BioProcess

International
eBooks

VIRAL VECTOR
PURIFICATION
A DISCUSSION OF CURRENT
CHALLENGES AND METHODS
Maribel Rios, with AleŠ Štrancar,
J. Michael Hatfield,
and Pete Gagnon

September 2020
Purification of Viral Vectors
A Discussion of Current Challenges and Methods

by Maribel Rios, with Aleš Štancar, J. Michael Hatfield, and Pete Gagnon

AAV Purification Challenges page 6

Capsid Removal page 6

Host-Cell DNA Removal page 8

Serotype Diversity page 10

Comparing Methods with IgG Purification page 14

AAV-Specific Analytics page 20

A Universal Purification Platform? page 21

About the Authors page 21

3 BioProcess International 18(9)e2 September 2020 E-Book

 Introducing the
SNAP® column range
• Hi Pressure Rating (up to 40 bars)
• Linear Motion of Pistons avoids Resin Disruption
• Calibrated Glass Tubing (accuracy +/- .0005”)
• True Frits
• Double Ended Adjustment Standard
• Robust Inlet and Outlets
• Graduated Glass with MM Scale
• Quick Release Ends
• Patented Safety Design

Size options: Pressure ratings:

Diameter (mm) Bed Length Maximum (mm) Diameter (ID-mm) Pressure (bars) Pressure (psig)

10 125 250 350 500 750 1000 10 40 580

15 125 250 350 500 750 1000 15 35 508

25 125 250 350 500 750 1000 25 24 348

35 125 250 350 500 750 1000 35 18 261

50 125 250 350 500 750 1000 50 13 188

Columns are configurable for Aqueous or Solvent Resistant applications

Heating and cooling options available.

Visit: www.essential-life.net for more information

D
iscussions about the progress of gene therapy development
and clinical success are incomplete without including the
advancements in viral vector production and purification.
Currently, adenoassociated viral (AAV) vectors are the
most common platform for in vivo gene delivery. They are produced
either by adherent-dependent and suspension cell culture systems.
Both mammalian and insect-cell culture systems have been used
to produce AAV vectors. However, purification from such cultures
can be complicated and problematic. Over the past decade, different
biomanufacturers have developed unique purification strategies
for yielding good manufacturing practice (GMP)-grade AAV vectors
with high quality, safety, and efficacy profiles. Capture typically is
performed using affinity or cation-exchange chromatography.
Polishing is most often performed with anion-exchange
chromatography. Each approach offers particular benefits and
each involves compromises.
Current challenges and methods of AAV purification were the
topics of a virtual discussion, presented below. Participating were
Pete Gagnon (chief scientific officer at BIA Separations), J. Michael
Hatfield (Atticus BioConsulting), and AleŠ Štrancar (chief executive
officer, BIA Separations). I also asked Suparna Sanyal (innovation
and commercial integration manager in the business unit at Lonza
Pharma and Biotech) and Nigel Shipston (director of technical Back to Contents
marketing at Fujifilm Diosynth) to provide their perspectives on
some key topics.

5 BioProcess International 18(9)e2 September 2020 E-Book

AAV Purification Challenges
What are the three biggest challenges in AAV purification?
Gagnon: The three I would pick are material handling at the
upstream–downstream interface, removal of contaminating DNA,
and chromatographic separation of empty and full capsids. The
interface is problematic because the debris and soluble contaminant
load at that step is high, especially with lysates. That grossly
interferes with filtration and concentration. Chromatographic
removal of DNA is difficult for a number of reasons, the simplest
being that the amount of DNA to be removed is high. High amounts
of DNA also foul chromatography media and impair their fractionation
performance. There also is increasing evidence that DNA binds to
the exteriors of AAV capsids. Chromatographic removal of empty
capsids is a challenge because the industry has not yet developed an
adequate toolbox to accommodate all serotypes.
Štrancar: I would add a fourth challenge: a lack of fast, sensitive,
and robust in-process analytics. Without proper analytical methods,
proper purification protocols cannot be developed. Manufacturers
should take into account that the amount of the AAV produced is
low, so an analytical mistake can have large consequences that
result in serious product safety problems.
Are there hidden challenges as well?
Gagnon: There are both hidden challenges and hidden
opportunities. Chromatin is the most well-known example. It hides
in plain sight. However, it is easy to detect and characterize for
developers with the foresight to measure DNA content across process
steps. The only requirement is that industry analysts use polymerase
chain reaction (PCR) technology as cleverly as they are using mass
spectrophotometry. Several published studies on IgG purification
describe a range of methods to extract chromatin from harvests
before chromatography process steps. Those strategies include the
use of charged particles, flocculating agents, depth filters, and
combinations of other methods. Some studies focus on the histone
component, some on the DNA component, and some target both.
Preemptive chromatin extraction is trickier with AAV because many
IgG methods remove viruses along with DNA, but already there are
AAV data that demonstrate practical feasibility of chromatin removal.
Ultimately, any reduction in chromatin improves purification
performance and reproducibility — and the more removed, the
better.
Why is chromatographic separation of empty and full Back to Contents
capsids difficult?
Štrancar: Empty and full capsids are similar in charge and size,
and most chromatographic resins do not provide sufficient resolution
to separate the two. Centrifugation provides better separation, but
scale-up and sampling are much more challenging. New
chromatographic ligands can provide better resolution, but their
adoption by the industry will take some time.
Gagnon: Chromatography is all about surface interactions.
Separation of species requires that exploitable differences exist on
6 BioProcess International 18(9)e2 September 2020 E-Book
 first
fda-approved
gene therapy
partner

leading expertise & proven

commercial success in gene therapy
Catalent Cell & Gene Therapy is your most qualified and capable partner deep & extensive expertise
in aav viral vectors
for the development and manufacturing of your advanced biotherapeutic.
fast & flexible process
Our deep experience and technical expertise in viral vectors combined with development
our ever-expanding clinical through commercial manufacturing capacity, secure clinical through
will provide you with full life cycle support of your life-saving therapeutic. commercial facilities
commercial-ready
aav processes

