REVIE W Open Access
Aqueous two-phase system (ATPS): an
overview and advances in its applications
Mujahid Iqbal
1
, Yanfei Tao
1*
, Shuyu Xie
1
, Yufei Zhu
1
, Dongmei Chen
1
, Xu Wang
1
, Lingli Huang
1
, Dapeng Peng
1
,
Adeel Sattar
1
, Muhammad Abu Bakr Shabbir
2
, Hafiz Iftikhar Hussain
2
, Saeed Ahmed
2
and Zonghui Yuan
1,2*
Abstract
Aqueous two-phase system (ATPS) is a liquid-liquid fractionation technique and has gained an interest because of
great potential for the extraction, separation, purification and enrichment of proteins, membranes, viruses, enzymes,
nucleic acids and other biomolecules both in industry and academia. Although, the partition behavior involved in
the method is complex and difficult to predict. Current research shows that it has also been successfully used in
the detection of veterinary drug residues in food, separation of precious metals, sewage treatment and a variety
of other purposes. The ATPS is able to give high recovery yield and is easily to scale up. It is also very economic and
environment friendly method. The aim of this review is to overview the basics of ATPS, optimization and its applications.
Keywords: Aqueous two-phase system (ATPS), Biomolecule separation, Solvent extraction, Veterinary drug residues
History and background
In 1896, Martinus Willem Beijerinck accidently found
the ATPS while mixing an aqueous solution of starch
and gelatin. However, its real application was discovered
by Per-Åke Albertsson. Since then, the ATPS has been
used for a range of purposes [1–3]. These systems can
be formed by mixing a variety of components in water
[4]. But two-polymer and polymer-salt (e.g., phosphate,
sulfate or citrate) systems have grown rapidly a nd a
lot of work has been put into studying this technique
using these types of ATPSs [2]. This method ha s ad-
vantages over conventional extraction techniques like,
environment-friendly, lo w cos t, capable of continuous
operation, ease of scaling-up and is efficient for many
kinds of experiments specially for the concentration
and purification of biomolecules [1, 2, 4, 5]. The use
of affinity ligands in ATPS can result in the higher
recovery yields and higher purification folds of bio-
logical products as it is a primary stage re covery tech-
nique [6]. Affinity ligands can be covalently attached
to polymer or polymer can also be modified with
hydrophobic groups [5] Interested readers about
aqueous two-phase affinity partitioning (ATPAP) are
referred to an excellent review by Ruiz-Ruiz et al. [6].
Water as the main component of both pha ses in ATPS
forms a gentle environment for biomole cules to separ-
ate and polymers stabilize their structure and biological
activities [3, 4, 7–9] while other liquid-liquid extraction
(LLE) methods could damage biological products
because of the process conditions a nd organic solvents
[1, 7]. The purpose of this review article is to over view
the technique extensively and it s applications in detail.
Types of aqueous two-phase system (ATPS)
The most common biphasic systems are formed by two
polymers (usually polyethylene glycol (PEG) and dextran)
or a polymer and a salt (e.g., phosphate, sulfate or citrate).
Other types include, ionic liquids and short-chain alcohols
[1, 2, 4, 6, 7, 10, 11]. In addition to this, ionic and/or non-
ionic surfactants are used for the formation of micellar
and reverse micellar ATPSs [6, 12, 13]. Polymer – poly-
mer/salt systems have been studied for more than five de-
cades. Polymer – polymer systems are preferably used for
the separation, recovery and purification solutes sensitive
to the ionic environment as these systems pose low ionic
strength. While, high ionic strength is the only disadvan-
tage of polymer – salts system [14]. Alcohol – salt ATPS
are inexpensive as compared to polymers and copolymers.
