Enhancing recombinant protein yield and quality using novel CHO GT cells in high density fed batch cultures 1

Enhancing Recombinant Protein Yield & QualityUsing Novel CHO GT Cells in High Density Fed-batch Cultures W ONG C HEE F URNG A thesis submitted for the degree of Doctor of Philosophy Depa

Enhancing Recombinant Protein Yield & Quality

Using Novel CHO GT Cells in High Density Fed-batch Cultures

W ONG C HEE F URNG

A thesis submitted for the degree of Doctor of Philosophy

Department of Paediatrics Faculty of Medicine National University of Singapore

2006

Enhancing Recombinant Glycoprotein Yield & Quality

Using Novel CHO GT Cells in High Density Fed-batch Cultures

SUMMARY

Chinese Hamster Ovary (CHO) cells are regarded as one of the ‘work-horses’ forcomplex biotherapeutics production Currently, batch (BC) and fed-batch (FBC) culturesare the main culture modes for a vast majority of industrial bioprocesses due to their ease

of operation and reliability During both BC and FBC, loss in viability attributed toapoptosis often results in lower recombinant protein yield and affects protein quality It ishypothesized that extension of culture life can potentially improve recombinantglycoprotein yield and quality Using an ‘in-house’ developed CHO cDNA array and amouse oligonucleotide array for time profile expression analysis of CHO BC and FBC, thegenetic circuitry that regulates and executes apoptosis induction were examined Genes

such as Fadd, Faim, Alg-2, and Requiem were identified to be key apoptosis signaling

genes during CHO cell cultures Four CHO GT (Gene Targeted) cell lines were developed,

in which each of these early apoptotic genes was either knocked down or overexpressed.These novel cell lines were shown to be effective in prolonging culture life resulting inhigher cell densities and significantly enhancing glycoprotein yield and quality

me to be more than just a ‘lab-rat’ and that there is more to science than just the lab.

Special thanks too to my friends in the Animal Cell Technology group at BTC,who welcomed me into the group with open arms and had shared some exciting andchallenging times in the research arena with me Kathy for her guidance and support, ChunLoong for successfully ‘knocking’ bioreactor principles into my brain and his ‘weird’insights, Yih Yean for his endless help and fish-tales, Vesna for keeping things inperspective, Niki for her friendship and inputs and of course not forgetting, Victor, YanYing, Janice, Sanny and Poh Choo

My heartfelt appreciation to BTI’s Analytics group led by Dr Goh Lin-Tang and

Dr Lee May May for their support in glycosylation analysis Many thanks to Sim LynChiin, Ong Boon Tee and Tracy

The BTI Microarray group: Dr Peter Morin Nissom, Jennifer Lo, Tan Kher Shing,Ong Peh Fern, Breanna Cham and Chuah Song Hui who had helped so much in getting theCHO and mouse chips up and going

Of course, there’s the help given by the undergraduate students whom worked with

me for their industrial attachment Many thanks to Andrew Wu, Wei Jan, Wong Ju Wei,

Nick Lim, Zeon Na, Amanda Lanza, Ang Pei Ling, Emily Lau and Dennis Goh for theirhelp and enthusiasm!

Last but not least, my family for being so understanding and supportive throughoutthe years Thanks especially to my wife, Winnie for being beside me through the good andthe bad times Many thanks to my parents for their love and nurture and allowing me topursue my own passion

Thanks everyone! This project would not be what it is without all of you All theresearch work described in this thesis was carried out in the Bioprocessing TechnologyInstitute (BTI), funded by the Biomedical Research Council (BMRC) established underthe Agency for Science, Technology and Research (A*STAR)

CHAPTER 2 Literature Review

2.1 Biotherapeutics Production In Mammalian Cell Culture

2.1.1 Batch Cultures

2.1.2 Fed-Batch Cultures

(A) Feed Media Design

(B) Feeding Strategy

2.1.3 Accumulation of Toxic Waste Metabolites

2.1.4 Reduction of Metabolite Waste Production

2.2 The Importance of Protein Glycosylation

2.2.1 N-Glycosylation

2.2.2 Heterogeneity in N-Glycosylation

2351518191921212323252526262728283030

2.2.3 Factors Affecting Glycosylation

(A) Host Expression System

(B) Culture Environment

(C) Extracellular Degradation

2.3 Cell Death In Bioreactors

2.3.1 Apoptosis vs Necrosis

2.3.2 Triggers of Apoptosis in Bioprocesses

2.3.3 Caspases, the central executioners of apoptosis

2.3.4 Apoptosis Signaling

2.3.5 Suppressing Apoptosis in Culture

2.4 Transcriptome Analysis

2.4.1 Microarray Technology

2.4.2 Transcription Expression Profiling

CHAPTER 3 Materials & Methods

3.2 Cell Culture Maintenance

3.2.1 Working Cell Bank

3.2.2 Culture Maintenance

3132333434353637384041414244444444444444454545

3.3 Bioreactor Culture Operations

3.3.1 Batch Culture Operations

3.3.2 Fed-batch Culture Operations

3.4.3 Amino Acid Analysis

3.5 Recombinant Glycoprotein Yields Determination

3.5.1 Determining IFN-γ Yield by ELISA

3.5.2 Average Specific Rates Calculations

3.6 N-Glycosylation Quality Of IFN-γ

3.6.1 Immunoaffinity Purification of IFN-γ

3.6.2 IFN-γ Macro-heterogeneity: Site-occupancy

3.6.3_ IFN-γ Micro-heterogeneity: Structural Composition of

Oligosaccharides

(A) Determination of Oligosaccharide Species using Mass

Spectrometry

(i) Tryptic digestion and glycopeptides separation

(ii) Reverse phase HPLC separation of IFN-γ glycopeptides

4545464848

4949494950505050505151

52

5252

(iii) Glycopeptides analysis using MALDI/TOF mass

spectometry

(B) Quantification of Oligosaccharide Species Using High pH

anion-exchange chromatography (HPAEC)

