Article

pubs.acs.org/IECR

Graphene Oxide Nanocomposite Incorporated Poly(ether imide)

Mixed Matrix Membranes for in Vitro Evaluation of Its Efficacy in
Blood Purification Applications
Noel Jacob Kaleekkal,† A. Thanigaivelan,† M. Durga,† R. Girish,‡ Dipak Rana,§ P. Soundararajan,‡
and D. Mohan*,†
†
Membrane Laboratory, Department of Chemical Engineering, Alagappa College of Technology, Anna University, Chennai 600025,
India
‡
Department of Nephrology, Sri Ramachandra University, Porur, Chennai 600116, India
§
Department of Chemical and Biological Engineering, University of Ottawa, 161 Louis Pasteur Private, Ottawa, Ontario K1N 6N5,
Canada

ABSTRACT: Hemodialysis is one of the most commonly used treatments for patients suffering from irrecoverable kidney
damage. In our present work, we investigate poly(ether imide) (PEI) mixed matrix membranes (MMMs) as a potential candidate
for hemodialysis applications due to their efficient clearance and high biocompatibility. Graphene oxide (GO) was synthesized by
the modified Hummers’ method and was then confirmed by X-ray diffraction spectroscopy, Fourier transform infrared
spectroscopy, Raman spectroscopy, and high-resolution transmission electron microscopy. The GO-polyvinylpyrrolidone
nanocomposite incorporated PEI MMMs were fabricated by a semiautomatic casting unit using the nonsolvent induced phase
separation technique. The effect of the nanocomposite loading ratio was evaluated by water content, ultrafiltration rate, and
porosity, which were all found to increase as the nanocomposite content increased. Cross-sectional and top surface morphology
was visualized using scanning electron microscopy and atomic force microscopy. The hydrophilicity of these membranes was in
consonance with contact angle values. These MMMs demonstrated an increase in biocompatibility: reduced protein adsorption,
suppressed platelet adhesion, and lower complement activation. Furthermore, the prolonged blood clotting time is an indication
of the heparin mimic anticoagulant properties of these membranes. The cytocompatibility results by 3-(4, 5-dimethyl-2-
thiazolyl)-2, 5-diphenyl-tetrazolium bromide assay and live cell/dead cell staining indicated that there was an increase in cell
viability. The membranes with 0.1 wt % GO showed an excellent clearance of the model uremic toxins, namely urea, vitamin B-
12, and cytochrome-c in vitro. The diffusive permeability of these membranes could be comparable to the existing commercial
hemodialysis membranes. Thus, it can be concluded that these membranes containing a composite of both functional nanosheets
and bioactive polymers have a tremendous potential to be utilized commercially in hemodialysis modules if shown successful in
further in vivo studies with an animal model.

1. INTRODUCTION
Membrane technology is a necessary component in a number
of medical applications and is crucial in common life-saving
treatments. Research on biomimicking membranes for
biomedical applications such as blood purification, artificial
organs, and other clinical medical devices is gaining impetus in
the biomedical sector.1 With the advancement of research,
artificial polymeric materials, such as cellulose acetate (CA),2
polyacrylonitrile (PAN),3 polysulfone (PSf),4 poly(ether
sulfone) (PES),5 polyvinylidene fluoride (PVDF),6 polyvinyl
chloride (PVC),7 nylon-66,8 polyurethane (PU),9 and poly-
tetrafluoroethylene (PTFE),10 are being widely applied in
various biomedical fields, namely hemodialysis, cardiopulmo-
nary bypass, artificial oxygenators, plasma collection, etc.
Considering the annual increase of about 7−8% in the
number of patients that suffer from end stage renal disease
(ESRD), hemodialysis membranes are a great area of interest.11
ESRD is defined as an irreversible loss of kidney function where
a patient cannot survive without any of the renal replacement
therapies (RRTs). Hemodialysis is the most popular and viable
of the RRTs, wherein accumulated uremic toxins, excess ions,
and water from the blood are removed, and insufficient
essential ions are replenished from the dialysate.12 As for any
other device in contact with blood, the appropriate physical and
chemical interface modifications are necessary for these
membranes to achieve a favorable biological response or
biocompatibility.13 Otherwise, platelet aggregation leads to the
formation of thrombus, which could possibly cause an artery
occlusion, resulting in severe injuries including heart attacks
and even death.14
Poly(ether imide) (PEI) is a novel membrane forming
polymer that has been validated for the potential use in
biomedical applications.15,16 PEI is an amber-transparent
amorphous polymer that is thermally, mechanically, and
chemically stable as well as resistant to many solvents. It is
capable of forming flexible thin films, which can be attributed to
the repeated phenyl and imide groups interspersed with ether
Received: May 4, 2015
Revised: July 24, 2015
Accepted: August 3, 2015
Published: August 3, 2015