* Correspondence: tyf@mail.hzau.edu.cn; yuan5802@mail.hzau.edu.cn
1
National Reference Laboratory of Veterinary Drug Residues (HZAU)/MOA Key
Laboratory of Food Safety Evaluation, Huazhong Agricultural University,
Wuhan, Hubei 430070, China
Full list of author information is available at the end of the article
© 2016 The Author(s). Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and
reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to
the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver
(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Iqbal et al. Biological Procedures Online (2016) 18:18
DOI 10.1186/s12575-016-0048-8
These systems are also characterized by low viscosity,
easy constituent recovery, and reduced settling times,
but a major drawback of using this type of ATPS is
that many proteins are not compatible with alcohol-
rich phase [15, 16]. The aqu eous micellar tw o-pha se
system was first introduced by Bordier for the separ-
ation of integral membrane proteins [17]. These sys -
tems are also useful for ionic environment sensitive
solutes as nonionic surfactants can be use d for t he
formation of these systems. Mixed micellar systems
are be coming popular for showing se lectivity features
[18]. Most recently, ionic liquids (ILs) ba sed ATPSs
are being developed [19]. Poly-phase systems (three
or four polymer pha ses) also have been constructed
for the separation of biomolecules [7]. One-polymer
ATPSs have also been reported, which utilize only
one polymer for the formation of ATPS in water [20].
PEGs of differe nt molecular weight s are widely u sed
polymers in ATPS due to their low toxicity, low price
and low volatile nature [21]. Table 1 shows different
types of ATPS with representative examples.
Two-phase formation, thermodynamics and
partitioning
Miscibility of solutions containing polymers is not a com-
mon phenomenon, this property of polymers results in
the formation of two phases [3, 4, 7]. Similar incompatibil-
ity can be observed upon mixing a polymer and a high
ionic strength salt. Polymer – polymer system forms large
aggregates and because of steric exclusion, polymers start
to separate between two different phases. In polymer –
salt ATPS, salt absorbs large amounts of water and a same
exclusion phenomena can be observed [5, 22].
Phase separation in ATPS is af fected by different fac-
tors like, concentration and molecular weight (MW) of
polymer, concentration and composition of salt [21, 23].
The presence of salt also influences phase behavior
which also bring changes according to the type and con-
centration. Although, the mechanism through which salt
influence ATPS in poorly understood [1, 7]. Generally,
three forces: gravitational, flotation and frictional, act on
a drop during phase separation and the balance between
these forces determine its movement. The gravitational
force depends on the density of drops while flotation
and frictional forces depend on the flow properties of
phases [10, 24]. Surface properties of materials and com-
ponents of ATPS determine the partitioning between
two phases [7]. Poorly understood partition behavior is a
major barrier in widely adaptation of ATPS on commer-
cial levels for the purification of biomolecules [25].
Phase diagram (see Fig. 1) is like a fingerprint to a sys-
tem under specific conditions (e.g., temperature and pH)
which is unique and shows the potential working area of
ATPS. It provides a set of information like concentration
of components for two phase formation and their con-
centration in the top and bottom phases [4, 26]. The dia-
gram (see Fig. 1) shows a binodal curve (TCB), which
divides the region of component concentrations. This
curve splits the concentrations which form two immis-
cible aqueous phases (above the binodal curve) from
Table 1 Types of ATPS with representative examples
Types of ATPS Representative examples Reference
Composition of ATPS Product Results
Polymer – polymer PEG – dextran Chitinase Successful partitioning of chitinase
towards bottom phase
[161]
PEG – dextran Nanospheres, nanowires and
DNA derivatized nanowires
Successful In situ binding Au
nanospheres with Au nanowires
[162]
Polymer – salt PEG – K
2
HPO
4
B-phycoerythin Recovery yield = 90 % [163]
Purification factor = 4
PEG 4000 – sulfate + 8.8 % NaCl α-Amylase Purification = 53 fold [164]
Purity = 86 %
Alcohol – salt 2-propanol – K
2
HPO
4
Lipase Recovery yield = 99 % Purification
factor = 13.5
[165]
Ethanol – K
2
HPO
4
2,3-butanediol Recovery yield= >98 % [16]
Micellar/reverse
micellar ATPS
n-Decyl tetra (ethylene oxide) Bacteriophages Bacteriophages partitioning
towards micelle poor phase
[12]
Isooctane/ethylhexanol/methyltrioctyl
ammoniumchloride
Plasmid DNA Successful purification of DNA
and RNA removal
[166]
Ionic liquids (ILs) –
based ATPS
1-Butyl-3-methylimidazolium
chloride – salt
Codeine and papaverine Recovery yield= >90 % (codeine),
>99 % (papaverine)
[167]
Imidazolium – K
2
HPO
4
Curcuminoids Extraction yield = 96 % [168]
Purity= >51 %
Iqbal et al. Biological Procedures Online (2016) 18:18 Page 2 of 18
those that make one phase (below the binodal curve).
