Separation and zeta-potential determination of proteins and their oligomers using electrical asymmetrical flow field-flow fractionation (EAF4)

Journal of Chromatography A 1633 (2020) 461625

Contents lists available at ScienceDirect

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

Separation and zeta-potential determination of proteins and their
oligomers using electrical asymmetrical flow field-flow fractionation
(EAF4)
Jaeyeong Choi a,∗, Catalina Fuentes a, Jonas Fransson b, Marie Wahlgren a, Lars Nilsson a
a
b

Department of Food Technology, Engineering and Nutrition, Lund University, 22100 Lund, Sweden
Swedish Orphan Biovitrum AB (publ.), 11276 Stockholm, Sweden

a r t i c l e

i n f o

Article history:
Received 1 August 2020
Revised 11 October 2020
Accepted 12 October 2020
Available online 14 October 2020
Keywords:
Electrical asymmetrical flow field-flow
fractionation (EAF4)
Electrical characteristics
Zeta-potential

Effective net charge
Proteins
Separation

a b s t r a c t
Electrical asymmetrical flow field-flow fractionation (EAF4) is an interesting new analytical technique that
separates proteins based on size or molecular weight and simultaneously determines the electrical characteristics of each population. However, until now, the research using EAF4 has not been published except
for the proof-of-concept in the original publication by Johann et. al. in 2015 [1]. Hence the methods capabilities and optimized conditions need to be further investigated, such as composition of the carrier
liquid, pH stability and effect of the electric field strength.
The pH instability was observed in the initial method of EAF4 due to the electrolysis products when
applied electric field. Therefore, we have investigated and provided a modified method for rapid pH stabilization through additional focusing step with the electric field. Then, the electrical properties such as
the zeta-potential and effective net charge of the monomer and oligomers of three different proteins
(GA-Z, BSA, and Ferritin) were determined based on their electrophoretic mobility from EAF4. The results
showed that there were limitations to the applicability of separation by EAF4 to proteins. Nevertheless,
this study shows that EAF4 is an interesting new technique that can examine the zeta-potential of individual proteins in mixtures (or monomers and oligomers) not accessible by other techniques.
© 2020 The Authors. Published by Elsevier B.V.
This is an open access article under the CC BY license ( />
1. Introduction
Analytical separation for quantification and characterization of
proteins is important for many applications in life science. The
tasks can involve separation of monomer, oligomer and aggregates
of a protein as well as separation of different protein species in a
mixture. In addition to the amount of different protein species as
well as their size and molecular weight (MW), the charge properties of proteins are important as they affect protein characteristics in relation to structure, oligomerization and aggregation. The
zeta-potential is one of the electrical properties which is commonly determined due to its experimental accessibility, and high
importance for protein stability [2,3]. The zeta-potential reflects
the range over which electrostatic interaction occurs in a solution or dispersion and is related to the surface charge of a pro-

∗

Corresponding author.
E-mail addresses: , (J. Choi),
(C. Fuentes), (J. Fransson),
(M. Wahlgren), (L. Nilsson).

tein and the ionic strength, among others. Hence, it can be related
to the inter-molecular electrostatic interaction between proteins in
solution and, thus, to their physical stability. The zeta-potential is
commonly determined from the electrophoretic mobility [4] and
the most widely utilized method of measuring zeta-potential is
phase analysis light scattering (PALS) [5]. As a batch-type analysis method, it provides an average value while zeta-potential values for individual components in mixtures or over broad size-range
distributions cannot be obtained [1,6].
A growing separation technique for proteins is asymmetrical
flow field-flow fractionation (AF4) [7,8] which can be coupled online with various detectors. AF4 is a sized-based separation technique which, in Brownian mode, will separate analytes according to their diffusion coefficient (i.e. hydrodynamic radius). For a
more detailed description of the technique, interested readers can
find information elsewhere [9–11]. Electrical asymmetrical flow
field-flow fractionation (EAF4) is a new sub-technique of AF4 first
described in 2015 [1]. It is a combination of asymmetrical flow
field-flow fractionation (AF4) and electrical field-flow fractionation
(ElFFF) in a separation channel. The combination enables separa-

/>0021-9673/© 2020 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license ( />

J. Choi, C. Fuentes, J. Fransson et al.

Journal of Chromatography A 1633 (2020) 461625

channel spacer was a 350 μm thick Mylar spacer and a regenerated cellulose (RC) membrane (molecular weight cut-off of 10
kDa, Millipore, Bedford, USA) constituted the channel accumulation
wall. The actual channel thickness (w) was determined to be 309

