J. Sep. Sci.

2016, 39, 655–662 655

Vivek Rajendran1 Research Article

Kirubhakaran Puvendran2
Bharath Raja Guru1
Guhan Jayaraman2
1 Department of Biotechnology,
Manipal Institute of Technology,
Manipal, Karnataka, India
2 Department of Biotechnology,
Bhupat and Jyoti Mehta School
of Biosciences, Indian Institute
of Technology Madras, Chennai,
India
Received August 28, 2015
Revised November 14, 2015
Accepted November 16, 2015
Design of aqueous two-phase systems for
purification of hyaluronic acid produced by
metabolically engineered Lactococcus lactis
Hyaluronic acid has a wide range of biomedical applications and its commercial value is
highly dependent on its purity and molecular weight. This study highlights the utility of
aqueous two-phase separation as a primary recovery step for hyaluronic acid and for removal
of major protein impurities from fermentation broths. Metabolically engineered cultures of
a lactate dehydrogenase mutant strain of Lactococcus lactis (L. lactis NZ9020) were used to pro-
duce high-molecular-weight hyaluronic acid. The cell-free fermentation broth was partially
purified using a polyethylene glycol/potassium phosphate system, resulting in nearly 100%
recovery of hyaluronic acid in the salt-rich bottom phase in all the aqueous two-phase sep-
aration experiments. These experiments were optimized for maximum removal of protein
impurities in the polyethylene glycol rich top phase. The removal of protein impurities re-
sulted in substantial reduction of membrane fouling in the subsequent diafiltration process,
carried out with a 300 kDa polyether sulfone membrane. This step resulted in considerable
purification of hyaluronic acid, without any loss in recovery and molecular weight. Diafil-
tration was followed by an adsorption step to remove minor impurities and achieve nearly
100% purity. The final hyaluronic acid product was characterized by Fourier-transform IR
and NMR spectroscopy, confirming its purity.

Keywords: Aqueous two-phase Systems / Diafiltration / Hyaluronic acid / Lacto-

coccus lactis
DOI 10.1002/jssc.201500907

 Additional supporting information may be found in the online version of this article
at the publisher’s web-site

1 Introduction Commercial grade HA is traditionally produced by ex-

Hyaluronic acid (HA) is a nonsulfated glycosaminoglycan
composed of repetitive, alternating units of glucuronic acid
(GlcUA) and N-acetyl glucosamine. In animal tissues, HA
is involved in cell proliferation, morphogenesis, metastasis,
inflammation, wound healing, signal transduction, and cel-
lular interactions through cell surface receptors [1,2]. HA is a
high-molecular-weight hydrophilic polymer with high water-
binding capacity and biocompatibility. These unique prop-
erties have made HA a commercially important biopolymer
with applications in cosmetics, food, and healthcare indus-
tries [3]. The purity of HA is therefore an important criterion
for its commercial applications.
Correspondence: Dr. Guhan Jayaraman, Department of Biotech-
nology, Bhupat and Jyoti Mehta School of Biosciences, Indian
Institute of Technology Madras, Chennai 600 036, India
E-mail: guhanj@iitm.acin
Abbreviations: ATPS, aqueous two-phase systems; GPC, gel-
permeation chromatography; HA, hyaluronic acid; NMWCO,
nominal molecular weight cutoff refers to the lowest molec-
ular weight solute (Da) in which 90% of the solute is retained
by the membrane; UF, ultrafiltration
traction from rooster combs. However, this method has its
limitations due to the high purification cost, cross-viral con-
tamination, immunogenic, and ethical issues. HA is also
naturally present in microbes such as Streptococcus and Lac-
tobacillus species [3-7]. Due to the potential pathogenicity of
many natural HA producers, microbial production of HA has
been extended to metabolically engineered organisms such
as Bacillus subtilis [8, 9], Escherichia coli species [10-12], Pichia
pastoris [13], and Lactococcus lactis [14-19].
Previous work in our laboratory has focused on metabolic
engineering of the GRAS microorganism, L. lactis, for HA
production [16-18]. The HA biosynthetic pathway genes were
isolated from the has-operon of Streptococcus zooepidemicus
and cloned into nisin-inducible expression systems in L. lac-
tis [17, 18]. However, the yield and molecular weight of HA
obtained from the recombinant L. lactis cultures was lower
than from S. zooepidemicus. Moreover, the yield of HA was
found to vary inversely with the yield of lactate produced
in the L. lactis cultures [20]. Therefore, the heterologous
has-genes were transformed into an ldh-mutant strain of
L. lactis (L. lactis NZ9020). It was found that the recombi-
nant L. lactis NZ9020 strain carrying heterologous hasABD
genes produced HA with much higher molecular weight