The line (TB) in the diagram (see Fig. 1) is a tie line; it
connects two nodes, which lie on the binodal curve. All
the potential systems (e.g., S1, S2, S3) have same top
phase and bottom phase equilibrium composition be-
cause of being on the same tie line. Point C on binodal
is called as a critical point, just above this point the vol-
ume of both phases is theoretically equal. At point C the
value tie line length (TLL) is equal to zero. The tie line
length and component concentration has same units.
The tie line length can be estimated by using the weight
ratio as shown in equation below;
V
t
ρ
t
V
b
ρ
b
¼
SB
ST
ð1Þ
Where V and ρ stands for volume and density of top (t)
and bottom (b) phases while SB and ST are segments
lengths as shown in Fig. 1.
Or by the analysis of top and bottom phase, which is a
more precise method;
TLL ¼
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ΔX
2
þ ΔY
2
p
ð2Þ
The tie lines are commonly straight and the slope of tie
line (STL) can be calculated with the help of equation 3.
This is also helpful in the construction of further tie lines.
STL ¼
ΔY
ΔX
ð3Þ
Binodal can be determined by three methods; turbido-
metric titration method, cloud point method and node de-
termination method [4, 26]. The researchers new to
technique are referred to references [3, 4, 8, 27] for prede-
termined phase diagrams and methods for construction.
The equilibrium relationship between the top and the
bottom phase of the ATPS determine the partition of bio-
molecules and the partition coefficient (K) is defined as;
K ¼
Conc:
AT
Conc:
AB
ð4Þ
Where Conc.
AT
is the concentration of component A
in top phase and Conc.
AB
is the concentration of A in
the bottom phase at equilibrium [26]. So far, different
models have been devised by the researchers to under-
stand the partitioning in ATPS [25, 28–32]. There is no
good comprehensive theory of liquid and liquid mix-
tures. In result, these models are based on the combin-
ation of different theories which makes a range of
possible out-comes possible [33]. Albertsson’s model has
been used commonly to describe ATPSs. He suggested
six different kinds of partitions, each having a different
kind of driving force [34]. According to his model,
Fig. 1 Schematic representation of phase diagram. Concentrations above binodal curve (TCB) forms aqueous two-phase system
Iqbal et al. Biological Procedures Online (2016) 18:18 Page 3 of 18
partition behavior is determined by these factors, separ-
ately or collectively and the manipulation of some of
these factors would dominate the overall behavior.
i. Electrochemical – where electrical potential drives
the partition
ii. Hydrophobicity – where hydrophilic properties of
molecules and phases determine the separation
iii. Bio-specific affinity – this kind of partition occurs
when required molecules binds to a specific site on
polymer
iv. Size – molecular size or surface area of molecules is
dominating factor
v. Conformation dependent – where partition depends
on the conformation of the molecule
The logarithmic form of the factors of partition coeffi-
cient can be expressed in equation (5).
ln K ¼ ln K
o
þ ln K
elec
þ ln K
hfob
þ ln K
affinity
þ ln K
size
þ ln K
conf
ð5Þ
Where elec stands for electrochemical, hfob, affinity,
size and conf denote as hydrophobic partitioning, affinity
partitioning and conformation while K
o
include all other
factors (e.g., environmental factors) [1, 3–5, 34]. Differ-
ent theoretical and experimental models have been pub-
lished by researchers. As, Andrews and Asenjo consider
hydrophobicity a s the main ruling factor of partition in
polymer – pol ymer and polymer – salt ATPSs for pro-
tein [5, 35, 36].