© 2015 American Chemical Society 7899 DOI: 10.1021/acs.iecr.5b01655

Ind. Eng. Chem. Res. 2015, 54, 7899−7913
Industrial & Engineering Chemistry Research
linkages and angular bonds.17 However, its suitability for
hemodialysis membranes has yet to be investigated. The
hydrophobicity of PEI is similar to commercially available
membrane-forming polymers; this causes an insufficient
wettability that induces incomplete coating18 and provides for
a high adsorption of proteins and blood cells onto the
membrane in any blood contact application.
Various modifications have been carried out to increase the
hydrophilicity or biocompatibility. Ran et al. improved the
biocompatibility of PES membranes by blending an amphiphilic
triblock copolymer of poly(vinylpyrrolidone)−b-poly(methyl
methacrylate)−b-poly(vinylpyrrolidone) with it.19 Nie et al.20
used heparin-mimicking polymer brush grafted carbon nano-
tubes, whereas the introduction of heparin-mimicking polyur-
ethane was the technique chosen by Ma et al.9 Grafting of
zwitterion from polysulfone membrane via surface-initiated
atom transfer radical polymerization (ATRP),21 surface
zwitterionization of hemocompatible poly(lactic acid),22 and
the introduction of vitamin E tocopherol polyethylene glycol
succinate (vitamin E TPGS)23 were few of the other methods
chosen to enhance the biocompatibility of polymeric
membranes.
Although each of these modifications imparts unique desired
properties to the membrane, blending of polymers with some
nanoparticles is favored because of the facile fabrication and the
resulting permanent modification.24 Here we have used
graphene oxide (GO), an environmentally and biologically
friendlier inorganic nanomaterial, to improve the properties of
the pristine membrane. GO is produced by chemical
introduction of oxygen containing groups into pristine graphite
material. The epoxide and hydroxyl groups deck the basal plane
carbon atoms, whereas the carbonyl and carboxyl group its edge
atoms. This ensures that GO is highly hydrophilic, and the
presence of these groups reduces interplanar forces, which
improves their interfacial interaction.25 In our study, we have
incorporated this hydrophilic GO along with PVP-K90 into the
PEI matrix to prepare mixed matrix membranes (MMMs) that
synergistically mimic the chemical groups of heparin. Further,
introduction of hydrophilic GO maintains the hydrophobic/
hydrophilic balance, improves antifouling properties, and
increases mechanical stability of the membranes.26
However, one of the challenges that must be circumvented
when GO is used is to prevent its agglomeration in the polymer
matrix. To overcome this shortcoming, we used an inexpensive,
biocompatible polyvinylpyrrolidone (PVP-K90) as a polymeric
stabilizer.25 The hereditary hydrophobic carbon skeleton of GO
attaches to the polymer matrix via hydrophobic−hydrophobic
interactions and π−π stacking, whereas the functional groups
(viz. −OH, −COOH, and CO) give it the required
hydrophilic character and attract hydrophilic substances by
dipole−dipole interaction, hydrogen bonding, and dispersion
forces.27
This study aims to produce membranes with a high flux,
optimum biocompatibility, and efficient removal of uremic
toxins by the incorporation of GO-PVP nanocomposite into
the PEI matrix forming MMMs. Commercially obtained
graphite was chemically oxidized by the modified Hummers’
method and confirmed using X-ray diffraction (XRD), Raman
spectroscopy, high-resolution transmission electron microscopy
(HR-TEM), electron diffraction X-ray (EDX), and Fourier
transform infrared (FTIR) spectroscopy. The influence of
increasing concentrations of this nanocomposite in MMMs was
evaluated in terms of mean pore size, the contact angle, and
7900
Article
work of adhesion. The morphology was evaluated using
scanning electron microscopy (SEM) and atomic force
microscopy (AFM). The mechanical stability of the modified
membranes was evaluated in terms of tensile strength. Blood
compatibility was evaluated with plasma protein adsorption,
platelet adhesion, plasma recalcification time, complement
activation, activated partial thromboplastin time (aPTT), and
prothrombin (PT). Furthermore, the cytocompatibility was
evaluated by cell culture (hepatocyte cell line), live/dead cell
imaging, and 3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-tetrazo-
lium bromide (MTT) assay.
2. MATERIALS AND METHODS
2.1. Materials. Commercial grade polymers PEI from
Sigma-Aldrich Inc., St. Louis, MO (melt index 9 g/10 min) was
used after being dried in a vacuum oven at 80 °C for 48 h. The
solvent N-methyl-2-pyrrolidone (NMP) from SRL Chemicals
Ltd., Mumbai, India was dried using 4 Å sieves. Molecular-
grade urea (MW 60.06 Da), vitamin B-12 (MW 1355 Da),
cytochrome-c (MW 11.8 kDa), and bovine serum albumin
(BSA) (MW 64 kDa) purchased from Himedia Laboratories
Pvt. Ltd., Mumbai, India were used in the solute diffusion
studies. Urea nitrogen (Diacetyl) reagents from Erba
Mannheim, Mnnheim, Germany were used for the determi-
nation of the solute concentration. Graphite (fine powder 98%)
extra pure, sodium nitrate, hydrogen peroxide, and potassium
permanganate were obtained from Loba Chemie Pvt. Ltd.,
Mumbai, India. These materials were used as-is. Milli-Q water
was used for all solution preparations and studies.
2.2. Preparation of GO. Three grams of graphite powder
was added to a round-bottom flask containing 70 mL of conc.
H2SO4 and stirred in an ice bath. To this mixture, 9 g of
KMnO4 was added slowly to ensure that the temperature was
maintained lower than 15 °C. After 30 min, the flask was
transferred into a 40 °C bath and vigorously stirred for 1 h.
Then 150 mL of water was added dropwise, and the solution
was maintained at 95 °C for 1 h. The solution changed color
from dark brown to yellow when an excess 500 mL of water
and 15 mL of H2O2 (30%) was added. This solution was
allowed to settle, and the solids were washed with dilute HCl
(250 mL) to remove metal ions.28 The solution was filtered and
centrifuged in hot conditions to remove the precipitates soluble
in warm water. This yellowish brown residue was then
ultrasonicated to get fully exfoliated GO. Next it was dried in
a vacuum oven overnight and then used for further character-
ization and membrane preparation.
2.3. Characterization of GO.
(1) Wide-angle X-ray scattering (WAXS): the X-ray powder
diffraction patterns were recorded on a Bruker-AXS
diffractometer (Bruker, Germany) using Cu Kα radiation
(λ = 0.154 nm).
(2) Confocal Raman spectroscopy: Raman measurements
were performed on an In Via/Reflex Laser Micro-Raman
spectroscopy (Renishaw, Gloucestershire, England) with
an excitation laser beam that had a wavelength of 514
nm. The powders of graphite and GO were then placed
on a clean SiO2/Si substrate that was used for the Raman
measurement.
(3) HR-TEM with EDX microscopy: TEM images (includ-
ing high-resolution images), selected area electron
diffraction (SAED) patterns, and EDX were acquired
using a JEM-3010 TEM (Jeol Ltd., Tokyo, Japan)
DOI: 10.1021/acs.iecr.5b01655
Ind. Eng. Chem. Res. 2015, 54, 7899−7913
Industrial & Engineering Chemistry Research
operated at 200 kV. Freshly sonicated GO (∼0.02 mg/
mL) was spread on carbon-coated copper grids and
blotted after 30 s.
(4) FTIR spectroscopy: powdered samples were spread on
the attenuated total reflection (ATR) crystal, and percent
transmittance between 400 and 4000 cm−1 was measured
on a spectrophotometer (Nicolet Avatar 370, Thermo
Electron Corp., Madison, WI).
(5) UV−visible spectroscopy: the absorption spectra were
recorded on a UV−vis spectrometer (Libra S50,
Biochrom Ltd., Cambridge, England). The aqueous
suspensions of GO were used as the UV−vis samples
along with pure water taken as reference.
2.4. Preparation of the MMMs. Initially, GO (0, 0.025,
0.050, 0.1, and 0.2 wt %) was sonicated in the solvent (NMP)
for 1 h, after which 2 wt % PVP-K90 was added and further
homogenized by vigorous stirring for 2 h. To this, 16 wt % PEI
was added, and a homogeneous dope solution was prepared by
mechanical stirring for 24 h. The dope solution was then
degassed by applying a vacuum. A semiautomatic casting unit
was used to prepare these MMMs. The dope solution was
casted on a dust free glass plate maintaining a thickness of 200
μm at a relative humidity (RH) of 25%. The glass plate was
then immediately immersed in a water bath containing 0.2%
solvent. The nascent membranes were washed with warm water
(35 ± 2 °C) and then stored in distilled water until further
testing.
2.5. Porosity and Average Pore Size. Membrane
porosity (ε %) was measured by the gravimetric method. The
membrane porosity was obtained by
(Ww − Wd)/Dw
ε (%) = Ww − Wd
− Wd /Dp
Dw (1)
where Ww and Wd are the weights (g) of the wet and dry
membrane, respectively, and Dw and Dp are densities (g/cm3)
of the water and polymer membrane, respectively. These tests
were replicated thrice, and their average value has been
reported.
The average pore radius rm (m) is determined by the
filtration velocity method using the Guerout-Elford−Ferry
equation:29
(2.9 − 1.75ε)8ηlQ
rm =
εAP (2)
where, η is the water viscosity at 25 °C (Pa s), l is the
membrane thickness (m), Q is the water flux (m3/s), A is the
membrane area used for filtration (m2), and P is the
predetermined operating pressure (Pa).
2.6. FTIR Spectra and Surface Streaming Potential.
Attenuated total reflection−Fourier transform infrared (FTIR/
ATR) spectra for prepared MMMs were compared to the
pristine membrane using a spectrophotometer (Nicolet Avatar
370, Thermo Electron Corp., Madison, WI). The measurement
of the surface streaming potential was determined using a
surface analyzer (SurPass, Anton Paar, Austria).
2.7. Contact Angle and Work of Adhesion. The static
CA is a measure of the surface hydrophilicity of the membrane.
The CA of the prepared membranes was obtained by a contact
angle goniometer (OCA15, DataPhysics, Filderstadt, Germany)
equipped with video capture. Five microliters of double-distilled
water was carefully placed on the 2 × 2 cm2 membrane sample,
7901
Article
which was affixed on a glass slide. The reading was taken after
