Hemocompatible poly(lactic acid) membranes prepared by immobilizing carboxylated graphene oxide via mussel-inspired method for hemodialysis

State Key Laboratory of Separation Memb Polytechnic University, Tianjin 300387, Chi School of Textiles, Tianjin Polytechnic Univ School of Materials Science and Engineerin 300387, China. E-mail: zhangyz2004cn@vip † Electronic supplementary information procedures for preparation of the origina reduced graphene oxide (rGO), carboxy coating on the PLA membrane surface, and carboxylic graphene oxide (GO–C graphene oxide, XPS and TEM analysis o of rGO, SEM analysis, water contact an plasma recalcication time analysis, h modied PLA membranes and the stabili See DOI: 10.1039/c7ra11091j Cite this: RSC Adv., 2018, 8, 153


Introduction
Biodegradable polymers have attracted much attention due to environment and sustainability issues. 1 As a promising biomedical material, poly(lactic acid) (PLA) is biodegradable, bioabsorbable, renewable and extensively investigated. [1][2][3] However, PLA is relatively hydrophobic and lacks reactive sidechain groups 1 when it is used as a human blood-contacting material. Li et al. improved the hemocompatibility of polypropylene by attaching silver nanoparticles capped with TPGS. 4 Jiang et al., 5 You et al. 6 and Gao et al. 7 graed heparin onto PE porous membrane surface, PU substrates and PLA membrane surface to develop their hemocompatibility, respectively. Chang et al. graed zwitterionic poly(sulfobetaine methacrylate) (PSBMA) onto PVDF membrane and PP brous membrane via plasma-induced surface copolymerization to improve the materials' blood compatibility. 8,9 Zhu et al. graed PSBMA to PLA membrane surface by atom transfer radical polymerization (ATRP) to improve the fouling resistance and hemocompatibility. 2 Xiong et al. improved the blood compatibility of PLA membrane by bonding heparin via surface cross-linking and glycidyl ether reaction. 10 Li et al. immobilized natural hirudin on PLA membrane to develop a hemocompatible PLA membrane through the hydrogen bonding interaction. 11 However, plasma treatment and ATRP are somewhat complex and may be not suitable for large-scale surface modication. 12 In recent years, graphene oxide (GO), which contains carboxyl, hydroxy and epoxy functional groups 13,14 has attracted many researchers' attention. Much research has been performed to investigate the biocompatibility and toxicity of GO in vivo and in vitro when it is used for biotechnological application. [14][15][16][17] Cheng et al. prepared a biopolymer adhered GO or rGO by a facile biomimetic method and the biopolymer exhibited controllable 2D morphology and excellent biocompatibility. 18 Hu et al. 19 and Chae et al. 20 suggested that GO performed good bio-antifouling property on account of its hydrophilicity, negative charge, and surface smoothness.
Huang et al. modied graphene oxide with aptamerconjugated gold nanoparticles and heparin to obtain a targeted anticoagulant. 21 He et al. fabricated a hemocompatible and antibacterial polymeric membrane by coating a top layer of graphene oxide and a sulfonated polyanion co-doped hydrogel thin lm. 22 Pinto et al. investigated the biocompatibility of PLA with blending graphene oxide and graphene nanoplatelets respectively. 23 However, to better understand the hemocompatibility of PLA membrane surface immobilizing GO, further research is urgently needed.
Dopamine (DA), which can be easily deposited on inorganic and organic substrates, has been widely investigated. [24][25][26][27][28] Furthermore, DA can be graed onto various materials, and the as-prepared materials exhibit excellent adhesive ability to solid surfaces. 12,18,[29][30][31] In the present work, DA molecules were grafted onto carboxylic graphene oxide (GOCOOH) nanoplatelets to obtain mussel inspired adhesive nanoplatelets (DA-g-GOCOOH) by carbodiimide chemistry method. Subsequently, DA-g-GOCOOH were adhered onto the PLA membrane surface via DA-g-GOCOOH's adhesive ability in tris(hydroxymethyl) aminomethane hydrochloride buffer (Tris-buffer, pH 8.5). The surface chemistry, morphology, charge, water contact angle and the hemocompatibility (platelet adhesion, clotting time and hemolysis ratio) of the PLA membranes aer modication were studied. Furthermore, the dialysis performances of the PLA/(DA-g-GOCOOH) membranes were investigated. The procedure of preparing PLA/(DA-g-GOCOOH) membrane, blood anticoagulation and hemodialysis was shown in Scheme 1. Herein, the present work aims to explore a novel method to improve the hemocompatibility of PLA membranes for hemodialysis.

