In vitro cytocompatibility and blood compatibility of polysulfone blend, surface-modified polysulfone and polyacrylonitrile membranes for hemodialysis

Abstract

The fabrication of dialysis membranes with significant biocompatibility is an active area of research. In this context, three types of asymmetric flat sheet membranes were fabricated and compared for potential use as hemodialysis membranes. A polysulfone–polyvinylpyrrolidone and polyethylene glycol-based polymer blend membrane, a polysulfone membrane surface-modified with trimesoyl chloride and m-phenylene diamine, and a polyacrylonitrile membrane were synthesized. All three types of membrane were characterized in terms of their surface morphology, permeability, hydrophilicity, surface charge, porosity and mechanical strength. They were then subjected to comprehensive cytocompatibility and hemocompatibility tests as well as analysing the transport of uremic toxins. On the basis of protein adsorption, oxidative stress, cell proliferation and adhesion, all three membranes were comparable. However, the blend and surface-modified membranes showed excellent results for hemolysis, platelet adhesion, blood cell aggregation and degree of thrombus formation. All these results indicated the suitability of the blend and surface-modified membranes for possible dialysis applications.

---

1. Introduction

Dialysis in the wake of kidney failure, or acute kidney injury (AKI),^1–3 is a lifeline for survival of the patient. Replacing the kidney function with a membrane module in an extracorporeal circuit has resulted in shifting the focal point of nephrological research to the development of biocompatible membranes. In the past several decades, importance has been given to polymeric materials, blends and surface modifications to achieve better dialysis-grade membranes with greater efficacy and improved biocompatibility. In this endeavour, cellulose acetate (CA) has been used since the 1850’s, following the pioneering work of Graham⁴ and Fick,⁵ exhibiting its potential in ultrafiltration and dialysis. However, a whole gamut of problems, viz. low flux⁶ and complement activation,^7,8 was encountered by practising clinicians using CA membranes. Ever since the Artifical Kidney-Chronic Uremia program was launched by NIAMD (National Institute of Arthritis and Metabolic Diseases) in the year 1966, materials scientists and engineers have started looking into the possibility of developing synthetic polymers for dialysis with significant hemocompatibility.

This led to the emergence of different materials with improved hemocompatibility. Therefore, polyacrylonitrile^9–11 (PAN) followed by polysulfone^12–14 (PSf) based membranes started to gain popularity in hemodialysis applications. Polyethersulfone (PES) blended with citric acid-grafted polyurethane¹⁵ has been reported as a possible dialysis membrane. Carbon nanotube-grafted PES composite membranes,¹⁶ chemically modified PSf,¹⁷ vitamin E–TPGS (D-α-tocopheryl polyethylene glycol 1000 succinate) composite membranes,¹⁸ and polyamide and monosodium glutamate blend membranes have also been explored for dialysis applications.^19,20 However, biocompatibility is not the only issue for the selection of membranes by practising clinicians. The development of high-performance membranes has become a pivotal issue in dialysis treatment, since higher and faster clearance of uremic toxins increases the patients’ longevity.^21,22 In fact, PSf-based membranes have a higher clearance rate for uremic toxins²³ and have become a prime choice for clinicians administering dialysis, taking into account both its biocompatibility and transport properties.²⁴

In view of the above discussion, it is clear that variants of PSf and PAN are suitable materials for hemodialysis applications. Therefore, the present article details an attempt to formulate three types of dialysis grade membranes based on PSf and PAN, and to evaluate their in vitro cell and blood compatibility as well as uremic toxin transport capabilities. The underlining step in synthesizing these membranes is attaining a specific molecular weight cut-off (MWCO) of around 6–16 kDa. Here, this was achieved through three different methodologies. First, a polymer blend membrane (S1) was synthesized using polysulfone (PSf) as a base material, blended with polyvinylpyrrolidone (PVP) and polyethylene glycol (PEG) with 6 kDa MWCO. Second, a PSf and PVP blend membrane (S2) was synthesized via surface treatment using trimesoyl chloride (TMC) and m-phenylene diamine (MPD) to yield a similar cut-off. Interestingly, TMC and MPD are in widespread use in synthesizing reverse osmosis membranes,²⁵ but have not yet been reported for their potential use in dialysis membranes. Lastly, a polyacrylonitrile (PAN) homopolymer membrane (S3) was prepared with a dialysis grade MWCO.²⁶ All the membranes were characterized for permeability, hydrophilicity, porosity, membrane morphology and mechanical strength. Detailed cytocompatibility and hemocompatibility analyses were carried out to evaluate biological activity. Finally, the performances of the membranes were quantified in terms of their urea and creatinine permeances. The results were interpreted and a final recommendation was made in terms of performance. Therefore, the novelty of this work includes the formulation of PSf–PVP–PEG, surface modified PSf–PVP and PAN homopolymer membranes and exploring their suitability for hemodialysis purposes.

2. Materials and methods

2.1 Membrane synthesis

2.1.1 Polymer blend membrane. Polysulfone (18 wt%, average molecular weight 22 400 Da, supplied by Solvay Chemicals, Mumbai, India), PVP (1, 2 and 3 wt%, molecular weight 40 000 Da, supplied by Sigma Aldrich, Missouri, USA) and PEG ( 3 wt%, supplied by S R Ltd., Mumbai, India) were dissolved in dimethyl formamide (DMF) (supplied by Merck (India) Mumbai Ltd.), by stirring over a magnetic stirrer (supplied by Anupam enterprises, Kharagpur, India) at around 60 °C for over 10 h. The solution was then kept overnight for degassing and was cast the next day over a non-woven fabric support (118 ± 22.8 μm thickness, supplied by Hollytex, India Inc., New York, USA). A film of thickness 150 μm was cast using a doctor’s blade (fabricated and supplied by Gurpreet Engineering Works, Kanpur, India).