Comprehensive solutions. Advanced technologies. Better biologics.™

us + 1 888 SOLUTION (765-8846) eur 00800 8855 6178 biologics.catalent.com
© 2020 Catalent, Inc. All rights reserved.
their respective surfaces. That is a challenge when those surfaces
are essentially identical. For AAV in particular, it raises the question
of how empty and full capsid surfaces actually differ. One hypothesis
is that empty capsids are empty because they are defective. The
hope is that those defects serendipitously modify surface chemistry
enough so it can be exploited by chromatography. Another notion is
that the negative charge of DNA radiates through full capsids walls
enough to influence ion-exchange retention. Either way, the
difference between full and empty capsids is slight.
Hatfield: The degree of chromatographic separation of empty and
full capsids also depends strongly on the serotype being purified.
For instance, chromatographic separation of empty and full AAV8
capsids is much easier than for AAV9, for example.
The percentage of empty capsids also affects separation. A low
level of empty capsids is easier to separate from full capsids than a AAV Host-Cell Platforms
high level of empty capsids. Wild-type AAV produces about one Both mammalian and insect cell lines
empty capsid for each full capsid. It is not unusual for AAV vectors have been used to express AAV
to be produced with 10 empty capsids for each full capsid. viruses. Typical platforms consist of
In addition, the heterogeneity of the surface charge on empty and transient transfection of suspension-
full capsids caused by different levels of posttranslational processing adapted mammalian human
embryonic kidney (HEK) 293 cells and
can depend on the mode of upstream vector production (e.g., flat-
the infection of Spodoptera frugiperda
stock cultures, attached cell bioreactors, and suspension cultures). (Sf9) insect cells using the baculovirus
Those factors have not been well investigated. DNA fragments on or expression vector system (BEVS). The
within AAV capsids also will add to capsid heterogeneity and thus former requires using large cell banks
add to separation challenges. and large amounts of plasmid DNA.
Even with equilibrium-density separation (e.g., CsCl The latter requires that two
recombinant baculoviruses (rBV) be
ultracentrifugation), there are well known subpopulations of full
engineered.
capsids that are biologically active. The reasons for such different
density populations are only now beginning to be understood based
on next-generation sequencing (NGS) of DNA within vector
preparations.
Why is it so difficult to remove host-cell DNA? Back to Contents
Gagnon: Host-cell DNA is not simply DNA. When we think about
DNA, we usually envision a lone independent double helix on a black
or white background. The reality is much more complex. Host-cell
DNA is the degraded remnant of host-cell chromosomes, also known
as chromatin. The DNA is very strongly associated with histones and
other nucleoproteins. The proteins protect the DNA against nuclease
enzymes. DNA-histone complexes persist for months after harvest in
the form of clustered nucleosomal arrays and fragments. Before
membrane filtration, they range in size up to several microns. After
membrane filtration, they range in size from 2 nm to 400 nm. The
ability of host-cell DNA and protein components to participate in
ionic, hydrophobic, metal-affinity, and hydrogen bonding makes
such complexes “sticky” from a chemical perspective. That stickiness
causes chromatin to foul the surfaces of all filtration and
chromatography media. Chromatin often forms stable complexes
with viruses as well.
The persistence of chromatin complexes in harvests and lysates
is the reason you find DNA in chromatography fractions where it
8 BioProcess International 18(9)e2 September 2020 E-Book
Forward thinking

Partnering with Everything at our flexible biomanufacturing facility was

designed with a complex gene therapy process in mind.

you to deliver We understand your challenges and anticipate the product’s

needs, using every inch of experience to help you deliver
Gene Therapies meaningful, patient-centred outcomes.

As your clinical and commercial demands change, we offer

you the ability to scale up and produce high throughput and
large volume gene therapy products. We simplify your supply
chain and support your full product lifecycle, from pre-clinical
development to commercial production. We partner with you
to produce both drug substance and drug product.