Factors influencing partitioning in ATPS
Since the partitioning mechanism in ATPS is still un-
known. Most of the ATPSs are optimized according to the
physicochemical properties of solutes of biomolecule. Dif-
ferent review articles [1, 5] and books [3, 27, 37–39] dis-
cuss these factors in detail. Main factors influencing
partition behavior in ATPS are:
Molecular weight (MW) and concentration of polymer
As most of the ATPSs are composed of polymer – poly-
mer/salt. MW of polymers greatly influence the parti-
tion. Generally,
↑ MW of polymer →↓ concentration of polymer
required for phase formation
↑ Differences between the MW of polymers →↑
asymmetrical curve of the phase diagram
↑ MW of PEG →↓ value of K
In a polymer – salt system, partition towards polymer-
rich phase decreases upon increasing the concentration
of polymer while in a polymer – polymer system
partition decreases towards phase having high MW poly-
mer. The main reason behind this phenomenon is the
increase in the steric exclusion of biomolecule from that
phase or because of changes in the hydrophobicity of
phases [1, 5, 27] As increase in the MW of polymer in-
creases hydrophobicity by reducing the hydrophilic
groups/hydrophobic area [5].
Hydrophobicity
Hydrophobicity play an important role in the partition-
ing of protein. Two main effects: phase hydrophobicity
effect and salting out effect, are involved in hydrophobic
interactions [5, 40, 41]. In polymer – salt system s,
hydrophobicity may be man ipulated by varying TLL,
MW of polymer and by adding a salt (e.g., NaCl). The
low NaCl concentrations (<1 M) do not affect ATPS
however, high salt concentrations (>1 M) changes the
phase diagram [35]. The addition of salt in ATPSs ha s a
significant effect on the partition coefficient [42]. These
salts contain ions of different hydrophobicities and the
hydrophobic ions force the partitioning of counter ions
to phase with higher hydrophobicity and vice versa. The
salting-out effect moves the biomolecule from salt-rich
phase to polymer-rich phase [26].
pH
The pH of ATPS may alter the charge and surface prop-
erties of solute which affects the partitioning of biomol-
ecule. The net charge of the protein turns negative in
case of higher pH than the isoelectric point (pI) and
positive if lesser than pI. If the pH is equal to pI, net
charge will be zero [26]. It has been reported that the
partitioning of negatively charged biomolecule in a
higher pH system increases the partition coefficient and
target biomolecule prefers top phase. Higher pH values
than pI of protein induce an affinity towa rds PEG-rich
phase because of the positive dipole moment [40, 43].
Temperature
Temperature greatly affects the composition of two phases
in an ATPS, hence, the phase diagram. The changes in
temperature also affect partition through viscosity and
density. Therefore, it is always recommended by the re-
searchers to have a strict control of temperatures in ATPS
related experiments. In general, phase separation is ob-
tained at lower temperature in a polymer – polymer ATPS
with lower concentrations of polymer, however, an oppos-
ite effect is seen in polymer – salt system [27].
Partitioning behavior of biomole cule and phase sep-
aration rate is also influenced by the physico-chemical
properties (i.e., density, vi scosity and interfacial ten-
sion) of ATPS. Measurement of such proper ties have
been explained by Albert sson [3], Za slavsky [37] and
Hatti-Kaul [4].
Iqbal et al. Biological Procedures Online (2016) 18:18 Page 4 of 18
Optimization of aqueous two-phase system
Since the partitioning behavior of biomolecules in ATPS
is complex, many laborious trials have to be performed
for the optimization of these systems. This optimization
leads to an increased overall-cost [44]. One conventional
way to optimize ATPS is a one-factor/variable-at-a-time
(OFAT) in which specific factors are identified to study,
but the major disadvantage of OFAT approach is not
considering the interaction between the factors as the
name indicate, one factor is studied at a time while
keeping all factors consta nt. This usually results in the
poor and false optimal conditions [26, 44, 45]. Now-
adays, a multivariate statistical tec hnique is used for
the optimization of ATPS called “Design of Experi-
ments (DoE)” . DoE consist s of fe w experiment s at a
specific factor level combination [26, 44]. The general
steps in a DoE are:
Screening of variables
First step in a DoE process is the screening of significant
factors (k), which demands further investigation because
of their great influences on the out responses [26, 45,
46]. This is usually done by full factorial design (FFD)
and fractional factorial design (fFD). In these designs all
factors (k) are assigned two levels, high (+) and low (− ).
