with 200 μl BCA solution. After the plate was incubated at 37 for 10min, absorbance

was measured at 562 nm by a microplate reader.

2.9.6 NuPAGE Bis-Tris gel electrophoresis

Frozen sample was thawed at room temperature and centrifuged at 15000 RPM for 10 min to

remove precipitation. Supernatant was mixed with 4•Lithium Dodecyl Sulfate (LDS) sample

loading buffer (Thermo Fisher Scientific, Cramlington, UK) and heated at 100 for 10 min.

Samples were centrifuged at 15000 RPM for 10 min to remove precipitation. The 4–12%

gradient NuPage SDS Novex precast gel (Invitrogen, Paisley, UK) was set up in gel tank and

immersed into 3-(N-morpholino) propanesulfonic acid) (MOPS) buffer. 5 μl Mark12TM

Unstained Standard (Thermo Fisher Scientific, Cramlington, UK) or 10 μl supernatant was

loaded into each well of the gel. Electrophoresis was performed at a constant voltage of 200 V

for 35 min using Bio-rad Powerpac Basic (Bio-rad, Watford, UK). After the electrophoresis,

gel was immersed into Coomassie Brilliant Blue (Sigma-Aldrich, Poole, UK) and rotated

overnight for staining. Staining solution was discarded, and the gel was washed twice using

Milli-Q water. It was destained by a buffer containing 10% (v/v) acetic acid and 10% (v/v)

ethanol. Protein bands were visualized using Amersham Imager 600 after gel destaining (GE

Life Sciences, Little Chalfont, UK).

2.9.7 Cell viability measurement

Cellular viability was measured using BD Accuri™ C6 cytometer (BD Biosciences, Oxford,

UK). Fresh cell samples were diluted to a cell concentration with OD600 of 0.5 using 0.9% (w/v)

NaCl. Then 7 μl propidium iodide was added into 1ml sample and the sample was incubated in

dark for 10min. Percentage of cells stained by propidium iodide was measured by flow

cytometer. Threshold setting was set at 10000 to exclude cell debris from data acquisition.

Staining was measured through the FL3 (red channel) filter and 5•104 cells were analysed per

sample.

2.9.8 Two-dimensional gel electrophoresis

Cell culture medium was changed by an ultrafilter with a cut-off of 3000 Da. Soluble proteins

were precipitated using 13% (v/v) trichloroacetic acid (TCA) in cold acetone. Briefly, 0.1 ml

sample and 1 ml TCA solution were pipetted into a 1.5 ml Eppendorf tube and mixed well using

a thermomixer. Dithiothreitol (DTT) was added to a final concentration of 20 mM and the

mixture was kept at -20 overnight. The mixture was centrifuged at 15000 RPM for 10 min to

Page 58: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

58

remove TCA solution and collect precipitation. In order to thoroughly remove residual TCA,

the precipitate was washed three times using cold acetone. After acetone was discarded by

centrifugation, the precipitate was dried at room temperature until residual acetone volatilized.

Protein pellet was re-dissolved using a rehydration buffer containing 8 M urea, 2% (v/v) 3-[(3-

Cholamidopropyl) dimethylammonio]-1-propanesulfonate hydrate (CHAPS), 0.002% (v/v)

bromophenol blue and 0.5% (v/v) ampholytes. Protein concentration was measured using BCA

method as described in the Section 2.9.5, Chapter 2. 200 μg protein was pipetted into a new

tube and added up to 150 μl using the rehydration buffer.

The first dimensional gel electrophoresis was done using the IPGphor Isoelectric Focusing

System (Stockholm, Sweden). Immobilized pH gradient (IPG) strip was put in the cassette and

protein sample was added in to immerse the gel. Finally, 300 μl sealing oil was added to cover

the IPG strip. The cassette was put on IPGphor and electro focusing was performed using the

voltage as shown in the table below.

Page 59: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

59

Voltage Time

0 V 12 h

200 V 20 min

450 V 15 min

750 V 15 min

2000 V 120 min

Table 2.2 Voltage and time used in IPG electro focusing.

After the electro focusing, IPG strip was incubated in 5 ml 1•NuPAGE LDL buffer with 50 mM

DTT for 15 min. Strip was taken out from DTT solution and washed twice using Milli-Q water.

It was then incubated in 1•NuPAGE LDL buffer containing 125 mM iodoacetamide for 15 min

to remove residual DTT.

NuPAGE 4-12% Bis-Tris ZOOM Protein Gel was set up in gel tank and immersed in MOPS

buffer. IPG strip was loaded to the ZOOM gel and 500 μl 0.5% (m/v) agarose was added in the

well to seal the gap between IPG strip and ZOOM gel. After the agarose solidified, gel

electrophoresis was performed at 200 V for 40 min using Bio-rad Powerpac Basic. After the

electrophoresis, gel was stained by coomassie blue overnight. Gel was destained by washing in

a buffer with 10% (v/v) methanol and 7% (v/v) acetic acid for 30 min. Finally, protein dots

were visualized by Amersham Image 600.

2.9.9 Proteomics assay

Host cell proteins in the supernatant were identified using a method reported before (Kumar et

al., 2016). Supernatant containing 5 μg soluble proteins was loaded to a 20% SDS-PAGE gel

and electrophoresis was run until loading samples were concentrated. Gel slice that contained

all proteins were cut down and proteins in gel were digested into peptides by acid cyanogen

bromide (CNBr) in 70% (v/v) formic acid. After peptide mixture was suspended in 0.1% (v/v)

formic acid, it was analysed by electrospray liquid chromatography-mass spectrometry (LC-

MS/MS) (Surveyor, ThermoFinnigan, CA). Spectrum was processed using Proteome

Discoverer (Thermo Fisher Scientific Inc.) and searched against Uniprot database using Mascot

search algorithm (Matrix Science, London, UK). Protein identification was conducted using

Scaffold (Proteome Software Inc., Portland, OR, USA). Protein identification was considered

Page 60: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

60

acceptable if threshold could be established over 95% probability and the protein contained at

least one identified unique peptide.

2.9.10 Cell size measurement

Cell size was measured by Malvern Mastersizer 3000 laser diffraction particle size analyser

(Malvern Panalytical Ltd, Malvern, UK) according to the manufacturer’s instruction. After

initialization and background measurement, samples were slowly dropped into the Hydro MV

liquid dispenser until light obscuration reached 15%. Each sample was measured 5 times to

obtain an average value for particle size. Water and P. pastoris cell culture were assumed to

have refractive indexes of 1.33 and 1.56, respectively.

2.9.11 Rheology measurement

Viscosity of sample was measured by Malvern Kinexus rheometer (Malvern Panalytical Ltd,

Malvern, UK) according to the user manual. After initialization of the rheometer and

temperature setting at 30, 1.2ml sample was loaded on the sample plate. Sample’s viscosity

in terms of Pa•s-1 between shear rates of 100 s-1 to 1000 s-1 was measured automatically.

2.9.12 AOX activity measurement

AOX activity was measured using a method recommended by Sigma-Aldrich in Enzymatic

Assay of Alcohol Oxidase (Sigma-Aldrich-b). Absorbance of the following reaction was

measured consecutively and AOX activity was determined from the reaction rate.

Methanol + ORSTU<⎯> Formaldehyde + HROR Eq 2.3

HROR + ABTSZT[<⎯> 2HRO + ABTS(oxidized) Eq 2.4

100nM phosphate buffer was obtained by preparing 13.6 mg•ml-1 solution of KH2PO4. Buffer

pH was calibrated to 7.5 with 1M KOH at room temperature. 2 mM 2,2’-Azino-bis (3-

ethylbenzthiazoline-6-sulfonic acid) (ABTS) solution was obtained by preparing 1mg•ml-1

solution of ABTS in phosphate buffer. 1.0% (v/v) methanol solution was obtained by preparing

1.0 mg•ml-1 HPLC grade methanol using Milli-Q water. 250 U•ml-1 peroxidase solution was

prepared immediately before use.

The following reagents were added into a 96 wells plate:

Page 61: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

61

Reagent Test samples (μl) Blank (μl)

ABTS solution 280 280

Peroxidase solution 1 1

Phosphate buffer - 10

Methanol solution 10 10

Table 2.3 Components of the reaction mixtures in quantification of AOX enzyme.

After the reagents were mixed well, 10 μl test sample was added to each well except the blank

well. Absorbance at 405 nm was measured consecutively every two minutes. Three replicates

were used for each sample and blank. Activity was defined as:

Units/mg =(∆405nm/minTest − ∆405nm/minBlank) ∗ 301 ∗ df

36.8 ∗ 10 Eq 2.5

Where ∆405 nm/min is the increasing rate of mixture’s absorbance at 405 nm. Constant of 301

refers to the total reaction volume (in microlitres), df is the dilution factor, constant of 36.8

refers to extinction coefficient of ABTS at 405 nm, constant of 10 is volume (in microlitres) of

enzyme used.

Page 62: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

62

Chapter 3 Impact of sorbitol/methanol mixed induction on cell

growth and product expression

3.1 Introduction

P. pastoris is becoming a popular host for the production of heterologous proteins. However,

P. pastoris cultivation faces challenges especially at industrial scale. As a methylotrophic yeast,

it uses methanol as the promoter inducer (Cereghino and Cregg, 2000). Methanol usage is

constrained by the high oxygen demand and need for heat removal in large scale bioreactors

(Hensing et al., 1995) which impose potential design restrictions. Besides, using methanol

imposes challenge to strict health and safety regulations. Thus, reducing methanol consumption

is potentially advantageous to process scale-up.

Partially replacing methanol with sorbitol has been shown to reduce drawbacks of pure

methanol induction and benefit P. pastoris cultivation (Çelik et al., 2009). Sorbitol has a lower

enthalpy of combustion and thus sorbitol/methanol mixed induction reduces the oxygen

consumption rate significantly. Besides, sorbitol/methanol mixed induction was reported to

reduce formation of toxic formaldehyde and enhance cellular viability (Wang et al., 2010).

Effect of sorbitol/methanol mixed induction on product yield is strain dependent. Celik and co-

workers reported that productivity of recombinant human erythropoietin was enhanced 1.8

times by using sorbitol as a co-substrate compared to linearly feeding methanol (Çelik et al.,

2009). Niu and co-workers found that product yield of β-galactosidase was comparable when

mole fraction of Cmethanol was maintained in the range of 45% ~100% (Niu et al., 2013). It was

also reported that using sorbitol/methanol (4:6, C-mol/C-mol) mixed induction reduced the

volumetric productivity of Rhyzopus oryzae lipase compared to methanol induction (Berrios et

al., 2017).

In a previous study, a sorbitol/methanol mixed induction strategy was developed for the

production of recombinant aprotinin, a competitive inhibitor of trypsin and related proteases

(Woodhouse, 2016). Aprotinin production using the sorbitol/methanol mixed induction was

successfully scaled up to pilot scale. It was found that the mixed induction strategy reduced the

heat generation and proteases co-released with the product. Despite the study, aprotinin yields

from methanol and mixed induction strategies have not been quantitatively compared, and the

best harvest time for the fermentation has not been determined. To better evaluate the

sorbitol/methanol mixed induction strategy, a quantitative assay of aprotinin was developed to

Page 63: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

63

quantitively determine product yield in this study. Impact of mixed induction on cell growth,

cell viability, oxygen uptake and HCPs co-released with the product was studied.

The main objectives of this chapter were to

1) Developing the best ratio of methanol/sorbitol mixture for P. pastoris induction using

microplate scale culture.

2) Study the impact of sorbitol/methanol mixed induction on cell growth and viability at

bioreactor scale.

3) Compare the product yields of methanol and sorbitol/methanol mixed induction strategies.

4) Study the HCPs co-released with the product and predict the impact of mixed induction on

product purification.

Page 64: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

64

3.2 Theoretical considerations

3.2.1 Specific growth rate

The specific growth rate (µ) is defined as the increasing rate of a cell population per unit of cell

concentration (Eq 3.1). Cell growth rate (μ) was calculated using Eq 3.2 (Clarke et al., 2011),

which is the slope of line between lnX and t. DCW2 and DCW1 refer to the dry cell weight at

sampling time of t2 and t1.

µ =lnXXt Eq 3.1

µ =ln[ijR − ln[ijk

tR − tk Eq 3.2

3.2.2 Cell respiration

The oxygen uptake rate (OUR) is one of the fundamental physiological characteristics of cell

culture. It refers to the rate of oxygen consumption by the cells. OUR can be determined by

measuring oxygen concentration in inlet and outlet gas phase. Correspondingly, carbon

evolution rate (CER), the rate of CO2 production by cells, can be determined by the mass

balance of CO2 in inlet and outlet gas. A gas analyser can be connected to the exhaust outlet of

bioreactor, and OUR and CER can be calculated using Eq 3.3 and Eq 3.4.

OUR =(pTR:; − pTRmno ∗ ppR:;/ppRmno) ∗ F ∗ ATM

RTVt Eq 3.3

CER =(piTRmno ∗ ppR:;/ppRmno − piTR:;) ∗ F ∗ ATM

RTVt Eq 3.4

Where PO2in and PO2out refer to volumetric fraction of oxygen in gas intake and outlet; PN2in and

PN2out refer to volumetric fraction of nitrogen in gas intake and outlet; PCO2in and PCO2out refer

to volumetric fraction of carbon dioxide in gas intake and outlet. F refers to the gas flow rate;

ATM refers to the standard atmosphere defined as 101 kPa; R refers to the gas constant with

value of 8.314 J•mol−1•K−1; V refers to the volume of cell culture; t refers to the duration of

measurement.

Page 65: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

65

3.3 Results and discussion

3.3.1 Quantitative assay of aprotinin

A quantification method of aprotinin was developed by Sigma-Aldrich (Sigma-Aldrich-a). In a

reaction buffer containing Ca2+, BAPNA is degraded by trypsin and the degradation rate is

calculated by consecutively measuring OD405 of the reaction mixture. Trypsin activity is

inhibited after the addition of aprotinin. The inhibition rate is determined using the equation as

described in the Section 2.9.1, Chapter 2. In this method, the reaction mixture has a total volume

of 3.0 ml and OD405 of the reaction mixture is measured by spectrophotometer.

When this method is used to measure aprotinin in the BSM medium, it was noticed that the

reaction buffer precipitated after being mixed with the medium. The precipitation was formed

by the phosphate and sulphate ion in the BSM medium and the Ca2+ in the reaction buffer. The

precipitation could not be eliminated even the medium was diluted over ten times. In the

reaction, Ca2+ plays a role in stabilizing the structure of trypsin and in turn increasing its activity.

In order to develop an assay compatible to the BSM medium, Ca2+ was removed from the

formula of reaction buffer. Trypsin activity slightly decreased without the Ca2+ but precipitation

was avoided.

In the original method, samples can only be measured one by one using spectrophotometer. It

is quite challenging to consecutively measure optical density of several samples at the same

time. Therefore, the reaction volume was scaled down linearly with a factor of 10.0 to increase

throughput of the measurement. The reaction could be done in 96 well plates after the scale-

down and optical density of the reaction mixture could be measured by a microplate reader.

In the original method, 1000U of trypsin was recommended to use. Since removing Ca2+

significantly reduced the reaction rate, it is better to screen the trypsin concentrations and find

the optimal one for measurement. As shown in Fig 3.1, highest reaction rate was obtained when

3000 U of trypsin was used. OD405 of the mixture reached stationary phase after 14 min of

reaction. The reaction rates decreased significantly at concentrations of 2000 U, 1000 U and

500 U. At these concentrations, OD405 of the reaction mixture increased linearly within 20 min.

When this assay is used to measure aprotinin, 2000 U of trypsin was used because it resulted in

a comparable reaction rate with the original method developed by Sigma.

Page 66: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

66

Figure 3.1 Effect of trypsin concentration on reaction rate of the enzymatic assay. Four

concentrations of trypsin, 500, 1000, 2000 and 3000 U were used in the reaction. Reaction

mixture’s absorbance at 405 nm was measured by microplate reader consecutively every two

minutes. Three replicates were used for each trypsin concentration and the data was shown by

mean�SD (n=3).

Effect of the culture medium on measuring accuracy was studied. A series of dilution of BMMY

and BSM mediums were added to the reaction mixture and inhibition rate was calculated. As

shown in Fig 3.2, the undiluted BMMY and BSM mediums inhibited the reaction by 40% and

23% respectively. The inhibitory effect was weakened by diluting the medium, and it was

negligible after BMMY and BSM mediums were diluted five and four times, respectively.

The reaction has an optimal pH of 7.8 (Sigma-Aldrich, n.d.-b) and the reaction rate was

apparently influenced by pH as noticed in a preliminary test. The BMMY and BSM mediums

have pH of 5.0 with phosphate buffer. Adding medium changed the of the reaction mixture and

thus inhibited the reaction. Aprotinin is measured by determining its inhibition rate to the

reaction. Thus, measuring aprotinin in the undiluted medium will be not accurate due to the

interference of medium. It is required to dilute the BMMY or BSM mediums for at least five

and four times in order to eliminate their interferences. The dilution is likely to make aprotinin

undetectable if the aprotinin concentration is very low which is common in early stage of

fermentation.

0 2 4 6 8 10 12 14 16 18 200

1

2

3

4

5

Reaction time min

Abs

orba

nce

at 4

05nm

500U 1000U

2000U 3000U

Page 67: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

67

Figure 3.2 Inhibitory effect of BMMY and BSM mediums on the trypsin catalyzed reaction.

BMMY and BSM mediums were diluted up to five times and added to the reaction mixture.

Same volume of 0.85% (v/v) NaCl was added in the uninhibited reaction. Inhibition rates were

calculated as described in the Section 2.9.1, Chapter2. Three replicates were used for each point

and data was shown by mean�SD (n=3).

0 1 2 3 4 5 60

20

40

60

Dilution factor

Inhi

bitio

n %

BMMY mediumA

0 1 2 3 4 5 60

10

20

30

Dilution factor

Inhi

bitio

n %

BSM mediumB

Page 68: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

68

Bovine aprotinin from Sigma-Aldrich was used to build a standard curve between the inhibition

rate and aprotinin concentration (Caprotinin). A series of standard arpotinin solutions with

concentration of 0.005, 0.010, 0.020, 0.040, 0.060, 0.080, 0.100 and 0.200 g•L-1 were prepared

and their inhibition rates on the reaction were determined.

As shown in Fig 3.3, aprotinin inhibited the reaction in a concentration dependent but not

linearly manner. The curve between aprotinin concentration and inhibition rate can be divided

into three parts: when the aprotinin concentration was below 0.06 g•L-1, it was related to the

inhibition rate with a function of Caprotinin=0.002•Inhibition%+0.0048, R2=0.99; when the

aprotinin concentration was in the range of 0.06~0.1 g•L-1, it was related to the inhibition rate

with a function of Caprotinin=0.0006•Inhibition%+0.045, R2=0.99; when aprotinin concentration

was over 0.1 g•L-1, the reaction was completely inhibited.

When the aprotinin concentration is unknow, it is necessary to dilute samples properly to

maintain the inhibition rate within the range of 0%~90%. It emphasizes the importance of

scaling down the assay to 96 well plate level. By using the plate, one sample can be diluted to

several concentrations and the inhibition rates can be measured in one plate. The suitable

inhibition rate for aprotinin determination can be picked.

Figure 3.3 Standard curve between inhibition rate and aprotinin concentration. Standard

aprotinin solutions with concentration of 0.005, 0.010, 0.020, 0.040, 0.060, 0.080, 0.100, 0.200

g•L-1 were prepared using bovine aprotinin. Inhibition rates of these aprotinin solutions were

calculated as described in the Section 2.9.1, Chapter 2. Three replicates were used for each

point and data was shown by mean�SD (n=3).

0 20 40 60 80 1000.00

0.05

0.10

0.15

0.20

0.25

Inhibition %

Sta

ndar

d ap

rotin

in g•L-1

Page 69: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

69

3.3.2 Cell culture at small scale

In order to understand cell growth behaviour on sorbitol/methanol mixture and determine the

best sorbitol/methanol ratio for product expression, cells were grown on sorbitol/methanol

mixture with different ratios at microplate scale.

Firstly, cells were grown on a buffered complex medium containing 5•10-4 mol Cmethanol to

establish the feasibility of cell culture in microplate. The microplate had a flat bottom and a

working volume of 2 ml. In order to minimize evaporation of medium, the microplate was

cultured in a humidified shaker. Agitation slower than 250 RPM could not suspend cells

completely. The microplate was agitated at speeds of 250 RPM and 400 RPM by humidified

shaker, respectively. When the cells were cultured at 250 RPM, OD600 of the cell culture

reached a maximum value of ~30.0 after 12 hours of cultivation. The cells cultured at 400 RPM

grew much faster and OD600 reached the maximum after 6 h cultivation. Oxygen transfer rate

of the microplate system increases with agitation speed and thus the cells had higher growth

rate at 400 RPM.

The cells were grown on different ratios and concentrations of sorbitol/methanol mixture.

Agitation speeds of 250 RPM and 400 RPM were used and OD600 of the cell culture was

measured after 12 or 6 hours of cultivation, respectively. As shown in Fig 3.4B and 3.4C, when

P. pastoris was cultured in a medium containing 5•10-4 mol or 10•10-4 mol carbons, OD600 of

the cell culture was not affected by the ratio of Cmethanol:Csorbitol. When higher carbon

concentration such as 20•10-4 mol or 30•10-4 mol was used, cells grown on carbons with higher

sorbitol ratio had higher OD600 values. The cell growth was significantly inhibited by 30•10-4

mol methanol but was not affected by the concentration of sorbitol (5~30•10-4 mol).

The result was consistent with previous reports where high concentration of sorbitol did not

inhibit cell growth (Çelik et al., 2009), whilst over 0.4% (v/v) methanol significantly inhibited

cell growth (Trinh et al., 2003, Zhang et al., 2000). Real time monitoring of methanol

concentration is critical at large scale to avoid its over-accumulation (Ramon et al., 2004).

Using sorbitol as a co-substrate will minimize the challenge of methanol monitoring and makes

process control easier.

Aprotinin expression was measured using the enzymatic assay. However, no product was

detected by the assay which may be due to the short cultivation time and low biomass

concentration.

Page 70: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

70

Figure 3.4 Study of cell growth on different sorbitol/methanol mixtures. P. pastoris cells were inoculated into medium to obtain an initial OD600

of 1.0. 2 ml medium was added into each well of microplate. (A) 5•10-4 mol CMethanol was added to each well and the plates were cultured at two

agitation speed, 250 RPM and 400 RPM, respectively. OD600 of cell culture was measured using spectrophotometer. (B) Cells were cultured on

medium with 5•10-4, 10•10-4, 20•10-4, 30•10-4 mol of carbon, respectively. The carbon contains different ratios of Cmethanol:Csorbitol ranging from

100:0 to 10:90. After the plate was agitated at 250 RPM for 12 hours, OD600 of cell culture was measured. (C) After the plate was agitated at 400

RPM for 6 hours, OD600 of cell culture was measured. Three replicates were used for each point and data was shown by mean�SD (n=3).

0 4 8 12 16 200

10

20

30

40

Time h

OD

600

nm

250RPM 400RPMA

1 2 3 4 5 6 7

24

28

32

36

!

!

! !

!!

!

! !

!

! !! !

!

!

! !

!!

!!

!

! ! !

!

!

Cmethanol:Csorbtol

OD

600

nm

5*10-4 mol C!10*10-4 mol C!

20*10-4 mol C!30*10-4 mol C!

M 85:15 70:30 55:45 40:60 25:75 10:90

B

1 2 3 4 5 6 7

24

28

32

36

!

!

!

!

!!

!

!

!!

! !! !!

!

! !!

!!! !

! !

!!

!

Cmethanol:Csorbtol

OD

600

nm

5*10-4 mol C!10*10-4 mol C!

20*10-4 mol C!30*10-4 mol C!

M 85:15 70:30 55:45 40:60 25:75 10:90

C

Page 71: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

71

The optimal ratio of sorbitol/methanol mixture for product expression was not obtained

from the cell culture in microplate. Therefore, a mixed induction strategy developed in

a previous study was used (Woodhouse, 2016). The sorbitol/methanol (1:1, C-mol/C-

mol) mixed induction effectively induced product expression and the fermentation was

scaled up to 20 litre scale.