(i) Enzymatic release of glycans from IFN-γ

(ii) Preparation of Glycan Standards

(iii) Purification of Released Glycans

(iv) HPAEC Operation

3.6.4 Sialylation Assay

3.7 Cell Viability and Apoptosis Detection

3.7.1 Trypan Blue Exclusion Viability Assay

3.7.2 Morphological Detection of Apoptosis

3.7.3 Biochemical Detection of Apoptosis

3.8 Transcriptome Analysis

3.8.1 Total RNA Purification

3.8.2 Microarray Construction

(A) Slide coating

(B) Printing of DNA probes

(C) DNA Immobilization & Preparation of Microarray Slides

3.8.4 Microarray Image and Data Analysis

3.8.5 Real-Time PCR Validation

3.9 Cloning of Apoptotic Genes

3.9.1 Bacterial Culture

(A) Bacterial Cells

(B) Culture Broth and Agar Plates

(A) Gene specific Polymerase Chain Reaction (PCR)

(B) Rapid Amplification of cDNA ends (RACE)

(C) DNA Restriction & Ligation

(D) DNA Sequencing

(i) Cycle Sequencing PCR

(ii) Ethanol Purification of Cycle Sequencing PCR Product

3.10 Vector Construction For the Creation of CHO GT cells

3.10.1 pcDNA3.1(+) cg FADD Dominant Negative

3.10.2 pcDNA3.1(+) cg FAIM

3.10.3 pSUPER.neo cg ALG-2

3.10.4 pSUPER.neo cg REQUIEM

3.11 Creation of Stable Cell Lines

3.11.1 Selecting for Stable Expression Transfected Pool

3.11.2 Selecting for Single Cell Cloning

3.12 Quantitative Real Time PCR

3.12 Determination of Statistical Significance

CHAPTER 4 High Density Fed-batch Cultures of CHO Cells

INTRODUCTION

RESULTS

4.1 CHO Cell Growth & Metabolism in FBC

4.1.1 Glutamine-controlled FBC (FBC0.1, FBC0.3 and FBC0.5)

(A) Tight control of glutamine concentrations

(A) Higher viable cell density and specific growth rates

(A) Reduced ammonia production

(A) Reduced lactate production

4.1.2 Glucose-controlled FBC (FBC0.3/0.35 and FBC0.3/0.70)

(A) Tight control of glucose concentrations

(A) Viable cell density and specific growth rates

(A) Reduction of lactate production

7373747475757576767779798081818182848587888989

4.2 Enhanced Recombinant IFN-γ yields during FBC

4.3 N-glycosylation Quality of Recombinant IFN-γ

(A) Glutamine-controlled FBC maintained the distribution of

major glycan species but not the minor glycan species

(A) Glucose-controlled FBC maintained the distribution of the

major glycan species but not the minor glycan species4.3.1 Sialylation of recombinant IFN-γ

4.3.1 Impact of culture viability on N-glycosylation Quality

( A ) Low culture viability does not alter IFN-γ

macro-heterogeneity

( A ) Low culture viability led to increased number of low

molecular weight N-glycans

(A) Low culture viability led to decreased IFN-γ sialic acid

contentDISCUSSIONS

4.4.1 Improving FBC through use of dynamic online feeding

4.4.1 N-Glycosylation in FBC

CONCLUSION

CHAPTER 5 Transcriptional Profiling of Apoptotic Pathways in

Batch and Fed-batch CHO Cell Cultures

INTRODUCTION

92949495

98

99

101

101102103

103

106

107107108

109111

111

5.1 Apoptosis Signaling During BC and FBC

(A) Up-regulation of FasL and Fadd in the Death

Receptor-mediated Apoptosis Signaling during BC and FBC

(A) Up-regulation of Bim and Bad in the Mitochondrial-mediated

Apoptosis Signaling during BC and FBC

(A) Down-regulation of Ire-1 and up-regulation of Alg-2 in the

Endoplasmic Reticulum (ER)-mediated Apoptosis Signalingduring BC and FBC

(D) Differential expression of inhibitors of apoptosis proteins

DISCUSSIONS

5.2.1 Apoptosis-related Cell Death in CHO cell BC and FBC

5.2.2 Strategies to delay onset of apoptosis in culture

CONCLUSIONS

CHAPTER 6 Targeting Early Apoptotic Genes in Batch and

Batch CHO Cell Cultures

INTRODUCTION

RESULTS

6.1 Cloning of Apoptotic Homologs from CHO cells

6.2 Creation of Gene Targeted CHO (CHO GT) cell lines

6.3 CHO GT Cells in BC

6.3.1 Growth of CHO GT Transfected Pools in BC

6.3.2 Proteolytic Activities of Caspases in BC

111111115

117

118

119120120122124126

126127127129131131135

6.4 CHO GT Cell Lines in FBC

6.5 N-glycosylation Quality of IFN-γ Produced by CHO GT

cells during FBC

6.5.1 Macro-heterogeneity of recombinant IFN-γ

6.5.1 Micro-heterogeneity of recombinant IFN-γ

6.5.1 Sialylation of recombinant IFN-γ

7.2.1 Impact of improved N-glycosylation IFN-γ quality on its

bioactivity and pharmocokinetic properties

7.2.2 Characterization of recombinant IFN-γ cross-reactivity with

CHO cells

7.2.3 Analysis of differentially expressed genes

7.2.4 Characterizing the role of Fadd, Faim, Requiem and Alg-2

7.2.5 Combinatorial gene targeting to further enhance apoptosis

140141143144144

145146148148150150

151151

152153

153

155

Appendix 1 Ammonia and Lactate effects on CHO cell growth and

viability

Appendix 2 Fas Associated Death Domain (Fadd)

Appendix 3 Fas Apoptisis Inhibitory Molecule (Faim)

Appendix 4 Apoptosis Linked Gene 2 (Alg-2)

Appendix 5 Requiem

Appendix 6 Publications

Wong CFD, Wong TKK, Goh LT, Heng CK and Yap MGS 2005 Impact

of Dynamic Online Fed-Batch Strategies on Metabolism, Productivity

and N-Glycosylation Quality in CHO Cell Cultures Biotechnol Bioeng

89: 164-177

Wong CFD, Wong TKK, Lee YY, Nissom PM, Heng CK and Yap MGS

2006 Transcriptional Profiling of Apoptotic Pathways in Batch and

Fed-batch CHO Cell Cultures Biotechnol Bioeng Accepted 27th December

2005

Wong CFD, Wong TKK, Heng CK and Yap MGS 2006 Targeting Early

Apoptotic Genes in Batch and Fed-Batch CHO cell Cultures Biotechnol

Bioeng Accepted 27th December 2005

175175

176177178180182

LIST OF FIGURES:

Page

Figure 2.2 Morphological differences between apoptosis and necrosis 36Figure 2.3 Example of scanned image of a section of a hybridized microarray

Figure 3.1 Schematic representation and picture of a dynamic online fed-batch

culture bioreactor system set-up

4246

Figure 3.2 Example chromatogram of oligosaccharides from IFN-γ resolved by

Figure 3.4 Semi-automatic DNA dispensing cell used to print DNA probes

onto glass slides

62

Figure 3.5 Restriction map of pCR®2.1 (Adapted from Invitrogen product

brochure)

67

Figure 3.6 Restriction map of pcDNA3.1 (+) and (-) (Adapted from

Invitrogen product brochure)

Figure 4.2 Viable cell densities of FBC controlled at 0.1, 0.3 and 0.5mM

glutamine and control BC

82

Figure 4.3 Ammonia accumulations during FBC controlled at 0.1mM, 0.3mM

and 0.5 mM glutamine compared to control BC

84

Figure 4.4 Glucose concentrations during FBC controlled at 0.1mM, 0.3mM

and 0.5 mM glutamine compared to control BC

85

Figure 4.5 Lactate accumulations during fed-batch cultures controlled at

0.1mM, 0.3mM and 0.5 mM glutamine compared to control BC

86

Figure 4.6 Residual glucose concentrations in FBC maintained at 0.70mM and

Figure 4.12 Sialic acid content of IFN-γ harvested during high viability

(>95%) in BC and FBC

Figure 4.13 Glycan site-occupancy during BC and FBC at low culture

viability (70-80%)

Figure 4.14 Comparison between sialic acid content of IFN-γ harvested during

low viability, 70-80% and high viability, >95% in BC and FBC

101

102

106

Figure 5.1 BC and FBC for expression profiling using microarrays 112

Figure 5.2 Apoptosis-related genes regulated during BC and FBC of CHO

cells

113

Figure 5.3 Validation of microarray expression profiles of apoptosis signaling

genes across exponential, stationary and death phases of BC and FBC

114

Figure 5.4 Apoptosis signaling in BC and FBC of CHO cells 116

Figure 6.1 Apoptosis signaling via death receptor-, mitochondria- and

ER-mediated apoptosis signaling pathways during CHO cell culture

Figure 6.2 Schematic representation of gene cloning approach using gene

specific PCR and RACE methods

Figure 6.3 FADD Dominant Negative Strategy For Apoptosis Suppression

126

128

130

Figure 6.4 CHO GTO cells in BC The viable cell density, viability and

percentage of apoptotic cells of CHO GTO FADD DN pool and CHO GTO

FAIM pool in BC

132

Figure 6.5 CHO GTKD cells in BC The viable cell density, viability and

percentage of apoptotic cells of CHO GTKD ALG-2 pool and CHO GTKD

REQUIEM pool in BC

133

Figure 6.6 Recombinant human IFN-γ yields of CHO GT cells in BC 134

Figure 6.7 Specific productivities of recombinant human IFN-γ in CHO GT

cells in BC

135

Figure 6.8 Caspase-8, -9 and –3 activities during BC of CHO GTO cell lines

Figure 6.9 Caspase-8, -9 and –3 activities during BC of CHO GTKD cell lines

135137Figure 6.10 Viable cell densities of CHO GT cell lines in FBC 139

Figure 6.11 Enhanced recombinant human IFN-γ yields in CHO GT cell lines

during FBC

140

Figure 6.12 Site-occupancy of IFN-γ purified from FBC of CHO GT cells 141

Figure 6.13 Micro-heterogeneity of complex-type glycans of recombinant

human IFN-γ harvested from CHO GT cells

142

Figure 6.14 Sialylation of recombinant IFN-γ in CHO GT cell lines during

mid-exponential, stationary and death phase of FBC

143

Table 2.2 Effects of the anti-apoptosis genetic engineering on CHO cell lines 41

Table 3.1 List of N-linked oligosaccharide standards and their volumes used

in the preparation of standard sets

55

Table 3.2 Gene specific primers used in quantitative real time PCR 77

Table 4.1 Setpoint concentrations of glucose and glutamine used for Batch

and Fed-batch cultures

Table 4.2 Specific growth rates and consumption/production rates of

metabolites for FBC compared to BC

Table 4.3 Yields and Specific Productivity of Recombinant IFN-γ during BC

and FBC

78

80

93Table 4.4 Sugar compositions and glycan structures of Asn25 96Table 4.5 Sugar compositions and glycan structures of Asn97 97

Table 4.6 Micro-heterogeneity of IFN-γ glycans on Asn25 harvested during

Table 4.8 Micro-heterogeneity differences between IFN-γ glycans on Asn25

harvested during high viability and low viability (70-80%) of BC and FBC0.3

104

Table 4.9 Micro-heterogeneity differences between IFN-γ glycans on Asn97

harvested during high viability and low viability (70-80%) of BC and FBC0.3

Table 6.1 Summary of similarity between C griseus genes (Fadd, Faim,

Alg-2 and Requiem) with equivalent homologs from M musculus and H sapien.

