Application of cation exchange chromatography in bind and elute and flowthrough mode for the purification of enteroviruses

Journal of Chromatography A 1676 (2022) 463259

Contents lists available at ScienceDirect

Journal of Chromatography A
journal homepage: www.elsevier.com/locate/chroma

Application of cation exchange chromatography in bind and elute and
flowthrough mode for the purification of enteroviruses
Spyridon Konstantinidis a,∗, Murphy R. Poplyk a, Andrew R. Swartz a, Richard R. Rustandi b,
Rachel Thompson b, Sheng-Ching Wang a
a
b

Vaccine Process Research and Development, Merck & Co., Inc., Rahway, NJ, USA
Analytical Research and Development, Merck & Co., Inc., Rahway, NJ, USA

a r t i c l e

i n f o

Article history:
Received 7 April 2022
Revised 15 June 2022
Accepted 16 June 2022
Available online 17 June 2022
Keywords:
Enterovirus
Empty capsids
Cation exchange chromatography
High throughput

Oncolytic virus

a b s t r a c t
Members of the enterovirus genus are promising oncolytic agents. Their morphogenesis involves the generation of both genome-packed infectious capsids and empty capsids. The latter are typically considered
as an impurity in need of removal from the final product. The separation of empty and full capsids can
take place with centrifugation methods, which are of low throughput and poorly scalable, or scalable
chromatographic processes, which typically require peak cutting and a significant trade-off between purity and yield. Here we demonstrate the application of packed bed cation exchange (CEX) column chromatography for the separation of empty capsids from infectious virions for a prototype strain of Coxsackievirus A21. This separation was developed using high throughput chromatography techniques and
scaled up as a bind and elute polishing step. The separation was robust over a wide range of operating
conditions and returned highly resolved empty and full capsids. The CEX step could be operated in bind
and elute or flowthrough mode with similar selectivity and returned yields greater than 70% for full mature virus particles. Similar performance was also achieved using a selection of other bead based CEX
chromatography media, demonstrating general applicability of this type of chromatography for Coxsackievirus A21 purification. These results highlight the wide applicability and excellent performance of CEX
chromatography for the purification of enteroviruses, such as Coxsackievirus A21.
© 2022 Merck Sharp & Dohme Corp., a subsidiary Merck & Co., Inc., Kenilworth, NJ, USA . Published by
Elsevier B.V.
This is an open access article under the CC BY-NC-ND license
( />
1. Introduction
Enteroviruses belong to the picornavirus family and are nonenveloped viruses with a diameter of 27–30 nm. They have a positive sense single stranded RNA genome of ∼7.4 kb long encoding
structural and non-structural proteins necessary for their replication [1,2]. Such viruses are attractive oncolytic agents employed
in cancer treatment [2–4]. Recently, a prototype strain of Coxsack-

Abbreviations: Abs., Absorbance; AC, Affinity capture; BSA, Bovine serum albumin; Csalt , Salt concentration in gradient; Csalt,o , Starting salt concentration in gradient; CCCH, Clarified cell culture harvest; CEX, Cation exchange; CV, Column volume;
CVelution , Column volume number in the elution phase of a column; CVA, Coxsackievirus; E1 – E3, Elution pools 1 – 3; ED, Effective dose; FT1 – FT5, Flowthrough
pools 1 – 5; GSH, Glutathione; HCP, Host cell protein; HT, High throughput; IEX,
Ion exchange; PAGE, Polyacrylamide gel electrophoresis; SDS, Sodium dodecyl sulfate; S, Strip pool; sd, Standard deviation; VP0 – 4, Viral polypeptide 0 - 4; W, Wash
pool.
∗
Corresponding author.
E-mail address: (S. Konstantinidis).

ievirus A21 (CVA21) was also demonstrated as a potentially novel
therapeutic for bladder cancer [5] in addition to other cancers involving tumors overexpressing the cell surface receptor intercellular adhesion molecule 1 [6]. The capsids of full mature enterovirus
virions are composed of 60 copies of four viral proteins (VP) VP1–
VP4 arranged in a shell that packages the RNA genome. The generation of such full mature virus virions is the result of a complex morphogenesis comprised of multiple steps [7,8]. Upon receptor binding and delivery of the RNA genome, the nascently
expressed P1 polyprotein is cleaved by virally encoded proteases
into VP0, VP1 and VP3. These associate to form protomers ([(VP0,
VP1, VP3)1 ]) which are then assembled into pentamers ([(VP0,
VP1, VP3)5 ]). Pentamers can assemble into either empty procapsids ([(VP0, VP1, VP3)5 ]12 ) or encapsidate the replicated genome to
form provirions ([(VP0, VP1, VP3)5 ]12 RNA)). Provirions undergo an
autocatalytic cleavage of VP0 into VP2 and VP4, and the accrued
re-arrangement of capsid proteins results into stable icosahedral
full mature virions ([(VP1, VP2, VP3, VP4)5 ]12 RNA).

/>0021-9673/© 2022 Merck Sharp & Dohme Corp., a subsidiary Merck & Co., Inc., Kenilworth, NJ, USA . Published by Elsevier B.V. This is an open access article under the CC
BY-NC-ND license ( />

S. Konstantinidis, M.R. Poplyk, A.R. Swartz et al.

Journal of Chromatography A 1676 (2022) 463259

Salisbury, UK) were cultured on 1 g L−1 Cytodex-1 microcarriers
(Cytiva) using GIBCOTM William’s E Medium (Thermo Fisher Scientific Inc., MA, USA), supplemented with 10% (v/v) HyCloneTM
Bovine Calf Serum (Cytiva), 0.1% P188, 4 mM L-glutamine, and 20
mM glucose. Cell culture took place at controlled conditions of 37
°C and pH 7.2. At 3 days post bioreactor batching, and once the
cells had reached >90% confluency, the bioreactor underwent a
80% media exchange into serum-free cell culture media. The temperature of the reactor was then controlled to either 37 °C (process
A) or 34 °C (process B). Upon media exchange, the cells were infected with a multiplicity of infection of 0.05 using a CVA21 virus
stock (cat. VR-860). At 4 days post infection, and at an observed
cytopathic effect of >90%, the reactor was harvested. The collected

viral fluid was filtered with a Sartopure® GF+ depth filter (Sartorius, Gưttingen, Germany) and clarified with a Sartoclean® CA 3
μm|0.8 μm filter (Sartorius) before it was processed further. Alternatively, the clarified harvest was stored at 4 °C or -70 °C for short
and long term storage, respectively, until further testing. Results
presented hereafter employ CVA21 material generated from process B, unless stated otherwise.

Virus maturation can be affected by a plethora of factors [8] and
in vitro cell culture production of enteroviruses, such as CVA21,
may not lead exclusively to the assembly of full mature virions,
which are the infectious particles displaying the desired oncolytic
activity; instead non-infectious particles may be assembled, such
as empty procapsids. The latter can potentially elicit undesired
immune responses and are often the subject of scrutiny from
regulatory authorities [9]. The separation of full mature virions
from empty procapsids, or full particles/capsids from empty particles/capsids, is a challenging task typically achieved by differential centrifugation which exploits buoyancy density differences
between the particles [10,11]. Such separations are, however, not
desirable for large scale bioprocessing [12,13]. Hence, alternative
routes for purifying full mature virions from empty procapsids, in
addition to process (e.g., host cell proteins) and product related impurities, are sought. Empty procapsids are routinely encountered
during the processing of adeno-associated virus vectors and here
their separation from full particles has been achieved by employing
ion exchange chromatography (e.g., [14–16]). However, this separation method results in closely-eluting full and empty particle peaks
and requires peak cutting which is challenging to implement at
manufacturing scale and can result in yield losses in favor of purity
[17]. Similarly, chromatography-based purifications of enteroviruses
return low product yields and focus predominantly on the reduction of process related impurities [18,19].
Recently, a novel glutathione affinity chromatography (GSH AC)
capture step has been shown to purify CVA21 from clarified cell
culture harvests [20]. However, upstream process conditions at infection can challenge this step due to an undesired co-elution of
both CVA21 full mature virions and empty procapsids. Here, we
report the deployment of cation exchange (CEX) chromatography

as a polishing step for the purification of CVA21. The virus is produced in adherent cell culture and purified in a three-column process, which employs the GSH AC step, an intermediate ion exchange (IEX) chromatography step, and the CEX polishing step.
High throughput chromatography techniques were employed to
develop the CEX-based polishing step and to generate information
regarding its wide applicability. It is demonstrated that CEX chromatography can be deployed robustly in either bind and elute or
flowthrough mode, returning in both cases mature virions in high
yields while eliminating empty procapsids from the resultant product pool. When deployed in bind and elute mode, the polishing
step eluted full mature virions in a concentrated form and the separation displayed baseline resolution from empty procapsids. The
scalability of the CEX polishing step was also demonstrated and it
was furthermore shown that efficient and effective purification of
the CVA21 full particles from empty procapsids could be obtained
using a diverse selection of cation exchange resins. These results
serve to demonstrate the value of cation exchange chromatography
in the purification of enteroviruses, such as CVA21.