Preparation DA-g-GOCOOH nanosheets
Briey, 0.5 g GOCOOH (dispersing in 75 mL H 2 O) was sonicated for 30 min, and then, EDC (0.1917 g), NHS (0.0575 g) and 125 mg DA were added sequentially. Stir the mixture vigorously under N 2 protection (to prevent the self-polymerization of dopamine) at 30 C. The resultant DA-g-GOCOOH were washed with DI water aer centrifuging for several times to remove impurities and kept in N 2 protection. For comparison, DA-g-GO was also prepared by the same method.
2.2 DA-g-GOCOOH coating on the PLA membrane surface and characterization of the DA-g-GOCOOH coating layer Briey, the prepared PLA membranes were immersed in DA-g-GOCOOH solution with different concentrations (0.5 mg mL À1 , 1.0 mg mL À1 , 1.5 mg mL À1 , 2.0 mg mL À1 ) dissolved in tris(hydroxymethyl)aminomethane hydrochloride buffer (10 mM, pH 8.5, containing 5 mM, Fe(SO 4 ) 2 (NH 4 ) 2 ) for 24 h. The DA-g-GOCOOH was adhered onto the PLA membrane due to its excellent adhesive ability. Then, the membranes coated with DA-g-GOCOOH were washed with deionized water to remove unrm DA-g-GOCOOH. The resultant DA-g-GOCOOH coated PLA membranes were marked by CG-0.5, CG-1.0, CG-1.5, CG-2.0. For comparison, DA-g-GO immobilized on the PLA membrane surface was also obtained by the same method. PLA membrane was immersed in 2.0 mg mL À1 DA solution (Trisbuffer, 10 mM, pH 8.5) for 24 h to obtain PLA/PDA membrane.
Scanning electron microscopy (SEM) images were acquired by utilizing a scanning electron microscope (SEM, Hitachi S4800, Japan). Attenuated total reectance Fourier transform infrared spectra (ATR-FTIR, Thermo-Nicolet iS50, America) and X-ray photoelectron spectroscopy (XPS, K-Alpha, ThermoFisher, America) were utilized to analyse the surface chemical composition of the modied PLA membranes. The water contact angle was obtained by using SL200KB (KINO, America). A SurPASS electrokinetic analyser (Anton Paar GmbH, Austria) was utilized to study the zeta potentials of the membranes (1 mmol L À1 KCl solution, 25 C).

Blood compatibility assays
2.3.1 Platelet adhesion. The anticoagulant sheep whole blood was purchased from Beijing Pingrui Biotechnology Company, China. All the blood experiments were performed in accordance with the guidelines and regulations for the care and use of laboratory animals of the National Institutes of Health. All procedures were approved by the Institutional Animal Care and Use Committee of Tianjin Polytechnic University (Tianjin, China). 5 mL anticoagulant sheep whole blood was centrifuged at 200 g for 10 min. The supernatant was platelet-rich plasma (PRP). 100 mL of the fresh PRP was added to each well containing one piece of membrane sample (1 cm Â 1 cm) washed with PBS buffer solution (pH 7.4) and then maintained at 37 C for 1 h. Aer rinsing in PBS twice gently, the membrane samples were processed with glutaraldehyde (2.5 wt%) to fasten the platelets adhesion. Then the samples were gradually dehydrated with changing weight percentage of ethanol/water solutions (10, 30, 50, 70, 90, and 100 wt%). SEM was applied to investigate the adhesion of platelets on membranes and the average number of platelets adhered on samples was calculated.
2.3.2 Plasma recalcication time. 5 mL anticoagulant sheep whole blood was centrifuged at 3000 rpm for 15 min. The platelet-poor plasma (PPP) was obtained by taking out the surpernatant. 100 mL PPP was pipetted onto the surface of each sample in a 24-well cell culture plate and incubation (37 C) was performed for 1 min. Subsequently, 100 mL CaCl 2 (0.025 M) aqueous solution was added and the addition time was accurately recorded. Stirred the mixture gently with a stainless steel hook to observe the appearance of the rst brin thread. Then the plasma recalcication time (PRT) was recorded.