μm from retention time of BSA in 50 mM NaNO3 using the FFFHydRad 2.2 [15]. The EAF4 carrier liquid was pumped into the channel using an Agilent 1200 HPLC pump (Agilent Technologies, Waldbronn, Germany) equipped with an auto-sampler. The conductivity
and pH of the solvent were measured online, after passing the last
detector, in flow cells which are part of the Mobility unit. The carrier liquids were prepared as 50 mM NaNO3 and 50 mM phosphate
buffer at pH 7.0 for BSA, and 25 mM phosphate buffer at pH 7.0
for GA-Z and ferritin experiments. All experiments were performed
with detector flow rate of 1.0 mL/min and constant cross-flow rate
of 4.0 mL/min. The channel was rinsed with the carrier liquid for
20 min without cross-flow and electric field at the end of each
run. All EAF4 experiments were performed at room temperature.
The collection and processing of detector signals was performed
using the ASTRA software (Wyatt Technology, Germany), and electrical data processing was performed using VISION CSH (Superon,
Germany).

tion based on both diffusion coefficient (based on AF4) and, to
some extent, the surface charge of analytes (based on ElFFF). These
two fields can be applied separately or together in EAF4 and in
this study, both of them were used. Ideally, EAF4 would provide
charge-size dependent separation of samples with different charge
or charge density, even with the same size. If the charge density
is different between sample components, it could potentially improve the resolution between components, in comparison to conventional AF4, due to the utilization of the electric field. Another
interesting aspect is that the zeta-potential could be determined
for multiple components while the size distribution is simultaneously determined [1]. A potential disadvantage of EAF4 is the increased number of parameters affecting the result of the separation due to the combination of the two fields applied. For example,
the pH in the channel can be changed by electrolysis products (i.e.
OH− or H+ ) from the electrodes when the electric field is applied
[1,12]. Obviously, such a change in pH can cause changes in the
size and structure/conformation of the sample components [13,14].
To date no investigation for protein characterization utilizing
EAF4 has been published, except for the proof-of-concept in the
original publication [1]. Hence, the method’s capabilities needs to
be investigated. In this study, the purpose is to investigate the application of EAF4 to the separation and characterization of proteins.

The electric field-induced pH change in the separation channel was
investigated as changes in pH may have a strong influence on protein properties and should, ideally, be minimized. The second aim
is to investigate whether the resolution in protein separations can
be improved using the electric field. The third aim is to use EAF4
for the determination of zeta-potentials of different populations
(monomer and oligomers) in a protein.

2.3. Theory
In EAF4 without electric field (i.e. AF4), the retention ratio (R)
in Brownian mode is given by the general expression [7]

V

R=

=

v

t0
1
= 6λ coth
− 2λ
tr
2λ

(1)

where V is the migration velocity of the component zone, <v> is
the average longitudinal carrier velocity, t0 is the void time, tr is

the retention time, and λ is the retention parameter.
At the limit λ → 0, Eq. (1) can be approximated by [7,16]

2. Materials and methods
2.1. Materials

R=

Sodium dihydrogen phosphate monohydrate (NaH2 PO4 H2 O),
disodium phosphate dihydrate (Na2 HPO4 2H2 O), sodium nitrate
(NaNO3 ), bovine serum albumin (BSA), and ferritin (equine spleen)
were purchased from Sigma-Aldrich (Darmstadt, Germany). The
GA-Z is a recombinant protein including GA-domain (albumin
binding site) and Z-domain (target molecule binding site) consisting of 108 amino acids (MW=11.5 kDa, isoelectric point, pI=4.2)
and was provided by Swedish Orphan Biovitrum AB (publ.) (Stockholm, Sweden). The carrier liquid for EAF4 and solution for sample preparation was prepared with water purified through a
Milli-Q Plus purification system (Millipore Co. Ltd., Billerica, USA,
resistance=18.2 M /cm).

t0
= 6λ
tr

(2)

The retention parameter λ is defined by [7]

λ=

l
D

DV 0
=
=
w
|u0 |w Vc w2

(3)

where l is the center of gravity distance from the accumulation
wall of the sample zone concentration distribution, w is the channel thickness, D is the diffusion coefficient of a specific analyte, u0
is the cross-flow velocity at the accumulation wall surface, V0 is
the volume of the channel (void volume), and Vc is the cross-flow
rate.
Substituting Eq. (3) into Eq. (2) yields

w2Vc t 0
6trV 0

2.2. Methods

D=

Electrical asymmetrical flow field-flow fractionation (EAF4)
used in this work was an Eclipse 3+ system (Wyatt Technology,
Dernbach Germany) connected with a Mobility electric field module included conductivity and pH sensor (Superon GmbH, Dernbach, Germany). The EAF4 system was coupled online with a
multi-angle light scattering (MALS) detector (DAWN HELEOS II,
Wyatt Technology), Agilent 1100 diode array detector (DAD, Agilent Technologies, Waldbronn, Germany) with wavelength set at
280 nm, and a differential refractive index (dRI) detector (Optilab
T-rEX, Wyatt Technology).
The EAF4 channel (Superon GmbH) was trapezoidal with a tipto-tip length of 26.5 cm and the inlet and outlet widths of 2.2 and

0.6 cm, respectively. The two electrodes are made of platinized
stainless steel and were opposed to each other in parallel in the
top and bottom block respectively with a distance of 3.7 mm. The

which yields the relationship between tr and D, for a specific analyte. Using the Stokes-Einstein equation the diffusion coefficient
can be transformed into the hydrodynamic radius (Rh ) [17]

