A porous hybrid imprinted membrane for selectively anchoring target proteins from a complex matrix

Abstract

A novel porous hybrid imprinted membrane (CP/CNT/DA-MIM) was prepared that could selectively anchor and separate target proteins from a complex matrix. CP/CNT/DA-MIM exhibits many of the advantages of molecularly imprinted polymers and membranes, including the high selectivity of MIPs, the lower energy consumption and the ability to continuously separate mixtures via membrane separation. The surface morphologies and physical/chemical properties of the different membranes were investigated using FTIR, XRD, DSC, XPS and SEM. The results showed that the different molecules contained within CP/CNT/DA-MIM were homogeneous; two different sizes of imprinted cavities were observed in CP/CNT/DA-MIM, which facilitate the selective anchoring property. The adsorption capacities, swelling behaviors and mechanical properties of the different constituent membranes were also compared. The results show that the adhesion and nonspecific adsorption properties of the membrane were manifestly reduced through the addition of PVP. The binding capacity and adsorption selectivity of the membrane were apparently improved because of the presence of dopamine. MWCNTs obviously improved the mechanical strength of the membranes. CP/CNT/DA-MIM was successfully applied to separate bovine serum albumin from bovine blood. CP/CNT/DA-MIM is an economical, hydrophilic and ecofriendly membrane and is a promising separation material for the large-scale continuous selective separation of target proteins from complex matrices in commercial applications.

---

1. Introduction

To further study the structure and function of living organisms at the molecular level, research into proteomics and genomics has recently been widely developed.¹ With the development of proteomics research, it is essential to selectively recognize and separate target proteins from complex matrices for industrial protein purification, basic biomedical research and clinical diagnostics.² However, the detection and separation of proteins are complicated processes because of their complicated three dimensional structures, labilities and denaturation tendencies.³ Proteins that have significant biological functions are often low abundance proteins, which increases the difficulty of detection and separation.⁴ Although antibodies have been the most widely used tools for capturing target proteins, the generation of antibodies is time-consuming and expensive, and many proteins are difficult to raise antibodies against.⁵ Additionally, antibodies are produced by living cells, and it is occasionally difficult to control their quality.⁶ Therefore, an ideal protein-capture agent should not only have high specificity but also be an economical, stable, robust, available and non-biological material. In this respect, molecularly imprinted polymers (MIPs), also called “antibody mimics”, are considered to be promising and facile separation materials. MIPs have wonderful application potential in the fields of amino acid, protein and nucleotide derivative recognition.⁷

MIPs are formed by the polymerization of a desired template and functional monomer with cross-linkers. After removing the template, a customized three-dimensional nanocavity is formed with complementary binding sites for the target.⁸ MIPs can act as artificial antibodies and exhibit high selectivity towards the imprinted molecules.⁹ The significant advantages of MIPs include their easy preparation, predictable specific recognition, low cost, high chemical/physical stability, robustness and reusability, which enable MIPs to be applied in areas such as separation,¹⁰ drug delivery,¹¹ chemical sensing,¹² catalysis,¹³ phytoextraction,¹⁴ biological analysis,¹⁵ and so on. Their most extensively investigated application has been as separation materials for the analysis of various compounds, including drugs,¹⁶ pesticides,¹⁷ trace elements,¹⁸ proteins⁶ and nucleotides.¹⁹ In particular, molecularly imprinted membrane (MIM) separation technology has been widely used in the fields of separation and purification, which benefit from the advantages of membranes and MIPs, including the high selectivity of MIPs, their lower energy consumption and their ability to continuously separate membrane separation mixtures.²⁰ These characteristics are helpful for large-scale continuous separation operations, especially in industrial applications. Therefore, it is believed that MIM is one of the most promising materials in the separation and purification fields.

To date, molecular imprinting has been proven to be particularly successful for low-molecular-weight compounds.^12–14 Although several imprinting techniques for creating protein MIPs have been developed,^1,2 protein imprinting still faces challenges due to the inherent properties of proteins, including their complicated physicochemical properties, large sizes, poor solubilities, sensitivities to chemical processes and their limited viabilities in organic solvents, which are usually essential for MIP formation.²¹

