High cell density perfusion process development for antibody producing Chinese Hamster Ovary

cells

Ye Zhang

Doctoral Thesis in Biotechnology Stockholm, Sweden 2017

©Ye Zhang School of Biotechnology Royal Institute of Technology AlbaNova University Center SE-106 91 Stockholm Sweden Printed by Universitetsservice US-AB TRITA-BIO Report 2017:14 ISSN 1654-2312 ISBN 978-91-7729-450-4

Abstract

Perfusion operation mode is currently under fast expansion in mammalian cell based

manufacturing of biopharmaceuticals, not only for labile drug protein but also for

stable proteins such as monoclonal antibodies (mAbs). Perfusion mode can

advantageously offer a stable cell environment, long-term production with high

productivity and consistent product quality. Intensified high cell density culture

(HCDC) is certainly one of the most attractive features of a perfusion process due to

the high volumetric productivity in a small footprint that it can provide. Advancements

in single-use technology have alleviated the intrinsic complexity of perfusion processes

while the maturing in cell retention devices has improved process robustness. The

knowledge for perfusion process has been gradually built and the “continuous” concept

is getting more and more acceptance in the field.

This thesis presents the development of robust perfusion process at very high cell

densities in various culture systems. Four HCDC perfusion systems were developed

with industrial collaborators with three different mAb producing Chinese Hamster

Ovary (CHO) cell lines: 1-2) WAVE Bioreactor™ Cellbag prototype equipped with cell

separation by hollow fiber filter utilizing Alternating Tangential Flow (ATF) and

Tangential Flow Filtration (TFF) techniques; 3) Fiber matrix based CellTank™

prototype; 4) Glass stirred tank bioreactor equipped with ATF. In all the systems,

extremely high viable cell densities above 130 million viable cells per milliliter

(MVC/mL) up to 214 MVC/mL were achieved. Steady states were maintained and

studied at 20-30 MVC/mL and 100-130 MVC/mL for process development. Perfusion

rate selection based on cell specific perfusion rate (CSPR) was systematically

investigated and exometabolome study was performed to explore the metabolic

footprint of HCDC perfusion process.

Keywords: WAVE Bioreactor, Cellbag, ATF, TFF, hollow fiber, CSPR, CellTank, SUB,

single-use-bioreactor, disposable, perfusion rate

Sammanfattning Perfusionssystem för produktion av biobaserade proteinläkemedels, via cellodling, är

ett område som just nu expanderas och utvecklas i snabbt takt, inte bara med avseende

på labila proteiner men också mer stabila produkter så som monoklona antikroppar

(mAb). Den stora fördelen med perfusionssystem, över andra odlingstekniker, är dess

kompakta formfaktor, en kontinuerligt hög produktivitet under lång tid, konsekvent

hög produktkvalitet och stabila odlingsförhållanden. Perfusionsteknikens främsta fördel

är möjligheten att erhålla extremt hög celltäthet, vilket resulterar i mycket hög

volymetrisk produktivitet. Utveckling inom engångsodlingssystem och

cellretentiontekniker har skapat möjligheter för att underlätta nyttjandet av

perfusionsprocesser samt ökat robustheten. Allt eftersom kunskapen om

perfusionssystem har ökat, har kontinuerliga produktionsystem blivit mer accepterade

inom forskningsområdet.

Den här avhandlingens fokus är utveckling av en robust perfusionsprocess som kan nå

oerhört höga celltäthet, i flertalet odlingssystem. Fyra olika system för hög celltäthet

har utvecklats, med industriella samarbetspartners, för produktion av tre olika mAb

genom olika Chinese Hamster Ovary (CHO) cellinjer: 1-2) WAVE Bioreactor™ Cellbag,

vågbaserad reaktor prototyp med cellseparation genom varierande tangentialt flöde

(ATF) och tangential flödes filtrering (TFF); 3) Fiber matris baserad CellTank™

prototype; 4) Glasbioreaktor med ATF, för cellretention. I samtliga system nås extremt

höga celltäthet, över 130 miljoner viabla celler per milliliter (MVC/mL), och som högst

214 MVC/ml. ”Steady state” förhållanden upphölls under processutvecklingen på

cellkoncentrationer motsvarande 20-30 MVC/ml respektive 100-130 MVC/ml. En

systematisk studie för optimering av perfusionsflöden utfördes, baserad på cellspecifika

perfusionshastigheter (CSPR), samt en exometabolomestudie för att utreda

metaboliska avtryck under höga cellmassor i perfusionssystem.

To my parents

List of publications This thesis is based on the following publications, which are referred to in the text by

their roman numerals:

I Clincke M-F, Mölleryd C, Zhang Y, Lindskog E, Walsh K, Chotteau V. Very High

Density of CHO Cells in Perfusion by ATF or TFF in WAVE BioreactorTM. Part I. Effect of

the Cell Density on the Process. Biotechnology Progress. 2013; 29(3): 754-767

II Zhang Y, Stobbe P, Orrego-Silvander C, Chotteau V. Very high cell density perfusion

of CHO cells anchored in a non-woven matrix-based bioreactor. Journal of

Biotechnology, 2015; 213: 28-41

III Zhang Y, Zhan CJ, Girod P-A, Martiné A, Chotteau V. Optimization of the cell

specific perfusion rate in high cell density perfusion process. (Manuscript)

IV Zamani L, Zhang Y, Aberg M, Lindahl A, Mie A, Chotteau V. Metabolic footprinting

of CHO cell culture bioprocess data in fed-batch and perfusion mode using LC-MS data

and multivariate analysis. (Manuscript)

The following publication is not included in this thesis:

V Al-Khalili L, Gillner K, Zhang Y, Åstrand C, Shokri A, Hughes-Brittain N, McKean

R, Robb B, Chotteau V. Characterization of Human CD133+ Cells in Biocompatible

Poly(l-lactic acid) Electrospun Nano-Fiber Scaffolds. Journal of Biomaterials and Tissue

Engineering. 2016; 6(12): 959-966(8)

Author's contributions Paper I: Took part in cultivation and sample analysis. Performed viscosity study.

Paper II: Experimental design and operation. Wrote the manuscript.

Paper III: Experimental design and operation. Wrote the manuscript.

Paper IV: Participated in the work related to the bioprocessing and cell biology. The

manuscript was collectively written.

List of abbreviations

ACD Average Cell Diameter

Amm 1) Ammonia; 2) Ammonia concentration in the culture †

API Active Pharmaceutical Ingredients

ATF Alternating Tangential Flow

ATF Wave bioreactor equipped with ATF perfusion system

CE Capillary Electrophoresis

CHO Chinese Hamster Ovary

CMO Contract Manufacturing Organization

COGS Costs Of Goods

CSPR Cell Specific Perfusion Rate

CSPR_min minimum CSPR

CT CellTank perfusion system

CV Viable Cell density

CVMAX Maximum Viable Cell density

D Perfusion rate

DHFR Dihydrofolate Reductase

DO Dissolved Oxygen

DSP Down Stream Process

ESI-QTOF Electrospray Ionization-Quadrupole Time-Of-Flight (MS)

Glc* 1) Glucose; 2) Glucose concentration in the culture †

GlcNAc N-acetyl-glucosamine

Gln* 1) Glutamine; 2) Glutamine concentration in the culture†

GS Glutamine Synthetase

GSH Glutathione

HCDC High Cell Density Culture

HCP Host Cell Protein

HF Hollow Fiber

HPLC High-Performance Liquid Chromatography

HTtot Accumulated production in harvest

IgG Immunoglobulin G

Lac 1) Lactate; 2) Lactate concentration in the culture †

mAb Monoclonal Antibody

MCB Master Cell Bank

MWCO Molecular Weight Cut Off

MVC/mL Million Viable Cells per milliliter

MS Mass Spectrometry

OUR Oxygen Uptake Rate

PBR Product concentration in bioreactor

PCA Principal Component Analysis

PHT Product concentration in harvest

PLS Partial Least Square

PLS-DA Partial Least Squares-Discriminant Analysis

qamm Cell specific ammonia production rate

qglc Cell specific glucose uptake rate

qgln Cell specific glutamine uptake rate

qlac Cell specific lactate production rate

qp Cell specific productivity

Redox Reduction-oxidation reaction

rdegr Degradation rate (of glutamine)

ROS Reactive Oxygen Species

RV/day Reactor Volume per day

Smedium Substrate concentration in fresh medium

Sstock Substrate stock solution concentration

SUB Single Use Bioreactor

SDS-PAGE Sodium Dodecyl Sulfate-Polyacrylamide Gel Electrophoresis

ST Stirred Tank bioreactor equipped with ATF perfusion system

TCA TriCarboxylic Acid

TFF Tangential Flow Filtration

TFF Wave bioreactor equipped with TFF perfusion system

TNF Tumor Necrosis Factor

t-PA Tissue Plasminogen Activator

USP Up Stream Process

V (Working) Volume

Vbleed Volume of (daily) cell bleed

VHT Volume of (daily) harvest

Vshoot Volume of (daily) supplement

WCB Working Cell Bank

µ Growth rate †In Table 9

Table of Contents 1 Introduction………...…………………………………………….......................... 1

1.1 Outline of this thesis………………………………………………………….......... 1

1.2 Recombinant protein production in mammalian cell systems……………......... 2

1.2.1 Monoclonal Antibodies (mAbs) manufacturing………………………….......... 4

1.2.2 Immunoglobulin G (IgG)…………………………………………………........... 7

1.2.3 CHO cell cellular metabolism………………………………………………........ 8

1.2.3.1 Energy and amino acid metabolism…………………………………….......... 8

1.2.3.2 Metabolomics profiling of extracellular metabolites………………….......... 11

1.2.4 Cultivation techniques………………………………………………………........ 12

1.3 Perfusion process………………………………………………………………........ 14

1.3.1 Perfusion systems……………………………………………………………........ 16

1.3.2 High Cell Density Culture (HCDC)………………………………………........... 20

1.3.3 Single-Use-Bioreactor (SUB) systems……………………………………........... 23

1.4 Perfusion process development and optimization…………………………......... 26

1.4.1 Perfusion rate D……………………………………………………………........... 27

1.4.2 Cell Specific Perfusion Rate (CSPR)………………………………………......... 28

2 Present investigation………………………………………………….................. 30

2.1 Aim of investigation…………………………………………………………........... 30

2.2 HCDC perfusion system development (Paper I-III)……………………............. 31

2.2.1 Perfusion system general set-up…………………………………………........... 32

2.2.2 WAVE Bioreactor™ Cellbag prototype perfusion systems with ATF and TFF

(Paper I) ………………………………………………………………………............... 37

2.2.3 CellTank™ prototype perfusion system (Paper II)……………………............. 38

2.2.4 Glass stirred tank bioreactor with ATF (Paper III)…………………................ 40

2.3 HCDC perfusion process development (Paper I - IV) ………………….............. 40

2.3.1 Maximal viable cell density (Paper I - III)…………………............................. 41

2.3.2 HCDC established and stabilized in the perfusion systems (Paper I-III)......... 44

2.3.3 Perfusion rate optimization (Paper III) …………………................................ 46

2.3.4 Product quality at HCDC (Paper II, III) …………………................................ 48

2.3.5 Cellular metabolism in HCDC (Paper I-IV) …………………........................... 51

2.3.5.1 Energy metabolism (Paper I-III) …………………........................................ 51

2.3.5.2 Metabolomics profiling of extracellular metabolites (Paper IV).................. 52

3 Concluding remarks…………………………………………………................... 55

Acknowledgements…………………………………………………........................

.

57

References…………………………………………………........................................

.

59

1

1 Introduction

1.1 Outline of this thesis

This thesis begins with an introduction to the background of current cultivation

techniques in recombinant protein production by mammalian cell-based systems.

Using Chinese Hamster Ovary (CHO) cells producing monoclonal antibodies (mAbs) as

model system, the thesis focuses on High Cell Density Culture (HCDC) perfusion

process development and covers mainly two parts.

The first part is HCDC perfusion system development with our industrial collaborators.

The second part addresses the HCDC perfusion process development and highlights

investigations of high cell densities and perfusion rate tuning from a more general

process perspective. The cell metabolism and product quality at HCDC perfusion

process are also discussed.

2

1.2 Recombinant protein production in mammalian cell

systems

Following on the recombinant DNA technique breakthrough by biochemist Herbert

Boyer and geneticist Stanley Cohen in the early 1970s, recombinant therapeutic

proteins have been under fast development. Before 1994 the approved

biopharmaceutical products were 25 in United States and European Union markets,

and this number increased significantly to 246 products by 2014 (including 34

products that have been withdrawn after approval in both markets)1.

Host cells are genetically engineered so that the gene encoding for a protein of interest

is incorporated and expressed during cultivation of the host cells. To be fully biological

functional, the expressed target proteins have to be correctly folded and processed with

proper post-translational modifications2, 3. There has been an ongoing shift from non

mammalian-based to mammalian-based expression system: from 50%/50% used in

biologics manufacturing in 1990-1994, to mammalian cell systems occupying 60% for

biologics approved between 2010-20141. Expression based on mammalian cells have

been favored due to the increase of products that require post-translational

modifications, particularly glycosylation. Microbial systems such as prokaryotic cells

Escherichia coli lack such machinery, but still dominate the production of approved

products quantitatively.

The CHO cells were first used in laboratory work in 1919 but the culture technique was

established in 1957, when Dr. Theodore T. Puck identified conditions for good viability

and fast growth. After that, CHO cells began to be intensively used as host cells. They

grow fast with good viability and are easy to culture. The post-translational

modifications of CHO cells provide folded proteins close to human proteins. They can

3

be genetically manipulated to reach high expression level and obtain other attributes

such as metabolic modification. Robust and effective gene amplification systems are

available for CHO cells such as dihydrofolate reductase (DHFR) and glutamine

synthetase (GS). They are also a safe choice with low risk for virus contamination4.

Since the recombinant biopharmaceutical t-PA (tissue plasminogen activator)

expressed in CHO cells was approved in 1987, CHO cells as host cells have

cumulatively occupied 35.5% of the total biopharmaceutical product approvals in US

and EU markets by 20141.

There are also other mammalian host cells available such as human cell lines HEK293

and Per C6 cells, mouse myeloma cell lines NS0 and Sp2/0. Human cell lines are

mainly used in products demanding specific post-translational modifications1. NS0 cells

have competitive high cell specific productivities similar to CHO cells, but have

exhibited less “human glycosylation pattern” compared to CHO cells, with potential

risk of non-human glycosylation and sialylation5. In addition, NS0 cells require

cholesterol supplementation6.

From a bioprocess point of view, CHO cells are relatively easier to be adapted to

suspension growth in serum-free and animal components free media compared to

other mammalian expression systems. Suspension cell culture is advantageous in

homogenous culture environment allowing easy scaling up of the process6, 7.

4

1.2.1 Monoclonal Antibodies (mAbs) manufacturing

Among the 54 recombinant biologics (biologically produced pharmaceuticals)

approved from 2010 to 2014 in US and EU markets, 17 were mAbs. The others were

hormones, enzymes, vaccines etc. and one gene therapy product1. Therapeutic

indications for mAbs range from cancers, inflammatory diseases to autoimmune

diseases. Table 1 presents the top 6 prescription drug biopharmaceuticals in 2015

showing that 5 of them are mAbs. The expiring patents of many of these blockbusters

makes them very attractive targets for biosimilars, copies of approved original mAbs

out of patents.

Table 1 Top 6 Best-selling biologics of 2015

Rank Biologics Expression

System Company

2015

Revenues

billion$

Patent expiry

(EU)

Patent expiry

(US)

1 Humira®

(adalimumab) CHO AbbVie 14.01 2018 2016

2 Rituxan®

(rituximab) CHO Roche 7.33 2013 2016

3

Lantus®

(insulin

glargine)

Escherichia

coli Sanofi 7.09 2014 2014

4 Avastin®

(bevacizumab) CHO Roche 6.95 2019 2017

5 Herceptin®

(trastuzumab) CHO Roche 6.8 2014 2019

6 Remicade®

(infliximab) CHO Janssen 6.56 2015 2018

5

Taking Remicade® as example: Remicade® is the trade name for infliximab, an anti-

TNFα (Tumor Necrosis Factor alpha) chimeric mAb drug for autoimmune diseases

treatment. It was originally approved by the US Food and Drug Administration (FDA)

in 1998. In 2016, two biosimilars of Remicade®, Inflectra® (Pfizer) and Remsima®

(Celltrion) have been registered by FDA. Another biosimilar, Flixabi® (Biogen), has

been approved in a few countries in EU. The race about titer, time, and cost is no

longer the privilege of larger pharmaceutical corporations, but many small companies

are present with a lot of efforts devoted to process development and optimization8.