C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.jss-journal.com
656 V. Rajendran et al.
Table 1. Literature reports on hyaluronic acid purification
Strains used Process strategy
S. zooepidemicus Solvent precipitation, multiple adsorption,
diafiltration, microfiltration
S. zooepidemicus Diafiltration
S. zooepidemicus Solvent precipitation, acid precipitation,
charcoal adsorption, diafiltration
S. zooepidemicus Diafiltration, adsorption
S. zooepidemicus Diafiltration
Streptococcus Adsorption using imprinted beads,
equi RSKK 679 desorption of HA
L. lactis NZ9020 ATPS, diafiltration, adsorption
(2–2.5 MDa) than the lactate-producing recombinant strain
L. lactis NZ9000, carrying the same combination of has-
genes [Mandeep Kaur and Guhan Jayaraman, unpublished
data]. Therefore, the HA purification study reported in this
manuscript is based on the high molecular weight HA pro-
duced by the recombinant ldh-mutant strain L. lactis NZ9020,
expressing the three heterologous hasABD genes obtained
from the S. zooepidemicus biosynthetic pathway [21]. Prior
reports on HA purification have been from cultures of the
natural producer, S. zooepidemicus (Table 1). To the best of
our knowledge, this is the first literature report on purification
of HA produced by a recombinant bacterial culture.
There are relatively few literature reports on HA purifi-
cation from microbial fermentation broths (Table 1). Com-
monly reported downstream processing techniques for HA
purification include adsorption, precipitation, and diafiltra-
tion. Oueslati et al. and Zhou et al. [22, 23] have exclusively
used diafiltration technique for HA purification, with 90% re-
covery and purity. Our investigations with ultrafiltration (UF)
as the initial HA purification step showed that the mem-
branes are susceptible to fouling due to the high amount
of protein impurities from the fermentation broth. Protein
loading leads to pore plugging, which can cause decline in
flux during the purification process [24, 25]. Removal of ma-
jor amounts of protein impurities before diafiltration reduces
or eliminates membrane fouling. Other workers have used
prepurification steps such as adsorption [26], and acid precip-
itation followed by adsorption [27] before the diafiltration step
(Table 1). However, these methods have resulted in poor HA
recovery (65%). Direct processing of fermentation broth
by diafiltration followed by adsorption for HA purification
results in recovery of about 60% [28]. Another study has ex-
ploited the interaction of hydroxyl groups of D-glucuronic acid
with metal ions to form metal–chelate complex. This tech-
nique was used to separate out HA using imprinted beads,
which had adsorption capacity of about 810 mg/g [29]. How-
ever, synthesis of imprinted beads is a complex and expen-
sive process. This study investigates the application of aque-
ous two-phase systems (ATPS) as a preliminary extraction
step before diafiltration. Most downstream processes involv-
ing ATPS follow it up with a diafiltration step to remove the