The cell growth on sorbitol/methanol (1:1, C-mol/C-mol) mixture was further evaluated

at shake flask scale. In 250 ml shake flasks, P. pastoris cells were cultured in the

buffered complex medium supplemented with 1% (v/v) glycerol, methanol, sorbitol or

sorbitol/methanol (1:1, C-mol/C-mol) mixture. As shown in Fig 3.5, cells grown on

glycerol had the highest growth rate and OD600 of the cell culture reached ~50 after 48

hours of cultivation. It was consistent with a previous report where cells grown on

glycerol had higher specific growth rate than that grown on methanol (Cos et al., 2005).

Cell growth rates on methanol, sorbitol and sorbitol/methanol (1:1, C-mol/C-mol)

mixture were comparable. Biomass concentrations at the harvesting time were also

similar, which was different from the study in microplate cell culture. This is possible

because the shaker was not humified in the shake flask culture and evaporation was

more significant.

DNA release to the culture medium is a key factor indicating cell viability (Newton et

al., 2016). After 48 hours of culture, released DNA in the supernatant was quantified

using a PicoGreen dsDNA assay kit. As shown in Fig 3.9A, a maximum amount of

DNA was released when the cells were grown on methanol, while the DNA release was

negligible when glycerol was used. Compared to cell culture on methanol, growing

cells on sorbitol/methanol (1:1, C-mol/C-mol) mixture apparently reduced the DNA

concentration in medium.

Soluble protein profile in the medium was analysed using protein gel. After 24 hours

of culture, protein profiles from different carbons were similar. Much more proteins

were released after 48 of hours. Protein profile from glycerol culture was much simpler

than the others, whilst protein profile from methanol culture was the most complex. It

indicates that using methanol as the solo carbon caused cell lysis in late stage of cell

culture and using sorbitol/methanol mixed carbons reduced cell lysis.

Aprotinin concentration in the medium was determined using the enzymatic assay as

established in the Section 3.3.1, Chapter 3. However, it was undetectable due to its low

concentration.

Page 72: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

72

Figure 3.5 Growth and lysis of P. pastoris cells cultured on medium containing different carbons. P. pastoris cells were inoculated into medium

to obtain an initial OD600 of 1.0 ml and 75 ml medium was added into each shake flask. 1% (v/v) of glycerol, methanol, sorbitol or sorbitol/methanol

mixture (1:1, C-mol/C-mol) was added into each flask and flasks were agitated at 250RPM. (A) OD600 of cell culture was measured after 0, 12,

24, 36 and 48 hours of cultivation. One sample was analysed for each point. (B) Released DNA in medium was quantified using Picogreen DNA

assay after 48 hours of cultivation. Three replicates were used for each point and data was shown by mean�SD (n=3). (C) After 24 and 48 hours

of cultivation, soluble protein profile was analyzed using protein gel. 7.5 μl of sample was loaded into each well.

12 24 36 480

20

40

60

Time h

OD

600

nm

Glycerol MethanolMixture Sorbitol

A

1 2 3 4

0

25

50

75

100

ds D

NA

µg•ml-1

Glycerol Sorbitol Methanol Sor/Met

B

Page 73: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

73

3.3.3 Cell culture at bioreactor scale

Methanol and sorbitol/methanol (1:1, C-mol/C-mol) mixed induction strategies were

compared at bioreactor scale. The bioreactor has one litre total volume. Fermentation

was done using the procedure recommended by Invitrogen (Invitrogen, 2002). P. pastoris has a high cell growth rate by culturing on glycerol. Therefore, cells were

initially grown on 4% (m/v) glycerol in the medium. After glycerol in medium was

consumed, 50% (m/v) glycerol was fed to further enhance biomass. Production was

induced by feeding methanol or sorbitol/methanol (1:1, C-mol/C-mol) mixture at a

constant rate of 10.8 ml•L-1•h-1 (270 mmol Carbon•L-1•h-1). The cell growth, respiration

and viability were studied.

As shown in Fig 3.6, the initial glycerol in medium was consumed after 18 hours of

culture and DCW reached 25.8 g•L-1 and 24.7 g•L-1, respectively. DCW was further

enhanced to 57.9 g•L-1 and 57.8 g•L-1 after 50% (m/v) glycerol feeding. It is known that

glycerol inhibits pAOX1 activity (Tschopp et al., 1987). Thus, cells were starved for

one hour after stopping glycerol feeding to thoroughly exhaust residual glycerol. In

early stage of methanol induction, an adaptation phase which lasted for about 4 hours

was observed where biomass did not increase. No adaptation was noticed in the mixed

induction and biomass increased immediately once sorbitol/methanol (1:1, C-mol/C-

mol) mixture was fed in. Biomass reached a stationary phase after 72 hours of

cultivation. At the harvest time, DCW reached 129.4 g•L-1 and 152.4 g•L-1 in methanol

and mixed induction, respectively.

AOX enzyme oxidises methanol to formaldehyde in peroxisome. AOX enzyme

synthesis took several hours after methanol was fed instead of glycerol (Cámara et al.,

2017). In contrast, sorbitol is utilized by the same pathway as glycerol and thus no

adaptation phase was observed when the mixed induction was used. It has been reported

that sorbitol co-substrate complemented energy production and increased methanol

fraction in biosynthesis pathway (Çelik et al., 2010). Therefore, biomass in mixed

induction was higher than that in methanol induction at the harvest time.

Specific growth rate (μ) was calculated using Equation 3.1 (Clarke et al., 2011) and

shown in Fig 3.5C. μ reached maximum (0.18 h-1) in the end of glycerol feeding and

dropped to below 0.05 h-1 after 72 hours of fermentation. As the accumulation of

biomass, carbons available per biomass became very limited in the end of fermentation

and thus the cell growth nearly stopped.

Page 74: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

74

Figure 3.6 Comparison of methanol and sorbitol/methanol mixed induction strategies at bioreactor scale. Cells were grown on glycerol and then

induced by methanol (A) or sorbitol/methanol (1:1, C-mol/C-mol) mixture (B). OD600, WCW and DCW of cell culture were measured at each

sampling time. In measurement of WCW and DCW, three replicates were used for each point and data was shown by mean�SD (n=3). (C) Specific

growth rate (μ) was calculated using Eq 3.1 based on data of DCW.

0 24 48 72 96 1200

200

400

600

800

1000

0

40

80

120

160

200

Time h

Abs

orba

nce

at 6

00nm

WC

W g

•L-1

OD600nm WCW DCW

Induction

DC

W g•L

-1

A

0 24 48 72 96 1200

200

400

600

800

1000

0

40

80

120

160

200

Time h

Abs

orba

nce

at 6

00nm

WC

W g

•L-1

OD600nm WCW DCW

Induction

DC

W g•L

-1

B

0 24 48 72 960.00

0.05

0.10

0.15

0.20

Time (h)

Spe

cific

gro

wth

rate

(h-1

)

Methnaol MixtureC

Page 75: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

75

During the fermentation, oxygen fraction in gas-in was recorded by control system of

the bioreactor. Gas exhaustion of the fermentation was measured by a gas analyser.

Oxygen uptake rate (OUR) and CO2 evolution rate (CER) were calculated using Eq 3.3

and Eq 3.4.

As shown in Fig 3.7. OUR of two fermentations were similar prior to induction. An

OUR spike was observed in methanol induction in the first 24 hours of induction while

the spike in the mixed induction was not apparent. In both of the induction methods,

the OUR maintained stable after 24 hours of induction. The average OUR were 256.42

mmol•L−1•h−1 and 150.2 mmol•L−1•h−1, respectively.

An CER spike was also noticed in the CER profile of methanol induction (Fig 3.7B).

The CER dropped to about 78 mmol•L−1•h−1 after the spike and slightly increased to

100 mmol•L−1•h−1 during the induction. Compared to the methanol induction, the

sorbitol/methanol (1:1, C-mol/C-mol) mixed induction also had a lower average CER

value (79 mmol•L−1•h−1 versus 96 mmol•L−1•h−1).

P. pastoris was unable to metabolize methanol in the first few hours of induction and

thus methanol was accumulated in the medium. After cells adapted themselves to

methanol, large amount of oxygen was required to oxidize the accumulated methanol

and caused the OUR and CER spikes. Methanol accumulation was reduced by using

mixed induction and thus the spikes was not apparent. The difference of OUR spike in

two induction strategies was consistent to another study (Woodhouse, 2016).

The CER increased slightly in both of the inductions. It is because more carbons were

used for cell maintenance as the increase of biomass concentration. Compared to the

methanol induction, the OUR was reduced by 41.4% in the mixed induction. Sorbitol

has a relative lower enthalpy of combustion than methanol (-504.3 kJ•C-mol−1 versus

−727 kJ•C-mol−1) and thus the mixed induction reduced oxygen uptake rate (Niu et al.,

2013, Çelik et al., 2009). Large scale bioreactors have average OTR of 150-250

mmol•L−1•h−1 and it is quite challenging to maintain cell growth with an OUR over 250

mmol•L−1•h−1 (Carly et al., 2016). Using sorbitol/methanol mixed induction will make

scale-up of P. pastoris fermentation easier.

From the aspect of energy metabolism, sorbitol is a better carbon source than methanol

because 1mol CSorbitol produced more ATPs than 1mol CMethanol (2.6 mol versus 2.0 mol)

(Niu et al., 2013). As a result, less carbon was used for energy metabolism and less CO2

was produced in mixed induction.

Page 76: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

76

Figure 3.7 Comparison of OUR (A) and CER (B) of the fermentations from methanol

and sorbitol/methanol (1:1, C-mol/C-mol) mixed induction strategies. The OUR and

CER were calculated using Eq 3.2 and Eq 3.3.

0 24 48 72 960

200

400

600

800

1000

Time h

OU

R mmol•L-1•h-1

Methanol MixtureA

Induction

0 24 48 72 960

50

100

150

200

Time h

CE

R m

mol

•L-1

•L-1

Methanol Mixture

Induction

B

Page 77: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

77

Carbon utilization in methanol and sorbitol/methanol (1:1, C-mol/C-mol) mixed

induction was studied (Fig 3.8). Sorbitol and methanol concentrations in the medium

was determined after 18, 42, 66 and 90 hours of induction. Methanol concentration was

nearly undetectable in the fermentation induced by either methanol or

sorbitol/methanol (1:1, C-mol/C-mol) mixture. Around 2.0 g•L-1 of residual sorbitol

was detected during mixed induction.

Sorbitol uptake rate is much lower than that of methanol in P. pastoris. In a bioreactor

with maximum kLa values of 800 h-1, maximum methanol and sorbitol uptake rates

were 0.077 g•g DCW−1•h−1 and 0.044 g•g DCW−1•h−1, respectively (Carly et al., 2016).

Slower uptake rate resulted in the slight accumulation of sorbitol. However, its effect

on process control is negligible since sorbitol does not inhibit production even at a

concentration over 50 g•L−1 (Çelik et al., 2009).

Figure 3.8 Concentration of residual carbons during methanol and mixed induction.

Samples were taken after 18, 42, 66 and 90 hours of induction and concentrations of

methanol and sorbitol were determined using HPLC. One sample was analysed for each

point.

0 24 48 72 960.0

0.5

1.0

1.5

2.0

2.5

Time of induction h

Res

idua

l car

bon g•L-1

Methanol Mixture (Residual methanol)Mixture (Residual sorbitol)

Page 78: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

78

As found in shake flask study, the cells grown on methanol had more lysis than that on

glycerol or sorbitol (Fig 3.5). Here viabilities of the cell culture induced by methanol

and sorbitol/methanol (1:1, C-mol/C-mol) mixture were compared using flow

cytometry. Propidium iodide, a red-fluorescent nuclear staining, was used to stain cells.

The staining is impenetrable to live cells but is capable to enter nucleus and bind DNA

when cell membrane is not intact.

Fresh cells were incubated with the propidium iodide and analysed by flow cytometry.

As shown in Fig 3.9, the cell viabilities were over 98% before induction. After 18 hours

of induction, the viability dropped to 95.8% in methanol induction which was much

lower than that in mixed induction. After 90 h induction, the cell viabilities in methanol

and mixed induction were 92.5% and 97.6%, respectively.

The finding was consistent with a previous report where using sorbitol as a co-substrate

enhanced cell viability of a P. pastoris strain expressing recombinant interferon-α

(Wang et al., 2010). The enhancement of cell viability will minimize HCPs and

proteases co-released with the product during induction.

Figure 3.9 Comparison of the cell viabilities in methanol and sorbitol/methanol (1:1,

C-mol/C-mol) mixed induction strategies. Samples were taken after cells were induced

by methanol or sorbitol/methanol (1:1, C-mol/C-mol) mixture for 18, 42, 66, 90 and

114 hours. After cells were spun down by centrifuge and stained by propidium iodide

in 0.9 (m/v) NaCl solution, cell viabilities were measured by flow cytometry. One

sample was analysed for each point.

0 24 48 72 9690

92

94

96

98

100

Induction time h

Cel

lula

r via

bilit

y %

Methanol Mixture

Page 79: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

79

3.3.4 Impact of mixed induction on product yield and purity

After getting an understanding of how the sorbitol/methanol mixed induction affected

cell growth and viability, its impact on the product yield and purity was further studied.

Aprotinin in the supernatant was quantified using the method described in the Section

2.9.1, Chapter 2. HCPs co-released with the product were studied using 2D gel and LC-

MS/MS.

In a previous study, protein gel analysis showed that cells produced less aprotinin when

mixed induction was used (Woodhouse, 2016). Here aprotinin expression was studied

by protein gel and enzymatic assay. Samples were analysed using protein gel after they

were induced for 24, 48, 72 and 96 hours. In both methanol and mixed induction,

aprotinin bands were visible after 24 hours of induction and product amount kept

accumulating until harvest. Meanwhile, only a few HCP bands were visible on the gel

which indicated the high product purity. The major HCP band has a molecular weight

of ~65 kDa and it also accumulated over time during induction.

It was reported that the major contaminating HCP in P. pastoris culture was an

extracellular protein X1 (EPX1) with a molecular weight of 65 kDa (Heiss et al., 2013).

Expression of EPX1 was affected by temperature, osmolality and substrates. It suggests

that changing induction condition affects product purification by affecting the secretion,

especially when the product has similar properties as EPX1.

In order to quantitatively compare product yields of two induction strategies and

determine the best harvest time of fermentation, product levels in two induction

strategies were determined using the enzymatic assay. Cells were induced for 96 h and

samples were taken after 18, 36, 42, 60, 66, 84 and 90 hours of induction, respectively.

As shown in Fig 3.10B, the product titre of methanol induction increased from 0.09

g•L-1 to 1.69 g•L-1 from 18 h to 90 h, while it raised from 0.20 g•L-1 to 1.05 g•L-1 in

mixed induction. Product yields of two induction strategies were comparable from 18

h to 42 h. However, it nearly stopped increasing after 60 h in mixed induction, while it

raised from 1.24 g•L-1 to 1.69 g•L-1 in methanol induction. Complete depletion of

methanol is not preferable during P. pastoris induction because it impairs pAOX1

activity (Vogl et al., 2018). As the accumulation of biomass, methanol available per

biomass became limited in mixed induction and thus product synthesis nearly stopped.

Since product yield nearly stopped increasing after 72 hours of induction, it is a good

time to harvest products (Heiss et al., 2013).

Page 80: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

80

Figure 3.10 Comparison of product yields of methanol and sorbitol/methanol (1:1, C-

mol/C-mol) mixture induced fermentations. (A) Samples were taken after cells were

induced for 24, 48, 72 and 96 hours. Soluble proteins in supernatant were analyzed

using protein gel. 5 μl sample was loaded into each well. Aprotinin was shown by the

symbol of arrow. (B) Samples were taken after cells were induced for 18, 36, 42, 60,

66, 84 and 90 hours. Aprotinin in medium was quantified using the enzymatic assay.

Three replicates were used for each point and data was shown by mean�SD (n=3).

0 24 48 72 960.0

0.5

1.0

1.5

2.0

Induction time h

Apr

otin

in g

•L-1

Methanol MixtureB

Page 81: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

81

While high density fermentation enhances product expression, it also secretes a large

amount of host cell proteins (HCPs). Besides, intracellular proteins are release into the

medium and especially in the late stage of fermentation. As shown in the cell viability

study (Fig 3.9), nearly 8.0% of P. pastoris cells died at the harvest time in methanol

induction. The cell mortality is not very high compared to that in CHO cell fermentation.

However, biomass of the dead cells was high after the high density was considered. In

terms of DCW, 10.5 g of cells were dead after 96 hours of methanol induction, while it

dropped to 3.0 g•L-1 when sorbitol/methanol mixed induction was used. Therefore, cell

cultures from the two induction strategies was likely to have different profiles of HCPs.

The profiles of HCPs were firstly studied by 2D protein gel to show a visible

distribution of all proteins.

Prior to the 2D gel assay, soluble proteins are required to be extracted from the culture

medium. The extraction is usually performed by mixing the culture medium with

solvents that reduce solubility of proteins. Cold acetone, trichloroacetic acid (TCA) and

ammonium persulfate solution are popularly used solvents. However, the protein

extraction from P. pastoris medium was not very successful which was probably

because of the high salt concentration. The ammonium persulfate solution failed to

extract any protein from the medium. And less than 10% (m/m) of proteins were

precipitated when acetone or 13% (v/v) TCA in water was used. Besides, it was

observed that acetone was prone to precipitate proteins with large molecular weights

while 13% (v/v) TCA in water extracted more proteins with small molecular weights,

as it was analysed by protein gels. In order to enhance the extraction, the culture

medium was changed by ultrafiltration with a cut-off of 3000 Da. A new solvent was

made by preparing 13% (v/v) TCA in acetone. The extraction rate was enhanced to over

40% (v/v) using the new protocol. During the extraction, 20mM dithiothreitol (DTT)

was added to prevent any proteolytic degradation. Finally, the precipitation was re-

dissolved by urea solution.

Isoelectric points (pI) of most nature proteins are in the range of 4-7 (Ciborowski and

Silberring, 2016) which are much lower than that of aprotinin (pI: 10.5). Therefore,

IPG strips with the pH range of 3-11 were used as the separating gel for first dimension.

Protein amount up to the loading capacity was used on each gel to visualize as many

spots as possible. In the second dimension, the proteins were separated by NuPAGE 4-

12% Bis-Tris Protein Gel based on their different molecular weights. Finally, the

protein gels were stained by coomassie blue.

Page 82: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

82

As shown in Fig 3.11, about eight spots were visible on the gel before induction. These

proteins had MW and pI within the range of 40~200 kDa and 3.0~6.0. The gel was

quite clear outside this area. This finding was consistent to a previous report where P. pastoris grown on glucose only secreted a few proteins into the culture (Ciborowski

and Silberring, 2016).

After 24 hours of methanol induction, over 20 protein spots were visible on the gel. pI

of these proteins located within the range of 3.0~6.0. The spot of aprotinin was also

clear visible on the right bottom corner of the gel. On the contrast, protein spots were

fewer on the gel analysis of sorbitol/methanol mixed induction. After 72 hours of

methanol induction, intensity of aprotinin spot increased to a very strong level. Besides,

several new spots were detected within the pI range of 5.0~7.0.

In gel analysis of both induction strategies, the intensity of aprotinin spot was much

stronger than these of HCPs. Besides, the MW and pI of aprotinin are apparently

different from most of the HCPs. It indicates that the secreted product has a high purity

and most of the co-released HCPs are likely to be easily removed in chromatographic

steps.

Compared to the 2D gel analysis of CHO cell culture (Tait et al., 2012), the HCP profile

of P. pastoris was much simpler. One possible reason is that CHO cells are much easier

to lose cell integrity due to the lack of cell wall. As a result, a larger population of HCPs

can be released into the medium. In addition, the product had a high abundance in the

protein mixture as it was shown on the gel. The HCPs loaded on the gel had relatively

lower amount and thus many spots were not visible. Columns with equal affinity to

most proteins have been developed to remove the high-abundance proteins (Bellei et

al., 2011). In the future work, these columns can be applied to remove the aprotinin,

which is likely to increase the number of visible spots on 2D gel.

Page 83: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

83

Figure 3.11 Analysis of soluble proteins in supernatant using 2D protein gel. Samples were taken before induction and after cells were induced by

methanol or sorbitol/methanol (1:1, C-mol/C-mol) mixture for 24 and 72 hours. Proteins were recovered from the medium and re-suspended by

rehydration buffer. Buffer containing 200 μg protein was loaded on each gel. Protein spot of aprotinin was indicated by the symbol of arrow.

Page 84: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

84

Tandem mass spectrometry, also known as LC-MS/MS analysis, was used to identify

HCPs in the culture medium. The analysis mainly has two approached: top–down and

bottom–up analyses (Feist and Hummon, 2015). The top-down approach analyses the

whole proteins and the bottom-up approach analyses the digested proteins. In both of

the approaches, proteins or peptides are separated by reversed-phase chromatography

based on polarities before they are ionized by an ion source. Mass-to-charge ratios of

the proteins or peptides are measured, and their species can be identified by comparing

with a standard database. The top-down method is used to identify a purified protein,

whilst the bottom-up approach can analyse a protein mixture. Identification of HCPs in

P. pastoris culture using the bottom-up approach has been reported (Huang et al., 2011).

In this report, the protein mixtures were digested to peptides using trypsin before being

analysed by LC-MS/MS. However, trypsin would be inhibited by the product in this

study. Thus, the protein mixtures from P. pastoris culture were digested by acid

cyanogen bromide (CNBr) in 70% (v/v) formic acid.

As shown in Table 3.1, 96 proteins and 393 peptides were identified from the cell

culture of methanol induction. The numbers decreased to 72 and 262 in the culture of

mixed induction. Besides, more types of protease were identified from the culture of

methanol induction (3 verses 1). Name, MW, pI, localization and function of the

identified proteins were obtained from the database of Uniport and listed in Appendix

Table 8.1.

Fig 3.12 shows the number of identified HCPs located at different cell substructures.

Proteins were identified from the secretion, cell wall, cell membrane and intracellular

organelles. Expect the unknow proteins, proteins from the cytoplasm had the largest

number, while proteins from secretion, cell wall and membrane are quite fewer.

Compared to the mixed induction, the cell culture from methanol induction contained

more intracellular proteins from cytoplasm and nucleus, which indicated that more cells

lost integrity in the methanol induction.

The result was comparable to a previous study where 75 proteins were identified from

a P. pastoris culture producing a protein named Sm14-C64V (Huang et al., 2011). As

reported by Huang and co-authors, both intracellular and extracellular proteins were

identified, and proteins from cytoplasm had the largest number. Compared to CHO cell

culture, the P. pastoris has much simpler HCP profiles. The number of HCPs in CHO

culture was nearly 500 (Tait et al., 2012). It shows the advantage of P. pastoris as a

production host.

Page 85: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

85

Protein identified

Peptide identified Protease

Methanol 96 393 3

Mixture 72 262 1

Table 3.1 The number of host cell proteins, peptides, proteases and stress related

proteins identified from the methanol and sorbitol/methanol (1:1, C-mol/C-mol) mixed

induction strategies.

Figure 3.12 Localization of the HCPs identified from methanol and sorbitol/methanol

(1:1, C-mol/C-mol) mixed induction strategies. Localization of these protein was

searched from the database of Uniprot.com.

Secre

ted

Cell W

all

Membra

ne

Cytoplas

m

Nucleus

Unknow

0

20

40

60

Numbers

Methanol Mixture

Page 86: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

86

In order to study the impact of mixed induction on purification, distributions of

molecular weight (MW) and isoelectric point (pI) of these proteins were compared (Fig

3.13).

57 proteins were identified in both of the cultures from methanol and mixed induction.

39 proteins were only identified from the culture of methanol induction, while the

number dropped to 15 in the mixed induction. Within the MW range of 0~24 kDa, 8

proteins were only identified from the methanol induction, while only one was from

the mixed induction. Within the MW range of 24~80 kDa, 26 and 12 unique proteins

were identified from the two cultures, respectively. These two ranges can be defined as

‘sweet ranges’ where products from the mixed induction are easier to be purified.

Within the pI range of 6.0~14, the number of unique proteins in two cultures were 21

and 4, respectively. The range can be defined as a ‘sweet range’ for the ion exchange

chromatography.