105

128

Table 6.2 Fold increase/decrease in gene expression in CHO GT cells and

control cell lines

129

1 INTRODUCTION

1.1 Background

One of the challenges faced by large-scale production of therapeutic proteins is theneed to achieve high cell density while maintaining high culture viability in order to obtainhigh recombinant glycoprotein yield and quality In most of these proceses, batch (BC)and fed-batch (FBC) cultures are the main culture modes used for recombinant proteinproduction The major cause of viability loss in BC is nutrient depletion By addressingnutrient depletion through nutrient feeding, FBC offer a solution towards higher celldensity and extended culture viability However, overfeeding can lead to increasedaccumulation of toxic metabolites such as ammonia and lactate that are detrimental to cellgrowth and viability This can be alleviated through the control of glucose and glutamine atlow levels in the culture medium allowing for a metabolic shift towards lower lactateproduction and an energetically more efficient glutamine metabolism without any loss in

productivity (Cruz et al., 1999; Europa et al., 2000, Lee et al., 2003) Thus, achieving high

cell density FBC through detailed analysis of nutrient consumption and feeding strategy ishighly desirable in terms of product yield improvement

Currently, the influence of metabolic shifts on product glycosylation remainsrelatively unknown The structural heterogeneity of glycans on glycoproteins is sensitive

to culture environment such as nutrient starvation, metabolic waste accumulation, cultureviability, pH and temperature (Goochee and Monica, 1990; Yang and Butler, 2000;

Andersen et al., 2000; Baker et al., 2001) As glycans on glycoproteins are often critical for

a myriad of functions, some of which are crucial for its pharmacokinetic properties (Varki,

1993; Jenkins et al., 1996; Jefferis, 2005), it is important to examine the impact of the

production process on glycosylation patterns to ensure product efficacy and consistency

Despite nutrient feeding, FBC are still susceptible to loss in culture viability albeit

at a later time point compared to BC The implication of this is that either a criticalnutrient is still missing or an insult is triggering cell death In addition, viability loss notonly lowers productivity but may affect recombinant protein quality as well For example,degradative enzymes released during cell death can detrimentally affect the sialylation of

the recombinant protein resulting in reduced circulatory half-life of biotherapeutics in vivo (Varki, 1993; Gramer et al., 1995).

It has been shown the major mode of cell death in culture is apoptosis, a genetically

controlled form of cellular suicide (Singh et al., 1994; Goswami et al., 1999) The most

common anti-apoptotic manipulation is BCL-2 protein overexpression (Fusseneger and

Bailey, 1998’ Laken and Leonard, 2001; Vives et al., 2003a; Arden and Betenbaugh, 2004).

However, in most cases, this resulted in limited protection from apoptosis induction(Laken and Leonard, 2001) Furthermore, in spite of being the predominant mode of celldeath, apoptosis signaling during bioreactor cultures has not been examined extensively.Investigation of apoptosis signaling is therefore crucial for a better understanding of celldeath in bioprocesses

First described by Schena and co-workers (1995), DNA microarray technology is

based on the simultaneous hybridization of two different DNA populations (each labeled

either with red or green fluorescence) onto microarrays containing thousands of distinctgene sequences The ratio of fluorescence intensity then represents the ratio of expressionbetween the two different populations Transcriptional profiling using DNA microarray

offers a tool to simultaneously characterize these multiple gene expression changes in aglobal manner Consequently, genes that are involved in apoptosis signaling in culture can

be identified and specifically targeted to suppress apoptosis in BC and FBC of CHO cells

1.2 THESIS OBJECTIVES

The main goal of the thesis is to enhance the recombinant glycoprotein yield andquality in FBC of CHO cells through gene targeting It is hypothesized that transcriptomeanalysis could be used to decipher cell death signaling in BC and FBC, enabling genesassociated with the early onset of apoptosis to be identified and targeted to prolongculture

The scope of the thesis involved:

(1) Developing an enhanced high-density FBC based on a dynamic online feedingstrategy for CHO cells producing recombinant human interferon gamma (IFN-γ) anddetermining the impact on IFN-γ production and glycosylation quality,

(2) Examining the signaling pathways that are responsible for apoptosis induction

in BC and FBC processes using transcriptome analysis and

(3) Based on the knowledge gained, develop novel apoptosis gene-targeted cell linesthrough the targeting of key early apoptosis genes to extend culture viability

1.3 THESIS ORGANIZATION

There are eight chapters in this thesis Chapter 1 provides a brief introduction anddefines the objective and scope of the thesis Chapter 2 reviews the literature on cellculture processes, protein glycosylation and apoptotic cell death, as well as transcriptomeanalysis A detailed description of the materials and methods used is covered in Chapter 3

and in Chapter 4, the development of an enhanced high-density fed-batch process and itsimpact on cellular metabolism and protein N-glycosylation are described In Chapter 5, theresults of the transcriptome profiling of apoptosis signaling pathways for BC and FBC arereported Chapter 6 describes the cloning of four early apoptosis signaling genes identifiedusing transcriptional profiling, and the subsequent targeting of the four genes through gene

‘knock-down’ or overexpression to generate novel apoptosis resistant CHO GT cells Thischapter also detailed CHO GT’s ability to significantly improve recombinant glycoproteinyield and quality Chapter 7 summarizes the important conclusions resulting from thisstudy and provides recommendations for future work

2 LITERATURE REVIEW

2.1 Biotherapeutics Production In Mammalian Cell Culture

The early 1980s saw the biopharmaceutical industry starting to use recombinantDNA technology for the production of large quantities of therapeutic proteins During thisperiod, human growth hormone (hGH) was extracted from human pituitaries for thetreatment of dwarfism Discovery of possible Creutzfeldt-Jacob disease transmission usinghuman pituitaries purified hGH in 1985 quickly shifted treatment to the use ofrecombinant hGH which offers a much safer treatment alternative The biopharmaceuticalindustry’s global market value now stands in excess of US$30 billion and has more than

500 biopharmaceuticals undergoing clinical trials (Walsh, 2003) These newbiopharmaceuticals offers treatment for diseases ranging from cancer, autoimmunity,cardiovascular and infectious diseases

Today, 60-70% of all the recombinant protein pharmaceuticals are produced inmammalian cells (Chu and Robinson, 2001; Wurm, 2004) Mammalian cells have been usedextensively for complex biologics production due to their ability to produce properlyfolded and glycosylated version of the proteins Although yeast, insect and plant cells arecapable of glycosylating proteins, only mammalian cells are capable of producingglycoforms similar to those required for human therapeutics Among the mammalian cells,Chinese hamster ovary (CHO) cells are one of the most commonly used cell lines for theproduction of therapeutically important proteins (Table 2.1) CHO cells have beensuccessfully used for the production of important biologics such as erythropoietin (Sasaki

et al 1987), Factor VIII (Kaufman et al 1988) and follicle stimulating hormone (Gerbert &

Gray 1994) They are well studied and characterized, have high transfection efficiency andgood amplification mechanisms like DHFR system (Simonsen and McGrogan, 1994).