C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
J. Sep. Sci. 2016, 39, 655–662
HA recovery (%) HA purity (%) Research group
[Ref. no.]
64 >99 [26]
89 NA [23]
72 99.2 [27]
60 >99 [7]
90 90 [22]
NA NA [29]
83 >99 [This study]
excessive amount of phase-forming polymers or salts. Since
HA is a high-molecular-weight polymer, it can be easily sep-
arated from the phase-forming salts as well as remaining
protein impurities during diafiltration.
ATPS involve separation of two predominantly aque-
ous, polymer/salt-rich phases where product recovery, en-
richment, and purification can be achieved in a single step.
It is particularly advantageous for separation of proteins and
hydrophilic polysaccharides such as HA. Aqueous two-phase
separation of proteins and monoclonal antibodies using a
PEG/salt system has been made selective by adding natural
compounds such as betaine [30] and sodium chloride [31].
ATPS is usually coupled with techniques such as diafiltra-
tion, dialysis, and chromatographic techniques to improve
the purity of the target molecule. It is an effective method
for large-scale processing of proteins, enzymes, antibodies,
and antibiotics extraction [32-36]. Continuous countercur-
rent ATPS has been employed in separation of human im-
munoglobulin G from the supernatant of Chinese hamster
ovary (CHO) cultures [37, 38]. Multistage extraction using
ATPS in combination with UF has been used for purification
of C-phycocyanin [39]. Partitioning of biomolecules in ATPS
is influenced by many variables such as type of salts and poly-
mer used for phase separation, molecular weight of polymer,
pH, temperature, ionic strength and physicochemical proper-
ties of the biomolecule [36–38]. Optimization of these factors
is essential to achieve effective use of ATPS for biomolecular
separations [37]. Central composite design based optimiza-
tion made ATPS a promising primary step in monoclonal
antibody separation from culture supernatants [34].
Most literature reports on applications of ATPS have fo-
cused on protein purification. There are very few reports on
separation of polysaccharides using ATPS and none on HA
purification using ATPS. Two reports on polysaccharide pu-
rification using ATPS include purification of high molecular
weight polysaccharides from the root of Achyranthes biden-
tate, with 55% recovery [40] and extraction of aloe polysac-
charides from leaf extract using ATPS coupled with UF [41].
The present study will be the first report on incorporation
of ATPS in a downstream processing scheme for HA purifi-
cation. In this study, ATPS systems with varying PEG/salt
www.jss-journal.com
J. Sep. Sci. 2016, 39, 655–662
compositions and phase ratios were studied to optimize HA
recovery and removal of protein impurities. The optimally
designed ATPS step was followed by diafiltration with a
300 kDa UF membrane and finally, an adsorption step to
remove minor impurities.
2 Materials and methods
2.1 Strains and plasmids
The strain used was L. lactis NZ9020, a lactate dehydroge-
nase (ldh) mutant in which two of three lactate producing
genes have been knocked out [42]. It was procured from
NIZO Food Research (Netherlands). The plasmid (pSJR6)
is a nisin-inducible expression (NICE) system that contained
hasA, hasB, and hasD heterologous genes, cloned from the
has-operon of S. zooepidemicus [17]. This plasmid was trans-
formed successfully by electroporation method in L. lac-
tis NZ9020. Chloramphenicol (10 ␮g/mL) and tetracycline
(2 ␮g/mL) were used for strain selection.
2.2 Culture media
Reagents and media components used for fermentation and
analysis were procured from Himedia Laboratories (India).
The culture medium was composed of brain heart infusion
5 g/L, yeast extract 5 g/L, ascorbic acid 0.5 g/L, dipotassium
hydrogen phosphate 1.5 g/L, potassium dihydrogen phos-
phate 0.5 g/L, and magnesium sulfate 0.5 g/L. The medium
pH was maintained at 7.0 throughout the fermentation us-
ing 5 M potassium hydroxide. All the fermentation experi-
R 
ments were carried out in a 2.4 L Bioengineering bioreactor
(Switzerland) with 1.0 L working volume, 1 vvm aeration and
200 rpm agitation.
2.3 Pretreatment of fermentation broth
Fermentation broth with HA and other impurities was sep-
arated from cell debris by centrifugation at 10 000 rpm for
20 min at room temperature. The supernatant was then sub-
jected to microfiltration using 0.45 ␮m membrane (Sartorius,
India) for effective removal of micro particles.
2.4 ATPS
Chemicals used for ATPS were procured from Sisco Re-
search Laboratories (Mumbai, India). PEG 6000/potassium
phosphate combination system was used for phase separa-
tion and the binodal curve was constructed as described [43].
Experiments were carried out with different combinations of
PEG/salt concentration at three different tie lines (S, M, L)
as shown (Fig. 1). Four different phase points on each tie
line (S1, S2, S3, S4, M1, M2, M3, M4, L1, L2, L3, L4) were