2.2. Large scale GSH affinity column chromatography
GSH affinity chromatography (GSH AC) was performed as described in [20] using a GSH Sepharose® 4 FF column. Briefly, the
column was typically loaded with 200 column volumes (CVs) of
clarified cell culture harvest (CCCH) and eluted in 5 CVs with a
15 mM Tris, pH 8.0, 100 mM NaCl, 1 mM Dithiothreitol, 1 mM
GSH buffer. At large scale, the collected elution pool (i.e., GSH AC
product) was purified further with the application of an intermediate IEX step, and a bind and elute CEX step. Conversely, at high
throughput (HT) scale, the GSH AC product was employed directly
for the development of the CEX step. Purified intermediates were
stored at 4 °C or –70 °C for short and long term storage, respectively. All buffers contained 0.005% polysorbate-80 (PS80). The GSH
AC capture of CVA21 produced from upstream process B, led to a
co-elution of full and empty CVA21 particles. Application of this
step to CVA21 material generated from upstream process A led to
reduced empty procapsids in the elution product pool.
2.3. High throughput chromatography
2.3.1. RoboColumn chromatography

Miniature column HT chromatography experiments employed
Opus® RoboColumns® (Repligen, MA, USA) on a Tecan EVO® 150
robotic station (base unit), which was equipped with an 8–channel
liquid handling arm and an eccentric robot manipulator arm and
operated by EVOware® v2.8. The station was fitted with short
stainless-steel tips and integrated with Te-ChromTM , Te-ShuttleTM ,
and Infinite® M10 0 0pro reader devices. The described configuration allowed for up to eight chromatographic separations to
be executed in parallel in a process described in [21]. Here, all
buffers and solutions contained 0.005% PS80, and all chromatography phases were run at a residence time of 2 min.
A total of 16 RoboColumn-based separations were performed
using GSH AC product as feed (Table 1). Separations #1–14 and
#16 employed 200 μL RoboColumns and they aimed to evaluate
the separation of full mature virions from empty procapsids on a
selection of ion exchange resins. Separation #15 employed 600 μL
RoboColumns and sought to evaluate the separation of full mature
virions from process related impurities. Furthermore, separations
#1–6 and #10–16 were run in bind and elute mode, whereas separations #7–9 were run in flowthrough mode.
For all separations, fractions were collected every 200 μL in fullarea UV transparent 96 well microplates (Corning Life Sciences)
and read on the M10 0 0pro reader at 260 nm, 900 nm and 990

2. Materials and methods
In this work, unless specified otherwise, chemicals and buffers
were from Sigma-Aldrich® (MO, USA), chromatography resins,
columns, stations, and 96 well PreDictorTM plates were from Cytiva (Uppsala, Sweden), robotic stations and pertinent components
were from Tecan Group Ltd. (Männedorf, Switzerland), and virus
stocks were acquired from ATCC (VA, USA).
2.1. Material generation
2.1.1. Generation of CVA21 enterovirus
CVA21 was produced using 3 L BioFlo® vessels operated
through a BioFlo 320 bioprocess control station (Eppendorf, NY,

USA). Human lung fibrobast MRC-5 cells (cat. 05072101; ECACC,
2


S. Konstantinidis, M.R. Poplyk, A.R. Swartz et al.

Journal of Chromatography A 1676 (2022) 463259

Table 1
Details of high throughput RoboColumn chromatography separations carried out to screen the polishing purification of Coxsackievirus A21 via
bind and elute and flowthrough mode cation exchange chromatography and anion exchange chromatography as a function of resin, mobile phase
conditions, phase duration, in column volumes (CVs), and gradient slope. In all separations the pH value was kept constant across all phases other
than the stripping of the columns. [NaCl] depicts the NaCl concentration in the equilibration (Equil.), load, and wash buffer in each separation.
This concentration was also the starting concentration in the elution gradient where applicable. Separations #1–6, #10–14, and #16, employed
200 μL RoboColumns and clarified cell culture harvest (CCCH) from upstream process B. Separation #15 employed 600 μL RoboColumns and
CCCH from upstream process A. In all separations, the collected fractions had a nominal volume of 200 μL.
Separation
#1
#2
#3
#4
#5
#6
#7
#8
#9
#10
#11
#12
#13

#14
#15
#16

Resin
Poros 50 HS
Poros 50 HS
Poros 50 HS
Poros 50 HS
Poros 50 HS
Poros 50 HS
Poros 50 HS
Poros 50 HS
Poros 50 HS
Capto S ImpAct
Capto SP ImpRes
Capto S
Nuvia HR-S
Nuvia S
Poros 50 HS
Nuvia HP-Q

pH
3.8
4.0
4.2
4.5
5.0
6.0
3.8

4.0
4.5
4.0
4.0
4.0
4.0
4.0
4.0
9.0

[NaCl] (mM)
450
450
300
50
50
50
1000
1000
550
50
50
50
50
50
250
50

Equil. CVs

Load CVs
a

10
10
10
10
10
10
10
10
10
10
10
10
10
10
5
10

60
60a
60a
60a
60a
60a
20b
20b
20b
30a

30a
30a
30a
30a
20c
30a

Elution CVs
d

19
19d
19d
16e,f
16e,f
16e,f
NAg
NAg
NAg
24e
24e
24e
24e
24e
13e
16e

Strip CVs
h

5
5h
5h
5h
5h
5h
10h
5h
10j
5h
5h
5h
5h
5h
5j
5h

Gradient slope (mM CV−1 )
55.3
55.3
63.2
59.4
59.4
59.4
NAg
NAg
NAg
60.4
60.4
60.4

60.4
60.4
57.7
59.4

a, load prepared by diluting glutathione affinity chromatography elution (GSH AC) product 3-fold into concentrated equilibration buffer; b load
prepared by adjusting GSH AC product to desired conditions with small additions of 1 M acetic acid and 4 M NaCl stocks; c, prepared as in (a)
with the addition of bovine serum albumin and λ-DNA spikes; d, [NaCl] at end of gradient was 1500 mM; e, [NaCl] at end of gradient was 10 0 0
mM; f, gradient followed by a 3 CV step elution at 10 0 0 mM [NaCl]; g, Not applicable (NA); h, stripped with a pH 7.0, 100 mM Tris, 10 0 0 mM
NaCl buffer; i, stripped with a pH 7.5, 100 mM Tris, 10 0 0 mM NaCl buffer; j, stripped with a pH 6.0, 50 mM citrate, 10 0 0 mM NaCl buffer.

nm. The latter two wavelengths were used for pathlength correction purposes [22]. The made measurements were employed to
construct chromatographic traces. These were used to design the
pooling of the collected fractions and to identify fractions in need
of further analysis. Here, the fractions were pooled in a fashion
yielding up to five pools containing flowthrough fractions (FT1–
FT5), one pool containing wash fractions (W), and one pool containing strip fractions (S). The fractions collected during the elution of the RoboColumns were typically pooled in up to three different ways (i.e., E1–E3), unless stated otherwise. Pools E1 and E2
contained the fractions collected in approximately the first and
second half of the main elution peak, respectively. Pool E3 contained all fractions included in pools E1 and E2 in addition to
a few fractions collected after the last fraction included in pool
E2. Pooling was carried out on a separate Tecan EVO 200 robotic
station, operated by EVOware v2.8, which was equipped with an
8-channel disposable tip liquid handling arm. Here, pools were
generated at a desired volume by mixing equal volumes of fractions of interest, per RoboColumn, in separate wells of Thermo
ScientificTM Armadillo PCR 96-well plates (Thermo Fisher Scientific Inc.). The fractions included in each pool are detailed in Table
S1 (Supporting information). Generated pools and fractions were
either analyzed immediately or stored at 4 °C or –70 °C until
their analysis. Plates containing fractions and pools were sealed
with Thermo ScientificTM NuncTM sealing tape (Thermo Fisher
Scientific Inc.).

For bind and elute chromatography, the RoboColumns were
eluted in NaCl gradients with slopes of ∼55–63 mM CV−1 following their washing. At the end of a gradient, the RoboColumns were
stripped for 5 CVs with a 100 mM Tris, pH 7.0, 10 0 0 mM NaCl
buffer unless stated otherwise. The same buffer was also used in
flowthrough chromatography based separations to strip the RoboColumns at the end of their wash, with the exception of separation
#9 which employed a 100 mM Tris, pH 7.5, 1500 mM NaCl buffer
(Table 1). Linear elution salt gradients were simulated by multistep gradients wherein each step had a size of 1 CV and a salt level
(Csalt ) determined by the equation Csalt = Csalt,o + gradient slope ×
CVelution . Here, Csalt,o is the salt level in the employed equilibration