Chitosan (CS), which contains abundant amino and hydroxyl groups, has remarkable affinity for proteins.²² CS is a linear hydrophilic polysaccharide derived from the deacetylation of chitin, which is one of the most abundant natural biopolymers. CS has a relatively good film-forming ability, it is eco-friendly and safe for humans and the environment.²³ Furthermore, it is a biocompatible and biodegradable material with attractive properties, including hydrophilicity, biocompatibility, nontoxicity and high permeability toward water.²⁴ These endow CS with a wide usage in molecular separation, wastewater treatment, packaging materials, artificial skin, cosmetics, bone substitutes, and so on.²⁵ Moreover, CS has been widely used as a supporter or functional monomer for the recognition and immobilization of proteins.^26,27 However, the specific adsorption of CS for target proteins is unsatisfactory because of the abundant functional groups in CS.²² Additionally, the fragility, uncontrollable porosity and adhesive property of chitosan matrices limit their feasibility and practicability in membrane separation.²⁵ To overcome its disadvantages, CS is typically mixed with other compounds, such as polytetrafluoroethylene,²⁸ polyvinyl alcohol,²⁹ polycaprolactone,³⁰ polyvinyl pyrrolidone (PVP),³¹ agarose³² and carbon nanotubes (CNTs).²⁵ CS and PVP can form a homogeneous phase due to the strong hydrogen bonding forces between these two types of molecules.³³ PVP is also biocompatible, non-toxic and easily fabricated into a membrane; it possesses relatively good lubricity and anti-adhesive properties.³⁴ Additionally, PVP has been shown to control the porosity, reduce the adhesion and improve the elasticity of chitosan structures.³⁵ In particular, the addition of CNTs into a chitosan matrix can significantly increase the tensile modulus and strength.²⁵

In this work, an imprinted hybrid membrane is examined that provides the merits of selectivity, economy, stableness and hydrophilicity for the anchoring of target proteins from biological samples (such as blood samples). The membrane was prepared using a mixture of chitosan and polyvinyl pyrrolidone as the supporter, dopamine and chitosan as the bi-functional monomers, polyethylene glycol as the porogen and bovine serum albumin (BSA) as the model template. To further improve the physical/chemical properties of the membrane, functionalized MWCNTs with multiple surface hydroxyl and carboxyl groups were added into the membranes. The mechanical properties and swelling behaviors of the different membranes were compared. The surface morphologies and physical/chemical properties of the different membranes were also investigated using Fourier transform infrared spectroscopy, X-ray diffraction, differential scanning calorimetry, X-ray photoelectron spectroscopy and a scanning electron microscope. The adsorption capacities and imprinted factors of the different membranes were also studied. Then, the chitosan/polyvinyl pyrrolidone/multi-walled carbon nanotube/dopamine molecularly imprinted membrane (CP/CNT/DA-MIM) was applied as a separation matrix to anchor BSA from bovine blood.

2. Experimental method

2.1 Materials and reagents

CS with a 90% degree of deacetylation (MW = 600 000 g mol⁻¹), PVP (MW = 1 300 000 g mol⁻¹) and polyethylene glycol (PEG, MW = 20 000 g mol⁻¹) were purchased from Sinopharm Chemical Reagent Co., Ltd. MWCNTs were purchased from Beijing DK Nano Technology Co., Ltd. Their diameters and lengths were approximately 20–30 nm and 10–20 μm, respectively. MWCNTs were purified by thermal oxidation (at 500 °C for 30 min in air) and acid treatment (refluxed in concentrated sulfuric acid and a nitric acid solution (1 : 3, v/v) for 12 h and washed with water) before use. Dopamine hydrochloride was obtained from J&K Scientific, Ltd. (China). Bovine serum albumin (BSA), human serum albumin (HSA) and lysozyme (Lyz) were obtained from Sigma-Aldrich. Glutaraldehyde (25% aqueous solution), absolute acetic acid (99%), sodium chloride (99.5%), sodium borohydride (98%) and other chemicals were purchased from Sinopharm Chemical Reagent Co., Ltd. Ultrapure water purified with a Milli-Q system (Millipore) was used in all of the experiments. The bovine blood sample was collected from the local slaughter house.