Table 2.1 Examples of Recombinant Therapeutics Proteins Produced in CHO cells(Adapted from Walsh, 2003)

Protein

Therapeutic Usage Commercial Drug

NameFactor VIII Hemophilia A ReFacto, Bioclate

Blood factors

Tissue plasminogenactivator

Myocardial infarction Tenecteplase,

TNKase, ActivaseFollicle-stimulating

hormone

Infertility Follistim, Puregon,

Gonal FChoriogonadotrophin Assisted reproductive

techniques

OvitrelleHormones

Thyrotrophin-α Detection/ trement of

multiple sclerosis

Rebif, Avonex

rIgG1kMab thatbinds to IgE

Zevalin

α-galactosidase Fabry disease FabrazymeTNF receptor-IgG

fragment fusionprotein

Rheumatoid arthritis EnbrelOthers

To be economically feasible, recombinant protein production processes often need

to achieve high cell yield, steady productivity and consistent glycosylation Industrialprocesses for large-scale production using mammalian cell culture are suspension serum-free batch and fed-batch cultures in stirred-tank reactors (Chu and Robinson, 2001;Andersen and Krummen, 2002; Wurm, 2004) The popularity of these culture modes stemfrom their relative ease and simplicity of operation Perfusion cultures have recently gained

popularity as they can achieve higher cell densities than BC and FBC and can bemaintained for many weeks, even months However, perfusion cultures are verysophisticated processes, which require a high degree of control and the need to maintaincontamination-free conditions for longer periods of time.

2.1.1 Batch Cultures (BC)

The usual practice in BC is to supply all the nutrients needed by the cells for theduration of the culture at the beginning of a culture Unlike bacterial cells, mammalian cells

have complex nutritional requirements for cell growth in vitro Different cell lines also have

different nutritional requirement, necessitating the development of different basal mediathat contains different amounts of essential nutrients Detailed spent medium analysisallowed for the development of highly fortified basal media enriched in multiple nutrientcomponents that were limiting in the basal media This allows for further improvement incell density and protein yield However, many medium components can inhibit cell growthwhen added at concentrations higher than those commonly found in basal medium (Hassell

et al., 1991, Bibila and Robinson, 1995, Lao and Toth, 1997).

2.1.2 Fed-batch Cultures (FBC)

FBC allows for periodic nutrient feeding to prevent nutrient depletion Periodicfeeding also prevents exposure to inhibitory concentrations of excessive nutrients Due tothe complexity of mammalian cell metabolism, optimization of timing and mode of addition

of nutrient feeds is often performed empirically (Bibila and Robinson, 1994) Most FBCstrategies rely upon a combination of physiological reasoning, nutrient depletion analysisand reiterative feed design to maximize cell growth, culture longevity and recombinantprotein production

(A) Feed Media Design

FBC efforts have evolved from feeding with single limiting components such asglucose or glutamine to complex feeding with multiple limiting components Concentratedsupplemental medium and feeding are often designed based on nutritional demands for cellgrowth by experimentally measuring nutrient consumption rates of key nutrients such asglucose and amino acids from BC (Bibila and Robinson, 1994) Although it is obvious thatnutrient consumption rate is affected by culture conditions and therefore consumptionrates are different in BC compared to FBC, this approach provides an initial step towardsreiterative designs in feed composition

Concentrated complete media without salts for feed media can also be used forfeeding This eliminates the need for time consuming experimental determination of limitingnutrients However, concentrated feed media has a higher risk of increasing theaccumulation of inhibitory media components and osmolarity of a culture This is becausenutrient ratios in the basal medium may not necessarily balance with cellular demands andleads to accumulation of certain nutrients and by-products Reiterative analysis of spentmedium is therefore needed to readjust the nutrient ratio if the process were to beimproved Alternatively, balanced feed media can be designed using mathematical modellingbased on estimated cellular composition, product composition and energy demand (Xie andWang, 1994)

Development of a feeding strategy is the next important step towards an effectiveFBC process Feeding strategies can be generally classified into two categories: feedforward and feed back control In feed forward systems, the culture is fed based on optimal

feeding trajectories Measured cell densities and estimated growth rates are used to

determine culture feed rates (Bibila and Robinson, 1994; Lee et al., 1999) The feeding is

therefore dependent on the development of kinetic models that describe cell growth

(Glacken et al., 1989).

The use of feed back control eliminates any requirement for a process model.Instead, cultures are fed based on online measurements of culture performance OnlineHPLC system can be used to monitor nutrient concentrations and automatic feeding can beinitiated to prevent nutrient depletion The important roles played by glucose in centralmetabolism makes it a popular nutrient for online measurement Other indications of cellgrowth such as dynamic oxygen uptake rates that can be measured online through the use

of a dissolved oxygen probe has also been used successfully for automatic feeding (Zhouand Hu, 1994)

2.1.3 Accumulation of Toxic Waste Metabolites during Culture

Besides nutrient limitation, accumulation of waste metabolites in the cell culturemedium can inhibit recombinant protein production and cell growth Excessive lactateaccumulation can result in increased medium osmolarity and decreased culture pH in the

absence of pH control (Omasa et al., 1992) Lactate is mainly generated via glucose

metabolism Even under aerobic conditions, the molar ratio of lactate to glucose generallyranges from 1 to 2 when glucose is not maintained at low concentrations When p Hcontrol is present, lactate itself has been shown to have little negative influence but it is the

osmolarity increase that is responsible for the inhibitory effects (Lao et al., 1997).