10 s. The measurement was done on five different locations of
the membrane surface to get reliable data.
The work of adhesion (W) explains the interactive forces
between liquid and solid surfaces. This was calculated using the
Young−Dupree equation:
W = γL(1 + cos θ )
(3)
where γL is the surface tension of the water (72 mJ), and θ is
the contact angle.30
2.8. Morphological Analysis. The cross-sectional and top
surface morphologies of all prepared membranes were
determined by SEM and AFM, respectively. A 1 cm2 sample
of the membrane was dried using a series of ethanol solutions,
then fractured in liquid nitrogen, and then loaded onto the
scanning electron microscope (FEI Quanta-400 FEG) to
observe the cross-sectional morphology.
AFM in the noncontact mode analyzed the surface
morphology and roughness of the membranes. The AFM
apparatus used was a XE-100, Park Systems Corp., Suwon,
South Korea. The prepared membranes were cut into small
squares (1 cm2) and affixed onto the metal substrate. The
roughness values reported are the average of three different
scan areas of 5 μm × 5 μm each.
2.9. Ultrafiltration Rate (UFR). In dialysis practice, to
evaluate and control the performance, the UFR is evaluated by
using the following equation:
⎛ ⎞
mL V
UFR ⎜ 2 ⎟=
⎝ m h mmHg ⎠ PSt (4)
where V is the quantity of permeate collected in mL, A is the
membrane surface area (m2), P is the operating pressure
(mmHg), and t is the time for which permeate is collected (h).
2.10. BSA Retention and Antifouling. It is imperative
that the membrane should prevent loss of albumin from the
bloodstream during dialysis. The BSA concentrations in the
original (Co) and the corresponding final mimic blood (Cf)
were determined with a UV−vis spectrophotometer (Libra S50,
Biochrom Ltd., Cambridge, England) at a wavelength of 280
nm and calculated utilizing a standard curve. The BSA retention
(R) of the membrane was then calculated by
⎛C ⎞
R (%) = ⎜ f ⎟ × 100
⎝ Co ⎠
(5)
Three different membrane areas are used, and the retention
is given as the standard deviation of the same. The fouling-
resistant ability was evaluated by the measurement of the time-
dependent fluxes of phosphate buffer solution (PBS) followed
by that of BSA solutions. Then the membranes were cleaned
using water for 20 min, and both fluxes were measured again.
The flux recovery ratio (FRR) was calculated using the
following expression.31 A total of three such runs were carried
out:
⎡J⎤
FRR (%) = ⎢ i ⎥ × 100
⎢⎣ J0 ⎥⎦ (6)
where J0 and Ji are solution fluxes for the ith run before and
after protein solution ultrafiltration, respectively.
2.11. Biocompatibility Characteristics. 2.11.1. Protein
Adsorption. Membrane samples (1 ×1 cm2) were equilibrated
DOI: 10.1021/acs.iecr.5b01655
Ind. Eng. Chem. Res. 2015, 54, 7899−7913
Industrial & Engineering Chemistry Research
in PBS for 12 h before use. Then these samples were incubated
in 1.0 or 0.05 mg/mL protein (BSA or FnG, respectively)
solution at 37 °C for 2 h and subsequently rinsed with PBS
solution followed by double-distilled water. Following this, the
membranes were then immersed in 2 mL of 2 wt % sodium
dodecyl sulfate (SDS) aqueous solution at 37 °C for 1 h to
remove the adsorbed protein. The Micro BCA Protein Assay
Kit (Thermo Scientific) was used to determine the protein
concentration in the SDS solution, which is the quantity of
adsorbed protein. The experiment was carried out in triplicate.9
2.11.2. Platelet Adhesion. Healthy fresh human blood (26
years, male) was collected using sodium citrate lined vacutainer.
Here, the ratio of the anticoagulant to blood was 1:9 v/v. The
blood was centrifuged at 1000 rpm for 15 min to obtain
platelet-rich plasma (PRP). The membrane was then immersed
in a PBS solution and equilibrated at 37 °C for 1 h. The PBS
was then removed, and 0.5 mL of fresh PRP was added. Lastly,
they were then incubated at a 5% CO2 atmosphere at 37 °C for
1 h.
The lactate dehydrogenase (LDH) method was used to
determine the number of platelets adhered to the samples. The
concentration of adhered platelets was calculated from the
calibration plot.32 Here, a linear relationship was obtained
between the LDH activity of aliquots, and the known
concentration of platelets was counted using a hemocytometer.
Another set of these membranes, the platelet adhesion, was
evaluated using SEM. After contact with the PRP, the
membranes were fixed using glutaraldehyde in PBS. The
samples were dried using the solvent exchange method. The
platelet adhesion was observed using a SEM (FEI Quanta-400
FEG).33
2.11.3. Thrombin−Antithrombin III (TAT) Generation. The
membrane (1 cm2) was incubated with 1 mL of healthy human
fresh whole blood (man, 26 years old, sodium citrate
anticoagulant) at 37 °C for 1 h, followed by centrifugation at
4000 rpm for 15 min to obtain the platelet-poor plasma (PPP).
TAT was measured by an enzyme-linked immunosorbent assay
(ELISA) with commercially available kits, Enzygnost TAT
micro (USA).20 Duplicate runs were carried out to ensure
reliability.
2.11.4. Clotting Time. aPTTs and PTs for the membranes
were measured by a semiautomated blood coagulation analyzer
(CA- 50, Sysmex Corp., Kobe, Japan) to evaluate the contact
activation pathway and the abnormality in the conversion of
fibrinogen to fibrin in coagulation, respectively. Here, the
membrane sample discs of 0.5 cm2 were incubated in 0.20 mL
of PPP at 37 °C for 30 min, and then the aPTT and PT were
measured. To get a reliable value, the test was repeated three
times for each sample.23
2.11.5. Plasma Recalcification Time (PRT). Membrane discs
of 0.5 cm2 were immersed in the PBS solution and equilibrated
at 37 °C for 1 h. A 100 μL sample of PPP was placed on the
sample film attached to a glass slide and incubated statically at
37 °C, and then 100 μL of 25 mM CaCl2 aqueous solution was
added to the PPP. The plasma solution was then monitored for
clotting by a UV−vis spectrophotometer. To get a reliable
value, the test was repeated three times for each sample.
2.11.6. Complement Activation. Twenty milliliters of
human blood was drawn from healthy adult volunteers without
the addition of any anticoagulant. The collected blood was
incubated for 1 h at 37 °C. The resulting clotted blood was
then centrifuged at 2000 rpm for 20 min to separate the serum.
Membranes (1 × 1 cm2) were immersed in 2 mL of serum and
7902
Article
further incubated at 37 °C for 1 h. The concentrations of
complement components, C3a and C5a, left in the serum after
incubation with membranes were determined with the Human
ELISA kit for C3a and C5a (Assay Pro, USA), respectively.
2.12. Toxic Solute Removal. The efficiencies for the
removal of uremic toxins and other middle molecular weight
solutes were tested in a single stage test cell in a counter current
manner. The membranes having an area of 38 cm2 were placed
between the solute and dialysate chamber, each having a
volume of 10 mL. The solutes chosen were urea (60 Da),
vitamin B-12 (1355 Da), and cytochrome-c (∼12 kDa). Freshly
prepared PBS solution was used in the dialysate side. Constant
flow rates of 50 mL/min and 100 mL/min were maintained for
solute and dialysate reservoirs correspondingly. The samples
were drawn at fixed intervals and were then analyzed. The
concentration of urea was determined using an Erba Mannheim
(Urea BUN) kit; the concentrations of vitamin B-12 and
cytochrome-c were directly measured using a UV−vis
spectrophotometer at wavelengths 360 and 550 nm,
respectively. Three runs were carried out for each solute to
ensure the consistency of performance for the membranes.
The overall mass transfer coefficient K0 is determined by the
solute component mass balance between the two chambers:
ln(ΔC1/ΔC2)
K0 =
S× { 1
Vb
+
1
Vd } × (t 2 − t1)
(7)
where t1 and t2 are the sampling times, ΔC is the difference
between the solute concentrations of both cells at each
sampling time, Vb and Vd are the solution volumes in each
cell (Vb = 100 cm3, Vd = 200 cm3), and S is the effective
membrane area (38 cm2).
The overall resistance to mass transfer is composed of two
components, the bulk fluid resistance and the membrane
resistance, as given by eq 8:
1 1 2
= +
K0 pmem kb (8)
The factor “2” in the above equation accounts for the
presence of the boundary layer in both of the chambers. The
bulk mass transfer coefficient kb was evaluated using eq 9:
k br
= αRe 0.567Sc 0.33
D∞ (9)
Here, r is the radius of the reservoirs, D∞, is the diffusion of the
solute at 25 °C, Re is the Reynolds number, and Sc is the
Schmidt number. The parameter α depends on geometry and
was determined to be 0.26.34 From these equations, the
diffusive permeability of the solute through the membrane
(pmem) can be determined.
2.13. Cell Cytocompatibility. 2.13.1. Cell Culture.
Hepatocytes were cocultured from Chang Liver cell line with
supplementation of Dulbecco Eagle’s Minimum essential
medium (MEM) and 10% fetal bovine serum. Hepatocytes
were isolated by a two-step gradient centrifugation, and PBS
was used to wash the cells and then finally diluted to 104 cells/
μL, and the viability was then confirmed by tryphan blue.
2.13.2. MTT Assay. Membranes of equal sizes (1 cm2) were
cut, equilibrated in PBS, and sterilized under UV light
overnight. These cultured cells were seeded on to 24-well
culture plates at an equal density of 104 cells/μL and incubated
for 24 h at 37 °C in a 5% CO2 incubator until the cells attained
DOI: 10.1021/acs.iecr.5b01655
Ind. Eng. Chem. Res. 2015, 54, 7899−7913
Industrial & Engineering Chemistry Research
Figure 1. FTIR and Raman spectroscopy of pristine graphite and GO.
Figure 2. HR-TEM, XRD, and UV spectroscopy of the prepared GO.
the required morphology. The membranes were then incubated
on to the cells and incubated in a CO2 incubator for 24 h at 37
°C, and cell viability was determined by MTT assay. Cells
cultures in wells without membranes served as controls.5 MTT
was prepared at a concentration of 1 mg/mL, and 50 μL was
added to each well and incubated for 4 h at 37 °C in 5% CO2.
Mitochondrial dehydrogenases of viable cells cleaved selectively
to the tetrazolium ring. Then 50 μL of DMSO was added to