Hemolysis ratio (HR).
The membranes (1 cm Â 1 cm) were preprocessed in 0.9 wt% NaCl at 37 C for 40 min. Then, 200 mL sheep whole blood was added, incubated at 37 C for 1 h. The mixture was centrifuged at 800g for 10 min and the absorbance of the supernatant was obtained by UV/visible spectrophotometer (Ultrospec 2100 pro, GE, America) at 545 nm. 0.9 wt% NaCl solution was selected as a negative control and DI water was selected as a positive control. The HR was calculated by the following equation: where AS was the absorption value of samples, AN was the absorption value of negative control (saline), and AP was the absorption value of positive control (water).

Hemodialysis evaluation tests
Urea (1.5 g L À1 ), lysozyme (0.04 g L À1 ) and bovine serum albumin (BSA, 1.0 g L À1 ) in physiological saline (0.9 wt%) were used as the mimic blood. The ows of the mimic blood and dialysate (pure water) were 100 mL min À1 and 300 mL min À1 , respectively. As the UV adsorption wavelength of BSA was similar to lysozyme, the retention to BSA was measured separately. Each sample was tested for 6 h (effective membrane area was 7.0 cm 2 ). 2.4.1 Water ux. The water ux was measured at 25 C under 0.1 MPa. Pure water was ltrated through a membrane sample and the liquid ux was named as J, which was calculated by where J (L m À2 h À1 ) was the ux, V (L) was the volume of permeated solution, S (7.0 cm 2 ) was the effective ltration area, t (h) was the operation time.
2.4.2 Urea clearance. 1.5 mL mimic blood sample was mixed with 0.8 mL PDAB solution (0.1 M) and 1.7 mL pure water and incubated at 37 C for 10 min. The UV/vis spectrophotometer (Ultrospec 2100 pro, GE, America) (at a wavelength of 410 nm) was used to measure the concentration of urea. Then the urea clearance (clearance urea ) was calculated as follows: where C original urea was the original urea concentration before dialysis in mimic blood and C nal urea was the nal urea concentration aer dialysis in mimic blood.
2.4.3 Lysozyme clearance. A UV/vis spectrophotometer was applied to determine the lysozyme concentration (at a wavelength of 278 nm). The lysozyme clearance (clearance lysozyme ) was calculated as follows: where C original lysozyme was the original lysozyme concentration before dialysis in mimic blood and C nal lysozyme was the concentration aer membrane dialysis in mimic blood.
2.4.4 BSA retention. A UV/vis spectrophotometer was used to obtain the BSA concentration in the mimic blood (at a wavelength of 280 nm). The BSA retention (retention BSA ) was obtained as follows: where C original BSA and C nal BSA were the concentrations in the mimic blood before and aer membrane dialysis.