This study indicated that using sorbitol/methanol mixed induction is an efficient tool to

simplify purification of products with specific MWs and pIs. For instance,

commercialized products including aprotinin (MW/pI, 6.5 kDa, 10.5), Glucagon-like

peptide-1 (MW/pI, 3.3 kDa, 9.3), Interferon gamma (MW/pI, 18.0 kDa/8.72),

Interferon beta (MW/pI, 22.0 kDa/ 9.69) and Keratinocyte growth factor (MW/pI, 22.5

kDa/9.29) locates within the ‘sweet range’ of MW and pI (0~24 kDa, 6.0~14). Some

antibody fragments, (MW/pI, 50 kDa and ~8.5) are within the ‘sweet range’ of MW

and pI (24~80 kDa, 6.0~14). The HCP profiles from different cell strains are likely to

be different. In the future, it will be interesting to compare the HCPs of methanol and

mixed induction using other strains and verify if the ‘sweet range’ will change.

Page 87: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

87

Figure 3.13 Distribution of the molecular weights (A) and isoelectric points (B) of

HCPs identified from methanol and sorbitol/methanol (1:1, C-mol/C-mol) mixed

induction strategies.

Page 88: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

88

3.4 Conclusion

In this chapter, an enzymatic assay of aprotinin that was compatible with culture

mediums of P. pastoris was developed. The undiluted medium interfered accuracy of

the enzymatic assay, but the interference could be minimized by diluting the medium

sufficiently. Cell culture at microplate scale showed that sorbitol did not inhibit cell

growth even at high concentrations, while methanol did. At shake flask scale, cell lysis

was reduced by growing cells on sorbitol rather than methanol. Aprotinin produced at

the microplate and shake flask scales were undetectable which was likely due to the

short cultivation time and low biomass.

Standard methanol induction and sorbitol/methanol mixed induction were compared

using parallel bioreactor. At bioreactor scale, sorbitol/methanol mixed induction

improved cell growth and cell viability. The mixed induction also benefited the

fermentation by reducing oxygen uptake rate. Since it is quite challenging to provide

sufficient oxygen at large scale, scaling-up P. pastoris fermentation with mixed

induction will be easier.

The mixed induction decreased product yield compared to standard methanol induction.

But fewer host cell proteins were co-released with the product into medium when the

mixed induction was used. Therefore, the mixed induction is likely to simplify the

product purification. Besides, fewer proteases were identified in the culture from the

mixed induction and thus product degradation will be minimized by using mixed

induction when the products are sensitive to proteolytic degradation.

Page 89: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

89

Chapter 4 Development of induction strategies in a bioreactor with limited oxygen transfer rate

4.1 Introduction

It was shown that standard methanol induction achieved higher product yield compared

to the sorbitol/methanol mixed induction. However, supplying sufficient oxygen for

standard methanol induction is challenging in large-scale bioreactors.

Methanol is a high degree reductant with high heat of combustion. It consumes 0.8~1.1

mole O2 to utilize one mole methanol by P. pastoris (Jahic et al. 2003).

Correspondingly, the bioreactor requires an OTR value over 230~290 mmol•L-1•h-1 for

standard methanol induction where it is fed at the rate of 10.8 ml•L-1•h-1 (Invitrogen

2002). However, the large-scale bioreactors only have an average OTR of 150~200

mmol•L-1•h-1 (Benz 2011). Using standard methanol induction in the large-scale

bioreactors will cause oxygen depletion and methanol accumulation.

In order to maintain the DO at 20%~30% and avoid methanol over-accumulation, the

methanol feeding rate has to be declined in the OTR-limited bioreactors. Consequently,

the cell growth will be decreased and product yield may be compromised compared to

the standard methanol induction (Maccani et al., 2014). Sorbitol has a relatively lower

enthalpy of combustion and its metabolism by P. pastoris consumes less oxygen.

Therefore, more carbons can be fed by using sorbitol/methanol mixed induction in the

OTR-limited bioreactor. It is likely to enhance the cell growth and product yield

compared to slower methanol induction.

In this paper, several induction strategies (Table 4.1) were compared in a benchtop

bioreactor that has a similar oxygen transfer rate (OTR) as large-scale ones.

1) kLa values of the vessels were measured using dynamic method and a fermentation

process with OTR of 150 mmol•L-1•h-1 was defined.

2) Induction in an oxygen limited condition was studied, and the impact of residual

methanol concentration on cell growth, viability and product yield was studied.

3) Induction in an oxygen unlimited condition was studied, and methanol and

sorbitol/methanol mixed inductions were compared.

4) AOX enzyme activities of the different induction strategies were compared.

Page 90: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

90

Abbr. of induction strategies

shown in figures Inducer feeding rate ml•h−1•L−1

Desired DO%

Pure methanol induction 100% Methanol 6.7 30%

Sorbitol/methanol (60% Cmethanol) mixed induction 60% Methanol 7.7 30%

Sorbitol/methanol (50% Cmethanol) mixed induction 50% Methanol 8.0 30%

Sorbitol/methanol (40% Cmethanol) mixed induction 40% Methanol 8.3 30%

Induction with residual methanol of 0~5 g•L-1 Met0-5 Variable 0%

Induction with residual methanol of 5~10 g•L-1 Met5-10 Variable 0%

Table 4.1 A summary of the induction strategies used in the bioreactor with an OTR of 150 mmol•L−1•h−1

Page 91: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

91

4.2 Theoretical considerations

4.2.1 Determining OTR of the bioreactor

Determining kLa of the bioreactor is required to define a fermentation process with a

limited OTR. Methods to measure the kLa include dynamic method, OTR measurement

and sulphite method, etc (Schlüter and Deckwer, 1992). Dynamic method is commonly

used when the bioreactor has no biological oxygen uptake. This method was used to

measure the kLa of one litre bioreactor in this study. In the measurement, dissolved

oxygen in the medium is eliminated to zero by sparging nitrogen into the vessels. Then

air is ventilated at a constant rate and dynamic change of the dissolved oxygen is

measured. The kLa values were determined from the slope of ln f(CL) and time graph

as described in Eq 4.1 (Garcia-Ochoa and Gomez, 2009).

k"a = −1tln(1 −

C,C∗) Eq 4.1

Where C* is the saturated concentration of dissolved oxygen in % when air is aerated;

Ct is the concentration of oxygen concentration in % at time t.

The kLa is influenced by conditions such as temperature, air flowrate and agitation. The

same temperature and air flowrate are usually used for kLa measurement and cell culture.

When the desired OTR cannot be reached with the maximum agitation, enriching the

air inlet with pure oxygen is required. The oxygen fraction in air inlet can be calculated

using Eq 4.2 and Eq 4.3. Both air and pure oxygen flows are connected to the bioreactor

and the desired oxygen fraction can be achieved automatically by control system of the

bioreactor.

C∗∗ =OTRk"a

− C" Eq 4.2

p34 = p34(567)C∗∗C34

Eq 4.3

Where OTR is defined as 150 mmol•L-1•h-1; C** is the theoretical concentration of

dissolved oxygen in mmol•L-1 when air/O2 gas mixture is aerated; CL is the

concentration of dissolved oxygen in mmol•L-1 in medium; pO2 is the maximum

fraction of oxygen in % in gas inlet; pO2(air) is the oxygen fraction in % in air; CO2 is the

saturated concentration of dissolved oxygen in mmol•L-1 at 30�.

Page 92: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

92

4.2.2 Determining feeding rates of the induction solutions

From average OUR of a fermentation where pure methanol was fed at a rate of 10.8

ml•L-1•h-1 (270 mmol Carbon•L-1•h-1), it is estimated that utilizing per millilitre

methanol consumed 22.2 mmol O2 by P. pastoris. Based on enthalpies of combustion

of methanol and sorbitol (-504.3 kJ•C-mol−1 versus −727 kJ•C-mol−1), it is derived that

15.4 mmol O2 is consumed by each millilitre 0.75 g•L-1 sorbitol solution. Pure methanol

and sorbitol/methanol mixtures with 60%, 50% and 40% (C-mol/C-mol) of methanol

were prepared and used as induction solutions. Their feeding rates were determined

using Eq 4.4

F =1000 • OTR

22.2r>?, + 15.4rCD7 Eq 4.4

Where F is the feeding rate of induction solution; OTR is defined as 150 mmol•L-1•h-1;

rMet and rSor represent fractions of Cmethanol and Csorbitol in the induction solution.

4.3 Results and discussion

4.3.1 Measurement of oxygen transfer coefficient

In order to define a fermentation process with an OTR of 150 mmol•L-1•h-1, kLa of the

Infors bioreactor was measured using the dynamic method. Dynamic method is ideal

to study the change of kLa under different operation conditions (Garcia-Ochoa and

Gomez, 1998, Garcia-Ochoa and Gomez, 2009). In this method, liquid is deoxygenated

and oxygenated by nitrogen and air flows, respectively. DO profile is recorded in

oxygenation phase, and the kLa value is determined from the slope of ln(1-Ct/C*) vs

time graph.

The bioreactor has four one-litre vessels. Each vessel has two impellers with the

diameter of 0.038 m. The same DO probe was used for the kLa measurement of four

vessels. Response time of the DO probe was determined by putting the probe in a beaker

containing deoxygenated water (DO = 0%) and transferring it rapidly to the vessel that

was filled with oxygenated water (DO=100%). The response time is defined as the time

period that the reading reached ~80%. The probe was determined to have a response

time of 2.6 sec.

Page 93: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

93

Each vessel was filled with 700 ml BSM medium before the kLa measurement.

Temperature and air flow were set at 30� and 1.4 vvm. Fig 4.1A shows the change of

DO profile with time during kLa measurement. From 5 sec to 45 sec, the DO dropped

linearly from 105.6% to 13.7%. It nearly dropped to 0% after 50 s of deoxygenation.

Once air flow was ventilated, DO rapidly increased to over 80% in 25 sec and then

slowly increased to 100% afterward.

kLa values were measured at agitation speeds of 300, 500, 800 and 1100 RPM,

respectively. As shown in Fig.4.1B, the kLa values of four vessels were similar at the

same agitation speed. The kLa enhanced from 50.3±7.5 h-1 to 391.3±25.3 h-1 when

agitation speed increased from 300 RPM to 1100 RPM. A linear equation with variables

of kLa and agitation speed was obtained: kLa=(0.42±0.08)N-(49.48±20.43), R2=0.94.

The kLa obtained in this study was comparable to that in previous reports (Garcia-

Ochoa and Gomez, 2004).

The impeller of Infor bioreactor has a maximum agitation speed of 1100 RPM. When

the air flow is ventilated at 1.4 vvm, the OTR is calculated to be 95.45 mmol•L-1•h-1.

In order to obtain an OTR of 150 mmol•L-1•h-1, it is calculated that the maximum

oxygen fraction in aeration should be set at 33% (v/v). During the fermentation, air

flow was maintained at the constant rate of 1.4 vvm. The agitation initially started from

300 RPM and gradually increased to 1100 RPM by correlating with DO. When the

desired DO could not be maintained with the maximum agitation, pure oxygen was

merged into the air flow until the oxygen fraction reached 33% (v/v).

Page 94: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

94

Figure 4.1 kLa measurement using dynamic method. (A) shows the representative

profile of DO in the kLa measurement. The profile is divided into deoxygenation and

oxygenation phases where N2 and air are ventilated, respectively. The vessel was filled

with 700 ml BSM medium. Temperature, aeration and agitation were set at 30�, 1.4

vvm and 300 RPM, respectively. (B) The kLa values of four vessels as a function of

agitation speed. Temperature and aeration were set at 30� and 1.4 vvm, and kLa was

measured at agitation speeds of 300, 500, 800 and 1100 RPM.

0 20 40 60 80 100 1200

20

40

60

80

100

120

Time sec

DO

% N2 O2

A

0 300 600 900 12000

100

200

300

400

500

Agitation RPM

k La

h-1

Bio-1 Bio-2

Bio-3 Bio-4

B

Page 95: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

95

4.3.2 Induction in an oxygen limited condition

When the OTR of bioreactor is constrained under 150 mmol•L-1•h-1, standard methanol

feeding will cause oxygen depletion and methanol accumulation. Although maintaining

DO at 20%~30% is desirable in most studies, induction in an oxygen limited condition

where DO dropped to nearly 0% (OLFB fermentation) was also reported. It even had

higher product yield and purity compared to oxygen unlimited fermentation

(Charoenrat et al., 2005, Khatri and Hoffmann, 2006). Besides, it was found that

residual methanol concentration affected both cell growth and product yield and the

effect depended on cell strains (Khatri and Hoffmann, 2006, Barrigón et al., 2013).

In this study, the OLFB fermentation was performed in the bioreactor with an OTR of

150 mmol•L-1•h-1. In the Chapter 3, it has been shown methanol over accumulation

inhibited the cell growth (Fig 3.). Therefore, concentration of residual methanol was

controlled by a feedback control system, and the OLFB fermentations with residual

methanol concentrations of 0~5 g•L-1 and 5~10 g•L-1 were studied, respectively.

Fig 4.2A shows the fermentation profile of OLFB fermentation with residual methanol

concentration of 0~5 g•L-1. The temperature and pH were maintained at 30� and 5.0

during the fermentation. The DO decreased from 100% to 30% after cells were

inoculated for several hours. The agitation gradually increased to 1100 RPM to

maintain the DO at 30%. Once glycerol feeding was started, the DO sharply decreased

from 30% to nearly 0% and maintained at nearly 0% during the glycerol feeding and

induction stages. Two spikes of agitation speed were observed which indicated the

consumption of glycerol in batch and fed-batch phases. Similar fermentation profile

was also observed in the OLFB fermentation with residual methanol concentration of

5~10 g•L-1.

Fig 4.2B shows the growth curves of OLFB fermentation with residual methanol

concentration of 0~5 g•L-1 and 5~10 g•L-1. The biomass concentration reached 49.4

g•L-1 and 56.2 g•L-1, respectively, before induction. When the residual methanol

concentration was maintained within 0~5 g•L-1, the DCW kept increasing in the whole

induction and reached 108.2 g•L-1 after 96 hours of induction. Whilst the DCW nearly

stopped increasing after 48 hours of induction and a final DCW of 79.1 g•L-1 was

obtained when the residual methanol concentration was maintained within 5~10 g•L-1.

Page 96: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

96

Figure 4.2 Fermentation profile (A) and growth curve (B) of the OLFB fermentation.

Change of temperature, pH, agitation speed and DO were shown in the fermentation

profile. Growth curves of P. pastoris in the OLFB fermentation with residual methanol

concentration of 0~5 g•L-1 and 5~10 g•L-1 were shown. Three replicates were used for

each point and data was shown by mean±SD (n=3).

0 24 48 72 96 1200

30

60

90

120

150

Time h

DC

W g•L-1

Met0-5 Met5-10

Inducton

B

Page 97: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

97

It is critical to control the residual methanol concentration in the OLFB fermentation.

The residual methanol can be measured by on-line and off-line tools. The online tool

such as methanol sensor based on photoelectric plethysmograph was reported

previously (Khatri and Hoffmann, 2006). The offline tools include HPLC, gas

chromatography and enzymatic reaction-based method (Minning et al., 2001,

Parpinello and Versari, 2000, Kučera and Sedláček, 2017).

Since on-line methanol sensor was not available in this study, an off-line HPLC based

method was used. Samples were taken intervally, and residual methanol was measured

by HPLC. Methanol feeding rate was turned up and down based on the residual

methanol concentration. As shown in Fig 4.3, the residual methanol concentration was

successfully controlled within the aimed ranges, 0~5 g•L-1 and 5~10 g•L-1. However, it

failed to maintain the residual methanol at a constant concentration. In a previous report,

residual methanol was maintained at constant level by using a feedback control system

with a commercial on-line methanol analyzer (MC168) (Ramon et al., 2004). The

residual methanol concentration was successfully maintained at 4.0 g•L-1 for over 60

hours. It was shown that the residual methanol concentration significantly affected cell

growth in the OLFB fermentation. Thus, using the on-line methanol analyser can be

considered in the future work.

Figure 4.3 Residual methanol concentration in the OLFB fermentations. Samples were

taken after 16, 24, 40, 48, 64, 72, 88 and 96 hours of induction and the methanol

concentration was determined using HPLC. One sample was analysed for each point.

0 24 48 72 960

3

6

9

12

Induction time h

Res

idua

l met

hano

l g•L-1

Met0-5 Met5-10

Page 98: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

98

Cell viabilities in the OLFB fermentations were studied. As shown in Fig 4.4, the cell

viabilities decreased to ~95.0% after 24 h of induction in both of the OLFB

fermentations. At the harvest time, viabilities decreased to 94.3% and 92.8%,

respectively, in the OLFB fermentation with the residual methanol concentration of

0~5 g•L-1 and 5~10 g•L-1.

At the harvest time, the cell viability in the OLFB fermentation with residual methanol

concentration of 5~10 g•L-1 was comparable to the viability of standard methanol

induction, as described in the Section 3.3.3, Chapter 3. The proteomic study in the

Section 3.3.4, Chapter 3 showed that loss of the viability caused release of the

intracellular HCPs and proteases. Therefore, the product purity is likely to be one of

the critical attributes of the OLFB fermentation.

The viability at harvest time was higher when the residual methanol concentration of

0~5 g•L-1 was used, which was consistent to a previous study. In a cell line expressing

recombinant scFv, less than 2.0% cells died when the residual methanol was maintained

below 3g•L-1, while the mortality increased to 4.0% when the residual methanol was

over 3g•L-1 (Khatri and Hoffmann, 2006). Therefore, it is necessary to control the

residual methanol concentration in OLFB fermentation which influences cell growth

and viability.

Figure 4.4 Change of the cell viability in the OLFB fermentations. The residual

methanol concentrations were maintained within 0~5 g•L-1 and 5~10 g•L-1, respectively.

Samples were taken after 16, 24, 40, 48, 64, 72, 88 and 96 hours of induction and cell

viability was measured by flow cytometry. One sample was analysed for each point.

0 24 48 72 9690

92

94

96

98

100

Time of induction h

Cel

lula

r via

bilit

y %

Met5-10Met0-5

Page 99: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

99

Product yields in the OLFB fermentations with residual methanol concentration of 0~5

g•L-1 and 5~10 g•L-1 were studied using protein gel and enzymatic assay. As shown in

Fig 4.5, the product bands were clearly visible after 24 h of induction in both of the

OLFB fermentations. The intensity of product bands kept increasing from 24 h to 96 h

induction. At the same time point, the intensity in the OLFB fermentation with residual

methanol concentration of 0~5 g•L-1 was higher than that with residual methanol of

5~10 g•L-1. Only a few HCP bands were visible on the gel which indicated the high

product purity. The major HCP band has a molecular weight of ~65 kDa, which is likely

to be the extracellular protein X1 (EPX1). Intensities of the EPX1 band were

comparable in both of the OLFB fermentations.

Result of the enzymatic assay was in agreement with the protein gel assay. The

aprotinin concentration kept increasing from 0.59 g•L-1 to 1.10 g•L-1 in OLFB

fermentation with the residual methanol concentration of 0~5 g•L-1 and increasing from

0.62 g•L-1 to 0.87 g•L-1 in the OLFB fermentation with the residual methanol

concentration of 5~10 g•L-1. The product yield was enhanced by 26% by maintaining

the residual methanol concentration below 5 g•L-1. Therefore, it is necessary to

optimize the residual methanol in OLFB fermentations to obtain a high product yield.

In future work, the impact of residual methanol concentration on product yield can be

better studied by maintaining methanol concentration at constant levels by using a

methanol sensor.

The impact of residual methanol concentration on product yield seems to vary on strains.

In one study, 3.0% (m/v) of methanol was the optimal concentration for the production

of recombinant scFV (Khatri and Hoffmann, 2006). However, 1.0% (m/v) of residual

methanol was observed to inhibit the expression of recombinant Rhizopus oryzae lipase

in another study (Barrigón et al., 2013).

Compared to the standard methanol induction as described in the Section 3.3.3, Chapter

3, the product yields in OLFB fermentations were reduced by 35% and 47%,

respectively. It is likely be attributed to the lower biomass concentration obtained in

the OLFB fermentations. The biomass concentrations at harvest time were reduced by

17% and 38%, respectively, compared to the standard methanol induction. However,

the high methanol feeding rate used in standard induction cannot be maintained in the

OTR-limited bioreactor. In that situation, the product yield in OLFB fermentation with

the residual methanol concentration of 5 g•L-1 looks acceptable.

Page 100: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

100

Figure 4.5 Product yields in the OLFB fermentations. The residual methanol

concentrations were maintained within 0~5 g•L-1 and 5~10 g•L-1, respectively. (A)

Samples were taken after cells were induced for 24, 48, 72 and 96 hours. Soluble

proteins in the supernatant were analyzed using protein gel. 5 μl sample was loaded

into each well. (B) Samples were taken after cells were induced for 16, 24, 40, 48, 64,

72, 88 and 96 hours. Aprotinin in the supernatant was quantified using the enzymatic

assay. Three replicates were used for each point and data was shown by mean�SD

(n=3).

0 24 48 72 960.0

0.4

0.8

1.2

Time h

Apr

otin

in g

•L-1

Met0-5 Met5-10B

Page 101: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

101

4.3.3 Induction in an oxygen unlimited condition

In the small scale fermentation, DO can be maintained at an optimal value, normally

20% or 30% of saturation (Singh et al., 2008, Invitrogen, 2002) by supplementing

sufficient pure oxygen into the gas inlet. However, it is difficult to be achieved at large

scale. In the bioreactor with a limited OTR, using the standard methanol induction will

decrease the DO to ~0%. Therefore, reducing the methanol feeding rate is required.

However, it may compromise the cell growth and product yield.

Based on the measured OUR of standard methanol induction (Fig 3.7), it is estimated

that consuming per millilitre methanol will require 22.2 mmol O2 by P. pastoris. In the

bioreactor with an OTR of 150 mmol•L-1•h-1, a maximum methanol feeding rate of 6.7

ml•L-1•h-1 can be used to maintain the DO at 30%. Compared to the standard methanol

induction, the feeding rate is declined by 38%.

Sorbitol has a lower enthalpy of combustion than methanol. Based on enthalpies of

combustion of methanol and sorbitol (-504.3 kJ•C-mol−1 versus −727 kJ•C-mol−1), it is

calculated that 15.4 mmol O2 is consumed by each millilitre 0.75 g•L-1 sorbitol solution.

Theoretically, partially replacing methanol with sorbitol is able to enhance the carbon

feeding rate while maintain the DO at 30%. As calculated by Eq 3.4, the rate can be

enhanced to 7.7 ml•L-1•h-1 by using sorbitol/methanol (60% Cmethanol) mixture. The

higher carbon feeding rate is likely to enhance the cell growth and product yield

compared to slower methanol induction. Besides, increasing sorbitol fraction in the

sorbitol/methanol mixture will further enhance the carbon feeding rate. The rate can be

enhanced to 8.0 ml•L-1•h-1 and 8.3 ml•L-1•h-1, respectively, by using sorbitol/methanol

(50% Cmethanol) and sorbitol/methanol (40% Cmethanol) mixtures. However, reducing the

methanol fraction it is likely to reduce the promoter activity, which will decrease the

product yield.

In the bioreactor with an OTR of 150 mmol•L-1•h-1, sorbitol/methanol mixtures with

100%, 60%, 50% and 40% (Cmethanol) of methanol were used for P. pastoris induction.

The DO is maintained at 30% by using the feeding rates shown in Table 4.2. Cell

growth, viability, product yield of these induction strategies was compared. Besides,

induction with the DO of 30% was compared with the OLFB fermentations, as

described in the Section 4.3.1, Chapter 4.

Page 102: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

102

Induction solution

Feeding rate

ml•L-1•h-1

100% Methanol 6.7

60% Methanol 7.7

50% Methanol 8.0

40% Methanol 8.3

Table 4.2 Feeding rates of different induction solutions

Fig 4.6A shows the representative fermentation profile of P. pastoris in the OTR

limited bioreactor. The cell culture was initially grown on glycerol and then induced by

pure methanol. Temperature and pH were maintained at 30� and 5.0 in the

fermentation. The DO decreased dramatically in the first few hours of fermentation.