2.1.2 Polysulfone surface-modified membrane. Polysulfone (18 wt%) and PVP (1 wt%) were dissolved in DMF with the help of a stirrer as described in the previous section. The solution was kept overnight for degassing and was cast in the same manner as described in the previous section. The cast membrane underwent surface treatment using TMC (supplied by Merck (India) Ltd.) and MPD (supplied by Merck (India) Ltd.). The membrane was first heated at 75 °C for 10 min. Then it was immersed in 2% aqueous MPD solution for 5 min and air dried for 15 min. It was taken out and immersed in a 0.1% TMC solution (dissolved in hexane) for 5 min and then exposed to air for 15 min. Finally, it was immersed in distilled water and left overnight.

2.1.3 Polyacrylonitrile (PAN) membrane. PAN (supplied by Technorbital Advanced Materials Pvt. Ltd., Kanpur, India) was dissolved in DMF at around 60 °C and was mechanically stirred (stirrer supplied by Anupam Enterprises, Kharagpur, India) for around 2 hours. Then, it was cast as described in the previous section.

2.2 Membrane characterization

2.2.1 Permeability and MWCO. The MWCO and permeability of the cast membrane was determined using a stirred batch cell.²⁷ Firstly, the nascent membrane was compacted at 690 kPa for 3 hours and the permeate flux flow rate noted at five different transmembrane pressures.²⁸ The permeate flux was calculated by:where, v_w is the pure water flux, Q is the volumetric flow rate of permeating water, A is the effective filtration area (33.16 cm²) and ΔT is the sampling time. A plot of v_w against transmembrane pressure drop resulted in a straight line passing through the origin, the slope of which gave the hydraulic permeability of the membrane.

Polyethylene glycol (PEG) of various molecular weights was supplied by S R Ltd., Mumbai, India. The molecular weights were 1000, 4000, 10 000, 20 000, 70 000 and 100 000 Da, and they are essentially neutral polymers. A 10 kg m⁻³ solution of each, prepared by separately dissolving the polymers in distilled water, was fed to a stirred batch cell.²⁶ A low transmembrane pressure (70 kPa) and high stirring speed (2000 rpm) were applied to minimize the concentration polarization layer, and the permeate was collected at intervals of five minutes and the percentage rejection (%R) measured:

where, C_P is the concentration of the permeate and C_F is the concentration of the feed. The rejection was calculated (eqn (2)) and plotted against the molecular weight of the solutes. The point of 90% rejection of solutes corresponds to the MWCO of the membrane.

2.2.2 Porosity. Porosity is measured from the difference between the wet weight and dry weight of the membrane. Membranes of specific dimensions were cut (2 cm × 2 cm), immersed in distilled water and taken out after 5 min. The surface water was dried off and their weights were measured (w₀). After this, they were placed in an air-circulating oven at 60 °C for 24 h and further dried in a vacuum oven. Their dry weight (w_i) was then measured and the porosity was calculated:²⁹where ε is the membrane porosity, A is the area of the membrane, l is the membrane thickness and ρ_w is water density. The membrane porosity was measured three times and average values are reported.

2.2.3 Tensile strength. The mechanical strengths of the membranes, as a function of yield stress, were studied using a universal testing machine, procured from Tinius Olsen Ltd., Redhill, England, model number H50KS. All measurements were carried out at room temperature and at a strain rate of 20 mm min⁻¹. The set of experiments consisted of three repetitions for each sample.

2.2.4 Contact angle. Contact angles were measured with a goniometer (New Jersey, USA, rame’-hart, model no. 200-F4) using the sessile drop method.³⁰ Contact angles at six different locations were measured and the average value was reported.

2.2.5 Membrane morphology. The cast membranes were dried in a desiccator overnight and then were dipped in liquid nitrogen and fractured. They were gold coated and placed on stubs for SEM imaging (model: ESM-5800, JEOL, Japan) at the desired magnification.

2.2.6 Surface charge measurement. The surface charges of the three membranes were measured in an electroultrafiltration cell.³¹ The operating conditions were: temperature = 298 ± 2.0 K, transmembrane pressure = 0–2 bar, solution pH = 7.4, NaCl concentration = 0.01 M and cross flow velocity = 0.12 m s⁻¹. The streaming potential coefficient (V_P) was determined from the slope of the plot between the potential difference (ΔV) and the pressure differential (ΔP) applied when the net current is zero. The relevant equations are:where ζ is the membrane zeta potential, ε₀ is the permittivity in vacuum, D_i is the dielectric constant of the medium, and μ and λ are the viscosity and conductivity of the feed solution.

2.3 Biological assessments of membranes

For in vitro biological assessments, NIH3T3 (mouse embryonic fibroblast cell line) cells were procured from the National Centre for Cell Science (NCCS) Pune, India. NIH3T3 cells were grown up to confluence in medium containing alpha-modified essential medium (αMEM) (12561-056, Invitrogen Life Sciences, India) with 1% antibiotics, antimyotic solution (penicillin 100 μg ml⁻¹, streptomycin 10 μg ml⁻¹, and amphotericin-B 25 μg ml⁻¹; A002A Himedia, India) and 10% fetal bovine serum (Himedia, India), at 37 °C, 95% humidity and 5% CO₂ (Heracell150i, Thermo, USA).

For the assessment of biological activity, the membranes were cut with identical dimensions, sterilized and soaked in cell culture medium overnight. For each experiment, 1 × 10⁴ cells per cm² were seeded on the samples and a control in a 12 well cell culture plate. Subsequently, the required volume of medium was added to each well and cultured for a particular time interval with respect to the assay type. All the assays were performed in triplicate and their mean values reported. Commercially available dialysis fiber (Fresenius F6) was used as a control in the experiments.