C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Other Techniques 657
chosen to study the effect of phase composition on partition
behavior.
Predetermined weights of PEG 6000 and potassium phos-
phate were added to cell-free fermentation broth containing
HA, so as to make the system 100% on w/w basis. All the
components were mixed well with continuous stirring. Ex-
periments were conducted at room temperature and pH was
maintained at 7 (dibasic: monobasic, 1.82:1). The system was
allowed to equilibrate by keeping the mixture in separating
funnel. Phase volume ratio was calculated by measuring both
top and bottom phase volume and samples from both the
phases were collected for further analysis.
2.5 Diafiltration
Polyethersulfone UF membrane cassette (Minimate TFF cap-
sule, PALL, USA) (50 cm²) with a nominal molecular weight
cutoff (NMWCO) of 300 kDa was used for diafiltration. The
concentrated bottom phase from the ATPS was diluted back
to its original volume so as to maintain high permeate flux.
This solution was subjected to continuous mode diafiltration
at room temperature with water as diluent. Both the trans-
membrane pressure and cross-flow rate were maintained at
around 1.0–1.5 bar and 30–40 mL/min, respectively. Sam-
ples were collected after each dia-volume for analysis. During
the course of diafiltration process parameters such as reten-
tion and purity were calculated [22]. Membrane washing and
storage was done using 0.1 N sodium hydroxide.
2.6 Adsorbents treatment
Diafiltered broth was treated with 1% activated charcoal for
2–3 h with continuous stirring. The adsorbent was then re-
moved by passing the solution through a 0.22 ␮m cellulose
acetate filter (Pall, USA). The final treated HA solution was
precipitated using iso-propyl alcohol (1:2) and then subjected
to lyophilization using freeze drier (Lark, India) to obtain
powdered form of pure hyaluronic acid.
2.7 Estimation of impurities
Protein estimation was done by using BCA protein assay kit
(Thermo scientific, US). The amount of residual nucleic acid
was determined by measuring the optical density of solution
at 260 nm using Bio photometer (Eppendrof, Germany).
2.8 Hyaluronic acid analysis and characterization
Samples were subjected to ethanol precipitation at 4⬚C for
1 h and the precipitate was dissolved in 0.9% sodium chlo-
ride. The concentration of HA was estimated by modified
carbazole method [44]. The molecular weight of hyaluronic
acid was estimated using HA-standards (CalBiochem, San
www.jss-journal.com
658 V. Rajendran et al. J. Sep. Sci. 2016, 39, 655–662

Figure 1. Phase diagram for PEG: potassium phosphate ATPS; S: small tie line; M: medium tie-line; L: long tie-line.

Table 2. Effect of phase composition on separation of proteins and hyaluronic acid

Tie line PEG/ Volume Total-protein Protein removal HA concentration HA purity HA recovery
points salt (g/g) ratio partition coefficient (top phase, %) (bottom phase, g/l ) (%) (%)

S1 18/7 2.90 0.72 71.12 2.9 29.4 97

S2 15/8.4 1.63 0.99 58.09 1.9 21.2 95
S3 11.7/10 0.87 0.53 31.56 1.4 16.0 99
S4 7.5/12 0.43 0.63 21.26 1.1 13.9 99
M1 26.2/7 2.90 0.56 63.24 2.7 25.2 99
M2 18.1/11 1.14 0.77 56.44 1.5 15.5 98
M3 14.1/13 0.75 0.82 42.15 1.1 14.1 88.7
M4 9/15.5 0.40 0.83 28.85 1.0 12.3 97.5
L1 25/11 1.73 0.77 57.07 1.6 21.4 90
L2 18.5/14 0.86 0.79 40.43 1.3 17.2 98
L3 12/16.8 0.48 0.81 27.97 1.1 13.8 99
L4 8/18.8 0.32 0.63 16.48 1.0 11.9 97

Diego, USA) with molecular weights ranging from 0.6 to
1.8 MDa. Molecular weight analysis was carried out by gel-
permeation chromatography (GPC) using a PolySep-GFC-P
6000 300 × 7.8 mm column and detected using refractive in-
dex detector prefitted within the Shimadzu HPLC (Model
CBM 20A, Japan). The mobile phase used was degassed
0.2 M NaNO3 , and operated at a flow rate of 0.6 mL/min.
The column temperature was maintained at 28⬚C.
Final purified HA samples were characterized by using
FTIR and NMR spectroscopy. Sample preparation for FTIR
and NMR spectroscopy was done using the protocol described
previously [27]. FTIR spectrometer (Perkin Elmer, USA) with
scan range of 450–4500cm−1 was used for IR spectroscopy.
Proton (1 H) NMR was carried out using a Bruker AVANCE
III 500 MHz (USA), and deuterated water (D2 O) was used as
solvent.