Rh =

(4)

V 0 kT
t
π ηt 0Vc r

(5)

w2

where k is the Boltzmann constant, T is the absolute temperature,
π is the ratio of the circumference of a circle to its diameter, and
η is the dynamic viscosity of the solvent.
The void time (t0 ) of the trapezoidal channel, can be calculated
by [18]

⎛

⎛

V0

Vc
⎝1 −
t0 =
ln ⎝1 +
Vc
Vout
2

w b0 z − z
V0

2 b −b
0
L
2L

−y

⎞⎞
⎠⎠

(6)


J. Choi, C. Fuentes, J. Fransson et al.

Journal of Chromatography A 1633 (2020) 461625

where Vout is the detector flow rate, b0 and bL are the breadths of
the trapezoid at the inlet and outlet respectively, z is the position

of the focusing point, y is the area lost from the trapezoid by the
tapered inlet and outlet ends, and L is the channel length.
When injecting a standard sample with a known diffusion coefficient (D) or hydrodynamic radius (Rh ), the actual channel thickness (w) can be determined from the retention time (tr ) of the
standard sample by Eqs. (4) or (5).
The electrophoretic mobility (μ) is defined by [19]

μ=

vEP

respectively. The resulting zeta-potentials reported in this paper
are based on Eq. (13) by using the approximated Henry’s function
(Eq. (12)).

3. Results and discussions
3.1. Optimization of EAF4 method for pH stability
The focusing step of conventional AF4 is a sample relaxation
process which is dependent on the external field (cross-flow) and
the diffusion coefficient of analytes resulting in a characteristic size-dependent concentration profile in the separation channel prior to elution [23–25]. EAF4 has an additional external field
i.e. the electric field. As previously reported, the electric field of
EAF4 is applied in the elution mode and is kept constant during
the analysis [1]. However, electrolysis by-products or components
from non-Faradaic processes (e.g., ions adsorbed or accumulated
on electrodes) can potentially appear when an electrical current
is applied, which, in turn, can cause pH-changes in the solution
[26]. For example, when a negative electric field is applied, the
bottom and top electrodes are the anode and cathode, respectively.
Conversely, the opposite applies when a positive electric field is
applied. The electrolysis products such as OH− or H+ at the top
electrode (cathode or anode) will have a larger effect on channel

pH than electrolysis products from the bottom electrode as it is
located under the accumulation wall membrane and its supporting frit. As a result, the channel pH can be shifted which makes
the use of a buffer in the carrier liquid crucial (It should be noted
that using a buffer for conventional AF4-separations should, in any
case, be the rule of thumb in order to have a defined pH and ionic
strength). Hence, pH-changes occurring in EAF4 will also be dependent on the buffering capacity of the buffer.
Fig. 1 (a) and (b) show the pH vs. elution time observed at various electric fields in two types of carrier liquid (50 mM NaNO3
or 50 mM phosphate buffer at pH 7.0) during an EAF4 run. The pH
changed rapidly during the initial 5 min after which the pH leveled
off in both carrier liquids. In addition, the range of changes in pH
was larger in the non-buffered carrier liquid (50 mM NaNO3 ) compared to the 50 mM phosphate buffer. Moreover, in Fig. 1(a) and
(b), the pH did not always return to the initial pH value even after
20 min of channel flushing, indicating that longer channel conditioning times may be needed between runs.
Fig. 1(c) shows the BSA fractograms and pH of repeated runs
and it can be seen that the void peak of the first experiment is relatively large compared to the second experiment. This is probably
due to the more extensive pH-change during the first experiment
using the non-buffered carrier liquid, which gave rise to changes
in carrier liquid composition. Very slight changes in the retention
times of BSA sub-populations were observed, which could be due
to the differences in pH-change.
Based on these results, the reproducibility in the obtained pH
was investigated by changing the point where the electric field
is switched on by adding an additional focusing step i.e. between
the initial focusing step and elution mode (Fig. 2a and b). It was
expected that the modified method could stabilize the pH faster
compared to the initial method as the electrolysis products are already generated during the additional focusing step, rather than
starting to be formed at the onset of elution. This would allow for
some time to equilibrate the concentration of electrolysis products
and, hence, to stabilize pH.
The result showed that the pH stabilized faster and in more reproducible manner when using the modified method, as shown in

Fig. 2(c) for the NaNO3 carrier liquid and 2(d) for the phosphate
buffer carrier liquid. As would be expected, the effects were more
pronounced for the non-buffered (NaNO3 ) carrier liquid but the re-

(7)

E

where vEP is the drift velocity due to the electric field, and E is the
electric field strength.
In EAF4 vEP can be calculated by [1]
t ln 1+ Vf Vc

vEP = v − vc = e ri

out

/tr

−

1+

f Vc
Vout

Vout
Ael f

(8)

where v is total drift velocity, vc is the drift velocity caused by the
cross-flow without electric field, tri is the retention time with electric field, tr is the retention time without electric field, f is the ratio
between the actual channel separation area (i.e. downstream from
the focusing point) and the total channel area, Vout is the detector flow rate, and Ael is the electrode area in the channel, which is
identical to the total channel area.
The electric field strength (E) can be obtained by [1]