2.3.1 Metabolic activity. Preliminary cytocompatibility of the prepared membranes was assessed to evaluate the leaching of any toxic chemicals or any other adverse effect on the cells by measuring the metabolic activity using the MTT (3-[4,5-dimethylthiazol-2-yl]-2,5 diphenyltetrazolium bromide) dye reduction assay on days three and seven, according to a previous report.³²

On the respective day of the assay, the cell-seeded membranes were rinsed with phosphate buffered saline (PBS), further incubated with 200 μl of 5 mg ml⁻¹ MTT solution (M5655, Sigma), in the dark under standard cell culture conditions. The dehydrogenase enzymes of metabolically active cells reduced the pale yellow MTT reagent to soluble purple-colored formazan crystals. The formazan product was dissolved in dimethyl sulfoxide (DMSO) and the absorbance was measured at 570 nm on a microplate reader (Recorders and Medicare Systems, India). The absorbance was considered as proportional to living and growing cells.

2.3.2 Cell proliferation. DNA quantification assays were carried out on days three and seven to evaluate the cell proliferation response of the seeded cells on the membranes. After seeding, the DNA content was measured using a DNA Quantitation Kit, Fluorescence Assay (DNAQF, Sigma), according to the manufacturer’s protocol. Double-stranded DNA binds primarily with the fluorescent dye, bisbenzimide H 33258 (Hoechst 33258), which was measured fluorometrically at an excitation wavelength of 350 nm and an emission wavelength of 460 nm. A standard DNA concentration curve was plotted using a standard solution of calf thymus DNA (D4810, Sigma).

2.3.3 Oxidative stress analysis. Cell sample interactions may enhance intercellular reactive oxygen species (ROS) production, which affects the cellular microenvironment. Therefore, the release of ROS was quantitatively measured using a di-chloro di-hydrofuran fluorescein di-acetate (DCFH-DA) assay according to a previously mentioned report.³³ Cell-seeded wells with samples were rinsed with PBS and incubated with a 1 mM methanolic DCFH-DA solution (Sigma-Aldrich) at 37 °C for 1 h. Subsequently, the fluorescence intensity of the cells interacting with DCFH-DA was measured with excitation at 485 nm and emission at 530 nm on a fluorescence spectrometer (Perkin Elmer, UK).

2.3.4 Estimation of total protein content. Protein adsorption is an important parameter for evaluating the biological response towards dialysis membrane application. Protein adsorption was measured by two different methods. In each method, an equal size (2 cm² surface area) of membrane samples was used.
2.3.4.1 Indirect method. NIH3T3 cells were seeded on the membranes for three and five days. The total protein concentration of the cell-cultured samples was quantified using a bicinchoninic acid (BCA) protein assay.³⁴ Briefly, PBS-rinsed cell-seeded samples were incubated with the BCA working solution (50 parts of BCA reagent with one part of 4% copper sulfate pentahydrate, green colored solution) at 37 °C for 30 min. During the incubation, free amino acids were reduced and formed a crimson-colored complex with BCA. The concentration of this colored complex was assessed from the absorbance at 562 nm on a microplate reader (Recorders and Medicare Systems, India). A standard protein concentration curve was plotted with known concentrations of bovine serum albumin.
2.3.4.2. Direct method. The membranes were incubated in phosphate buffer solution (0.02 M, pH 7.4), containing bovine serum albumin (BSA, 5 g dl⁻¹), human γ-globulin (1.5 g dl⁻¹) and human fibrinogen (0.45 g dl⁻¹), at 37 °C for 2 hours. Subsequently, the samples were gently rinsed with PBS three times. The samples were kept in a 1 wt% aqueous solution of sodium dodecyl sulfate (SDS) for 60 min at room temperature on a shaker. Adsorbed proteins were removed from the samples and measured using a bicinchoninic acid (BCA) protein assay.³⁴

2.3.5 Cell attachment and morphology. Cell attachment and morphology of NIH3T3 cells were evaluated by scanning electron microscopy (SEM) and fluorescence microscopy on days three and seven.³⁵ For SEM analysis, cell-seeded film samples were rinsed gently with PBS, fixed with 4% paraformaldehyde at 37 °C, further dehydrated with gradient ethanol solution and vacuum dried overnight. Prior to SEM, the samples were gold-coated (Polaron, UK).

For fluorescence imaging, cell-seeded membranes were washed thrice with PBS and the cells were fixed with 4% paraformaldehyde followed by permeabilization of the cells using cell lysis solution (0.1% Triton-X in PBS). Cells fixed on the films were stained with Hoechst dye (H1399, Invitrogen Life Sciences) according to the manufacturer’s instructions. Images were acquired using an Axio Observer Z1 (Carl Zeiss, Germany).

2.4 Hemocompatibility tests

Hemocompatibility analysis of the prepared membranes is a requisite towards dialysis membrane applications. For the hemocompatibility assays, whole blood was collected from healthy donors in polyethylene disposable syringes containing 4.9% citrate–phosphate–dextrose–adenine (CPDA) solution. The blood was mixed well with anticoagulant solution and the following tests were performed as presented in the subsections. Commercially available fiber as a control was not included or compared in the hemocompatibility studies due to the difficulty in obtaining active (inner) surface blood contact. The sample size was kept similar in each test and each sample was equilibrated with normal saline via incubation for one hour before testing. All the assays were performed in triplicate and their mean values reported.

2.4.1 Hemolysis assay. Hemolysis assays were carried out to evaluate the red blood cell (RBC) compatibility of the samples. Normal saline-equilibrated samples were added to freshly collected uncoagulated blood and incubated for 1 hour at 37 °C, 95% humidity and 5% CO₂ (Heracell150i, Thermo, USA). Subsequently, the RBC lytic activity was quantified by measuring the optical density at 540 nm. Normal saline and 1% Triton-X solution were used as positive and negative controls.