dissolve the formazan crystals, which yields a purple color. The
level of the reduction of MTT into a colored formazan salt by
the living cells was measured spectrophotometrically at 570 nm
using a micro plate reader (PerkinElmer ENFIRE, multimode
reader). Assay was performed in triplicates, and the statistical
significance was determined by the student t-test with the level
of significance at P < 0.05.
2.13.3. Cell Viability Using Acridine Orange/Ethidium
Bromide (AO/EtBr) Staining. For cell staining, the membranes
and medium were removed through following the incubation of
membrane in seeded cells at 37 °C in the CO2 incubator. Cells
were then stained with 1 mL of AO/EtBr prepared in PBS and
incubated for 5 min. Acridine orange cleaves the membrane and
7903
Article
stains the intact viable cells with the color green, while ethidium
bromide, which is membrane impermeable, stains the apoptotic
cells in red. Fluorescent stained cells were viewed under a
fluorescent microscope, and images were captured with a Nikon
camera built in the microscope.
3. RESULTS AND DISCUSSION
3.1. Characterization of GO. We have employed the
modified Hummers’ method for the synthesis of GO from
graphite using sulfuric acid and KMnO4 as oxidizing agents.28,35
The FTIR spectra (Figure 1A) revealed the existence of OH
stretching vibrations (3400 cm−1), stretching vibrations from
CO (1720 cm−1), CC skeletal vibrations from unoxidized
graphitic domains (1640 cm−1), C−OH (1224 cm−1), and C−
O (1050 cm−1) functional groups, suggesting that oxygen-
containing groups are introduced into the graphene. Disorder
of structures or site defects of G or GO are identified and
characterized using the powerful Raman spectroscopy techni-
que. The Raman spectrum, as seen in Figure 1, panel B,
displayed a fortify peak assigned to the vibration of sp2-banded
DOI: 10.1021/acs.iecr.5b01655
Ind. Eng. Chem. Res. 2015, 54, 7899−7913
Industrial & Engineering Chemistry Research
Figure 3. Optical images of the membrane samples.
carbon atoms at 1580 cm−1 (G band) and another strong peak
assigned to the vibration of disordered sp3 carbon at 1350 cm−1
(D band).36 The D peak arises due to first-order zone boundary
phonons, which are absent in pristine graphite. Figure 2, panel
A shows a TEM image of exfoliated GO sheets. It shows a
number of stacked GO nanosheets wherein each of the single
sheets appears transparent and is folded over or curled on one
edge. EDX analysis gave a C:O ratio of 3:1. The 2θ peak seen
from the diffraction analysis is seen to be centered at 11.36°
(Figure 2B), corresponding to the diffraction from the (002)
plane, which indicates an interplanar spacing of 7.78 Å
compared to 3.4 Å for the pristine graphite material. This
reveals the intercalation of oxygen functionalities into the
ordered graphite structure, which facilitates the hydration and
exfoliation of GO sheets in aqueous media.37 The spectra are
plotted in the wavelength range from 200−500 nm for both
water and NMP. In the case of NMP, the data are plotted from
>260 nm as a result of the impossibility of properly
compensating for the strong absorption of the solvent at
smaller wavelengths. The π−π transitions of the doubly bonded
aromatic carbon atoms38 present on the GO backbone are
responsible for the highest absorption peak at 231 nm (Figure
2C) observed from the UV spectra. Further from this spectra at
300 nm, a shoulder peak was observed to be attributed to n →
π* transitions of the CO moieties. These results are
comparable to earlier studies.
3.2. Preparation of PEI/GO MMMs. A preweighed
quantity of the dried GO was sonicated with the PVP to
form a nanocomposite. These nanocomposites were incorpo-
rated into the polymer dope (PEI + NMP) solution by physical
blending. The MMMs were prepared by first casting the
solution with a preset thickness using a semiautomatic casting
unit followed by the nonsolvent induced phase separation
(NIPS) technique. The optimized concentrations of PEI and
PVP were found to be 16 and 2 wt %, respectively. M0 indicates
the pristine membrane without any GO, where M1, M2, M3
and M4 are membranes with increasing concentrations of GO
(0.025, 0.05, 0.1, and 0.2 wt %, respectively). The coagulation
bath contained distilled water with 0.2% w/v NMP, and the
bath temperature was maintained at 20 °C. After 2 h, the
nascent membranes were transferred to a bath containing warm
7904
Article
water and stored overnight. Furthermore, no elution of GO was
observed into the bath. These membranes were washed and
stored until further use. It can be seen that the surface of the
membranes becomes darker as the GO loading increases, as
seen in Figure 3.
3.3. Surface Chemical Structure and Surface Stream-
ing Potential. The ATR-FTIR spectroscopy was used to
analyze the surface enrichment of GO on the membranes.
Infrared spectra of the MMMs are shown in Figure 4. For the
pristine membrane, the bands at 1778 and 1718 cm−1
correspond to the asymmetric and symmetric stretching
vibrations of the carbonyl, respectively. The absorption band
at 1350 cm−1 is assigned to the bond stretching vibration of C−
N in phthalimide rings. The aryl ether bonds are confirmed by
the vibrations at 1270, 1232, 1072, and 1018 cm−1.39 In
comparison with pristine PEI membrane, the GO modified
membranes exhibited intense and wider peaks at ∼3400 cm−1,
which indicated that the surface hydrophilicity was obviously
improved due to the resident GO on the top surface. Similar
observations were also made for GO incorporated PVDF
membranes.40 The pristine M0 membrane showed a surface
charge of −18.20 mV, which increased to −30.60 mV for
membrane M3 at a pH of ∼7.2. This was done using KI
solution and was then adjusted for pH. The decrease in surface
charge is an indication of GO nanocomposites being oriented at
the membranes surface, which is in accordance with the optical
images of the MMMs.
3.4. Morphology Analysis. The microstructure cross-
sectional morphologies of the pure and MMMs were analyzed
using SEM, as seen in Figure 5. All of the membranes had an
asymmetric structure composed of dense skin layers and porous
substructure. Both pore size and pore numbers have notably
increased with increasing concentration of GO. Also, it was
seen that the skin layer thickness decreases with an increase in
GO concentration, which in turn explains the pure water flux.
In the composite membranes, the amphiphilic GO-PVP
nanocomposite migrated and was concentrated on the
membrane surfaces. Consequently, it can be hypothesized
that the increase in pore size and relatively compact
architecture of the MMMs would reduce the time required
for removal of the toxins as the flow rate, which is a limiting
DOI: 10.1021/acs.iecr.5b01655
Ind. Eng. Chem. Res. 2015, 54, 7899−7913
Industrial & Engineering Chemistry Research
Figure 4. Surface chemical structures (FTIR) of the membranes.
factor, increases. The presence of GO makes the system
thermodynamically unstable, leading to a quicker L−L phase
separation with a more porous structure.41
The use of biomaterials in applications where they come in
contact with the blood cells or other tissues could lead to a
cascade of biological responses originating with an a protein
Figure 5. Cross-sectional morphology of the membranes using SEM.
7905
Article
adsorption (viz. fibrinogen), which is followed by attachment
and activation of serum platelets and also cell cytotoxicity.20
The top surface morphologies of the MMMs are, as shown
from AFM analysis, seen in Figure 6. From the topography, it
can be understood that as the GO nanocomposite concen-
tration in the MMMs increased, the surface roughness would
also increase. The light and dark regions seen from the AFM
mapping correspond to the high and low (membrane pores)
areas of the membrane sample. The Ra, Rq, and Rz parameters
given in Table 1 are the average values of three 5 μm × 5 μm
areas scanned. This possibly demonstrates that the hydrophilic
nature of GO leads to a faster exchange of S-NS during the
NIPS process. The rate of solvent exchange increases, which
accounts for the spheres or nodules of polymer on the top
surface, which accounts for the greater roughness. This
phenomenon could explain the increase in porosity and has
been observed by other researchers.42,43 However, the surface
roughnesses of all the MMMs were lower than that of the
pristine PEI membrane. Therefore, it can be seen that the
incorporation of GO caused the huge “peaks” and “valleys” to
be replaced by numerous small ones, which led to an overall
smoother membrane surface. The surface roughness parameters
are as shown in Table 1.
3.5. Contact Angle and Work of Adhesion. The water
contact angle measurements can analyze the hydrophilicity of
the membrane surface. A hydrophilic surface will exhibit a
smaller contact angle as it permits the spreading of water on its
surface. Figure 7 shows sessile drop contact angle of the
fabricated MMMs. As shown, a significant decline of the
contact angle is observed when a greater amount of GO was
incorporated into the membrane matrix. During the vitrification
of the MMMs, in the course of phase inversion, the
nanocomposites migrate spontaneously to the water interface
to reduce the interfacial free energy.44 The darker color on the
top surface of the membranes in comparison to that at the
bottom indicates a migration to the membrane surface. The
large amount of −OH and −COOH groups of the GO orient
themselves in a way to impart their hydrophilic properties to
the membrane thereby increasing the adsorption of water and,
therefore, improving the membrane water permeability. The
adhesion on hydrophobic surfaces is greater compared to
surfaces possessing high concentration of polar groups. Lower
interfacial free energy surface corresponds to lower fouling in
most membrane separation processes. In the MMMs, to lower
the surface free energy of the system, during the membrane
formation, lower surface energy nanoparticles preferentially
migrate to the top surface thereby creating a gradient between
DOI: 10.1021/acs.iecr.5b01655
Ind. Eng. Chem. Res. 2015, 54, 7899−7913
Industrial & Engineering Chemistry Research Article