and wash buffers or the load, and CVelution corresponds to the number of CVs for which a RoboColumn was eluted for. The steps in
the gradient were generated by mixing low and high NaCl concentration buffers, per pH, in Axygen® 2.2 mL 96-well deep square
well plates (Corning Life Sciences) at different ratios to obtain the
desired Csalt .
Separation of full mature CVA21 virions from empty procapsids. The
full/empty CVA21 particle separation was tested in a range of mobile phase conditions for CEX resin PorosTM HS 50 (Thermo Fisher
Scientific Inc.). Additional CEX resins CaptoTM S ImpAct, Capto SP
ImpRes, Capto S, NuviaTM HR-S (Bio-Rad, CA, USA), and Nuvia S
(Bio-Rad) were evaluated, along with the AEX resin Nuvia HP-Q
(Bio-Rad). The CEX-based separations (i.e., #1–14) employed a 50
mM citrate buffer system at a pH range of 3.8–6.0, whereas the
AEX-based separation (#16) employed a 50–70 mM Tris, pH 9.0
buffer system. In both cases, the employed mobile phases included
NaCl concentrations ([NaCl]) of 50 mM–1500 mM (Table 1). The
equilibration, loading, and wash chromatography phases were carried out at pH, buffer, and [NaCl] conditions matching those of the
equilibration buffer.
For separations #1–6, #10–14, and #16, the GSH AC product (feed) was diluted 3-fold in concentrated buffers to provide
the load to the RoboColumns; this was loaded for 30–60 CVs
(Table 1). The concentrated buffers were prepared at compositions
(pH, buffer, and [NaCl]) matching those of the equilibration buffers

post the 3-fold dilution of the GSH AC product. For the AEX-based
separation (#16), the Tris concentration was increased to 70 mM in
the load compared to 50 mM in the equilibration buffer. For separations #7–9, the GSH AC product was adjusted to match the equilibration buffer with the addition of small amounts of 1 M citrate,
pH 4.0 and 5 M NaCl solutions it was loaded to the RoboColumns
for 20 CVs (Table 1).
Separation of full mature CVA21 virions from process related impurities. The described CEX bind and elute RoboColumn methodology was also employed to perform column challenge experiments.
These were carried out by increasing the levels of impurities pre3


S. Konstantinidis, M.R. Poplyk, A.R. Swartz et al.

Journal of Chromatography A 1676 (2022) 463259

sented to the resin and determining their impact on the separation
of full mature virus particles from impurities, such as host cell DNA
and bovine serum albumin (BSA). For this purpose, 600 μL Poros
50 HS RoboColumns were employed in separation #15 (Table 1).
They were equilibrated for 5 CVs before they were loaded for 20
CVs and washed for 5 CVs with equilibration buffer. The RoboColumns were then eluted for 13 CVs in a multi-step NaCl gradient with a slope of 57.7 mM CV−1 and stripped for 5 CVs. The
mobile phases employed during the equilibration and wash of the
RoboColumns were comprised of a 50 mM citrate, pH 4.0, 250 mM
NaCl, 0.005% PS80 buffer system. The strip employed a 50 mM citrate, pH 6.0, 10 0 0 mM NaCl, 0.0 05% PS80 buffer. The generation of
elution buffers took place as described in Section 2.3.1 to return
1/3 CV steps, each at an increasing NaCl level, while using a 50
mM citrate, pH 4.0, 10 0 0 mM NaCl, 0.0 05% PS80 buffer. In these
experiments, the GSH AC product was diluted 3-fold in concentrated buffers to match the equilibration buffer composition post
dilution, as described in Section 2.3.1.1. The final material loaded
to the RoboColumns was then generated by spiking the diluted
GSH AC product with small volumes of concentrated BSA and λDNA (Thermo Fisher Scientific Inc.) stocks to final concentrations
of 0.1 g L−1 and 200 ng mL−1 , respectively. These corresponded to

loading 1.2 mg BSA and 2.4 μg λ-DNA to the Poros 50 HS RoboColumns which represented a >100-fold increase of such impurities in a typical GSH AC product. Absorbance measurements of the
collected 200 μL fractions and their pooling took place as described
in Section 2.3.1.

small amounts of 1 M citrate, pH 4.0 and 5 M NaCl solutions. The
residence time across all steps was set to 3 min, instead of 2 min
used in HT scale.
2.6. Sucrose density gradient centrifugation analysis of CVA21 process
intermediates
The presence of CVA21 empty procapsids and full mature virions in CEX chromatography loads and fractions from a large scale
purification was verified via sucrose density gradient centrifugation
performed. Continuous sucrose gradients were prepared at 11 mL
in Polyclear ultracentrifuge tubes (Seton Scientific, CA, USA) using
15 mM Tris, pH 8.0, 150 mM NaCl, 0.005% PS80 buffers containing sucrose at 15% (w/v) and 45% (w/v). Upon application of 1 mL
samples to the top of the tubes, the gradients were centrifuged
at 360 0 0 rpm for 10 0 min at 4°C using an OptimaTM -SE Ultracentrifuge (Beckman Coulter, CA, USA). Twelve fractions of equal
volumes were then collected from the top of the gradients using
a piston gradient fractionator (Biocomp Instruments, Canada) and
stored at 4 °C until further analysis.
2.7. Analytical methods
2.7.1. Quantitative western blotting
CVA21 full mature virion (VP4) and empty procapsid (VP0) contents in samples were determined via quantitative western blotting using a Sally SueTM system and a 12–230 kDa Sally SueTM
Separation Module kit (Protein Simple, CA, USA). Here, it needs to
be emphasized that while VP0 is included in both provirions and
empty procapsids, the presence of the former in the purified CCCH
samples is expected to be negligible [20]. Therefore the VP0 measurements were indicative of the presence of empty propcapsids in
tested samples. Samples were prepared using an Anti–Rabbit Detection Module (Protein Simple) according to the manufacturer’s
protocol and were denatured in a Mastercycler® Gradient (Eppendorf) for 5 min at 95 °C. For their analysis, an anti–VP4 rabbit pAb
(Lifetein LLC, NJ, USA), diluted to 20 μg mL−1 in Antibody Diluent
2 (Protein Simple), was used. Upon their preparation, the samples

were loaded to the capillaries for 9 sec, separated for 40 min at
250 V, and immobilized for 250 sec. This was followed by their
exposure to antibody diluent for 23 min, to anti–VP4 rabbit primary antibody for 30 min, and to the anti–rabbit secondary antibody for 30 min. The capillaries were then imaged with the chemiluminescence detection settings and a 8 s exposure time setting.
Peaks were integrated with a dropped lines method. All samples
were diluted with a concentrated Tris, pH 7.5 buffer, 0.005% PS80
to a final composition of ∼150 mM Tris, pH 7.5, 0.005% PS80 prior
to their analysis. Assay results were employed to determine yields
via mass balancing.

2.4. Stability of CVA21 in chromatography mobile phase conditions
The impact of three factors on the stability of CVA21 was investigated: pH, [NaCl], and time. For this purpose, GSH AC product
was used, and it was diluted 3-fold in concentrated buffers to yield
a final composition of 50 mM citrate buffers at 18 combinations of
pH (3.8, 3.9, 4.0, 4.1, 4.2, 4.5) and [NaCl] (100 mM, 400 mM, 700
mM) conditions. The starting GSH AC product was also included in
this study as a control. The 18 conditions and control were prepared in triplicate in separate wells of an Axygen 2.2 mL 96-well
deep square well plate and upon their preparation the plate was
sealed with Thermo Scientific Nunc sealing tape and shaken at
1100 rpm for 1.5 h at room temperature. At the end of the incubation period, an aliquot was taken from each well of the plate and
added to 0.5 mL MatrixTM 2D barcoded tubes (Thermo Fisher Scientific Inc.) which were stored at -70 °C until their analysis. The
plate was then sealed and left at room temperature for one day,
under shaking, until a new aliquot was transferred to a second set
of 0.5 mL Matrix 2D barcoded tubes, also stored at -70 °C until
their analysis. All used buffers contained 0.005% PS80.
2.5. Large scale cation exchange column chromatography

2.7.2. Infectivity assay
An automated, high-throughput viral imaging infectivity assay
was used to measure CVA21 potency. Briefly, in this assay, the
tested samples and a CVA21 positive reference control were used

to infect confluent 384 well tissue culture cell plates, which were
planted with SK-MEL-28 cells (cat. HTB-72; ATCC). Upon their infection and incubation, the plates were fixed, permeabilized, and
stained with Hoechst 33342 (nuclei stain) (cat. H3570; Thermo
Fisher Scientific Inc.). Subsequently, the cells in the plates were
immunostained with purified rabbit anti-CVA21 pAb (National Biologics Laboratory) and labeled with Alexa Fluor® 488 AffiniPure
donkey anti-rabbit IgG (cat. 711-545-152; Jackson ImmunoResearch
Inc, PA, USA). The plates were then imaged for the stained nuclei
and the fluorescently tagged viral protein on a BioTek CytationTM 3
reader (Agilent Technologies, CA, USA). The images were analyzed
on the reader’s software to count total (nuclei stain) and infected

The Poros 50 HS CEX microscale purification method was
scaled-up large scale using a 10 cm bed height column with a bed
volume of 200 mL which was connected to an ÄKTA chromatography station, controlled by UNICORNTM v7. The column was first
equilibrated for 4 CVs using a 50 mM citrate, pH 4.0, 400 mM NaCl,
0.005% PS80 buffer, followed up by its loading for 27 CVs. The column was then washed for 4 CVs with a 25 mM citrate, pH 4.0,
50 0 mM NaCl, 0.0 05% PS80 buffer before it was eluted for 4 CVs
with a 25 mM citrate, pH 4.0, 800 mM NaCl, 0.005% PS80 buffer.
Finally, the column was stripped for 4 CVs using a PBS buffer at
pH 7.0, 10 0 0 mM NaCl and 0.005% PS80. Here, the load to the column was a process intermediate obtained by purifying clarified cell
culture harvest with the GSH AC and intermediate IEX steps. The
intermediate IEX product was adjusted to match the composition
of the equilibration buffer for the CEX step with the addition of
4


S. Konstantinidis, M.R. Poplyk, A.R. Swartz et al.

Journal of Chromatography A 1676 (2022) 463259

of pH 4.0, 540 mM NaCl on Capto SP ImpRes led to ∼100% and
∼13% flowthrough yields for full (Fig. 1B) and empty (Fig. 1E) particles, respectively. Likewise, the same condition on Capto S ImpAct
led to flowthrough yields of ∼100% and ∼0% for full (Fig. 1A) and
empty (Fig. 1D) particles, respectively. The aforementioned operating space became narrower with increased pH values suggesting
that an optimal separation would need to employ mobile phases
with a low pH. While enteroviruses can be stable across a wide
range of conditions [24], the employment of an acidic condition
for separating full mature virus particles and empty procapsids,
via CEX chromatography, led to concerns over potential infectivity
losses for CVA21. These were addressed by the execution of a stability study at room temperature which evaluated the relationship
between CVA21 infectivity and factors including liquid conditions
(pH and NaCl concentration) and hold duration (two time points).