2.4.2 Blood cell aggregation. To ascertain changes in the surface properties of blood cells, a blood cell aggregation study was conducted. Blood cell aggregation was carried out by modification of a previously reported method.² For the RBC aggregation study, freshly collected blood was centrifuged at 700 rpm. The collected pellet was resuspended with normal saline in a 1 : 9 volume ratio. Subsequently, 100 μl of this solution was mixed with 600 μl of normal saline. Equal sized-membranes were incubated with the prepared suspension for 1 hour at 37 °C. For the white blood cell (WBC) aggregation study, WBCs were isolated from uncoagulated freshly isolated blood using the Ficoll-Paque mononuclear cell isolation principle with HiSep™ LSM-1077 (Himedia) according to the manufacturer’s instructions. Isolated WBCs were mixed with normal saline and incubated with membranes as previously. After incubation, the cell suspension was smeared on a glass slide and observed under a microscope (Axio Observer Z1Carl Zeiss, Germany).

2.4.3 Platelet adhesion. Fresh blood with anticoagulant was centrifuged at 1500 rpm to collect platelet rich plasma (PRP). Normal saline pre-equilibrated samples were incubated with PRP blood for 2 hour at 37 °C, 95% humidity and 5% CO₂ (Heracell150i, Thermo, USA). After incubation, the samples were rinsed gently with normal saline, fixed using 4% paraformaldehyde at 37 °C, further dehydrated with gradient ethanol solution and vacuum dried overnight. Before optical characterization using SEM, samples were gold coated (Polaron, UK).

2.4.4 Thrombus formation. Normal saline-equilibrated samples were incubated with freshly collected whole human blood in a 24 well plate for 2 hour at 37 °C, 95% humidity and 5% CO₂ (Heracell150i, Thermo, USA). Subsequently, the samples were rinsed gently three times with normal saline, fixed using 4% paraformaldehyde at 37 °C, further dehydrated with gradient ethanol solution and vacuum dried overnight. The degree of thrombosis (DOT) was measured using a published method³⁶ as follows:where W_t is the weight of the blood-treated sample and W_d is the dry weight of the sample before blood treatment.

2.5 Urea and creatinine transport

The cast membranes were tested in terms of their urea and creatinine permeances. The set up is described in Fig. 1. Flat sheet membranes, which were cast as described in the previous sections, were cut to fit into the cross flow membrane module. The peristaltic pump drives the feed fluid from the feed tank through the rotameter into the membrane module. The dialysate side fluid is pumped from the dialysate tank by the peristaltic pump through the rotameter into the membrane module. The urea and creatinine permeate through the membrane from the feed to the dialysate side, flowing in a cross flow pattern. The flow rate of the feed side was maintained at 250 ml min⁻¹ and that of the dialysate side was 250 and 500 ml min⁻¹. The concentrations of urea and creatinine in the feed were 500 mg l⁻¹ and 20 mg l⁻¹, respectively.

Fig. 1 Experimental set up for measuring urea and creatinine permeances: (a) feed tank; (b) peristaltic pump; (c) rotameter; (d) pressure gauge (sphygmomanometer); (e) cross flow membrane module; (f) pressure gauge (sphygmomanometer); (g) rotameter; (h) dialysate tank; (i) peristaltic pump.

2.6 Statistical analysis

All the experiments were carried out in triplicate. A two-tailed Student’s t-test was carried out for all the data sets and expressed as a mean ± standard deviation (SD). Differences were considered to be significant at p < 0.05. The SD value of each measurement is presented in different figures.

3. Results and discussion

3.1 Permeability, MWCO and contact angle

The permeabilities and contact angles for the three membranes are shown in Fig. 2. It is evident from this figure that PAN (S3) is the most hydrophobic membrane with the highest contact angle of 80°. This higher degree of hydrophobicity is linked to depleted hydrogen-bond interaction sites with water. Due to this, hydrophobic solutes experience a spontaneous adsorption onto the membrane,³⁷ resulting in fouling of the membrane surface. However, in the case of dialysis, this problem of protein adsorption leads to bigger complications, such as complement activation. This phenomenon also results in lower flux. The S3 membrane has a permeability of around 0.2 × 10⁻¹⁰ m s⁻¹ Pa, which is the least of the three membranes. The surface-modified membrane (S2) has a higher contact angle than S1, but lower than S3. This can be attributed to the fact that adding PVP to PSf induces hydrophilicity, which reduces the contact angle to 73°. The contact angle of S1 is 69°, the lowest of the three membranes, since it has two hydrophilic polymers in its blend, viz. PVP and PEG. This hydrophilic nature is reflected in the permeability results as well. The permeability of S2 is 0.6 × 10⁻¹⁰ m s⁻¹ Pa, and that of S1 is 1.4 × 10⁻¹⁰ m s⁻¹ Pa. The MWCO of all the membranes is 6 kDa, as presented in Fig. 2(b).

Fig. 2 (a) Permeabilities and contact angles of the three membranes. (b) MWCO of S1, S2 and S3 membranes.

3.2 Porosity, surface morphology and tensile strength

The cross section images show the typical formation of a phase inversion membrane, i.e. a thin skin followed by a porous sub structure, and a spongy bottom layer. The surface morphology can be attributed to the addition of hydrophilic polymers to the blend. During phase inversion, the addition of hydrophilic polymers induces more water flux into the membrane structure, thereby increasing the number of macrovoids. This is reflected in the SEM images (Fig. 3), where S1 has a more porous structure than S2 and S3. However, the skin thicknesses of all three membranes are almost the same and this shows that even though the porous nature of the membranes varies, the MWCO of the membranes being the same, the skin thickness of the membranes is comparable. The porosities and tensile strengths are shown in Fig. 4. It is evident from the above discussion that the porosity of the membranes increases in the order of S3 < S2 < S1. While S3 has a porosity of 54%, S1 has a porosity of 62%. S2 has an intermediate porosity of 60%. The porosity is an exact mirror reflection of the breaking stress relationship. It follows a similar trend, i.e. an increase in the extent of porosity reduces the mechanical strength and thus reduces the breaking stress. Hence, the failure stresses of the S1, S2 and S3 membranes are 6 MPa, 7 MPa and 11 MPa, respectively.