E=

I
Ael kc

(9)

where, I is the electrical current, and kc is the specific conductivity
of the carrier liquid.
Therefore, the electrophoretic mobility (μ) from EAF4 can be
calculated through Eq. (7) from at least two experiments measuring the retention times with and without the electric field.
The effective net charge (Z) of an analyte is defined by [20]

Z=

μ6 π η R h ( 1 + κ R h )
e
f ( κ Rh )

(10)

where, e is the elementary charge, f(κ Rh ) is Henry’s function, and

κ is the inverse of the Debye-Hückel length that is defined by

2e2 NA Ic
0 kT

κ=

(11)

where NA is the Avogadro number, Ic is the ionic strength, is the
relative dielectric constant of the solvent, and 0 is the permittivity
of vacuum.
In EAF4, Henry’s function (f(κ Rh )) assumes a relatively simple
empirical approximation described as [21]

f ( κ Rh ) =

16 + 18κ Rh + 3(κ Rh )2
16 + 18κ Rh + 2(κ Rh )2

(12)

The zeta-potential (ζ ) can be derived from the electrophoretic
mobility (μ) by [21]

ζ=

2

3ημ

0 f ( κ Rh )

(13)

In the limit of small analytes in relation to the DebyeHückel length, κ Rh 1, f(κ Rh ) approaches 1 and Eq. (13) reduces
to the Hückel equation. In the limit of large analytes in relation to the Debye-Hückel length, κ Rh 1, f(κ Rh ) approaches 1.5
and Eq. (13) reduces to the Helmholtz-Smoluchowski equation.
Eq. (13) is valid for determinations of zeta-potentials ≤|50 mV|
[22].
Finally, if the hydrodynamic radius (Rh ) and the electrophoretic
mobility (μ) is determined from EAF4, the effective net charge (Z)
and the zeta-potential (ζ ) can be calculated by Eqs. (10) and (13),
3


J. Choi, C. Fuentes, J. Fransson et al.

Journal of Chromatography A 1633 (2020) 461625

Fig. 1. pH vs. elution time at various electrical currents (-20 to 20 mA) in two types of carrier liquids. (a) 50 mM NaNO3 , (b) 50 mM phosphate buffer at pH 7.0, and (c)
BSA fractograms and pH in duplicate (in 50 mM NaNO3 ).

producibility in the obtained pH was also improved for the phosphate carrier liquid when using the modified method.
It is clear that buffers should be employed to minimize the
change in pH and subsequent effects on the sample as well as for
reproducibility in pH between runs. Thus, a phosphate buffer was
utilized for further experiments in our study.

bly due to attraction between the membrane surface and analytes
caused by the anode at the bottom electrode. In addition, a decreased peak area was observed when the membrane was positively charged, most clearly observed at -15 mA and -20 mA electrical currents (Fig. 3), which is likely to be caused by sample adsorption resulting from the opposite charge of GA-Z and the bottom electrode (anode). Accordingly, analyses should be carried out

carefully to avoid sample adsorption when applying higher electrical currents in the case that the sample and the bottom electrode
are oppositely charged.
In Brownian mode AF4 separation, the GA-Z monomer will
elute before the dimer, following AF4 theory [7]. In order to obtain
somewhat more accurate data for the monomer and dimer, data
points were taken at 10% peak height at the front as well as the
tail of the peak. The MW of 10% height at the front and tail were
determined as 13 kDa and 25 kDa, respectively, which shows that
these fractions were mainly composed of monomer (theoretical
MW=11.5 kDa) and dimer (theoretical MW=23 kDa), respectively.
These points were, thus, used to determine the electrophoretic mobility and zeta-potential of GA-Z monomer and dimer (Table 1).
The fractograms of GA-Z at -15 mA and -20 mA were excluded
from the calculation of electrophoretic mobility as the retention
times were the same for -10 mA to -15 mA and GA-Z was not
eluted at -20 mA. Most likely, strong interaction was present already at -15 mA giving rise to deviations in the relationship between drift velocity and electric field strength.
The results showed that the zeta-potential was -11.2 mV of the
10% height at peak front (mostly monomer) and -7.7 mV for the
10% height at peak tail (mostly dimer) of GA-Z. The lower magnitude of the dimer zeta-potential could possibly be explained by
that the dimer was formed through association of the Z-domains
[27]. The Z-domain contains a higher number of charged amino