In FFD, experiments are carried out at different combi-
nations of the factors with a total number of 2
k
. For in-
stance, the number of factor is 2 (e.g., A and B). Then
the possible number of experiments to be conducted is
2
2
= 4 resulting in the combinations of (−,−), (+,−), (−,+)
and (+,+). No doubt, this factorial design gives high ac-
curacy results along with the possible interaction be-
tween the factors, but the number of experiments will
be more in case of more factors (e.g., 2
5
) [26, 44–46]. To
control this drastic increa se in the number of experi-
ments to one-half, one- quarter or a higher fraction of
full fraction, fFD is used, which is denoted as 2
k-1
,2
k-2
and 2
k-4
. Another method used for the screening is
Plackett-Burman design (PBD), a linear screening ap-
proach, used when only main influences are of interest
[26, 44]. This could be represented in the form of the
equation as follows
y ¼ β
0
þ
X
k
i¼1
β
i
X
i
þ ε ð6Þ
(Here, y = predicted response variable, β
0
and β
i
=
coefficient of regression, X
i
=experiment factor and ε =
random error).
Finally, in all three screening designs (FFD, fFD and
PBD) the magnitude of the significant factors is analyzed
by using analysis of variance (ANOVA) [26, 44, 47].
Initial optimization
After the screening of factors/variables, the next step is
to confirm the optimum level of these factors. As it is
important to check that the factors are ne ar to the opti-
mal experimental region. This is done by an analysis of
the model curvature after adding few cente r points
experiments to screening model. Upon significant differ-
ence between these center point experiments and aver-
age out-put responses a model curvature occurs. This
means the responses are situated in the optimal region
and can be optimized in the next step which is not pos-
sible in case of no significant difference [26, 44, 45]. If
no curvature exists, steepest ascent and steepest descent
experiments are performed to increase or decrease the
out-put responses to reach the proximity of optimal re-
gion [26, 44, 48]. Steepest ascent/descent experiments
are useful in the determination of experimental di rec-
tion. These experiments are initially performed at the
center point of the significant factors and each factor
level is increased or decreased in accordance with the
magnitude of main effects. [26, 44]. In addition to this,
these experiments have to be performed until no more
increase in the out responses is observed, thus the gen-
eral vicinity of the optimal experimental region can be
drawn from the maximum response point of these ex-
periments. Finally, these points can be taken as center
points for final optimization [26, 44].
Final optimization
Response surface methodology (RSM) is used for the
final optimization of significant factors. Box and Wilson
first reported this optimization approach [49]. This
methodology is useful in the determination of optimal
operating conditions and significant independent factors
or their interactions with dependent output responses in
the multivariate complex system (e.g., ATPS) [44, 50].
RSM is helpful in the prediction of respons es by inves-
tigating an optimal experimental region and collecting
experimental data which fits quadratic equation/sec-
ond-order polynomial model [26, 44]. In this context,
regression analysis is performed to selec t the best data
representing equation and then output responses are
analyzed by surface or contour plots [51]. Central com-
posite de sign (CCD) and Box-Behken design (BBD) are
two different multilevel designs of RSM but the CCD is
most commonly used because of rotatability and uni-
form precision [45, 51]. CCD is also an ex tensively used
model for the optimization of ATPS (PEG – Salt) [44, 51].
However, according to Raja et al., BBD has advantages
over CCD, like, less number of experiments are done in
BBD as compared to CCD and there are 3 factor levels in
BBD while 5 in CCD [26]. Finally, the experimental data
of these methods are used for fitting a full quadratic
model and analyzed by regression analysis [26, 44].
Iqbal et al. Biological Procedures Online (2016) 18:18 Page 5 of 18