Fig. 3 SEM images of the three membranes: (a) cross section of S1; (b) cross section of S2; (c) cross section of S3; (d) skin image of S1; (e) skin image of S2; (e) skin image of S3.

Fig. 4 Breaking stresses and porosities of the three membranes.

3.3 Surface charge measurement

The surface charges of the three membranes were determined at neutral pH and it was found that S1 and S2 were nearly neutral (0.0 to 0.1 mV). S3 was slightly negative in charge (−0.03 mV), which can be attributed to presence of nitrile groups.

3.4 Biological assessments of the membranes

3.4.1 Metabolic activity and cell proliferation. Cytocompatibility of dialysis membranes is an important parameter owing to their applications in plasmapheresis, hemodialysis, hemodiafiltration and related blood and body fluid purification. In this context, NIH3T3 cell metabolic activity was evaluated through MTT assays on days three and five and the results are displayed in Fig. 5. The synthesized formazan absorbances of the prepared membranes are higher than the control (0.48 ± 0.01 and 0.80 ± 0.01 on days three and five), demonstrating satisfactory cytocompatibility. Notably, S2 and S3 displayed similar growth kinetics of 0.50 ± 0.03 and 0.56 ± 0.03 on day three and 0.83 ± 0.04 and 0.88 ± 0.04 on day five, while S1 displayed higher cell metabolic activities of 0.67 ± 0.05 and 1.06 ± 0.05 on days three and five, respectively. Therefore, it can be concluded that all three membranes exhibited cytocompatibility and have favourable biological activities.

Fig. 5 Cell metabolic activity of the three membranes.

These results further confirmed by the cell proliferation analysis. The cell proliferation activity of the seeded cells was measured by DNA quantification assay as summarized in Fig. 6. This displayed a similar growth pattern to the MTT assay. The cell proliferation rates of all three membranes were significantly higher than the control (39.2 ± 0.1, 55 ± 0.3 on days three and five, respectively). The absorbance value for the S1 seeded cells was 49.4 ± 0.3 on day three of cell seeding followed by 78.08 ± 0.1 on day five; however, the DNA-bound dye absorbance values for seeded cells on S2 and S3 were 41.16 ± 0.7 and 43.59 ± 0.2 on the third day and 60.46 ± 0.3 and 67.15 ± 0.2 on the fifth day, respectively. Here, similar to the metabolic activity, S1 displayed a higher cell proliferation rate, and subsequently higher cell viability, which is expected due to the fact that PSf–PVP membranes are reported to be biocompatible.³⁸ Interestingly, S2, being a surface modified membrane, also exhibited comparable biocompatibility. This fact is further corroborated in the sections discussed below.

Fig. 6 Cell proliferation assay of the three membranes.

3.4.2 Oxidative stress analysis. Cell cytocompatibility was further verified by measuring oxidative stress on seeded cells in the presence of the membrane samples. The previously described di-chloro di-hydrofuran fluorescein di-acetate (DCFH-DA) assay was used to quantify reactive oxygen species (ROS). During incubation with seeded cells, DCFH-DA becomes oxidized and forms fluorescence active 20,70-dichlorofluorescein (DCF). The concentration of DCF is directly proportional to the newly formed ROS, finally measured as fluorescence intensity as displayed in Fig. 7. The DCF fluorescence intensity was found to increase with the duration of incubation. Interestingly, the control sample displayed the highest ROS activity among all the samples on day three. S2 and S3 initially displayed lower ROS activity but it increased and became similar to the control with increasing duration of incubation. S1 displayed lower activity compared to the control on both days three and five. This can be attributed to the general cytocompatibility exhibited by PSf–PVP blend membranes. However, S2 exhibited comparable results. This may be owing to the phenylenediamine. Phenylenediamine derivatives have well reported antioxidant activity,³⁹ which is primarily due to three mechanisms: (i) free-radical scavenging ability,⁴⁰ (ii) inhibition of oxidative glutamate toxicity and (iii) acting as peroxide decomposers by eliminating oxidative catalyst to avoid further oxidation.⁴¹

Fig. 7 Oxidative stress by the three membranes.

3.4.3 Estimation of total protein content. The protein adsorption study of prepared membranes is prerequisite towards blood dialysis applications. It has been reported that the adsorption of plasma proteins is one of the key issues in evaluating the hemocompatibility of a respective material. Here, protein adsorption was measured through direct (using human plasma proteins such as albumin, γ-globulin and fibrinogen) and indirect (via NIH3T3 cell incubation) methods.

According to earlier studies, proteins are preferentially adsorbed on hydrophobic surfaces. In an aqueous system, the initial surface hydration of a hydrophobic material governs the subsequent protein adsorption.⁹ In this direction, hydrated protein molecules displace the interfacial water through electrostatic interactions to achieve thermodynamic equilibrium.^42,43 Therefore, the prepared sampled are supposed to have good protein adsorption properties owing to their nearly hydrophobic nature. Using the direct method, the protein adsorption behavior of the sample membranes incubated with a predefined human plasma protein solution is displayed in Fig. 8. The control sample displayed very high protein absorption (70.41 ± 0.05). S3, being the most hydrophobic membrane (contact angle 80°) among all the prepared membranes, exhibited high protein adsorption. Similar results were obtained for both S2 (contact angle 73°) and S1 (contact angle 69°), having a less hydrophobic nature. S1 displayed the lowest protein adsorption (13.33 ± 0.03) compared to S2 (21.66 ± 0.04) and S3 (32.91 ± 0.05).

Fig. 8 Total protein adsorption.