2.2 Preparation of hybrid membranes

The CP/CNT/DA-MIM was prepared by a phase inversion method. The proposed route of polymerization is demonstrated in Fig. 1. Five milligrams of functional MWCNTs was dissolved in 20 mL of water and sonicated for 3 h to obtain a uniform dispersion of the MWCNT solution. Then, 0.4 mL of acetic acid, 0.25 g of chitosan powder, 0.25 g of PVP, 0.1 g of PEG and 0.2 mL of glutaraldehyde (25%) were added one by one and consecutively stirred for 2 h. The mixer was labeled as A. BSA (0.75 μmol) was dissolved in 5 mL of a PBS solution (5 mM, pH = 7.0); then, 60 μmol of DA was added with consecutive stirring for the pre-assembly for roughly 1 h, generating mixture B. Solutions A and B were mixed together and 0.2 mL of sodium borohydride (0.2 mol L⁻¹) was added. The mixture was stirred continuously for 2 h. The solution was cast on a clean glass plate (9 × 12 cm) and dried at 30 °C. Then, the dried sample was immersed into 4% sodium hydroxide for 3 h to neutralize the residual acetic acid, and washed with distilled water. The membrane was then boiled for 1 h in 300 mL of water to extract the porogen. It was washed with 100 mL of 5 mM phosphate buffer that contained 1 M NaCl (pH = 7.0) to remove the embedded template protein until no BSA was detected in the supernatant (using a RF-5301PC fluorescence spectrophotometer (SHIMADZU, Japan)). The membrane was subsequently rinsed with distilled water. After drying in air, CP/CNT/DA-MIM was stored in a desiccator. The proposed presentation and the anchoring protocol of CP/CNT/DA-MIM are shown in Fig. 2.

Fig. 1 The proposed route of polymerization of CP/CNT/DA-MIM.

Fig. 2 The proposed presentation and the recognition protocol of CP/CNT/DA-MIM.

The chitosan/polyvinylpyrrolidone/MWCNTs/dopamine-non-molecularly imprinted membrane (CP/CNT/DA-NIM) was fabricated identically to the CP/CNT/DA-MIM membrane but without the template proteins. A chitosan/polyvinylpyrrolidone-molecularly imprinted membrane (CP-MIM) and a chitosan/polyvinylpyrrolidone/dopamine-molecularly imprinted membrane (CP/DA-MIM) were also prepared for comparison. The preparation protocols of the membranes are shown in Table 1.

Table 1 Optimization of the preparation conditions for membranes and their swelling degrees, adsorption capacities and imprinted factors

Membrane	MWCNTs (%)	CS : PVP	BSA (μmol)	DA (μmol)	Swelling degree (%)	Q (μmol cm^−3)	IF
CP-MIM1	—	5 : 1	0.75	—	256.8 ± 4.1	0.590 ± 0.018	1.42 ± 0.07
CP-NIM1	—	5 : 1	—	—	255.3 ± 5.6	0.415 ± 0.034	—
CP-MIM2	—	5 : 3	0.75	—	225.0 ± 3.9	0.493 ± 0.031	1.73 ± 0.06
CP-NIM2	—	5 : 3	—	—	227.9 ± 7.2	0.285 ± 0.017	—
CP-MIM3	—	5 : 5	0.75	—	215.4 ± 4.6	0.459 ± 0.011	1.81 ± 0.08
CP-NIM3	—	5 : 5	—	—	226.7 ± 6.7	0.254 ± 0.028	—
CP-MIM4	—	5 : 7	0.75	—	211.3 ± 4.8	0.204 ± 0.043	1.29 ± 0.17
CP-NIM4	—	5 : 7	—	—	231.6 ± 3.5	0.158 ± 0.022	—
CP/DA-MIM1	—	5 : 5	0.75	30	203.0 ± 2.9	0.557 ± 0.031	2.06 ± 0.15
CP/DA-NIM1	—	5 : 5	—	30	199.7 ± 5.2	0.270 ± 0.019	—
CP/DA-MIM2	—	5 : 5	0.75	60	185.3 ± 3.4	0.719 ± 0.017	2.75 ± 0.06
CP/DA-NIM2	—	5 : 5	—	60	190.4 ± 4.4	0.261 ± 0.013	—
CP/DA-MIM3	—	5 : 5	0.75	120	196.2 ± 4.1	0.505 ± 0.024	1.86 ± 0.09
CP/DA-NIM3	—	5 : 5	—	120	188.7 ± 3.9	0.271 ± 0.014	—
CP/CNT/DA-MIM1	0.5	5 : 5	0.75	60	191.1 ± 2.7	0.711 ± 0.018	2.68 ± 0.11
CP/CNT/DA-NIM1	0.5	5 : 5	—	60	201.3 ± 6.2	0.265 ± 0.029	—
CP/CNT/DA-MIM2	1	5 : 5	0.75	60	182.5 ± 4.9	0.726 ± 0.012	2.76 ± 0.06
CP/CNT/DA-NIM2	1	5 : 5	—	60	185.2 ± 3.6	0.263 ± 0.035	—
CP/CNT/DA-MIM3	1.5	5 : 5	0.75	60	165.2 ± 5.4	0.433 ± 0.038	2.01 ± 0.15
CP/CNT/DA-NIM3	1.5	5 : 5	—	60	175.7 ± 3.8	0.215 ± 0.015	—
CP/CNT/DA-MIM4	1	5 : 5	0.15	60	185.4 ± 6.3	0.395 ± 0.026	1.50 ± 0.12
CP/CNT/DA-MIM5	1	5 : 5	1.5	60	197.8 ± 2.9	0.687 ± 0.021	1.40 ± 0.15