Once it dropped to ~30%, the agitation speed increased gradually to maintain the DO

level. In the first 30 hours, two DO spikes were observed which indicated the

exhaustion of batch and fed glycerol. The agitation speed also decreased when the DO

spikes occurred. Once pure methanol was fed, the agitation immediately increased to

the maximum speed and worked at that speed in the whole induction. The DO was

successfully maintained at 30% when 6.7 ml•L-1•h-1 of pure methanol was fed. Similar

profiles were observed when sorbitol/methanol mixtures with 60%, 50% and 40%

(Cmethanol) of methanol were used.

Fig 4.6B shows the growth curve of P. pastoris in the OTR limited bioreactor. The

biomasses reached 42.9±3.0 g•L-1 after cell growth on batch and fed glycerol, and

different vessels had comparable values. When pure methanol was used, the DCW kept

increasing during the whole induction and reached 90.2 g•L-1 after 96 hours. The DCW

was enhanced to 114.6 g•L-1, which was 27% higher compared to methanol induction,

by using sorbitol/methanol mixture with 60% (Cmethanol) of methanol. It was further

enhanced to ~130.0 g•L-1 when the methanol fraction was declined to 50% and 40%

(Cmethanol).

Compared to the standard methanol induction described in the Section 3.3.2, Chapter

3, the DCW decreased by 30% when 6.7 ml•L-1•h-1 of pure methanol was used, whilst

comparable biomass was obtained by using sorbitol/methanol mixtures with 50% or 40%

(Cmethanol) of methanol. Compared to the OLFB fermentation with residual methanol

Page 103: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

103

concentration of 0~5 g•L-1, methanol induction with DO of 30% reduced the DCW by

16%, whilst the sorbitol/methanol (50%, Cmethanol) mixed induction enhanced it by 20%.

In the OTR-limited bioreactors, sorbitol/methanol mixed induction advantages over

methanol induction in terms of enhancing the biomass and maintaining the DO.

Figure 4.6 Fermentation profile (A) and growth curve (B) of P. pastoris in the oxygen

unlimited condition. After cell growth on glycerol, cells were induced by feeding

methanol at the rate of 6.7 ml•L-1•h-1. Growth curves of P. pastoris induced by limited

feeding of methanol or sorbitol/methanol mixtures. Three replicates were used for each

point and data was shown by mean�SD (n=3).

0 24 48 72 96 1200

30

60

90

120

150

Time h

DC

W g•L-1

100%Methanol 60%Methanol

50%Methanol 40%Methanol

B

Page 104: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

104

During the feeding of methanol and sorbitol/methanol mixtures, samples were taken

after 16, 24, 40, 48, 64, 72, 88, and 96 hours of induction. After the supernatant was

collected by centrifugation, residual methanol in the medium was measured using

HPLC. As shown in Fig 4.7, the residual methanol concentration was less than 1g•h-1

in both of the methanol and sorbitol/methanol mixed induction strategies. It was even

below the detecting limitation at some time points.

Differently from the OLFB fermentation (Fig 4.3), the methanol was nearly exhausted

when the DO was maintained at 30%. The residual methanol profile was more

comparable to that of standard methanol induction where the DO was also set at 30%

(Fig 3.8). As it is shown in the study of OLFB fermentation, the residual methanol

concentration could not be maintained stable without using online methanol sensor.

Although the HPLC based method is an alternative, it could only keep the residual

methanol within a specific range. There is the probability that the residual methanol

exceeds the desired range at some time points, which may cause failure of fermentation.

In that situation, it will make the process control much easier by maintaining the DO at

30% rather than at 0%. On the other aspect, nearly exhaustion of methanol may

compromise the promoter induction and decrease the product yield.

Figure 4.7 Residual methanol concentration in the oxygen unlimited condition. Pure

methanol and sorbitol/methanol mixtures with 60%, 50% and 40% (Cmethanol) of methanol were

used for induction. Samples were taken after 16, 24, 40, 48, 64, 72, 88 and 96 hours of

induction and residual methanol in the supernatant was measured using HPLC. One

sample was analysed for each point.

0 24 48 72 960

3

6

9

12

Induction time h

Res

idua

l met

hano

l g•L-1

100%Methanol 60%Methanol

50%Methanol 40%Methanol

Page 105: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

105

Cell viabilities in the methanol and sorbitol/methanol mixed induction strategies were

studied. As shown in Fig 4.8, the cell viabilities changed in a similar manner among

different induction strategies. The cell viabilities were over 98% in four vessels before

induction. They kept decreasing slowly with the feeding of induction solutions. After

96 hours of induction, an average viability of 97.5±0.6% was obtained. Although the

viability of sorbitol/methanol (60%, Cmethanol) mixed induction was lower, it has no

significant difference compared with viabilities of the other induction strategies.

Compared to the OLFB fermentations, induction with DO of 30% resulted in higher

cell viabilities at the harvest time (93.5±0.8 versus 97.5±0.6%, p=0.004). The

proteomic study in the Section 3.3.4, Chapter 3 showed that loss of the viability caused

release of the intracellular HCPs and proteases. Therefore, induction with DO of 30%

is likely to result in higher product purity comparing to the OLFB fermentations, which

can be validated in the future study.

Compared to standard methanol induction at the rate of 10.8 ml•L-1•h-1 as described in

the Section 3.2.3, Chapter 3, reducing methanol feeding rate to 6.7 ml•L-1•h-1 (167

mmol Carbon•L-1•h-1) significantly improved the cell viability. It is likely to be because

less toxic by-product was formed when the slower methanol feeding is used.

Figure 4.8 Cell viabilities in the oxygen unlimited condition. Samples were taken after

16, 24, 40, 48, 64, 72, 88 and 96 hours of induction by methanol or sorbitol/methanol

mixtures. Cell viabilities were measured by flow cytometry. One sample was analysed

for each point.

0 24 48 72 9690

92

94

96

98

100

Induction time h

Cel

lula

r via

bilit

y %

100%Methanol 60%Methanol

50%Methanol 40%Methanol

Page 106: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

106

Although sorbitol/methanol mixed induction enhanced cell growth in the OTR-limited

bioreactor, it would be not beneficial without increasing the product yield. The product

yields in the methanol and sorbitol/methanol mixtures induced fermentations were

studied using protein gel and enzymatic assay. As shown in Fig 4.9, the intensities of

product band in methanol and sorbitol/methanol (60%, Cmethanol) mixed induction

strategies were comparable to each other. When sorbitol/methanol mixtures with 50%

and 40% (Cmethanol) of methanol were used, the intensities were reduced. Only a few

bands of HCPs were visible in all the induction strategies. The major HCP band has a

molecular weight of ~65 kDa, which is likely to be the extracellular protein X1 (EPX1).

The product yields were quantitatively determined using the enzymatic assay. The

aprotinin concentration increased in a similar rate in the four inductions. It increased

from 0.29 g•L-1 to 0.85 g•L-1 when pure methanol induction was used and increased

from 0.29 g•L-1 to 0.79 g•L-1 when sorbitol/methanol (40%, Cmethanol) mixture was used.

After 96 hours of induction, the aprotinin concentration reached 0.85, 0.86, 0.78 and

0.79 g•L-1, respectively.

As described above, sorbitol/methanol (40%, Cmethanol) mixed induction enhanced the

biomass by 27% compared to the pure methanol induction. However, it reduced the

product yield by 8% (p=0.001). The specific productivity was much lower in the mixed

induction (0.062 mg•g DCW-1•h-1 versus 0.096 mg•g DCW-1•h-1). It is likely that the

sorbitol/methanol mixed induction compromised the induction of promoter. Since high

cell density increases challenges in product recovery (Salte et al., 2006, Lopes and

Keshavarz�Moore, 2012), induction using pure methanol is preferred rather than using

sorbitol/methanol mixtures. Impact of sorbitol/methanol mixed induction on product

synthesis depends on cell strains. It was reported that the specific productivity of β-

galactosidase was even slightly enhanced when sorbitol/methanol mixed induction was

used (Niu et al., 2013). If that cell line is cultured in an OTR-limited bioreactor, using

sorbitol/methanol mixed induction may enhance both biomass and product yield.

Compared to the OLFB fermentation with residual methanol of 0~5 g•L-1, the methanol

induction with DO of 30% reduced the product yield by 18%, while the

sorbitol/methanol mixed inductions reduced the yield by 26%. However, the protein

gel looks much clearer when the DO was maintained at 30%, which indicated a higher

product purity. Besides, the enhancement of viability may reduce the proteases released

into the medium.

Page 107: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

107

Figure 4.9 Product yields in the oxygen unlimited condition. (A) Samples were taken

after 96 hours of induction. Soluble proteins in the supernatant were analyzed using

protein gel. 10 μl sample was loaded into each well. Aprotinin was shown by the

symbol of arrow. (B) Samples were taken after 16, 24, 40, 48, 64, 72, 88 and 96 hours

of induction. Aprotinin in the medium was quantified using the enzymatic assay. Three

replicates were used for each point and data was shown by mean�SD (n=3).

16 24 40 48 64 72 960.0

0.2

0.4

0.6

0.8

1.0

Induction time h

Apr

otin

in g

•L-1

40%Methanol50%Methanol

60%Methanol100%MethanolB

Page 108: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

108

4.3.4 Measurement of AOX enzyme activity

In the last section, it was shown that partially replacing methanol with sorbitol reduced

the specific productivity of P. pastoris cell. It is likely to due to the lower activity of

promoter in the mixed induction compared to methanol induction. The activity of AOX

enzyme was used to show the induction level of pAOX1 under different induction

strategies. The activities in different induction strategies were compared.

AOX enzyme locates in peroxisome of P. pastoris cells (Cámara et al., 2017). In order

to extract the enzyme, the condition of cell breakage is optimized. Breakage of P. pastoris cell is challenging because of the mechanically rigid cell wall. High pressure

homogenizer is widely used where cells are disrupted by high velocity, cavitation, fluid

shear and decompression. This method needs a large volume of sample and is not

compatible for small scale study. In this study, adaptive focused acoustics (AFA),

which break cells through focused bursts of ultrasonic acoustic energy, was used as an

alternative option. The AFA has been shown to be an efficient approach to break P. pastoris cells (Woodhouse, 2016, Bláha et al., 2017). An AFA device, Covaris E210,

only requires one millilitre sample each time.

In this study, cell culture was taken from the bioreactor and the medium was removed

by centrifugation. The P. pastoris cells were washed three times by cold 200 mM

phosphate buffer. The cells were re-suspended using the buffer after washing.

Operating parameters of Covaris E210 such as duty factor, intensity, cycles per burst

were set on the maximum values to obtain the best efficiency of cell breakage.

The impact of biomass concentration and treating time on efficiency of cell breakage

was studied. The cells were diluted to concentrations of 5, 10, 20, 30, and 40 g WCW

•L-1 using the buffer and were treated with the ultrasonic acoustic for 300, 600, 1200,

1800 sec, respectively. After the treatment, the cells were centrifuged at 12000RM for

10min and the supernatant was collected. Total protein in the supernatant was

quantified using BCA method and specific protein release (mg protein • g WCW-1) was

calculated in each condition.

No protein was detected by the BCA in samples without treatment of ultrasonic acoustic.

A maximum specific protein release of 41.5 mg protein • g WCW-1 was obtained when

40.0 g WCW •L-1 was treated for 1800 sec (Fig 8.4, Appendix). When 20.0 g WCW

•L-1 was treated for 1200 sec, the specific protein release reached 37.5 mg protein • g

Page 109: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

109

WCW-1 which was comparable to the maximum value. Therefore, 20.0 g WCW •L-1

and 1200 s were selected as the cell break condition to save time sample and time.

The DCW and WCW of P. pastoris cells were correlated by DCW= 0.289WCW-4.99,

R2=0.99 (Fig 8.5, Appendix). The concentration of 20.0 g WCW•L-1 used in this study

was consistent with 4.06 g DCW•L-1. In a previous study, it was estimated that 48.3%

of contents in dry P. pastoris cells were proteins (Ihl and Tagle, 1974). Theoretically,

the maximum protein release is about 98 mg in each gram of wet cell. And the

efficiency of cell breakage by Covaris was about 38.5%. The efficiency was considered

to be acceptable since our aim was not to release all the intracellular proteins but to

release some of the AOX enzyme.

An enzymatic assay of AOX enzyme has been developed by Sigma-Aldrich (Sigma-

Aldrich-b). In the assay, the AOX enzyme catalyses the oxidation of methanol and

formed hydrogen peroxide (H2O2). H2O2 is reduced by ABTX and the oxidized ABTX

is determined by measuring OD405 of the reaction mixture. A total volume of 3.01 ml

is used in the assay and the samples can only be measured one by one using

spectrophotometer. In this study, the reaction volume was scaled down linearly with a

factor of 10. After the scale-down, the reaction could be done in 96 well plates and the

reaction rate could be measured by microplate reader.

The activity of AOX enzyme was determined from the reaction rate as described in the

Section 2.9.12, Chapter 2. However, the activity could not be determined because the

reaction reached stationary phase very fast. Therefore, diluting the samples to an

appropriate concentration is necessary prior to the assay.

In order to find the optimal concentration, the samples were diluted to different

concentrations of total protein. The reaction rates using the diluted samples were

measured. As shown in Fig 4.10, the reaction rate increased with the concentration of

total protein. When 0.48 mg•ml-1 and 0.24 mg•ml-1 of total proteins were used, OD405

of the reaction mixture reached stationary phase after 5 min and 7 min of reaction,

whilst the reaction rates were too slow if the concentration of total protein was lower

than 0.06 mg•ml-1. Therefore, samples would be diluted to a total protein concentration

of 0.12 mg•mL-1 prior to the activity assay in the future. The total activity assay of AOX

enzyme was obtained after multiplying the dilution factor of sample.

Page 110: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

110

Figure 4.10 Activity of AOX enzyme at different concentrations of total protein. 20

g•L-1 of wet cells were treated for 1200 s. The released proteins were diluted to different

concentrations varying from 0 mg•mL-1 to 0.48 mg•mL-1. The diluted samples were

added into the reaction mixture and absorbance at 405 nm was measured consecutively

by microplate reader every two minutes. Three replicates were used for each

measurement and data was shown by mean�SD (n=3).

Activities of AOX enzyme from different induction strategies used in the bioreactor

with an OTR of an OTR of 150 mmol•L-1•h-1 were compared. Fig 4.11A shows the

AOX enzyme activities in oxygen unlimited fermentation. It was observed that the

activities of AOX enzyme decreased over time during the induction. Compared to the

pure methanol induction, sorbitol/methanol mixed induction resulted in lower activities

of AOX enzyme. It is likely to be the reason that sorbitol/methanol mixed induction

only enhanced the cell growth but not product yield.

In the OLFB fermentations, activities of AOX enzyme also decreased over time. The

enzyme activity was also affected by the residual methanol concentration. After 48

hours of induction, the enzyme activity was much higher when the residual methanol

concentration was kept in the range of 0~5.0 g•L-1.

After 48 hours of induction, the activity of AOX enzyme was much higher in the OLFB

fermentations compared to the induction in oxygen unlimited condition. It is probably

due to the sufficient methanol in the medium in the OLFB fermentation.

0 1 2 3 4 5 6 7 8 90.0

0.5

1.0

1.5

2.0

2.5

Time min

Abs

orba

nce

at 4

05nm Blank

0.0035mg•ml-1

0.007mg•ml-1

0.015mg•ml-1

0.030mg•ml-1

0.060mg•ml-1

0.12mg•ml-1

0.24mg•ml-1

1 2 3 4 5 6 7 8

0.48mg•ml-1

Page 111: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

111

Figure 4.11 Comparing the activities of AOX enzyme from different induction

strategies. Samples were taken after 16, 24, 40, 48, 64, 72, 88 and 96 hours of induction.

20.0 g•L-1 of wet cells were sonicated for 1200 sec to release the AOX enzyme. The

released proteins were quantified by BCA assay and AOX enzyme activity was

measured using the method described in the Section 2.9.12, Chapter 2. The activities

of AOX enzyme from oxygen unlimited fermentations and OLFB fermentations were

shown in A and B, respectively. One sample was analysed for each point.

0 24 48 72 960

400

800

1200

1600

Induction time h

AO

X a

ctiv

ityun

its•m

g-1 to

tal p

rote

in

100% Methanol 60% Methanol

50% Methanol 40% Methanol

A

0 24 48 72 960

400

800

1200

1600

Induction time h

AO

X a

ctiv

ityun

its•m

g-1 to

tal p

rote

in

Met5-10B Met0-5

Page 112: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

112

4.4 Conclusion

Methylotrophic P. Pastoris has been developed into an efficient host for heterologous

protein production. However, supplying sufficient oxygen in standard methanol

induction is quite challenging in most large-scale bioreactors. In this paper, several

induction strategies were compared in a benchtop bioreactor having similar oxygen

transfer rate (OTR) as large-scale.

kLa of the bioreactor was characterized using dynamic method to define a fermentation

process with an OTR of 150 mmol•L-1•h-1. The kLa was found to correlate with the

agitation speed and was consistent in different vessels. In the OTR-limited bioreactor,

the standard methanol induction caused oxygen depletion and methanol accumulation.

It was observed that the residual methanol concentration significantly affected cell

growth and product expression. Therefore, strict control of the residual methanol is

required to avoid that the methanol accumulates to an inhibitive level.

Induction using slower methanol feeding avoided the oxygen depletion and eliminated

the need for control of residual methanol. But the biomass and product yield were

reduced by 15.0% and 19.0%, respectively, compared to the OLFB fermentation. The

carbon feeding rate could be enhanced by using sorbitol/methanol mixed induction.

Compared to pure methanol induction, the sorbitol/methanol mixed induction only

enhanced cell growth but not product yield, consequently, resulted in a lower specific

productivity. Since high cell density challenges the product recovery, pure methanol

induction seems to be preferable than sorbitol/methanol mixed induction in that

condition.

Activities of AOX enzyme from different induction strategies were compared.

Compared to the pure methanol induction, sorbitol/methanol mixed induction reduced

the enzyme activities. It may explain why the mixed induction strategies only enhanced

the cell growth but not product yield. The OLFB fermentations had higher activities of

AOX enzyme than oxygen unlimited fermentations in late stage of the induction.

This study helps to determine the best induction strategies of P. pastoris in OTR-limited

bioreactors. The OLFB fermentation with residual methanol concentration of 0~5.0

g•L-1 is the best in terms of obtaining the highest product yield. However, process

control of the OLFB fermentation is challenging. The pure methanol induction in an

oxygen unlimited condition is preferable when the online methanol sensor is not

available.

Page 113: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

113

Chapter 5 Prediction of cell robustness and centrifugal

dewatering using scale-down methodology

5.1 Introduction

In the first chapters, it has been established that sorbitol/methanol mixed induction

benefited upstream by reducing oxygen uptake, enhancing cell viability and improving

product purity. However, the impact of mixed induction on early downstream

processing still remains unclear.

Centrifugation is commonly used to separate product from the high density P. pastoris

culture in industry. However, the cells may be damaged by shear stress at the feeding

zone of large scale centrifuge (Boychyn et al., 2000). In the Chapter 3, it was shown

that around 8.0% cells lost viability in the end of methanol induction (Fig 3.9). Shear

stress in the centrifuge may cause further lysis of these dead cells. Cell lysis will release

cell debris, host DNA and proteins into the medium, which will reduce clarification and

product purity. It is necessary to study the impact of shear stress on the robustness of

P. pastoris cells.

Efficiency of centrifuge can be evaluated by clarification of cell culture. Clarification

refers to the efficiency of separating solid from liquid. Getting high clarification is

desirable in the centrifugation because any remaining cell solid or debris in the

supernatant will foul chromatographic resins in further downstream (Hutchinson et al.,

2006). Clarification efficiency is influenced by centrifugal conditions and properties of

the feeding liquid. Disc stack centrifuges is able to achieve higher clarification

efficiency than tubular bowl and scroll continuous centrifuges. Clarification is also

affected by operating condition of centrifuges such as bowl speed and liquid feeding

rate. Meanwhile, cell culture properties like cell size, shape, density and liquid viscosity

also affect clarification.

Dewatering of cell culture is another consideration to evaluate centrifugation.

Dewatering refers to the efficiency to purge liquid away cell solids by centrifugal force

(Lopes, 2013). High dewatering level is desired in P. pastoris centrifugation since any

liquid remaining in the sediment results in product loss. Dewatering at large scale is

affected by the centrifugation speed and liquid flow rate (Lee et al., 2003b). Besides, it

is influenced by cell culture properties such as cell morphology and liquid rheology

(Salte, 2006).

Page 114: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

114

In a previous study, an ultra scale-down methodology was established to predict the

dewatering in a pilot and industrial scale disc stack centrifugation (Lopes and

Keshavarz�Moore, 2012). It was shown that dewatering was affected by the choice of

P. pastoris strains (Lopes et al., 2012). Whether induction strategies influence the

centrifugal dewatering has not been studied.

The objective of this chapter was to

1) predict the cell robustness to shear stress in large scale centrifuges using an ultra

scale-down device.

2) characterize the properties of cell culture that may influence dewatering in

centrifugation.

2) predict the dewatering of cell culture using an ultra scale-down model of pilot and

industrial scale centrifuges.

3) study the impact of induction strategies on dewatering of cell culture.

Page 115: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

115

5.2 Theoretical considerations

5.2.1 Mimic of shear in large scale centrifuges

An ultra scale-down shear device was developed to produce energy dissipation rates (ε)

that were equivalent to those generated in the feeding zone of large scale centrifuges

(Boychyn et al., 2004). The device contains a stainless disc (diameter 40 mm, thickness

0.14 mm) in a cylindrical chamber (diameter 50 mm). The chamber has a capacity of

20 ml to hold cell culture. Energy dissipation rate (ε) and rotation speed (N) of the disc

are correlated by Eq 5.1, as being characterized by the computational fluid dynamics

(CFD) (Boychyn et al., 2001). By using this device, the shear stress generated in a large

scale centrifuge can be mimicked by rotating the disc at a specific speed. By using the

device, the impact of shear stress on cell integrity can be studied with a small volume

of cell culture.

ε=1.7•10-3•N3.71 Eq 5.1

5.2.2 Scale down of large scale centrifuges

Sigma (Σ) theory is commonly used to scale down large scale centrifuges (Ambler,

1959, Boychyn et al., 2000). Sigma (Σ) concept of equivalent settling area, which was

developed by Ambler, refers to the surface area required to achieve the same

centrifugation results as the gravitational force. It is used to predict the performance of

large scale centrifuges using benchtop ones and to compare centrifuges with different

sizes, designs and operating modes (Boychyn et al., 2004). By using Eq 5.2, flow rates

at scale can be mimicked by lengths of centrifugation time at benchtop scale.

QCΣ

=V"5N

t"5NC"5NΣ"5N Eq 5.2

where Q is the liquid flow rate at large scale, Σ and Σlab are settling area of large and

laboratory scale centrifuges, Vlab is the sample volume used in laboratory scale, tLab is

the settling time of sample at laboratory scale, C and CLab are correlation factors for

deviation of non-ideal liquid in large and laboratory scale centrifuge. For a laboratory

scale benchtop centrifuge, Σlab can be calculated by Eq 5.3 (Maybury et al., 1998).

Page 116: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

116

Σ"5N =V"5NO4(3 − 2x − 2y)

6gln(2R4R4+RT

) Eq 5.3

where ω is the angular velocity of centrifuge, R2 and R1 are outer and inner radius of

centrifuge rotor, x and y are fractional time of acceleration and deceleration in

centrifugation, g is the gravitational acceleration.

For a disc stack centrifuge, ΣDs can be calculated by Eq 5.4 (Boychyn, Yim et al. 2004).

ΣUV =2πXO4(R4

Y − RTY)

3gtanθ Eq 5.4

where n is the disc numbers, θ is the half disc angle.

5.2.3 Calculation of clarification and dewatering

Centrifugation speed and residence time are critical factors in predicting clarification

using scale-down approach, whilst dewatering is also affected by solid heights (Chu

and Lee, 2001). In the scale-down model, it is necessary to maintain the same relative

centrifugal force (RCF) as large scale. Liquid flow rate determines the residence time

of solids at large scale. In order to mimic flow rates, samples can be centrifuged for

different time periods at small scale. Solid height affects the dewatering by determining

the pressure applied to the solid. Thus, a cell concentration that would give the same

solid height as large scale should be used in scale-down model.