Here, it has to be noted that the surface charges, albeit present, are marginal in magnitude. Hence, surface hydrophilicity/hydrophobicity is the dominant factor influencing protein adsorption in such cases.^44,45 Hydrophilicity helps in forming a thin layer of aqueous film on the surface of the membrane, impeding the advance and deposition of proteins on the membrane surface, leading to less adsorption of protein. Hence, S1 and S2 experience lower protein adsorption than S3. Moreover, the zwitter-ionic/mixed-charge hydration phenomenon does not exist here owing to the high contact angle (73°) compared to zwitter-ionic surfaces (<20°). During the indirect protein adsorption study, similar results were obtained. S1 displayed the lowest protein adsorption profile compared to S2 and S3.

3.4.4 Cell attachment and morphology. Cell attachment and proliferation on membranes were evaluated by seeding cells for three and seven days. The cell-seeded membranes were observed by SEM and fluorescence microscopy, as displayed in Fig. 9. The S2 and S3 membranes displayed very little cell attachment initially on day three, and few cells were observed on day seven. As reported in the literature,⁴⁶ surface characteristics, such as surface charge, surface roughness and hydrophilicity, play an important role in improving the membrane performance for hemodialysis applications. In this context, hydrophilicity imparts a thin film of water near the surface of the membrane, thereby decreasing the interactions between proteins and also preventing cell adsorption. The S1 membrane does not show any cell attachment due to a higher hydrophilicity, near neutral surface charge and highly smooth surface. Commercially available fiber was not included as a control in this study due to the difficulty in obtaining active (inner) surface blood contact.

Fig. 9 Cell attachment and morphology study (inset – SEM; main – fluorescence): (a) S1, 3 days, (b) S1, 7 days, (c) S2, 3 days, (d) S2, 7 days, (e) S3, 3 days, (f) S3, 7 days.

3.5 Hemocompatibility tests

3.5.1 Hemolysis assay. To determine hemodialysis applicability, the membranes were incubated with fresh blood to test for RBC hemolysis and the results are reported in Fig. 10. It was observed that up to 2 hours, there was no significant change in hemolysis. All the samples were found to be comparable with the positive control, causing less than 2% hemolysis which suggests blood cell compatibility of the membranes. The hemolytic behaviour is further substantiated by the blood cell aggregation assay, as discussed in the succeeding section.

Fig. 10 Hemolysis assay of the membranes.

3.5.2 Blood cell aggregation. Blood cell aggregation tests were performed to ensure hemocompatibility of the membranes and the results are presented in Fig. 11. There was no significant RBC aggregation observed in any of the membranes. Few WBC were found ruptured in S3, with no significant WBC aggregation behavior being seen in any of the membranes. These further support the hemolysis assay results (discussed in 3.5.1) towards hemocompatibility of the membranes.

Fig. 11 RBC aggregation on the three membranes: (a) control; (b) S1; (c) S2; (d) S3.

It has been reported that antifouling properties, surface charge and smoothness control blood cell compatibility and aggregation activity during hemodialysis.⁴⁶ The promising blood cell compatibility of the S1 membrane is due to the incorporation of PVP, which enhances the hydrophilicity of the membrane, resulting in reduced adsorption of proteins. Further, blending with PSf–PEG causes reduced oxidative stress and limits the surface charge to near neutral, which is also in agreement with reported literature.⁴⁷ Similar reasons are given for the comparable blood cell compatibility of the S2 membrane. Additionally, phenylenediamine reduces the ROS activity, as discussed previously. S3 also has similar hemolytic activity, but the reduced hydrophilicity and negative surface charge lead to cell aggregation.

3.5.3 Platelet adhesion. The protein adsorption, platelet adhesion and activation on the membranes are closely interrelated phenomena. The SEM micrographs of the PRP-incubated membranes are displayed in Fig. 12. It has been observed that the greatest number of platelets adhered and aggregated on the S3 membrane. S2 exhibited comparatively lower platelet adhesion than S3, without any morphological changes and aggregation as found in S3. The platelet adhesion on the S1 membrane was significantly less than S2 and S3.

Fig. 12 Platelet adhesion on the three membranes: (a) S1, (b) S2, (c) S3.

The lower platelet adhesion on S1 is believed to be dependent on the protein adsorption behaviour. According to Ishihara et al. (1999), fibrinogen adsorption is prerequisite for platelet adhesion.⁴⁸ The neutral surface charge results in the repulsion of proteins and negatively charged platelets. Tanaka et al. (2000) observed that besides fibrinogen adsorption, its conformation also needs to be changed for platelet adhesion.⁴⁹ Adsorbed fibrinogen in native conformation never participates in platelet adhesion and activation. In this context, the S2 sample has lower fibrinogen adsorption without significant conformation changes compared to S3, resulting in low platelet adhesion and no aggregation.

3.5.4 Thrombus formation. The thrombus formation properties were evaluated for all three prepared membranes. The outcome is in line with the results obtained in the previous platelet adhesion and protein adsorption tests. Therefore, a significant difference in thrombus formation behaviour was observed between the three membranes. Fig. 13 displays the thrombus formation behavior of the different membranes. S1 and S2 exhibit similar thrombus formation activities. Similar sized samples (1 cm²) of the three membranes were taken. Initial weights of S1, S2 and S3 were 1.68, 1.82 and 2.82 μg respectively. After 1 h, S3 showed highest degree of thrombus (DOT) of 0.63 compared to those of S1 and S2 which are 0.31 and 0.37 (according to eqn (6)).

Fig. 13 Thrombus formation on the three membranes.

3.6 Urea and creatinine permeation

The urea and creatinine transport through the membranes are shown in Fig. 14. It is evident from the figure that S1 has a higher uremic toxin transport rate than S2 or S3. While the time taken for S1 and S2 to bring down the urea concentration in the feed tank to the desired level (400 mg l⁻¹) is around 120 minutes, the same for S3 is 180 minutes. Similarly, creatinine transport requires 120 minutes for S1 to bring down the concentration to 12 mg l⁻¹. However, the rates are much slower for S2 and S3. This can be explained on the basis of the permeability behaviour of the three membranes (Section 3.1). While the permeability of S1 is the highest, that of S3 is lowest. Given that the membranes have the same MWCO, this would automatically imply that the transport rate of the solutes would decrease as permeability decreases.