There are three production technologies commonly used in mAb manufacturing9:

transient expression, stable cell pool, and stable clonal cell line. All three methods

involve getting the foreign target gene into the cells. In the transient expression

method the foreign DNA does not integrate into the host genome. Therefore the target

gene does not replicate and eventually gets lost over several days. It yields ≈ 50 µg to

gram per liter level expression in a short time (6-10 days from the point of

transfection) depends on the host cell line and it is suitable for milligrams to grams of

mAb production under short time constraints, for instance in preclinical development10-

12. As an alternative for rapid protein production, stable cell pool refers to a

heterogeneous pool of transfected host cells where the foreign gene was integrated into

the host genome and reproduce along when the host cells proliferate. This method

yields ≈ 300 mg to over gram per liter and the turnaround time is also relatively fast as

early as 8 weeks13, 14. It is ideal for antibody reagents generation and has certain

advantages over the transient expression: it can be frozen down to working stock for

production to avoid repeated transfections; the medium change that is difficult to

operate at large scale in transient expression is eliminated.

Compared to cell pool method that contains cells with different expression levels, the

stable clonal cell line method screens and selects high producers in the pool, leading to

6

cell clones with multiple gram-level expressions. A single clone is then used to establish

a cell line that is then amplified and cryopreserved. The cryopreservation is called cell

banking, usually is a two-tiered system - a master cell bank (MCB) and a working cell

bank (WCB). A MCB usually contains 50-400 ampoules and a WCB is derived from one

ampoule of the MCB15. The establishment of stable production cell lines usually takes

14 weeks up to a whole year and they are used over many manufacturing cycles for

consistent manufacturing of the mAb products.

The mAb manufacturing is a process taking several weeks that can be separated into

upstream process (USP) and downstream process (DSP). In the USP cells are thawed

from the WCB and expanded in shake flasks placed in CO2 supplemented incubators.

The culture is then amplified in seed train in bioreactors in increasing sizes. The

production cultivation takes place in the production bioreactor by one of the following

modes: batch, fed-batch or perfusion. In batch/fed-batch processes, the harvested cell

broth is processed in DSP that usually starts with centrifugation to collect the cell-free

supernatant where the mAb product has been excreted, while in perfusion the cell-free

supernatant is available in the harvest tank. A series of filtration and chromatography

steps are carried out to purify, concentrate and finally put the product in an adequate

formulation buffer.

The current thesis focuses on USP only.

7

1.2.2 Immunoglobulin G (IgG)

IgG is the most abundant immunoglobulin in human serum (70-75%). Figure 1

illustrates IgG and its Fc N-glycosylation. Immunoglobulin glycosylation is one of the

most important post-translational modifications. For this, sugar moieties are covalently

added to specific amino acid residues of the protein. Correct glycosylation is critical to

protein solubility, functionality and stability to generate biologically effective

biotherapeutics. For IgG the N-glycosylation site on the Fc fragment is highly conserved

at CH2 domain Asn-297 of each heavy chain. It is a biantennary structure that is often

core-fucosylated. Figure 1 lists the most common four structures G0F, G1F, G2F, and

Man5 based on the degree of galactosylation. Variation in glycosylation profiles mainly

focus on the degree of fucosylation, N-acetyl-glucosamine (GlcNAc), galactosylation,

and sialylation.

Figure 1 IgG and Fc N-glycosylation

Fab

Fc

VL

C L

V H

CH1

CH2

CH3

N297

N297

N297

N297

G0F

G1F

G2FS1

Man5

GlcNAc Man Galactose Fucose Sialic acid

8

1.2.3 CHO cell cellular metabolism

Mammalian cells are complex systems with a large amount of biochemical reactions

taking place simultaneously. To be able to improve recombinant cell performance in

culture, it is important to understand the cellular metabolism.

1.2.3.1 Energy and amino acid metabolism

Figure 2 illustrates the energy metabolism pathways in CHO cells focusing on

glycolysis, TriCarboxylic Acid (TCA) cycle, and glutaminolysis. Glucose serves as the

main energy source for the cells and the aerobic complete oxidation (via TCA cycle) of

glucose generates 32 molecules of ATP per mole of glucose. However cultured cells

also utilize the anaerobic glycolysis that generates only 2 ATP known as Warburg

effect16, producing lactate in the cytosol before it is excreted. Lactate acidifies the

culture and is considered one of the most critical byproducts.

Previous studies have shown that the cells utilize glucose more efficiently when a low

glucose level is present in the culture medium17, 18. At a low environmental glucose

concentration, the cells tend to use the TCA cycle pathway to generate more energy,

while at a high environmental glucose concentration, the cell metabolism switches to

anaerobic glycolysis and produces mainly lactate17, 19, 20.

The other main byproduct ammonia comes from amino acids catabolism, mainly from

glutamine. Glutamine is an unstable essential amino acid that is the major source of

nitrogen, carbon and energy for the cells. Its oxidation (glutaminolysis) leads to

ammonia and lactate, as well as non-essential amino acids such as Ala and Asn21-24.

Glutamine produces ammonia from the metabolic degradation in the mitochondria but

a cell-free spontaneous degradation in culture medium is also taking place due to

9

instability at physiological pH. Both lactate and ammonia are known to be toxic to

mammalian cell growth above certain levels, but the actual inhibitory concentrations

vary with cell lines and processes25-28.

Figure 2 Simplified metabolic pathways for CHO cells showing mostly amino acids catabolism

To minimize the adverse effect of lactate and ammonia to the culture, controlled

feeding strategies of glucose and glutamine have been adopted29, 30. As an alternative,

the introduction of GS gene in host cell genome allows the cells to grow without

glutamine supplement31-33. This is now widely used as a highly efficient selection

system in different mammalian cell lines34-37. Alternatively dipeptides containing

glutamine, such as glutamax (AlaGln), are used.

10

Amino acids are essential for biomass building up, energy metabolism and also protein

products formation. They are important building blocks of cellular components such as

glutathione. Glutathione (GSH) is a tripeptide generated from glutamate, cysteine, and

glycine, a key survival antioxidant that protects cells from damaging caused by reactive

oxygen species (ROS) such as free radicals and peroxides. It maintains cellular

reduction-oxidation reaction (redox) balance and alleviates cellular oxidative stress38-40.

11

1.2.3.2 Metabolomics profiling of extracellular metabolites

Nowadays various omics analytical techniques/platforms e.g. transcriptomics,

genomics, proteomics, metabolomics and statistical tools have become available for

mammalian cell engineering application41-50. This has largely enhanced the

understanding of mammalian cell metabolism. Metabolomics profile of the culture

provides knowledge about numerous extracellular metabolites and has a great

potential in discovery and biomarker identification for cell engineering purpose51-60.

Compared with other omics based tools, metabolomics reveals the phenotype defining

processes that pinpoint the exact metabolites directly responsible for cell behavior61.

Metabolomics profiling is usually performed on High Performance Liquid

Chromatography Mass spectrometry (HPLC/MS) analytical platform and huge data sets

are generated. Intensive data-mining is then needed to obtain useful information and

interpret the data, which is still very technically challenging in reality62-65. Different

multivariate data analysis tools are therefore been employed to extract information

from the data, both the metabolomics analysis results and their correlation with the

bioprocess parameters in cultures65, 66. Among all the available tools Principal

Component Analysis (PCA), Partial Least Square (PLS) regression and Partial Least

Squares-Discriminant Analysis (PLS-DA) are the most commonly used in bioprocess

field65, 67, 68.

12

1.2.4 Cultivation techniques

There are four cultivation modes commonly used for mammalian cells based

recombinant protein manufacturing: batch, fed-batch, continuous and perfusion as

shown in Figure 3 below.

Figure 3 Four main cultivation techniques in recombinant protein production

Since the first approved recombinant therapeutic protein Activase (Genentech)

produced in CHO cells in 1987, the manufacturing techniques of recombinant proteins

have significantly evolved. Early manufacturing employed 7-day batch cultivation with

medium containing animal components such as serum. In batch process, the cells grow

in a nutrient-rich base medium with a fixed volume and the whole batch is harvested

when the production is terminated. The process usually lasts for a week and the whole

batch is harvested when the viability drops to below 30%69. Its simple setup makes it

relatively easy to perform compared with the other techniques, but there are several

drawbacks such as low productivity, low cell density, nutrients depletion, byproducts

accumulation, etc.

In fed-batch mode, a nutrient-rich feed medium is added to the cultivation with a

defined feeding profile to avoid nutrients depletion and to support higher cell densities

Batch Fed-batch Continuous Perfusion

Feed Feed Cell-free

Harvest

Cells

Feed Harvest

13

and longer period of production. The production processes are 10-21 days with an

average of 14 days. In the fed-batch processes the production bioreactors are reaching

25,000 liters. Drawbacks are however that byproduct accumulation is inevitable in fed-

batch mode and that the culture is changing with time, which can cause a product

quality variation with time70.

In chemostat, the feed medium is added at the same rate as part of the culture broth is

removed. In this mode a constant volume is maintained and steady state can be

maintained for a long time. This mode is sometimes adopted in process development

but not in production, due to intrinsic disadvantages of this technique, such as low

achievable cell densities due to the low growth rate of mammalian cells, and the

complex product recovery from collected cell broth71.

In perfusion process the cells are retained in the bioreactor, cell-free supernatant is

continuously collected and fresh medium is continuously added at the same rate. In

this cultivation mode, the cells are accumulated in the bioreactor, nutrients are

continuously fed and inhibiting waste products are continuously discarded so it is

possible to reach high cell densities with high cell viabilities and long duration of the

culture. Although process specific, protein synthesis is usually proportional to the

number of productive cells. Therefore a high cell density implies a high volumetric

productivity for both growth related and non-growth related products. Generally

speaking perfusion systems are able to achieve 4-10 times higher volumetric

production rates compared to fed-batch process due to potential high cell densities72.

The collected cell-free supernatant can be either stored at 4°C until preparation for

purification, or directly to a continuous DSP, depending on the facility infrastructure.

With the maturation of cell culture technology in the past decades, high titers with

desired quality have been successfully achieved in fed-batch processes. New progresses

14

in gene cloning and clone selection have largely enhanced the cell growth, viability,

cell specific productivity as well as stability. Chemically defined media can nowadays

support better culture performance with higher productivities, longer production time

and higher cell viability. Product titer of 10 g/L or even higher have been reported73.

Despite the high performances of fed-batch technique, there is today a shift of process

development motivated by cost reductions, process intensification and higher control of

the product quality and process consistency74.

1.3 Perfusion process

A perfusion process has a constant renewal of the culture medium, which provides a

stable environment to cells and the product as well as a short retention time to the

product. Due to the high cell densities, high volumetric productivities can be achieved

leading to higher space-time yields. Although the process is relatively more

complicated than fed-batch, perfusion technique has recently received an increasing

interest in the biopharmaceutical industry75 and is widely acknowledged nowadays as

an efficient cultivation technique for production, for seeding train, and/or for high cell

density banking76-78. Due to the shorter residence time of continuous renewal of the

culture medium, perfusion process is the ideal choice for unstable active

pharmaceutical ingredients (APIs), for example coagulation factor VIII Kogenate®

(Bayer), ReFacto® (Pfizer), and coagulation factor VIIa Novoseven® (Novo Nordisk).

Interestingly, although the majority of biopharmaceuticals are still manufactured using

fed-batch process, historically big pharmaceutical companies have successfully applied

perfusion process in their commercial biologics manufacturing7, 79. Some biologics

manufactured with perfusion systems are Protein C Xigris® (Eli Lilly), mAbs such as

15

Campath® (Sanofi), Simulect® (Novartis), ReoPro®, Remicade®, Simponi® and Stelara®

(Janssen Biotech, formerly Centocor).

Perfusion technology has several advantages over fed-batch and batch modes:

o Optimum and stable environment for the cells and protein products

o Low byproducts accumulation due to the continuous medium removal

o High cell density with potentially high viability

o High and consistent product quality

o Long-term production

o Reduced manufacturing facility cost capital expenditure and associated costs

o Much smaller size bioreactors to reach the same yearly yield of product

o Amenability to disposable equipment

o Manufacturing flexibility

o …

Some of these listed advantages correlate with each other, for instance an optimum

and stable environment for the cells leads to high viability and less apoptosis in the

culture, which minimizes the host cell protein (HCP) released into the culture that

simplifies downstream purification process.

Since a couple of years, there has been a paradigm shift towards continuous

manufacturing in biopharmaceutical industry80-82. Integration of perfusion process with

continuous downstream process has been explored and is under fast development83-85.

It has been described that compared with existing fed-batch production process, a fully

integrated continuous process of upstream and downstream can cut down the

recombinant protein manufacturing costs of goods (COGS) by an average of 55%, in

particular due to a smaller capital expenditure86.

16

1.3.1 Perfusion systems

The cell retention module is a unique feature in perfusion systems. Perfusion systems

can be generalized into two types (Fig.4A&B) based on the cell separation techniques:

i) cell separation ensured by a device and ii) cell immobilization/entrapment in a

system continuously perfused with circulation of medium.

Case i) is shown in Fig.4A: the cells are cultured in suspension in a bioreactor, and a

separate cell retention device is used to retain the cells in the bioreactor while the cell-

free supernatant is processed to the harvest tank. Such cell separation devices can be

connected with any kind of bioreactor to establish a perfusion system. The most

commonly used cell separation devices are filtration-based methods, acoustic settler,

gravity settler, centrifuge, etc. as in Fig.4A187.

Figure 4 Perfusion systems

A): Cell suspension with cell retention system outside (A1) and inside (A2) the bioreactor

B): Cell immobilized or entrapped in a matrix outside (B1) and inside (B2) the bioreactor

Feed Cell-free Harvest

Feed

Cell retention system

(A1) (B1)

Feed

Feed

(A2) (B2)

Cell suspension

Perfused tank

Cell loaded matrix

Cell loaded matrix

Cell suspension

Cell-free Harvest

Cell-free Harvest

Cell-free Harvest

17

The cell separation module can also be integrated into the bioreactor as shown in

Fig.4A2. This system has the advantages that there is no need to pump the cell broth

out, so that the potential pump shear stress is removed, and that the cells are less

subject to varying environment; for instance there is no temperature drop due to

circulation in the tubing and to/from the cell retention system. Some filtration-based

devices have been developed and are widely used for this case, for example internal

spin filters are used in the manufacturing of ReoPro® and Remicade® in 500 L

bioreactors in Janssen Biotech79. A summary of the cell retention devices is given in

Table 2.

Fig.4B demonstrates the other type of perfusion system (ii), where the cells are

anchored or entrapped in a support such as a membrane or a matrix/scaffold. The cell-

loaded matrix is often outside (Fig.4B1) of the perfused tank, but configurations inside

the tank have been developed (Fig.4B2). In this setting the homogenization of the

culture is performed by either magnetic stirrer or propellers in the perfused tank,

where the control of the culture parameters such as temperature, dissolved oxygen

(DO) and pH takes place. The concept is that as long as the medium circulation

between the cell-loaded matrix and the perfused tank is rapid and adequate, the

culture environment in the tank is similar to the environment of the matrix where the

cells are located.

Typical examples of such bioreactors are packed bed bioreactor and fibrous bed

bioreactors with different scaffolds for instance depth filter88, ceramics matrix89 and

fibrous matrix90. It can also be cells attached or captured on immobilized

microcarriers91, 92 or hollow fibers93. Gradients of nutrients, metabolites, gas exchange,

and dead cells accumulation are common drawbacks in such setting94, 95. Such

heterogeneous cell environment could potentially trigger cell population shift,

productivity and product quality variations, even necrosis in extreme cases. This

18

heterogeneity also implies scaling up limitation96, therefore this kind of perfusion

system has not yet been adopted in very large-scale production process, except for

Rebif®, an interferon β-1a by Merck-Serono, was reported to be manufactured using

multiple 75 L fixed bed bioreactors79. Table 2 Summary of cell retention devices

Type Principle Cell retention device Reference

Cells immobilized or

entrapped in a

matrix/scaffold

Packed bed

FibraCel™ nonwoven polyester

base carriers 97-99

iCELLis® bioreactor (polyester

microfibers)

Quantum hollow-fiber-based

(Beckman-Coulter)

Ceramic matrix 89

Porous glass sphere 92

3D annular porous scaffolds 100

Fibrous bed Polyester fabric sheet 90

Depth filter 88

Suspension cells with

individual separation

device

Centrifugation Centritech® 101-103

Sedimentation Inclined settler, gravity settler 104-109

High frequency resonant

ultrasonic waves Acoustic settler 110-118

Acceleration Hydrocyclone 119-121

Dielectrophoresis Dielectrophoretic separator 122

Filtration

Dynamic rotating disc filter 123

Floating membrane filter 124-128

Coiled porous fibers 129

Hollow fiber 130

TFF 77, 131, 132

ATF 77, 131, 133

Spin filter 134, 135

Vortex flow filtration 136, 137

Ceramic filter 138

19

Among all the available methods, filtration based systems have a very high retention

(100%) of the cells. A drawback of filtration based systems is the clogging and fouling

of filters, which may lead to retention of the product of interest or termination of the

cultivation, alternatively need of replacement by a new cartridge. The tangential flow

filtration (TFF) is also known as crossflow filtration, where the feed in flow travels

along the filter surface instead of passing through the filter (Fig. 5).