As shown in Eq 5.5, clarification efficiency is calculated by measuring optical densities

of the feeding stream and the supernatant after centrifugation.

%Clarification =

ODa??b − ODVcd?7e5,5e,ODa??b − OD7?a

• 100 Eq 5.5

where ODfeed is the optical density of fermentation broth before centrifuge, ODsupernatant

is the optical density of supernatant after centrifuge and ODref is the optical density of

supernatant that is equivalent to 100% clarification.

Page 117: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

117

Dewatering level can be calculated by Eq 5.6 (Salte et al., 2006).

%D = 100 −100(WCW− DCW/di7)

WCW Eq 5.6

dw7 =DCWa

WCWa Eq 5.7

where WCW is the weight of wet cell cake and DCW is the weight of dry cells. dwr is

the ratio of dry cell weight to wet cell weight after maximum removal of water in

extracellular space using filtration. DCWf is the weight of dry cells and WCWf is the

weight of wet cells after filtration.

5.3 Results and discussion

5.3.1 Determination of biomass and product

Cell culture for the dewatering study was prepared using the fermentation process as

described in the Chapter 3. Briefly, the cell culture was growth on glycerol firstly and

then induced by pure methanol or sorbitol/methanol (1:1, C-mol/C-mol) mixture at the

feeding rate of 10.8 ml•h-1•L-1 (Invitrogen, 2002). The cell cultures were induced 72

hours before harvest, and duplicate fermentations were performed for each induction

method using the parallel bioreactor.

Fig 5.1 shows the cell growth curve, cell viability and product expression during the

fermentation. After initial cell growth on 40 g•L-1 of batch glycerol, the DCW reached

20.1�1.5 g•L-1 in four vessels. The DCW was further enhanced to 59.7�5.6 g•L-1 by

feeding 50% (v/v) glycerol for six hours. After the growth on glycerol, cells were

induced by pure methanol or sorbitol/methanol (1:1, C-mol/C-mol) mixture. The DCW

nearly reached the maximum after 48 hours of induction in both of the strategies. At

the harvest time, the DCW reached 137.1�5.6 g•L-1 and 149.7�4.8 g•L-1 in methanol

and sorbitol/methanol (1:1, C-mol/C-mol) mixed induction, respectively. The harvested

biomass concentrations were comparable to the result described in the Section 3.2.3,

Chapter 3.

Page 118: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

118

After the harvest, the cell cultures were added to 15 ml falcon tubes and were

centrifuged at 4000RPM for 10min. Volumetric fraction of cell solids in the tubes were

determined. It was observed that the volumetric fractions of cell solids in methanol and

sorbitol/methanol (1:1, C-mol/C-mol) mixed induction strategies were 40.8% and

47.2%, respectively.

Cell viabilities were determined by staining the cells with propidium iodide and

measuring fraction of stained cells using flow cytometry. Prior to induction, the cell

viabilities were as high as ~98% in all the vessels. The viabilities dropped to ~92%

after 72 hours of methanol induction, which was much lower than the cell viabilities

(~97%) in sorbitol/methanol mixed induction.

Samples were taken after 24, 48, and 72 hours of induction and product expression was

determined using protein gel. In both of the induction strategies, the product bands

(~6.5 kDa) were clear visible after 24 hours of induction. The intensities of product

band kept increasing from until the harvest time. Compared to the methanol induction,

the intensity of product band in sorbitol/methanol mixed induction was lower at the

same time point. It was consistent to the finding in the Section 3.3.3, Chapter 3 where

sorbitol/methanol mixed induction decreased the product yield. The major HCP had a

molecular weight of 66 kDa. Besides, several other HCP bands were also visible, but

their amount were much lower.

Page 119: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

119

Figure 5.1 Cell growth, viability and product expression of the cell culture used in dewatering study. The cell cultures were firstly grown on

glycerol and then induced by methanol or sorbitol/methanol (1:1 C-mol/C-mol) mixture. Duplicate experiments were run for each induction method.

(A) shows the growth curve of cell cultures. Three replicates of DCW were measured at each sampling time and data was shown by mean ± SD

(n=3). (B) shows the change of cell viabilities during fermentation. Samples were taken before and after 24, 48 and 72 hours of induction. One

sample was analysed for each point. (C) shows the profile of soluble proteins after 24, 48 and 72 hours of induction by pure methanol or

sorbitol/methanol mixture.

Page 120: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

120

5.3.2 Prediction of cell robustness to shear

A rotating shear device was has been used to mimic the shear stress generated in large

scale centrifuges (Boychyn et al., 2004). The device contains a stainless disc and

rotation of the disc could mimic different energy dissipation rate (ε) generated by the

large scale centrifuges (Boychyn et al., 2001). Industrial scale centrifuges had a

maximum ε value of 1.3•106 W•kg-1 in the feeding zone, as predicted by the

computational fluid dynamics (Boychyn et al., 2004). Here the disc was rotated at

speeds of 167, 233, and 300 rps to generate ε values of 0.2, 0.75 and 1.3•106 W•kg-1,

respectively.

Prior to shear treatment, the cell cultures were diluted to a solid fraction of 20% (v/v)

using Milli-Q water because disc stack centrifuge can only process a fluid with solid

concentration up to 20% (v/v). Afterwards, the cell cultures were treated with the shear

for 20 sec to mimic the residence time of cells in the feeding zone of large scale

centrifuges. Protein release and cell viability were measured to determine the change

of cell integrity before and after shear treatment.

The supernatant was collected by a centrifuging the cell cultures at 4000RPM for 10min.

Concentration of soluble proteins in the supernatant was quantified using BCA assay.

As shown in Fig 5.2A, total protein concentrations in the cell cultures from methanol

and sorbitol/methanol (1:1, C-mol/C-mol) mixed induction strategies were comparable

before the shear treatment (1.46 ± 0.065 mg•ml-1 versus 1.52 ± 0.035 mg•ml-1, p=0.445).

Treating the cells with the shear did not cause an increase of the protein concentration.

The cell viabilities were determined using flow cytometry. Consistent to Fig 5.1B, the

cell cultures induced by methanol had higher mortalities than that induced by

sorbitol/methanol (1:1 C-mol/C-mol) mixture. After being treated with the shear of

1.3•106 W•kg-1, the mortality increased from 9.3 ± 1.2% to 11.5 ± 0.5% (p=0.236).

Change of the cell mortality after the shear treatment was negligible in the cultures

induced by sorbitol/methanol mixture.

Most of the dead cells stayed intact due to the protection of cell wall, and thus the shear

treatment caused little damage to the cell integrity. Compared to shear sensitive

mammalian cells, both live and dead P. pastoris cells are more resistant to the shear

stress generated in industrial centrifuges, which makes P. pastoris a robust host for

protein production at large scale.

Page 121: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

121

Figure 5.2 Predicting the cell robustness to shear stress in feeding zoon of the large

scale centrifuges. After 72 hours of induction, the cell cultures were harvested and

diluted to a cell solid concentration of 20% (v/v). Cells were then exposed to a series

of energy dissipation rates (0.2•106, 0.75•106, 1.3•106 W•kg-1) for 20 sec by a shear

device. Cell cultures were centrifuged at 4000 RPM for 10 min after the shear treatment.

(A) Soluble proteins in the supernatant were quantified using BCA assay. Data in the

graph are presented as mean ± SD (n = 3). (B) Cells were stained by propidium iodide

and percentage of dead cells were measured by flow cytometry. One sample was

analysed for each point.

Page 122: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

122

5.3.3 Characterization of cell culture properties

Cell culture properties, such as cell shape, size distribution, viscosity and density of

culture were shown to influence sedimentation and package of cells and thus affect

dewatering efficiency in centrifuges (Salte, 2006). Characterization of the cell culture

properties will help to predict the centrifugal dewatering of cell cultures. P. pastoris

cells are round as observed by an optical microscope. The densities of cell culture from

methanol and mixed induction methods were ignored. The cell size distribution and

culture viscosity were compared in this study.

Fig 5.3A shows the cell size distribution measured at different time points of

fermentation. No peak of cell debris was detected after 72 hours of induction by either

methanol or sorbitol/methanol mixture, which indicated that most cells penetrating to

propidium iodide still stayed intact. The cell size distribution in two induction strategies

changed in different manners. The cell diameter stayed stable during methanol

induction, whereas the cell diameter shifted to smaller values over time in

sorbitol/methanol mixed (1:1, C-mol/C-mol) induction. Dv50 value of the cell size

decreased from 3.85 ± 0.3 μm to 3.14 ± 0.2 μm (p=0.002) after 72 h of mixed induction,

while the change of D50 value in methanol induction was not so significant (3.80 ± 0.6

versus 3.74 ± 0.3 μm, p= 0.465). The Dv50 values of cell cultures from methanol and

sorbitol/methanol mixed induction had significant difference at the harvest time (p=

0.003).

It was reported that P. pastoris cultured on methanol had larger cell diameter than that

grown on glucose (Rebnegger et al., 2016). However, no report was found where the

cell diameters grown on methanol and sorbitol were compared. It is likely that the

carbon influences cell philosophy and thus changes the cell diameter. Similarly,

diameter of CHO cells were observed to change during the cell culture (Kiehl et al.,

2011).

A previous study showed that centrifugal dewatering of P. pastoris was significantly

influenced by the cell diameter (Lopes et al., 2012). As being predicted by a scale-down

model of large scale centrifuge, cells with diameter of 3~4 μm had much higher

dewatering levels than that with diameter of 4~6 μm. As a result, the smaller cells had

a higher product recovery in centrifugation. Thus, it becomes interesting to study

whether the sorbitol/methanol mixed induction affects centrifugal dewatering of the

cell culture.

Page 123: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

123

Figure 5.3 Cumulative cell size distributions in methanol and sorbitol/methanol (1:1 C-

mol/C-mol) mixed induction. (A) Samples were taken before and after 24, 48 and 72

hours of induction. Cell size distribution was measured using Mastersizer 3000. (B)

shows the cumulative cell size distributions at the harvest time.

Page 124: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

124

Viscosities of the cell cultures induced by methanol and sorbitol/methanol mixture were

measured by rheometer. Cell cultures were diluted to a solid fraction of 30% (v/v)

before measurement. During the measurement, shear rate of the disc gradually

increased from 100 sec-1 to1000 sec-1. It was observed that the viscosities decreased

slowly with the increasing of shear rate (Fig 5.4A), which was a typical behaviour of

non-Newtonian fluid. At the rate of 800 sec-1, cell culture from sorbitol/methanol mixed

induction had higher viscosities than that from methanol induction (0.0014 ± 0.00006

versus 0.0020 ± 0.00013, p=0.04). The viscosities of P. pastoris culture are quite low

which will make cell sedimentation easier in centrifugation (Körner, 2013).

Viscosities of the cell cultures were measured at different time points of fermentation

and shown at the shear rate of 800 sec-1 (Fig 5.4B). It was observed that the viscosities

increased with the biomass. The viscosities were below 0.001 in the batch phase and

increased to nearly 0.003 after 24 hours of induction. At the harvesting time, viscosities

of cell cultures from methanol and sorbitol/methanol mixed induction were similar

(0.0027 ± 0.00015 versus 0.0027 ± 0.00017, p=0.98).

The cell cultures used in dewatering study had quite low viscosities, and thus the impact

of viscosities on dewatering was ignored.

Page 125: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

125

Figure 5.4 Comparing the viscosities of cell cultures in methanol and sorbitol/methanol

(1:1 C-mol/C-mol) mixed induction. (A) After 72 hours of induction, the cell cultures

were diluted to a cell concentration of 30% (v/v) using Milli-Q water. Shear rate of

rheometer’s disc linearly increased from 100 sec-1 to1000 sec-1. And the viscosities

were recorded. (B) The cell cultures were taken before and after 24, 48 and 72 hours of

induction. The viscosities within shear rates of 100 sec-1 to1000 sec-1 were measured

and the viscosities at shear rate of 800 sec-1 was shown. One sample was analysed for

each point.

Page 126: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

126

5.3.4 Prediction of centrifugal dewatering

Here dewatering levels of cell cultures were evaluated using a scale-down model of

pilot scale CSA-1 centrifuge and industrial scale BTPX305 centrifuge (Lopes. 2013).

Centrifugation speed is critical to predict dewatering. In the scale-down model, it is

necessary to maintain the same relative centrifugal force (RCF) as large scale. Here

speeds of 9700 RPM and 7488 RPM were used in the benchtop centrifuges to mimic

the bowl speeds of CSA-1 and BTPX305 centrifuges. Liquid flow rates of large

centrifuges were mimicked by the ratio of sample volume to centrifugation time at

small scale. Dewatering levels at flow rates of 8 L•h-1~110 L•h-1 and 200 L•h-1~1600

L•h-1 were predicted in this study. Solid height affects dewatering by determining extra-

pressure applied to the solid at the bottom (Chu and Lee, 2001). In order to obtain the

average solid heights at scale, cell cultures were spun in 2.2 ml Eppendorf tube and 15

ml centrifuge tube by Eppendorf 5810R and Beckman Coulter Avanti J-E Centrifuge,

respectively. Dimensions of the centrifuges were shown in Table 5.1 and Σ values were

calculated using Eq 5.3 and Eq 5.4. By using Eq 5.2, flow rates of disc stack centrifuges

were transferred to different lengths of centrifuge time in benchtop centrifuges.

Centrifuge Dimensions N (r·sec-1) C Σ (m2)

Eppendorf 5810R

R1 (0.075 m) R2 (0.1 m)

149 1.0 0.66~0.77

Beckman Coulter

Avanti J-E

R1 (0.073 m) R2 (0.102 m)

92 1.0 1.12~1.82

CSA-1

R1 (0.026 m) R2 (0.055 m)

n (45)

θ (38.5�)

162 0.4 1444

BTPX-305

R1 (0.036 m) R2 (0.085 m)

n (82)

θ (40�)

125 0.4 7127

Table 5.1 Dimensions of centrifuges used in the dewatering study.

Page 127: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

127

Fig 5.5 shows the dewatering levels of cell cultures induced by methanol or

sorbitol/methanol mixture. Within the studied flow rates, dewatering levels were over

70% in both cell cultures. Dewatering levels at different flow rates were also

comparable. Compared to the cell culture induced by methanol, the cell culture from

mixed induction had higher dewatering levels in both centrifuges. Difference between

the dewatering levels were significant at some but not all the predicted flow rates. It is

likely to because of the variations. Average dewatering levels were improved from

77.3±4.6% to 83.0±3.8% (p<0.01) in CSA centrifuge and from 78.5±3.6% to 83.1±1.9%

(p<0.01) in BTPX305 centrifuge by using the mixed induction. In addition, clarification

was nearly 100% in both centrifuges (Fig 5.6).

Larger particles are more difficult to be packed in centrifugation and more liquid

accumulates in interstitial space (Wu et al., 1997). In a previous study, it was shown

that the centrifugal dewatering of P. pastoris was affected by cell size (Lopes et al.,

2012). The dewatering reached ~80% in disc stack centrifuges when cells with an

average size of 3~4 μm were used, while it was less than 60% for cells with an average

size of 4~6 μm. The difference of dewatering levels of methanol and sorbitol/methanol

mixture induced cultures is likely to be attributed to the different cell sizes.

P values calculated using student’s t-test indicates significant statistical differences

between the average dewatering levels in both centrifuges. This leads to a prediction of

a loss of 41.3±5.3 g product from a 1000 litre culture induced by methanol, whereas a

loss of 17.1±2.1 g if mixed induction is used. This indicates that changing induction

method is an effective way to minimize product loss in centrifugal separation. This

becomes a valuable process optimization tool specially when high value products are

manufactured.

Page 128: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

128

Figure 5.5 Dewatering levels of the cell cultures induced by methanol and

sorbitol/methanol (1:1 C-mol/C-mol) mixture as predicted by the scale-down model of

CSA-1 and BTPX305 disc stack centrifuges. Data in the graph are presented as mean

± SD (n = 3). The significance of dewatering between methanol and mixed induction

strategies were shown by * (p<0.05) and ** (p<0.01).

8 16 32 64 12860

70

80

90

100

Flow rate L·h-1

Dew

ater

ing

%

Methanol-1

Methanol-2

Mixture-1

Mixture-2

A

***

*

256 512 1024 204860

70

80

90

100

Flow rate L·h-1

Dew

ater

ing

%

Methanol-1

Methanol-2

Mixture-1

Mixture-2

B

** *** *

Page 129: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

129

Figure 5.6 Clarification levels of the cell cultures induced by methanol and

sorbitol/methanol (1:1 C-mol/C-mol) mixture as predicted by a scale down model of

CSA-1 and BTPX305 disc stack centrifuges. Data in the graph are presented as mean

± SD (n = 3).

8 16 32 64 12898

99

100

101

102

Flow rate L·h-1

Cla

rific

atio

n %

Methanol-1Methanol-2

Mixture-1Mixture-2

A

128 256 512 1024 204898

99

100

101

102

Flow rate L·h-1

Cla

rific

atio

n %

Methanol-1

Methanol-2

Mixture-1

Mixture-2

B

Page 130: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

130

Centrifugal dewatering of the cell cultures from an OTR-limited bioreactor was also

studied using the ultra scale-down method. As described in the Chapter 4, the cell

cultures were induced by methanol or sorbitol/methanol mixtures with 40%~60%

(Cmethanol) of methanol in a bioreactor with OTR of 150 mmol•L-1•h-1. The cell cultures

were harvested after 96 hours of induction. Cell size distribution and centrifugal

dewatering of the cultures were studied.

As shown in Fig 5.7, the cell culture from methanol induction had smaller cell diameter

than that from sorbitol/methanol mixed induction methods. The D50 values of cell

cultures from methanol and sorbitol/methanol mixed induction were 3.23 μm and 3.74

± 0.7 μm (p=0.011), respectively.

Consequently, the cell cultures from sorbitol/methanol mixed induction had higher

dewatering levels than that from methanol induction, as predicted by the scale-down

model of CSA-1 disc stack centrifuge. Difference between the dewatering levels were

significant at all the predicted flow rates. The average dewatering of cell cultures from

three mixed induction methods reached 84.2±1.6%, 81.8±1.1% and 82.3±1.3%,

respectively, while the average dewatering level was 79.1±1.0% in cell culture from

methanol induction. The result further indicated that sorbitol/methanol mixed induction

enhanced centrifugal dewatering of cell culture by influencing cell diameter.

Page 131: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

131

Figure 5.7 Cumulative size distribution and dewatering of cell cultures in fermentations

performed in an OTR-limited bioreactor. The cell cultures were induced by methanol

or sorbitol/methanol mixtures containing 40%~60% (Cmethanol) of methanol in a

bioreactor with OTR of 150 mmol•L-1•h-1. Samples were taken after 96 hours of

induction and cumulative size distributions of the cell cultures were measured using

Mastersizer 3000. (B) Dewatering of the cell cultures was predicted by a scale-down

model of CSA-1 centrifuge. Data in the graph are presented as mean ± SD (n = 3). The

significance of dewatering between methanol and mixed induction strategies were

shown by ** (p<0.01) and *** (p<0.001).

4 8 16 32 64 12870

80

90

Flow rate L·h-1

Dew

ater

ing

%

100%Methanol 60%Methanol

50%Methanol 40%Methanol

B

***** **

***

Page 132: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

132

5.4 Conclusion

Disc stack centrifuge is widely used to harvest product from high density P. pastoris

culture. Efficiency of the centrifugation is influenced by the cell culture properties.

Besides, shear stress of the large scale centrifuges is likely to damage the cells. In this

chapter, the cell cultures induced by methanol and sorbitol/methanol mixture were

compared in terms of dewatering levels and robustness to shear stress using ultra scale-

down method.

Cell mortality was much higher in the standard methanol induction. Despite the

mortality, the cell cultures were robust to shear stress in feeding zone of the disc stack

centrifuges, as predicted by an ultra scale-down device. The shear treatment caused

litter damage to the cell integrity due to the presence of cell wall. It makes P. pastoris

a good host for the production of recombinant proteins compared to shear sensitive

mammalian cells.

Cell culture properties affecting centrifugal dewatering were characterized. It was

observed that the cell diameter became smaller and smaller over time in the mixed

induction, while the diameter in methanol induction did not change. In an OTR-limited

bioreactor, cell size from the mixed induction was also smaller than that from methanol

induction. Besides, the cell cultures had low viscosities which was beneficial for the

cell sedimentation in centrifuges.

Due to the difference of cell diameter, the cell culture from mixed induction had higher

dewatering levels as predicted by an ultra scale-down model of pilot and industrial

centrifuges. Consequently, less product got lost in centrifugation of the methanol

induced cell culture. This finding establishes the sorbitol/methanol mixed induction as

an effective way to minimize product loss in centrifugation. This becomes a valuable

process optimization tool specially when high value products are manufactured.

Page 133: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

133

Chapter 6 Conclusions and future work

6.1 Conclusions

In this thesis, the impact of sorbitol/methanol mixed induction on P. pastoris

fermentation was studied in terms of cell growth, viability, product yield, host cell

protein profile and centrifugal dewatering efficiency. Besides, methanol and

sorbitol/methanol mixed induction strategies were compared in a bioreactor with

comparable OTR as large scale.

An enzymatic assay of the product that was compatible with the culture medium was

established. The cell culture in microplate showed that sorbitol did not inhibit cell

growth even at high concentrations while methanol did. The study in shake flasks

showed that using sorbitol as a substrate reduced cell lysis during cell culture.

Standard methanol induction and sorbitol/methanol (1:1, C-mol/C-mol) mixed

induction were compared using a parallel bioreactor. It was observed that the

sorbitol/methanol mixed induction enhanced cell growth, improved cell viability and

reduced oxygen uptake.

Using sorbitol/methanol (1:1, C-mol/C-mol) mixed induction resulted in a lower

product yield compared to the standard methanol induction. However, fewer HCPs

were co-released into the medium with the product. It is likely to make the product

purification easier. Fewer types of protease were identified from the mixed induction,

and thus product degradation may be reduced when the products are sensitive to

proteolytic degradation.

Oxygen transfer coefficient (kLa) of the parallel bioreactor was measured using

dynamic method. The kLa was observed to linearly correlated with the agitation speed.

By using the kLa, a fermentation process with an OTR of 150mmol•L-1•h-1 was defined

and was used to study the cell culture at large scale.

Induction strategies in an OTR-limited bioreactor was developed. Using standard

methanol induction caused oxygen limitation and methanol accumulation in the

bioreactor. In the oxygen limited fermentation, biomass concentration and product

yield were significantly influenced by the concentration of residual methanol. Thus,

strict control of residual methanol concentration is required to avoid that it rises to an

inhibitive level.

Page 134: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

134

Induction in an oxygen unlimited condition eliminated the need to strictly control the

residual methanol. Due to the slower methanol feeding rate, the biomass and product

yield were significantly reduced compared to the oxygen limited fermentation. Partially

replacing methanol with sorbitol enhanced the carbon feeding in oxygen unlimited

fermentation. Consequently, the biomass concentration was significantly improved.

However, the higher biomass did not result in higher product yields compared to

methanol induction. Since high cell density challenges product recovery, methanol

induction is preferable to sorbitol/methanol mixed induction in the oxygen unlimited

fermentation.

Activity of AOX enzyme in different induction strategies were compared. In the oxygen

unlimited fermentation, sorbitol/methanol mixed induction resulted in lower activities

of AOX enzyme compared to methanol induction. It is likely to be the reason that

sorbitol/methanol mixed induction only enhanced the cell growth but not product yield.

The activity of AOX enzyme was much higher in the oxygen limited fermentations

compared to that in oxygen unlimited condition. It is probably due to the sufficient

methanol in the medium in the OLFB fermentations.

Finally, the impact of sorbitol/methanol mixed induction on centrifugal dewatering was

studied. An ultra scale-down shear device was used to predict the cell robustness to

shear stress in large scale centrifuges. It was found that the cells induced by methanol

or sorbitol/methanol mixture were robust to the shear stress. While treating cell culture

from methanol induction resulted in a slight decline of cell viability, it did not cause

the release of intracellular proteins.

Cell culture properties influencing dewatering levels was studied. It was observed that

the cell diameter became smaller and smaller over time in sorbitol/methanol mixed

induction. As a result, cell cultures from sorbitol/methanol mixed induction had higher

dewatering levels, as predicted by an ultra scale-down model of pilot and industrial

scale disc stack centrifuges. This indicates that changing induction method is an

effective way to minimize product loss in centrifugal separation.