Fig. 14 Uremic toxin transport through the three membranes: (a) urea transport, (b) creatinine transport.

3.7 Comparison with state of the art literature

Tables 1 and 2 illustrate a detailed comparison between the present work and recently reported literature. It is evident from Table 1 that detailed membrane characterization has been carried out in the present work. Hydraulic permeability, MWCO and surface hydrophilicity have been investigated and the reported values are comparable with the literature data. However, mechanical strength, porosity and surface charge are a few of the additional aspects that have been quantified in the present work. The important interplay of these characteristics with biocompatibility has already been discussed in earlier sections. Table 2 presents a detailed comparison of the present work in terms of cytocompatibility and hemocompatibility with the available literature. Two important aspects are evident from this information. First, exhaustive cytocompatibility and hemocompatibility tests have been carried out in this work totalling 9 parameters as described in Table 2. Second, the developed materials in the present work show competitive results with the reported literature.

Table 1 Comparison of membrane physiological properties with state of art literature

Membrane properties	Present work	State of art literature^15–20,36
Hydraulic permeability (m s^−1 Pa)	0.2–1.4 × 10^−10	3.12 × 10^−11 to 4.16 × 10^−10
Molecular weight cut off (kDa)	6	—
Contact angle	68°–80°	56°–72°
Tensile strength (MPa)	6–11	—
Porosity (%)	53–62	—
Surface charge (mV)	−0.03–0.1	—
Surface morphology (SEM)	Studied	Studied

---

Table 2 Comparison of membrane cytocompatibility and hemocompatibility with state of art literature

Material	Present work	State of art literature
		Li et al.^15	Nie et al.^16	Higuchi et al.^17	Dahe et al.^18	Shakaib et al.^19,20	Lin et al.^36
	PSf–PVP–PEG; surface modified PSf; PAN homopolymer	PES blended with citric acid-grafted polyurethane	Carbon nanotube-grafted PES composite	Chemically modified PSf	PSf–vitamin E–TPGS composite	Polyamide and monosodium glutamate blend	PAN immobilized with chitosan and heparin
Cytocompatibility
Metabolic activity (% better than control)	√√ (37%)	√√ (24%)	√√ (30%)	—	√√ (33%)	—
Cell proliferation	√√	—	—	—	√√	—	—
Oxidative stress	√√	—	—	—	√√	—	—
Total protein adsorption (μg cm^−2)	√√ 13–32	√√ 18–34	√√ 7–15	√√ 3–8	√√ 18–30	—	—
Cell attachment and morphology (SEM and confocal imaging)	√√	—	—	—	√√	—	—
Hemocompatibility
Hemolysis	√√	—	—	—	√√	—	—
Blood cell aggregation	√√	—	—	—	—	—	—
Platelet adhesion	√√	√√	√√	√√	√√	—	—
Thrombus formation	√√	—	—	—	√√	—	—

---

4. Conclusions

(i) Three types of dialysis grade membrane of identical MWCO were synthesized and characterized in detail. The permeability, mechanical strength, contact angle, surface charge, porosity and membrane morphology were analyzed. It was found that the S1 (PSf–PVP–PEG) blend membrane had the highest permeability (1.2 × 10⁻¹⁰ m Pa⁻¹ s), smallest contact angle (69°) and highest porosity (62%). The surface charge of the membrane was found to be near neutral.

(ii) The biological assessment of the three membranes was carried out comprehensively. It was found that S1, S2 and S3 were cytocompatible, displaying promising proliferation and metabolic activity as well as minimal cell adhesion and ROS activity.

(iii) The hemocompatibility performances of the synthesized membranes were comparable. The blood cell aggregation activities were similar in all three membranes. Interestingly, the prepared membranes exhibited very low hemolysis activity compared to the control. Further, a varying trend of platelet adhesion was found among the three membranes owing to their protein adhesion nature. S1 adhered less platelets compared to S2 and S3, which followed in the thrombus formation activity. Overall S1 and S2 have comparable hemocompatibilities and are better than S3 and the commercial membrane.

(iv) The transport of uremic toxins indicated that S1 and S2 facilitated the permeation more than S3. While it took only 120 minutes for S1 and S2 to bring down the urea concentration from 500 to less than 400 mg l⁻¹, the same for S3 was 180 minutes. Similar results were also obtained for creatinine transport.

Acknowledgements

This work is partially supported by a grant from the Department of Science and Technology, Government of India, and Forus under the scheme no. IDP/MED/7/2011, dt. 05.03.2012. Any opinions, findings and conclusions expressed in this paper are those of the authors and do not necessarily reflect the views of the DST.