(tagged viral protein) cells and these counts were used to calculate
the percentage of infected cells in each well of a tested plate. A
dose response curve was then generated from the estimated percentage of infected cells for each sample and for the CVA21 reference standard in order to calculate the associated effective dose
(ED) 50. Finally, the relative potency for each test sample was determined by taking the ratio between a sample’s ED50 to the reference control’s ED50 and reporting it as a percentage (%Response).
2.7.3. SDS-PAGE
Samples were analyzed via gel electrophoresis using NuPAGETM
12% Bis-Tris 1.0 mm gels (Invitrogen, CA, USA) to track CVA21
empty procapsids and full mature virus particles (VP0 and VP2, respectively; VP4 could not be reliably tracked due to its molecular
weight being close to the low limit of the gel) and proteinaceous
impurities. For this purpose, 700 μL of denaturing buffer was prepared by mixing 200 μL of NuPAGE Sample Reducing Agent (10X)
(Invitrogen) and 500 μL of NuPAGE LDS Sample Buffer (4X) (Invitrogen). 14 μL and 26 μL of denaturing buffer and sample, respectively, were mixed in wells of a Thermo Scientific Armadillo PCR
96-well plate. This was sealed with a Thermo Scientific Nunc sealing tape and centrifuged briefly at 30 0 0 rpm on a Sorvall Legend
XTR centrifuge (Thermo Fisher Scientific Inc.). The PCR plate was
then denatured in a Mastercycler Gradient (Eppendorf) for 10 min
at 70 °C. Following denaturation, up to 25 μL of sample and 2 μL of
Mark12 Unstained Standard (Invitrogen) were loaded into separate
lanes of a gel. The prepared gels were electrophoresed for 50 min

at 200 V in a 1X MOPS running buffer, prepared from NuPAGE
MOPS SDS Running Buffer (20X) (Invitrogen), and stained with a
PierceTM Silver Stain Kit (Thermo Fisher Scientific Inc.) according to
the manufacturer’s protocol, with a 2 min development time. Gel
images were generated on a Gel DocTM EZ System (Bio-Rad) with
a Silver Stain autoexposure scan protocol. VP0–VP4 were identified
based on their expected molecular weight and annotated where
possible by arrows.

3.1.1. Impact of acidic conditions on CVA21 infectivity
The performed stability study indicated an average decrease in
CVA21 infectivity of 15.2% ± 10.6% between the two time points
across the 18 liquid conditions tested (Fig. 1G vs. H). The pH and
[NaCl] effects on CVA21 infectivity were investigated based on regression analysis (Table S2) and they were found to differ between
the two time points; after the 1.5 h hold (Fig. 1G), only NaCl concentration had a significant and positive effect, whereas after the
28 h hold (Fig. 1H), both pH and NaCl concentration had positive
and almost equal effects on CVA21 infectivity. Moreover, the relationship between infectivity and these two factors was stronger at
the second time point compared to the first one (i.e., %R2 of ∼17%
and ∼49% at the first and second time points, respectively, in Table S2). This implied that pH and [NaCl] affected CVA21 infectivity
more prominently at increased hold times at room temperature.
The loss of infectivity as a function of time was also observed in
the GSH AC product control sample which was buffered at pH 8.0
(Fig. 1G and H). The employment of one-way analysis of variance
to compare between the measured infectivities of the control sample and of the tested 18 samples, at the first time point, showed no
significant difference between the 19 samples (Table S3). Hence,
the time dependent infectivity losses in Fig. 1G and H were not
specific to the tested acidic conditions alone; instead they also included inherent, short term infectivity losses for CVA21 at room
temperature. Based on these results, the application of CEX chromatography to purify full mature virions form empty procapsids at
acidic conditions was deemed to be a viable approach for CVA21
purification since no significant infectivity losses are expected over

the short duration of the CEX step (∼5 h at large scale).

2.7.4. Total protein, DNA, and bovine serum albumin analytics
Quant-iTTM PicoGreenTM dsDNA (Invitrogen, CA, USA) and
PierceTM Coomassie Plus (Bradford) (Thermo Fisher Scientific Inc.)
assays were deployed as per the manufacturer’s instructions. BSA
quantitative western blotting analysis was performed as described
in [23].
3. Results and discussion
3.1. Identification of cation exchange chromatography for separation
of CVA21 full mature virions and empty procapsids

3.2. Poros 50 HS chromatography for separation of CVA21 full mature
virions and empty procapsids

RoboColumn resin screening of GSH AC product from an early
static culture virus production process supported the application of
cation exchange chromatography for separating CVA21 empty procapsids from full mature virions, as opposed to anion exchange
and hydrophobic interaction chromatography (data not shown).
These early results were further corroborated via screening HT
batch chromatography experiments (Fig. 1A–F). Full mature virions bound to cation exchangers Capto S ImpAct and SP ImpRes
only at pH 4.0, 420 mM NaCl (Fig. 1A and B, respectively), whereas
binding of empty procapsids to these resins was stronger across
a wider range of tested conditions (Fig. 1D and E, respectively).
Conversely, the multimodal resin Capto MMC ImpRes bound more
strongly both types of particles across a wider range of test conditions (Fig. 1C and F).
The binding differences between the two particle types outlined
a pH and NaCl concentration operating space resulting in nearly
complete separation of CVA21 full mature virions from empty procapsids on each resin. For example, employing a binding condition

The early chromatography screening and stability testing results
were followed by the characterization of the CEX-based CVA21 purification using RoboColumns. Here, Poros 50 HS resin was employed in bind and elute mode (separations #1–6 in Table 1)
due to its large pore size characteristics [25], rendering it better suited to adsorb large solutes, such as CVA21 particles. The
chromatograms from separations #1–6 (Figs. 2A and S1) demonstrated the excellent repeatability of the RoboColumn technique
and yielded valuable information. At pH 5.0 and 6.0 (separations
#5 and #6, respectively), only a single peak was observed in the
elution gradient, whereas at pH 3.8–4.5 (separations #1–4, respectively) two peaks were observed; one in the gradient and
one in the strip. For each separation, the fractions in each peak
were pooled into elution (E3) and strip (S) pools (Table S1). The
areas of the elution and strip peaks in the chromatograms increased and decreased, respectively, with increasing pH. This indicated the presence of two solute populations with their retention
5


S. Konstantinidis, M.R. Poplyk, A.R. Swartz et al.

Journal of Chromatography A 1676 (2022) 463259

Fig. 1. Preliminary high throughput cation exchange batch chromatography screening results for Coxsackievirus A21 and its infectivity dependence on liquid conditions and
time: (A)–(C) Flowthrough yields for full mature virions, as a function of binding pH and NaCl concentration ([NaCl]), for resins Capto S ImpAct, Capto SP ImpRes, Capto
MMC ImpRes, respectively; (D)–(F) Flowthrough yields for empty procapsids, as a function of binding pH and NaCl concentration ([NaCl]), for resins Capto S ImpAct, Capto SP
ImpRes, Capto MMC ImpRes, respectively; (G) and (H) %Response, depicting CVA21 infectivity based on the deployed viral imaging infectivity assay, as a function of pH for
a 1.5 h and 28 h hold, respectively, at room temperature. In (A)–(F) the yields (z-axis) are averages of duplicates. The colorbar in (C) denotes the color scale across (A)–(C).
The colorbar in (F) denotes the color scale across (D)–(F). In (G) and (H), symbols (o), ( ), and (♦) correspond to NaCl concentrations of 100 mM, 400 mM, and 700 mM,
respectively, and symbol ( ) corresponds to a non-acidic control sample. Error bars correspond to ±1 standard deviation (sd).