Figure 6. AFM imaging of top surface morphology of the membranes.

Table 1. Physical Characterization of the Membranes

roughness parameters from AFM (nm)
membrane code GO (%) mean pore size (10−9 m) porosity (%) Ra Rq Rz
M0 0.00 32 62.8 20.6(±2.6) 32.6(±1.8) 126.6(±22.4)
M1 0.025 36 66.6 8.4(±1.8) 14.4(±2.4) 88.4(±14.6)
M2 0.05 44 68.4 9.6(±2) 16.6(±3.2) 92.6(±18.8)
M3 0.1 52 72.5 11.8(±1.4) 18.2(±2.4) 122.4(±25.5)
M4 0.2 76 77.2 13.6(±1.6) 21.2(±1.6) 144.8(±12.8)

the top surface and the bulk of the membrane.45,46 Thus, the
most promising alternative in preventing protein or platelet
adsorption is using materials that have a low surface energy and
low adhesive surface.
3.6. Porosity and Pore Size (Filtration Porosity
Method). The effects of the GO content on the porosity
and mean pore size were also listed in Table 1. Both porosity
and mean pore size of membranes had increased with the
increased addition of GO, which was consistent with the results
of SEM. The presence of GO nanocomposite in the dope
solution caused micro phase separations, which facilitated the
generation of a polymer poor phase, thus suggesting that the
7906
existence of GO is beneficial for membranes that have a high
porosity and mean pore size.40 The pristine M0 had a pore size
of 32 nm, which increased to 76 nm for membrane M4.
3.7. BSA Retention. Renal failure patients had an
occurrence of albumin loss associated syndrome. Albumin is
approximately 67 kDa and is lost during the hemodialysis.
Therefore, the ideal membrane should avoid albumin loss
during treatment. All of the prepared membranes were
subjected to a protein retention study. They showed a
retention ratio greater than 95% for the BSA protein. An
increase in the GO concentration (beyond 0.1%), however,
shows a small loss of BSA to the dialysate (Table 2). M4 shows
DOI: 10.1021/acs.iecr.5b01655
Ind. Eng. Chem. Res. 2015, 54, 7899−7913
Industrial & Engineering Chemistry Research Article

gamma-globulin, are also adsorbed onto hemodialysis mem-

branes. From protein adsorption studies, the protein concen-
trations were formulated to resemble physiological concen-
trations so that the concentration of albumin used would be
much greater than the fibrinogen. As shown in Figure 8, the

Figure 7. Sessile drop contact angle and work of adhesion for the
pristine PEI and modified membranes.