C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.jss-journal.com
J. Sep. Sci. 2016, 39, 655–662 Other Techniques 659

Figure 2. (A) Variation of HA and protein concentration with volume of liquid diafiltered (expressed as number of feed-reservoir volumes
passed through the UF membrane). () HA concentration (g/L); (•) total protein concentration (g/L). (B) Variation of purity of HA with volume
of liquid diafiltered.

Figure 3. FTIR spectra of standard HA (black line) and purified HA (red line).

3 Results and discussion
Batch bioreactor cultures of recombinant L. lactis NZ9020
produced around 0.8–1 g/L of HA. The molecular weight of
HA measured using GPC was found to be >1.8 MDa. The
GPC column gave a retention time of 13.2 min for the largest
HA standard used (1.8 MDa), and retention time of 12.8 min
for purified HA. Protein concentration in the final fermen-
tation broth was around 5–6 g/L. The initial purity of HA in
the fermentation broth was found to be 13–14%. A schematic
of the downstream processing scheme adopted in this study
is shown in Supporting Information Fig. S1, involving aque-
ous two-phase separation, diafiltration, and adsorption as the
major purification steps.


C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.jss-journal.com
660 V. Rajendran et al.
Figure 4. NMR spectra of standard HA (black line) and purified HA (red line).
3.1 Hyaluronic acid recovery by ATPS
The effect of ATPS phase composition on HA recovery, HA
purity, and removal of protein impurities is shown in Table 2.
HA purity was calculated based on the concentrations of HA
and total protein (major impurities) present in the salt-rich
bottom phase. The phase compositions were chosen based
on three different tie-lines having different lengths, short (S),
medium (M), and long (L) (Fig. 1). Four points were chosen
on each tie-line, corresponding to different phase composi-
tions (Table 2). It was observed with every ATPS experiment
that almost all the HA partitioned into the salt-rich bottom
phase. The HA recovery in the bottom phase was >95% in
all experiments, with negligible amount of HA detected in
the PEG-rich top phase. However, the phase compositions
had a significant effect on removal of protein impurities and
thereby, the purity of HA obtained in the bottom phase.
The partition coefficient for the total protein varied from
0.53 to 0.99 across all the tie-lines, indicating that most
of the protein also tends to partition preferentially into the
bottom phase. However, by manipulating the phase volume
ratios, one can ensure substantial removal of the total protein
into the top phase. For example, the S2 phase composition
has a higher partition coefficient than S1; however, the S1
has a higher total protein removal than S2 due to the higher
phase volume ratio of S1 (Table 2). With each tie-line, the