3.2. Separation and characterization of proteins by EAF4
In a previous study it was shown that GA-Z has a fast equilibrium between monomer and dimer forms [27]. Additionally,
the dimer is the dominant species at pH 7.0 (approximately 82%
dimer determined by small angle X-ray scattering, SAXS), and the
monomer and dimer of GA-Z cannot be resolved by either AF4 or
size exclusion chromatography (SEC). Therefore, it was investigated
if EAF4 could increase the resolution between the monomer and
dimer of GA-Z compared to conventional AF4.
Fig. 3 shows the fractograms of GA-Z with different electrical

currents (-15 mA to 20 mA) in 25 mM phosphate buffer at pH 7.0.
The results showed that the resolution was not improved when applying the electric fields and only electric field-dependent shifts in
the retention time of GA-Z was observed. In the positive electric
field, the retention time of GA-Z decreased from approximately 5.8
min to 5.5 min which could be due to an increased electrostatic
repulsion between the membrane surface and the analytes (GAZ pI=4.2), caused by the increased negative surface charge of the
membrane with increasing positive electrical current. Moreover, no
significant difference in the retention time of the peak maxima between 10 mA and 15 mA were observed. For this reason, positive
electrical current higher than 15 mA was not used.
Contrariwise, with the negative electrical current, the retention
time of GA-Z increased from 5.8 min to 6.2 min, which is proba4


J. Choi, C. Fuentes, J. Fransson et al.

Journal of Chromatography A 1633 (2020) 461625

Fig. 2. Separation methods for EAF4. (a) initial method, (b) modified method. pH vs. elution time with electrical current at -10 mA for two types of carrier liquids. (c) in 50
mM NaNO3 , and (d) in 50 mM phosphate buffer at pH 7.0, performing two runs using both the initial and modified method respectively.
Table 1
Electrical properties of GA-Z, BSA, and Ferritin from EAF4.
From EAF4
Electrophoretic mobility1 μ
(μmcm/(Vs))

Proteins
GA-Z2

BSA3

Ferritin3
1
2
3

10% height at front
Peak maxima
10% height at tail
Monomer
Dimer
Trimer
Monomer
Dimer

-0.614
-0.577
-0.420
-0.164
-0.159
-0.229
-0.383
-0.280

±
±
±
±
±
±
±

±

0.120
0.120
0.075
0.012
0.002
0.044
0.044
0.009

Theoretical
1

Zeta-potential (mV)

Effective net charge

-11.2 ± 2.1
-10.5 ± 2.2
-7.7 ± 1.4
-3.2 ± 0.2
-3.1 ± 0.1
-4.4 ± 0.9
-6.9 ± 0.8
-5.0 ± 0.2

-1.62
-1.70
-1.76

-0.77
-1.08
-1.88
-3.01
-3.50

±
±
±
±
±
±
±
±

0.31
0.35
0.31
0.01
0.02
0.36
0.35
0.11

1

Net charge at pH 7.0
-10

-16 ~ 18 [29,30]

-

The values include the standard error of the mean (±) based on two replicates.
The theoretical net charge of GA-Z at pH 7.0 was calculated based on the sequence [31].
The values of electrical properties from EAF4 were calculated at peak maxima of monomer and oligomers.

acids compared to the GA-domain. At pH 7.0, the Z-domain and
the GA-domain have 14 and 9 negatively charged amino acids, respectively. Thus, it is possible that the lower magnitude of the
zeta-potential for the dimer is due to that the negatively charged
amino acids of the Z-domain were shielded to a larger extent. The
zeta-potential of the peak maximum (i.e. a mixture of dimer and
monomer) was determined as -10.5 mV.
Fig. 4 shows the fractograms of BSA and Ferritin with different
electrical currents. The retention times of BSA and Ferritin were
decreased with positive electric field, and increased with negative electric field. No change in resolution between monomers and
oligomers was clearly observed when the electric field was applied.

Similarly, as for GA-Z mentioned, repulsion or attraction between
the analytes and the accumulation wall membrane is likely the
cause for the change in retention time as the pH of carrier liquid was above pI of both proteins (Ferritin pI=4.1-5.1 [31] and BSA
pI=4.5–5.5 [32]).
The MW of monomer and dimer of BSA without electric field
(0 mA) were determined as 66.2 kDa and 130.8 kDa, respectively.
However, the MW determined for the monomer and dimer of BSA
were different depending on the electric field applied. For example,
the MW decreased when negative field was applied, and increased
when the positive field was applied (Table 2). Probably, the differences observed in the determination of MW was due to the elec5