www.fujifilmdiosynth.com
should not occur. For example, purified DNA does not bind to
cation exchangers, and it does not bind to affinity-chromatography
media such as protein A. But host-cell DNA binds to both through
its histone component, and subsets of it elute in all fractions.
Beyond that, a major subset of chromatin binds more strongly than
the target product, thereby reducing column capacity, separation
performance, and cleanability. With protein A affinity
chromatography, nonspecific chromatin binding inflates host
protein contamination by a factor of 100 and DNA contamination
by a factor of 10,000. The problem is worse for AAV vectors because
of their much higher ratios of chromatin to product. However, that
also suggests that AAV purification has more to gain than IgG from
chromatin reduction before chromatography.
Hatfield: Host-cell DNA and plasmid DNA can bind to the outside
of capsids. If they cannot be dissociated from the capsids without
destroying the capsids, they will be part of the final product. Capsids
with host-cell DNA and plasmid DNA on the outside or within capsids
cannot be separated from vector-DNA containing capsids unless the
surface-charge difference is large enough to outweigh the
heterogeneity of the posttranslational processing of the capsid
proteins.
Host-cell DNA is bound in chromatin released by cell death
during the viral production phase and lysis during harvest.
Chromatin contains highly positive regions on nucleosomes that
bind DNA, highly negative DNA bound to nucleosomes, and very
hydrophobic regions on histone proteins that make up the AAV Serotypes
nucleosome protein complex. If DNA is removed from sections of Eleven serotypes of AAV vectors have
chromatin, the highly positive regions of the histones are available been identified so far, with AAV2
to bind negative regions of other proteins and columns. The highly being the most commonly used and
negative DNA bound to chromatin is available to bind positive best characterized. Serotypes differ in
the type of cells they infect (tropism).
regions of other proteins and columns. Exposed hydrophobic So a specific cell type can be
regions of chromatin histones can bind hydrophobic regions of transduced by selecting the
proteins and columns. That can lead to large agglomerations of appropriate serotype. For example,
chromatin and other cellular debris and binding to anion, cation, AAV2 tropism includes kidney tissue,
or hydrophobic columns and hydrophobic regions of column central nervous system, photoreceptor
structural polymers. All of these interactions can make separation cells, and retinal pigment epithelium.
Current research is focusing on
on columns multimodal rather than simply anionic, cationic, or elucidating the AAV serotype structure
hydrophobic. Conglomerates of chromatin and other cellular debris and developing purification methods
will have a wide range of charge density, which means that some independent of serotype.
conglomerates will coelute with the purified vector. The more that
chromatin and conglomerates can be removed before the capture
step, the better the remainder of the purification process will
perform with respect to yield and purity.
Many different natural AAV serotypes and recombinant Back to Contents
serotypes have been identified. Is this diversity part of the
problem or part of the solution?
Gagnon: Both. From a therapy perspective, diversity is part of the
solution because it expands the scope of treatment options. Different
serotypes have preferences for different tissue types, and different
abilities to cross membranes, including the blood–brain barrier.
10 BioProcess International 18(9)e2 September 2020 E-Book
Solving Three Lentiviral and AAV Viral Vector Bioanalysis Challenges
in Downstream Processing with One Miniaturized Immunoassay
Optimization of viral vector production and downstream processing methods 1 000

requires robust bioanalytical methods that provide answers in ever-shortened

timeframes to influence in-process decision-making. Ideally, methods that 100

streamline analytical workflows, keep pace with accelerated development,

and reduce sample consumption with widened dynamic ranges are needed
10

Response [RU]
AAV1

industry-wide (1). 1
AAV2
AAV3
AAV4
AAV5

Breaking Analysis Time and Volume Barriers

AAV6
0.1 AAV7
AAV8
AAV10

Gyrolab ®
systems perform microfluidic-format immunoassays utilizing 0

10 000 000 100 000 000 1 000 000 000 10 000 000 000 100 000 000 000 1 000 000 000 000

streptavidin–coated particles functionalized for specific assays within a 15 Concentration [VP/mL]

nanoliter affinity column (Figure 1). This column, the heart of the miniaturized Figure 2. Standard AAV capsid titer curves for serotypes 1–8, and AAVrh10.
immunoassay, is contained within each of 96 or 112 microstructures on a 10000

consumable Gyrolab Bioaffy™ CD. Capillary action is used to draw samples

and reagents through the sample inlet of the CD, while precisely applied
centrifugal force moves liquids through the microstructures during every
1000

step of the immunoassay. Each microstructure equates to one data point,

eliminating crosstalk concerns. The affinity flow-through column format

S/B
100

facilitates high binding capacity, providing a large dynamic range.

10
Gyrolab AAV2
ELISA AAV2 Figure 3. AAV2 vector capsid standard
curve with over 3-log dynamic range using
1
1,00E+07 1,00E+08 1,00E+09 1,00E+10 1,00E+11 1,00E+12 Gyrolab AAVX Kit compared to a 1-2 log
[VP/mL] range for ELISA.