C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
J. Sep. Sci. 2016, 39, 655–662
volume ratio of the top phase to bottom phase increased with
increasing total PEG/salt ratio in the two phases. Therefore,
it is important to increase the PEG/salt ratio in order enhance
the total amount of proteins removed into the top phase. It
can be seen from Table 2 that the PEG/salt ratio (and the
total amount of PEG and salt) used to achieve a given volume
ratio increases with the length of the tie-line, as one moves
from the S tie-line to the M and L tie-lines. Therefore, it is
desirable to use phase compositions in the S tie-line (or in
tie-lines close to the critical point), in attempting to maximize
phase volume ratio and protein partition coefficients.
HA purity was found to be highest at 30% (2.9 g/L HA
and 7 g/L total protein in the bottom phase) in the S1 phase
composition. Since the phase volume ratio is nearly 3:1, there-
fore removal of proteins is higher at S1 (70%), compared
to other phase compositions. The conditions S1 and M1 have
the same phase ratio; however, the usage of PEG is substan-
tially less in S1. Moreover, the higher partition coefficient in
S1 gives rise to higher removal of total protein. Finally, the
low salt concentration in the bottom phase of S1 allows for
lower number of diafiltration volumes in the subsequent step
of downstream processing. The RSD for protein removal was
around 43% for the phase compositions along the small (S)
and long (L) tie-lines, while it was around 28% for the phase
compositions along the medium (M) tie-line. However, phase
compositions are insensitive to HA recovery.
www.jss-journal.com
J. Sep. Sci. 2016, 39, 655–662
3.2 Diafiltration
The ATPS separation step improved the purity of HA from 12
to 30%, thereby reducing the protein load before diafiltration.
Diafiltration was effective in removing remaining protein im-
purities and salt present in the bottom phase of ATPS. The
protein concentration decreased more than 80-fold from 1.7
to 0.02 g/L, after seven volumes of diafiltration (Fig. 2A).
The purity of HA increased from 30 to 98% at the end of di-
afiltration (Fig. 2B), without any significant loss of HA (Fig.
2A). The small amount of protein impurities remaining af-
ter diafiltration gave rise to a slightly turbid solution, which
could be only clarified in a subsequent adsorption process.
At the end of diafiltration, retention of HA (RHA ) was
found to be 0.99, which is not surprising on considering the
large difference between the membrane NMWCO (300 kDa)
and HA molecular weight (>1.8 MDa) in the sample. Similar
results have been observed for retention of 1.5 MDa HA with
a 100 kDa membrane [23]. This study has shown for the
first time that a 300 kDa NMWCO membrane can be used
for diafiltration, which can potentially result in much higher
permeate fluxes than a 100 kDa membrane reported for other
HA diafiltration studies [19]. Membrane filtration does not
affect the polymer quality, unlike traditional methods such as
acid precipitation [45], and are easier to scale up.
3.3 Adsorption
The diafiltered solution was mixed with activated charcoal,
which was effective in removing remaining protein and trace
impurities by adsorption. This removed the turbidity and gave
rise to a transparent solution. The concentration of proteins
and nucleic acids in solution obtained after adsorbent treat-
ment was almost negligible. The purity of HA increased
further and reached nearly 100%. Although there is no
significant loss of HA during ATPS and diafiltration, the char-
coal treatment results in about 8–10% HA loss. The molecular
weight of HA remained unaffected throughout the purifica-
tion process. The final recovery of HA is around 83%, which is
much >60–65% recovery reported in the literature [26–28].
3.4 Characterization of hyaluronic acid
The HPLC chromatograms for HA standard and purified
HA [46] are shown in Supporting Information Fig. S2. We
also characterized the purified HA by FTIR and NMR spec-
troscopy. The superimposed FTIR spectra of the standard and
purified HA are shown in Fig. 3. The relative position of the
FTIR peaks due to various bond vibrations in the fingerprint
and diagnostic region were found to be similar for both stan-
dard and purified hyaluronic acid (Supporting Information
Table S1). The spectra confirm that the compound obtained
after purification is hyaluronic acid and provides information
about its purity.
The 1 H NMR spectrum of standard HA (Fig. 4) has four
sets of peaks, which includes a singlet at 1.9 ppm and broad