J. Choi, C. Fuentes, J. Fransson et al.

Journal of Chromatography A 1633 (2020) 461625

tric field-dependent change in pH which would slightly affect the
dn/dc value of BSA as the solvent composition was changed. Hence,
the utilization of the electric field introduces an uncertainty for the
determination of MW.
The zeta-potentials and electrophoretic mobility of monomers
and oligomers of BSA and Ferritin were determined from peak
maxima for the respective species and are shown in Table 1. The
zeta-potentials and electrophoretic mobility of BSA were determined as -3.2 mV and -0.164 μmcm/V−1 s−1 for monomer, -3.1
mV and -0.159 μmcm/V−1 s−1 for dimer, and -4.4 mV and -0.229
μmcm/V−1 s−1 for trimer, respectively.
The determined zeta-potentials of the monomer and dimer of
BSA were similar, while that of the trimer was slightly higher. The
slightly higher zeta-potential for the trimer is likely to be related
to the trimer structural properties but it is difficult to draw any
conclusion regarding this observation.
The electrophoretic mobility for monomer and dimer of BSA
was previously reported as -2.66 μmcm/V−1 s−1 and -3.77
μmcm/V−1 s−1 by EAF4 [1], which is approximately 15 times
higher than the values determined in this study. The 50 mM phosphate buffer at pH 7.0 for carrier liquid used in our study was
closer to the pI of BSA than the carrier liquid used in a previous
study (10 mM phosphate buffer at pH 8.0) [1], which could result in lower electrophoretic mobility. Another reason for the lower
electrophoretic mobility in our study is that the high conductivity (ionic strength) of the carrier liquid decreased the electric field
strength at a given electrical current (Eq. (9)). In our results, the retention time shift between runs in absence or presence of the electric field was lower than the previously reported results, (i.e. lower
vEP , Eq. (8)). Therefore, it is presumed that we obtained lower electrophoretic mobility (Eq. (7)).
The zeta-potentials of Ferritin were determined to be -6.9 mV
for the monomer and -5.0 mV for the dimer, respectively. Similar

to GA-Z, it is reasonable to suspect that the lower zeta-potential of
the dimer is related to shielding of charges when in the dimeric
form.
The theoretical net charge of GA-Z and BSA at pH 7.0 are 10 and -16 ~ -18, respectively [28,29]. However, the effective net
charge determined from EAF4 was much lower than the theoretical
values (see Table 1). The deviation probably, in part, arose from the
low electrophoretic mobility resulting in a low effective net charge
(Eq. (10)). Nevertheless, the approach can be advantageous when
used for relative comparison between analytes rather than as absolute value.
It is important to emphasize that zeta-potential and surface
charge reflects considerably different properties. The zeta-potential
is dependent on the charge at the surface but is influenced by several other parameters. This means that the zeta-potential will be
strongly dependent on the ionic strength of the surrounding solution as it will influence the Debye-Hückel length (Eq. (11)) as well
as, for instance, pH. Thus, the zeta-potential will decrease strongly
with increasing ionic strength and experimental determination of
the zeta-potential becomes sensitive to already small differences
in the ionic strength.
It should be noted that results with similar trend were observed
for the analyzed samples (GA-Z, BSA, and Ferritin) i.e. when positive current was applied, the retention time of the samples decreased. This is expected because the samples (pI range 4.1 to 5.5)
were negatively charged under these running conditions (i.e. close
to physiological pH). The negative charge of the cathode at the bottom electrode gave rise to repulsion between the sample and the
surface of the membrane resulting in shorter retention time. Conversely, it is expected that the retention time and separation behavior increase when negative electric filed was applied. It could
be thought that increment of the electrical current (higher electric
field) would allow for higher resolution. Nevertheless, this did not

Fig. 3. Fractograms of GA-Z with different electrical currents in 25 mM phosphate
buffer at pH 7.0.

Fig. 4. EAF4 fractograms with different electrical currents. (a) BSA in 50 mM phosphate buffer at pH 7.0, and (b) Ferritin in 25 mM phosphate buffer at pH 7.0.

Table 2
The molecular weight (MW) of the monomer and dimer of BSA determined
from EAF4-MALS (using dn/dc=0.185 mL/g).
Electrical current (mA)

Molecular weight of BSA at peak maxima
Monomer (kDa)
Dimer (kDa)

-20
-10
0
10
20

63.3
64.2
66.2
66.3
67.8

121.7
125.3
130.8
131.8
141.5

6

J. Choi, C. Fuentes, J. Fransson et al.

Journal of Chromatography A 1633 (2020) 461625

occur during the analysis of the samples. The elution behavior did
not show significant improvements when a negative current is applied (see Figs. 3 and 4) and at excessive electrical current, adsorption or immobilization to the accumulation wall membrane could
instead be observed.
An underlying limitation for the applicability of EAF4 to
proteins seems to be the relatively narrow window for EAF4
method parameters. This applies both for higher ionic strengths
(approx.>50 mM) which will quench the electric field and limits the investigation of therapeutic proteins in formulation or proteins under physiological conditions as well as for higher electric
fields as outlined above. Additionally, as the electrolyte concentration increases at higher ionic strengths, the value of zeta-potential
falls due to the shielding effect of the increased concentration of
counter ions which causes a strong decay in the electric potential
arising from analyte charges.

editing, Project administration, Supervision. Lars Nilsson: Conceptualization, Writing - review & editing, Supervision, Project administration.
Acknowledgments
The research in this study was performed with financial support from Vinnova-Swedish Governmental Agency for Innovation
Systems and the Swedish Research Council within the NextBioForm
Competence Centre (grant number 2018-04730). SOLVE Research &
Consultancy AB, Lund, Sweden is gratefully acknowledged for providing access to the EAF4 instrumnet. Swedish Orphan Biovitrum
AB, Stockholm, Sweden are acknowledged for providing the GA-Z
protein and information regarding the protein.
References
[1] C. Johann, S. Elsenberg, H. Schuch, U. Rösch, Instrument and method to determine the electrophoretic mobility of nanoparticles and proteins by combining
electrical and flow field-flow fractionation, Anal. Chem. 87 (8) (2015) 4292–
4298 />[2] W.R. Bowen, N.J. Hall, L.-C. Pan, A.O. Sharif, P.M. Williams, The relevance of
particle size and zeta-potential in protein processing, Nat. Biotechnol. 16 (8)
(1998) 785–787 />[3] A. Koulov, Protein stability and characterization, in: D.J.A. Crommelin, R.D. Sindelar, B. Meibohm (Eds.), Pharmaceutical Biotechnology: Fundamentals and