Impurity Analysis Accelerated: Process-related impurities in biologics

Sa
andwich assay Affinity-capture Bioaffy CD Gyrolab Bioaffy CD
on
n streptavidin column (15 nL) microstructure
b d
bead
Figure 1. Diagram of sandwich assay within flow-through structures on a
Gyrolab Bioaffy CD.
This format shortens sample-contact time, minimizing assay susceptibility to
matrix interference, with an assay run time of about an hour for 96 or 112
datapoints. Significant for viral vector production with limited manufacturing
batch sizes, sample and reagent consumption are reduced significantly,
with only 8 µL required per duplicate data-point analysis, compared with
the typical 100 µL required for plate-based methods. Gyrolab software is
designed for 21 CFR Part 11 compliance for use in regulated environments.
Rapid LV p24 Titer Analysis with Broad Analytical Range: Lentivirus (LV)
vectors are common delivery systems for ex vivo cell-mediated treatments
such as chimeric antigen receptor (CAR) T-cell therapies. Immunoassay
quantitation of free p24 antigen, a component of the LV capsid, is an accepted
method for LV vector quantitation during production and bioprocessing.
Gyrolab p24 Kits are optimized to measure total LV vector titers and include
labelled reagents, CD, buffer, and the Gyrolab method for 96 data points in
80 minutes. The kits work over a broad analytical range (0.2–1,000 ng/mL).
Prediluted standards are supplied ready to use as well, minimizing pipetting
errors and maximizing analytical consistency and quality (2).
Adeno-associated Virus (AAV) Vector Quantitation for Multiple Serotypes:
AAV vectors have risen in popularity for systemic delivery of genetic therapies
primarily because of their small size, tissue tropism, and low immunogenicity.
Quantitation of AAV vectors during manufacturing and downstream
processing typically require titer assay development of each AAV serotype.
Gyrolab AAVX Kit offers a ready-to-use solution for AAV vector titer
quantitation during bioprocess development for AAV serotypes 1–8 and
AAVrh10 (Figure 2). The kit uniquely uses a biotinylated Thermo Scientific™
CaptureSelect™ anti-AAVX (Thermo Fisher Scientific) ligand that binds with
including viral vector-based therapies and vaccines are highly regulated
because of their immunogenic potential and associated risks to product
safety, efficacy, and quality. Measuring complex, heterogeneous mixtures
of culture-related host-cell proteins (HCPs) require robust and reproducible
methods for bioanalysis of samples throughout manufacturing and
bioprocessing steps.
Gyrolab platform, an open system, where
assays can also be developed using
customer-sourced reagents, for example
for human embryonic kidney (HEK) 293
HCP analysis. Anti-HEK 293 HCP antibodies
(Cygnus Technologies) have been used as
capture and detection reagents to deliver
precision data for Gyrolab HCP analysis
over a broad working range of 2–10,000
ng/mL (customer data).
Streamlining Viral Vector Bioanalysis
The Gyrolab platform, together with software designed to support bioprocess
workflows and minimize in-lab time, streamlines and accelerates bioanalysis
of viral vectors in bioprocess development, facilitating fast decision-making.
The high throughput, unattended operation, wide dynamic ranges, and a
rapid assay completion time of about one hour make Gyrolab immunoassays
ideal for use in viral vector-based gene therapies and vaccines, addressing the
needs of vector manufacturing bioanalysis to keep development pipelines
moving forward.
1) Cell & Gene Therapy Insights 2020; 6(7), 871–875
2) Product Information Sheet: Gyrolab p24 Kit. Gyros Protein Technologies:
Uppsala, Sweden, 2019; https://www. gyrosproteintechnologies.com/
gyrolab-p24-kit
3) Gyrolab Assay Protocols. Gyros Protein Technologies: Uppsala, Sweden,
2020; https://www. gyrosproteintechnologies.com/gyrolab-assays
Gyrolab and Rexxip are registered trademarks and Gyros, Gyrolab xPlore, Bioaffy and Gyros logo are trademarks of Gyros
Protein Technologies Group. All other trademarks are the property of their respective owners. Products and technologies
from Gyros Protein Technologies are covered by one or more patents and/or proprietary intellectual property rights. All
infringements are prohibited and will be prosecuted. Please contact Gyros Protein Technologies AB for further details.
Products are for research use only. Not for use in diagnostic procedures. © Gyros Protein Technologies AB 2020.

high affinity and selectivity to AAV particles of various serotypes for capture,
and an Alexa® 647 (Thermo Fisher Scientific) labeled trimeric CaptureSelect
anti-AAVX conjugate as a detection reagent. The >2X wider dynamic range
compared to a plate-based enzyme-linked immunosorbent assay (ELISA)
shown in Figure 3 reduces repeats and sample dilutions.

www.gyrosproteintechnologies.com
From the perspective of process development, diversity compounds
the effort required to develop effective processes.
Hatfield: The diversity among the different AAV serotypes certainly
is part of the problem in developing a standard purification method.
However, the range of AAV serotypes is a major part of the power of
AAVs for gene therapy. As such, purification can be difficult, but that
diversity is not “part of the problem.” It is simply one of the
challenges.
Does AAV purification offer the same continuous-process Back to Contents
potential as IgG purification?
Štrancar: The key question for me would be whether it makes
sense to invest in continuous AAV manufacturing when we need
several orders of magnitude less product than we do IgG. I would say
no.
Gagnon: I would say AAV purification offers better continuous-
process potential than IgG purification. With antibodies, it requires
retrofitting processes that were not originally developed with
continuous processing in mind. That includes retrofitting