---

2.3 Characterization of the membranes

The morphologies of the membranes were observed by a TM-1000 Scanning Electron Microscope (SEM) (Hitachi, Japan) and a JEOL 7800F Field Emission Scanning Electron Microscope (FESEM) (Hitachi, Japan). The chemical compositions of the membranes were recorded on a Thermo Nicolet Nexus 330 FT-IR spectrometer (Madison, USA) with a scanning range from 400 to 4000 cm⁻¹ at room temperature. The calorimetry measurements were made using a Mettler Toledo 822e Differential Scanning Calorimeter (DSC) (Mettler-Toledo Inc., USA). All membranes were ground and tested in crimped aluminum pans at a heating rate of 10 °C min⁻¹ under dry N₂ gas (25 mL min⁻¹) over a temperature range from 25 to 300 °C. The mechanical properties of the membranes (2.0 × 4.0 cm) were assessed using AG-I model mechanical testing apparatus (Shimadzu Co., Japan) with an elongation rate of 1 mm min⁻¹ at room temperature. The crystalline structures of the membranes were analyzed by X-ray diffraction (XRD) (XRD-6100, Shimadzu Co., Japan) using Cu Kα radiation (λ = 0.154 nm) at 40 kV and 200 mA. For the chemical composition analysis, a Kratos Axis Ultra DLD X-ray photoelectron spectroscopy (XPS) (Kratos, US) device was used with a mono Al Kα X-ray source operating at 150 W.

2.4 Swelling tests

The swelling tests of the composite membranes were studied before and after hydration. The weight of a dry membrane (1 × 1 cm) was taken as W_d. The dry membrane was immersed in water for 1 h. The membrane was then removed and swept with a filter paper; the wet membrane (W_w) was subsequently weighed. The swelling degree W (%) of the membrane was calculated according to eqn (1):

2.5 Binding capacities of the membranes

Single protein solutions were prepared by dissolving BSA, HSA or Lyz in a phosphate buffer solution (50 mM, pH = 7.0). The protein binding was conducted by soaking and gently shaking the membrane sample in 10 mL of single protein solutions with different concentrations at 25 °C. After 2 h, the membrane sample was removed from the solution and washed with 20 mL of phosphate buffer (5 min, 3 times). The membrane was then soaked and shaken in 10 mL of 10% SDS : acetic acid (1 : 1, v/v) for 2 h to elute the bound protein at 25 °C. The concentration of the protein solution before and after absorption by the membrane were determined using a fluorescence spectrophotometer. The protein binding capacity (Q) was calculated according to eqn (2):where C₀ (μmol mL⁻¹) and C_f (μmol mL⁻¹) are the initial and final concentrations of the protein in solution, respectively; v (mL) is the total volume of the solution; and V (cm³) is the volume of the membrane.

The imprinted factors (IFs) of the membranes were calculated using the relationship given below:

where Q_MIM and Q_NIM are the adsorption amounts of the molecularly imprinted membrane (MIM) and non-imprinted membrane (NIM) for the target protein, respectively.

2.6 Sample analysis

To evaluate the applicability and separation of CP/CNT/DA-MIM, CP/CNT/DA-MIM was used to separate BSA from bovine blood samples. The bovine blood sample was diluted 100-fold with Tris–HCl (50 mM, pH = 7.0) and then treated with CP/CNT/DA-MIM for 2 h at 25 °C. Then, the adsorbed membrane was rinsed with 2 mL of water and 10% SDS : acetic acid (1 : 1, v/v) was applied for 2 h to elute the adsorbed protein. SDS-PAGE was employed to detect the diluted bovine blood sample, the diluted bovine blood sample after membrane treatment and the eluted sample.