References

1. The Renal Association, Normal GFR, 2013, http://www.renal.org/information-resources/the-uk-eckd-guide/normal-gfr#sthash.Qoxv8mmG.dpbs.
2. W. J. Kolff and H. T. J. Berk, Acta Med. Scand., 1944, 117, 121–134.
3. N. Alwall, Therapeutic and Diagnostic Problems in Severe Renal Failure, Munksgaard, Copenhagen, 1963.
4. T. Graham, Philos. Trans. R. Soc. London, 1854, 144, 177–228.
5. A. Fick, Ann. Phys. Chem., 1855, 94, 59–86.
6. W. R. Clark, R. J. Hamburger and M. J. Lysaght, Kidney Int., 1999, 56, 2005–2015.
7. T. Kinoshita, Immunol. Today, 1991, 12, 291–294.
8. S. Bowry and T. Rintellen, ASAIO J., 1998, 44, 579–583.
9. M. C. Yang and W. C. Lin, J. Polym. Res., 2002, 9, 201–206.
10. H. G. Hicke, P. Bohme, M. Becker, H. Schulze and M. Ulbricht, J. Appl. Polym. Sci., 1996, 80, 1147–1161.
11. M. Ulbricht and A. Papra, Enzyme Microb. Technol., 1997, 20, 61–68.
12. M. Ulbricht and G. Belfort, J. Membr. Sci., 1996, 111, 193–215.
13. M. Hayama, T. Miyasaki, S. Mochizuki, H. Asahara, K. Tsujioka, F. Kohori, K. Sakai, Y. Jinbo and M. Yoshida, J. Membr. Sci., 2002, 210, 45–53.
14. A. C. Yamashita, Contrib Nephrol, Karger, Basel, 2011, vol. 173, pp. 58–69.
15. L. Li, C. Cheng, T. Xiang, M. Tang, W. Zhao, S. Shun and C. Zhao, J. Membr. Sci., 2012, 405–406, 261–274.
16. C. Nie, L. Ma, Y. Xia, C. He, J. Deng, L. Wang, C. Cheng, S. Sun and C. Zhao, J. Membr. Sci., 2015, 475, 455–468.
17. A. Higuchi, K. Shirano, M. Harashima, B. O. Yoon, M. Hara, M. Hattori and K. Imamura, Biomaterials, 2002, 23, 2659–2666.
18. G. J. Dahe, R. S. Teotia, S. S. Kadam and J. Bellare, Biomaterials, 2011, 32, 352–365.
19. M. Shakaib, I. Ahmed, R. M. Yunus and A. Idris, Int. J. Polym. Mater. Polym. Biomater., 2013, 62, 345–350.
20. M. Shakaib, I. Ahmed, R. M. Yunus and M. Z. Noor, Int. J. Polym. Mater. Polym. Biomater., 2014, 63, 80–85.
21. A. Saito, Contrib Nephrol, Karger, Basel, 2011, 173, pp. 1–10.
22. S. K. Bowry, E. Gatti and J. Vienken, Contrib Nephrol., Karger, Basel, 2011, vol. 173, pp. 110–118.
23. A. M. MacLeod, M. K. Campbell, J. D. Cody, C. Daly, A. Grant, I. Khan, K. S. Rabindranath, L. Vale and S. A. Wallace, Cochrane Database of Systematic Reviews, 2005, issue 3, CD003234.
24. S. K. Bowry, Int. J. Artif. Organs, 2002, 25, 447–460.
25. S. Qiu, L. Wu, L. Zhang, H. Chen and C. Gao, J. Appl. Polym. Sci., 2009, 112, 2066–2072.
26. A. Roy and S. De, J. Food Eng., 2014, 126, 7–16.
27. P. Rai, G. C. Majumdar, S. Dasgupta and S. De, LWT–Food Sci. Technol., 2007, 40, 1765–1773.
28. S. Banerjee and S. De, J. Membr. Sci., 2012, 389, 188–196.
29. J. F. Li, Z. L. Xu, H. Yang, L. Y. Yu and M. Liu, Appl. Surf. Sci., 2009, 255, 4725–4732.
30. S. R. Panda and S. De, J. Polym. Res., 2013, 20, 179–195.
31. S. Chatterjee and S. De, Sep. Purif. Technol., 2014, 125, 223–238.
32. T. Mosmann, J. Immunol. Methods, 1983, 65, 55–63.
33. B. Das, P. Dadhich, P. Pal, P. K. Srivas, K. Bankoti and S. Dhara, J. Mater. Chem. B, 2014, 2(39), 6839–6847.
34. P. K. Smith, R. I. Krohn, G. T. Hermanson, A. K. Mallia, F. H. Gartner, M. D. Provenzano, E. K. Fujimoto, N. M. Goeke, B. J. Olson and D. C. Klenk, Anal. Biochem., 1985, 85, 76–85.
35. F. Pati, P. Datta, B. Adhikari, S. Dhara, K. Ghosh and P. K. Das Mohapatra, J. Biomed. Mater. Res., Part A, 2012, 100(4), 1068–1079.
36. W.-C. Lin, T.-Y. Liu and M.-C. Yang, Biomaterials, 2004, 25, 1947–1957.
37. W. Sun, J. Liu, H. Chu and B. Dong, Membranes, 2013, 3, 226–241.
38. M. Hayama, K. Yamamoto, F. Kohori and K. Sakai, J. Membr. Sci., 2004, 234, 41–49.
39. M. Matsumoto, M. Yamaguchi, Y. Yoshida, M. Senuma, H. Takashima, T. Kawamura and A. Hirose, Food Chem. Toxicol., 2013, 56, 290–296.
40. Chemicalland21, 2012, N,N′-Diphenyl-p-Phenylenediamine.
41. T. Satoh and M. Izumi, Neurosci. Lett., 2007, 418(1), 102–105.
42. M. M. Santore and C. F. Wertz, Langmuir, 2005, 21, 10172–10178.
43. K. Nakanishi, T. Sakiyama and K. Imamura, J. Biosci. Bioeng., 2001, 91(3), 233–244.
44. T. Godjevargova and A. Dimov, J. Membr. Sci., 1992, 67, 283–287.
45. S. Y. Wong, L. Han, K. Timachova, J. Veselinovic and N. Hyder, Biomacromolecules, 2012, 13(3), 719–726.
46. Z.-G. Wang, L.-S. Wan and Z.-K. Xu, J. Membr. Sci., 2007, 304, 8–23.
47. C. R. Hasler, G. R. Owen, W. Brunner and W. H. Reinhart, Nephrol., Dial., Transplant., 1998, 13, 3132–3137.
48. K. Ishihara, K. Fukumoto, Y. Iwasaki and N. Nakabayashi, Biomaterials, 1999, 20, 1553–1559.
49. M. Tanaka, T. Motomura, M. Kawada, T. Anzai, K. Yuu, T. Shiroya, K. Shimura, M. Onishi and A. Mochizuki, Biomaterials, 2000, 21, 1471–1481.

---