Page 135: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

135

Induction strategy Biomass

g•L-1

Cell viability

%

Average OUR

mmol•L-1•h-1

Product yield

g•L-1 Identified

HCP Dewatering in CSA

centrifuge %

Standard methanol induction 129.4 92.5 256.4 1.69 96 77.3%

Sorbitol/methanol (1:1, C-mol/C-mol) mixed induction 152.4 97.6 150.2 1.05 72 83.0%

Fermentation in an OTR-

limited bioreactor

OLFB (Met0-5) 108.2 94.3 152.2 1.04 - -

OLFB (Met5-10) 79.1 92.8 155.7 0.87 - -

100% Methanol 90.2 97.9 130.6 0.86 - 79.1%

Mixed induction (60% Methanol) 114.6 96.6 135.0 0.86 - 84.2%

Mixed induction (50% Methanol) 130.0 98.1 127.1 0.79 - 81.8%

Mixed induction (40% Methanol) 130.2 98.0 129.2 0.78 - 82.3%

Table 6.1 A summary of the fermentation and centrifugal dewatering investigated in this thesis.

Page 136: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

136

6.2 Future work

In this project, sorbitol/methanol mixed induction strategy was developed to offset the

drawbacks of pure methanol induction. Future works will extend the findings described

in this thesis.

In the production of recombinant aprotinin, sorbitol/methanol mixed induction strategy

benefited the cell culture by reducing oxygen uptake and enhancing cell viability.

Nevertheless, its product yield was much lower compared to using standard methanol

induction, which made the production less efficient. In the OTR-limited bioreactor,

methanol induction advantages over sorbitol/methanol mixed induction by producing

comparable yield with fewer cells. On the other hand, the impact of sorbitol/methanol

mixed induction on product yield varies on cell strains. As it was summarized in Table

1.3, sorbitol/methanol mixed induction significantly improved the product yield of

recombinant Interferon-α, porcine Circovirus cap protein, Thermomyces lanuginosus

lipase, β-Glucosidase, β-Mannanases, Erythropoietin and Rhizopus oryzae lipase.

Therefore, the advantages of sorbitol/methanol mixed induction are likely to be shown

better by applying it to other strains in the future. In addition, sorbitol’s price is over

ten times higher than that of methanol. A mixing tank is also required to prepare the

sorbitol solution. The mixed induction strategy is likely to increase cost of goods of the

cell culture especially at large scale. Although it diminishes the usage of expensive pure

oxygen, its impact on the whole capital cost needs to be well calculated in the future.

It was shown that sorbitol/methanol mixed induction reduced the number of HCPs and

types of proteases identified from the cell culture. In a specific range of molecular

weight (MW) and isoelectric point (PI), fewer HCPs were identified from the cell

culture induced by sorbitol/methanol mixture. When the products have MW and PI

within the range, using sorbitol/methanol mixed induction is likely to make the

purification much easier. In the future work, it will be interesting to apply the mixed

induction strategy to production of recombinant Interferon-γ, Interferon-β and

Keratinocyte growth factor, etc., and investigate whether the mixed induction is an

efficient tool to simplify the purification. Although proteolytic degradation of aprotinin

was not observed in this study, it was observed in expression of other products such as

recombinant interferon-τ (Sinha et al., 2005). In the future, sorbitol/methanol mixed

induction can be applied to the production of protease-sensitive proteins and investigate

whether the mixed induction will protect the products from degradation.

Page 137: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

137

Residual methanol concentration has a significant impact on cell growth and product

expression in oxygen limited fermentation. It was observed that residual methanol

higher than 5 g•L-1 reduced both biomass and product yield in the OTR-limited

bioreactor. However, the concentration was not maintained at a constant level by using

a HPLC-based feedback control. In the future, commercial methanol sensor can be

applied to control the residual methanol better. An optimal methanol concentration can

be found to get the maximum product yield.

Sorbitol/methanol mixed induction was observed to enhance centrifugal dewatering of

the cell culture, as predicted by a scale down model of disc stack centrifuge. In the

future, the finding can be further verified by using other cell strains and studying the

dewatering in a pilot or large scale centrifuge. Product loss in methanol and

sorbitol/methanol mixture induced cell cultures can be calculated. Besides, it is

interesting to study whether shear stress in the centrifuges influences the product quality.

The soluble protein profile can be analysed using protein gel or proteomics before and

after the shear treatment. It was reported that shear stress caused aggregation and

conformation change of some proteins (Di Stasio and De Cristofaro, 2010, Nesta et al.,

2017). In the future, more complex product, such as monoclonal antibody, can be used

and whether the shear stress in centrifugation induces protein aggregation can be

studied. The finding will guide the selection of product recovery options for these

proteins.

Currently, high density fermentation is popularly used in P. pastoris culture. It results

in high oxygen uptake in upstream and challenges product harvest in downstream.

Alternatively, continuous fermentation can be used where product is produced with a

relatively lower cell density. In continuous fermentation, a smaller scale of bioreactor

and centrifuge can be used, which will cut the cost of facility installation in industry.

In the future, a small scale continuous culture can be developed at small scale and its

productivity, efficiency of product recovery and cost of goods can be compared with

the traditional fed-batch fermentation. In addition, sorbitol/methanol mixed induction

enhanced the cell viability in this study. That is beneficial to a continuous culture in

which the induction is performed for a longer period. It will be interesting to apply the

mixed induction to the continuous culture and study its impact on the cell viability and

product quality.

It was reported that adding sorbitol as a co-substrate causes shift of methanol flux

distribution. It was found that the methanol flux decreased in energizing pathway while

Page 138: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

138

increased in biomass synthesis pathway. The shift of methanol flux was likely to

influence product yield as reported by several studies (Wang et al., 2010, Çelik et al.,

2009). In the future, the intracellular metabolic fluxes in sorbitol/methanol mixed

induction can quantitatively studied by labelling methanol with C13. Change of

metabolic flux under different sorbitol/methanol ratios and feeding rates can be studied.

From the deeper insights into the metabolic fluxes, potential metabolic bottlenecks for

protein production can be revealed, which will further guide the optimization of carbon

feeding strategy.

Page 139: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

139

Chapter 7 References

AHMAD, M., HIRZ, M., PICHLER, H. & SCHWAB, H. 2014. Protein expression in

Pichia pastoris: recent achievements and perspectives for heterologous protein

production. Applied microbiology and biotechnology, 98, 5301-5317.

AHN, J., HONG, J., LEE, H., PARK, M., LEE, E., KIM, C., CHOI, E., JUNG, J. &

LEE, H. 2007. Translation elongation factor 1-α gene from Pichia pastoris:

molecular cloning, sequence, and use of its promoter. Applied microbiology and

biotechnology, 74, 601-608.

ALFORD, J. 2008. Automation applications in bio-pharmaceuticals, Softbound Book,

2008, ISA.

AMBLER, C. M. 1959. The theory of scaling up laboratory data for the sedimentation

type centrifuge. Journal of Biochemical and Microbiological Technology and

Engineering, 1, 185-205.

BARRIGÓN, J. M., MONTESINOS, J. L. & VALERO, F. 2013. Searching the best

operational strategies for Rhizopus oryzae lipase production in Pichia pastoris

Mut+ phenotype: Methanol limited or methanol non-limited fed-batch cultures?

Biochemical engineering journal, 75, 47-54.

BATRA, J., BERI, D. & MISHRA, S. 2014. Response surface methodology based

optimization of β-glucosidase production from Pichia pastoris. Applied

biochemistry and biotechnology, 172, 380-393.

BERRIOS, J., FLORES, M.-O., DÍAZ-BARRERA, A., ALTAMIRANO, C.,

MARTÍNEZ, I. & CABRERA, Z. 2017. A comparative study of glycerol and

sorbitol as co-substrates in methanol-induced cultures of Pichia pastoris:

temperature effect and scale-up simulation. Journal of industrial microbiology

& biotechnology, 44, 407-411.

BLÁHA, B. A., MORRIS, S. A., OGONAH, O. W., MAUCOURANT, S.,

CRESCENTE, V., ROSENBERG, W. & MUKHOPADHYAY, T. K. 2017.

Development of a high‐throughput microscale cell disruption platform for

Pichia pastoris in rapid bioprocess design. Biotechnology progress. 34, 130-

140

BOYCHYN, M., DOYLE, W., BULMER, M., MORE, J. & HOARE, M. 2000.

Laboratory scale down of protein purification processes involving fractional

Page 140: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

140

precipitation and centrifugal recovery. Biotechnology and bioengineering, 69, 1-10.

BOYCHYN, M., YIM, S., BULMER, M., MORE, J., BRACEWELL, D. & HOARE,

M. 2004. Performance prediction of industrial centrifuges using scale-down

models. Bioprocess and Biosystems Engineering, 26, 385-391.

BOYCHYN, M., YIM, S., SHAMLOU, P. A., BULMER, M., MORE, J. & HOARE,

M. 2001. Characterization of flow intensity in continuous centrifuges for the

development of laboratory mimics. Chemical Engineering Science, 56, 4759-

4770.

BRACEWELL, D. G., FRANCIS, R. & SMALES, C. M. 2015. The future of host cell

protein (HCP) identification during process development and manufacturing

linked to a risk‐based management for their control. Biotechnology and

bioengineering, 112, 1727-1737.

BRETTHAUER, R. K. & CASTELLINO, F. J. 1999. Glycosylation of Pichia pastoris‐

derived proteins. Biotechnology and applied biochemistry, 30, 193-200.

ÇALIK, P., İNANKUR, B., SOYASLAN, E. Ş., ŞAHIN, M., TAŞPINAR, H., AÇIK,

E. & BAYRAKTAR, E. 2010. Fermentation and oxygen transfer characteristics

in recombinant human growth hormone production by Pichia pastoris in

sorbitol batch and methanol fed‐batch operation. Journal of chemical

technology and biotechnology, 85, 226-233.

CÁMARA, E., LANDES, N., ALBIOL, J., GASSER, B., MATTANOVICH, D. &

FERRER, P. 2017. Increased dosage of AOX1 promoter-regulated expression

cassettes leads to transcription attenuation of the methanol metabolism in Pichia

pastoris. Scientific reports, 7:44302.

CANALES, C., ALTAMIRANO, C. & BERRIOS, J. 2015. Effect of dilution rate and

methanol‐glycerol mixed feeding on heterologous Rhizopus oryzae lipase

production with Pichia pastoris Mut+ phenotype in continuous culture.

Biotechnology progress, 31, 707-714.

CARLY, F., NIU, H., DELVIGNE, F. & FICKERS, P. 2016. Influence of

methanol/sorbitol co-feeding rate on pAOX1 induction in a Pichia pastoris

Mut+ strain in bioreactor with limited oxygen transfer rate. Journal of industrial

microbiology & biotechnology, 43, 517-523.

Page 141: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

141

ÇELIK, E., ÇALIK, P. & OLIVER, S. G. 2009. Fed‐batch methanol feeding strategy

for recombinant protein production by Pichia pastoris in the presence of co‐

substrate sorbitol. Yeast, 26, 473-484.

ÇELIK, E., ÇALIK, P. & OLIVER, S. G. 2010. Metabolic flux analysis for

recombinant protein production by Pichia pastoris using dual carbon sources:

Effects of methanol feeding rate. Biotechnology and bioengineering, 105, 317-

329.

CEREGHINO, J. L. & CREGG, J. M. 2000. Heterologous protein expression in the

methylotrophic yeast Pichia pastoris. FEMS microbiology reviews, 24, 45-66.

CHANDLER, M. A. & ZYDNEY, A. L. 2005. Clarification of yeast cell suspensions

by depth filtration. Biotechnology progress, 21, 1552-1557.

CHAROENRAT, T., KETUDAT-CAIRNS, M., JAHIC, M., ENFORS, S.-O. &

VEIDE, A. 2006. Recovery of recombinant β-glucosidase by expanded bed

adsorption from Pichia pastoris high-cell-density culture broth. Journal of

biotechnology, 122, 86-98.

CHAROENRAT, T., KETUDAT-CAIRNS, M., STENDAHL-ANDERSEN, H.,

JAHIC, M. & ENFORS, S.-O. 2005. Oxygen-limited fed-batch process: an

alternative control for Pichia pastoris recombinant protein processes.

Bioprocess and biosystems engineering, 27, 399-406.

CHEN, G. H., YIN, L. J., CHIANG, I. H. & JIANG, S. T. 2007. Expression and

purification of goat lactoferrin from Pichia pastoris expression system. Journal

of food science, 72, 67-71.

CHERYAN, M. 1998. Ultrafiltration and microfiltration handbook, CRC press.

CHU, C. & LEE, D. 2001. Experimental analysis of centrifugal dewatering process of

polyelectrolyte flocculated waste activated sludge. Water Research, 35, 2377-

2384.

CIARKOWSKA, A. & JAKUBOWSKA, A. 2013. Pichia pastoris as an expression

system for recombinant protein production. Postepy biochemii, 59, 315-321.

CLARKE, C., DOOLAN, P., BARRON, N., MELEADY, P., O'SULLIVAN, F.,

GAMMELL, P., MELVILLE, M., LEONARD, M. & CLYNES, M. 2011.

Predicting cell-specific productivity from CHO gene expression. Journal of

biotechnology, 151, 159-165.

Page 142: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

142

COS, O., RAMÓN, R., MONTESINOS, J. L. & VALERO, F. 2006. Operational

strategies, monitoring and control of heterologous protein production in the

methylotrophic yeast Pichia pastoris under different promoters: a review.

Microbial cell factories, 5, 17.

COS, O., SERRANO, A., MONTESINOS, J. L., FERRER, P., CREGG, J. M. &

VALERO, F. 2005. Combined effect of the methanol utilization (Mut)

phenotype and gene dosage on recombinant protein production in Pichia

pastoris fed-batch cultures. Journal of biotechnology, 116, 321-335.

CREGG, J. M., CEREGHINO, J. L., SHI, J. & HIGGINS, D. R. 2000. Recombinant

protein expression in Pichia pastoris. Molecular biotechnology, 16, 23-52.

CREGG, J. M., SHEN, S., JOHNSON, M. & WATERHAM, H. R. 1998. Classical

genetic manipulation. Pichia protocols. Springer.

CUNHA, A., CLEMENTE, J., GOMES, R., PINTO, F., THOMAZ, M., MIRANDA,

S., PINTO, R., MOOSMAYER, D., DONNER, P. & CARRONDO, M. 2004.

Methanol induction optimization for scFv antibody fragment production in

Pichia pastoris. Biotechnology and bioengineering, 86, 458-467.

D'ANJOU, M. C. & DAUGULIS, A. J. 2001. A rational approach to improving

productivity in recombinant Pichia pastoris fermentation. Biotechnology and

bioengineering, 72, 1-11.

DEAKIN, L. 2012. Scaling Up Disposable Systems for Depth Filtration in Cell-Culture

Clarification. Biopharm international, S16-S19.

DI STASIO, E. & DE CRISTOFARO, R. J. B. C. 2010. The effect of shear stress on

protein conformation: Physical forces operating on biochemical systems: The

case of von Willebrand factor. 153, 1-8.

DING, J., ZHANG, C., GAO, M., HOU, G., LIANG, K., LI, C., NI, J., LI, Z. & SHI,

Z. 2014. Enhanced porcine circovirus Cap protein production by Pichia pastoris

with a fuzzy logic DO control based methanol/sorbitol co-feeding induction

strategy. Journal of biotechnology, 177, 35-44.

DONEANU, C., XENOPOULOS, A., FADGEN, K., MURPHY, J., SKILTON, S. J.,

PRENTICE, H., STAPELS, M. & CHEN, W. Analysis of host-cell proteins in

biotherapeutic proteins by comprehensive online two-dimensional liquid

chromatography/mass spectrometry. MAbs, 2012, 24-44.

Page 143: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

143

DONG, Y., ZHANG, F., WANG, Z., DU, L., HAO, A., JIANG, B., TIAN, M., LI, Q.,

JIA, Q. & WANG, S. 2012. Extraction and purification of recombinant human

serum albumin from Pichia pastoris broths using aqueous two-phase system

combined with hydrophobic interaction chromatography. Journal of

Chromatography A, 1245, 143-149.

ECKER, D. M., JONES, S. D. & LEVINE, H. L. The therapeutic monoclonal antibody

market. MAbs, 2015, 9-14.

ELLIS, S. B., BRUST, P. F., KOUTZ, P. J., WATERS, A., HARPOLD, M. M. &

GINGERAS, T. R. 1985. Isolation of alcohol oxidase and two other methanol

regulatable genes from the yeast Pichia pastoris. Molecular and cellular

biology, 5, 1111-1121.

FANG, Z., XU, L., PAN, D., JIAO, L., LIU, Z. & YAN, Y. 2014. Enhanced production

of Thermomyces lanuginosus lipase in Pichia pastoris via genetic and

fermentation strategies. Journal of industrial microbiology & biotechnology, 41, 1541-1551.

FEIST, P. & HUMMON, A. J. I. J. O. M. S. 2015. Proteomic challenges: sample

preparation techniques for microgram-quantity protein analysis from biological

samples. 16, 3537-3563.

FLORES, M. V., CUELLAS, A. & VOGET, C. E. 1999. The proteolytic system of the

yeast Kluyveromyces lactis. Yeast, 15, 1437-1448.

GAO, M.-J., LI, Z., YU, R.-S., WU, J.-R., ZHENG, Z.-Y., SHI, Z.-P., ZHAN, X.-B. &

LIN, C.-C. 2012. Methanol/sorbitol co-feeding induction enhanced porcine

interferon-α production by P. pastoris associated with energy metabolism shift.

Bioprocess and biosystems engineering, 35, 1125-1136.

GAO, M.-J., ZHAN, X.-B., GAO, P., ZHANG, X., DONG, S.-J., LI, Z., SHI, Z.-P. &

LIN, C.-C. 2015. Improving performance and operational stability of porcine

interferon-α production by Pichia pastoris with combinational induction

strategy of low temperature and methanol/sorbitol co-feeding. Applied

biochemistry and biotechnology, 176, 493-504.

GARCIA-OCHOA, F. & GOMEZ, E. 1998. Mass transfer coefficient in stirred tank

reactors for xanthan gum solutions. Biochemical Engineering Journal, 1, 1-10.

Page 144: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

144

GARCIA-OCHOA, F. & GOMEZ, E. 2004. Theoretical prediction of gas–liquid mass

transfer coefficient, specific area and hold-up in sparged stirred tanks. Chemical

Engineering Science, 59, 2489-2501.

GARCIA-OCHOA, F. & GOMEZ, E. 2009. Bioreactor scale-up and oxygen transfer

rate in microbial processes: an overview. Biotechnology advances, 27, 153-176.

GIBBS, R. A., WEINSTOCK, G. M., METZKER, M. L., MUZNY, D. M.,

SODERGREN, E. J., SCHERER, S., SCOTT, G., STEFFEN, D., WORLEY,

K. C. & BURCH, P. E. 2004. Genome sequence of the Brown Norway rat yields

insights into mammalian evolution. Nature, 428, 493-521.

GOLDNER, M. G. 1972. History of insulin. Annals of internal medicine, 76, 329-329.

GOODRICK, J., XU, M., FINNEGAN, R., SCHILLING, B., SCHIAVI, S., HOPPE,

H. & WAN, N. 2001. High‐level expression and stabilization of recombinant

human chitinase produced in a continuous constitutive Pichia pastoris

expression system. Biotechnology and bioengineering, 74, 492-497.

HANKO, V. P. & ROHRER, J. S. 2004. Determination of amino acids in cell culture

and fermentation broth media using anion-exchange chromatography with

integrated pulsed amperometric detection. Analytical biochemistry, 324, 29-38.

HARDER, W. & VEENHUIS, M. 1989. Metabolism of one-carbon compounds. The

yeasts, 3, 289-316.

HEISS, S., MAURER, M., HAHN, R., MATTANOVICH, D. & GASSER, B. 2013.

Identification and deletion of the major secreted protein of Pichia pastoris.

Applied microbiology and biotechnology, 97, 1241-1249.

HÉLÈNE, B., CÉLINE, L., PATRICK, C., FABIEN, R., CHRISTINE, V., YVES, C.

& GUY, M. 2001. High-level secretory production of recombinant porcine

follicle-stimulating hormone by Pichia pastoris. Process Biochemistry, 36, 907-

913.

HELLWIG, S., EMDE, F., RAVEN, N. P., HENKE, M., VAN DER LOGT, P. &

FISCHER, R. 2001. Analysis of single‐chain antibody production in Pichia

pastoris using on‐line methanol control in fed‐batch and mixed‐feed

fermentations. Biotechnology and bioengineering, 74, 344-352.

Page 145: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

145

HENSING, M., ROUWENHORST, R., HEIJNEN, J., VAN DIJKEN, J. & PRONK, J.

1995. Physiological and technological aspects of large-scale heterologous-

protein production with yeasts. Antonie van Leeuwenhoek, 67, 261-279.

HEYLAND, J., FU, J., BLANK, L. M. & SCHMID, A. 2011. Carbon metabolism limits

recombinant protein production in Pichia pastoris. Biotechnology and

Bioengineering, 108, 1942-1953.

HIGGINS, D. R. 2001. Overview of protein expression in Pichia pastoris. Current

protocols in protein science, 5.7.1-5.7. 18.

HILT, W. & WOLF, D. H. 1992. Stress‐induced proteolysis in yeast. Molecular

microbiology, 6, 2437-2442.

HUANG, C.-J., DAMASCENO, L. M., ANDERSON, K. A., ZHANG, S., OLD, L. J.

& BATT, C. A. 2011. A proteomic analysis of the Pichia pastoris secretome in

methanol-induced cultures. Applied microbiology and biotechnology, 90, 235-

247.

HUANG, L.-Y., DUMONTELLE, J. L., ZOLODZ, M., DEORA, A., MOZIER, N. M.

& GOLDING, B. 2009. Use of toll-like receptor assays to detect and identify

microbial contaminants in biological products. Journal of clinical microbiology,

47, 3427-3434.

HUTCHINSON, N., BINGHAM, N., MURRELL, N., FARID, S. & HOARE, M. 2006.

Shear stress analysis of mammalian cell suspensions for prediction of industrial

centrifugation and its verification. Biotechnology and bioengineering, 95, 483-

491.

IGUNNU, E. T. & CHEN, G. Z. 2012. Produced water treatment technologies.

International Journal of Low-Carbon Technologies, 9, 157-177.

IHL, M. & TAGLE, M. A. 1974. Estimation of protein in yeast. Journal of the science

of food and agriculture, 25, 461-464.

INVITROGEN. 2002. Pichia Fermentation Process Guidelines [Online].

https://tools.thermofisher.com/content/sfs/manuals/pichiaferm_prot.pdf.

[Accessed].

JAHIC, M., ROTTICCI-MULDER, J., MARTINELLE, M., HULT, K. & ENFORS,

S.-O. 2002. Modeling of growth and energy metabolism of Pichia pastoris

Page 146: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

146

producing a fusion protein. Bioprocess and Biosystems Engineering, 24, 385-

393.

JAHIC, M., WALLBERG, F., BOLLOK, M., GARCIA, P. & ENFORS, S.-O. 2003.

Temperature limited fed-batch technique for control of proteolysis in Pichia

pastoris bioreactor cultures. Microbial cell factories, 2, 6.

JENZSCH, M., LANGE, M., BÄR, J., RAHFELD, J.-U. & LÜBBERT, A. 2004.

Bioreactor retrofitting to avoid aeration with oxygen in Pichia pastoris

cultivation processes for recombinant protein production. Chemical

Engineering Research and Design, 82, 1144-1152.

JOHNSON, I. S. 1983. Human insulin from recombinant DNA technology. Science,

219, 632-637.

JONES, E. W. 1991. Three proteolytic systems in the yeast Saccharomyces cerevisiae.

J. Biol. Chem, 266, 7963-7966.

JUNGO, C., MARISON, I. & VON STOCKAR, U. 2007. Mixed feeds of glycerol and

methanol can improve the performance of Pichia pastoris cultures: A

quantitative study based on concentration gradients in transient continuous

cultures. Journal of biotechnology, 128, 824-837.