being strongly affected by pH; a weaker binding one, eluting in
the salt gradient, and a stronger binding one, eluting in the strip.
The composition of the two populations was determined via SDSPAGE analysis of pools E3 and S (Fig. 3A). Here, it is noted that
this analysis did not employ concentration normalizations during
gel loading and hence its results also reflected volumetric concentrations as depicted by the generation of pools from the collected

fractions (Table S1). Furthermore, silver stain also stains single and
double stranded DNA and RNA (e.g., [26]) and for samples which
were rich in full mature virions this resulted to the observation
of a band at the top of the loaded gel lanes which was attributed
to genomic RNA of CVA21. For separations #1–4, E3 contained primarily full mature CVA21 particles (abundant VP2 band and little
to no presence of VP0 band), whereas S contained empty procapsids and small amounts of full mature virions (abundant VP0 band
and presence of VP2 band) (Fig. 3A). In contrast, for separations
#5–6, E3 contained both full mature CVA21 particles and empty
procapsids, and S contained neither of the two (Fig. 3A).
The composition of these two populations was also investigated
by quantitative western blotting analyses of the FT, W, E3 and S
pools (Fig. 2B). For all six separations, the FT and W pools contained ∼0% of the full and empty particles included in the loaded
GSH AC product. Hence, all binding conditions, spanning a pH
range of 3.8–6.0, resulted in high binding of CVA21 particles to

Poros 50 HS. Elution yields, as depicted by the E3 pool yields, varied between ∼75%–∼100% and increased with increasing pH. Strip
yields, as depicted by the S pool yields, decreased with increasing pH (Fig. 2B). These led to mass balance closures well in excess
of 80% for full mature CVA21 virions. Hence, the elution yields of
CVA21 full particles were high and varied within a narrow range.
However, for separations #1–6, the E3 pool yields for the empty
procapsids varied between ∼0% and 60% and increased rapidly
with increasing pH (Fig. 2B). Mass balance closures for these particles (∼30%–∼70%) were poorer than those observed for the full
CVA21 particles and the unaccounted empty particles were considered to be irreversibly bound to the resin. While this would have
a negative effect on the re-use of a column, unless the empty procapsids were removed through a cleaning in place strategy, here
their irreversible binding was considered to be a desirable feature
of this step. Hence, a separation between CVA21 full mature virions
and empty procapsids was also achieved using packed bed column
chromatography, and its resolution depended on pH. This agreed
with the early batch-based chromatography HT screens (Fig. 1A–F).
The results generated from separations #1–6 provided sufficient

information to derive the conclusion that attempting to purify full
mature CVA21 particles from empty particles at less acidic pH conditions led to their co-elution in the salt gradient. This was primarily due to pH effects on the retention of the empty procap-

6


S. Konstantinidis, M.R. Poplyk, A.R. Swartz et al.

Journal of Chromatography A 1676 (2022) 463259

Fig. 2. Bind and elute Poros 50 HS high throughput RoboColumn chromatography results for the separation of Coxsackievirus A21 full mature virions from empty procapsids:
(A) 3D plot of chromatographic traces of recorded absorbance at 260 nm as a fraction number (y-axis) from six separations (#1–6), each at a different pH (x-axis). Lines
(-) and (—) denote absorbances (Abs.) and salt levels, respectively, normalized by their maximum, and symbols (◦) and ( ) denote duplicated experiments (R1 and R2).
The z-axis is a normalized scale from 0 to 1 where 1 denotes the maximum; (B) Bar plot of elution and strip yields and mass balances for full mature virions and empty
procapsids (each bar corresponds to a different pH/separation). Error bars correspond to ±1 standard deviation (sd); (C) Elution salt levels of main elution peak as a function
of pH. The salt level was determined by identifying the fraction associated to the beginning of the elution peak.

sids which decreased more as a function of the pH compared to
the retention of the full particles. This suggested that the encapsidation of genome in the full particles led to an increase in their
negative net charge which led to electrostatic repulsions with CEX
resins and reduced retention compared to the empty particles. This
agrees with earlier studies for a selection of picornaviruses, such
as enterovirus 71, which observed the presence of fewer negatively
charged surface patches for empty particles compared to full particles [27]. The difference in the retention between the two CVA21
particle types on Poros 50 HS was exploited to define operating
windows for establishing the polishing purification step for CVA21
and to deploy it at large scale.

sufficiently high to enable the selection of conservative NaCl concentrations in binding conditions (e.g., 200 mM–650 mM) to avoid
any loss of CVA21 particles in the flowthrough, while allowing for

the robust preparation of mobile phases and ease of implementation. The latter is particularly important when taking into consideration that at large scale processing the product of the GSH AC step,
eluted in a 100 mM NaCl buffer, was loaded directly to the following IEX step, which was run in flowthrough mode. Hence, adopting
a pH binding condition within 3.8–5.0 for the CEX polishing step
would only require the pH adjustment of the intermediate product
and not its dilution (typically undesired at large scale processing).
While the Poros 50 HS operating pH range of 3.8–5.0 led to optimal and robust conditions for binding of GSH AC purified CVA21
particles, the optimal elution pH range for separating full mature
virions from empty procapsids was narrower. Pool E3 contained
quantifiable amounts of empty procapsids (Fig. 2B) at a pH between 4.2 and 5.0 (∼16%–∼60%). Conversely, elution pH values between 3.8 ≤ pH < 4.2 led to the complete separation of CVA21 full
particles from empty particles (Figs. 2B and 3A). This supported the
selection of these pH values as the optimal elution pH range. Such
elution pH conditions had additional benefits that rendered them
well suited to large scale processing. SDS-PAGE analysis for separations #1 (Fig. 3B) and #2 (Fig. 3C) supported that within this
elution pH range, full mature CVA21 virions could be eluted in a
concentrated form, and collected with robust collection windows,
via the application of a single step elution method with a step at
high salt level.
Finally, the near optimal separation of CVA21 mature virions
from empty procapsids at a pH range of 4.2–4.5 could also render this pH range as a viable alternative for eluting CVA21. At such

3.2.1. Operating windows for separating CVA21 full mature virions
and empty procapsids via bind and elute chromatography
Retention trends were generated (Fig. S2) using the bind and
elute chromatograms in Fig. 2A to describe the interaction between
the GSH AC purified CVA21 particles and the Poros 50 HS resin.
A quadratic relationship was, therefore, derived describing the dependence of elution salt on pH for the main elution peak (Fig. 2C)
and binding at a salt level below the fitted line would result to the
binding of CVA21 particles, in the GSH AC product, to the Poros
50 HS resin. Here, it needs be emphasized that in Fig. 2C the salt
levels represent their concentration in elution buffers at the inlet

of a column. Hence, a safety factor of at least ∼60 mM NaCl (i.e.,
approximately equal to the employed gradient slope in mM CV−1 )
would need be considered when choosing the binding NaCl concentration based on these results.
At a pH range of 3.8–5.0, the required elution salt levels varied between ∼350 mM and ∼800 mM NaCl (Fig. 2C). These were
7


S. Konstantinidis, M.R. Poplyk, A.R. Swartz et al.

Journal of Chromatography A 1676 (2022) 463259

Fig. 3. Bind and elute Poros 50 HS high throughput RoboColumn chromatography SDS-PAGE results for the separation of Coxsackievirus A21 full mature virions from empty
procapsids: (A) Gel images of GSH affinity chromatography product (feed), 3-fold diluted feed in concentrated equilibration buffer (load), elution pool 3 (E3), and strip pool
(S) for six separations (#1–6), each at a different pH; (B)–(D) Gel images of fractions comprising pool E3 for six separations (#1–6), each at a different pH, respectively. In
(A) and (B) text above each lane denotes the identity of the tested sample.

pH conditions, pool E3 corresponded to empty procapsid elution
yields of ∼16% and ∼35%, respectively (Fig. 2B). However, for these
two conditions, the E3 pools were comprised of fractions 68–84
and 71–81, respectively (Table S1), which included fractions at the
tail of the corresponding elution peaks (Figs. 2A and S1). The late
eluting fractions contained decreasing and increasing amounts of
full mature virions and empty procapsids, respectively (Fig. 3D and

E as indicated by the intensity of the VP2 and VP0 bands). Excluding a few fractions from the tail of the elution peaks (e.g., fractions
76, 77 and 79–81 for separations #3 and #4, respectively) would
result in product pools free of empty procapsids, at the cost of a
marginal reduction in the elution yields for full CVA21 particles;
the majority of the full particles eluted in a small number of fractions at lower salt levels than the empty procapsids (Fig. 3D and

8


S. Konstantinidis, M.R. Poplyk, A.R. Swartz et al.

Journal of Chromatography A 1676 (2022) 463259

Fig. 4. Flowthrough mode Poros 50 HS high throughput RoboColumn chromatography SDS-PAGE results for the separation of Coxsackievirus A21 full mature virions from
empty procapsids: (A) Gel images of 3-fold diluted GSH affinity chromatography product adjusted to the desired pH and NaCl concentration conditions (load), flowthrough
pool (FT), wash pool (W), and strip pool (S) for three separations (#7–9, respectively), each at a different binding condition (pH and NaCl concentration); (B) Gel images of
load and 13 fractions collected during the loading of the column for separation #9 employing a binding condition of pH 4.5 and 550 mM NaCl. In (A) and (B) text above
each lane denotes the identity of the tested sample.

E). Hence, purifying CVA21 full particles at a range of 4.2 ≤ pH
≤ 4.5 would be near optimal albeit with the requirement of more
stringent peak collection control to avoid co-purifying empty procapsids in the CEX CVA21 product pool.

bind and step elution. However, the latter offers the significant advantage of product concentration via volumetric reduction, which
is desirable for subsequent downstream unit operations. This contributed to the selection of bind and elute Poros 50 HS chromatography for the polishing of CVA21 at large scale.