Figure 5 TFF and ATF process for IgG production

In this flow pattern the filter cake that is clogging the filter is largely removed, leading

to a substantially increased operation time of the filter. It uses hollow fiber (HF) that

packs numerous long porous filaments in parallel inside a cartridge. Crossflow

filtration alleviates fouling, but the increased flow rate in the filter leads to increased

shear rate. The alternating tangential flow filtration (ATF) shares the same principal as

TFF, except that the cell broth is drawn through the filter and pushed back to the

Permeate

Diaphragm

Controller

Feed/Retentate

Permeat

e

Feed

Retentate

Cell suspension Membrane

Permeate

TFF ATF

Bioreactor Bioreactor

Section

20

reactor by a diaphragm pump, which functions cyclically using vacuum and

compressed air. The rapid flow back and forth helps preventing fouling, and the

pressure variation over the membrane occurring during pump cycles leads to back flush

of the filtration system.

Hollow Fiber cartridges used in cell culture perfusion processes are generally divided

into two main types based on the membrane pore size: microfiltration 0.1 to 0.65 µm

and ultrafiltration 30 to 750 kDa of molecular weight cut off (MWCO). The selection of

the HF membrane depends on the application. In the case of a production process for

IgG with approximate molecular weight of 150 kDa, a microfilter allows the retention

of all the cells in the bioreactor and the product is collected in the permeate side of the

cartridge as shown in Fig.5.

An ultrafiltration filter with much smaller pore size retains both the cells and the

protein product inside the bioreactor. The product accumulation can reach very high

titer due to the fast increasing cell density and is suitable for stable protein products

with no or little degradation issues.

1.3.2 High Cell Density Culture (HCDC)

By implementing a cell retention device, a high cell density culture (HCDC) is

achievable and definitely one of the most advantageous features of a perfusion process

if accompanied by a high volumetric productivity. Compared to cell densities 5-25

MVC/mL reached in a typical fed-batch mode139, cell densities of 20 to 100 MVC/mL in

perfusion cultivations (homogenous suspensions) have been previously reported76, 123,

126, 127, 132, 135. HCDC is very challenging from both the culture system performances and

process techniques perspectives. Limiting oxygen and mass transfer are commonly

21

observed140 and excessive accumulation of CO2 in HCDC large bioreactors is also a

problem for the cell growth and product quality.

In the case of a non-growth related production, a perfusion process can be divided into

two phases: a growth phase where the cell density reaches a desired high number,

followed by a production phase where the high cell density is maintained by cell

bleed141-145 or hypothermia146-150. Cell bleed refers to partial removal of cell broth and

contains as well dead cells and cell debris. Hypothermia induces cell cycle arrest at the

G0/G1 phase151 and slows down the cell growth and cellular metabolism. In many

cases, cell arrest is accompanied by an enhanced productivity146-150.

As part of the perfusion system development, the feasible HCDC for production phase

needs to be determined. Earlier studies have recommended an optimum viable cell

density (Cv) of 40-80 MVC/mL to reach a high volumetric production152 but Wright et

al have reported that only very few processes were able to achieve such high Cv145.

It is unclear which cell density can be (or should be) called 'high cell density', and this

value is certainly evolving due to the progresses of perfusion processes enabling higher

cell densities. In this context, the present thesis reports extremely high cell densities

well above 100 MVC/ml. Table 3 summarizes published data of HCDCs restricted to ≥

40 MVC/mL and using commonly used cell lines. These HCDCs have been mainly

established during process development and not in manufacturing.

It is worth pointing out that some of the high cell densities reported were not from

direct cell sample analysis. For example Oh et al. reported a density of 187 MVC/mL

hybridoma cells achieved in a hollow fiber bioreactor, calculated from the production

data by assuming a constant cell specific productivity qp. This is however not always

correct due to cellular metabolism variation. The same group compared two

22

approaches for CV calculation, one based on a constant cell specific glucose

consumption rate qglc and another based on a constant qp and obtained 60 MVC/mL

using the first method and 30 MVC/mL using the second method on exactly the same

culture153.

Table 3. HCDC of mammalian/human cells above 40 MVC/mL reported in the literature

Cell line Product CV

(MVC/mL)

D

(RV/day)

Cell retention device Reference

S2 mAb 104 9.9 Floating filter 126

Sf21 55 6 HF

154

CHO mAb 68.3 1 ATF 155

CHO mAb 132 6 ATF 131 present thesis

CHO mAb 214 10 TFF 131 present thesis

CHO mAb 2001 10 Fibrous bed 156 present thesis

CHO mAb 50-60 ATF 85

CHO TNK-tPA 60 6 Acoustic filtration

system

157

Hybridoma mAb 1872 39 Dual hollow fiber

bioreactor

153, 158

Hybridoma mAb 1003 Packed bed 99

Hybridoma mAb 101 Fibrous bed 159

SP20 IgM 45 11 Vortex flow filter 136

HEK293 100 ~6.4 HF 160

1cell density given by online biomass probe 2cell density calculated from culture productivity 3cell density per cm3 of packed bed volume

23

1.3.3 Single-Use-Bioreactor (SUB) systems

Despite several obvious advantages, perfusion operation mode has still limited

application due to its inherent technical complexity. Disposable systems can help to

alleviate the technical operational complexity and decrease the risk of contamination.

Disposable technologies are under fast development in the biopharmaceutical industry,

covering the whole manufacturing process from vial thaw to the final product. The

disposable systems offer benefits such as161:

o Elimination of contamination risk due to pre-sterilization

o Minimization of turnover time between cultivations due to absence (or high

reduction) of cleaning or sterilization

o Elimination of clean validation

o Reduction of staff training and simplified set up

o The process can be replicated or transferred between different facilities due to small

footprint and close systems

o Low risks of cross contamination. Various products can be produced at the same

facility, a factor critical for contract manufacturing organizations (CMOs).

o …

Three kinds of SUB systems based on different agitation principles can be

distinguished: 1) disposable culture bags on a rocking table; 2) conventional stirred

tank bioreactors where the stainless steel or glass vessel is replaced by polymer-based

containers, including plastic stirred tanks and stirred bags that are placed in a

supporting tank; 3) Hollow fiber bioreactors. Table 4 summarizes the SUB systems

used in mammalian cell based upstream processes for scale ≥ 1 L. Here traditional

shake flasks, roller bottles and (multi-layer) culture dishes are not included.

24

Table 4 SUB systems used in mammalian cell based upstream processes for scale ≥ 1 L

Type Name Vendor Scale Stirring Ref.

Plastic stirred tank or stirred

bag

UniVessel® SU Sartorius-Stedim 2.6 L

Impeller

162, 163

Mobius® CellReady Applikon/Merck

Millipore 3 L 164

Brunswick™ CelliGen® BLU

Eppendorf 5L, 14L 165

CellVessel™ Cercell 500 mL - 30

L

Bio Bench SUB Solida Biotech 250 mL - 75

L

BIOne Distek HyClone S.U.B. Thermo Fisher 50 - 2000 L 166

Xcellerex XDR-50 - 2000 GE Healthcare 50, 200, 500, 1000, 2000L

167

Biostat® STR Sartorius-Stedim 12-1000 L 168

BaySHAKE® Bayer 50 L - 1000 L Rotary

oscillation 169

SB10/50/200-X (OrbShake)

Kühner 10L, 50L,

200L Orbital shaking

Nucleo™ Bioreactor/PadReactor®

ATMI/PALL 25 - 1200 L Paddle 163

Disposable bags on a rocking

table

WAVE Bioreactor™ GE Healthcare mL - 1000 L

Rocking

124-127,

170

Biostat® RM Sartorius-Stedim 100 mL - 300

L

162, 171-

173 AppliFlex Applikon 10 L - 50 L

Cell-tainer® Celltainer biotech 20 L, 200 L 2-dimensional

movement 174

Hollow fiber bioreactor

FiberCell Systems FiberCell Systems 175 CellMax Spectrum labs

Others

PBS PBS 3, 15, 80,

500 L Vertical-Wheel™

176

iCellis Applikon/PALL 1 - 70 L Magnetic

drive impeller

CellMaker Plus Cellexus

Biosystems 1 - 50 L

Airlift+magnetic stirring

177

25

Since the introduction of the disposable cell culture bag178, this idea has been rapidly

adapted to existing technologies and platforms. One of the most successful SUB, the

wave bioreactor, has a wave like rocking motion instead of traditional mechanical

mixing. It facilitates the gas and mass transfer, and generates less damaging bubbles

compared to a sparged bioreactor, leading to a significantly reduced shear stress

environment for the cells. It is widely used mainly for the seeding train but also as

production bioreactor in various biologics manufacturing from different cell lines124, 179-

182as well as for cell banking.

To provide a closer mimic of conventional stirrer tank bioreactors and to address the

oxygen transfer challenge at high cell densities in large-scale cultivation, large

disposable stirred tank bioreactors are made of a culture bag placed in a support tank,

and are called “stirred bag”. They are available in different scales up to 2000 L, and

reported to have comparable performance to their stirred tank counterparts183-186 with

all the advantages offered by disposable systems.

The benefits from SUBs make them the ideal support for perfusion process.

Development of both perfusion process and disposable techniques are currently

pushing the industry towards a next-generation continuous manufacturing process.

26

1.4 Perfusion process development and optimization

Upstream process (USP) development includes different aspects, is commonly

platform-based and preferably performed in-house. The ultimate goal is to reduce the

cost per product9, 187, 188 and obtain a robust process ensuring a safe product with the

targeted quality. The factors involved in perfusion process optimization are described

in Table 5. Table 5 Perfusion process optimization

Target Optimization goal Candidates

Cell line Gene cloning Clone selection

o Cell growth o Productivity o Cell line stability o Product quality o …

Medium formulation

Media/additives

o Cell growth and survival o Productivity o Byproducts o Product quality o …

o Carbon sources o Amino acids o Vitamins o Inorganic salts o Trace metals o Lipids

Equipment Cell retention set up

o Cell retention efficiency o Product retention o Viability o …

o Selection of device o Flow rate o Shear stress o Specific operational

parameters o …

Process control and scale up

o Temperature o pH o Perfusion rate o Working cell density o Dissolved oxygen o pCO2 o …

Analytical methods

o Cell density/Biovolume o Metabolites o Product titer o Product quality o HCP o DNA

27

1.4.1 Perfusion rate D

The perfusion rate is one of the most important parameters in a perfusion process. An

optimal perfusion rate is vital in delivering nutrients and removing byproducts in

accordance to the cell needs to support high productivity and product quality. It is also

essential for the product cost, since large media expenses in perfusion mode is always a

big concern. In some applications where the protein quantity is prioritized within a

short time restraint, such as research and preclinical studies, excessive medium feeding

can be preferred to minimize the risk. However in most cases, the medium

consumption needs to be taken into consideration, and mainly three strategies are

considered for the selection of the perfusion rate: nutrients/metabolites based, cell

density based and in-house experience based, depending on the experience of the

company or the scientist developing the process (Table 6).

Table 6 Perfusion rate selection strategies

Adjustment principle Based on References

Cell density-based Cell specific perfusion rate (CSPR) 131, 189, 190

Metabolite-based Residual nutrients/metabolites 191-194

OUR 160, 195

Experience-based Previous know-how 98

Among all the metabolite-based strategies, although oxygen uptake rate (OUR) is the

most accurate online analysis for cellular metabolism196, the residual glucose

concentration in the culture is often used approach due to its simple implementation.

28

1.4.2 Cell specific perfusion rate (CSPR)

A strategy for the perfusion rate selection based on CSPR (unit nL/cell/day) assumes

that the medium is exchanged at a rate proportional to the cell density. It provides a

consistent microenvironment to the cells in the culture, regardless of the cell density. It

is a straightforward strategy that links the perfusion rate to the cell density, a culture

parameter that is easily accessible and highly reliable. This strategy does not take

potential metabolic variations in consideration.

The CSPR is the ratio of the perfusion rate D (unit Reactor volume/day, RV/day) to the

viable cell density CV (unit MVC/mL), as shown in eq. 1:

CSPR = ! !!

eq. 1

For a defined CSPR, D is given as a function of CV. CSPR based feeding strategy can be

applied automatically by the control station with online biomass measurement. It

allows minor and steady regulation of D in response to CV variations and can be

preferred instead of step-wise change of D.

A key factor is the minimum CSPR (CSPR_min) that delivers nutrients meeting cell

needs and supports high productivity. Not to mention that the lower CSPR is, the lower

medium amount is spent, which becomes substantial at high cell densities.

Konstantinov et al. suggested that a CSPR below 50 pL/cell/day offers batch-like titer

and that CSPR of 40-50 pL/cell/day has been successfully used to support HCDC 85, 131.

However the CSPR value is determined by the “richness” of the culture media, also

known as “medium depth”. Fortified media are able to sustain HCDC at much lower

CSPR, for example Xu S et al. maintained steady states at 42.5 and 68.3 MVC/mL at

CSPR 23 and 15 pL/cell/day respectively in two processes155.

29

To determine the CSPR_min for a process it is usual to employ a steady state or

pseudo-steady state screening strategy. A perfusion culture at a reference CSPR is first

stabilized, then the CSPR is stepwise varied and at each step a new steady state (or

pseudo-steady state) is established. The culture performances are monitored to decide

whether the CSPR needs to be further varied 152, 197. During this procedure, the

perfusion medium can be optimized to improve the performances.

Steady states investigation is rather time-consuming as it usually takes over a week to

establish steady states152, 198. This not only raises development cost, but also brings in

concerns for potential cell population shift due to the long time span. Transient

responses to process parameter variation have been reported previously to show a

qualitative prediction similar to steady states198. Therefore in our study a strategy based

on pseudo-steady states was adopted for process parameter screening and

optimization. Each process parameter variation was usually maintained for minimum 5

days and the first 2 days of each variation was considered as "transient" and thus was

not taken into data analyses. For simplicity in the text, pseudo-steady state is simply

named steady state.

30

2 Present investigations

2.1 Aim of investigation

Monoclonal antibodies are used intensively in biomedical science and as effective

therapeutic treatments against autoimmune disorders, cancers. Since the first use of

CHO cells in recombinant protein production, the manufacturing techniques have been

steadily developed. There has been a changing processing paradigm towards

continuous manufacturing and leading pharmaceutical companies such as Janssen,

Genzyme, Novartis have already adopted perfusion process in their mAb/recombinant

protein products pipeline. As one of the key advantages that perfusion process offers,

high achievable cell density is attractive but very demanding in equipment advances

and process knowledge.

The aim of this thesis is to contribute to the understanding and evolution of

mammalian cell based perfusion process in general. We target at high cell density

perfusion process and intend to provide know-how for its establishment as well as

optimization.

By working closely with our industrial collaborators, we develop high cell density

perfusion systems for antibody producing CHO cells that includes both equipment

development from industrial prototypes and process knowledge advancement. The

present investigation thus starts with HCDC perfusion system development, followed

by HCDC perfusion process development.

31

2.2 HCDC perfusion system development

The thesis includes three projects and four perfusion system configurations (Table 7):

WAVE bioreactor™ with a Cellbag prototype connected with ATF or TFF, named 'ATF'

and 'TFF' respectively; CellTank™ prototype named 'CT' ; stirred tank connected with

ATF named 'ST' . Three cell lines, exhibiting different productivities, were used in the

thesis. Cell line #1 (used in ATF and TFF runs) was a research cell line CHO DHFR-

producing IgG1, named K4. Cell line #2 (used in CT and ST#1) was CHO DP-12 clone

#1934 (ATCC), a research cell line with low productivity. Cell line #3 (used in ST) was

an industrial high producer CHOM provided by Selexis (Switzerland). All the cell lines

were adapted to suspension growth in serum-free and medium free of animal derived

component. Table 7 Summary of the perfusion runs

Perfusion system Exp. Cell line Collaborations Paper

WAVE™ + ATF

'ATF'

ATF#4

K4 GE Healthcare

(Sweden, USA) I, IV

ATF#5

ATF#8

ATF#9

ATF#15A

ATF#15B

WAVE™ + TFF

'TFF'

TFF#6

TFF#10

Fiber matrix based – CellTank™

'CT'

CT#1

DP-12 Belach (Sweden)

PerfuseCell (Denmark) II CT#2

CT#3

Glass stirred tank + ATF

'ST'

ST#1 DP-12 Belach (Sweden)

Iprabio (Belgium)

Selexis (Switzerland)

III ST#2 CHOM

ST#3

32

2.2.1 Perfusion system general set-up

The perfusion was performed by either the control systems WAVEPOD™ for ATF&TFF,

and CytoSys™ for some ST runs, or peristaltic pumps with manual tuning. Except one

run (ST#2) where the CytoSys™ was used to perform automatic bleeding at given cell

density according to online biomass sensor, the cell bleeding was done with peristaltic

pump based on daily cell growth information. The hollow filter cartridges were

ReadyToProcess™ filters (GE Healthcare) with 0.2 micron pore size and surface area

850 cm2 for ATF&TFF runs and 420 cm2 for ST runs.