Preparation and characterization of DA-g-GOCOOH nanosheets
The chemical compositions and morphology of GO-COOH were analysed by FTIR spectroscopy (Fig. S1 †) and AFM (Fig. S2 †). As shown in Fig. S1, † the peaks at 1730, 1640 and 1100 cm À1 in the FT-IR spectrum were characteristic of the C]O and C-O stretches of the ketone groups and epoxy groups, respectively, on GO. Compared with GO, the FT-IR spectrum of GO-COOH showed a stronger -COOH peak and a bigger peak width, while the -OH peak became weaker and a smaller peak width. This implied that partial epoxy and hydroxyl were activated into carboxyl. As shown in Fig. S2 † and 2 the thickness of GO-COOH nanosheet was about 1.66 nm and a 2D surface morphology was observed. The results indicated the GO-COOH nanosheets successfully prepared.
The chemical compositions of the DA-g-GOCOOH nanosheets were analysed by FTIR (Fig. 1A) and XPS (Fig. 1B). As shown in Fig. 1A, the peaks at 1730, 1630 and 1380 cm À1 appeared in the FTIR spectra of GOCOOH and DA-g-GOCOOH were related to the stretching vibration of C]O in carboxy group, asymmetric deformation of COO À in carboxy group, and C-OH blending vibration in carboxy group respectively. Compared with GOCOOH, a new weak peak at 1280 cm À1 (C-N stretching vibrations) appeared in spectra of DA-g-GOCOOH.
The peak of N-H bending vibrations (1610 cm À1 ) was overlapped with COO À asymmetric deformation. Therefore, the DAg-GOCOOH component was further characterized by XPS. The XPS and the molar percentages of element were shown in Fig. 1B and Table 1. Compared with GOCOOH, the new peak of nitrogen (N 1s) at 400 eV attributed to the C-N group could be observed and the molar percentage of N 1s increased from 0 to 1.82%. Both FTIR and XPS results conrmed that the dopamine molecules were successfully graed onto GOCOOH nanosheets.
TEM was also employed to detect the morphologies of DA-g-GOCOOH nanosheets. As shown in Fig. 2, the wrinkle-like thin sheets, which were the characteristic structure of GO nanosheet, were observed in both GOCOOH and DA-g-GOCOOH. The TEM image of DA-g-GOCOOH exhibited a 2D surface morphology. The FTIR spectra, XPS spectra and TEM image  indicated that the DA-g-GOCOOH nanosheets were successfully prepared. DA-g-GO nanosheets were shown in Fig. S3. † The molar percentage of N 1s increased from 0 to 2.02% (Table S2 †) and DA-g-GO exhibited a 2D surface morphology. The photographs of rGO suspension with different microwave treatment time (0, 20, 40, 60 min) and XPS wide scans of C 1s in rGO were shown in Fig. S4. † And the -COOH group contents of DA-g-GOCOOH, DA-g-GO and rGO were shown in Table S1. † Compared with DA-g-GO (2.430 mM g À1 ), the -COOH group density was 5.470 mM g À1 in DA-g-GOCOOH. This was related to the carboxylation of GO. However, the -COOH group density of rGO with 60 min reduction treatment was 0.310 mM g À1 . The -COOH density had closer relationship with the hydrophilicity and electronegativity of the membranes.

Characterization of the modied membranes
The surface chemical components of the membranes were detected by XPS. The XPS spectra and the molar percentages of element were shown in Fig. 3 and Table 2. There was no nitrogen element in original PLA membrane just as shown in Fig. 3, whereas, the peak of nitrogen (N 1s, molar ratio 3.46%) at 400 eV attributed to the C-N group, was observed on DA-g-GOCOOH-immobilized membrane.
The morphologies of membrane surface were observed by SEM. As shown in Fig. 4, the surface of pristine PLA membrane was relatively smoother and the pores could be observed. The DA-g-GOCOOH adhered onto the PLA membranes surface due to the polymerization of dopamine molecules in alkaline solution. The pore size and pore number of the DA-g-GOCOOH modied PLA membranes remarkably decreased. It was related to the immobilizing of DA-g-GOCOOH layer. And the wrinklelike structures on membrane surface became more and more apparent with the increasing immersing concentration of the DA-g-GOCOOH. The DA-g-GOCOOH nanosheets were distinctly observed when the coating concentration of DA-g-GOCOOH was 2.0 mg mL À1 (Fig. 4, CG-2.0). The same situation occurred for DA-g-GO coated PLA membranes (Fig. S5 †). The SEM image for rGO coated PLA membranes were shown in Fig. S5. † The rGO nanosheets with different microwave treatment time could be obviously observed the PLA membranes.
The hydrophilic-hydrophobic properties of membrane surfaces were detected by water contact angle. As shown in Fig. 5A, the contact angle of pure PLA membrane was as high as 77.4 . The contact angle decreased to 66.5 of the polydopamine coating PLA membrane (PLA/PDA). It was attributed to the hydrophilic groups (-OH and -NH 2 ) in polydopamine. 26 Compared with the native PLA membrane and the PLA/PDA membrane, the DA-g-GOCOOH modied PLA membranes showed remarkably improvement of hydrophilicity, and the water contact angle decreased from 60.3 to 36.2 with the increasing immersion concentration of DA-g-GOCOOH (from 0.5 mg mL À1 to 2.0 mg mL À1 ). A hydration layer was able to form by the electrostatic interaction and hydrogen bonds. 13,32 It was related to the abundant carboxy group of DA-g-GOCOOH nanosheets.
The membrane surface charge was another key factor to inuence the hemocompatibility of the membrane. The zeta   This journal is © The Royal Society of Chemistry 2018 potential (z) of the membranes was investigated (1 mM KCl solution at 25 C). And as shown in Fig. 5B, the zeta potential of the pure PLA membrane was À26.6 mV at pH 7.4 (simulating the vivo environment). The PLA/PDA membrane (À25.0 mV) exhibited less electronegative than the native PLA membrane, which might be induced by the amine groups of polydopamine layer. For the DA-g-GOCOOH immobilized PLA membranes (CG-0.5, CG-1.0, CG-2.0), the zeta potential notably decreased from À47.7 mV to À59.9 mV. However, the rGO and DA-g-GO modied PLA membrane exhibited less hydrophilic or electronegative (Fig. S6 †). It might be related to more -COOH group at DA-g-GOCOOH. The zeta potential demonstrated that the surfaces of DA-g-GOCOOH modied PLA membrane become more negative than that of the original PLA membrane, which was attributed to the more negative groups (-COOH and -OH) in the immobilized layer of DA-g-GOCOOH. The adsorption of negative platelets was restrained by the electrostatic repulsion for the DA-g-GOCOOH modied PLA membranes. In brief, both the surface hydration and electrostatic was able to benet the hemocompatibility of PLA membranes by immobilizing DA-g-GOCOOH.