3.2.2. Separation of CVA21 full mature virions from empty procapsids
via flowthrough chromatography
The stronger binding of CVA21 empty procapsids to the Poros
50 HS resin at acidic conditions, compared to its full mature
virions, made possible the purification of the GSH AC product
via flowthrough mode CEX chromatography. Separations #7–9
(Table 1) were carried out at pH conditions of 3.8, 4.0 and 4.5, respectively (Fig. S3). Here, the employed salt level was determined
based on the retention trends elucidated from the bind and elute
experiments (Fig. 2C) and the SDS-PAGE analysis results in Fig. 3B,
C and E. High flowthrough yields for full mature CVA21 particles

were obtained for separations #7–9 based on quantitative western
blotting analysis (i.e., 92.7% ± 2.9%, 91.7% ± 4.1% and 84.8% ± 8.0%,
respectively). The highly robust nature of CVA21 full and empty
particle separation at pH ≤ 4.0 was also supported by separations
#7 and #8; at this pH range, the tested flowthrough pools were
free of empty procapsids (Fig. 4A). Separation #9, carried out at a
pH of 4.5, led to the inclusion of a small amount of empty procapsids in the flowthrough product pool (13.6%±7.3% and Fig. 4A). This
corresponded to a ∼3-fold reduction in the co-purified empty procapsids in the E3 pool of the bind and elute separation #4 (Figs. 2B
and 3A). The breakthrough of empty procapsids for separation #9
was tracked via SDS-PAGE analysis (Fig. 4B), which showed that
empty procapsids were flowing through at low amounts during
early stages of column loading and their abundance increased with
increasing loading. Hence, their further reduction would require
fine-tuning of both pH and [NaCl] in the binding conditions instead
of modulating the amount of the loaded GSH AC product alone.
Despite this, those pH conditions that were found to be optimal
in a bind and elute based purification (i.e., pH < 4.2) were also
optimal when deployed in a flowthrough mode based purification.
Purifying CVA21 GSH AC product with Poros 50 HS, with optimal bind and elute or flowthrough mode chromatography conditions, led to product pools with high full mature CVA21 particle yields and free of empty procapsids. Hence, both purification
modes were viable. Considering large scale unit operations, implementing a polishing step in flowthrough mode is easier than in

3.3. Large scale CVA21 full mature virion purification via Poros 50 HS
bind and elute chromatography
The polishing of CVA21 via Poros 50 HS bind and elute chromatography, at pH of 4.0, was verified at large scale using a 200
mL column (Fig. 5A). The large scale purification process deployed
the CEX polishing step after the preceding IEX and GSH AC steps.
Bound CVA21 particles were step-eluted from the Poros 50 HS column at 800 mM NaCl. This led to the observation of a single peak
containing concentrated full mature CVA21 particles and no empty
particles. The latter were recovered during the stripping of the column with a neutral pH and high salt buffer. This behavior, along
with the absence of any particles in the collected flowthrough

(Fig. 5B), agreed with the observation made from the HT scale experiments (Figs. 2B and 3A, C). Good agreement was also observed
across scales for full mature virion yields with large scale yields of
∼91% and ∼84% yields, based on quantitative western blotting and
infectivity assays, respectively.
The high product volumes generated from the large scale run
enabled the use of sucrose density gradient centrifugation analysis to verify the CEX-based separation of full CVA21 particles from
empty particles. The process intermediate, which was loaded to
the 200 mL Poros 50 HS column, was shown to contain both particle types (Fig. 6A), whereas the Poros 50 HS elution product was
free of empty particles (Fig. 6B). These results demonstrated further the scalability of the HT scale results and provided additional
confirmation for the performance of the selected conditions; they
led to high elution yields and a complete separation of full CVA21
particles from empty ones in a robust and easy to implement purification at large scale.
3.4. Purification of CVA21 full mature virions from process related
impurities via Poros 50 HS bind and elute chromatography
Apart from separating full mature virions from empty procapsids, the CEX step was also determined to flow through small
9


S. Konstantinidis, M.R. Poplyk, A.R. Swartz et al.

Journal of Chromatography A 1676 (2022) 463259

Fig. 5. Bind and elute Poros 50 HS large scale chromatography results for the separation of Coxsackievirus A21 full mature virions from empty procapsids: (A) Chromatographic trace at 280 nm on the left-hand side y-axis and conductivity, pH traces on the right-hand side y-axis. The x-axis represents column volumes; (B) SDS-PAGE analysis
of fractions collected across the entire loading phase (FT), elution phase (E) and strip phase (S) of the chromatogram in (A). Text above each lane denotes the identity of the
tested sample.

Fig. 6. SDS-PAGE analysis of fractions collected during sucrose density gradient centrifugation for: (A) Starting material (CEX Load) purified by the large scale bind and
elute Poros 50 HS polishing step; (B) Elution product pool (CEX Elution) from the polishing step. In both (A) and (B) the second lane shows the sample that was analyzed
by sucrose density gradient centrifugation and lanes B1–B12 show the fractions collected during their sucrose density gradient centrifugation analysis from the top to the
bottom collected layers. In (A) and (B), text above each lane denotes the identity of the tested sample. Boxes with dashed lines denote the location of empty procapsids and

full mature virions in the collected fractions.

amounts of persistent high molecular weight proteinaceous impurities, which were not removed by the preceding GSH AC and
intermediate IEX steps. This was observed while purifying CCCH
from upstream process A and testing the generated products with
overexposed SDS-PAGE gels (Fig. S4). This trend was also observed
when purifying CCCH from upstream process B; for separation #9
(Fig. 4B) the collected 20 CV flowthrough pool contained both
CVA21 particles and faint bands of protein impurities at similar
molecular weights to those observed in Fig. S4. This supported further the operation of the Poros 50 HS step in bind and elute mode
rather than flowthrough mode for CVA21 purification.
The observation that the bind and elute CEX step contributed
to a further reduction of process related impurities led to the execution of studies aiming to challenge its purification potential.
Here, large amounts of BSA and λ-DNA were spiked to GSH AC
product, as described in Section 2.3.1.2, and the CEX purification
was performed with a binding condition of pH 4.0, 250 mM NaCl
and a NaCl gradient of 58 mM CV−1 . BSA represented a major

process related impurity to be removed by the downstream process, present due to the inclusion of bovine calf serum in the cell
culture, whereas λ-DNA represented a molecularly-distinct type of
contaminant (i.e., host cell DNA) in need of removal from the final
purified product. It needs be highlighted that the employed GSH
AC product purified in this study was generated by upstream process A and as shown in Fig. S4 it included a low amount of empty
procapsids.
The overlay of chromatographic traces from the HT total protein and DNA assays led to the observation of three peaks in the
salt gradient and two peaks in the neutral pH column strip (Fig. 7).
Their pooling (Fig. 7) was followed by analytical testing to determine the presence of full CVA21 particles and BSA via quantitative
western blotting. The former were observed only in elution pool
E2, leading to an elution yield of 96.7% ± 5.2%. BSA was observed
in pools E2 and E4 with increased presence at higher salt levels

(i.e., 17.6 μg ± 1.0 μg and 370.0 μg ± 29.3 μg, respectively, based
on quantitative western blotting analysis). Pool E1 was not tested

10


S. Konstantinidis, M.R. Poplyk, A.R. Swartz et al.

Journal of Chromatography A 1676 (2022) 463259

Fig. 7. High throughput RoboColumn chromatography results from challenging the bind and elute Poros 50 HS step by applying it to the purification of GSH affinity chromatography Coxsackievirus A21 product spiked with large amounts of BSA and λ-DNA (separation #15). The left hand-side y-axis shows overlaid concentration traces (-)
plotted against the number of collected fractions as generated by the Bradford (Protein, (◦)) and PicoGreen (dsDNA, ( )) assays. The right hand-side y-axis shows the NaCl
concentration (-) during the separation. Insert at the top left hand-side corner shows a zoomed-out view of the entire separation to focus on the column load and wash
phases. Double headed arrows (↔) denote the beginning and end points of flowthrough (FT), wash (W), elution (E1–E4) and strip (S) pools, also separated by dashed vertical
lines. Reported concentrations are averages of two replicates.

only for <2% of the loaded 2.4 μg of the λ-DNA spike (Section
2.3.1.2).
The results from these challenge studies demonstrated that the
Poros 50 HS bind and elute step was successful in removing large
amounts of impurities from the GSH AC purified full mature CVA21
virions. It successfully resolved full CVA21 particles from BSA even
when the latter was present in significantly higher amounts than
normally encountered during processing, and it also effected removal of λ-DNA which was mostly recovered in the strip. These
results, along with the fact that the Poros 50 HS step was a polishing step preceded by the GSH AC and intermediate IEX steps, supported the conclusion that the CEX step improved the robustness
of the CVA21 purification process and rendered it more capable of
dealing with process variability.

with the BSA quantitative western blotting assay since the HT total
protein assay yielded a low signal for fractions in this pool (Fig. 7).