Table 2. Permeation, Rejection, and Fouling Parameters of

the Membranes
FRR (%)
membrane BSA UFR first second third
code retention (%) (mL/h m2 mmHg) run run run
M0 99 (±0.5) 6.8 76 71 64.5
Figure 8. BSA and FnG adsorption onto the membranes.
M1 99 (±0.5) 8.4 79 76 75.5
M2 98 (±0.5) 14.6 84 81 78
M3 98 (±0.5) 26.8 88 86 86 pristine PEI membrane (M0) has shown the highest protein
M4 95 (±1) 35.0 92 89 87.5 adsorption of both BSA and FnG (30.6 and 17.4 μg/cm2,
respectively). It can be seen that the incorporation of GO into
the membrane matrix results in lower protein adsorption. The
a retention of only 95%, which could be explained by the larger membrane M4 shows the highest resistance to protein
pores as visualized in the SEM or due to the presence of micro adsorption, which can be explained by the increase in
cracks on the skin layer due to the high concentration of GO. hydrophilicity and the lower surface roughness. BSA protein,
3.8. Ultrafiltration Rate and Antifouling Properties. at a pH of 7.4, is negatively charged so it has a lower adsorption
During ultrafiltration, the permeability of a membrane is of tendency on more negative surfaces because of charge
pronounced importance. Membranes with excellent antifouling repulsion.50
and reasonable protein rejection ratios are favorable in Generally, membranes are prepared using the NIPS
hemodialysis. The UFR of the pristine membrane was observed technique, and the hydrophilic components (here GO)
to be 6.8 mL/h m2 mmHg, and the most efficient membrane preferentially migrate to the solution−membrane interface
M3 had a UFR of 26.8 mL/h m2 mmHg. The antifouling due to the low interfacial energy between the water and
properties could be characterized in terms of flux recovery ratio hydrophilic components. The GO present on membrane
during the ultrafiltration of the protein solution. It can be seen surfaces extends its functional group into the surrounding
that membrane M0 possessed very low flux recovery owing to aqueous environment, and hydrogen bonds could be formed
its inherently hydrophobic nature. However, even a small between water and the −OH/−COOH groups present on the
amount of the nanocomposites showed significant improve- GO sheets. This forms a hydration layer, and because of the
ment in FRR values, as seen in Table 2. The improvement of effect of steric exclusion, protein adsorption could be
PBS flux and a small decline in BSA rejection ratio could be inhibited.51 The amounts of protein adsorption that were
attributed to the following reasons: (1) the addition of GO reported in previous publications showed a significant differ-
causes a formation of a larger number of interconnected pores ence from not more than 1.0 μg/cm2 to several hundreds of
with greater size, (2) the existence of micro surface defects due μg/cm2. Because of modifications where hydrophilic groups are
to aggregation of the more hydrophilic GO onto the surface. introduced onto the membrane surface, a sharp decline has
3.9. Biocompatibility Studies. 3.9.1. Protein Adsorption. been observed by other researchers.52−54 Further, it was
A key parameter in evaluating blood compatibility is the explained that the protein adsorption on hydrophilic surfaces
amount of protein adsorbed onto the membrane surface, and occurred in one of the two modes. One is that the surface is
this is one of the main initial events that leads to blood aggressively covered by the protein it comes in contact with,
clotting.47 Numerous factors influence the protein adsorption and the other involves an inhibition of adsorption by the
onto the membrane surface: protein size, shape, temperature, initially deposited layer to prevent further adsorption (surface
pH, surface charge, hydrophilicity/hydrophobicity, and surface passivation).5
topology. Other factors in consideration are coadsorption of 3.9.2. Platelet Adhesion. Another vital tool for the appraisal
low-molecular-weight ions, the strength of functional groups, of blood compatibility is platelet adhesion. Any foreign material
intermolecular forces between the adsorbed molecules, the that is in contact with blood initially adsorbs the blood plasma
composition of the protein solution, and the chemistry of the proteins. This is followed by platelet adhesion and activation of
surface.48 The adsorption of protein fibrinogen in blood plasma the platelets, which in turn induces thrombus formation.55,56 A
is important since it paves the way for platelet adhesion so that crucial step in the process of thrombus formation is the extent
it can bind to the platelet GP IIb/IIIa receptor.49 Apart from of platelet adhesion and platelet aggregation. It was observed
FnG, many other plasma proteins, including albumin and that the platelets adhering on the pristine PEI membrane were
7907 DOI: 10.1021/acs.iecr.5b01655
Ind. Eng. Chem. Res. 2015, 54, 7899−7913
Industrial & Engineering Chemistry Research
Figure 9. Platelet adhesion by SEM: (A) M0, (B) M1, (C) M2, (D) M3, (E) M4, and (F) LDH assay results.
aggregated in large numbers on the membrane surface and
possessed a network-like structure with deformed and extended
pseudopodium. However, in the modified MMMs, the platelets
retained their original shape without deformation, and their
numbers were fewer. We also noticed that the conformational
change of the platelet was suppressed even if it adsorbed onto
the membrane surface. The amounts sharply declined with an
increase in percentage of GO, as seen in Figure 9. It is
understood that the fibrinogen is a known mediator for platelet
adhesion and aggregation.57 The platelet adhesion results were
consistent with the previous FnG adsorption results. The
results from SEM were also correlated with those obtained by
LDH assay. The number of platelets adhered decreased from
22 × 104 cells/cm2 for pristine PEI membrane to almost 5 ×
103 cells/cm2 for M4 membrane. Other researchers have also
shown that upon inclusion of GO or GO-g-pMPC, platelet
adhesion is suppressed.58
3.9.3. TAT Generation. The adhered platelets have different
types of interactions with various coagulation factors; one of
them, the coagulation product thrombin, is known to be a
potent platelet activating agonist.
7908
Article
The platelet activation level could be reflected by the platelet
factor 4 (PF4) expressed by the activated platelets and the
thrombin that resulted from the activation of coagulation
cascade combined with antithrombin III to generate TAT
complexes and, in effect, demonstrated the extent of thrombin
generation.59,60
The level of thrombin generation in the plasma after it had
contact with blood was also measured by TAT complexes.
These complexes are formed following the neutralization of
thrombin by antithrombin III and are used as a surrogate
marker for thrombin generation.61 Thrombin generation
kinetics are measured by the appearance of plasma TAT
complexes, demonstrating the anticoagulant property of the
GO, when incorporated into the membrane matrix. It was
observed that the TAT concentration for the M0 membrane
remained higher than the plasma (control) values (Figure 10).
M1 and M2, which had lower concentrations of GO, also
displayed significant TAT generation. Membranes M3 and M4
showed suppressed generation of TAT complexes. It is
assumed that an augmentation of GO concentration increased
the hydrophilicity of the membrane surface and this, in turn, led
to a lower thrombus generation. The results determined that
DOI: 10.1021/acs.iecr.5b01655
Ind. Eng. Chem. Res. 2015, 54, 7899−7913
Industrial & Engineering Chemistry Research
Figure 10. TAT generation.
the modified membranes obviously showed lower platelet
activation and the minimal coagulation cascade activation when
compared with those of pristine PEI membrane.
3.9.4. Clotting Time. When blood is made to flow through
any extra corporeal circuit, it comes in contact with artificial
surfaces, and the hemostatic system may respond in diverse
ways depending on the nature and the duration of stimuli.
These surfaces will rapidly adsorb plasma proteins at the
blood−material interface, which will mediate subsequent
thrombotic events and clotting. The aPTT and PT are
principally used to examine mainly the contact activation
pathways and common coagulation pathways of blood
coagulation factors.62 aPTT, the global screening procedures
used to evaluate coagulation abnormalities in the intrinsic
pathway, can also be used to detect functional deficiencies in
Factors II, III, V, VIII, X, or fibrinogen.63 Clotting times of the
modified membranes were seen to have improved compared to
the pristine membrane, as seen in Figure 11. The aPTT of
Figure 11. aPTTs and PTs of the membranes.
membrane M4 was found to be twice as much as that of M0.
Moreover, the PT was also slightly increased with the addition
of GO into the polymer matrix. M4 has an aPTT of 75 s, which
demonstrates that there is a delay in clotting. These results
were in accordance with other researchers who introduced
functionalized MWCNTs into the polymer matrix.20
Blood clotting time was considerably improved in PEI
MMMs. The result might have partial contribution from its
surface hydrophilicity since this results in less protein
adsorption and platelet adhesion on the membrane surface.
7909
Article
Thrombosis is usually initiated by the adsorption of plasma
proteins followed by the adhesion of platelets. The high
hydrophilic surface formed on the PEI mixed matrix surface
extended clotting time, lowered protein adsorption, and
suppressed platelet adhesion and activation.
3.9.5. PRT. In the interaction of blood with negatively
charged surfaces, the intrinsic pathway is triggered by surface
contact, and this in turn sets off a cascade of reactions requiring
Ca2+. Plasma recalcification profiles serve as a measure of the
intrinsic coagulation system.64 The presence of turbidity is an
indication of clot formation. Citrated platelet poor plasma
(without the addition of CaCl2) serves as a negative control
since it should not form a clot.
The PRT basically has the same clinical significance as the
whole blood coagulation time (CT). Though it is generally
more sensitive and accurate than CT, and is comparable to the
in vivo experiments.65 The PRT of M0 is similar to the positive
control values. Upon increasing the GO content, the PRT is
found to increase in accordance with the aPTT values, as
observed from Figure 12.
Figure 12. PRT of the different membranes.
3.9.6. Complement Activation. Complement activation is
the body’s defense mechanism against pathogens or any “non-
self” entities. The human complement system consists of 20 or
more different plasma proteins that function either as enzymes
or binding proteins. Complement activation is initiated by
classical or alternative pathways where the terminal pathway is
common to both. Both pathways contain an initial enzyme that
catalyzes the formation of the C3 convertase, which
consecutively generates the C5 convertase allowing the
assembly of the terminal complement complex.66 Complement
activation triggers the host defense mechanism by generating
the localized inflammatory mediator. This could be quantified
by the measurement of the produced anaphylatoxins C3a, C4a,
and C5a. We have determined the concentration of C3a and
C5a using the ELISA assay for the complement activation
study. The assay was also used as a way to characterize the
blood−membrane compatibility of the GO incorporated
MMMs.
Both markers of complement activation C3a and C5a are
elevated in M0 manifesting the generation of acute
inflammatory response of the pristine membranes. C3a and
C5a concentrations were 42 ng/mL and 1.6 ng/mL,
respectively (Figure 13). The membranes loaded with GO
tend to inhibit the complement activation and the membrane
M4 exhibited the least ability to activate the complement
DOI: 10.1021/acs.iecr.5b01655
Ind. Eng. Chem. Res. 2015, 54, 7899−7913
Industrial & Engineering Chemistry Research
Figure 13. Classical pathway and alternate pathway complement activity of the membrane exposed to the serum.
system. These detailed studies might provide useful information
to other studies on material with negatively charged surfaces
and extend the applications of the “heparin-like modification
method”.
3.10. Solute Permeation through the MMMs. In
hemodialysis, the solutes with low molecular weights are
removed from the solution by allowing them to diffuse into a
region of low concentration. Hence, for the MMMs, the
permeability coefficients of the solutes are in concordance with
their molecular weight permeability coefficient of urea >
vitamin B-12 > cytochrome-c. Figure 14 shows the permeability
Figure 14. Permeation coefficient of urea, vitamin B-12, and
cytochrome-c through the membranes.
coefficients of the chosen solutes urea, vitamin B-12, and
cytochrome−c. Urea is a standard solute marker and has a high
clearance during hemodialysis. Its movement in and out of the
RBCs is rapid. The dialysis clearance of urea is greater than the
plasma flow. Therefore, the clearance of urea is higher than
other solutes. Some researchers hypothesize that membrane
solute clearance is not directly correlated with its ultrafiltration
rate;67 however, our results do indicate a correlation between
the two, as seen by Gao et al.68 Though the permeability
coefficients of urea through the membranes M3 and M4 are of
a magnitude smaller than the commercial high flux F-60
dialyzer,69 it is of significance to note that vitamin B-12
clearance is indeed higher than commercial AN-69 and
Cuprophan membranes.4 Therefore, membranes with higher
loading ratio of GO in the nanocomposites demonstrate higher
clearance of even the middle molecular weight surrogate
marker, cytochrome-c. The presence of a greater number of
7910
Article
pores (SEM) and the increase of UFR could explain the greater
clearance of these solutes.
3.11. Cell Cytocompatibility. 3.11.1. MTT Assay. Upon
contact with any biomaterials, cells will undergo certain