C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Other Techniques 661
peaks at 3.5, 4.3, and 4.8 ppm (D2 O). A singlet at 1.9 ppm
is due to highly shielded three protons present in the methyl
(CH3 ) group and the peaks in the range 2.9–3.9 and 4.3 ppm
are due to C–H ring protons, which are slightly deshielded.
Similar peaks with same chemical shift are present in the
spectra of purified HA (Fig. 4). However, the spectra of pu-
rified HA also have three additional peaks at 1, 2.6, and 3.8
ppm due to 2-propanol, which was used during sample prepa-
ration (precipitation) of HA. No other interfering peaks are
present in the spectra of purified HA, which indicates that the
final product is free from impurities present in fermentation
and ATPS separation processes.
4 Concluding remarks
This study demonstrated the utility of ATPS for purification
of HA produced by recombinant microbial cultures and is
the first report of its kind in literature. It was observed that
ATPS followed by diafiltration and adsorption (Supporting
Information Fig. S1) is effective in obtaining pure HA with a
higher recovery than erstwhile reported in literature. The use
of ATPS as a primary purification step avoids the complete
dependence on membrane filtration for removal of proteins,
which are present in much higher concentration as compared
to HA in the fermentation broth. A large fraction of protein
impurities can be removed using ATPS, while recovering al-
most all the HA in this step. Removal of substantial amount
of protein by ATPS will have a positive impact on the sub-
sequent diafiltration step, in terms of increasing permeate
flux (which varies inversely with solute concentration) as well
as reduction of membrane fouling. Although diafiltration is
commonly used as a stand-alone technique for purification of
high MW HA, the coupled system of ATPS diafiltration is a
better method, especially when the entire product partitions
into the salt-rich bottom phase. Furthermore, since HA re-
covery is nearly 100% in the ATPS step, a multistage ATPS
extraction process can be used to maximize protein removal.
This would make the diafiltration step even more efficient.
Finally, the ATPS separation process is independent of up-
stream process conditions, i.e. it could be applied for protein
removal from any kind of fermentation process or media as
well as for 100% HA recovery in this step. This method can
also be broadly translated to purification of other polysaccha-
rides produced by fermentation.
The use of 300 kDa UF membrane is the highest
NMWCO reported so far and allowed nearly 100% HA re-
covery. Such membranes will also provide higher permeate
fluxes and lower processing times than 100 kDa UF mem-
branes, which have been used in other studies [23]. Permeate
flux can be increased using high tangential flow velocity in
a lower MW cut off membrane unit, but some reports have
shown that increasing flow rates have resulted in lowering the
HA MW due to shearing of hyaluronic acid [23]. Therefore,
the use of a higher MW cut off membrane is desirable.
The characterization studies by FTIR and NMR spec-
troscopy showed that highly pure HA was obtained by the
www.jss-journal.com
662 V. Rajendran et al.
three-stage purification process outlined in this work. Further
process optimization and scale-up studies would be carried
out in future to make it a commercially attractive downstream
process for HA purification.
We acknowledge the Department of Biotechnology (Gov-
ernment of India) for the funding under Project No.
BT/IN/Finland/30/GJ/2013, which also provided a Fellowship
to Mr. P. Kirubhakaran. We thank the Sophisticated Analyti-
cal Instrumentation Facility (IIT Madras) for providing FTIR
and NMR facilities. We thank NIZO (Netherlands) for provid-
ing us L. lactis NZ9020 strain and Dr. Shashi Bala Prasad for
recombinant plasmid containing has-operon genes.
The authors have declared no conflict of interest.
5 References
[1] Juhlin, L., J. Intern. Med. 1997, 242, 61–66.
[2] Kogan, G., Šoltés, L., Stern, R., Gemeiner, P., Biotechnol.
Lett. 2007, 29, 17–25.
[3] Chen, W. Y., Marcellin, E., Steen, J. A., Nielsen, L. K., Mol.
Biotechnol. 2013, 56, 47–156.
[4] Wu, T., Huang, W., Chen, Y. T. Ã., Proteomics. 2009, 9,
4507–4518.
[5] Izawa, N., Hanamizu, T., Iizuka, R., Sone, T., Mizukoshi,
H., Kimura, K., Chiba, K., J. Biosci. Bioeng. 2009, 107,
119–123.
[6] Izawa, N., Hanamizu, T., Sone, T., Chiba, K., J. Biosci.
Bioeng. 2010, 109, 356–360.
[7] Choi, S.-B., Lew, L.-C., Hor, K.-C., Liong, M.-T., Appl.
Biochem. Biotechnol. 2014, 173, 129–142.
[8] Widner, B., Behr, R., Von Dollen, S., Tang, M., Heu, T.,
Sloma, A., Sternberg, D., Deangelis, P. L., Weigel, P. H.,
Brown, S., Appl. Environ. Microbiol. 2005, 71, 3747–3752.
[9] Chien, L. J., Lee, C. K., Biotechnol. Prog. 2007, 23, 1017–
1022.
[10] Yu, H., Stephanopoulos, G., Metab. Eng. 2008, 10, 24–32.
[11] Mao, Z., Shin, H. D., Chen, R., Appl. Microbiol. Biotech-
nol. 2009, 84, 63–69.
[12] Hoshi, H., Nakagawa, H., Nishiguchi, S., Iwata, K., Ni-
ikura, K., Monde, K., Nishimura, S.-I., J. Biol. Chem. 2004,
279, 2341–2349.
[13] Jeong, E., Shim, W. Y., Kim, J. H., J. Biotechnol. 2014,
185, 28–36.
[14] Sheng, J., Ling, P., Wang, F., J. Ind. Microbiol. Biotechnol.
2014, 42, 197–206.
[15] Chien, L.-J., Lee, C.-K., Appl. Microbiol. Biotechnol. 2007,
77, 339–346.
[16] Chauhan, A. S., Badle, S. S., Ramachandran, K. B., Ja-
yaraman, G., Biochem. Eng. J. 2014, 90, 73–78.
[17] Prasad, S. B., Jayaraman, G., Ramachandran, K. B., Appl.
Microbiol. Biotechnol. 2010, 86, 273–283.
[18] Prasad, S. B., Ramachandran, K. B., Jayaraman, G., Appl.
Microbiol. Biotechnol. 2012, 94, 1593–1607.