The accumulation of ammonia during mammalian cell culture is another main cause

of growth inhibition and decline in productivity (Ryll et al., 1994; Hansen and Emborg,

1994; Yang and Butler, 2000) Ammonia accumulation can result either directly from enzymatic breakdown of glutamine in the medium or via enzymatic glutamine cellularmetabolism In the first case, glutamine is normally degraded to generate ammonium at

non-physiological temperature and pH (Ozturk et al., 1990) In the second case, ammonia is a

secreted waste metabolite of glutamine metabolism Ammonium groups are released viaglutaminolysis due to the enzymatic actions of glutaminase and glutamate dehydrogenase.Although different cell lines can tolerate different concentrations of ammonia, no growthinhibition was observed in CHO cells exposed to 10mM ammonia (Hansen and Emborg,1994)

2.1.4 Reduction of Metabolite Waste Production

Various approaches have been used to reduce the amount of metabolite waste inculture The most effective method involves manipulation of culture environment Culturescan be kept at low glucose or glutamine concentrations to reduce metabolite waste

production of lactate and ammonia (Glacken et al., 1986; Ljunggren and Haggstrom, 1994; Kurokawa et al., 1994) A low glucose concentration reduces the glycolytic flux that

generates pyruvate, a substrate for lactate production while low glutamine reducesglutaminolysis and overflow of metabolism of other amino acids Substitution of glucosewith other carbon source such as galactose, fructose or mannose and glutamine withalternative amino acids such as glutamic acid also can reduce lactate and ammoniaaccumulation (Bibila and Robinson, 1995)

2.2 The Importance of Protein Glycosylation

The majority of proteins secreted by mammalian cells are glycosylated.Glycosylation, which involve the addition of sugar residues to a peptide backbone, is one

of the most frequent post-translational modifications It is not surprising then that theseoligosaccharides moiety play critical roles in determining biotherapeutics activity andquality (Varki, 1993) Glycosylation can influence both the structural and functionalproperties of the protein, such as protein folding and conformation, stability todenaturation, solubility and resistance to proteolysis as well key biological properties such

as receptor binding, modulation of enzyme activity, and cellular recognition events (Varki,

1993; Helenius, 1994; Fielder and Simons, 1995; Berg et al., 1995; Saraneva et al., 1995).

Oligosaccharides can be attached to proteins via N-glycosidic linkages, via glycosidic linkages or are glycosylphosphatidylinositol anchored N-glycosylation occursspecifically at Asparagine-X-Serine/Threonine tripeptide sequence while O-glycosylation

O-is less specific but must occur at a serine or threonine residue.Glycosylphosphatidylinositol sites meanwhile are indicated by a hydrophobic carboxyl-terminal peptide, which is concurrently cleaved when the glycosylphosphatidylinositol isattached to the polypeptide N-glycosylation represents the majority of glycosylationoccurring in mammalian cell-derived glycoproteins

The majority of recombinant proteins manufactured for human therapy areglycoproteins derived from animal cells, and it is essential to fully characterize and ifpossible to control the glycosylation profiles of these products Indeed, the Food and DrugAdministration (FDA) in the United States and the Committee for Proprietary MedicalProductions (CPMP) of the European Community are demanding increasinglysophisticated carbohydrate analysis on all new glycoproteins destined for human therapy(Liu 1992) Currently, most recombinant proteins intended for human therapy areaccepted with some degree of glycosylation heterogeneity However, the FDA is placing

increasing emphasis on the importance of defining what critical recombinant proteinheterogeneity is important to the efficacy of a product and to design an appropriateanalytical test to ensure that the heterogeneity is produced consistently between lots(Jenkins & Curling 1994).

ER lumen This core structure is trimmed to varying degrees by ER glucosidases, and then

by mannosidases in the ER and cis Golgi In some case the high-mannose core is left

relatively intact with no further additions of other sugar residues Alternatively, complexouter structures can be built on a trimmed glycan core by a series of GlcNAc, galactose,sialic acid or fucose additions using Golgi-resident glycosyltransferases and nucleotide-sugar intermediates Fucose may be added to the core structure via a α-1,6-linkage andtends to be glycosylation site specific

2.2.2 Heterogeneity in N-Glycosylation

Due to the many different steps and modifications involved in proteinglycosylation, glycoproteins often exhibit differences in the macro- and micro-

heterogeneity in their oligosaccharide structures Macro-heterogeneity refers to variability

in the location and number of oligosaccharide attachment sites while micro-heterogeneityrefers to the variability in the oligosaccharide structure at specific glycosylation site

Different N-glycan forms can be classified into: High-mannose type: only mannose

is present in the outer arms; Complex type: containing the disaccharide Galβ(1,4)GlcNAc and Hybrid type: mannose residues in one arm and complex structures on the other arm

(Figure 2.1) Further classification of complex-type glycans is based upon the number ofarms (antennae) emanating from the core structure and the types of sugar residue present

(Rademacher et al., 1988).

High

antennary antennaryTri- antennaryTetra-

Bi-Complex

Figure 2.1 N-glycan structures High-mannose, Hybrid and Complex type structures

of N-linked oligosaccharides (Mannose , Galactose , N-acetylglucosamine ,Neuramic Acid and Fucose )

2.2.3 Factors Affecting Glycosylation

Heterogeneity in both the glycosylation site occupancy (macro-heterogeneity) andsugar structures (micro-heterogeneity) occurs extensively in glycoproteins produced by

mammalian cells (Hooker et al., 1995; Jenkins et al., 1996) Given the potential effects of

glycosylation, it is important to understand the factors that influence glycosylation duringthe production of glycoproteins for biotherapeutics use.

(A) Host Expression System

The host cells used for glycoprotein expression play a critical part in influencingprotein glycosylation Common bacterial expression systems are incapable of proteinglycosylation Yeast, plant, insect and mammalian cells have the same core oligosaccharideprocessing in the endoplasmic reticulum but have different oligosaccharide processing inthe Golgi apparatus Yeast cells often contain elaborate high-mannose type structureswhile plant glycoproteins are not sialylated and frequently contain xylose, amonosaccharide not typically found in mammalian glycoproteins Insect cells are limited toproducing simply high-mannose type glycans

It is mammalian cells that produce glycans most similar in structure to humans.However, even in mammalian cells, there are disparities in the glycans derived from mouseand human cells Antibodies produced from mouse cells often contain N-glycolneuraminicacid (NeuGc) rather than N-acetylneuraminic acid (NeuAc), which is the common sialicacid in human High amount of NeuGc can elicit an antigenic immune response in humans

(Noguchi et al., 1995) CHO cells is known to lack the functional enzyme sialyltransferase, leading to exclusively α-2,3-linked terminal sialic acid residues (Lee et al

α-2-6-1989) Furthermore, CHO cell is unable to sulfate GalNAc residues, which are a common

motif on certain glycoprotein hormones and may modulate their clearance in vivo The

absence of a functional α-1,3-fucosyltransferase in CHO cells also prevents the addition of

peripheral fucose residues (Potvin et al 1990) Endomannosidase activity has also been reported to be absent in CHO cells (Moore and Spiro 1990; Hiraizumi et al 1993; Dairaku

and Spiro 1997) Fortunately, because CHO cell line can readily express a wide variety ofrecombinant proteins, its glycosylation machinery can be modified to resemble more

closely the human profile by transfection of the appropriate glycosyltransferases (Lee et al

1989 and Potvin et al 1990) For example, CHO cells have been engineered to express

α-2-6-sialyltransferase for production of human-like sialylated recombinant glycoproteins

(Minch S L et al 1995; Monaco et al 1996; Bragonzi et al 2000).