KALIDAS, C., JOSHI, L. & BATT, C. 2001. Characterization of glycosylated variants

of β-lactoglobulin expressed in Pichia pastoris. Protein engineering, 14, 201-

207.

KATAKURA, Y., ZHANG, W., ZHUANG, G., OMASA, T., KISHIMOTO, M.,

GOTO, Y. & SUGA, K.-I. 1998. Effect of methanol concentration on the

production of human β2-glycoprotein I domain V by a recombinant Pichia

pastoris: a simple system for the control of methanol concentration using a

semiconductor gas sensor. Journal of Fermentation and Bioengineering, 86, 482-487.

KHATRI, N. K. & HOFFMANN, F. 2006. Impact of methanol concentration on

secreted protein production in oxygen‐limited cultures of recombinant Pichia

pastoris. Biotechnology and bioengineering, 93, 871-879.

KIEHL, T. R., SHEN, D., KHATTAK, S. F., JIAN LI, Z. & SHARFSTEIN, S. T. J. C.

P. A. 2011. Observations of cell size dynamics under osmotic stress. 79, 560-

569.

Page 147: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

147

KIM, J. Y., KIM, Y.-G. & LEE, G. M. 2012. CHO cells in biotechnology for production

of recombinant proteins: current state and further potential. Applied

microbiology and biotechnology, 93, 917-930.

KOBAYASHI, K., KUWAE, S., OHYA, T., OHDA, T., OHYAMA, M., OHI, H.,

TOMOMITSU, K. & OHMURA, T. 2000. High-level expression of

recombinant human serum albumin from the methylotrophic yeast Pichia

pastoris with minimal protease production and activation. Journal of bioscience

and bioengineering, 89, 55-61.

KÖRNER, F. 2013. Evaluation of a chemically defined medium for Pichia pastoris

high cell density fermentation process [Online]. http://edoc.sub.uni-

hamburg.de/haw/volltexte/2014/2193/pdf/lsab13_65_BA_BT.pdf. [Accessed].

KUČERA, I. & SEDLÁČEK, V. 2017. An Enzymatic Method for Methanol

Quantification in Methanol/Ethanol Mixtures with a Microtiter Plate

Fluorometer. Food Analytical Methods, 10, 1301-1307.

KUMAR, A., BICER, E. M., MORGAN, A. B., PFEFFER, P. E., MONOPOLI, M.,

DAWSON, K. A., ERIKSSON, J., EDWARDS, K., LYNHAM, S. & ARNO,

M. 2016. Enrichment of immunoregulatory proteins in the biomolecular corona

of nanoparticles within human respiratory tract lining fluid. Nanomedicine:

Nanotechnology, Biology and Medicine, 12, 1033-1043.

KUNERT, R. & REINHART, D. 2016. Advances in recombinant antibody

manufacturing. Applied microbiology and biotechnology, 100, 3451-3461.

LAGASSÉ, H. D., ALEXAKI, A., SIMHADRI, V. L., KATAGIRI, N. H.,

JANKOWSKI, W., SAUNA, Z. E. & KIMCHI-SARFATY, C. 2017. Recent

advances in (therapeutic protein) drug development. F1000Research, 6.

LAI, T., YANG, Y. & NG, S. K. 2013. Advances in mammalian cell line development

technologies for recombinant protein production. Pharmaceuticals, 6, 579-603.

LAUER, B., OTTLEBEN, I., JACOBSEN, H.-J. & REINARD, T. 2005. Production of

a single-chain variable fragment antibody against fumonisin B1. Journal of

agricultural and food chemistry, 53, 899-904.

LEE, C. Y., NAKANO, A., SHIOMI, N., LEE, E. K. & KATOH, S. 2003a. Effects of

substrate feed rates on heterologous protein expression by Pichia pastoris in

Page 148: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

148

DO-stat fed-batch fermentation. Enzyme and Microbial Technology, 33, 358-

365.

LEE, S., CHU, C., TAN, R., WANG, C. & LEE, D. 2003b. Consolidation dewatering

and centrifugal sedimentation of flocculated activated sludge. Chemical

Engineering Science, 58, 1687-1701.

LEVIN, D., GOLDING, B., STROME, S. E. & SAUNA, Z. E. 2015. Fc fusion as a

platform technology: potential for modulating immunogenicity. Trends in

biotechnology, 33, 27-34.

LEVY, N. E., VALENTE, K. N., CHOE, L. H., LEE, K. H. & LENHOFF, A. M. 2014.

Identification and characterization of host cell protein product‐associated

impurities in monoclonal antibody bioprocessing. Biotechnology and

bioengineering, 111, 904-912.

LI, X. & LI, J. 2015. Dead-end filtration. Encyclopedia of Membranes, 1-3.

LIM, H.-K., CHOI, S.-J., KIM, K.-Y. & JUNG, K.-H. 2003. Dissolved-oxygen-stat

controlling two variables for methanol induction of rGuamerin in Pichia

pastoris and its application to repeated fed-batch. Applied microbiology and

biotechnology, 62, 342-348.

LIM, H. C. & SHIN, H. S. 2013. Fed-Batch Cultures, Cambridge University Press.

LIU, N., BREVNOV, M., FURTADO, M. & LIU, J. 2012. Host cellular protein

quantification. Bioprocess Int, 10, 44-50.

LIU, W.-C., GONG, T., WANG, Q.-H., LIANG, X., CHEN, J.-J. & ZHU, P. 2016.

Scaling-up Fermentation of Pichia pastoris to demonstration-scale using new

methanol-feeding strategy and increased air pressure instead of pure oxygen

supplement. Scientific reports, 6.

LIU, Y. Y., WOO, J. H. & NEVILLE, D. M. 2005. Overexpression of an anti-CD3

immunotoxin increases expression and secretion of molecular chaperone

BiP/Kar2p by Pichia pastoris. Applied and environmental microbiology, 71, 5332-5340.

LOPES, A. & KESHAVARZ‐MOORE, E. 2012. Prediction and verification of

centrifugal dewatering of P. pastoris fermentation cultures using an ultra scale‐

down approach. Biotechnology and bioengineering, 109, 2039-2047.

Page 149: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

149

LOPES, A. G. 2013. Development of ultra scale-down methodologies for the prediction

of centrifugal dewatering of high cell density yeast cultures. UCL (University

College London).

LOPES, A. G., KHAN, N., LIDDELL, J. & KESHAVARZ‐MOORE, E. 2012. An ultra

scale‐down approach to assess the impact of the choice of recombinant P.

pastoris strain on dewatering performance in centrifuges. Biotechnology

progress, 28, 1029-1036.

MACAULEY‐PATRICK, S., FAZENDA, M. L., MCNEIL, B. & HARVEY, L. M.

2005. Heterologous protein production using the Pichia pastoris expression

system. Yeast, 22, 249-270.

MACCANI, A., LANDES, N., STADLMAYR, G., MARESCH, D., LEITNER, C.,

MAURER, M., GASSER, B., ERNST, W., KUNERT, R. & MATTANOVICH,

D. 2014. Pichia pastoris secretes recombinant proteins less efficiently than

Chinese hamster ovary cells but allows higher space‐time yields for less

complex proteins. Biotechnology journal, 9, 526-537.

MATTANOVICH, D., GRAF, A., STADLMANN, J., DRAGOSITS, M., REDL, A.,

MAURER, M., KLEINHEINZ, M., SAUER, M., ALTMANN, F. & GASSER,

B. 2009. Genome, secretome and glucose transport highlight unique features of

the protein production host Pichia pastoris. Microbial cell factories, 8, 29.

MATTHEWS, C. B., KUO, A., LOVE, K. R. & LOVE, J. C. 2018. Development of a

general defined medium for Pichia pastoris. Biotechnology and bioengineering,

115, 103-113.

MAYBURY, J., MANNWEILER, K., TITCHENER-HOOKER, N., HOARE, M. &

DUNNILL, P. 1998. The performance of a scaled down industrial disc stack

centrifuge with a reduced feed material requirement. Bioprocess Engineering,

18, 191-199.

MCGREW, J. T., LEISKE, D., DELL, B., KLINKE, R., KRASTS, D., WEE, S.,

ABBOTT, N., ARMITAGE, R. & HARRINGTON, K. 1997. Expression of

trimeric CD40 ligand in Pichia pastoris: use of a rapid method to detect high-

level expressing transformants. Gene, 187, 193-200.

Page 150: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

150

MINNING, S., SERRANO, A., FERRER, P., SOLÁ, C., SCHMID, R. D. & VALERO,

F. 2001. Optimization of the high-level production of Rhizopus oryzae lipase in

Pichia pastoris. Journal of Biotechnology, 86, 59-70.

MOCHIZUKI, S., HAMATO, N., HIROSE, M., MIYANO, K., OHTANI, W.,

KAMEYAMA, S., KUWAE, S., TOKUYAMA, T. & OHI, H. 2001.

Expression and characterization of recombinant human antithrombin III in

Pichia pastoris. Protein expression and purification, 23, 55-65.

NESTA, N., HE, H., SHPUNGIN, R. & SWEDER, B. 2017. Aggregation from Shear

Stress and Surface Interaction: Molecule-Specific or Universal Phenomenon?

Bioprocess Intl. Ebook

NIU, H., JOST, L., PIRLOT, N., SASSI, H., DAUKANDT, M., RODRIGUEZ, C. &

FICKERS, P. 2013. A quantitative study of methanol/sorbitol co-feeding

process of a Pichia pastoris Mut+/pAOX1-lacZ strain. Microbial cell factories,

12, 33.

OBRIEN, T. P., BROWN, L. A., BATTERSBY, D. G., RUDOLPH, A. S. & RAMAN,

L. 2012. Large-Scale, Single-Use Depth Filtration Systems. Bioprocess

International.

OLSON, B., MARKWELL, J., COLIGAN, J., DUNN, B., SPEICHER, D. &

WINGFIELD, P. 2007. Current protocols in protein science. Detection and

Assay Method, 48, 1-3.4.

OVERTON, T. W. 2014. Recombinant protein production in bacterial hosts. Drug

discovery today, 19, 590-601.

PARPINELLO, G. P. & VERSARI, A. 2000. A simple high-performance liquid

chromatography method for the analysis of glucose, glycerol, and methanol in

a bioprocess. Journal of chromatographic science, 38, 259-261.

PHILIPPIDIS, A. 2017. The Top 15 Best-Selling Drugs of 2016 [Online]. Genengnews.

[Accessed].

PLA, I. A., DAMASCENO, L. M., VANNELLI, T., RITTER, G., BATT, C. A. &

SHULER, M. L. 2006. Evaluation of Mut+ and MutS Pichia pastoris

phenotypes for high level extracellular scFv expression under feedback control

of the methanol concentration. Biotechnology progress, 22, 881-888.

Page 151: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

151

POLEZ, S., ORIGI, D., ZAHARIEV, S., GUARNACCIA, C., TISMINETZKY, S. G.,

SKOKO, N. & BARALLE, M. 2016. A Simplified and Efficient Process for

Insulin Production in Pichia pastoris. PloS one, 11, e0167207.

POTGIETER, T. I., CUKAN, M., DRUMMOND, J. E., HOUSTON-CUMMINGS, N.

R., JIANG, Y., LI, F., LYNAUGH, H., MALLEM, M., MCKELVEY, T. W. &

MITCHELL, T. 2009. Production of monoclonal antibodies by glycoengineered

Pichia pastoris. Journal of biotechnology, 139, 318-325.

POTGIETER, T. I., KERSEY, S. D., MALLEM, M. R., NYLEN, A. C. & D'ANJOU,

M. 2010a. Antibody expression kinetics in glycoengineered Pichia pastoris.

Biotechnology and Bioengineering, 6, 918-927.

POTGIETER, T. I., KERSEY, S. D., MALLEM, M. R., NYLEN, A. C. & D'ANJOU,

M. 2010b. Antibody expression kinetics in glycoengineered Pichia pastoris.

Biotechnology and Bioengineering, 106, 918-927.

POTVIN, G., AHMAD, A. & ZHANG, Z. 2012. Bioprocess engineering aspects of

heterologous protein production in Pichia pastoris: a review. Biochemical

Engineering Journal, 64, 91-105.

PRIELHOFER, R., MAURER, M., KLEIN, J., WENGER, J., KIZIAK, C., GASSER,

B. & MATTANOVICH, D. 2013. Induction without methanol: novel regulated

promoters enable high-level expression in Pichia pastoris. Microbial cell

factories, 12, 5.

PYBUS, L. P., JAMES, D. C., DEAN, G., SLIDEL, T., HARDMAN, C., SMITH, A.,

DARAMOLA, O. & FIELD, R. 2014. Predicting the expression of recombinant

monoclonal antibodies in Chinese hamster ovary cells based on sequence

features of the CDR3 domain. Biotechnology progress, 30, 188-197.

RADER, R. A. 2018. The Commercial Expression Systems Market — What Has

Changed in the Past Decade. Bioprocess International. Ebook

RAMON, R., FELIU, J., COS, O., MONTESINOS, J., BERTHET, F. & VALERO, F.

2004. Improving the monitoring of methanol concentration during high cell

density fermentation of Pichia pastoris. Biotechnology letters, 26, 1447-1452.

RAMÓN, R., FERRER, P. & VALERO, F. 2007. Sorbitol co-feeding reduces

metabolic burden caused by the overexpression of a Rhizopus oryzae lipase in

Pichia pastoris. Journal of Biotechnology, 130, 39-46.

Page 152: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

152

REBNEGGER, C., VOS, T., GRAF, A. B., VALLI, M., PRONK, J. T., DARAN-

LAPUJADE, P. & MATTANOVICH, D. 2016. Pichia pastoris Exhibits High

Viability and a Low Maintenance Energy Requirement at Near-Zero Specific

Growth Rates. Applied and environmental microbiology, 82, 4570-4583.

RUPA, P., NAKAMURA, S. & MINE, Y. 2007. Genetically glycosylated ovomucoid

third domain can modulate immunoglobulin E antibody production and

cytokine response in BALB/c mice. Clinical & Experimental Allergy, 37, 918-

928.

SALTE, H. 2006. Rapid evaluation of options for the primary recovery of antibody

fragments expressed in high cell density cultures. University of London.

SALTE, H., KING, J. M., BAGANZ, F., HOARE, M. & TITCHENER‐HOOKER, N.

J. 2006. A methodology for centrifuge selection for the separation of high solids

density cell broths by visualisation of performance using windows of operation.

Biotechnology and bioengineering, 95, 1218-1227.

SAMBROOK, J., FRITSCH, E. F. & MANIATIS, T. 1989. Molecular cloning: a

laboratory manual, Cold spring harbor laboratory press.

SAMPATH, M., SHUKLA, A. & RATHORE, A. S. 2014. Modeling of filtration

processes—microfiltration and depth filtration for harvest of a therapeutic

protein expressed in Pichia pastoris at constant pressure. Bioengineering, 1, 260-277.

SANCHEZ-GARCIA, L., MARTÍN, L., MANGUES, R., FERRER-MIRALLES, N.,

VÁZQUEZ, E. & VILLAVERDE, A. 2016. Recombinant pharmaceuticals

from microbial cells: a 2015 update. Microbial cell factories, 15, 33.

SCHENK, J., BALAZS, K., JUNGO, C., URFER, J., WEGMANN, C., ZOCCHI, A.,

MARISON, I. W. & VON STOCKAR, U. 2008. Influence of specific growth

rate on specific productivity and glycosylation of a recombinant avidin

produced by a Pichia pastoris Mut+ strain. Biotechnology and bioengineering,

99, 368-377.

SCHILLING, B. M., GOODRICK, J. C. & WAN, N. C. 2001. Scale‐Up of a High Cell

Density Continuous Culture with Pichia pastoris X‐33 for the Constitutive

Expression of rh‐Chitinase. Biotechnology progress, 17, 629-633.

Page 153: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

153

SCHLÜTER, V. & DECKWER, W.-D. 1992. Gas/liquid mass transfer in stirred vessels.

Chemical engineering science, 47, 2357-2362.

SENGER, R. S. & KARIM, M. N. 2003. Effect of Shear Stress on Intrinsic CHO

Culture State and Glycosylation of Recombinant Tissue‐Type Plasminogen

Activator Protein. Biotechnology progress, 19, 1199-1209.

SHI, X., KARKUT, T., CHAMANKHAH, M., ALTING-MEES, M., HEMMINGSEN,

S. M. & HEGEDUS, D. 2003. Optimal conditions for the expression of a single-

chain antibody (scFv) gene in Pichia pastoris. Protein expression and

purification, 28, 321-330.

SHUKLA, A. A. & KANDULA, J. R. 2008. Harvest and recovery of monoclonal

antibodies from large-scale mammalian cell culture. BioPharm International,

Ebook

SIGMA-ALDRICH-a. Enzymatic Assay of Aprotinin [Online].

https://www.sigmaaldrich.com/china-mainland/zh/technical-

documents/protocols/biology/enzymatic-assay-of-aprotinin.html. [Accessed].

SIGMA-ALDRICH-b. Enzymatic Assay of Alcohol Oxidase (EC 1.1.3.13) [Online].

https://www.sigmaaldrich.com/china-mainland/zh/technical-

documents/protocols/biology/enzymatic-assay-of-alcohol-oxidase.html.

[Accessed].

SINGH, S., GRAS, A., FIEZ-VANDAL, C., RUPRECHT, J., RANA, R., MARTINEZ,

M., STRANGE, P. G., WAGNER, R. & BYRNE, B. 2008. Large-scale

functional expression of WT and truncated human adenosine A 2A receptor in

Pichia pastoris bioreactor cultures. Microbial cell factories, 7, 28.

SINHA, J., PLANTZ, B. A., INAN, M. & MEAGHER, M. M. 2005. Causes of

proteolytic degradation of secreted recombinant proteins produced in

methylotrophic yeast Pichia pastoris: Case study with recombinant ovine

interferon‐τ. Biotechnology and bioengineering, 89, 102-112.

SPELLMAN, M. W., BASA, L. J., LEONARD, C. K., CHAKEL, J., O'CONNOR, J.,

WILSON, S. & VAN HALBEEK, H. 1989. Carbohydrate structures of human

tissue plasminogen activator expressed in Chinese hamster ovary cells. Journal

of Biological Chemistry, 264, 14100-14111.

Page 154: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

154

STRATTON, J., CHIRUVOLU, V. & MEAGHER, M. 1998. High cell-density

fermentation. Pichia protocols, 107-120.

STROHL, W. R. 2015. Fusion proteins for half-life extension of biologics as a strategy

to make biobetters. BioDrugs, 29, 215-239.

TAIT, A. S., HOGWOOD, C. E., SMALES, C. M. & BRACEWELL, D. G. 2012. Host

cell protein dynamics in the supernatant of a mAb producing CHO cell line.

Biotechnology and bioengineering, 109, 971-982.

THOMAS PURKARTHOFER, I. D., EVELYN TRUMMER-GÖDL 2017. what about

pichia. Achievements in Protein Manufacture. Ebook

THOR, D., XIONG, S., ORAZEM, C. C., KWAN, A.-C., CREGG, J. M., LIN-

CEREGHINO, J. & LIN-CEREGHINO, G. P. 2005. Cloning and

characterization of the Pichia pastoris MET2 gene as a selectable marker.

FEMS yeast research, 5, 935-942.

TOLNER, B., SMITH, L., BEGENT, R. H. & CHESTER, K. A. 2006. Production of

recombinant protein in Pichia pastoris by fermentation. Nature protocols, 1, 1006-1021.

TRINH, L., PHUE, J. & SHILOACH, J. 2003. Effect of methanol feeding strategies on

production and yield of recombinant mouse endostatin from Pichia pastoris.

Biotechnology and Bioengineering, 82, 438-444.

TSCHOPP, J. F., BRUST, P. F., CREGG, J. M., STILLMAN, C. A. & GINGERAS, T.

R. 1987. Expression of the lacZ gene from two methanol-regulated promoters

in Pichia pastoris. Nucleic acids research, 15, 3859-3876.

UNIPROT. N.A. UniProtKB - P00974 (BPT1_BOVIN) [Online].

http://www.uniprot.org/uniprot/P00974. [Accessed].

USMANI, S. S., BEDI, G., SAMUEL, J. S., SINGH, S., KALRA, S., KUMAR, P.,

AHUJA, A. A., SHARMA, M., GAUTAM, A. & RAGHAVA, G. P. 2017.

THPdb: Database of FDA-approved peptide and protein therapeutics. PloS one,

12, e0181748.

VALERO, F. 2013. Bioprocess engineering of Pichia pastoris, an exciting host

eukaryotic cell expression system. Protein Engineering-Technology and

Application. InTech.

Page 155: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

155

VAN DEN HAZEL, H. B., KIELLAND‐BRANDT, M. C. & WINTHER, J. R. 1996.

Biosynthesis and function of yeast vacuolar proteases. Yeast, 12, 1-16.

VAN DER VALK, J., BRUNNER, D., DE SMET, K., SVENNINGSEN, Å. F.,

HONEGGER, P., KNUDSEN, L. E., LINDL, T., NORABERG, J., PRICE, A.

& SCARINO, M. 2010. Optimization of chemically defined cell culture media–

replacing fetal bovine serum in mammalian in vitro methods. Toxicology in

vitro, 24, 1053-1063.

VOGL, T., STURMBERGER, L., FAULAND, P. C., HYDEN, P., FISCHER, J. E.,

SCHMID, C., THALLINGER, G. G., GEIER, M. & GLIEDER, A. 2018.

Methanol independent induction in Pichia pastoris by simple derepressed

overexpression of single transcription factors. Biotechnology and

bioengineering, 115, 1037-1050.

WALLS, P. L., MCRAE, O., NATARAJAN, V., JOHNSON, C., ANTONIOU, C. &

BIRD, J. C. 2017. Quantifying the potential for bursting bubbles to damage

suspended cells. Scientific Reports, 7, 15102.

WANG, A., LEWUS, R. & RATHORE, A. S. 2006. Comparison of different options

for harvest of a therapeutic protein product from high cell density yeast

fermentation broth. Biotechnology and bioengineering, 94, 91-104.

WANG, F., RICHARDSON, D. & SHAMEEM, M. 2015. Host-cell protein

measurement and control. Biopharm Int, 28, 32-38.

WANG, X., HUNTER, A. K. & MOZIER, N. M. 2009. Host cell proteins in biologics

development: Identification, quantitation and risk assessment. Biotechnology

and bioengineering, 103, 446-458.

WANG, Z., WANG, Y., ZHANG, D., LI, J., HUA, Z., DU, G. & CHEN, J. 2010.

Enhancement of cell viability and alkaline polygalacturonate lyase production

by sorbitol co-feeding with methanol in Pichia pastoris fermentation.

Bioresource technology, 101, 1318-1323.

WATERHAM, H. R., DIGAN, M. E., KOUTZ, P. J., LAIR, S. V. & CREGG, J. M.

1997. Isolation of the Pichia pastoris glyceraldehyde-3-phosphate

dehydrogenase gene and regulation and use of its promoter. Gene, 186, 37-44.

Page 156: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

156

WOODHOUSE, S. A. 2016. An upstream platform for the production of high grade

heterologous proteins in the yeast Pichia pastoris. UCL (University College

London).

WU, C. C., HUANG, C. & LEE, D. 1997. Effects of polymer dosage on alum sludge

dewatering characteristics and physical properties. Colloids and Surfaces A:

Physicochemical and Engineering Aspects, 122, 89-96.

YAMANÈ, T. & SHIMIZU, S. 1984. Fed-batch techniques in microbial processes.

Bioprocess parameter control. Springer.

YANG, M. & BUTLER, M. 2000. Enhanced erythropoietin heterogeneity in a CHO

culture is caused by proteolytic degradation and can be eliminated by a high

glutamine level. Cytotechnology, 34, 83-99.

YAVORSKY, D., BLANCK, R., LAMBALOT, C. & BRUNKOW, R. 2003. The

clarification of bioreactor cell cultures for biopharmaceuticals. Pharmaceutical

technology, 27, 62-77.

YE, J., LY, J., WATTS, K., HSU, A., WALKER, A., MCLAUGHLIN, K.,

BERDICHEVSKY, M., PRINZ, B., SEAN KERSEY, D. & D'ANJOU, M.

2011. Optimization of a glycoengineered Pichia pastoris cultivation process for

commercial antibody production. Biotechnology progress, 27, 1744-1750.