Hemocompatibility
Blood coagulation was a vital factor in hemocompatibility. In the present work, the anticoagulation properties of the original PLA membrane and the DA-g-GOCOOH modied PLA membranes were evaluated by the platelet adhesion and plasma recalcication time. Platelets adhesion on the original and modied PLA membranes was investigated by SEM. And the results were shown in Fig. 6. The average number of adhered platelets on each sample was calculated based on SEM images. It was found that remarkable aggregation of the platelets on the native PLA and polydopamine coated membranes occurred. Besides, the platelets spread in irregular shapes on the surface of the PLA and PLA/PDA membranes were observed. This indicated that the platelets were deformed and activated. More platelets adhered on the PLA/PDA membrane surface (about 2.789 Â 10 6 cell per cm 2 ) than that of the native PLA membrane (about 2.324 Â 10 6 cell per cm 2 ). It might be induced by the less surface charge of PLA/PDA membrane than the original PLA membrane. For the DA-g-GOCOOH immobilized PLA membranes, the number of adhered platelets notably decreased from 1.046 Â 10 6 cell per cm 2 (CG-0.5) to 0.116 Â 10 6 cell per cm 2 (CG-2.0). Rarely platelets were observed and the platelets expressed with nearly no deformation on CG-2.0 membrane as shown in Fig. 6. It indicated that immobilizing DA-g-GOCOOH nanosheets on PLA membranes was able to obviously suppress the activation and transmutation of platelets. The hemocompatibility of PLA membranes efficiently improved with the hydrophilic and electronegative membrane surfaces modied by DA-g-GOCOOH.
The blood-clotting of PLA membranes before and aer modication was investigated by the plasma recalcication time (PRT). As shown in Fig. 7A, the PRT of the DA-g-GOCOOH immobilized PLA membranes was prolonged to 220 s (Fig. 7, CG-2.0) comparing with original PLA membrane (120 s). The increased PRT indicated that the plasma coagulation on DA-g-GOCOOH immobilized PLA membranes was restrained, which was related to the improvement of the hydrophilicity and electronegativity of the DA-g-GOCOOH modied membranes.
Hemolysis ratio (HR) was used to investigate the damage caused by materials on erythrocyte. And the results were shown in Fig. 7B. It was found that the HR value of the original PLA and PLA/PDA membrane was 10.7% and 5.9%, respectively, which were out of safety level for biomaterials. Compared with DA-g-GOCOOH modied membranes, the PLA/PDA membrane is less hydrophilic and electronegative, leading a higher hemolysis ration. The HR value of DA-g-GOCOOH immobilized membranes notably reduced to less than 0.3% (CG-0.5: 0.23%; CG-1.0: 0.2%; CG-2.0: 0.13%, respectively). This was related to the enhancement of hydrophilicity and electronegativity of the modied membranes. All results indicated that coating DA-g-GOCOOH on PLA membrane surface not only prevented platelets adhesion and blood coagulation but also weakened the damage to erythrocyte.
The SEM images of platelets adhesion, recalcication time and hemolysis ratio data of rGO and DA-g-GO modied membrane surfaces were shown in Fig. S7 and S8. † More platelets adhered on the surface, shorter recalcication time and higher hemolysis ratio of rGO and DA-g-GO modied PLA membranes than those of DA-g-GOCOOH modied PLA membranes. It was attributed to the less hydrophilicity and    electronegativity than that of DA-g-GOCOOH modied membrane.
The stability of the DA-g-GOCOOH coating layers was also investigated. The UV-vis absorbance of DA-g-GOCOOH for PLA and modied membranes aer incubating different time in water was shown in membrane. This phenomenon might be attributed to deterioration of the water. As shown in Fig. S10, † the PLA/(DA-g-GOCOOH) membranes still restrained the platelet adhesion on membrane surface compared with the original PLA membrane aer incubating in water for 48 h. It indicated that the DA-g-GOCOOH coating layers were relatively stable.