On the other hand, the strong HT total protein assay signal for fractions in the S pool (Fig. 7) was attributed to BSA without relying
on the specificity of the BSA quantitative western blotting assay.
This was based on the observations that the feed-stream did not
contain empty procapsids that could elute in the S pool, the majority of the CVA21 full particles eluted in pool E2, and the column
was stripped before collecting the entire amount of BSA in the gradient (total protein trace in Fig. 7). Hence, BSA was shown to be a
strongly retained solute on Poros 50 HS at pH 4.0 and to elute at
higher salt levels than the full CVA21 particles. Here, the resolution
between the two solutes was not complete since pool E2 contained
small amounts of BSA. However, the co-eluted BSA amount corresponded to < 1% of the loaded 1.2 mg BSA spike, with the latter
representing a >2 log increase of protein content in a typical GSH
AC product (Section 2.3.1.2).
Similarly, due to the absence of additional solutes that could
elute in the strip and also return a strong HT DNA assay signal,
the DNA rich peak in the strip (Fig. 7) was attributed to λ-DNA.
Strong binding of DNA to Poros 50 HS has been reported previously [28] without an explanation of the underlying binding mechanism. The HT DNA assay trace also displayed a small peak for
fractions coinciding with pool E2, suggesting the co-elution of DNA
and CVA21 full particles (Fig. 7). However, earlier studies indicated
that the HT DNA assay could track the presence of full mature
CVA21 particles during the CEX polishing step and the generated
signal in pool E2 was therefore inclusive of the eluting CVA21 particles. Despite this, the HT DNA assay signal in pool E2 accounted

3.5. CVA21 full mature virus particle purification with alternative
CEX media
Although early batch chromatography screening results demonstrated that multiple CEX stationary phases could lead to a good
separation between full and empty CVA21 particles (Fig. 1A–F),
these experiments employed starting material from an early virus
production process. In order to confirm this important observation. GSH AC product from virus production process B was employed to assess five alternative cation exchangers in separations
#10–14 (Table 1), carried out at a pH of 4.0. Similar to the Poros
50 HS bind and elute separation at a pH 4.0 (separation #2 in
Table 1), these separations resulted in one peak in the salt gradient and a second one in the strip (Figs. 8A and S5). This behavior resembled the one observed in separations #1–6 (Figs. 2A

11


S. Konstantinidis, M.R. Poplyk, A.R. Swartz et al.

Journal of Chromatography A 1676 (2022) 463259

Fig. 8. Bind and elute high throughput RoboColumn chromatography results for the separation of Coxsackievirus A21 full mature virions from empty procapsids with resins
Capto S ImpAct, Capto SP ImpRes, Capto S, Nuvia HR-S, Nuvia S, and Nuvia HP-Q (separations #10–14 and #16, respectively): (A) 3D plot of chromatographic traces of
recorded absorbance at 260 nm as of fraction number (y-axis) from six separations, each with a different resin (x-axis). Lines (-) and (—) denote absorbances (Abs.) and
salt levels, respectively, normalized by their maximum, and symbols (◦) and ( ) denote duplicated experiments (R1 and R2). The z-axis is a normalized scale from 0 to
1 where 1 denotes the maximum; (B) Bar plot of elution and strip yields and mass balances for full mature virions and empty procapsids (each bar corresponds to a
different resin/separation). Error bars correspond to ±1 standard deviation (sd); (C) SDS-PAGE gel images of GSH affinity chromatography product (feed), 3-fold diluted feed
in concentrated equilibration buffer (load), elution pool 3 (E3), and strip pool (S) for six separations, each at a different resin. In (C), the gel image of Nuvia HP-Q shows the
load and the fractions comprising of the main elution peak for this separation (#16) instead of the feed, E3 and S. In (C) text above each lane denotes the identity of the
tested sample.

and S1). For separations #10–14, the fractions in the observed two
peaks were combined in elution and strip pools, which were then
analyzed via quantitative western blotting (Fig. 8B) and SDS-PAGE
(Fig. 8C). These showed that all alternative cation exchange resins
led to high elution yields for the full mature CVA21 particles and
to elution product pools virtually free of empty procapsids as they
eluted in the strip. The estimated elution yields were higher for the
stationary phases with bead sizes similar to Poros 50 HS (∼50 μm)
(i.e., Capto S ImpAct, Capto SP ImpRes, and Nuvia HR-S led to >90%
yields), compared to those with bead sizes of ∼90 μm (i.e., Capto
S and Nuvia S led to ∼70% yields). The retention trends from these
separations (Fig. S6) showed that the latter group of resins eluted

the full mature CVA21 particles at lower salt levels compared to
the former group which eluted them at similar NaCl concentrations to Poros 50 HS. Consequently, the lower elution yields for
the full CVA21 particles for resins Capto S and Nuvia S could not
be attributed solely to their strong electrostatic binding to these
resins since this would have resulted in elution of the full particles
at higher salt levels compared to resins Capto S ImpAct, Capto SP
ImpRes and Nuvia HR-S. The elution behavior of this type of particle was therefore most likely affected by multiple factors in addition to a resin’s bead size. Full mature CVA21 particle mass balances for separations #10–14, based on the elution and strip pool
yields alone, were in excess of 95% (Fig. 8B). Hence, similarly to
the Poros 50 HS purification (separation #2), the tested alternative
cation exchangers bound nearly all full mature CVA21 particles in
the loaded GSH AC product at a pH of 4.0.

The complete binding of the CVA21 particles and their separation into empty and full particle populations, for five different
cation exchangers, suggested that the cation exchange modality itself was the main driving force enabling the highly efficient and
effective purification of CVA21 particles, more so than the intrinsic
properties of the different stationary phases. This enables its wide
applicability, with respect to the applied chromatography media,
and could extend it to include membrane absorbers and monoliths, in addition to the diffusive media tested here. On the contrary, AEX chromatography was determined to have poor applicability for the purification of CVA21 particles. The AEX resin Nuvia
HP-Q, run at pH 9.0 in bind and elute mode (separation #16 in
Table 1), returned high elution yields for full particles (Fig. 8B),
which co-eluted, however, with the empty particles, also shown
via SDS-PAGE analysis (Fig. 8C). Furthermore, the binding of CVA21
particles to AEX resins decreased significantly as the pH decreased
to 8.0, even at low binding salt levels, indicating that binding could
occur only within a narrow range of conditions (data not shown).
This rendered AEX chromatography as potentially unfavorable for
the separation of full CVA21 particles from empty.
The purification of GSH AC product, with the additional CEX
resins at a pH of 4.0, was determined to be very similar to the
Poros 50 HS based purification; all CEX resins returned high binding of CVA21 particles, which were separated into full mature virions and empty procapsids. The former eluted in the salt gradient,

with high yields, whereas the latter eluted at a high pH and high
salt concentration buffer. The difference in the retention of the two
12


S. Konstantinidis, M.R. Poplyk, A.R. Swartz et al.

Journal of Chromatography A 1676 (2022) 463259

particle types would potentially enable the deployment of these
resins in a similar fashion to Poros 50 HS at large scale (i.e., step
elution at a high salt level). These observations supported the wide
applicability of CEX for purifying CVA21 and had a direct impact
on the large scale CVA21 purification process since the identification of multiple resins can mitigate process disruptions due to, for
example, procurement challenges.

Murphy R. Poplyk: Methodology, Investigation, Data curation,
Visualization, Writing – original draft. Andrew R. Swartz: Methodology, Validation, Resources, Investigation, Data curation, Visualization, Writing – review & editing. Richard R. Rustandi:
Methodology, Investigation, Data curation, Writing – review &
editing. Rachel Thompson: Methodology, Investigation, Data curation, Writing – original draft. Sheng-Ching Wang: Resources,
Project administration.

4. Conclusions
Acknowledgments

Empty viral particles represent a challenging product related
impurity in viral vaccines and gene therapy drug substance production. Their separation from full mature particles typically requires techniques that are not amenable to large scale processing. While anion exchange chromatography based purifications can
be used, these are often limited by a low selectivity between the
two particle types and a significant purity and yield trade-off due
to peak cutting. The application of anion exchange chromatography was shown to be ineffective in separating full mature CVA21

virions from empty procapsids. Conversely, cation exchange chromatography, applied in either bind and elute or flowthrough mode,
led to high yields of full mature CVA21 particles which were free
of empty procapsids. In bind and elute mode, the CEX step was
able to be readily deployed at large scale, due to its robustness,
ease of implementation, and enabling efficient subsequent processing. While the main outcome of this step was separation of empty
CVA21 procapsids from full mature particles, it was also shown
that the CEX step contributed to the robustness of the downstream purification process; it removed proteinaceous impurities,
which were not entirely cleared by previous purification steps,
and could separate the full mature particles from prominent process related impurities in the presence of feed-stream variability.
These observations supported the establishment of cation exchange
chromatography as a polishing step in the purification of CVA21.
Furthermore, preliminary evidence that this effective purification
modality extended to multiple cation exchange resins and alternate enteroviruses, support its applicability as a general technique
for enterovirus-based therapeutic products.
Supporting Information: Additional supporting information may
be found in the online version of this article. This includes Tables
S1–S3 and Figs. S1–S6.