(B) Culture Environment

The structural heterogeneity of oligosaccharides (glycans) on glycoproteins issensitive to culture environment including nutrient starvation, metabolic wasteaccumulation, culture viability, pH and temperature (Goochee and Monica, 1990; Jenkins

and Curling, 1994; Yang and Butler, 2000; Andersen et al., 2000) There has been

increasing awareness among both academia and industry that the culture environment can

change the glycosylation profile of a protein (Curling et al., 1990; Goochee, 1990; Goochee

et al., 1991; Goochee, 1992; Andersen and Goochee, 1994).

Ammonium ions are natural byproduct of glutamine metabolism or breakdown incell culture It can reach concentrations of 5-10mM and significantly reduce the sialylation

of recombinant product at even 2mM (Andersen and Gochee, 1994) Carbon dioxide is

another metabolic byproduct of cellular metabolism that can accumulate in poorlyventilated cultures Elevated pCO2 can decrease sialylation of tissue plasminogen activator

in culture (Kimura and Miller, 1997) Nyberg et al (1998) found that glucose and

glutamine limitation reduced glycosylation site occupancy

(C) Extracellular Degradation

Glycosidase degradation, especially by sialidase, has also been implicated as asource of glycoform heterogeneity in cell culture Fucosidase, β-galactosidase and β-hexomindase activities have also been detected in cell culture supernatants of 293, CHOand hybridoma cell lines (Gramer and Goochee 1993; Gramer and Goochee 1994; Gramer

et al 1994) These researchers hypothesized that these enzymes may be responsible for

significant oligosaccharide degradation especially under operations with high cell densitiesand low cell viabilities

2.3 Cell Death In Bioreactors

Cell death is a critical factor limiting the productivity of cultured mammalian cells.Cells in culture are constantly exposed to changing environmental conditions anddepending on these conditions, cells will either die or live In many large-scale cultures,apoptosis has been implicated as the principle form of cell death (Laken and Leonard,2001) For many who work with cell cultures, the main goal is to keep the cells alive and toensure continuous production of the desired product Therefore, the role played byapoptosis in limiting culture performance is now being investigated

Within the past ten years, it has been demonstrated that under both physiological

and in vitro conditions, cells that experience mild environmental stress die not by necrosis,

but rather by the genetically controlled process known as apoptosis (Kerr 1972) In scale culture, apoptosis and proliferation are often regulated such that cessation ofproliferation coincides with the engagement of apoptosis Apoptosis induction following

large-growth arrest is thus a major barrier to the development of culture strategies that optimizespecific productivity by reducing growth rates.

Cell death will also contribute to an increase in the amount of cell-derivedsubstances liberated into the medium following loss of membrane integrity and disruption

of the cells This will inevitably lead to a reduction in the performance of downstream

processing equipment (Maiorella et al., 1991) Moreover, proteases released from

non-viable or lysed cells will inactivate proteins and may even have negative influence on cellproliferation

2.3.1 Apoptosis vs Necrosis

Apoptosis was originally distinguished from necrosis on the basis of its

ultrastructure (Kerr et al 1972) The striking similarity of the morphological changes of apoptosis across cell types and species make its identification applicable to a variety of in

vitro studies (Figure 2.2).

Apoptotic cells initially undergo highly characteristic condensation andfragmentation of chromatin and cell shrinkage They then undergo intensive ‘blebbing’,during which numerous protrusions form at the surface of the cells (Steps 1-2) Theseprotusions eventually break off from the plasma membrane, forming intact cytoplasmic

packages called apoptotic bodies (Step 3) In vitro, in the absence of macrophages, the

apoptotic cell subsequently enters a degenerative phase during which it undergoesstructural collapse, a phase known as secondary necrosis (Step 4) Necrosis on the otherhand, occurs in response to extreme environmental stress Nuclear morphology does notundergo the condensation observed during apoptosis and the chromatin undergoes non-

specific cleavage (Steps 5-7) Membrane damage leads to the failure of the cell to maintainionic homeostasis, resulting the swelling and eventual bursting of the cell (Step 8).

A1 A2 A3 A4

N1 N2 N3 N4

Figure 2.2 Morphological differences between apoptosis and necrosis Thesequence of ultra-structural changes in apoptosis (Steps A1-A4) and necrosis (StepsN1-N4) (Adapted from www.roche-applied-science.com/sis/apoptosis)

2.3.2 Triggers of Apoptosis in Bioprocesses

Despite the best efforts to maintain a stable and uniform environment forbiotherapeutics production through the use of bioreactors, cells are still subjected to avariety of stimuli that could induce apoptosis Nutrient limitation of key substrates forenergy metabolism such as glucose and glutamine (Mercille and Massie, 1994), depletion

of either essential or non-essential amino acids can induce apoptosis (Simpson et al.,

1998) Hydrodynamic stresses due to stirrer, gaseous sparging for bioreactor oxygenation,Necrosis

Apoptosis

osmolarity, pH, metabolic waste accumulation and nutrient depletion has all been shown

to contribute to cell death (Laken and Leonard, 2001)

2.3.3 Caspases (Central Executioners of Apoptosis)

Most of the morphological changes seen in apoptosis are caused by caspases.Caspases are a conserved family of cysteine protease enzymes that irreversibly commit a

cell to die in the apoptosis response (Degterev et al., 2003; Riedl and Shi, 2004) Caspases

are capable of mediating the cleavages of substrates such as iCAD, nuclear lamins andPAK2 which are responsible for DNA laddering, shrinking and blebbing respectively

(Nagata, 2000; Buendia et al., 1999; Rudel et al., 1997) With more than 100 other caspases

substrates identified, caspases definitely play a central role in the apoptotic process

(Earnshaw et al., 1999).

