Process Development for Cell-Based Influenza Vaccine

Downstream Processing
Process Development for
Cell-Based Influenza Vaccine
Influenza virus vaccines have traditionally been produced
by infection of fertilized hen eggs. this labor-intensive
approach requires large facilities, which has led to the
development of large-scale mammalian cell culture
methods for future virus vaccine production processes.
the main focus of these studies was to use generic,
established, and scalable techniques for industrial production
of live influenza vaccines (Figure 1) and potentially also for
other viruses independent of type or size. development of
three different focus areas in live influenza processing are
described: optimization of a disposable cell culture step; a
downstream purification process for efficient removal of
genomic dNa (gdNa) from live influenza virus cultured in
Madine-darby Canine Kidney (MdCK) cells; and development of
a fast, accurate live influenza assay.
Ce l l Cu lt u r e De v e l o p m e n t
Wave Bioreactor systems were originally designed for culturing
cells in suspension. However, several cell types used for cell
therapy and vaccine production are anchorage-dependent and
require a surface to grow on. in this study, Cytodex
microcarriers were used in a Wave Bioreactor to expand
MdCK and vero cells. different mixing conditions were tested
for both cell lines in order to optimize cell attachment.
We found that composition of the cell culture medium had a
significant effect on cell attachment; for example, MdCK and vero
cells cultured in low serum-containing medium attached to the
microcarriers with intermittent and continuous mixing. However,
under serum-free conditions, high attachment of vero cells could
be obtained only by using intermittent mixing. MdCK cells were
grown on Cytodex 1 and Cytodex 3 microcarriers (Figure 2) in a
Wave Bioreactor system at working volumes of 2 l and 10 l. the
cells reached a maximum concentration of 3 × 10 6 cells/ml. vero
cells were cultivated on Cytodex 1 and Cytodex 3 carriers at a
working volume of 2 l and reached a maximum concentration of
1.4 × 10 6 cells/ml — the same concentration of vero cells was
obtained with experiments using spinner flasks.
the data show that the Wave Bioreactor system is a versatile
platform for microcarrier cultivation of MdCK and vero cells,
and it is also a fast and convenient alternative to conventional
stirred-tank bioreactors or roller-bottle systems.
Do w n s t r e a m p r o C e s s D e v e l o p m e n t
the method of vaccine administration determines the
requirements for residual dNa in downstream process
development. the WHo recommendation for orally administered
vaccines is 100 µg dNa/dose (1 dose is assumed to be 45 µg
hemagglutinin), whereas parenteral vaccines allow only
Table 1: Hemagglutinin, DNA, and protein yield, with recovery of viable
virus (TCID 50 ) for A/Solomon Islands influenza virus after purification on
Capto ViralQ; a dose is assumed to be 45 µg.
HA Yield
(Biacore T100)
DNA/Dose
(qPCR)
Protein/Dose
(Bradford) TCID 50 /mL
19% 22.5 ng 52.5 µg 9.5
Table 2: Comparison of Biacore biosensor assay with SRID for analysis of
influenza virus hemagglutinin
Analytical Performance Biacore SRID
Standard curve range 0.5–10 µg/ml 8–30 µg/ml
Sensitivity:
limit of detection (lod)
limit of Quantitation (loQ)
0.3 µg/ml
0.8 µg/ml
6 µg/ml
13 µg/ml
Precision 97% 18%
time for 100 Samples:
Hands-on time
total
1–2 hours
15–16 hours
6–8 hours
20–22 hours
10 ng dNa/dose. the recommendation for protein is
100 µg/strain (300 µg/dose). to achieve these criteria, a
downstream method to remove host-cell gdNa from human
influenza virus based on ultrafiltration chromatography using
Capto viralQ medium and sterile filtration was developed (Figure
2). three seasonal 2007/2008 influenza virus strains —
a/Solomon islands/ 3/2006 (H1N1), a/Wisconsin/67/2005 (H3N2),
and B/Malaysia/2506/2004 — were tested in downstream
process development, with only minor process adjustments.
table 1 shows process results with strain a/Solomon islands.
the dNa reduction was 3.2 log over the Capto viralQ step. the
remaining dNa from the process was roughly twice the WHo
limit for parenteral vaccines, although the requirement for
protein content per dose was achieved. Further optimization,
addition of another purification step, or the use of benzonase,
would be needed to achieve the regulatory level of dNa/dose.