The authors acknowledge Shieh Yvonne for carrying out the
sucrose density gradient centrifugation analysis, Jimmy Devlin for
carrying out the infectivity analysis, Michael A. Winters, Joseph G.
Joyce and Rebecca A. Chmielowski for their assistance in preparing
this manuscript.
Supplementary materials
Supplementary material associated with this article can be
found, in the online version, at doi:10.1016/j.chroma.2022.463259.
References
[1] J. Baggen, H.J. Thibaut, J.R. Strating, F.J. van Kuppeveld, The life cycle of non-polio enteroviruses and how to target it, Nat. Rev. Microbiol. 16 (6) (2018)
368–381.
[2] P.M. Chumakov, V.V. Morozova, I.V. Babkin, I.K. Baikov, S.V. Netesov,

N.V. Tikunova, Oncolytic enteroviruses, Mol. Biol. 46 (5) (2012) 639–650.
[3] H. Fukuhara, Y. Ino, T. Todo, Oncolytic virus therapy: a new era of cancer treatment at dawn, Cancer Sci. 107 (10) (2016) 1373–1379.
[4] D. Masemann, Y. Boergeling, S. Ludwig, Employing RNA viruses to fight cancer:
novel insights into oncolytic virotherapy, Biol. Chem. 398 (8) (2017) 891–909.
[5] N.E. Annels, D. Mansfield, M. Arif, C. Ballesteros-Merino, G.R. Simpson,
M. Denyer, S.S. Sandhu, A.A. Melcher, K.J. Harrington, B. Davies, G. Au,
M. Grose, I. Bagwan, B. Fox, R. Vile, H. Mostafid, D. Shafren, H.S. Pandha, Phase
I trial of an ICAM-1-targeted immunotherapeutic-coxsackievirus A21 (CVA21)
as an oncolytic agent against non muscle-invasive bladder cancer, Clin. Cancer
Res. 25 (19) (2019) 5818–5831.
[6] S. Bradley, A.D. Jakes, K. Harrington, H. Pandha, A. Melcher, F. Errington-Mais,
Applications of coxsackievirus A21 in oncology, Oncolytic Virotherapy 3 (2014)
47.
[7] N. Jayawardena, J.T. Poirier, L.N. Burga, M. Bostina, Virus–receptor interactions
and virus neutralization: insights for oncolytic virus development, Oncolytic
Virotherapy 9 (2020) 1.
[8] P. Jiang, Y. Liu, H.C. Ma, A.V. Paul, E. Wimmer, Picornavirus morphogenesis,
Microbiol. Mol. Biol. Rev. 78 (3) (2014) 418–437.
[9] A. Mullard, Gene therapy community grapples with toxicity issues, as pipeline
matures, Nat. Rev. Drug Discov. (2021).
[10] M.M. Hankaniemi, O.H. Laitinen, V.M. Stone, A. Sioofy-Khojine, J.A. Määttä,
P.G. Larsson, V. Värjomaki, H. Hyöty, M. Flodström-Tullberg, V.P. Hytönen, Optimized production and purification of Coxsackievirus B1 vaccine and its preclinical evaluation in a mouse model, Vaccine 35 (30) (2017) 3718–3725.
[11] C.C. Liu, M.S. Guo, F.H.Y. Lin, K.N. Hsiao, K.H.W. Chang, A.H. Chou, Y.C. Wang,
Y.C. Chen, C.S. Yang, P.C.S. Chong, Purification and characterization of enterovirus 71 viral particles produced from vero cells grown in a serum-free
microcarrier bioreactor system, PLoS One 6 (5) (2011) e20 0 05.
[12] M.G. Moleirinho, R.J. Silva, P.M. Alves, M.J. Carrondo, C. Peixoto, Current challenges in biotherapeutic particles manufacturing, Expert Opin. Biol. Ther. 20
(5) (2020) 451–465.
[13] M.M. Segura, A.A. Kamen, A. Garnier, Overview of current scalable methods for
purification of viral vectors, Viral Vectors Gene Ther. (2011) 89–116.
[14] W. Qu, M. Wang, Y. Wu, R. Xu, Scalable downstream strategies for purification

of recombinant adeno-associated virus vectors in light of the properties, Curr.
Pharm. Biotechnol. 16 (8) (2015) 684–695.
[15] T. Okada, M. Nonaka-Sarukawa, R. Uchibori, K. Kinoshita, H. Hayashita-Kinoh,
Y. Nitahara-Kasahara, S. Takeda, K. Ozawa, Scalable purification of adeno-associated virus serotype 1 (AAV1) and AAV8 vectors, using dual ion-exchange
adsorptive membranes, Hum. Gene Ther. 20 (9) (2009) 1013–1021.
[16] M. Urabe, K.Q. Xin, Y. Obara, T. Nakakura, H. Mizukami, A. Kume, K. Okuda,
K. Ozawa, Removal of empty capsids from type 1 adeno-associated virus vector stocks by anion-exchange chromatography potentiates transgene expression, Mol. Ther. 13 (4) (2006) 823–828.
[17] M. Hebben, Downstream bioprocessing of AAV vectors: industrial challenges &
regulatory requirements, Cell Gene Ther. Insights 4 (2018) 131–146.
[18] A.R.K. Venkatachalam, M. Szyporta, T.K. Kiener, P. Balraj, J. Kwang, Concentration and purification of enterovirus 71 using a weak anion-exchange monolithic column, Virol. J. 11 (1) (2014) 1–8.

Declaration of Competing Interest
Spyridon Konstantinidis reports a relationship with Merck &
Co Inc that includes: employment and equity or stocks. Murphy
R Poplyk reports a relationship with Merck & Co Inc that includes: employment and equity or stocks. Andrew R Swartz reports a relationship with Merck & Co Inc that includes: employment and equity or stocks. Richard R Rustandi reports a relationship with Merck & Co Inc that includes: employment and equity
or stocks. Rachel Thompson reports a relationship with Merck &
Co Inc that includes: employment and equity or stocks. ShengChing Wang reports a relationship with Merck & Co Inc that includes: employment and equity or stocks. Spyridon Konstantinidis has patent #US20210187049A1 pending to Merck Sharp and
Dohme Corp. Murphy R Poplyk has patent #US20210187049A1
pending to Merck Sharp and Dohme Corp. Andrew R Swartz has
patent #US20210187049A1 pending to Merck Sharp and Dohme
Corp.
CRediT authorship contribution statement
Spyridon Konstantinidis: Conceptualization, Methodology,
Software, Formal analysis, Investigation, Data curation, Visualization, Writing – original draft, Writing – review & editing.
13


S. Konstantinidis, M.R. Poplyk, A.R. Swartz et al.

Journal of Chromatography A 1676 (2022) 463259

[19] J.Y. Chang, C.P. Chang, H.H.P. Tsai, C.D. Lee, W.C. Lian, I.H. Sai, C.C. Liu,
A.H. Chou, Y.J. Lu, C.Y. Chen, P.H. Lee, J.R. Chiang, P.C.S. Chong, Selection and
characterization of vaccine strain for enterovirus 71 vaccine development, Vaccine 30 (4) (2012) 703–711.
[20] E.M. Curtis, S. Konstantinidis, M. Poplyk, A.R. Swartz, M.D. Wenger, inventors; Merck Sharp and Dohme Corp, assignee. Purified compositions of enteroviruses and methods of purification with glutathione affinity chromatography. US patent 17/125002. 2020 Dec 17.
[21] S. Konstantinidis, H.Y. Goh, J.M. Martin Bufájer, P. de Galbert, M. Parau, A. Velayudhan, Flexible and accessible automated operation of miniature chromatography columns on a liquid handling station, Biotechnol. J. 13 (3) (2018)
e1700390 Mar.
[22] E.L. McGown, D.G. Hafeman, Multichannel pipettor performance verified by
measuring pathlength of reagent dispensed into a microplate, Anal. Biochem.
258 (1) (1998) 155–157.
[23] J.W. Loughney, C. Lancaster, S. Ha, R.R. Rustandi, Residual bovine serum albumin (BSA) quantitation in vaccines using automated Capillary Western technology, Anal. Biochem. 461 (2014) 49–56.

[24] R.J. Salo, D.O. Cliver, Effect of acid pH, salts, and temperature on the infectivity
and physical integrity of enteroviruses, Arch. Virol. 52 (4) (1976) 269–282.
[25] Y. Wu, J. Simons, S. Hooson, D. Abraham, G. Carta, Protein and virus-like particle adsorption on perfusion chromatography media, J. Chromatogr. A 1297
(2013) 96–105.
[26] M.J. Berry, C.E. Samuel, Detection of subnanogram amounts of RNA in polyacrylamide geis in the presence and absence of protein by staining with silver,
Anal. Biochem. 124 (1) (1982) 180–184.
[27] M. Strauss, N. Jayawardena, E. Sun, R.A. Easingwood, L.N. Burga, M. Bostina,
Cryo-electron microscopy structure of Seneca Valley virus procapsid, J. Virol.
92 (6) (2018) e01927 -17.
[28] H.F. Liu, B. McCooey, T. Duarte, D.E. Myers, T. Hudson, A. Amanullah, R. van
Reis, B.D. Kelley, Exploration of overloaded cation exchange chromatography for monoclonal antibody purification, J. Chromatogr. A 1218 (39) (2011)
6943–6952.

14