Production, Purification, and Contamination Control with Nigel Shipston

BPI reached out to Nigel Shipston, PhD (director of technical
marketing at Fujifilm) to discuss AAV production in suspension
culture, purification challenges, and best practices for
eliminating process. (Shipston acknowledges support from
Steve Pincus, Mike Baker, and Ramesh Koukuntla.)
Production Platforms: “Upstream process scale-up is
practically and logistically more straightforward — and
accompanied by a smaller equipment footprint — than adherent
cell culture processes that rely on scale-out. With certain genes
of interest (GoI) or vectors that cause syncytia, final titers often
are less significantly affected than when using adherent cells.
Animal-component–free media used for suspension cultures
are readily available, and cells are generally easier to handle
(e.g., recovery procedures do not require the use of reagents
such as trypsin). The industry also has an abundance of
experience and expertise in using suspension culture, so
training and technology/process transfer are straightforward.
But not all cells can be adapted to suspension culture, which
can present difficulties for certain viruses and vectors. Adapting
cells to suspension culture can be laborious and unsuccessful.
Adapted cells need to be characterized and tested for both
productivity and safety, which takes time and expense. Titers
achieved in suspension cell culture can be lower than those
gained with using an adherent cell system. Challenges also are
encountered in maintaining titers from transfection-based
systems, because the process is scaled up, particularly >500 L.
Overall, provided cells can either be grown or adapted to grow
in suspension culture. The long-term advantages of suspension
culture over adherent systems in terms of cost of goods (CoG)
and process portability are compelling.”
Preventing Contamination:: “It is very important to implement
appropriate procedures for control of raw materials and
consumables, to ensure they have been prepared properly and
12 BioProcess International
tested before use. Microfiltration and nanofiltration ensures
removal of adventitious agents from critical process
components such as growth media components and buffers.
Use of presterilized, single-use consumables with closed
operations as much as possible is highly recommended. If
closed operations are not possible, then processing and
handling should be performed in a biological safety cabinet
(BSC) where practicable. If the virus cannot be sterile filtered
and a process must be performed closed, then aseptic process
simulation (APS) will be needed to qualify that process as being
aseptic. Finally, it is essential to ensure that all staff involved in
all aspects of processing operations and material handling
receive comprehensive, documented training in specific
procedures and best practices such as aseptic handling.”
Process Challenges: “One principal difficulty is making viral
vectors in sufficient quantity to enable an efficient downstream
process to be configured to support good recovery of the
desired full capsid product. Efficient lysis of cells and removal
of cell debris before the application of chromatography
purification methods (e.g., affinity and ion-exchange
chromatography) also can be problematic. Biomanufacturers
have had difficulty separating full-capsid products from empty-
capsid subpopulations without loss of the enriched-capsid
fraction. A major challenge is the ability to monitor a process
adequately using rapid analytics. Using established methods
such as genome copy-number determination by quantitative
polymerase chain reaction (PCR) and total particle
determination by enzyme-linked immunosorbent assay (ELISA)
offers certain disadvantages in terms of time to execute and
possible effects of matrix or cellular residual components. Such
technologies can be replaced by high-performance liquid
chromatography (HPLC) methods, which measure the titer of
both full and empty vector particles in a single assay.”
18(9)e2 September 2020 E-Book
ASSESSING CRITICAL QUALITY ATTRIBUTES
OF VIRAL VECTORS
Mark Redfern - Cell and Gene Therapy Team Leader, Intertek Pharmaceutical Services, describes key considerations
for the characterization of viral vectors
Challenges • Aggregation analysis (AUC, DLS, TEM/Cryo-TEM)
The inherent complexity of viral vector-based • Empty vs full capsid analysis (AUC, TEM/Cryo-
products makes their physical and biological TEM)
characterization highly challenging. From a • Charge heterogeneity - icIEF
regulatory perspective, an understanding of • Transgene expression (Flow cytometry, RTqPCR
viral vector critical quality attributes that impact and ELISA)
product safety, purity, and potency, mean that • Liquid chromatography and mass spectrometry
specific, robust analytical assays are needed. (LC-MS) studies on whole virus species to
characterise the viral proteome
Addressing Key Attributes Sedimentation coefficient(s) • Digestion of isolated proteins followed by LCMS
Adeno-associated viruses (AAVs) are the most and / or MALDI-MS to give detailed information to
Figure 1: AUC data for an AAV sample containing (A)
commonly used type of viral vector applied in assist identification of viral proteins
empty capsids, (B) full capsids and (C) other species
gene therapies in trials to date (1) and determining are partially filled capsids containing an incomplete • Potency assays
the presence of capsids that are either empty or DNA load. • Infectious dose and genome titre (qPCR or
contain DNA other than the desired full-length ddPCR)
vector genomes (including wild-type AAV Analytical techniques • Cell based potency with a range of read outs
[wtAAV] DNA sequences) is key to ensuring Analytical ultracentrifugation (AUC) enables (Biochemical endpoints and/or flow cytometry)
purity and for improving the downstream process characterisation of the homogeneity of a vector