Figure 6 Perfusion system general set-up

x

Harvest

Cell bleeds

bubble flask

CO2 O2 Air/N2

Alkali

Feed

Intermediate flask

Glucose

Glutamine

50 mL syringe 10 mL syringe

Sampling

Bioreactor

Sampling

Antifoam

Controller

33

A general perfusion system scheme is given in Figure 6, showing the bioreactor

connected with ATF system as an example. The bioreactor was a stirred tank equipped

with propellers or magnetic stirrer, or a disposable culture bag on a rocking table.

Alternatively a CellTank™ prototype system was used where a matrix was immersed in

a reservoir (Paper II). Daily feed was prepared in an intermediate flask with feed

medium, glutamine and glucose fulfilling cells’ daily needs, calculated from the cell

consumption. Antifoam was supplemented if needed. The intermediate flask helped to

deliver a homogenized and accurate feed according to the needs of the cells, as well as

minimize glutamine degradation in the feed.

In the case of suspension culture, daily samples were taken via the sampling line,

which was also used as cell bleeding line. Cell bleed was performed manually to reach

the target CV or continuously via a peristaltic pump that was calibrated to keep a given

CV. An alternative was that online biomass probe was installed in the system, with

control based on the biomass signal to carry out the cell bleed according to a CV set

point. This bleeding set up was used in experiment ST#2 in in stirred tank with

CytoSys® control system from Iprabio, Belgium. For simplicity, the inoculation line is

not represented in Figure 6. The inoculation was performed using the feed or bleeding

line depending on the connections available and experimental settings. A “bubble

flask” is a sterile flask, half filled with water, where the exhaust air enters and the

bubbles are monitored. This is a practical setting to watch the aeration of the whole

system, as well as to maintain a slight overpressure of the bioreactor system to lower

the risk of contamination. In the configurations involving ATF and TFF systems, the HF

cartridges (GE Healthcare, Uppsala, Sweden) were mounted vertically on an ATF-2

station (Refine Technology) or installed with a Watson-Marlow 620S pump as the

recirculation pump. Cell free harvest was pumped out continuously from the permeate

side of the HF and collected in a harvest tank. The perfusion was performed by the

34

control station with assigned perfusion rates or manually through two peristaltic

pumps, one for feed in and one for the harvest.

In all the experiments of ATF, TFF and ST, cell samples were taken twice a day from the

bioreactor and harvest. For CT runs, samples from the matrix, reservoir and harvest

were daily taken. The cell density, viability, average cell diameter (ACD), pH, pCO2,

pO2, osmolality, concentrations of glucose, lactate, glutamate, glutamine, ammonia

were analyzed offline with Bioprofile FLEX (Nova Biomedical). In CT runs, cell samples

were not available, so the viable cell number was monitored by on online biomass

sensor and the culture viability was calculated from the activity of lactate

dehydrogenase (LDH, Promega) in the supernatant. The perfusion rate was calculated

from daily weight of feed medium, harvest and cell bleeds.

The process parameters dissolved oxygen (DO), pH, biomass etc. were on-line

monitored and controlled as listed in Table 8.

Table 8 Monitoring and feedback control of the bioreactor

Control Set-point Up regulation Down regulation ATF TFF CT ST

Agitation -

Propeller �

Rocking on a rocking table � �

Magnetic stirring �

Tempera

ture 37°C

Heating jacket around the bioreactor � �

Heating element on the rocking table � �

DO 40%

O2, Air to headspace � � � �

Open tube/ porous sparger � �

N2 to headspace �

pH 7.0

Addition of NaCO3 � � � �

CO2 to

headspace � � � �

The product IgG concentration was quantified by high-performance liquid

chromatography (HPLC) protein A method (Waters). The IgG concentrations in the

35

bioreactor and the harvest line – or the matrix, the reservoir and the harvest in CT

experiments - were analyzed to study the product retention by the filter. Reducing

sodium dodecyl sulfate-polyacylamide gel electrophoresis (SDS-PAGE) was performed

on products and glycan analysis was performed using capillary electrophoresis (CE)

method for some of the experiments to investigate the product quality.

The calculation of the cell specific perfusion rate (CSPR), cell specific

consumption/production rates of glucose (qglc), lactate (qlac), glutamine (qgln), ammonia

(qamm), cell growth rate (µ), cell specific productivity (qp) are given in table 9.

Table 9 Main process parameters

Parameter Equation

µ µ = 1 C!(dC!dt

+V!"##$V ∆T

C!)

qglc q!"# = 1 C!(D Glc!"#$%! − Glc −

dGlcdt

+V!"##$ Glc!"#$%

V ∆T)

qgln q!"# = 1 C!(D Gln!"#$%! − Gln −

dGlndt

+V!"##$ Gln!"#$%

V ∆T−r!"#$Gln)

qlac q!"# = 1 C!(D Lac +

dLacdt

)

qamm q!"" = 1 C!(D Amm − Amm!"#$%! +

dAmmdt

−r!"#$Gln)

qp q! = 1 C!(dP!"dt

+ D −V!"##$V

P!" +V!"##$

VP!")

Accumulated

production in

harvest

HT!"! = P!"V!"dt!

!

For abbreviations, see List of Abbreviations.

For the metabolomics study, UHPLC was performed on low molecular weight (<10

kDa) fractions of the culture media and each sample was analyzed by both positive and

36

negative ionization mass spectrometry to increase the range of ions detected. The

components in the elutes were then analyzed by an ESI-QTOF MS in both positive and

negative ion modes.

After synchronizing the data from different experiments using peak detection and

alignment, peaks found in all the samples were chosen and multivariate statistical

analyses were performed using Unscrambler-X (CAMO, Norway). PCA was used in

dimensionality reduction and visualization of the complex dataset. PLS regression

models were used to identify the correlation between LC-MS data and bioprocess

variables in cultures.

37

2.2.2 WAVE Bioreactor™ Cellbag prototype perfusion system

with ATF and TFF (Paper I)

Figure 7 shows part of the perfusion systems developed for the Cellbag prototype.

Perfusion system with ATF (Figure 7A) and TFF (Figure 7B) were established through

a series of experiments and process parameters were gradually established and

optimized. Before the perfusion cultivations listed in Table 7, water model ATF#1 was

carried out using the first version of Cellbag prototype to identify basic parameters

such as working volume, rocking speed and rocking angle etc. Supplement of antifoam

at 0/10/50 ppm were screened to determine working volume that eliminated bubble

occurrence in the dip tube, which was detrimental to both cells and filters. ATF#2 and

ATF#3 were two batch cultures to build fundamental understanding of the cells,

medium, and the system. Several fed-batch runs were also carried out in the same

experimental setting to gain knowledge of the system from comparative studies

(Paper IV).

Figure 7 Perfusion system development (part) for the Cellbag prototype

(B) (A)

38

Through a series of HCDC perfusion runs, various technical obstacles were identified

and addressed accordingly. Here take ATF system for example, more details can be

found in Paper I. In ATF system the ATF alternated flow stopped after CV reached 132

MVC/mL in ATF#15A. This CV was then confirmed by ATF#15B, where the ATF

function was interrupted at 123 MVC/mL again. By pressurizing the Cellbag with 0.02-

0.03 bar the ATF function was successfully restored and the culture was continued.

This clearly suggested the vacuum limit in present settings to pull the highly viscous

cell broth at high CV. Potential answers to that were: 1) lower the vacuum effect to <-

0.53 bar; 2) pressurize the bioreactor to facilitate the vacuum; 3) increase the total

lumen section area. These solutions were verified experimentally and thus contributed

to the perfusion system understanding. Furthermore several other bottlenecks were

overcome such as oxygen transfer limitation at extremely high cell densities and

rocking conditions suitable for ATF operations.

2.2.3 CellTank™ prototype perfusion system (paper II)

In the CT system, suspension cells were entrapped inside a non-woven polyester matrix

that was sealed in a cassette. The cassette was immersed in a 2 L tank (reservoir)

where the actual perfusion took place. Figure 8A shows the perfusion system

configuration and Figure 8B is a closer look of the system design and how the cells are

located in the fibrous matrix observed under microscope. Limited access to the cell

sample necessitated the implementation of an online biomass probe based on dielectric

spectroscopy. Unlike other cell immobilized or entrapped fibrous bed bioreactor

systems that always suffer from nutrients/metabolites gradient and gas transfer

heterogeneity, CellTank™ advantageously provides a very homogeneous environment.

Furthermore no product retention was observed.

39

Figure 8 Perfusion system for the CellTank™ prototype

The CellTank™ prototype design, stirrer table prototype, as well as the control system

were optimized during the perfusion system development.

(A)

(B)

40

2.2.4 Glass stirred tank bioreactor with ATF (Paper III)

HCDC perfusion system was developed in a bench top

glass stirred tank bioreactor using HF cartridge and

ATF as cell separation module. Figure 9 is a close look

at the connection between the bioreactor and HF

mounted on an ATF station. CSPR based perfusion

rate selection strategy was systematically studied at

steady states. A CSPR optimization strategy was

experimentally proposed and two CSPR adjustment

approaches, D variation and CV variation, were

performed and compared. The critical role of the

glucose concentration per cell was emphasized and

discussed (Section 2.3).

2.3 HCDC perfusion process development (Paper I - IV)

As described in Section 2.2, four HCDC perfusion systems were developed. The

maximal CV (CVMAX) supported by each system configuration was achieved in one of the

runs. The goal was to explore the system limitations in our perfusion process settings

and to overcome various challenges during this operation by optimizing industrial

prototype design as well as process parameters. Other than pushing the system limit by

reaching the CVMAX, steady states at 20-30 MVC/mL and at 100-130 MVC/mL were also

established and studied for process development. Steady state at 20-30 MVC/mL is

known as a highly suitable system to screen the process parameters152, 197 and we were

Figure 9 ST Perfusion system

development (part)

41

able to perform the same study at 100-130 MVC/mL to explore further the potential of

perfusion processes. Table 10 summarizes the experiments in various perfusion systems

for HCDC perfusion process development and more details are presented separately.

Table 10 Summary of the perfusion runs

Perfusion

system Exp. CVMAX

Steady state study at

20-30 MVC/mL

Steady state study at

100-130 MVC/mL

ATF

ATF#4 �

ATF#8 �

ATF#9 �

ATF#15A �

TFF TFF#6 �

TFF#10 � � �

CT

CT#1 �

CT#2 �

CT#3 �

ST

ST#1 � �

ST#2 � �

ST#3 � � �

2.3.1 Maximal viable cell density (Paper I-IV)

A summary of the CVMAX achieved in all the perfusion systems and some process details

are given in Table 11. The perfusion rate selection was based on a CSPR of 50-55

42

pL/cell/day, except in ST experiments where the CSPR effect was systematically

studied. Table 11 CVMAX achieved in the different perfusion systems

Perfusion system ATF TFF CT ST

Cell line K4 K4 DP-12 DP-12 CHOM

Experiment no. ATF#15A TFF#10 CT#1 ST#1 ST#3

Working volume (L) 4 4 150 mL 1 1

Culture duration (day) 11 44 27 37 30

CVMAX (MVC/mL) 132 214 ≥200 (200

pF/cm) 188 165

CtotMAX (106 cells/mL) 137 224 - 198 174

Cell viability (%) 96.3 - 99.2 85.1 - 98.3 87.4-97.9 91-99.3 92.9-98.2

ACD (µm) (a) 20-30

MVC/mL; (b) 100 MVC/mL;

(c) above 150 MVC/mL

16.4a; 17.3b 17.1a;

17.1b; 16.7c -

16.7a;

16.5b;

17.7c

18.5a;

17.3b;

17.9c

Average CSPR

(pL/cell/days) 50 55 50 varied varied

Maximal D (RV/day) 6 10 10 3.7 6.5

Average µ during perfusion

(day-1) (a) 20-30 MVC/mL;

(b) 100 MVC/mL; (c) above

150 MVC/mL

0.57a; 0.26b 0.35a;

0.28b; 0.15c

0.17a; 0.26b;

0.14c

0.54a;

0.26b;

0.11c

0.38a;

0.34b;

0.14c

Average qp during perfusion

(pg/c/d) (a) 20-30

MVC/mL; (b) 100 MVC/mL;

(c) above 150 MVC/mL

6.0a; 7.8b 9.6a; 16.5b;

13.6c

2.5a; 1.6b;

1.3c

2.8a;

1.5b; 0.9c

39.2a;

32.7b;

26.7c

pCO2 (kPa) 2.4 - 21.9 3.1 – 28.8 2.7-8.6 4.9-10.6 2.5-13.8

CtotMAX = maximal total cell density; ACD = average cell diameter; qp = cell specific IgG productivity

In ATF#15A, CV reached 132 MVC/mL but no higher value due to the vacuum power

limitation of the diaphragm pump to pull highly viscous cell broth from the bioreactor

into the HF. This maximum CV was later confirmed with a new HF in run ATF#15B

43

(Paper I). In a TFF system we were able to reach a higher CV of 214 MVC/mL, which

is the currently highest CHO cell density ever reported according to our knowledge.

This cell density is definitely approaching the theoretical limit of biovolume using cells

with am average cell diameter of 17 µm (both K4 and DP-12). However this high cell

density above 200 MVC/mL lasted only for 2 days before the system collapsed due to

high pressure in the recirculation loop caused by high viscosity and a very high pCO2 of

31 kPa that was deleterious for the cells (Paper I).

In the CT the cells were entrapped in the matrix and, therefore, it was not possible to

take cell samples for offline cell density analysis. An independent experiment was

carried out in Paper II to compare the online biomass probe analysis with offline

image-based cell density analysis of Bioprofile instrument. It was confirmed that the

biomass sensor reading of 1 pF/cm equaled the offline measurement of 1 MVC/mL up

to CV 160 MVC/mL, above which the online sensor gave lower cell densities and

saturated at around 203 pF/cm. Therefore we believe that CV above 200 MVC/mL was

reached in CT#1 when the online biomass sensor indicated 200 pF/cm (Paper II).

In ST experiments, high CV 188 MVC/mL was achieved in ST#1 at ultra low CSPR 26

pL/cell/day however HF fouling and clogging led to termination of the run. 165

MVC/mL was achieved in ST#3 using another cell line with slightly larger size, at

CSPR 34 pL/cell/day. Compared with the ATF runs performed in the disposable

Cellbag, the working volume was smaller relatively to the filter area, resulting in higher

CV (Paper III).

All the perfusion cultures had high viabilities despite the extremely high cell densities.

None of the runs were terminated due to cell death, but intrinsic limitation due to the

cell retention were encountered such as high viscosity and excessive bubble formation.

44

2.3.2 HCDC established and stabilized in the perfusion systems (Paper I-III)

Table 12 gives a summary of HCDC at steady states in all 4 perfusion systems.

Table 12 HCDC at steady states

Sys. Exp.no.

Days at

20-30

MVC/mL

Days at

100-130

MVC/mL

Approach to

maintain cell

density

CSPR

(pL/cell/

day)

Cell

viability

(%)

Purpose of study

ATF

ATF#4 4

Bleeding

80 92.6-98.9

Perfusion system

development,

Process consistency

and reproducibility

ATF#8 9 60 89-98

ATF#9 16 60 88.3-98

TFF TFF#6 16 60 88.5-98

TFF#10 14 19 60 85.1-98.3

CT CT#2 14

Hypothermia 60 90.6-97.1

CT#3 15 60 71-97.9

ST

ST#1 44

Bleeding

40/60/80 92.7-99.3 Perfusion system

development, CSPR

optimization

strategy

ST#2 12+8* 5 45/65/85 91-98.1

ST#3 11 9 45/65 92.4-98.2

* Cv was maintained at 15 MVC/day (12 days) and 35 MVC/days (8 days) respectively.

Cell arrest by hypothermia is known to improve the DP-12 cell productivity. In CT#2

we studied the possibility of arresting the cell growth and maintain the high CV at 100-

130 MVC/mL to explore the potential of operation at such high CV with elevated IgG

production. The culture temperature was gradually lowered from 37°C to 32°C, 31°C,

30°C, finally to 29°C which led to a complete cell growth arrest (Figure 10A). By doing

so we were able to maintain the culture at 100-130 pF/cm for 14 days and the viability

was ≥90% during the whole run. The cell specific productivity at hypothermia

increased by 47% compared to 37°C. CT#3 was a high CV production process based on

45

information collected from CT#1&2. The culture temperature was decreased directly to

31°C on day 13 when CV reached 100 MVC/mL (Figure 10B)C. After a stable

production phase at 100-130 pF/cm for 15 days, the temperature was set back to 37°C

and the cell growth successfully resumed.

Figure 10 Hypothermia applied to CellTank cultures for cell growth arrest

Packed bed bioreactor and fibrous bed bioreactor have intrinsic drawback of culture

heterogeneity as presented in Section 1.3.1, but the CellTank was able to overcome this

nutrients/metabolites gradient due to its very efficient recirculation system. The

successful implementation of hypothermia enables its application in fast product

manufacturing and the collected product is completely cell free, which alleviates the

downstream processing as well.