3. Results and discussion

3.1 Optimization of the preparation conditions for the membranes

The preparation processes of the membranes were optimized by changing crucial factors, including the mass ratio of CS to PVP, and the amounts of DA, MWCNTs and template. The results are given in Table 1 and Fig. 3.

Fig. 3 Photographs of the membranes prepared with different amounts of MWCNTs. (A) CP-MIM3 (0%); (B) CP/DA-MIM2 (0%); (C) CP/CNT/DA-MIM1 (0.5%); (D) CP/CNT/DA-MIM2 (1%); (E) CP/CNT/DA-MIM3 (1.5%).

To control the adhesion and porosity of the CS membrane, PVP was mixed with CS and CS/PVP (CP) served as the supporter in the matrix. Fig. 3(A) shows that CS and PVP were compatible and that the mixture of CS/PVP was a homopolymer due to the strong hydrogen bonding forces between them. These results are in agreement with results from previous studies.³¹ As shown in Table 1, with the increase in the ratio of PVP in the mixture, the swelling degrees and adsorption capacities of the membranes gradually decreased, but the IF values of the membranes increased accordingly. This likely occurred because the addition of PVP reduced the size of the inner pores of the CS membrane; therefore, the retaining property of the membrane was decreased. Meanwhile, the presence of PVP covered part of the functional groups of CS and then inhibited the nonspecific adsorption properties of the membrane. This result is in accordance with the findings of Lim et al.³⁶ They studied the interconnectivity of CS/PVP scaffolds, which could be increased by increasing the amount of PVP in the mixtures. However, excess PVP could aggregate and break the minute pore structures; then, the adsorption capacity and IF of the membrane were simultaneously decreased in CP-MIM4.

Fig. 3 shows the photographs of the different membranes with different transparencies. Comparing Fig. 3(A) with Fig. 3(B), the color of CP/DA-MIM is darker than that of CP-MIM, indicating that DA was cross-linked with chitosan.³⁷ These results are important because the presence of DA significantly improved the adsorption capacities and imprinted factors of the membranes for the target proteins (Table 1). DA has multifunctional groups (catechol and amine groups), and good hydrophilicity and biocompatibility properties; these merits make it appropriate for imprinting proteins.³⁸ Dopamine was pre-assembled with the target protein through a noncovalent interaction before polymerization. The dopamine-protein unit was fixed in the matrix after polymerization, and the dopamine was covalently bound to CS through glutaraldehyde crosslinking. Then, a particular imprinting cavity was formed in the membrane after eluting the template protein; meanwhile, dopamine was retained in the cavity as the specific recognition site, which provided the anchoring structure for the protein. Conversely, excess DA self-polymerized and blocked the movement of proteins in and out of the pores, which resulted in a reduction of the adsorption capacity for CP/DA-MIM3 (Table 1).

The mechanical properties of a membrane are crucial to its performance; membranes should not only be flexible but also strong enough to be handled in practical applications.³⁹ To verify the enhancing effect of MWCNTs on the mechanical properties of a CP membrane, membranes with different MWCNT content were investigated. The tensile strengths of all of the membranes decreased after hydration, while the elongation increased, as shown in Table 2. Fig. 3 shows that the transparency of the membranes decreased gradually with the increase in MWCNTs. The swelling degrees of the membranes were also reduced significantly with the increase in MWCNTs (Table 1). These results are important because the tensile strength and elongation were simultaneously improved with the increase in MWCNTs, indicating that MWCNTs can extensively improve the mechanical properties of the membrane whether the membrane is dry or wet. However, the excessive addition of MWCNTs (reaching 1.5% in CS/PVP) resulted in reductions in the swelling degree and adsorption capacity. Excessive addition might have resulted in excessive MWCNTs filling the holes of the membrane, which influenced the water absorption and mass transfer of proteins into the membrane. In general, the optimal membrane obtained was CP/CNT/DA-MIM2, which had the highest adsorption capacity and imprinted factor, and a relatively superior mechanical strength.

Table 2 The tensile strength and elongation at breaking of the various membranes

	MWCNTs (%)	Tensile strength (N mm^−2)	Elongation (%)
Dry membranes	0	31.4	50%
	0.5	42.1	53%
	1	42.7	108%
	1.5	55.2	90%
Wet membranes	0	3.0	105%
	0.5	4.6	243%
	1	5.2	227%
	1.5	7.1	205%

---

3.2 Characterization of the membranes

It is well known that the miscibility of the molecules in a polymer blend can be judged by its morphology and/or its solid state properties, such as its crystalline state, melting temperature, glass transition temperature, mechanical relaxation, and so on.³³ Consequently, the morphologies and physical/chemical properties of the membranes were characterized.