Generally, the hemagglutinin (Ha) recovery variability ranged
from 15 to 60% between runs and strains. loss of measurable Ha
during the initial microfiltration and ultrafiltration is the main
bottleneck of the process, but no loss of Ha in the permeate was
detected during ultrafiltration. the recovery of retained viable
virus was determined by tCid 50 to 12%. tCid 50 measurement of
live virus suffers from very high variability, however, and
recoveries can range from 20%.
74 BioProcess International In d u s t r y ye a r b o o k 2010 - 2011 advertorial

Figure 1: Overview of a cell-based influenza vaccine process showing
reduction of total DNA content (purple)
HiPrep 26/10
Cell culture
MDCK cells on Cytodex 3 carriers
116.4 × 10 6 ng DNA
Clari�cation
ULTA Prime GF 2 µm -
ULTA Prime GF 0.6 µm
39.3 × 10 6 ng DNA
Ultra�ltration/Dia�ltration
Cross�ow �ltration, UFP-500-C
9.50 × 10 6 ng DNA
Nonbinding Anion-Exchange
Chromatography
Capto ViralQ �ltration
5.06 × 10 3 ng DNA
Ultra�ltration/Dia�ltration
Cross�ow �ltration, UFP-500-C
3.99 × 10 3 ng DNA
Sterile �ltration
ULTA Pure SF 0.2 µm
2.06 × 10 3 ng DNA
In f l u e n z a an a ly s I s De v e l o p m e n t
the potency of influenza vaccines are mainly determined by
quantitation of Ha using the single radial immunodiffusion (Srid)
assay. this method, although approved by both Fda and eMea, is
labor-intensive and suffers from low precision and sensitivity.
Biacore biosensor assays offer an alternative to Srid for
vaccine development and manufacturing. Quantitation of
influenza Ha using Biacore biosensor assay is performed in an
indirect manner as an inhibition assay (Figure 3). Biacore
biosensor quantitation of influenza Ha using Biacore t100 system
shows significantly higher sensitivity and precision compared
with Srid (table 2). in addition, the analysis time is shorter.
Co n C l u s I o n s
optimal microcarrier conditions in Wave Bioreactor were
developed demonstrating the potential for fully disposable
influenza vaccine manufacturing from adherent MdCK cells. a
generic and efficient purification process was developed for the
removal of gdNa and host cell–derived impurities from MdCK
cell-based influenza. a further purification step is, however,
required to achieve vaccine purity according to WHo
recommendations. a Biacore based influenza analysis was
developed and shown to be more accurate and faster than Srid.
advertorial In d u s t r y ye a r b o o k 2010 - 2011
Figure 2: Morphology of Vero cells on Cytodex 1 (top) and Cytodex 3
(bot tom) microcarriers after four days’ growth in a WAVE bioreactor system
Cytodex 1
Cytodex 3
Figure 3: Principle of inhibition in Biacore T100 system; HA is first
immobilized on a dextran matrix (red-filled circles); virus is then mixed with a
fixed concentration of serum and injected over the surface. Free antibodies
(not bound to virus at equilibrium) bind to the surface HA, giving a response.
Low concentration of virus in the sample a) gives high antibody binding,
whereas high virus concentration b) results in low binding level.
A) B)
Björn Lundgren is the vaccine marketing manager at GE Healthcare
Bio-Sciences AB, Björkgatan 30, 75184 Uppsala, Sweden, bjorn.
lundgren@ge.com; www.gelifesciences.com/bioprocess.
BioProcess International 75