40

80

120

160

0 5 10 15 20

Cv (

MVC

/mL)

20

25

30

35

40

0 10 20 30

Tem

pera

ture

(˚C

)

Time (Days)

CT#2 CT#3

46

2.3.3 Perfusion rate optimization (Paper III)

CSPR based perfusion rate selection was applied in all the studies. In ATF, TFF and CT

runs, a constant CSPR 50-60 pL/cell/day was applied. It supported all the cultures very

well and extremely high cell densities were achieved. In ST runs this approach was

reviewed and investigated in details.

CSPR is calculated from D and CV (eq.1). This indicates that the regulation of CSPR can

be done via varying either of the two values and the outcome should be the same for a

given CSPR from a mathematical point of view. This assumption was investigated in

ST#1 by implementing the same CSPR using both approaches. According to discussion

in the previous section, a perfusion culture stable at 20-30 MVC/mL is a perfect tool to

study various process parameters. The idea was that starting from a steady state at 30

MVC/mL, the CSPR was varied by changing D at constant CV or changing CV at

constant D. Steady states were then established for each variation and the average cell

growth rate µ in each state was monitored to determine whether another variation was

needed (in case an improvement/deterioration was observed). The µ was used as the

optimization indicator because it was easily accessible and highly reliable. µ could be

replaced by another important process parameter such as qp to be used as the

optimization target.

As part of the results, Figure 11 shows the culture performances obtained for CSPR

increase by increasing D (from 1.5 RV/D) at constant CV of 30 MVC/mL or decreasing

CV (from 30 MVC/mL) at a constant D 1.5 RV/mL. Both approaches suggested the

same optimum CSPR 60 pL/cell/day for cell growth or culture maintenance while

minimizing the medium consumption. The highest qp was observed at 60 pL/cell/day

when increasing D, but at 80 pL/cell/day when decreasing Cv. As mentioned above, if

qp had been selected as the reference parameter instead of µ, the CV would have been

47

lowered further attempting to reach even higher qp. The optimum of 60 pL/cell/day

CSPR should however be put in regards to the increased medium cost brought by the

relatively high resulting perfusion rate.

Figure 11 CSPR variation study by (A) increasing D at constant CV of 30 MVC/mL;

(B) decreasing CV at constant D = 1.5 RV/day

0

1

2

3

D (

RV/d

ay)

0

0.2

0.4

0.6

µ (d

ay-1

)

2

4

6

0 10 20 30

q p (

pg/c

/d)

Time (days)

0

20

40

60

80

100

CSPR

(pL

/cel

l/da

y)

(A)

I II III IV

0

10

20

30

40Cv

(M

VC/m

L)

2

4

6

8

0 5 10 15 20

q p (

pg/c

/d)

Time (days)

0.1

0.2

0.3

0.4

µ (d

ay-1

)

0

20

40

60

80

100

CSPR

(pL

/cel

l/da

y) (B)

I II III

48

It is worth noticing that these two approaches do not deliver identical effect in terms of

productivity and cell metabolism (Paper III). A potential explanation to this

observation is the glucose metabolism switch between TCA cycle and anaerobic

glycolysis, determined by the instantaneous glucose concentration available per cell,

and not by the glucose mass available per cell, which relates only to the stoichiometric

consumption. The same calculated CSPR from both D and CV variation by eq. 1 gives

the same stoichiometric glucose availability (mass per cell), not the same glucose

concentration per cell. Based on this we suggest that glucose concentration per cell is

used to analyze the influence of a component.

2.3.4 Product quality at HCDC (Paper II, III)

As the cell productivity is highly cell line specific, here we take a closer look at DP-12

cells because they have been used both in CT and ST runs. The average qp of DP-12

cells in a 7-day fed-batch process is 1.5 pg/c/d and the average qp at different cell

densities in perfusion cultures CT#1 and ST#1 are presented in Table 11. At 20-30

MVC/mL the average qp was 67%-87% higher in perfusion processes and the average

qp at 100 MVC/mL was almost identical as in fed-batch cultures, where the peak cell

density hardly reaches 5 MVC/mL in the same medium. From this, one can easily

figure out the volumetric productivity benefit that HCDC perfusion process provides,

not to mention the additional 47% increased qp when hypothermia was applied as

discussed in section 2.3.2.

At the same time, product quality is certainly as important as productivity. As

presented in the introduction, perfusion process generally provides a stable

environment to have a more consistent product. Despite that it still raises concern for

product quality at very high CV due to limited publishing of such information.

49

Reduced SDS-PAGE was performed in CT experiments to study the product quality in

HCDC (Paper II). The product was very consistent in CT#1 for all the CV's, even when

CV reached 200 x106 cells/mL. However fragment variant of the light chain appeared in

CT#2 after hypothermia was applied. It was postulated that the light chain variation

appeared as a result of different glycosylation pattern, potentially due to lower

temperature.

An industrial high producer CHOM cell was used in ST runs and N-glycosylation

related product quality profiles were analyzed (Paper III). Figure 12 shows the main

glycan of G0F, G1F, G2F and Man-5 with corrected peak area in percentage. ”Other”

refers to other structures such as afucosylated G0, G1, etc. Only the steady states are

shown for better visualization. In ST#2, the cell density was maintained at 15, 35 and

100 MVC/mL at CSPR 45, 65 and 80 pL/cell/day respectively (Figure 12A). G0F, G1F,

G2F and Man-5 were very consistent at CV 15 and 35 MVC/mL at both CSPR 45 and 65

pL/cell/day, but obvious decrease of G1F and G2F and increase of Man-5 and G0F

could be seen at 92 MVC/mL under 80 pL/cell/day CSPR. This could be due to the

elevated concentration of lactate with this CSPR. A different glycosylation profile has

been also observed by the company Selexis in fed-batch cultivations with high lactate

concentrations.

The same glycan analysis was applied to ST#3, where the CV was maintained at 20 and

100 MVC/mL and a much lower CSPR of 45 pL/cell/day was applied at 100 MVC/mL

(Fig.12B). After the CSPR variation was studied at 100 MVC/mL the cell bleeding was

stopped and the CV was maximized at low CSPR. It is very clear that the glycan pattern

was very consistent at different CV, indicating that the process was very robust with no

product quality shift from beginning to the end at these different CV's. As this

commercial cell line is a cell line for the manufacturing of a biosimilar, it is very

important that the product quality closely resemble the originator drug.

50

Figure 12 Glycosylation profiles of CHOM cells. (A) 5 steady states in ST#2: CV 15MVC/mL (i&ii), CV 35

MVC/mL (iii&iv), at CSPR 45 (i&iii) & 65 (ii&iv) pL/cell/day respectively. CV 92 MVC/mL at CSPR 85

pL/cell day (v). (B) ST#3, where CV was 15, 20, 88, 106, 86, 131, 165 MVC/mL at low CSPRs.

20

40

60

80

100

-40

0

40

80

120

160

5 10 15 20 25 30 35

Viab

ility

(%

)

Cv (

MVC

/mL)

CS

PR (

pL/c

ell/

day)

Cultivation time (days)

20

40

60

80

100

-40

-20

0

20

40

60

80

100

Viab

ility

(%

)

Cv (

MVC

/mL)

CS

PR (

pL/c

ell/

day)

i

ii

iii

iv v

b

(A)

(B)

51

2.3.5 Cellular metabolism in HCDC 2.3.5.1 Energy metabolism

A CSPR around 50 pL/cell/day was applied in all the perfusion runs other than the ST

experiments where the CSPR strategy was investigated. In all the runs the cell specific

uptake rates of glucose and glutamine, as well as cell specific production rates of

lactate and ammonia were calculated and monitored as part of the daily routine. The

medium used contained 25 mM pre-mixed glucose so it was impossible to lower the

glucose concentration in the feeding. Here we take DP-12 cells for example because

they have been used in different systems. In CT experiments, the maximum residual

lactate and ammonia concentrations were 86 mM and 8 mM respectively (CT#2). Due

to the high and probably inhibiting concentrations of both byproducts, CT#3 was

performed with a low substrate feeding strategy that led to moderate production of

lactate (40 mM) and ammonia (5 mM) with very low residual levels of glucose and

glutamine. In ST, by applying lower CSPR's, the residual glucose and glutamine in the

cultures were maintained at 1-2 mM and ≈ 0 respectively. The lactate was mainly

maintained below 10 mM with peaks at 21.9 mM at high CSPR during CSPR variation

study. The ammonia concentration was comparable to the other experiments with a

maximum value of 4.9 mM in ST#1.

Our results are consistent with previous reports describing that at low glucose

concentrations, the cells consume glucose more efficiently and utilize mainly the TCA

cycle pathway. We have also demonstrated that qglc was linear to CSPR independently

of the cell density, and that the cells unnecessarily consumed more glucose when more

glucose was fed, leading to lactate accumulation in the culture. This speaks against the

alternative method to tune the perfusion rate based on the residual glucose level in the

culture in the development of a perfusion process.

52

2.3.5.2 Metabolomics profiling of extracellular metabolites (Paper IV)

An exometabolome study, i.e. metabolomics of the extra-cellular components in the

culture, was performed in perfusion culture TFF#10 (Figure 13A) where the CV

reached 214 MVC/mL in TFF system, in comparison to two fed-batch runs (Figure 13B)

FB#11 (37°C) and FB#16 (37°C à 35.5°C on day 7). The cell growths in both process

modes are shown in Figure 13. Both fed-batch cultures at different temperatures were

compared to HCDC perfusion process, especially at extremely high cell densities.

Figure 13 Growth profiles of TFF#10 and two fed-batch runs, illustrating the samples selected for

metabolomics profiling study. (A) Sample events (red), at different cultivation stages in TFF#10

perfusion run and cell density (blue). (B) Sample events in fed-batch runs FB#11 (37°C) and FB#16

(37°C à 35.5°C on day 7).

Figure 14 shows the PCA score plots for the perfusion run TFF#10 (red) and for both

fed-batch runs (blue). Figure 14A is a score plot where the model has been calibrated

and validated with the data from the three runs. Figure 14B is a score plot showing a

model calibrated with the data from the fed-batch runs only, and then used for the

predictions of the perfusion run and fresh medium. Figure 14A clearly exhibits that fed-

batch samples are mostly defined by PC1 while perfusion samples are mostly defined

by PC2 with much less spreading than fed-batch samples. This indicates that the

0 5 10 150

4

8

12

16

20

Cv (

MVC

/mL)

FB#11

FB#16

50

100

150

200

250

0 10 20 30 40

Cv (

MVC

/mL)

Time (days)

(A) (B)

53

extracellular metabolites profile was more constant in the perfusion run despite the

huge difference in cell density and the much longer cultivation time. In Figure 14B it is

also very interesting to observe that the data from the perfusion run process cluster

with fresh medium samples and are close to fed-batch at day 3 to 5. It suggests that

throughout the whole perfusion process regardless of the cell density, the extracellular

metabolites might be growth-related similarly to day 3 to 5 in a fed-batch process.

Figure 14 PCA score plot for TFF#10 and FB#11&16. (A) A model calibrated and validated with the

data from three runs. (B) A model calibrated with the data from the fed-batch runs only.

!""#$%&

'(&

)*+&,-&

)*+&-&

)*+&.&)*+&$/& "0#$$&1&"0#$2&

*&

34567&859:*&

!"#$%&'#()*&

(*+&,-(*+&.&

/&

(A)

(B)

54

In perfusion mode, almost all the detected metabolites were highly consistent

irrespective of the cell density due to the fact that the same CSPR had been applied

throughout the culture. However we identified some potential extracellular biomarkers

that varied with the cell density. They were all involved in glutathione metabolism and

potentially imply an escalation of oxidative stress at cell density above 110 MVC/mL

even though a constant CSPR was applied (Paper IV). This hypothesis needs further

investigation in the future.

55

3 Concluding remarks HCDC perfusion process is very attractive due to the high volumetric productivity it can

offer, which allows much smaller production footprints with improved productivity and

product quality. However outcome can be uncertain and the process development can

be complex due to many unknowns at high cell densities. This thesis is about HCDC

perfusion process development and it is divided into two main parts.

The first part of the work presents four HCDC perfusion systems developed with our

industrial collaborators: WAVE Bioreactor™ Cellbag prototype coupled with ATF and

TFF respectively, a non-woven polyester matrix based CellTank™ prototype, and a

bench top stirred tank bioreactor equipped with ATF. In the HCDC perfusion system

development, industrial prototypes were optimized and perfusion processes were

gradually established.

The second part of the work focuses on HCDC perfusion process characterization and

optimization. As part of the perfusion system development, maximal viable cell

densities were achieved in all the systems to identify the capacity of the different

system configurations to support HCDCs. In WAVE Bioreactor™ coupled with TFF and

ATF, the CVMAX was 214 MVC/mL and 132 MVC/mL respectively. 214 MVC/mL is

definitely approaching the higher physical limit of the possible CV where the cells are

compressed against each other with no liquid between cells131. In the ATF equipped

system, the achieved cell density was lower due to the limitation of vacuum pump to

draw the highly viscous cell broth at such high cell density. This was later alleviated by

a filter size larger relatively to the culture volume, and the extremely high CV of 188

MVC/mL was achieved in the stirred tank bioreactor equipped with ATF. In the

CellTank™ system the maximum CV was 200 pF/cm measured from online biomass

56

probe based on dielectroscopy principle. This has been verified to be at least 200

MVC/mL during the study156.

During HCDC perfusion process development, steady states at 20-30 MVC/mL and 100-

130 MVC/mL were established and maintained by cell bleeding or hypothermia. These

were used to screen the key process variables, as well as to verify the process

consistency and reproducibility.

As one of the key process variables in perfusion processes, perfusion rate is important

not only for the culture performances but also for the process cost. CSPR based

perfusion rate selection has been widely used because it ensures a constant

environment to the cells, and it has been successfully applied to all our perfusion

processes to support the HCDCs. Two approaches to vary the CSPR via CV and D do not

deliver the same effect despite the identical value of CSPR. This is probably due to the

glucose concentration per the cell, resulting in different glucose consumption kinetics

despite similar glucose mass available per cell.

Note that a medium containing 25 mM glucose was used in all our experiments.

Balanced feeding of glucose and amino acids is very important regarding the cell

growth, productivity and glycosylation related product quality. The byproducts

accumulation observed in all the runs is related to sub-optimally balanced feed, which

is addressed in this thesis by perfusion rate variation. Medium development, outside

the focus of the current thesis, would definitely be helpful as part of the process

development, and associated to the perfusion rate strategy. We foresee that our studies

will add value to the development of perfusion technology in recombinant protein

production in mammalian cell systems.

57

Acknowledgements Without the support and guidance that I received from many people it would not have

been possible to do this PhD. I would like to thank all the people who have helped,

listened, encouraged, and pushed me throughout this study.

Foremost, I would like to express my deepest gratitude to my supervisor Véronique

Chotteau for her continuous support of my PhD study and research, for her

immeasurable guidance, advice and patience. I would not have imaged having a better

mentor and advisor for my PhD study.

I would like to thank my supervisor, Vincent Bulone for regular meetings to keep me

on the right track, and valuable advices.

My sincere thanks to all members of the Biotechnology School at Albanova, especially

the division of industrial biotechnology, both past and present, for support, discussion

and a nice work environment.

Thank you to VINNOVA for financial support.

I would like to acknowledge all collaborators for the hard work. Thank you to GE

Healthcare for sponsoring the Wave Bioreactor™ Cellbag project. Special thanks to Eva

Lindskog, Kieron Walsh, Craig Robinson and Eric Fäld from GE Healthcare. Thank you

to PerfuseCell for sponsoring the CellTank bioreactor. Special thanks to Per Stobbe and

Jacob Hiob Thilo for being always so supportive J. Thank you to Belach Biotechnology

for their support in the CellTank project and equipment for Stirred Tank project.

Special thanks to Christian Orrego Silvander. Thank you to David Sergeant from

Ipratech for lending us the CytoSys process control system and helping with all

troubleshooting. Thank you to Selexis for lending us the CHOM cell line and Pierre-

58

Alain Girod for our nice collaboration. Special thanks to Alexandra Martiné from

Selexis for glycosylation analysis.

Thank you Gen Larsson, for all best suggestions and discussions about research,

education and much more beyond that.

Thank you Andres Veide, as my internal advance reviewer, much more as a mentor for

bioprocess. Doesn’t matter if it is upstream or downstream, E. coli or CHO cells, it’s the

process that matters J

Thank you Atefeh Shokri and Carolina Åstrand for pushing me J, and for valuable

suggestions of course.

I would like to thank all colleagues and lab mates, both past and present, not only for

the support and inspiration, but also for your kindness and friendship. Especially Johan

Norén, Lubna Al-Khalili, Erika Hagrot, Marie-Françoise Clincke, Carin Mölleryd, Lena

Thoring, Caijuan Zhan, Mattias Leino, Karin Gillner, Fredrik Rönnmark, Jenny

Rönnmark, Hubert Schwarz, Nils Arnold Brechmann, Liang Zhang, Martin Gustavsson,

Gustav Sjöberg, Johan Jarmander, Magnus Lindroos, Antonius Van Maris, Gunaratna

Kuttuva Rajarao, Mónica Guevara-Martínez, Mariel Pérez-Zabaleta, Caroline

Bramstång, Berndt Björlenius, Eric Björkvall, Kaj Kauko, Anna Klara Lindgren. Special

thanks to David Hörnström, who was always supportive and agreed to revise the

Swedish abstract at the last minute. I am very grateful to all of you and it is my great

honor to be among the bioprocess family.