of large scale vector production. Empty capsids preparation and determination of complete, Your Total Quality Assurance partner
present a source of unnecessary, potentially empty, and partially packaged capsids (Figure 1). We can support your product development from
antigenic material, possibly inducing or elevating Benefits of AUC include good mass resolution early-stage, through to in-process control and
capsid-triggered anti-AAV immune responses (2). with samples analyzed directly in a buffer/ product release assays. Our experts are adept at
formulation of interest so there are no interactions developing, optimising, qualifying and validating
There are a range of methods that can be applied of sample with chromatography media to
to quantify full versus empty capsids (Table 1) methods. We also have significant experience in
consider when analysing data. It is possible to method transfer. With a heritage of supporting
which include Analytical Ultra Centrifugation look at noncovalent/ weak associations and
(AUC), cryo-TEM, qPCR/ELISA, A260/A280/ advanced pharmaceutical product development,
provide information about particle shape and coupled with a comprehensive range of analytical
absorption at 260nm, and HPLC [SEC, Ion conformation. Results depend on effective
exchange and affinity columns]. technology, our experts offer Total Quality
fitting of a model, input parameters and Assurance expertise to help you ensure the safety,
METHOD ADVANTAGE DISADVANTAGE fitting procedures so these must be carefully efficacy and quality of your product.
Electron Single Labour intensive, controlled.
microscopy (EM) experiment potential for complexity Cryo-TEM can accurately discriminate and References
/ Cryo TEM Robust, quantify particles containing partial genomes 1. Ginn, S.L. et al, Gene therapy clinical trials
Accurate good from full/empty particles and is now routinely
for comparative worldwide to 2017: An update, J Gene Med.,
used to characterize the composition of AAV 2018;20, :e3015
studies
vector preparations. Cryo-TEM can clearly 2. Chemistry, Manufacturing, and Control
qPCR/ELISA Indirect, relies on data visualize viral vectors in their native state and (CMC) Information for Human Gene Therapy
from two different distinguish between full and empty capsids
methods plus
Investigational New Drug Applications (INDs) Draft
allowing for accurate statistical analysis Guidance for Industry Last accessed April 01 2019
uncertainty stems from
whether the ELISA is (Figure 2).
only detecting complete About our expert
vector particles or if
it also interacts even Mark Redfern Cell and Gene
weakly with free capsid Therapy Team Leader
proteins
Mark has 20 years’ experience in the
A260/A280 / Rapid Measures DNA to protein analysis of biopharmaceuticals and vaccines,
absorption at ratio, assumes all DNA/ including method development, validation, and
260nm Protein is associated to
the capsids therefore QC release testing of licensed products. His
requires highly purified wide range of experience covers live viral and
capsids bacterial products, therapeutic DNA together
HPLC Single Very sensitive to with classical recombinant proteins and
[SEC, Ion experiment impurities and the nature monoclonal antibodies
exchange and of buffer formulation Figure 2: Cryo-TEM Image analysis of filled (red)
affinity columns] and empty (blue) capsids allowing for statistical
Analytical Ultra Superb mass- Validation can be
evaluation FOR MORE INFORMATION
Centrifugation resolution – can challenging
(AUC) distinguish / Low throughput
Intertek’s viral vector characterisation
quantify the expertise +44 161 721 5247
different AAV Our experts have worked on multiple studies
species involving Adenovirus, Adeno- Associated Virus biopharma@intertek.com
Does not (AAV) and Lentivirus based products. With
require
standards
a wide range of expertise and technology
in-house, we deliver comprehensive intertek.com/pharmaceutical
Table 1: Analytical methods to quantify full versus
empty capsids characterisation packages or specific services:
multibillion-dollar manufacturing facilities that were designed to
accommodate batch processing. With AAV, we have the opportunity
to apply our knowledge and experience from IgG to develop
processes and design facilities that are intended from the start to
accommodate continuous processing. That adds motivation for
developing chromatographic methods to separate empty and full
capsids. Ultracentrifugation will be difficult to adapt to continuous
processing.
Why does the industry not yet have a standard purification
method for AAV vectors?
Gagnon: Only one process is licensed, so it really is too early for a
standard to emerge. Among several capture options, cation-exchange
and affinity chromatography appear to be the most competitive
currently. Technology for removing empty capsids requires further
development. They can be removed by ultracentrifugation, but there
are some scale-up issues. Polishing with anion-exchange
chromatography supports good separation of empty and full capsids
for some serotypes, but not for all.
Hatfield: We are early in the cycle of AAV gene therapy approvals.
I believe it took quite a few IgGs before there was a typical
purification method for them. The affinity capture step looks good
on paper, but there are some issues with bioactivity of the captured
capsids and final product purity. With regard to the separation of
empty and full capsids, we are only now getting a signal from
regulatory agencies regarding the importance of empty capsid levels
based on total capsid dosing. We are still in early days of
commercialization of gene therapy.
Štrancar: It is too early to expect a standard process because even
AAV products as such are not yet fully defined. Questions to consider
include whether all capsids with DNA inserts can be regarded as
product, and what amount of the residual DNA can be tolerated.
Analytical methods to define purity also have not been standardized.