Of the 14 known mammalian caspases, 7 are known to have important roles inapoptosis These caspases are divided into initiator caspases that include caspases-2, -8, -9and -10 and executioner caspases that include caspases-3, -6 and -7 The specificity ofcaspases is typically determined by four amino acids residues, which are terminal to thecleavage site Typically, they exist as enzymatically inert zymogens composed of p10,p20 domains as well as an N-terminal prodomain Upon proteolytic cleavage of the p20and the p10 sites and usually also between the N-terminal prodomain and the p20 domain,

it creates an active caspase molecule that cleaves proteins between the Asp-xxx site.Interestingly, the proteolytic cleavage site of the inactive caspase is typically an Asp-xxxsite as well, leading to the possibility of autocatalytic activation of the caspase cascade

2.3.4 Apoptosis Signaling

The apoptotic signaling pathways that can lead to caspases activation can bemediated through either extrinsic or intrinsic pathways The extrinsic pathway involves thetransmission of apoptosis trigger from outside the cell through the binding of ligands to acell surface death receptor while intrinsic pathways are triggered in response to stimuligenerated from within the cell

Death receptors are characterized by the presence of conserved extracellular domaincontaining two to four cysteine-rich pseudo-repeats, a single transmembrane region and aconserved intracellular protein-protein interaction domains called death domain (Curtin andCotter, 2003) Upon activation, the death receptors multimerizes leading to the formation

of death inducing signaling complex (DISC) which is made up of adaptor proteins such asFas associated death domain (FADD), Tumour necrosis factor receptor I associated deathdomain (TRADD) or Receptor interacting protein (RIP) The subsequent recruitment andoligomerization of caspase-8 to the DISC then results in its autocatalytic activation (Lavrik

et al., 2003; Peter and Krammer, 2003).

Bcl-2 family members play critical roles in the mitochondria-mediated apoptosis

signaling pathway (Reviewed in Gross et al., 1999; Desagher and Martinou, 2000; van Gurp et al., 2003) Bcl-2 family proteins contained conserved domains called Bcl-2

homology (BH) domains BH3-only proteins such as BIM, BID and BAD possess onlyone of the BH domains and are responsible for the initial triggering of this pathway TheseBH3-only members can activate other pro-apoptotic Bcl-2 members such as Bax, Bak andBok that are localized on the outer mitochondrial membranes or cytosol of the cell Uponactivation, they oligomerize and insert into the mitochondria membrane, trigerring the

release of cytochrome c and other proteins Released cytochrome c then binds to APAF-1

to form an ‘apoptosome’ complex which in turn recruits and activates caspases-9 Thereare also Bcl-2 members such as Bcl-2 and Bcl-xL, which are anti-apoptotic and can block

cell death by preventing cytochrome c release.

More recently, another intrinsic pathway involving the endoplasmic reticulum (ER)

had been implicated in apoptosis regulation (Breckenridge et al., 2003; Szegezdi et al., 2003; Rao et al., 2004) The signaling pathway involved in ER-mediated apoptosis is still

poorly understood but it is clear that proteins such as IRE-1, PERK and ATF6 play keyroles These proteins respond to unfolded/misfolded protein accumulation and act as ERstress sensors that can lead to apoptosis induction ER-mediated apoptosis induction canoccur independently of the mitochondria and death receptors through caspase-12activation Upon activation of ER-mediated apoptosis, caspase-12 is translocated from the

ER to the cytosol where it cleaves procaspase-9, which in turn activates the executionercaspase, caspase-3 In addition, Bcl-2 family members such as Bcl-2, Bcl-xL, BAX, BAKand BIK are also involved as modulators in ER stress–induced apoptosis by controlling ER

Ca2+ homeostasis (Scorrano et al., 2003).

Interestingly, there are two distinct cell types, which utilize distinct apoptosis

signaling pathways (Scaffidi et al., 1998; Scaffidi et al., 1999) In Type I cells, apoptosis

can occur via CD95 death receptor signaling without the involvement of the mitochondria

In contrast, Type II cells require the mitochondria release of cytochrome c in order for

CD95 to exert its apoptotic effects At the molecular level, these two cell types differinthe amount of caspase-8 recruited to CD95 viathe adapter molecule FADD/Mort1 to formthe DISC (Reviewed in Roy and Nicholson, 2000) Type I cells contain large amounts of

DISC in response to anti-CD95 antibodies but Type II cells do notand thus are dependent

on stimulation of the intrinsic apoptotic pathway to undergo cell death These couldexplainwhy the extrinsic apoptotic pathway was insensitive to Bcl-2 overexpression inType I cells but sensitive to Bcl-2overexpression in other Type II cells

2.3.5 Suppressing Apoptosis in Culture

In general, there are three approaches that are commonly used to suppressapoptosis in cell culture processes (Leonard and Laken, 2001) These include alleviatingnutrient deprivation by feeding, addition of apoptosis-suppressing chemicals or metabolic

engineering using anti-apoptotic survival genes In BC, Simpson et al (1998) found that

simple replenishment of a single key amino acid was able to delay nutrient-deprivedhybridoma cells from initiating apoptosis Implementation of effective feeding strategiestherefore can overcome nutrient depletion that can induce apoptosis

In the second approach, various chemicals have been shown to provide protectionagainst apoptosis (Laken and Leonard, 2001; Arden and Betenbaugh, 2004) Bongkrekicacid is able to inhibit apoptosis induced by dexamethasone by blocking formation ofmitochondrial pore while addition of an antioxidant, N-acetylcysteine, has also been shown

to prevent apoptosis caused by reactive oxygen species generated during culture Caspaseinhibitors such as Z-VAD.fmk and YVAD.cmk have been used to inhibit apoptosis byacting as artificial substrates for caspase However, in large-scale reactors, continuousaddition of these chemicals may not be economically feasible and may be toxic when added

in high amounts