I would like to thank my family, especially MaMa, BaBa, Chao, and my parents-in-law

who have been by my side through this PhD, for your guidance, encouragement and

support. It is your loves that give me power, confidence and the most precious thing,

time. Finally to darling Tiana for being such an angel baby that makes it possible for

me to complete what I started. Happy 1st birthday!

59

References 1. Walsh, G., Biopharmaceutical benchmarks 2014.Naturebiotechnology2014, 32,

(10),992-1000.

2. Wurm, F. M., Production of recombinant protein therapeutics in cultivated

mammaliancells.Naturebiotechnology2004,22,(11),1393-8.

3. Walsh, G., Biopharmaceutical benchmarks 2010.Naturebiotechnology2010, 28,

(9),917-24.

4. KarthikP.Jayapal,K.F.W.,Wei-ShouHu,MirandaG.S.Yap,RecombinantProtein

Therapeutics from Cho Cells - 20 Years and Counting. CHO Consortium: SBE Special

Edition2007,40-47.

5. Muchmore, E. A.; Milewski, M.; Varki, A.; Diaz, S., Biosynthesis of N-

GlycolyneuraminicAcid-thePrimarySiteofHydroxylationofN-AcetylneuraminicAcidIs

theCytosolicSugarNucleotidePool.JBiolChem1989,264,(34),20216-20223.

6. Sato,J.D.;Kawamoto,T.;McClure,D.B.;Sato,G.H.,CholesterolrequirementofNS-

1mousemyelomacellsforgrowthinserum-freemedium.MolBiolMed1984,2,(2),121-

34.

7. Chu, L.; Robinson, D. K., Industrial choices for protein production by large-scale

cellculture.Currentopinioninbiotechnology2001,12,(2),180-7.

8. McCamish,M.;Woollett, G.,Worldwide experiencewith biosimilar development.

Mabs2011,3,(2),209-217.

9. Birch,J.R.;Racher,A.J.,Antibodyproduction.AdvDrugDelivRev2006,58,(5-6),

671-85.

10. Backliwal, G.;Hildinger,M.; Chenuet, S.;Wulhfard, S.;De Jesus,M.;Wurm, F.M.,

Rational vector design and multi-pathway modulation of HEK 293E cells yield

recombinantantibodytitersexceeding1g/lbytransient transfectionunderserum-free

conditions.NucleicAcidsRes2008,36,(15),e96.

60

11. Durocher, Y.; Perret, S.; Kamen, A., High-level and high-throughput recombinant

proteinproductionby transient transfectionof suspension-growinghuman293-EBNA1

cells.NucleicAcidsRes2002,30,(2),E9.

12. Geisse, S., Reflections onmore than 10 years of TGE approaches.ProteinExpres

Purif2009,64,(2),99-107.

13. Ye,J.X.;Alvin,K.;Latif,H.;Hsu,A.;Parikh,V.;Whitmer,T.;Tellers,M.;Edmonds,M.

C. D.; Ly, J.; Salmon, P.; Markusen, J. F., Rapid Protein Production Using CHO Stable

TransfectionPools.BiotechnolProgr2010,26,(5),1431-1437.

14. Hatton,D.;Forrest-Owen,W.;Dean,G.;Gibson,S.;Crook,T.;Lunney,A.;Ruddock,

S.;Davis,A.;Daramola,L.;Field,R.,High-yieldingCHOcellpoolsforrapidproductionof

recombinantantibodies.InCellsandCulture,Springer:2010;pp239-244.

15. Froud, S. J., Cell Bank Preparation and Characterization. In Animal Cell

Biotechnology:MethodsandProtocols, Jenkins,N., Ed.HumanaPress:Totowa,NJ, 1999;

pp99-115.

16. Warburg,O.,Ontheoriginofcancercells.Science1956,123,(3191),309-14.

17. Europa,A.F.;Gambhir,A.;Fu,P.C.;Hu,W.S.,Multiplesteadystateswithdistinct

cellular metabolism in continuous culture of mammalian cells. Biotechnology and

bioengineering2000,67,(1),25-34.

18. Zhou,W.;Rehm,J.;Europa,A.;Hu,W.-S.,Alterationofmammaliancellmetabolism

bydynamicnutrientfeeding.Cytotechnology1997,24,(2),99-108.

19. Cairns,R.A.;Harris,I.S.;Mak,T.W.,Regulationofcancercellmetabolism.NatRev

Cancer2011,11,(2),85-95.

20. Ronald Zielke, H.; Ozand, P. T.; Tyson Tildon, J.; Sevdalian, D. A.; Cornblath, M.,

Reciprocal regulation of glucose and glutamine utilization by cultured human diploid

fibroblasts.JournalofCellularPhysiology1978,95,(1),41-48.

21. Fitzpatrick, L.; Jenkins,H. A.; Butler,M., Glucose and glutaminemetabolism of a

murineB-lymphocytehybridomagrowninbatchculture.ApplBiochemBiotechnol1993,

43,(2),93-116.

61

22. Jenkins,H.A.;Butler,M.;Dickson,A.J.,Characterizationofglutaminemetabolism

intworelatedmurinehybridomas.Journalofbiotechnology1992,23,(2),167-82.

23. Ljunggren, J.; Haggstrom, L., Glutamine limited fed-batch culture reduces the

overflowmetabolismofaminoacidsinmyelomacells.Cytotechnology1992,8,(1),45-56.

24. Ozturk,S.S.;Palsson,B.O.,Growth,metabolic,andantibodyproductionkineticsof

hybridomacellculture:1.Analysisofdatafromcontrolledbatchreactors.Biotechnology

progress1991,7,(6),471-80.

25. Ozturk, S. S.; Riley, M. R.; Palsson, B. O., Effects of ammonia and lactate on

hybridoma growth, metabolism, and antibody production. Biotechnology and

bioengineering1992,39,(4),418-31.

26. Schneider, M.; Marison, I. W.; von Stockar, U., The importance of ammonia in

mammaliancellculture.Journalofbiotechnology1996,46,(3),161-85.

27. Xing, Z.; Li, Z.; Chow, V.; Lee, S. S., Identifying inhibitory threshold values of

repressing metabolites in CHO cell culture using multivariate analysis methods.

Biotechnologyprogress2008,24,(3),675-83.

28. Yang, M.; Butler, M., Effects of ammonia on CHO cell growth, erythropoietin

production,andglycosylation.Biotechnologyandbioengineering2000,68,(4),370-80.

29. Glacken, M. W.; Fleischaker, R. J.; Sinskey, A. J., Reduction of waste product

excretionvianutrientcontrol:Possiblestrategiesformaximizingproductandcellyields

onseruminculturesofmammaliancells.Biotechnologyandbioengineering1986,28,(9),

1376-89.

30. Wlaschin,K.F.;Nissom,P.M.;GattiMde,L.;Ong,P.F.;Arleen,S.;Tan,K.S.;Rink,A.;

Cham, B.; Wong, K.; Yap, M.; Hu, W. S., EST sequencing for gene discovery in Chinese

hamsterovarycells.Biotechnologyandbioengineering2005,91,(5),592-606.

31. Bebbington,C.R.;Renner,G.;Thomson,S.;King,D.;Abrams,D.;Yarranton,G.T.,

High-Level Expression of a Recombinant Antibody from Myeloma Cells Using a

Glutamine-SynthetaseGene as anAmplifiable SelectableMarker.Bio-Technol1992, 10,

(2),169-175.

62

32. Bell, S. L.; Bebbington, C.; Scott, M. F.;Wardell, J. N.; Spier, R. E.; Bushell, M. E.;

Sanders,P.G.,Geneticengineeringofhybridomaglutaminemetabolism.EnzymeMicrob

Technol1995,17,(2),98-106.

33. Birch,J.R.;Boraston,R.C.;Metcalfe,H.;Brown,M.E.;Bebbington,C.R.;Field,R.P.,

Selecting and designing cell lines for improved physiological characteristics.

Cytotechnology1994,15,(1-3),11-6.

34. Browne,S.M.;Al‐Rubeai,M.,AnalysisofanartificiallyselectedGS‐NS0variant

withincreasedresistancetoapoptosis.BiotechnolBioeng2011,108,(4),880-892.

35. Fan,L.;Kadura,I.;Krebs,L.E.;Hatfield,C.C.;Shaw,M.M.;Frye,C.C.,Improvingthe

efficiency of CHO cell line generation using glutamine synthetase gene knockout cells.

BiotechnolBioeng2012,109,(4),1007-1015.

36. Rendall, M.; Maxwell, A.; Tatham, D.; Khan, P.; Gay, R.; Kallmeier, R.; Wayte, J.;

Racher,A., Transfection tomanufacturing: reducing timelines for high yieldingGS-CHO

processes.AnimalCellTechnologyMeetsGenomics,Springer,Dordrecht2005,701-704.

37. Smales,C.M.;Dinnis,D.M.;Stansfield,S.H.;Alete,D.;Sage,E.;Birch,J.R.;Racher,

A. J.; Marshall, C. T.; James, D. C., Comparative proteomic analysis of GS‐NS0murine

myeloma cell lines with varying recombinant monoclonal antibody production rate.

BiotechnolBioeng2004,88,(4),474-488.

38. Anderson,M.E.,Glutathione:anoverviewofbiosynthesisandmodulation.Chem

BiolInteract1998,111-112,1-14.

39. Bray,T.M.;Taylor,C.G.,Tissueglutathione,nutrition,andoxidativestress.CanJ

PhysiolPharmacol1993,71,(9),746-51.

40. Lu,S.C.,Glutathionesynthesis.BiochimBiophysActa2013,1830,(5),3143-53.

41. Bailey, J. E.; Sburlati, A.; Hatzimanikatis, V.; Lee, K.; Renner, W. A.; Tsai, P. S.,

Inverse metabolic engineering: a strategy for directed genetic engineering of useful

phenotypes.Biotechnologyandbioengineering2002,79,(5),568-79.

42. Dinnis, D. M.; Stansfield, S. H.; Schlatter, S.; Smales, C. M.; Alete, D.; Birch, J. R.;

Racher,A.J.;Marshall,C.T.;Nielsen,L.K.; James,D.C.,Functionalproteomicanalysisof

63

GS-NS0 murine myeloma cell lines with varying recombinant monoclonal antibody

productionrate.Biotechnologyandbioengineering2006,94,(5),830-41.

43. Korke,R.;Gatti,M.D.;Lau,A.L.Y.;Lim,J.W.E.;Seow,T.K.;Chung,M.C.M.;Hu,W.

S.,Largescalegeneexpressionprofilingofmetabolicshiftofmammaliancellsinculture.J

Biotechnol2004,107,(1),1-17.

44. Korke,R.;Rink,A.;Seow,T.K.;Chung,M.C.;Beattie,C.W.;Hu,W.S.,Genomicand

proteomicperspectivesincellcultureengineering.Journalofbiotechnology2002,94,(1),

73-92.

45. Seow, T. K.; Korke, R.; Liang, R. C.M. Y.; Ong, S. E.; Ou, K.;Wong, K.; Hu,W. S.;

Chung, M. C. M., Proteomic investigation of metabolic shift in mammalian cell culture.

BiotechnolProgr2001,17,(6),1137-1144.

46. Seth,G.;Philp,R.J.;Lau,A.;Jiun,K.Y.;Yap,M.;Hu,W.S.,Molecularportraitofhigh

productivityinrecombinantNS0cells.BiotechnolBioeng2007,97,(4),933-951.

47. Smales, C. M.; Dinnis, D. M.; Stansfield, S. H.; Alete, D.; Sage, E. A.; Birch, J. R.;

Racher, A. J.; Marshall, C. T.; James, D. C., Comparative proteomic analysis of GS-NS0

murinemyeloma cell lineswith varying recombinantmonoclonal antibody production

rate.Biotechnologyandbioengineering2004,88,(4),474-88.

48. Stansfield, S.H.; Allen, E. E.; Dinnis, D.M.; Racher, A. J.; Birch, J. R.; James,D. C.,

DynamicanalysisofGS-NS0cellsproducingarecombinantmonoclonalantibodyduring

fed-batchculture.Biotechnologyandbioengineering2007,97,(2),410-24.

49. Tian,Q.;Stepaniants,S.B.;Mao,M.;Weng,L.;Feetham,M.C.;Doyle,M.J.;Yi,E.C.;

Dai,H.;Thorsson,V.;Eng,J.;Goodlett,D.;Berger,J.P.;Gunter,B.;Linseley,P.S.;Stoughton,

R. B.; Aebersold, R.; Collins, S. J.; Hanlon, W. A.; Hood, L. E., Integrated genomic and

proteomicanalysesofgeneexpressioninMammaliancells.MolCellProteomics2004,3,

(10),960-9.

50. Wlaschin, K. F.; Seth, G.; Hu, W. S., Toward genomic cell culture engineering.

Cytotechnology2006,50,(1-3),121-140.

64

51. Ahn, W. S.; Antoniewicz, M. R., Parallel labeling experiments with [1,2-

(13)C]glucose and [U-(13)C]glutamine provide new insights into CHO cellmetabolism.

MetabEng2013,15,34-47.

52. Carinhas,N.;Duarte,T.M.;Barreiro,L.C.;Carrondo,M.J.;Alves,P.M.;Teixeira,A.

P., Metabolic signatures of GS-CHO cell clones associatedwith butyrate treatment and

culturephasetransition.Biotechnologyandbioengineering2013,110,(12),3244-57.

53. Chong,W.P.;Reddy,S.G.;Yusufi,F.N.;Lee,D.Y.;Wong,N.S.;Heng,C.K.;Yap,M.

G.; Ho, Y. S., Metabolomics-driven approach for the improvement of Chinese hamster

ovary cell growth: overexpression ofmalate dehydrogenase II. Journalofbiotechnology

2010,147,(2),116-21.

54. Chong,W.P.;Yusufi,F.N.;Lee,D.Y.;Reddy,S.G.;Wong,N.S.;Heng,C.K.;Yap,M.

G.; Ho, Y. S., Metabolomics-based identification of apoptosis-inducing metabolites in

recombinantfed-batchCHOculturemedia.Journalofbiotechnology2011,151,(2),218-

24.

55. Dietmair,S.;Hodson,M.P.;Quek,L.E.;Timmins,N.E.;Chrysanthopoulos,P.;Jacob,

S. S.; Gray, P.; Nielsen, L. K., Metabolite profiling of CHO cells with different growth

characteristics.Biotechnologyandbioengineering2012,109,(6),1404-14.

56. Sellick, C. A.; Croxford, A. S.; Maqsood, A. R.; Stephens, G.; Westerhoff, H. V.;

Goodacre, R.; Dickson, A. J., Metabolite profiling of recombinant CHO cells: designing

tailored feeding regimes that enhance recombinant antibodyproduction.Biotechnology

andbioengineering2011,108,(12),3025-31.

57. Sellick, C. A.; Croxford, A. S.; Maqsood, A. R.; Stephens, G. M.;Westerhoff, H. V.;

Goodacre, R.; Dickson, A. J., Metabolite profiling of CHO cells: Molecular reflections of

bioprocessingeffectiveness.BiotechnolJ2015,10,(9),1434-45.

58. Selvarasu,S.;Ho,Y.S.;Chong,W.P.;Wong,N.S.;Yusufi,F.N.;Lee,Y.Y.;Yap,M.G.;

Lee, D. Y., Combined in silicomodeling andmetabolomics analysis to characterize fed-

batchCHOcellculture.Biotechnologyandbioengineering2012,109,(6),1415-29.

65

59. Sheikholeslami,Z.;Jolicoeur,M.;Henry,O.,Probingthemetabolismofaninducible

mammalian expression system using extracellular isotopomer analysis. Journal of

biotechnology2013,164,(4),469-78.

60. Templeton, N.; Dean, J.; Reddy, P.; Young, J. D., Peak antibody production is

associatedwithincreasedoxidativemetabolisminanindustriallyrelevantfed-batchCHO

cellculture.BiotechnolBioeng2013,110,(7),2013-+.

61. Čuperlović-Culf,M.,MetabolomicsinAnimalCellCulture.InAnimalCellCulture,Al-

Rubeai,M.,Ed.SpringerInternationalPublishing:Cham,2015;pp615-646.

62. Bajad,S.;Shulaev,V.,LC-MS-BasedMetabolomics. InMetabolicProfiling:Methods

andProtocols,Metz,T.O.,Ed.HumanaPress:Totowa,NJ,2011;pp213-228.