Comparing Methods with IgG Purification Back to Contents

Some industry experts suggest that IgG provides a good
template for AAV purification. Is that helpful or misleading?
Štrancar: It is important to point out that IgG purification deals with
separation of one of the major components in a fermentation broth,
but the typical amount of AAV produced is almost negligible
compared with other components. We are talking about several
orders of magnitude difference in the amount, so purification of AAV
is a very different task. The sensitivity of analytical methods for
AAVs also is vastly different than for IgGs. Using the same approach
to purify IgG and AAV might lead to problems with process
robustness, scale-up, and in many cases, products that are less pure.
Contaminants that elute with IgG from affinity columns could be
regarded as negligible, but that is definitely not the case when using
affinity methods for AAV purification.
Gagnon: Using IgG purification as a template is both helpful and
misleading, but neither in the ways that most people imagine. The
14 BioProcess International 18(9)e2 September 2020 E-Book
the next leap
in viral vector
gene therapy...
Let’s take it together.
We want to be your partner and add value
to your therapy development process.
We invest in enabling technologies and
build expertise to support the development
and commercialization of new innovative
therapies.
Our scientists and engineers bring decades-
long development experience across a broad
spectrum of cell types and technologies.
This builds the backbone of an extensive
service offering, providing you with tailored
process and analytical development,
manufacturing and regulatory services.
Visit pharma.lonza.com
North America +1 301 898 7025
Europe & rest of world +41 61 316 81 11
Email pharma@lonza.com
© 2020 Lonza. All rights reserved.
fact is that IgG is the molecule the biopharmaceutical industry
matured with — in every respect — from raw-material qualification
to final product testing, including development, scale-up, validation,
regulatory approval, and manufacturing facility design. That
knowledge infrastructure is a tremendous asset for any emerging
class of therapeutics.
The antibody sector also is the primary source of staffing for new
companies developing AAV-based therapeutics. Ex-IgG folks naturally
start with the lessons they learned from IgG. That creates a
disposition toward affinity capture followed by some variant of
anion-exchange polishing, but antibodies have much more to teach.
Production, Purification, and Contamination Control
BPI asked Suparna Sanyal (innovation and commercial
integration manager at Lonza) to provide her thoughts on
crucial issues of viral vector production and purification.
Production Platforms: Suspension cultures offer key
advantages over adherent platforms: Suspension platforms
enable efficient process scale-up, whereas only scale-out
options are available for adherent systems. Using suspension
cultures can lower manufacturing footprints, cost savings, and
number of batches required, and potentially cost per dose.
Closed automated processes with greater process control
enables development of robust and reliable production with
homogeneous cell growth and performance and increased
good manufacturing practice (GMP) and regulatory compliance.
But suspension platforms bring challenges as well. Because
most adenoassociated viral (AAV) vector production processes
use adherent platforms, longer development timelines are
required to establish and validate clonal cell lines adapted for
suspension culture and AAV production. Suspension cultures
have a higher propensity for cellular aggregation and although
scale-up is possible, it requires significant optimization and is
not plug-n-play.
Preventing Contamination: An optimal design space and
established practices are required to ensure segregation of
critical manufacturing operations and effectively mitigate the
potential risk for cross-contamination. That includes use of
designated areas with dedicated manufacturing staff,
segregated material handling, and use of manufacturing suites
dedicated to production of one product and one batch at a
time.
Sterility assurance can be established by using single-use,
disposable materials, and critical process steps are handled in
closed systems when possible. Aseptic processes use
appropriate biosafety cabinets to control potential particulate
and microbial contamination and are verified with aseptic
process simulations. Other best practices for preventing
contamination in include designing and constructing product-
contact surfaces for sterile operations and requiring personnel
to regown before entering a new suite or area. Facilities,
equipment, and utility systems should be routinely cleaned,
disinfected and validated. Such activities should be monitored,
and the data should be regularly reviewed for consistency with
approved specifications and acceptance criteria.
16 BioProcess International
with Suparna Sanyal
Process Challenges: There is a significant shortfall in AAV
production to meet current market demands for development
of AAV therapeutics. Particularly for indications that require
high doses and/or have a large addressable patient population.
One of key reasons is a lack of efficient, scalable, and
industrialized processes. Gene therapies based on viral vectors
are highly complex biological processes. Our understanding is
in its infancy for the underlying mechanisms for viral protein
assembly, genome packaging, infectivity, target specificity, and
gene delivery and expression driving overall therapeutic
efficacy. At best, we are leveraging learnings and efficiencies
from recent advances in mammalian protein production, but
those methodologies are not readily transferable for optimal
viral vector manufacturing. Optimization is required at each
step and unit operation to arrive at efficient, robust, scalable
and commercially viable production processes.
AAV therapeutics development processes also are highly
diverse. Much of this is because of the different production
platforms used to generate AAVs (e.g., transient transfection of
mammalian HEK293 cells, Baculovirus methods using insect
Sf9 cells, and stable producer cell lines). Production and
purification strategies also vary because they depend on the
type of AAV serotype, vector design, transgene sequence, and
host cell line.
Upstream challenges include low and inconsistent viral vector
productivity, which lead to low yield, and cellular aggregation
often is observed with suspension processes. Downstream
difficulties include the need to use different purification
processes (particularly column chromatography unit operations)
according to AAV serotype and transgenes.
Final AAV drug product is a heterogeneous mixture comprising
full, partially full, and empty capsids. Lack of high-throughput
analytics with quick turn-around time that can enable in-process
control is a huge barrier for developing a more defined drug
product enriched with a high percentage of biologically active
full capsids. The industry lacks in vitro potency assays, which
severely limits researchers’ ability to predict clinical efficacy of
AAV therapeutic production. Currently, in vitro cellular infectivity
is used to assess viral vector “potency,” but that does not
always correlate well with in vivo efficacy.
18(9)e2 September 2020 E-Book
 Large-Scale
Transfection for
Virus Production
TransIT-VirusGEN® SELECT
Transfection Reagent
Increase yields and reduce costs
associated with production of AAV
and lentiviral vectors in HEK 293 cells.

• High Performance – Higher

performance than PEI or PEIpro for
maximum virus titers
• Quality Tested – Tested for
performance, identity, sterility,
endotoxin and mycoplasma
® • Cost Effective – Higher virus
titers reduce the cost of materials
and labor