Applications, Springer International Publishing, Cham, 2019, pp. 33–56. https:
//doi.org/10.1007/978- 3- 030- 00710- 2_3.
[4] R.J. Hunter, Zeta Potential in Colloid Science: Principles and Applications, 2,
Academic press, 2013.
[5] W.W. Tscharnuter, Mobility measurements by phase analysis, Appl. Opt. 40
(24) (2001) 3995–4003 />[6] J.F. Miller, K. Schätzel, B. Vincent, The determination of very small electrophoretic mobilities in polar and nonpolar colloidal dispersions using phase
analysis light scattering, J. Coll. Interface Sci. 143 (2) (1991) 532–554 https:
//doi.org/10.1016/0021-9797(91)90286-H.
[7] K.G. Wahlund, J.C. Giddings, Properties of an asymmetrical flow field-flow fractionation channel having one permeable wall, Anal. Chem. 59 (9) (1987) 1332–
1339 />[8] A. Litzen, K.G. Wahlund, Improved separation speed and efficiency for proteins,
nucleic acids and viruses in asymmetrical flow field flow fractionation, J. Chromatogr. 476 (1989) 413–421 />[9] A. Litzén, Separation speed, retention, and dispersion in asymmetrical flow
field-flow fractionation as functions of channel dimensions and flow rates,
Anal. Chem. 65 (4) (1993) 461–470 0 052a025.
[10] K.G. Wahlund, Flow field-flow fractionation: critical overview, J. Chromatogr. A
1287 (2013) 97–112 />[11] K.G. Wahlund, L. Nilsson, Flow FFF – basics and key applications, in:
S.K.R. Williams, K.D. Caldwell (Eds.), Field-Flow Fractionation in Biopolymer
Analysis, Springer Vienna, Vienna, 2012, pp. 1–21. />978- 3- 7091- 0154- 4_1.
[12] K. Baler, O.A. Martin, M.A. Carignano, G.A. Ameer, J.A. Vila, I. Szleifer, Electrostatic unfolding and interactions of albumin driven by pH changes: a molecular dynamics study, J. Phys. Chem. B 118 (4) (2014) 921–930 />1021/jp409936v.
[13] R. Li, Z. Wu, Y. Wangb, L. Ding, Y. Wang, Role of pH-induced structural change
in protein aggregation in foam fractionation of bovine serum albumin, Biotechnol. Rep. 9 (2016) 46–52 />[14] T.M. Carlson, K.W. Lam, C.W. Lam, J.Z. He, J.H. Maynard, S. Cavagnero, Nakedeye detection of reversible protein folding and unfolding in aqueous solution, J. Chem. Educ. 94 (3) (2017) 350–355 />6b00507.
[15] A. Håkansson, E. Magnusson, B. Bergenståhl, L. Nilsson, Hydrodynamic radius
determination with asymmetrical flow field-flow fractionation using decaying
cross-flows. Part I. A theoretical approach, J. Chromatogr. A 1253 (2012) 120–
126 />[16] K.G. Wahlund, H.S. Winegarner, K.D. Caldwell, J.C. Giddings, Improved flow
field-flow fractionation system applied to water-soluble polymers: programming, outlet stream splitting, and flow optimization, Anal. Chem. 58 (3) (1986)
573–578 />[17] A. Einstein, Über die von der molekularkinetischen Theorie der Wärme
geforderte Bewegung von in ruhenden Flüssigkeiten suspendierten Teilchen,
Annalen der Physik 322 (8) (1905) 549–560 />19053220806.
[18] A. Litzén, K.G. Wahlund, Zone broadening and dilution in rectangular and
trapezoidal asymmetrical flow field-flow fractionation channels, Anal. Chem.

63 (10) (1991) 10 01–10 07 0 010a013.
[19] S.A. Palkar, M.R. Schure, Mechanistic study of electrical field flow fractionation.
1. Nature of the internal field, Anal. Chem. 69 (16) (1997) 3223–3229 https:
//doi.org/10.1021/ac9700134.