morphological changes to stabilize the interface with the
material. MTT is the most common and efficient method to
determine the cell viability after interaction with the membrane
surface. The mitochondrial dehydrogenases of viable cells were
usually cleaved selectively to the tetrazolium ring, yielding
blue−purple formazan crystals, while the level of the reduction
of MTT into the colored formazan was measured spectropho-
tometrically. This reflected the number of cells that were viable.
Figure 15 illustrates the MTT data for the control sample and
the membranes incubated in hepatocyte cells for 24 h in a CO2
incubator at 37 °C and 5% CO2. The formazan absorbance
indicates the quantity of viable cells present on the control (the
polystyrene cell-culture plate) as well as those on the prepared
membranes. It was observed that the viability of the cells on all
of the modified membranes increased compared with the
pristine PEI membranes, which indicated that the addition of
the GO nanoparticles might have a positive influence on
viability over the period of time, and this in turn confirms the
cytocompatibility of these MMMs.
3.11.2. Cell Viability. AO/EtBr staining is used to stain the
live and dead cells following the incubation of membranes in
the cells. The morphological indication of cell apoptosis will
infer the viability of cells upon the interaction with the
membrane surface. In principle, AO can cleave the intact cell
membrane and interact with components of the nucleus, and
hence only the viable cells will be stained green. The ethidium
bromide is a specific cell stain, which imparts the color red to
the apoptotic cells due to the impermeable nature of its cell
membrane. Figure 15 illustrates the AO/EtBr staining of
hepatocyte cells, and from the figures, it is clear that pristine
PEI membrane (M0) shows the highest number of apoptotic
cells. The increase in the concentration of GO (membranes M3
and M4) in the MMMs decreases the number of apoptotic
cells, which are visualized by the decrease in red color stained
cells. Hence, it can be concluded that the number of cells
adhered on to the membrane is less when compared to pristine
PEI, and further adhered cells are viable over the period of time
as confirmed by the increase of cells, which have imbibed the
green stain. Therefore, it can be hypothesized that these
membranes have a potential to be used not only as
hemodialysis membranes, but also other bioartificial organs or
organ supports due to their greater cytocompatibility.
DOI: 10.1021/acs.iecr.5b01655
Ind. Eng. Chem. Res. 2015, 54, 7899−7913
Industrial & Engineering Chemistry Research
Figure 15. MTT assay and live/dead cell staining images.
4. CONCLUSION
The goal of this study was to prepare novel MMMs by
incorporation of hydrophilic GO-PVP nanocomposite into the
PEI matrix. The presence of these nanocomposites on the
membrane surface confirmed by FTIR improved the hydro-
philicity of the membrane surface as characterized by lower
contact angles and the higher work of adhesion. All of the
prepared membranes showed a typical asymmetric structure
with lower surface roughness. The integration of the nano-
composite leads to an increase in porosity and water flux. This
was confirmed by SEM, which revealed a greater number of
longitudinal fully developed finger-like pores. The 0.2 wt % GO
exhibited higher fouling resistance and greater biocompatibility.
These MMMs also had greater biocompatibility with both
blood and cells. They exhibited greater protein adsorption
resistance, suppressed platelet adhesion, and prolonged clotting
time. It can be seen that the uremic toxin removal efficiency
increased as the GO loading concentrations increased.
Overall, the membrane M3 exhibited a high UFR, high
uremic toxin removal efficiency shown to increase cell viability,
prolonged clotting times, suppressed complement activation,
lowered protein adsorption, and low platelet activation, while at
the same time maintaining sufficient clearance of the uremic
toxins. These show a much lower loss of essential proteins like
serum albumin and smoother surface (than M4), which could
prevent it from damaging blood cells in an in vivo study. Hence,
it can be confirmed that the introduction of GO-PVP in the
membrane matrix as a hydrophilic modifier for the PEI
membrane is a safe and efficient method for the development of
novel materials for blood purification applications.
7911
Article
■ AUTHOR INFORMATION
Corresponding Author
*E-mail: mohantarun@gmail.com. Phone: 0091-44-2235-9136.
Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS
This study was supported by the Instrument Development
Program (IDP) of the Department of Science and Technology
(DST), Government of India. The authors would like to thank
Dr. S. Senthil Kumar for valuable discussions regarding the cell
culture studies. We gratefully acknowledge the help extended
by Dr. Saravana Babu, Center for Toxicology and Devel-
opmental Research (CEFT), SRMC for permitting use of lab
space for the cytocompatibility studies. We would also like to
thank Philip George T for the valuable input.
■ REFERENCES
(1) Stamatialis, D. F.; Papenburg, B. J.; Gironés, M.; Saiful, S.;
Bettahalli, S. N. M.; Schmitmeier, S.; Wessling, M. Medical
Applications of Membranes: Drug Delivery, Artificial Organs and
Tissue Engineering. J. Membr. Sci. 2008, 308 (1−2), 1.
(2) Senthilkumar, S.; Rajesh, S.; Mohan, D.; Soundararajan, P.
Preparation, Characterization, and Performance Evaluation of Poly-
(Ether-Imide) Incorporated Cellulose Acetate Ultrafiltration Mem-
brane for Hemodialysis. Sep. Sci. Technol. 2013, 48 (1), 66.
(3) Yokomatsu, A.; Fujikawa, T.; Toya, Y.; Shino-Kakimoto, M.; Itoh,
Y.; Mitsuhashi, H.; Tamura, K.; Hirawa, N.; Yasuda, G.; Umemura, S.
Loss of Amino Acids into Dialysate during Hemodialysis Using
Hydrophilic and Nonhydrophilic Polyester-Polymer Alloy and
DOI: 10.1021/acs.iecr.5b01655
Ind. Eng. Chem. Res. 2015, 54, 7899−7913
Industrial & Engineering Chemistry Research
Polyacrylonitrile Membrane Dialyzers. Ther. Apheresis Dial. 2014, 18
(4), 340.
(4) Mahlicli, F. Y.; Altinkaya, S. A. Immobilization of Alpha Lipoic
Acid onto Polysulfone Membranes to Suppress Hemodialysis Induced
Oxidative Stress. J. Membr. Sci. 2014, 449, 27.
(5) Li, L.; Cheng, C.; Xiang, T.; Tang, M.; Zhao, W.; Sun, S.; Zhao,
C. Modification of Polyethersulfone Hemodialysis Membrane by
Blending Citric Acid Grafted Polyurethane and Its Anticoagulant
Activity. J. Membr. Sci. 2012, 405−406, 261.
(6) Chan, K. H.; Wong, E. T.; Khan, M. I.; Idris, A.; Yusof, N. M.
Fabrication of Polyvinylidene Difluoride Nano-Hybrid Dialysis
Membranes Using Functionalized Multiwall Carbon Nanotube for
Polyethylene Glycol (hydrophilic Additive) Retention. J. Ind. Eng.
Chem. 2014, 20 (5), 3744.
(7) Frank, R. D.; Mueller, U.; Lanzmich, R.; Floege, J. Factor XII
Activation Markers Do Not Reflect FXII Dependence of Thrombin
Generation Induced by Polyvinylchloride. J. Mater. Sci.: Mater. Med.
2013, 24 (11), 2561.
(8) Shakaib, M.; Ahmad, I.; Yunus, R. M.; Noor, M. Z. Preparation
and Characterization of Modified Nylon 66 Membrane for Blood
Purification. Int. J. Polym. Mater. 2014, 63 (2), 80.
(9) Ma, L.; Su, B.; Cheng, C.; Yin, Z.; Qin, H.; Zhao, J.; Sun, S.;
Zhao, C. Toward Highly Blood Compatible Hemodialysis Membranes
via Blending with Heparin-Mimicking Polyurethane: Study in Vitro
and in Vivo. J. Membr. Sci. 2014, 470, 90.
(10) Hoshi, R. A.; Van Lith, R.; Jen, M. C.; Allen, J. B.; Lapidos, K.
A.; Ameer, G. The Blood and Vascular Cell Compatibility of Heparin-
Modified ePTFE Vascular Grafts. Biomaterials 2013, 34 (1), 30.
(11) United States Renal Data System. http://www.usrds.org/
(accessed July 1, 2015).
(12) Yang, Q.; Chung, T.-S.; Santoso, Y. E. Tailoring Pore Size and
Pore Size Distribution of Kidney Dialysis Hollow Fiber Membranes via
Dual-Bath Coagulation Approach. J. Membr. Sci. 2007, 290 (1−2),
153.
(13) Yoshida Kozai, T. D.; Langhals, N. B.; Patel, P. R.; Deng, X.;
Zhang, H.; Smith, K. L.; Lahann, J.; Kotov, N. A.; Kipke, D. R.
Ultrasmall Implantable Composite Microelectrodes with Bioactive
Surfaces for Chronic Neural Interfaces. Nat. Mater. 2012, 11 (12),
1065.
(14) Mao, C.; Qiu, Y.; Sang, H.; Mei, H.; Zhu, A.; Shen, J.; Lin, S.
Various Approaches to Modify Biomaterial Surfaces for Improving
Hemocompatibility. Adv. Colloid Interface Sci. 2004, 110 (1−2), 5.
(15) Rajagopalan, M.; Oh, I.-K. Fullerenol-Based Electroactive
Artificial Muscles Utilizing Biocompatible Polyetherimide. ACS Nano
2011, 5 (3), 2248.
(16) Kim, S.-B.; Jo, J.-H.; Lee, S.-M.; Kim, H.-E.; Shin, K.-H.; Koh,
Y.-H. Use of a Poly(ether Imide) Coating to Improve Corrosion
Resistance and Biocompatibility of Magnesium (Mg) Implant for
Orthopedic Applications. J. Biomed. Mater. Res., Part A 2013, 101A (6),
1708.
(17) Rajagopalan, M.; Jeon, J.-H.; Oh, I.-K. Electric-Stimuli-
Responsive Bending Actuator Based on Sulfonated Polyetherimide.
Sens. Actuators, B 2010, 151 (1), 198.
(18) Albrecht, W.; Seifert, B.; Weigel, T.; Schossig, M.; Holländer, A.;
Groth, T.; Hilke, R. Amination of Poly(ether Imide) Membranes
Using Di- and Multivalent Amines. Macromol. Chem. Phys. 2003, 204
(3), 510.
(19) Ran, F.; Nie, S.; Zhao, W.; Li, J.; Su, B.; Sun, S.; Zhao, C.
Biocompatibility of Modified Polyethersulfone Membranes by
Blending an Amphiphilic Triblock Co-Polymer of Poly(vinyl
Pyrrolidone)-B-Poly(methyl Methacrylate)-B-Poly(vinyl Pyrrolidone).
Acta Biomater. 2011, 7 (9), 3370.
(20) Nie, C.; Ma, L.; Xia, Y.; He, C.; Deng, J.; Wang, L.; Cheng, C.;
Sun, S.; Zhao, C. Novel Heparin-Mimicking Polymer Brush Grafted
Carbon nanotube/PES Composite Membranes for Safe and Efficient
Blood Purification. J. Membr. Sci. 2015, 475, 455.
(21) Yue, W.-W.; Li, H.-J.; Xiang, T.; Qin, H.; Sun, S.-D.; Zhao, C.-S.
Grafting of Zwitterion from Polysulfone Membrane via Surface-
7912
Article
Initiated ATRP with Enhanced Antifouling Property and Biocompat-
ibility. J. Membr. Sci. 2013, 446, 79.
(22) Zhu, L.-J.; Liu, F.; Yu, X.-M.; Gao, A.-L.; Xue, L.-X. Surface
Zwitterionization of Hemocompatible Poly(lactic Acid) Membranes
for Hemodiafiltration. J. Membr. Sci. 2015, 475, 469.
(23) Dahe, G. J.; Teotia, R. S.; Kadam, S. S.; Bellare, J. R. The
Biocompatibility and Separation Performance of Antioxidative
Polysulfone/vitamin E TPGS Composite Hollow Fiber Membranes.
Biomaterials 2011, 32 (2), 352.
(24) Yi, Z.; Zhu, L.; Xu, Y.; Jiang, J.; Zhu, B. Polypropylene Glycol:
The Hydrophilic Phenomena in the Modification of Polyethersulfone
Membranes. Ind. Eng. Chem. Res. 2011, 50 (19), 11297.
(25) Wajid, A. S.; Das, S.; Irin, F.; Ahmed, H. S. T.; Shelburne, J. L.;
Parviz, D.; Fullerton, R. J.; Jankowski, A. F.; Hedden, R. C.; Green, M.
J. Polymer-Stabilized Graphene Dispersions at High Concentrations in
Organic Solvents for Composite Production. Carbon 2012, 50 (2),
526.
(26) Mazaheri, M.; Akhavan, O.; Simchi, A. Flexible Bactericidal
Graphene Oxide−chitosan Layers for Stem Cell Proliferation. Appl.
Surf. Sci. 2014, 301, 456.
(27) Valcárcel, M.; Cárdenas, S.; Simonet, B. M.; Moliner-Martínez,
Y.; Lucena, R. Carbon Nanostructures as Sorbent Materials in
Analytical Processes. TrAC, Trends Anal. Chem. 2008, 27 (1), 34.
(28) Chen, J.; Yao, B.; Li, C.; Shi, G. An Improved Hummers
Method for Eco-Friendly Synthesis of Graphene Oxide. Carbon 2013,
64, 225.
(29) Abe, Y.; Mochizuki, A. Hemodialysis Membrane Prepared from
cellulose/N-Methylmorpholine-N-Oxide Solution. II. Comparative
Studies on the Permeation Characteristics of Membranes Prepared
fromN-Methylmorpholine-N-Oxide and Cuprammonium Solutions. J.
Appl. Polym. Sci. 2003, 89 (2), 333.
(30) Kee, C. M.; Idris, A. Permeability Performance of Different
Molecular Weight Cellulose Acetate Hemodialysis Membrane. Sep.
Purif. Technol. 2010, 75 (2), 102.
(31) Hadidi, M.; Zydney, A. L. Fouling Behavior of Zwitterionic
Membranes: Impact of Electrostatic and Hydrophobic Interactions. J.
Membr. Sci. 2014, 452, 97.
(32) Lin, W.-C.; Liu, T.-Y.; Yang, M.-C. Hemocompatibility of
Polyacrylonitrile Dialysis Membrane Immobilized with Chitosan and
Heparin Conjugate. Biomaterials 2004, 25 (10), 1947.
(33) Ran, F.; Niu, X.; Song, H.; Cheng, C.; Zhao, W.; Nie, S.; Wang,
L.; Yang, A.; Sun, S.; Zhao, C. Toward a Highly Hemocompatible
Membrane for Blood Purification via a Physical Blend of Miscible
Comb-like Amphiphilic Copolymers. Biomater. Sci. 2014, 2 (4), 538.
(34) Langsdorf, L. J.; Zydney, A. L. Diffusive and Convective Solute
Transport through Hemodialysis Membranes: A Hydrodynamic
Analysis. J. Biomed. Mater. Res. 1994, 28 (5), 573.
(35) Dimiev, A. M.; Tour, J. M. Mechanism of Graphene Oxide
Formation. ACS Nano 2014, 8 (3), 3060.
(36) Shang, J.; Ma, L.; Li, J.; Ai, W.; Yu, T.; Gurzadyan, G. G. The
Origin of Fluorescence from Graphene Oxide. Sci. Rep. 2012, 2, 792.
(37) Zhang, K.; Zhang, Y.; Wang, S. Enhancing Thermoelectric
Properties of Organic Composites through Hierarchical Nanostruc-
tures. Sci. Rep. 2013, 3, 3448.
(38) Paredes, J. I.; Villar-Rodil, S.; Martínez-Alonso, A.; Tascón, J. M.
D. Graphene Oxide Dispersions in Organic Solvents. Langmuir 2008,
24 (19), 10560.
(39) Romero, A. I.; Parentis, M. L.; Habert, A. C.; Gonzo, E. E.
Synthesis of Polyetherimide/silica Hybrid Membranes by the Sol−gel
Process: Influence of the Reaction Conditions on the Membrane
Properties. J. Mater. Sci. 2011, 46 (13), 4701.
(40) Zhao, C.; Xu, X.; Chen, J.; Yang, F. Effect of Graphene Oxide
Concentration on the Morphologies and Antifouling Properties of
PVDF Ultrafiltration Membranes. J. Environ. Chem. Eng. 2013, 1 (3),
349.
(41) Xu, Z.; Zhang, J.; Shan, M.; Li, Y.; Li, B.; Niu, J.; Zhou, B.; Qian,
X. Organosilane-Functionalized Graphene Oxide for Enhanced
Antifouling and Mechanical Properties of Polyvinylidene Fluoride
Ultrafiltration Membranes. J. Membr. Sci. 2014, 458, 1.
DOI: 10.1021/acs.iecr.5b01655
Ind. Eng. Chem. Res. 2015, 54, 7899−7913
Industrial & Engineering Chemistry Research Article