63. Idborg, H.; Zamani, L.; Edlund, P. O.; Schuppe-Koistinen, I.; Jacobsson, S. P.,

MetabolicfingerprintingofraturinebyLC/MSPart1.Analysisbyhydrophilicinteraction

liquidchromatography-electrosprayionizationmassspectrometry.JChromatogrBAnalyt

TechnolBiomedLifeSci2005,828,(1-2),9-13.

64. Idborg-Bjorkman, H.; Edlund, P. O.; Kvalheim, O. M.; Schuppe-Koistinen, I.;

Jacobsson,S.P.,ScreeningofbiomarkersinraturineusingLC/electrosprayionization-MS

andtwo-waydataanalysis.AnalChem2003,75,(18),4784-4792.

65. Chong,W.P.;Thng,S.H.;Hiu,A.P.;Lee,D.Y.;Chan,E.C.;Ho,Y.S.,LC-MS-based

metabolic characterization of high monoclonal antibody-producing Chinese hamster

ovarycells.Biotechnologyandbioengineering2012,109,(12),3103-11.

66. Devantier,R.;Scheithauer,B.;Villas-Boas,S.G.;Pedersen,S.;Olsson,L.,Metabolite

profiling for analysis of yeast stress response during very high gravity ethanol

fermentations.BiotechnolBioeng2005,90,(6),703-714.

67. Jackson,J.E.,Auser'sguidetoprincipalcomponents.JohnWiley&Sons:2005;Vol.

587.

68. Wold,S.;Sjöström,M.;Eriksson,L.,PLS-regression:abasictoolofchemometrics.

Chemometricsandintelligentlaboratorysystems2001,58,(2),109-130.

66

69. Mehra,S.;Chandrawanshi,V.;Prashad,K.,AdvancedBioprocessEngineering:Fed-

BatchandPerfusionProcesses.InAppliedBioengineering,Wiley-VCHVerlagGmbH&Co.

KGaA:2017;pp417-468.

70. Castilho, L.; Moraes, A.; Augusto, E.; Butler, M., Animal cell technology: from

biopharmaceuticalstogenetherapy.GarlandScience:2008.

71. Griffiths, J. B., Animal-Cell Culture Processes - Batch or Continuous. JBiotechnol

1992,22,(1-2),21-30.

72. Ozturk, S.; Hu, W.-S., Cell culture technology for pharmaceutical and cell-based

therapies.CRCPress:2005.

73. Huang, Y.-M.; Hu, W.; Rustandi, E.; Chang, K.; Yusuf-Makagiansar, H.; Ryll, T.,

Maximizing productivity of CHO cell-based fed-batch culture using chemically defined

mediaconditionsand typicalmanufacturingequipment.BiotechnolProgr2010,26, (5),

1400-1410.

74. Kelley, B., Industrialization of mAb production technology The bioprocessing

industryatacrossroads.Mabs2009,1,(5),443-452.

75. Langer, E. S., Trends in Perfusion Bioreactors.BioProcess International2011, 9,

(10),5.

76. Tao, Y.; Shih, J.; Sinacore, M.; Ryll, T.; Yusuf-Makagiansar, H., Development and

implementation of a perfusion-based high cell density cell banking process.Biotechnol

Progr2011,27,(3),824-829.

77. Clincke, M. F.; Molleryd, C.; Samani, P. K.; Lindskog, E.; Faldt, E.; Walsh, K.;

Chotteau,V.,VeryhighdensityofChinesehamsterovarycellsinperfusionbyalternating

tangential flowortangential flowfiltrationinWAVEBioreactor-partII:Applicationsfor

antibodyproductionandcryopreservation.Biotechnologyprogress2013,29,(3),768-77.

78. Heidemann,R.;Mered,M.;Wang,D.Q.;Gardner,B.;Zhang,C.;Michaels,J.;Henzler,

H.-J.; Abbas, N.; Konstantinov, K., A new seed-train expansionmethod for recombinant

mammaliancelllines.Cytotechnology2002,38,(1-3),99-108.

67

79. Pollock, J.; Ho, S. V.; Farid, S. S., Fed-batch and perfusion culture processes:

Economic, environmental, and operational feasibility under uncertainty. Biotechnol

Bioeng2013,110,(1),206-219.

80. Hernandez, R., Continuous manufacturing: a changing processing paradigm.

BioPharmInternational2015,28,(4).

81. Konstantinov,K.B.;Cooney,C.L.,Whitepaperoncontinuousbioprocessing.May

20-21,2014ContinuousManufacturingSymposium.JPharmSci2015,104,(3),813-20.

82. Langer,E.S.;Rader,R.A.,Continuousbioprocessingandperfusion:wideradoption

comingasbioprocessingmatures.BioProcessingJournal2014,13,43-9.

83. Godawat,R.;Konstantinov,K.;Rohani,M.;Warikoo,V.,End-to-endintegratedfully

continuous production of recombinant monoclonal antibodies. J Biotechnol2015, 213,

13-19.

84. Kamarck,M.E.,Buildingbiomanufacturingcapacity[mdash]thechapterandverse.

NatBiotech2006,24,(5),503-505.

85. Warikoo,V.;Godawat,R.;Brower,K.;Jain,S.;Cummings,D.;Simons,E.;Johnson,T.;

Walther, J.; Yu, M.;Wright, B.; McLarty, J.; Karey, K. P.; Hwang, C.; Zhou,W.; Riske, F.;

Konstantinov,K.,Integratedcontinuousproductionofrecombinanttherapeuticproteins.

BiotechnolBioeng2012,109,(12),3018-3029.

86. Walther, J.; Godawat, R.; Hwang, C.; Abe, Y.; Sinclair, A.; Konstantinov, K., The

businessimpactofanintegratedcontinuousbiomanufacturingplatformforrecombinant

proteinproduction.Journalofbiotechnology2015,213,3-12.

87. Voisard, D.; Meuwly, F.; Ruffieux, P. A.; Baer, G.; Kadouri, A., Potential of cell

retention techniques for large-scale high-density perfusion culture of suspended

mammaliancells.Biotechnologyandbioengineering2003,82,(7),751-65.

88. Lee, J. C.; Chang,H.N.; Oh,D. J., Recombinant antibody production by perfusion

culturesofrCHOcellsinadepthfilterperfusionsystem.Biotechnologyprogress2005,21,

(1),134-9.

89. Mitsuda, S.; Matsuda, Y.; Kobayashi, N.; Suzuki, A.; Itagaki, Y.; Kumazawa, E.;

Higashio,K.;Kawanishi, G., ContinuousProductionofTissuePlasminogen-Activator (T-

68

Pa) by Human Embryonic Lung Diploid Fibroblast, Imr-90 Cells, Using a Ceramic Bed

Reactor.Cytotechnology1991,6,(1),23-31.

90. Chen, C.; Huang, Y. L.; Yang, S. T., A fibrous-bed bioreactor for continuous

production of developmental endothelial locus-1 by osteosarcoma cells. Journal of

biotechnology2002,97,(1),23-39.

91. Bohak,Z.;Kadouri,A.; Sussman,M.V.;Feldman,A.F.,NovelAnchorageMatrices

forSuspension-CultureofMammalian-Cells.Biopolymers1987,26,S205-S213.

92. Looby,D.;Griffiths, J.B.,Fixedbedporousglasssphere(porosphere)bioreactors

foranimalcells.Cytotechnology1988,1,(4),339-46.

93. Knazek,R.A.;Gullino,P.M.;Kohler,P.O.;Dedrick,R.L.,Cell cultureonartificial

capillaries:anapproachtotissuegrowthinvitro.Science1972,178,(4056),65-6.

94. delaBroise,D.;Noiseux,M.;Massie,B.;Lemieux,R.,Hybridomaperfusionsystems:

acomparisonstudy.Biotechnologyandbioengineering1992,40,(1),25-32.

95. Piret, J. M.; Cooney, C. L., Mammalian-Cell and Protein Distributions in

UltrafiltrationHollowFiberBioreactors.BiotechnolBioeng1990,36,(9),902-910.

96. Piret, J. M.; Devens, D. A.; Cooney, C. L., Nutrient and Metabolite Gradients in

Mammalian-CellHollowFiberBioreactors.CanJChemEng1991,69,(2),421-428.

97. Kaufman,J.B.;Wang,G.;Zhang,W.;Valle,M.A.;Shiloach,J.,Continuousproduction

andrecoveryofrecombinantCa2+bindingreceptorfromHEK293cellsusingperfusion

throughapackedbedbioreactor.Cytotechnology2000,33,(1-3),3-11.

98. Meuwly, F.; von Stockar, U.; Kadouri, A., Optimization of themedium perfusion

rateinapacked-bedbioreactorchargedwithCHOcells.Cytotechnology2004,46,(1),37-

47.

99. Wang,G.;Zhang,W.; Jacklin,C.;Freedman,D.;Eppstein,L.;Kadouri,A.,Modified

CelliGen-packedbedbioreactorsforhybridomacellcultures.Cytotechnology1992,9,(1-

3),41-9.

100. Donato,D.; Labate,G. F.D.;Debbaut,C.; Segers,P.; Catapano,G.,Optimizationof

constructperfusion in radial-flowpacked-bedbioreactors for tissueengineeringwitha

2Dstationaryfluiddynamicmodel.BiochemEngJ2016,109,197-211.

69

101. Johnson, M.; Lanthier, S.; Massie, B.; Lefebvre, G.; Kamen, A. A., Use of the

Centritechlabcentrifugeforperfusioncultureofhybridomacellsinprotein-freemedium.

BiotechnolProgr1996,12,(6),855-864.

102. Kim, B. J.; Chang, H. N.; Oh, D. J., Application of a cell-once-through perfusion

strategy for production of recombinant antibody from rCHO cells in a centritech lab II

centrifugesystem.BiotechnolProgr2007,23,(5),1186-1197.

103. Kim, B. J.; Oh, D. J.; Chang, H. N., Limited use of Centritech Lab II Centrifuge in

perfusion culture of rCHO cells for theproductionof recombinant antibody.Biotechnol

Progr2008,24,(1),166-174.

104. Batt, B. C.; Davis, R. H.; Kompala, D. S., Inclined sedimentation for selective

retention of viable hybridomas in a continuous suspension bioreactor. Biotechnology

progress1990,6,(6),458-64.

105. Choo, C. Y.; Tian, Y.; Kim, W. S.; Blatter, E.; Conary, J.; Brady, C. P., High-level

productionofamonoclonalantibodyinmurinemyelomacellsbyperfusioncultureusing

agravitysettler.Biotechnologyprogress2007,23,(1),225-31.

106. Lipscomb, M. L.; Mowry, M. C.; Kompala, D. S., Production of a secreted

glycoproteinfromaninduciblepromotersysteminaperfusionbioreactor.Biotechnology

progress2004,20,(5),1402-7.

107. Nivitchanyong, T.; Martinez, A.; Ishaque, A.; Murphy, J. E.; Konstantinov, K.;

Betenbaugh, M. J.; Thrift, J., Anti-apoptotic genes Aven and E1B-19K enhance

performanceofBHKcellsengineeredtoexpressrecombinantfactorVIIIinbatchandlow

perfusioncellculture.Biotechnologyandbioengineering2007,98,(4),825-41.

108. Pohlscheidt,M.;Jacobs,M.;Wolf,S.;Thiele,J.;Jockwer,A.;Gabelsberger,J.;Jenzsch,

M.;Tebbe,H.;Burg,J.,Optimizingcapacityutilizationbylargescale3000Lperfusionin

seedtrainbioreactors.BiotechnolProgr2013,29,(1),222-229.

109. Searles,J.A.;Todd,P.;Kompala,D.S.,Viablecellrecyclewithaninclinedsettlerin

the perfusion culture of suspended recombinant Chinese hamster ovary cells.

Biotechnologyprogress1994,10,(2),198-206.

70

110. Coakley, W. T., Ultrasonic separations in analytical biotechnology. Trends

Biotechnol1997,15,(12),506-11.

111. Coakley,W. T.;Whitworth, G.; Grundy,M. A.; Gould, R. K.; Allman, R., Ultrasonic

manipulationofparticlesandcells.Ultrasonicseparationofcells.Bioseparation1994,4,

(2),73-83.

112. Dalm, M. C.; Jansen, M.; Keijzer, T. M.; van Grunsven, W. M.; Oudshoorn, A.;

Tramper, J.; Martens, D. E., Stable hybridoma cultivation in a pilot-scale acoustic

perfusion system: long-term process performance and effect of recirculation rate.

Biotechnologyandbioengineering2005,91,(7),894-900.

113. Dowd, J. E.; Kwok, K. E.; Piret, J. M., Glucose-based optimization of CHO-cell

perfusioncultures.Biotechnologyandbioengineering2001,75,(2),252-6.

114. Gaida,T.;Doblhoff-Dier,O.; Strutzenberger,K.;Katinger,H.;Burger,W.;Groschl,

M.; Handl, B.; Benes, E., Selective retention of viable cells in ultrasonic resonance field

devices.Biotechnologyprogress1996,12,(1),73-6.

115. Gorenflo,V.M.;Angepat,S.;Bowen,B.D.;Piret, J.M.,Optimizationofanacoustic

cellfilterwithanovelair-backflushsystem.Biotechnologyprogress2003,19,(1),30-6.

116. Gorenflo, V. M.; Smith, L.; Dedinsky, B.; Persson, B.; Piret, J. M., Scale-up and

optimization of an acoustic filter for 200 L/day perfusion of a CHO cell culture.

Biotechnologyandbioengineering2002,80,(4),438-44.

117. Ryll,T.;Dutina,G.;Reyes,A.;Gunson,J.;Krummen,L.;Etcheverry,T.,Performance

of small-scale CHO perfusion cultures using an acoustic cell filtration device for cell

retention: characterization of separation efficiency and impact of perfusion onproduct

quality.Biotechnologyandbioengineering2000,69,(4),440-9.

118. Trampler,F.;Sonderhoff,S.A.;Pui,P.W.;Kilburn,D.G.;Piret, J.M.,Acousticcell

filterforhighdensityperfusioncultureofhybridomacells.Biotechnology(NY)1994,12,

(3),281-4.

119. Elsayed,E.;Piehl,G.-W.;Nothnagel, J.;Medronho,R.;Deckwer,W.-D.;Wagner,R.,

Use of hydrocyclone as an efficient tool for cell retention in perfusion cultures.Animal

CellTechnologyMeetsGenomics2005,679-682.

71

120. Elsayed,E.A.;Wagner,R.InApplicationofhydrocyclonesforcontinuouscultivation

of SP-2/0 cells in perfusion bioreactors: Effect of hydrocyclone operating pressure, BMC

proceedings,2011;Springer:2011;pP65.

121. Jockwer,A.;Medronho,R.A.;Wagner,R.;Anspach,F.;Deckwer,W.-D.,Theuseof

hydrocyclones for mammalian cell retention in perfusion bioreactors. In Animal Cell

Technology:FromTargettoMarket,Springer:2001;pp301-306.

122. Docoslis,A.;Kalogerakis,N.;Behie,L.A.;Kaler,K.V.,Anoveldielectrophoresis‐

based device for the selective retention of viable cells in cell culturemedia.Biotechnol

Bioeng1997,54,(3),239-250.

123. Castilho, L. R.; Anspach, F. B.; Deckwer, W. D., An integrated process for

mammalian cell perfusion cultivation and product purification using a dynamic filter.

Biotechnologyprogress2002,18,(4),776-81.

124. Tang, Y. J.; Ohashi, R.; Hamel, J. F. P., Perfusion culture of hybridoma cells for

hyperproductionofIgG(2a)monoclonalantibodyinawavebioreactor-perfusionculture

system.BiotechnolProgr2007,23,(1),255-264.

125. Tao, Y.; Shih, J.; Sinacore, M.; Ryll, T.; Yusuf-Makagiansar, H., Development and

implementationofaperfusion-basedhighcelldensitycellbankingprocess.Biotechnology

progress2011,27,(3),824-9.

126. Wang,L.;Hu,H.;Yang,J.;Wang,F.;Kaisermayer,C.;Zhou,P.,Highyieldofhuman

monoclonal antibody produced by stably transfected Drosophila schneider 2 cells in

perfusioncultureusingwavebioreactor.Molecularbiotechnology2012,52,(2),170-9.

127. Yuk, I. H.; Baskar, D.; Duffy, P. H.; Hsiung, J.; Leung, S.; Lin, A. A., Overcoming

ChallengesinWAVEBioreactorswithoutFeedbackControlsforpHandDissolvedOxygen.

BiotechnolProgr2011,27,(5),1397-1406.

128. Wang,L.L.;Hu,H.X.;Yang,J. J.;Wang,F.;Kaisermayer,C.;Zhou,P.,HighYieldof

Human Monoclonal Antibody Produced by Stably Transfected Drosophila Schneider 2

CellsinPerfusionCultureUsingWaveBioreactor.Molecularbiotechnology2012,52,(2),

170-179.

72

129. Banik, G. G.; Heath, C. A., Hybridoma Growth and Antibody-Production as a

FunctionofCell-DensityandSpecificGrowth-RateinPerfusionCulture.BiotechnolBioeng

1995,48,(3),289-300.