4. Conclusion
In this study, we investigated methods for rapid stabilization of
pH in EAF4 to maintain high reproducibility of results. It was confirmed that using an additional focusing step including the electric
field gives a more rapid stabilization of pH. Furthermore, the results show that it is crucial to use appropriate buffers as carrier
liquids to avoid large pH-changes during analyses.
In addition, no significant increase in resolution between
monomers and oligomers was observed for the investigated proteins. EAF4 was also used to determine the electrical properties
based on electrophoretic mobility, such as zeta-potential and effective net charge, over the size distribution of the investigated proteins. There are limitations for the applicability of EAF4 to proteins
such as ionic strength and buffer composition. Another challenge
is of course that many proteins are relatively small and have relatively low number of charges which limits the effect of the electric
field. It is likely that the method would be more suitable for application to larger or highly charged analytes such as large proteins,
polyelectrolytes and charged nanoparticles. Nevertheless, the results show that EAF4 made it possible to determine differences in
zeta-potential between monomers and oligomers. This shows that
EAF4 is an interesting technique for probing zeta-potentials in protein mixtures (or oligomer mixtures) yielding valuable information
which is otherwise not accessible by other techniques. It could also
be a possibility to apply EAF4 to research questions where protein
charge properties are changed as a result of binding or interacting
with other molecules.
Contributor roles taxonomy (credit)
The contribution of each author who have participated in “Separation and zeta-potential determination of proteins and their
oligomers using electrical asymmetrical flow field-flow fractionation (EAF4)” based on CRediT is shown in the table below.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to
influence the work reported in this paper.
CRediT authorship contribution statement

Jaeyeong Choi: Investigation, Validation, Writing - original
draft, Writing - review & editing, Visualization. Catalina Fuentes:
Validation, Writing - review & editing. Jonas Fransson: Resources,
Writing - review & editing. Marie Wahlgren: Writing - review &
7


J. Choi, C. Fuentes, J. Fransson et al.

Journal of Chromatography A 1633 (2020) 461625

´
˙ M. Trojanowicz, A. Wilk, P. Garstecki, R. Hołyst, Net
[20] J. Szymanski,
E. Pobozy,
charge and electrophoretic mobility of lysozyme charge ladders in solutions
of nonionic surfactant, J. Phys. Chem. B 111 (19) (2007) 5503–5510 https://doi.
org/10.1021/jp067511d.
[21] J.W. Swan, E.M. Furst, A simpler expression for Henry’s function describing
the electrophoretic mobility of spherical colloids, J. Coll. Interface Sci. 388 (1)
(2012) 92–94 />[22] G.V. Lowry, R.J. Hill, S. Harper, A.F. Rawle, C.O. Hendren, F. Klaessig, U. Nobbmann, P. Sayre, J. Rumble, Guidance to improve the scientific value of zetapotential measurements in nanoEHS, Environ. Sci. 3 (5) (2016) 953–965 https:
//doi.org/10.1039/C6EN00136J.
[23] F.J. Yang, M.N. Myers, J. Calvin Giddings, Peak shifts and distortion due to solute relaxation in flow field-flow fractionation, Anal. Chem. 49 (4) (1977) 659–
662 />[24] K. Jayaraj, R.S. Subramanian, On relaxation phenomena in field-flow fractionation, Sep. Sci. Technol. 13 (9) (1978) 791–817 />01496397808057128.
[25] J. Janca, J. Chmelik, D. Pribylova, Optimization of field-flow fractionation with
respect to relaxation and use of stop-flow technique at constant field operation, J. Liq. Chromatogr. 8 (13) (1985) 2343–2368 />01483918508076575.
[26] R. An, K. Massa, D.O. Wipf, A.R. Minerick, Solution pH change in non-uniform
alternating current electric fields at frequencies above the electrode charging frequency, Biomicrofluidics 8 (6) (2014) 064126 />4904059.

[27] J. Choi, M. Wahlgren, V. Ek, U. Elofsson, J. Fransson, L. Nilsson, A. Terry,

C..A.G. Söderberg, Characterization of binding between model protein GA-Z and
human serum albumin using asymmetrical flow field-flow fractionation and
small angle X-ray scattering, Plos One (2020) in progress.
[28] K. Rezwan, L.P. Meier, M. Rezwan, J. Vörös, M. Textor, L.J. Gauckler, Bovine
serum albumin adsorption onto colloidal Al2O3 particles: a new model based
on zeta potential and UV−Vis measurements, Langmuir 20 (23) (2004) 10055–
10061 />[29] D.C. Carter, J.X. Ho, Structure of serum albumin, in: C.B. Anfinsen, J.T. Edsall, F.M. Richards, D.S. Eisenberg (Eds.), Advances in Protein Chemistry, 45,
Academic Press, 1994, pp. 153–203. />60640-3.
[30] L.P. Kozlowski, IPC – isoelectric point calculator, Biol. Direct 11 (1) (2016) 55
016- 0159- 9.
[31] P. Arosio, T.G. Adelman, J.W. Drysdale, On ferritin heterogeneity. Further evidence for heteropolymers, J. Biol. Chem. 253 (12) (1978) 4451–4458 PMID
659425.
[32] R. Kun, M. Szekeres, I. Dékány, Isothermal titration calorimetric studies of the
pH induced conformational changes of bovine serum albumin, J. Therm. Anal.
Calorim. 96 (3) (2009) 1009–1017 009- 0040- 5.

8