ZHA, D. 2012. Glycoengineered yeast as an alternative monoclonal antibody discovery

and production platform. Glycosylation. InTech.

ZHANG, Q., GOETZE, A. M., CUI, H., WYLIE, J., TRIMBLE, S., HEWIG, A. &

FLYNN, G. C. Comprehensive tracking of host cell proteins during monoclonal

antibody purifications using mass spectrometry. MAbs, 2014, 659-670.

ZHANG, W., BEVINS, M. A., PLANTZ, B. A., SMITH, L. A. & MEAGHER, M. M.

2000. Modeling Pichia pastoris growth on methanol and optimizing the

production of a recombinant protein, the heavy-chain fragment C of botulinum

neurotoxin, serotype A. Papers in Biotechnology, 18.

ZHANG, W., POTTER, K. J. H., PLANTZ, B. A., SCHLEGEL, V. L., SMITH, L. A.

& MEAGHER, M. M. 2003. Pichia pastoris fermentation with mixed-feeds of

glycerol and methanol: growth kinetics and production improvement. Journal

of Industrial Microbiology and Biotechnology, 30, 210-215.

Page 157: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

157

ZHANG, Y., LIU, R. & WU, X. 2007. The proteolytic systems and heterologous

proteins degradation in the methylotrophic yeast Pichia pastoris. Annals of

Microbiology, 57, 553.

ZHU, T., YOU, L., GONG, F., XIE, M., XUE, Y., LI, Y. & MA, Y. 2011.

Combinatorial strategy of sorbitol feeding and low-temperature induction leads

to high-level production of alkaline β-mannanase in Pichia pastoris. Enzyme

and microbial technology, 49, 407-412.

Page 158: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

158

Chapter 8 Appendix

Figure 8.1 Detection of glycerol, methanol and sorbitol using UltiMate 3000 HPLC

with Aminex HPX-87h column. 0.5% (v/v) trifluoroacetic acid at flow rate of 0.6 ml•h-

1 was used as the mobile phase. Standard solutions with 1.0 g•L-1 of glycerol, methanol

and sorbitol were prepared, respectively. 20 μl solution was loaded through the column

for 30min, and absorbance was detected by the RefractoMax 520 Refractive Index

Detector.

Page 159: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

159

Figure 8.2 Correlation between peak area and concentration of methanol (A) and

sorbitol (B). 1%, 5%, 10%, 20% and 30% (v/v) of methanol and sorbitol solutions

(0.75g•L-1) were prepared and assayed using the method described in the Section 2.9.3,

Chapter 3.

Page 160: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

160

Figure 8.3 Viscosities of methanol, 0.75 g•ml-1 sorbitol solution and sorbitol/methanol

(1:1, C-mol/C-mol) mixture at the shear rate of 800s-1.

Figure 8.4 The impact of biomass concentration and treating time on specific protein

release.

0 400 800 12000.000

0.005

0.010

0.015

0.020

Shear rate s-1

Vis

cosi

ty P

a s

Methanol Sorbitol Mixture

Page 161: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

161

Figure 8.5 Correlation between DCW and WCW of P. pastoris

Page 162: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

162

Table 8.1 A summary of names, accession number, molecular weight, isoelectric point, localization and function of the HCPs identified from the cell cultures induced by methanol or sorbitol/methanol (1:1, C-mol/C-mol) mixture.

Methanol Name No. MW PI Localization Function

1 Vacuolar proteinase B (YscB) C4QYT0 52 5.51 Secreted protease

2 Vacuolar proteinase B (YscB) C4QWH2 59 5.93 Vacuole protease, cellular response to starvationfamily

3 Vacuolar aspartyl protease (Proteinase A)

C4R6G8 44 4.64 Vacuole protease, cellular response to starvation

4 Uncharacterized protein C4R862 42 7.19 Cytosol Nucleus Peroxisome

aspartate biosynthetic process

5 Uncharacterized protein C4R8H7 64 3.96 Extracellular region or secreted

cellulase activity

6 Uncharacterized protein C4R743 34 5.19 N/A N/A

7 Uncharacterized protein C4R6P1 26 4.75 N/A calcium ion binding

8 Uncharacterized protein C4R3Q7 61 4.18 N/A aspartic-type endopeptidase activity

Page 163: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

163

9 Uncharacterized protein C4R6F1 13 8.60 N/A N/A

10 Uncharacterized protein C4R5N2 95 6.45 Membrane N/A

11 Uncharacterized protein C4R4B2 24 6.06 N/A N/A

12 Uncharacterized protein C4R489 32 4.44 N/A N/A

13 Uncharacterized protein C4R3H3 32 5.26 Extracellular region or secreted

N/A

14 Uncharacterized protein C4R3C4 63 4.26 N/A N/A

15 Uncharacterized protein C4R3B1 41 3.84 N/A N/A

16 Uncharacterized protein C4R3A0 63 3.90 N/A N/A

17 Uncharacterized protein C4R2D7 91 4.02 N/A N/A

18 Uncharacterized protein C4R2B9 45 4.15 Cell Wall N/A

19 Uncharacterized protein C4R2M0 27 6.83 N/A N/A

20 Uncharacterized protein C4R1Q1 31 4.17 Cell Wall structural constituent of cell wall

21 Uncharacterized protein C4R0W4 57 5.94 N/A oxidoreductase activity

Page 164: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

164

22 Uncharacterized protein C4QZQ5 93 6.28 N/A GTPase activity

23 Uncharacterized protein C4QZN3 29 4.80 N/A protein domain specific binding

24 Uncharacterized protein C4R0Z8 24 6.09 N/A N/A

25 Uncharacterized protein C4QY91 10 4.83 N/A atty-acyl-CoA binding

26 Uncharacterized protein C4QXM2 31 4.25 Extracellular region or secreted

N/A

27 Uncharacterized protein C4QUZ5 21 5.73 N/A catalytic activity

28 Uncharacterized protein C4QWA6 26 5.69 N/A catalytic activity

29 Uncharacterized protein C4QW56 46 7.05 N/A N/A

30 Uncharacterized protein C4QW08 45 5.40 Membrane N/A

31 Uncharacterized protein C4R9F6 45 5.08 N/A N/A

32 Transketolase, similar to Tkl2p C4R5Q0 79 6.03 N/A catalytic activity

33 Transketolase, similar to Tkl2p C4R5P8 79 6.37 N/A catalytic activity

34 Thioredoxin reductase C4R2E7 34 5.23 cytoplasm Catalytic activity

Page 165: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

165

35 Thiol-specific peroxiredoxin C4R0V9 19 8.63 N/A oxidoreductase activity

36 Suppressor protein STM1 C4QY12 30 9.73 N/A N/A

37 Superoxide dismutase [Cu-Zn] C4R8X7 16 5.92 N/A Catalytic activity

38 Superoxide dismutase C4QXC7 25 7.89 Mitochondrion Catalytic activity

39 Subunit of the RAVE complex (Rav1p, Rav2p, Skp1p)

C4R0E0 158 6.47 N/A N/A

40 Sortilin C4R564 167 4.97 Golgi apparatus hydrolase activity

41 S-adenosylmethionine synthase C4R5U7 42 6.06 Cytosol Nucleus Catalytic activity

42 S-(hydroxymethyl)glutathione dehydrogenase

C4R6A5 41 6.08 Cytosol Nucleus Catalytic activity

43 Putative chitin transglycosidase C4R894 50 4.31 Cell Wall hydrolase activity

44 Protein of the SUN family C4R2Z5 45 4.37 N/A Glycoside Hydrolase Family

45 Plasma membrane localized protein that protects membranes from desiccation

C4R8G2 12 4.81 Cytosol Endosome Plasma Membrane

cell adhesion cellular response to stress

Page 166: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

166

46 Phosphoglycerate mutase C4R5P4 28 6.03 Cytosol Mitochondrion Catalytic activity

47 Phosphatidylglycerol/phosphatidylinositol transfer protein

C4QZC2 19 4.49 introcellular intracellular sterol transport

48 Peptide hydrolase C4QWD7 46 5.14 N/A metal ion binding

49 Ornithine carbamoyltransferase C4R533 38 6.46 Mitochondrion amino acid binding

50 O-glycosylated protein required for cell wall stability

C4R7G9 34 4.50 Cell Wall structural constituent of cell wall

51 Nucleoside diphosphate kinase C4R300 17 6.13 Cytosol Nucleus Catalytic activity

52 Mucin family member C4QVR8 89 3.82 Membrane N/A

53 Mitochondrial protein involved in maintenance of the mitochondrial genome

C4QXV0 16 8.91 N/A N/A

54 Mitochondrial porin (Voltage-dependent anion channel)

C4R1Z2 30 9.02 Mitochondrion voltage-gated anion channel activity

55 Mitochondrial outer membrane and cell wall localized SUN family member

C4R6P9 36 4.90 N/A Glycoside Hydrolase Family

Page 167: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

167

56 Mitochondrial matrix ATPase C4R4C3 70 5.41 Mitochondrion ATPase activity

57 Major exo-1,3-beta-glucanase of the cell wall

C4R0Q7 48 4.53 N/A Hydrolase activity

58 Lysophospholipase C4R703 70 4.11 N/A Catalytic activity

59 Lectin-like protein with similarity to Flo1p

C4QYW7 51 4.36 N/A N/A

60 Histone H2B C4R0M7 14 10.1 Nucleus DNA bindingprotein heterodimerization activity

61 Histone H2A C4R0M8 14 10.3 Nucleus DNA bindingprotein heterodimerization activity

62 Glycerol kinase C4R8X4 68 5.25 Mitochondrion glycerol kinase activity

63 Glyceraldehyde-3-phosphate

dehydrogenase C4R0P1 36 6.24 N/A Catalytic activity

64 Fusion protein C4R0U2 15 9.87 ribosome structural constituent of ribosome

65 Fructose-1,6-bisphosphatase C4R5T8 38 6.12 Cytosol Nucleus cellular response to glucose starvation

Page 168: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

168

66 Fructose 1,6-bisphosphate aldolase C4QWS2 39 6.08 N/A fructose-bisphosphate aldolase activity

67 Fructose 1,6-bisphosphate aldolase C4QW09 40 6.02 N/A fructose-bisphosphate aldolase activity

68 Formate dehydrogenase C4R606 40 6.61 cytoplasm Catalytic activity

69 Enolase I C4R3H8 47 5.44 Cytosol magnesium ion binding

70 Endo-beta-1,3-glucanase C4QYF3 34 4.07 N/A hydrolase activity

71 Elongation factor 1-alpha C4QZB0 50 9.12 cytoplasm GTPase activity

72 Dihydrolipoyl dehydrogenase C4R312 52 6.30 N/A dihydrolipoyl dehydrogenase activity

73 Daughter cell-specific secreted protein with similarity to glucanases

C4QW71 110 4.65 N/A glucan endo-1,3-beta-glucanase activity

74 Cytosolic and mitochondrial glutathione oxidoreductase

C4R686 50 8.13 Cytosol Mitochondrion electron carrier activity

75 Cytoplasmic ATPase that is a ribosome-associated molecular chaperone

C4R5E4 67 5.09 cytoplasm ATPase activity

Page 169: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

169

76 Cytochrome c, isoform 1 C4R6L9 12 9.61 Mitochondrion electron carrier activity

77 Cytochrome c oxidase assembly protein/Cu2+ chaperone

C4R5F9 7 k 6.13 Mitochondrion copper chaperone activity

78 Chitin deacetylase C4QW42 35 5.14 N/A hydrolase activity

79 Cell wall protein with similarity to glucanases

C4QZH9 49 4.44 N/A Glycoside Hydrolase Family

80 Cell wall protein with similarity to glucanases

C4QVL7 36 4.96 N/A hydrolase activity

81 Cell wall protein that contains a putative GPI-attachment site

C4R2Q4 43 3.87 N/A N/A

82 Catalase C4R2S1 58 6.56 N/A Catalytic activity

83 Carboxypeptidase C4R546 61 4.88 N/A serine-type carboxypeptidase

activity

84 Branched-chain-amino-acid aminotransferase

C4R7A4 45 5.81 N/A Catalytic activity

Page 170: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

170

85 Aspartic protease C4R458 52 4.19 N/A aspartic-type endopeptidase activity

86 Aspartate aminotransferase C4QWE4 48 6.68 Mitochondrion aminotransferase activity

87 Amine oxidase C4R098 90 4.40 N/A copper ion bindingprimary amine

oxidase activity

88 Alanine: glyoxylate aminotransferase (AGT)

C4R7U0 45 7.77 N/A transaminase activity

89 Acid trehalase required for utilization of extracellular trehalose

C4R7L0 114 4.79 N/A catalytic activity

90 ATPase involved in protein import into the ER

C4QZS3 74 4.78 N/A ATP binding

91 ATPase involved in protein folding and

the response to stress C4R3X8 71 5.09 N/A ATP binding

92 ATPase involved in protein folding and nuclear localization signal directed nuclear transport

C4R887 70 4.89 Cell Wall Cytosol Nucleus ATPase activity

Page 171: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

171

93 6,7-dimethyl-8-ribityllumazine synthase C4R6B4 18 6.30 MitochondrionOther locations

Catalytic activity

94 1,3-beta-glucanosyltransferase C4QVL5 58 4.00 Plasma Membrane transferase activity

95 1,3-beta-glucanosyltransferase C4QVL4 57 4.02 Plasma Membrane transferase activity

96 1,3-beta-glucanosyltransferase C4R9F4 54 4.12 Plasma Membrane transferase activity

Mixture Name No MW PI Localization Function

1 Vacuolar aspartyl protease (Proteinase A) C4R6G8 44 4.64 Vacuole protease, cellular response to starvation

2 Uncharacterized protein C4R862 42 7.19 Cytosol Nucleus Peroxisome aspartate biosynthetic process

3 Uncharacterized protein C4R7K4 50 9.43 Membrane N/A

4 Uncharacterized protein C4R8H7 64 3.96 Extracellular region or secreted cellulase activity

5 Uncharacterized protein C4R885 36 5.57 N/A N/A

6 Uncharacterized protein C4R3Q7 61 4.18 N/A aspartic-type endopeptidase activity

Page 172: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

172

7 Uncharacterized protein C4R6F1 13 8.60 N/A N/A

8 Uncharacterized protein C4R4B2 24 6.06 N/A N/A

9 Uncharacterized protein C4R489 32 4.44 N/A N/A

10 Uncharacterized protein C4R3H3 32 5.26 Extracellular region or secreted N/A

11 Uncharacterized protein C4R3C4 63 4.26 N/A N/A

12 Uncharacterized protein C4R3B1 41 3.84 N/A N/A

13 Uncharacterized protein C4R3B0 72 4.12 N/A N/A

14 Uncharacterized protein C4R3A0 63 3.90 N/A N/A

15 Uncharacterized protein C4R2D7 91 4.02 N/A N/A

16 Uncharacterized protein C4R2B9 45 4.15 Cell Wall N/A

17 Uncharacterized protein C4R2M0 27 6.83 N/A N/A

18 Uncharacterized protein C4R1Q1 31 4.17 Cell Wall structural constituent of cell wall

19 Uncharacterized protein C4QZQ5 93 6.28 N/A GTPase activity

Page 173: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

173

20 Uncharacterized protein C4R0Z8 24 6.09 N/A N/A

21 Uncharacterized protein C4QZD0 29 5.67 N/A N/A

22 Uncharacterized protein C4QY91 10 4.83 N/A atty-acyl-CoA binding

23 Uncharacterized protein C4QXM2 31 4.25 Extracellular region or secreted N/A

24 Uncharacterized protein C4R9F6 45 5.08 N/A N/A

25 Translation elongation factor EF-1 gamma

C4R6E8 24 6.41 N/A translation elongation factor activity

26 Transaldolase C4R245 36 5.07 Cytosol Nucleus Catalytic activity

27 Superoxide dismutase [Cu-Zn] C4R8X7 16 5.92 N/A Catalytic activity

28 Superoxide dismutase C4QXC7 25 7.89 Mitochondrion Catalytic activity

29 Putative chitin transglycosidase C4R894 50 4.31 Cell Wall hydrolase activity

30 Protein of the SUN family C4R2Z5 45 4.37 N/A Glycoside Hydrolase Family

31 Plasma membrane ATPase C4QVS9 98 4.96 Membrane Catalytic activity

Page 174: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

174

32 Phosphatidylglycerol/phosphatidylinositol transfer protein

C4QZC2 19 4.49 introcellular intracellular sterol transport

33 Peroxisomal 2,4-dienoyl-CoA reductase C4R8U1 31 5.60 N/A oxidoreductase activity

34 Peptide hydrolase C4QWD7 46 5.14 N/A metal ion binding

35 O-glycosylated protein required for cell wall stability

C4R7G9 34 4.50 Cell Wall structural constituent of cell wall

36 Nucleoside diphosphate kinase C4R300 17 6.13 Cytosol Nucleus Catalytic activity

37 Nuclear protein required for transcription of MXR1

C4R6V3 47 6.12 N/A translation elongation factor activity

38 Mucin family member C4QVR8 89 3.82 Membrane N/A

39 Mitochondrial protein involved in maintenance of the mitochondrial genome

C4QXV0 16 8.91 N/A N/A

40 Mitochondrial porin (Voltage-dependent anion channel)

C4R1Z2 30 9.02 Mitochondrion voltage-gated anion channel activity

41 Mitochondrial outer membrane and cell wall localized SUN family member

C4R6P9 36 4.90 N/A Glycoside Hydrolase Family

Page 175: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

175

42 Mitochondrial alcohol dehydrogenase isozyme III

C4R0S8 37 5.84 N/A Oxidoreductase activity

43 Major exo-1,3-beta-glucanase of the cell wall

C4R0Q7 48 4.53 N/A Hydrolase activity

44 Lysophospholipase C4R703 70 4.11 N/A Catalytic activity

45 Lectin-like protein with similarity to Flo1p

C4QYW7 51 4.36 N/A N/A

46 Integral membrane protein localized to mitochondria (Untagged protein) and eisosomes

C4R441 32 5.95 Plasma Membrane N/A

47 Histone H4 C4R2J6 11 11.3 Nucleus DNA bindingprotein heterodimerization activity

48 Glycerol kinase C4R8X4 68 5.25 Mitochondrion glycerol kinase activity

49 Glyceraldehyde-3-phosphate dehydrogenase

C4R0P1 36 6.24 N/A Catalytic activity

50 Fusion protein C4R0U2 15 9.87 ribosome structural constituent of ribosome

Page 176: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

176

51 Fructose 1,6-bisphosphate aldolase C4QW09 40 6.02 N/A fructose-bisphosphate aldolase activity

52 Formate dehydrogenase C4R606 40 6.61 cytoplasm Catalytic activity

53 Endo-beta-1,3-glucanase C4QYF3 34 4.07 N/A hydrolase activity

54 Dihydroxyacetone kinase C4R5Q6 65 5.36 N/A ATP binding glycerone kinase activity

55 Daughter cell-specific secreted protein with similarity to glucanases

C4QW71 110 4.65 N/A glucan endo-1,3-beta-glucanase activity

56 Cytochrome c, isoform 1 C4R6L9 12 9.61 Mitochondrion electron carrier activity

57 Chitin deacetylase C4QW42 35 5.14 N/A hydrolase activity

58 Cell wall protein with similarity to glucanases

C4QZH9 49 4.44 N/A Glycoside Hydrolase Family

59 Cell wall protein with similarity to glucanases

C4QVL7 36 4.96 N/A hydrolase activity

60 Cell wall protein that contains a putative GPI-attachment site

C4R2Q4 43 3.87 N/A N/A

Page 177: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

177

61 Catalase C4R2S1 58 6.56 N/A Catalytic activity

62 Aspartic protease C4R8B8 63 4.45 N/A aspartic-type endopeptidase activity

63 Aspartic protease C4R458 52 4.19 N/A aspartic-type endopeptidase activity

64 Amine oxidase C4R098 90 4.40 N/A copper ion bindingprimary amine oxidase activity

65 Alanine:glyoxylate aminotransferase (AGT)

C4R7U0 45 7.77 N/A transaminase activity

66 Acid trehalase required for utilization of extracellular trehalose

C4R7L0 114 4.79 N/A catalytic activity

67 ATPase involved in protein import into

the ER C4QZS3 74 4.78 N/A ATP binding

68 ATPase involved in protein folding and the response to stress

C4R3X8 71 5.09 N/A ATP binding

69 ATP synthase subunit beta C4R2N5 54 5.15 proton-transporting ATP synthase complex

Catalytic activity,

70 1,3-beta-glucanosyltransferase C4QVL5 58 4.00 Plasma Membrane transferase activity

Page 178: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

178

71 1,3-beta-glucanosyltransferase C4QVL4 57 4.02 Plasma Membrane transferase activity

72 1,3-beta-glucanosyltransferase C4R9F4 54 4.12 Plasma Membrane transferase activity

Page 179: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

179

Python Code for mining protein properties from Uniport.com

from bs4 import BeautifulSoup

import requests

from xlrd import open_workbook,cellname

import xlwt

import xlutils

from xlutils.copy import copy

from lxml import html

n=116

existing_file="D:/Users/lanselibai/Downloads/WebCrawler/Baolong.xls"

new_file="D:/Users/lanselibai/Downloads/WebCrawler/Baolong_new.xls"

code_all=[None] * n #create empty list to store code, pI, location and function

pI_all=[None] * n

location_all=[None] * n

function_all=[None] * n

book = open_workbook('D:/Users/lanselibai/Downloads/WebCrawler/Baolong.xlsx')

sheet = book.sheet_by_index(0)

#store the code list

for row_index in range(n):

code_all[row_index]=sheet.cell(row_index+1,3).value

Page 180: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

180

#store the pI list

for row_index in range(n):

print("pI "+str(row_index))

url='https://web.expasy.org/cgi-bin/compute_pi/pi_tool1?'+code_all[row_index]+'@noft@average'

wb_data=requests.get(url)

soup=BeautifulSoup(wb_data.text,'lxml')

pI=soup.select(Potgieter et al.sib_body > p:nth-of-type(2)')

string_pI = str(pI)

pI_all[row_index]=string_pI[27:31]

#print(pI_all)

#store the location list

for row_index in range(n):

print("location " + str(row_index))

url='http://www.uniprot.org/uniprot/'+code_all[row_index]+'#subcellular_location'

root = html.parse(url)

location_1=root.xpath('//*[@id="table-go_annotation"]/div/ul/li[1]/h6/text()')

location_2 = root.xpath('//*[@id="table-go_annotation"]/div/ul/li[2]/h6/text()')

location_3 = root.xpath('//*[@id="table-go_annotation"]/div/ul/li[3]/h6/text()')

location_4 = root.xpath('//*[@id="table-go_annotation"]/div/ul/li[4]/h6/text()')

location_5 = root.xpath('//*[@id="table-go_annotation"]/div/ul/li[5]/h6/text()')

location_together=location_1+location_2+location_3+location_4+location_5

#print(location_together)

Page 181: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

181

location_all[row_index] = location_together

#print(location_all)

#store the function list

for row_index in range(n):

print("function " + str(row_index))

url = 'http://www.uniprot.org/uniprot/' + code_all[row_index] + '#function'

root = html.parse(url)

f1 = root.xpath('//*[@id="function"]/h4[1]/span/text()')

f2 = root.xpath('//*[@id="function"]/ul/li/a/text()')

f3 = root.xpath('//*[@id="function"]/h4[2]/span/text()')

f4 = root.xpath('//*[@id="function"]/h4[3]/span/text()')

f5 = root.xpath('//*[@id="function"]/h4[4]/span/text()')

f6=root.xpath('//*[@id="section_x-ref_family"]/text()')

f_together = f1 + f2 +f3+f4+f5+f6

function_all[row_index] = f_together

#The following is to write the data in "pI_all", "location_all", "function_all" into existing excel by copying&write.

rb = open_workbook(existing_file,formatting_info=True)

rs = rb.sheet_by_index(0)

wb = copy(rb)

ws = wb.get_sheet(0)

for row_index in range(n):

Page 182: DEVELOPMENT OF NEW CULTIVATION STRATEGIES TO …

182

ws.write(row_index + 1, 5, pI_all[row_index])

ws.write(row_index + 1, 6, location_all[row_index])

ws.write(row_index + 1, 7, function_all[row_index])

wb.save(new_file)