130. Blasey,H.D.;Jäger,V.,StrategiestoIncreasetheEfficiencyofMembraneAerated

andPerfusedAnimalCellBioreactorsbyanImprovedMediumPerfusion.InAnimalCell

Culture and Production of Biologicals: Proceedings of the Third Annual Meeting of the

Japanese Association for Animal Cell Technology, held in Kyoto, December 11–13, 1990,

Sasaki,R.;Ikura,K.,Eds.SpringerNetherlands:Dordrecht,1991;pp61-73.

131. Clincke,M. F.;Molleryd, C.; Zhang, Y.; Lindskog, E.;Walsh, K.; Chotteau, V., Very

highdensityofCHOcellsinperfusionbyATForTFFinWAVEbioreactor.PartI.Effectof

thecelldensityontheprocess.Biotechnologyprogress2013,29,(3),754-67.

132. Cortin, V.; Thibault, J.; Jaeob, D.; Garnier, A., High-titer adenovirus vector

productionin293Scellperfusionculture.BiotechnolProgr2004,20,(3),858-863.

133. Genzel,Y.;Vogel,T.;Buck,J.;Behrendt,I.;Ramirez,D.V.;Schiedner,G.;Jordan,I.;

Reichl,U.,Highcelldensitycultivationsbyalternatingtangentialflow(ATF)perfusionfor

influenzaAvirusproductionusingsuspensioncells.Vaccine2014,32,(24),2770-2781.

134. Deo,Y.M.;Mahadevan,M.D.;Fuchs,R.,Practicalconsiderationsinoperationand

scale-up of spin-filter based bioreactors for monoclonal antibody production.

Biotechnologyprogress1996,12,(1),57-64.

135. Werner, A.; Lutkemeyer, D.; Poggendorf, I.; Lehmann, J.; Muthing, J., Serum-free

production of a chimeric E-selectin-IgG protein from 1 to 100 l scale: Repeated batch

cultivationversuscontinuousspinfilterperfusion.Cytotechnology2002,38,(1-3),47-56.

136. Mercille, S.; Johnson, M.; Lanthier, S.; Kamen, A. A.; Massie, B., Understanding

factors that limit the productivity of suspension-based perfusion cultures operated at

highmediumrenewalrates.BiotechnolBioeng2000,67,(4),435-450.

137. Mercille, S.; Massie, B., Apoptosis-resistant E1B-19K-expressing NS/0 myeloma

cells exhibit increased viability and chimeric antibody productivity under perfusion

cultureconditions.BiotechnolBioeng1999,63,(5),529-543.

73

138. Dong,H.;Tang,Y.J.;Ohashi,R.;Hamel,J.F.P.,Aperfusionculturesystemusinga

stirredceramicmembranereactorforhyperproductionofIgG(2a)monoclonalantibody

byhybridomacells.BiotechnolProgr2005,21,(1),140-147.

139. Li,F.;Vijayasankaran,N.;Shen,A.Y.;Kiss,R.;Amanullah,A.,Cellcultureprocesses

formonoclonalantibodyproduction.Mabs2010,2,(5),466-79.

140. Ozturk, S. S., Engineering challenges in high density cell culture systems.

Cytotechnology1996,22,(1-3),3-16.

141. Deschenes,J.S.;Desbiens,A.;Perrier,M.;Kamen,A.,Useofcellbleedinahighcell

density perfusion culture and multivariable control of biomass and metabolite

concentrations.Asia-PacJChemEng2006,1,(1-2),82-91.

142. Fenge, C.; Fraune, E.; Schiigerl, K., Perfusionbioreactor performance at different

cellbleedrates.AnimalCellTechnology1992,365-372.

143. Kuystermans,D.;Al-Rubeai,M.,Bioreactor systems forproducing antibody from

mammaliancells.InAntibodyExpressionandProduction,Springer:2011;pp25-52.

144. Sherman,M.;Lam,V.;Carpio,M.;Hutchinson,N.;Fenge,C.,ContinuousCellCulture

Operationat2,000-LScale.BioProcessInt2016,14,(7).

145. Wright,B.;Bruninghaus,M.;Vrabel,M.;Walther, J.; Shah,N.;Bae,S.; Johnson,T.;

Yin, J.; Zhou, W.; Konstantinov, K., A novel seed-train process: using high-density cell

banking, a disposable bioreactor, and perfusion technologies. BioProcess International

2015,13.

146. Ahn,W.S.;Jeon,J.J.;Jeong,Y.R.;Lee,S.J.;Yoon,S.K.,Effectofculturetemperature

on erythropoietin production and glycosylation in a perfusion culture of recombinant

CHOcells.Biotechnologyandbioengineering2008,101,(6),1234-44.

147. Chen,Z.L.;Wu,B.C.;Liu,H.;Liu,X.M.;Huang,P.T.,Temperatureshiftasaprocess

optimizationstepfortheproductionofpro-urokinasebyarecombinantChinesehamster

ovarycell lineinhigh-densityperfusionculture.Journalofbioscienceandbioengineering

2004,97,(4),239-43.

74

148. Ducommun,P.;Ruffieux,P.;Kadouri,A.;vonStockar,U.;Marison,I.W.,Monitoring

oftemperatureeffectsonanimalcellmetabolisminapackedbedprocess.Biotechnology

andbioengineering2002,77,(7),838-42.

149. Rodriguez,J.;Spearman,M.;Tharmalingam,T.;Sunley,K.;Lodewyks,C.;Huzel,N.;

Butler, M., High productivity of human recombinant beta-interferon from a low-

temperatureperfusionculture.Journalofbiotechnology2010,150,(4),509-18.

150. Tsai, Y. S.; Yoon, S. J.; Chuppa, S.; Konstantinov, K.; Naveh, D., Fermentor

temperatureasatoolforcontrolofhigh-densityperfusionculturesofmammaliancells.

AbstrPapAmChemS1996,211,128-BIOT.

151. Sunley, K.; Butler, M., Strategies for the enhancement of recombinant protein

production frommammalian cells by growth arrest.Biotechnology advances2010, 28,

(3),385-94.

152. Konstantinov, K.; Goudar, C.; Ng, M.; Meneses, R.; Thrift, J.; Chuppa, S.;

Matanguihan,C.;Michaels, J.;Naveh,D.,The“push-to-low”approachforoptimizationof

high-density perfusion cultures of animal cells. In Cell Culture Engineering, Springer:

2006;pp75-98.

153. Oh,D. J.;Choi,S.K.;Chang,H.N.,High-densitycontinuousculturesofhybridoma

cells inadepth filterperfusionsystem.Biotechnologyandbioengineering1994,44, (8),

895-901.

154. Deutschmann,S.M.;Jager,V.,OptimizationofthegrowthconditionsofSf21insect

cells for high-density perfusion culture in stirred-tank bioreactors. Enzyme Microb

Technol1994,16,(6),506-12.

155. Xu, S.; Chen,H.,High-densitymammalian cell cultures in stirred-tankbioreactor

withoutexternalpHcontrol.Journalofbiotechnology2016,231,149-59.

156. Zhang,Y.;Stobbe,P.;Silvander,C.O.;Chotteau,V.,Veryhighcelldensityperfusion

ofCHOcellsanchored inanon-wovenmatrix-basedbioreactor. JBiotechnol2015,213,

28-41.

157. Ryll,T.;Dutina,G.;Reyes,A.;Gunson,J.;Krummen,L.;Etcheverry,T.,Performance

of small-scale CHO perfusion cultures using an acoustic cell filtration device for cell

75

retention:Characterizationof separationefficiencyand impactofperfusiononproduct

quality.BiotechnolBioeng2000,69,(4),440-449.

158. Oh,D. J.;Chang,H.N.,HighDensityCultureofHybridomaCells inaDualHollow

FiberBioreactor.BiotechnolTech1992,6,(1),77-82.

159. Zhu,H.;Yang,S.T.,Long-termContinuousProductionofMonoclonalAntibodyby

HybridomaCellsImmobilizedinaFibrous-BedBioreactor.Cytotechnology2004,44,(1-

2),1-14.

160. Kyung, Y. S.; Peshwa, M. V.; Gryte, D. M.; Hu, W. S., High density culture of

mammalian cells with dynamic perfusion based on on-line oxygen uptake rate

measurements.Cytotechnology1994,14,(3),183-90.

161. Eibl,R.;Kaiser,S.;Lombriser,R.;Eibl,D.,Disposablebioreactors:thecurrentstate-

of-the-art and recommended applications in biotechnology. Applied microbiology and

biotechnology2010,86,(1),41-9.

162. Imseng,N.;Steiger,N.;Frasson,D.;Sievers,M.;Tappe,A.;Greller,G.;Eibl,D.;Eibl,

R.,Single-usewave-mixedversusstirredbioreactors for insect-cell/BEVS-basedprotein

expressionatbenchtopscale.EngLifeSci2014,14,(3),264-271.

163. RodriguezR,C.J.,GiraudS,Demonstratedperformanceofadisposablebioreactor

withananchorage-dependentcellline.BioProcessInternational2010,8,4.

164. Kaiser, S. C.; Eibl, R.; Eibl, D., Engineering characteristics of a single-use stirred

bioreactoratbench-scale:TheMobiusCellReady3Lbioreactorasacasestudy.EngLife

Sci2011,11,(4),359-368.

165. Gossain, V.; Mirro, R., Linear scale-up of cell cultures. BioProcess International

2010.

166. Valasek, C.; Cole, J.;Hensel, F.; Ye, P.; Conner,M.A.;Ultee,M.E., Production and

purification of a PER. C6-expressed IgM antibody therapeutic. BioProcess Int2011, 9,

(11),28-37.

167. Minow, B.; Rogge, P.; Thompson, K., Implementing a fully disposable MAb

manufacturingfacility.BioProcessInt2012,10,(6),48-57.

76

168. Noack,U.;DeWilde,D.;Verhoeye,F.;Balbirnie,E.;Kahlert,W.;Adams,T.;Greller,

G.;Reif,O.-W.,Single-UseStirredTankReactorBIOSTATCultiBagSTR:Characterization

andApplications.InSingle-UseTechnologyinBiopharmaceuticalManufacture,JohnWiley

&Sons,Inc.:2010;pp225-240.

169. Kauling,J.;Brod,H.;Jenne,M.;Waldhelm,A.;Langer,U.;Bodeker,B.,Novel,Rotary

Oscillated,ScalableSingle-UseBioreactorTechnologyfortheCultivationofAnimalCells.

Chem-Ing-Tech2013,85,(1-2),127-135.

170. Clincke,M. F.;Molleryd, C.; Zhang, Y.; Lindskog, E.;Walsh, K.; Chotteau, V., Very

highdensityofCHOcellsinperfusionbyATForTFFinWAVEbioreactor.PartI.Effectof

thecelldensityontheprocess.BiotechnolProgr2013,29,(3),754-767.

171. Hundt,B.;Best,C.;Schlawin,N.;Kassner,H.;Genzel,Y.;Reichl,U.,Establishmentof

aminkenteritisvaccineproductionprocessinstirred-tankreactorandWaveBioreactor

microcarrierculturein1-10Lscale.Vaccine2007,25,(20),3987-95.

172. Slivac, I.; Srcek, V. G.; Radosevic, K.; Kmetic, I.; Kniewald, Z., Aujeszky's disease

virusproductionindisposablebioreactor.JBiosciences2006,31,(3),363-368.

173. Weber,W.;Weber,E.;Geisse,S.;Memmert,K.,Optimisationofproteinexpression

and establishment of the Wave Bioreactor for Baculovirus/insect cell culture.

Cytotechnology2002,38,(1-2),77-85.

174. Zijlstra, G. M.; Oosterhuis, N., Cultivation of PER.C6® Cells in the Novel CELL-

Tainer™High-PerformanceDisposableBioreactor.InCellsandCulture:Proceedingsofthe

20th ESACT Meeting, Dresden, Germany, June 17-20, 2007, Noll, T., Ed. Springer

Netherlands:Dordrecht,2010;pp807-808.

175. Cadwell, J. J., New developments in hollow-fiber cell culture. AMERICAN

BIOTECHNOLOGYLABORATORY2004,22,14-19.

176. Lee,B.;Fang,D.;Croughan,M.;Carrondo,M.;Paik,S.-H.,Characterizationofnovel

pneumatic mixing for single-use bioreactor application.BMCProceedings2011, 5, (8),

O12.

77

177. Taylor,I.,TheCellMakerPlus™single-usebioreactor:anewbioreactorcapableof

culturing bacteria, yeast, insect and mammalian cells. Biotechnica, Hannover, Germany

2007.

178. Singh, V., Disposable bioreactor for cell culture using wave-induced agitation.

Cytotechnology1999,30,(1-3),149-58.

179. Hami, L. S.; Green, C.; Leshinsky, N.; Markham, E.; Miller, K.; Craig, S., GMP

productionandtestingofXcelleratedTCells(TM)forthetreatmentofpatientswithCLL.

Cytotherapy2004,6,(6),554-562.

180. Hollyman,D.;Stefanski,J.;Przybylowski,M.;Bartido,S.;Borquez-Ojeda,O.;Taylor,

C.;Yeh,R.;Capacio,V.;Olszewska,M.;Hosey, J.,Manufacturingvalidationofbiologically

functionalTcellstargetedtoCD19antigenforautologousadoptivecelltherapy.Journal

ofimmunotherapy(Hagerstown,Md.:1997)2009,32,(2),169.

181. Birch,J.R.;Racher,A.J.,Antibodyproduction.AdvDrugDeliverRev2006,58,(5-

6),671-685.

182. Chartrain, M.; Chu, L., Development and Production of Commercial Therapeutic

Monoclonal Antibodies in Mammalian Cell Expression Systems: An Overview of the

CurrentUpstreamTechnologies.CurrPharmBiotechno2008,9,(6),447-467.

183. DeWilde,D.;Noack, U.; Kahlert,W.; Barbaroux,M.; Greller, G., Bridging the gap

fromreusabletosingle-usemanufacturingwithstirred,single-usebioreactors.BioProcess

Int2009,7,(4).

184. Ozturk, S., Comparison of product quality: disposable and stainless steel

bioreactor.BioProduction,Berlin2007.

185. Valentine,P. In Implementationofasingle-usestirredbioreactoratpilotandGMP

manufacturing scale formammalian cell culture, ESACT 2009 Meeting, Dublin, Ireland,

2009;2009.

186. Zambaux, J., How synergy answers the biotech industry needs. BioProduction,

Berlin2007.

78

187. Butler, M.; Meneses-Acosta, A., Recent advances in technology supporting

biopharmaceutical production from mammalian cells. Applied microbiology and

biotechnology2012,96,(4),885-94.

188. Shukla, A. A.; Thommes, J., Recent advances in large-scale production of

monoclonalantibodiesandrelatedproteins.TrendsBiotechnol2010,28,(5),253-61.

189. Dowd,J.E.;Jubb,A.;Kwok,K.E.;Piret,J.M.,Optimizationandcontrolofperfusion

culturesusingaviablecellprobeandcellspecificperfusionrates.Cytotechnology2003,

42,(1),35-45.

190. Ozturk, S. S., Engineering challenges in high density cell culture systems.

Cytotechnology1996,22,(1),3-16.

191. Dowd,J.E.;Kwok,E.K.;Piret,J.M.,Off-lineglucose-basedcontrolandoptimization

ofperfusionCHOculturefeedrates.AbstrPapAmChemS2000,219,U181-U181.

192. Dowd, J. E.; Kwok, K. E.; Piret, J. M., Glucose-based optimization of CHO-cell

perfusioncultures.BiotechnolBioeng2001,75,(2),252-256.

193. Link,T.;Backstrom,M.;Graham,R.;Essers,R.;Zorner,K.;Gatgens, J.;Burchell, J.;

Taylor-Papadimitriou, J.; Hansson, G. C.; Noll, T., Bioprocess development for the

productionofarecombinantMUC1fusionproteinexpressedbyCHO-K1cellsinprotein-

freemedium.JBiotechnol2004,110,(1),51-62.

194. Wang,M.D.;Yang,M.;Huzel,N.;Butler,M.,ErythropoietinproductionfromCHO

cellsgrownbycontinuouscultureinafluidized-bedbioreactor.BiotechnolBioeng2002,

77,(2),194-203.

195. Galvez, J.;Lecina,M.;Sola,C.;Cairo, J. J.;Godia,F.,OptimizationofHEK-293Scell

cultures for the production of adenoviral vectors in bioreactors using on-line OUR

measurements.Journalofbiotechnology2012,157,(1),214-22.

196. Wlaschin, K. F.; Hu, W. S., Fedbatch culture and dynamic nutrient feeding. Adv

BiochemEngBiotechnol2006,101,43-74.

197. Chotteau,V.,ContinuousBiomanufacturing:CurrentPracticeandFuturePotential.

RefineTechnology,27-39.

79

198. Angepat, S.; Gorenflo, V. M.; Piret, J. M., Accelerating perfusion process

optimization by scanning non-steady-state responses.BiotechnolBioeng2005, 92, (4),

472-478.