Request a sample:
www.mirusbio.com/SELECT

www.mirusbio.com
©2020 All rights reserved Mirus Bio LLC. Mirus Bio and TransIT-VirusGEN are registered
trademarks of Mirus Bio LLC. All trademarks are the property of their respective owners.
Any method you can think of has been applied to their purification.
Some methods provide important practical benefits. All this
information is accessible easily in published literature.
On the downside, all of those purification media and methods
were customized for antibodies. AAV imposes different constraints:
different size, different surface chemistry, and an immensely
different ratio of product to contaminants. Antibody titers of 2–5 g/L
are commonplace now whereas host-cell contaminants generally are
at least 10 times lower. Productivity of AAV is lower by orders of
magnitude, but contaminant levels remain on a par with IgG
cultures and are higher with AAV lysates. At best, contaminants are
at least 10 times more abundant than AAV — and often 100 times
more abundant. That makes the contaminant-to-product ratio 100–
1,000 times less favorable than it is with antibodies.
Hatfield: IgG has some common features with AAV vectors,
including extensive posttranslational processing that is difficult to
predict and control during IgG production, but IgGs are much
simpler than AAVs. AAVs consist of one genome, five VP1, five VP2,
and 50 VP3 capsid proteins. All 60 of those proteins are
posttranslationally modified to different extents. At this time, we
do not have much of an understanding of how such modifications
affect the biological activity of AAV viruses. There is an
understanding that AAV capsids produced in human HEK293 cells
are less modified than AAV capsids produced in Sf9 insect cells.
AAV produced in human cells were shown to be more potent than
AAV produced in insect cells. In addition, host-cell DNA and
plasmid DNA have been shown to coelute with AAV vectors. At least
some of that DNA is likely to be packaged within AAV capsids. That
incorporation of nonvector DNA adds an additional level of
heterogeneity not seen with IgGs.
Are current purification tools up to the task, or do we need
new ones? Back to Contents
Gagnon: Some current tools are adequate, but some are underused,
and more are needed. The idea that solutions designed for IgG
purification will be able to meet the needs of purifying viruses with
very different characteristics is nonsense. It is easy to forget but
important to remember that customization of purification tools to
meet the needs of IgG purification has been inching along for 40
years and continues to do so. Media for AAV purification will require
similar evolution, both in terms of media architectures and surface
chemistries. Our foundations are more advanced now than they
were for antibodies, so media optimization should not require
another 40 years, but there is still a lot to do.
Hatfield: One of the biggest regulatory issues that will need to be
determined is the level of empty capsids that can be tolerated safely
for a specific route of administration and dose. With that in mind,
we currently do not have the chromatographic tools to produce
highly purified AAV9 full caps without taking a large yield hit.
How are current purification systems ensuring the
elimination of adventitious agents and contamination?
18 BioProcess International 18(9)e2 September 2020 E-Book
 Hatfield: AAV is highly tolerant of surfactants and relatively low
pH. Both parameters are useful for eliminating enveloped viruses.
The challenge will be the multiple log reduction of nonenveloped
viruses without a large reduction in AAV yields.
Gagnon: DNA plasmids represent a potential source of adventitious
agents. Because these plasmids are produced in Escherichia coli,
there is no concern about their contributing human-infective
viruses, but they potentially could carry endotoxins and other host
contaminants. Otherwise, adventitious contaminant sources
basically are the same as with other products from mammalian cell
cultures, and the tools for managing them are the same. Cation-
exchange chromatography is well documented to reduce virus loads,
and lipid-enveloped viruses are extremely labile under the conditions
used for cation-exchange methods for AAV. Anion-exchange
chromatography is better documented in general for virus reduction
and reduces endotoxin contamination. Cation exchangers, anion
exchangers, and AAV all tolerate nonionic surfactant washes that
can further enhance adventitious virus reduction. Chromatin comes
back into the picture as well because it forms strong stable
associations with both viruses and endotoxins. To the extent that a
process reduces chromatin contamination, virus and endotoxin
levels tend to diminish in parallel.

AAV-Specific Analytics Back to Contents

It seems as though analytical methods for AAV vectors need
to address different issues from those addressed by analytics
for recombinant proteins. Is this a barrier to purification
process development?
Gagnon: I might characterize that as a burden rather than a
barrier. Some assays commonly used for AAV are procedurally
complex and take longer than desirable for in-process assays. Some
require days, such as PCR for evaluation of host-cell DNA and
quantitation of vector plasmids. Other methods such as electron
microscopy and analytical ultracentrifugation to measure relative
content of empty and full capsids can require weeks to months,
depending on the capacity of service providers. Electrophoresis is
conveniently rapid, but polyacrylamide gel electrophoresis (PAGE)
gels generally require staining with silver or ultrasensitive
fluorescent dyes to provide information suitable to guide process
development.
High-performance liquid chromatography (HPLC) assays with
simultaneous monitoring by different detectors represent a bright
spot. Monitoring UV at dual wavelengths has been common practice
for a decade and has special utility for AAV. You can estimate DNA
versus protein content across a chromatogram by calculating the
ratio of 260/280 absorbance. Empty capsids are 280 dominant
because of the absence of DNA. Full capsids are 260 dominant
because of their internal DNA.
Multiangle light scattering (MALS) preferentially detects large
species such as AAV capsids. So detecting which peak or peaks from
20 BioProcess International 18(9)e2 September 2020 E-Book
multipeak chromatograms contain AAV is relatively easy. MALS also
can be used to calculate capsid size and mass. Fluorescence detection
is extremely powerful. Intrinsic tryptophan fluorescence gives
15–20 times higher sensitivity protein detection than UV does, and
it does not detect DNA. That makes intrinsic tryptophan fluorescence
valuable for detecting AAV peaks from backgrounds with high
chromatin content. Samples can be prestained with fluorescent
intercalating dyes to selectively reveal DNA or RNA. All of these
detection methods can be combined with size-exclusion
chromatography and ion-exchange chromatography to provide
feedback in 15–30 minutes per assay.
Hatfield: Analytical ultracentrifugation (AUC) provides the most
honest measure of empty and full capsids, with the caveat that it
cannot distinguish capsids containing vector genomes from capsids
containing host-cell or plasmid DNA. However, AUC is difficult to
establish as a release assay. A reliable alternative to AUC would be
useful.

A Universal Purification Platform? Back to Contents

Is it reasonable to expect that one particular purification
platform could emerge?
Hatfield: I doubt you will see a single purification platform emerge,
particularly for the separation of empty and full capsids.
Gagnon: An effective platform first and foremost requires
uniformity of the product class to be purified. Variations within AAV
serotypes appear to be of the same magnitude as within IgG
subclasses while variations among serotypes are more similar in
magnitude to variations among IgG subclasses. That represents
much more variation than IgG platforms currently manage.
Biopharmaceutical companies have overwhelmingly pursued IgG1.
That has enhanced the effectiveness of platform approaches, but a
similarly narrow focus on a single AAV serotype bears a risk of
excluding too many therapeutic opportunities. We need to work with
what nature has given us. That means diversity, and diversity tends
to require a bigger toolbox.

About the Authors

Maribel Rios is managing editor of BioProcess International, a part
of Informa Connect; maribel.rios@informa.com. Aleš Štrancar is
chief executive officer at BIA Separations, Pete Gagnon is chief
scientific officer at BIA Separations and a member of BioProcess
International’s Editorial Advisory Board, and J. Michael Hatfield
is founder of Atticus BioConsulting.

21 BioProcess International 18(9)e2 September 2020 E-Book