Dialysis performance
The pure water uxes of PLA and modied membranes were evaluated, and the results were shown in Fig. 8A. The water ux of the original PLA membrane was 125.3 L m À2 h À1 , and the water ux of modied membranes decreased from 59.2 to 32.6 L m À2 h À1 with the increment of DA-g-GOCOOH immersing concentration. It indicated that the immobilized DA-g-GOCOOH layer was able to decrease the pore size of the membranes leading to the increment of water permeation resistance.
The urea clearance (clearance urea ), lysozyme clearance (clearance lysozyme ) and BSA retention (retention BSA ) of the membranes were utilized to evaluate the hemodialysis performances of the membranes. As shown in Fig. 8B, for the pure PLA membranes, clearance urea , clearance lysozyme and retention BSA were 82%, 44.3% and 76.4%, respectively. And clearance urea and clearance lysozyme of the PLA/(DA-g-GOCOOH) membranes decreased with the increment of DA-g-GOCOOH immersing concentration, while the retention BSA increased. The clearance urea of the DA-g-GOCOOH immobilized membranes decreased from 75% to 65%, and the clearance lysozyme decreased from 33% to 24.5%, while the retention BSA increased from 91.2% to 95.6% when immersing concentration of DA-g-GOCOOH increased from 0.5 mg mL À1 to 2.0 mg mL À1 . Both the pore size and the charge of the membrane surface affected the dialysis performances. The molecular weight of urea, lysozyme and BSA is 60.06 Da, 14 kDa and 66.4 kDa, respectively. The pore size of DA-g-GOCOOH coated PLA membranes decreased signicantly comparing with that of the pure PLA membrane, and thus the clearance of urea and lysozyme decreased while the retention of BSA increased. Besides, the electronegativity of the modied PLA membranes increased with the DA-g-GOCOOH immersing concentration (see Fig. 6) and therefore the retention of BSA increased. In brief, the dialysis performances of the DA-g-GOCOOH modied PLA membranes with negative charge were inuenced by the pore size and space charge effect. The performance of hemocompatibility and dialysis in this work compared with references was shown in Table 3.

Conclusions
The fabrication of hemocompatible PLA hemodialysis membranes by a facile mussel-inspired method was developed in this work. A stable DA-g-GOCOOH layer was formed via immobilizing DA-g-GOCOOH nanosheets onto PLA membranes. The obtained PLA/(DA-g-GOCOOH) membranes exhibited good hydrophilicity and enhanced electronegativity. Moreover, the hemocompatibility of the modied PLA membranes was improved involving less platelets adhesion, longer plasma recalcication time and lower hemolysis ratio. In addition, the DA-g-GOCOOH immobilized PLA membrane cleared almost 65% urea, 24.5% lysozyme and retained 95.6% BSA. Immobilizing DA-g-GOCOOH on the surface of membranes provided a promising avenue to improve the hemocompatibility of hemodialysis membranes.

Conflicts of interest
There are no conicts to declare.