(42) Ganesh, B. M.; Isloor, A. M.; Ismail, A. F. Enhanced (60) Siljander, P.; Carpen, O.; Lassila, R. Platelet-Derived Micro-
Hydrophilicity and Salt Rejection Study of Graphene Oxide- particles Associate with Fibrin during Thrombosis. Blood 1996, 87
Polysulfone Mixed Matrix Membrane. Desalination 2013, 313, 199. (11), 4651.
(43) Zhao, H.; Wu, L.; Zhou, Z.; Zhang, L.; Chen, H. Improving the (61) Hong, J.; Nilsson Ekdahl, K.; Reynolds, H.; Larsson, R.; Nilsson,
Antifouling Property of Polysulfone Ultrafiltration Membrane by B. A New in Vitro Model to Study Interaction between Whole Blood
Incorporation of Isocyanate-Treated Graphene Oxide. Phys. Chem. and Biomaterials. Studies of Platelet and Coagulation Activation and
Chem. Phys. 2013, 15 (23), 9084. the Effect of Aspirin. Biomaterials 1999, 20 (7), 603.
(44) Vatanpour, V.; Esmaeili, M.; Farahani, M. H. D. A. Fouling (62) Su, B.; Fu, P.; Li, Q.; Tao, Y.; Li, Z.; Zao, H.; Zhao, C.
Reduction and Retention Increment of Polyethersulfone Nano- Evaluation of Polyethersulfone Highflux Hemodialysis Membrane in
filtration Membranes Embedded by Amine-Functionalized Multi- Vitro and in Vivo. J. Mater. Sci.: Mater. Med. 2008, 19 (2), 745.
Walled Carbon Nanotubes. J. Membr. Sci. 2014, 466, 70. (63) Flanders, M. M. Comparison of Five Thrombin Time Reagents.
(45) Chen, Z.; Rana, D.; Matsuura, T.; Yang, Y.; Lan, C. Q. Study on Clin. Chem. 2003, 49 (1), 169.
(64) Zhao, W.; Mou, Q.; Zhang, X.; Shi, J.; Sun, S.; Zhao, C.
the Structure and Vacuum Membrane Distillation Performance of
Preparation and Characterization of Sulfonated Polyethersulfone
PVDF Composite Membranes: I. Influence of Blending. Sep. Purif.
Membranes by a Facile Approach. Eur. Polym. J. 2013, 49, 738−751.
Technol. 2014, 133, 303. (65) Jiang, J.-H.; Zhu, L.-P.; Li, X.-L.; Xu, Y.-Y.; Zhu, B.-K. Surface
(46) Rajesh, S.; Shobana, K. H.; Anitharaj, S.; Mohan, D. R. Modification of PE Porous Membranes Based on the Strong Adhesion
Preparation, Morphology, Performance, and Hydrophilicity Studies of of Polydopamine and Covalent Immobilization of Heparin. J. Membr.
Poly(amide-Imide) Incorporated Cellulose Acetate Ultrafiltration Sci. 2010, 364 (1−2), 194.
Membranes. Ind. Eng. Chem. Res. 2011, 50 (9), 5550. (66) Gorbet, M. B.; Sefton, M. V. Biomaterial-Associated
(47) Tsai, W.-B.; Grunkemeier, J. M.; McFarland, C. D.; Horbett, T. Thrombosis: Roles of Coagulation Factors, Complement, Platelets
A. Platelet Adhesion to Polystyrene-Based Surfaces Preadsorbed with and Leukocytes. Biomaterials 2004, 25 (26), 5681.
Plasmas Selectively Depleted in Fibrinogen, Fibronectin, Vitronectin, (67) Clark, W. R.; Hamburger, R. J.; Lysaght, M. J. Effect of
or von Willebrand’s Factor. J. Biomed. Mater. Res. 2002, 60 (3), 348. Membrane Composition and Structure on Solute Removal and
(48) Willem, N.; Charles, A. H. Proteins at Interfaces II; Horbett, T. Biocompatibility in Hemodialysis. Kidney Int. 1999, 56 (6), 2005.
A., Brash, J. L., Eds.; ACS Symposium Series; American Chemical (68) Gao, A.; Liu, F.; Shi, H.; Xue, L. Controllable Transition from
Society: Washington, DC, 1995; Vol. 602. Finger-like Pores to Inter-Connected Pores of PLLA Membranes. J.
(49) Chen, H.; Yuan, L.; Song, W.; Wu, Z.; Li, D. Biocompatible Membr. Sci. 2015, 478, 96.
Polymer Materials: Role of Protein−surface Interactions. Prog. Polym. (69) Eloot, S.; Vierendeels, J.; Verdonck, P. Optimisation of Solute
Sci. 2008, 33 (11), 1059. Transport in Dialysers Using a Three-Dimensional Finite Volume
(50) Tang, M.; Xue, J.; Yan, K.; Xiang, T.; Sun, S.; Zhao, C. Heparin- Model. Comput. Methods Biomech. Biomed. Engin. 2006, 9 (6), 363.
like Surface Modification of Polyethersulfone Membrane and Its
Biocompatibility. J. Colloid Interface Sci. 2012, 386 (1), 428.
(51) Zhao, X.; Ma, J.; Wang, Z.; Wen, G.; Jiang, J.; Shi, F.; Sheng, L.
Hyperbranched-Polymer Functionalized Multi-Walled Carbon Nano-
tubes for Poly (vinylidene Fluoride) Membranes: From Dispersion to
Blended Fouling-Control Membrane. Desalination 2012, 303, 29.
(52) Liu, T.-Y.; Lin, W.-C.; Huang, L.-Y.; Chen, S.-Y.; Yang, M.-C.
Hemocompatibility and Anaphylatoxin Formation of Protein-Immo-
bilizing Polyacrylonitrile Hemodialysis Membrane. Biomaterials 2005,
26 (12), 1437.
(53) Senthilkumar, S.; Rajesh, S.; Jayalakshmi, A.; Aishwarya, G.; Raju
Mohan, D. Preparation and Performance Evaluation of Poly (ether-
Imide) Incorporated Polysulfone Hemodialysis Membranes. J. Polym.
Res. 2012, 19 (6), 9867.
(54) Zhao, C.; Liu, X.; Rikimaru, S.; Nomizu, M.; Nishi, N. Surface
Characterization of Polysulfone Membranes Modified by DNA
Immobilization. J. Membr. Sci. 2003, 214 (2), 179.
(55) Grunkemeier, J. M.; Tsai, W. B.; McFarland, C. D.; Horbett, T.
A. The Effect of Adsorbed Fibrinogen, Fibronectin, von Willebrand
Factor and Vitronectin on the Procoagulant State of Adherent
Platelets. Biomaterials 2000, 21 (22), 2243.
(56) Corum, L. E.; Hlady, V. The Effect of Upstream Platelet-
Fibrinogen Interactions on Downstream Adhesion and Activation.
Biomaterials 2012, 33 (5), 1255.
(57) Bailly, A. L.; Laurent, A.; Lu, H.; Elalami, I.; Jacob, P.; Mundler,
O.; Merland, J. J.; Lautier, A.; Soria, J.; Soria, C. Fibrinogen Binding
and Platelet Retention: Relationship with the Thrombogenicity of
Catheters. J. Biomed. Mater. Res. 1996, 30 (1), 101.
(58) Jin, S.; Zhou, N.; Xu, D.; Wu, Y.; Tang, Y.; Lu, C.; Zhang, J.;
Shen, J. Synthesis and Anticoagulation Activities of Polymer/
functional Graphene Oxide Nanocomposites via Reverse Atom
Transfer Radical Polymerization (RATRP). Colloids Surf., B 2013,
101, 319.
(59) Blezer, R.; Willems, G. M.; Cahalan, P. T.; Lindhout, T.
Initiation and Propagation of Blood Coagulation at Artificial Surfaces
Studied in a Capillary Flow Reactor. Thromb. Haemost. 1998, 79 (2),
296.

7913 DOI: 10.1021/acs.iecr.5b01655

Ind. Eng. Chem. Res. 2015, 54, 7899−7913