C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
J. Sep. Sci. 2016, 39, 655–662
[19] Sheng, J. Z., Ling, P. X., Zhu, X. Q., Guo, X. P., Zhang, T.
M., He, Y. L., Wang, F. S., J. Appl. Microbiol. 2009, 107,
136–144.
[20] Badle, S. S., Jayaraman, G., Ramachandran, K. B., Biore-
sour. Technol. 2014, 163, 222–227.
[21] Chen, W. Y., Marcellin, E., Hung, J., Nielsen, L. K., J. Biol.
Chem. 2009, 284, 18007–18014.
[22] Oueslati, N., Leblanc, P., Bodin, A., Harscoat-Schiavo, C.,
Rondags, E., Meunier, S., Marc, I., Kapel, R., J. Membr.
Sci. 2015, 490, 152–159.
[23] Zhou, H., Ni, J., Huang, W., Zhang, J., Sep. Purif. Technol.
2006, 52, 29–38.
[24] Jones, K. L., R., C. O’Melia, J. Membr. Sci. 2000, 165,
31–46.
[25] Carić, M. D. X., Milanović, S. D., Krstić, D. M., N., M.
Tekić;, J. Membr. Sci. 2000, 165, 83–88.
[26] Rangaswamy, V., Jain, D., Biotechnol. Lett. 2008, 30, 493–
496.
[27] Reddy, K. J., Karunakaran, K. T., J. Biosci. Biotechnol.
2013, 2, 173–179.
[28] Choi, S., Choi, W., Kim, S., Lee, S.-Y., Noh, I., Kim, C.-W.,
Biomater. Res. 2014, 18(6).
[29] Akdamar, H. A., Sarıözlü, N. Y., Özcan, A. A., Ersöz, A.,
Denizli, A., Say, R., Mater. Sci. Eng. C 2009, 29, 1404–
1408.
[30] Zeng, C.-X., Xin, R.-P., Qi, S.-J., Yang, B., Wang, Y.-H., J.
Sep. Sci. 2015.
[31] Wu, Q., Lin, D.-Q., Zhang, Q.-L., Gao, D., Yao, S.-J., J.
Sep. Sci. 2014, 37, 447–453.
[32] Tanuja, S., Srinivas, N. D., Raghava Rao, K. S. M. S.,
Gowthaman, M. K., Process Biochem. 1997, 32, 635–641.
[33] Azevedo, A. M., a J. Rosa, P., Ferreira, I. F., R., M. Aires-
Barros, J. Biotechnol. 2007, 132, 209–217.
[34] Rosa, P. A. J., Azevedo, A. M., Aires-Barros M. R., J. Chro-
matogr. A 2007, 1141, 50–60.
[35] Naganagouda, K., Mulimani, V. H., Process Biochem.
2008, 43, 1293–1299.
[36] Boland, M. J., Mol. Biotechnol. 2002, 20, 85–93.
[37] Gu, Z., Glatz, C. E., J. Chromatogr. B Anal. Technol.
Biomed. Life Sci. 2007, 845, 38–50.
[38] Liu, Y., Yu, Y. L., Chen, M. Z., Xiao, X., Can. J. Chem. Eng.
Technol. 2011, 2, 1–7.
[39] Patil, G., Raghavarao, K. S. M. S., Biochem. Eng. J. 2007,
34, 156–164.
[40] Chao, Z., Shibusawa, Y., Shindo, H., Ito, Y., J. Liq. Chro-
matogr. Relat. Technol. 2003, 26, 1895–1903.
[41] Xing, J.-M., Li, F.-F., Nat. Prod. Res. 2009, 23, 1424–1430.
[42] Arends, J. G. a, a Arends, J. G., Hugenholtz, J., Hugen-
holtz, J., Kleerebezem, M., Kleerebezem, M., Microbiol-
ogy 2003, 185, 4499–4507.
[43] Albertsson, P. A., Johansson, G., Tjerneld, F., Bioprocess
Technol. 1990, 9, 287–327.
[44] Bitter, T., Muir, H. M., Anal. Biochem. 1962, 4, 330–334.
[45] Rezaei, H., Ashtiani, F. Z., Fouladitajar, A., Braz. J. Chem.
Eng. 2014, 31, 503–518.
[46] Ruckmani, K., Shaikh, S. Z., Khalil, P., Muneera, M. S., a.
Thusleem, O., J. Pharm. Anal. 2013, 3, 324–329.
www.jss-journal.com