3.2.1 SEM and FESEM. Fig. 4 shows the SEM and FESEM images of the front and back side of CP/CNT/DA-MIM and its cross-section. It can be observed that the thickness of the membrane is approximately 100 μm and that the membrane has an asymmetric structure with three features. (1) The front side of the membrane is more crudely assembled than the back side (Fig. 4(A and B)), which favors the adsorption of templates onto the front side. (2) The funnelled thru-holes are widely distributed on the membrane (Fig. 4(A, B and D)), which was caused by the porogen. The diameters of the upstream sides of the funnelled thru-holes are approximately 5–10 μm (Fig. 4(A1)) and those of the downstream sides are approximately 1–2 μm (Fig. 4(B1)). The existence of funnelled thru-holes is a benefit for solute diffusion through the membrane. (3) Inhomogeneous nano-holes still exist inside the membrane (Fig. 4(C1 and E)). Note that a higher porosity inside the membrane might help to improve solute diffusion into the membrane and enhance the adsorption capacity of the membrane because the adsorption sites are both on the surface and inside the membrane. The possible interactions between the imprinted cavity and the target proteins are shown in Fig. 4(F). In general, the overall porosity of the membrane increased the contact area and facilitated the adsorption of target proteins by the membrane. Fig. 4(C1) shows that the MWCNTs were dispersed inside the membrane and cross-linked with the matrix, which improved the mechanical strength of the membrane. Activation of the MWCNTs introduced hydroxyl and carboxyl groups onto the surface of the MWCNTs, which not only improved the water-solubility and dispersibility of the MWCNTs but also facilitated crosslinking. The results are in line with the findings of Shieh et al.⁴⁰

Fig. 4 SEM and FESEM images of the front (A, A1) and back side (B, B1) of CP/CNT/DA-MIM, the cross-section of the membrane at different magnifications (C, C1), a model of a funnelled thru-hole (D), binding sites (E) and the imprinted cavity (F).

3.2.2 FT-IR. The FTIR spectra of CP/CNT/DA-MIM, CP-MIM and the activated MWCNTs are given in Fig. 1 of the ESI.† The absorption peaks at approximately 3443 cm⁻¹ (O–H or N–H), 1281 cm⁻¹ (C–N) and 1101 cm⁻¹ (–C–O–C–) are the characteristic peaks of CP-MIM (ESI Fig. 1(b)†), and all of the peaks were also observed in CP/CNT/DA-MIM. As shown in Fig. 1(a) of the ESI,† the bands at 1580 cm⁻¹ (the vibration of the benzene skeleton) and 683 cm⁻¹ (the out-of-plane bending of C–H) were both attributed to the characteristic peaks of a phenyl group, indicating the presence of dopamine; the peak at 1646 cm⁻¹ was attributed to the stretching and bending vibrations of –C=N–, suggesting that dopamine was cross-linked with chitosan using glutaraldehyde. As seen in Fig. 1(c) of the ESI,† the absorption peaks at approximately 3418 cm⁻¹ (O–H) and 1129 cm⁻¹ (C–O) were the characteristic peaks of the hydroxyl groups in the MWCNTs; however, these characteristic peaks were covered by CP/CNT/DA-MIM, indicating that the MWCNTs has been successfully activated and were all immersed in the hybrid membrane. The observations in this study agree well with the findings of Venkatesan et al.⁴¹

3.2.3 XPS. XPS was applied to characterize the surface chemical compositions of the membranes (Fig. 5 and 6). As shown in Fig. 5, it was calculated that the C/N atomic ratio in CP-MIM was 9.8, close to the theoretical value of 8.5. After adding MWCNTs into the membrane, the C/N atomic ratio of CP/CNT-MIM greatly increased (C/N = 26.6). The addition of dopamine reduced the atomic ratio of C and N in CP/CNT/DA-MIM (C/N = 19.5), suggesting that the MWCNTs and dopamine were well-wrapped and cross-linked into the membranes and that a decreased loss occurred during the manufacturing procedure. Fig. 6(A) shows that the peak at 284.8 eV originates from the CS or PVP sp² carbon atoms (C–C bonds); however, the peak at 286.7 eV is attributed to the C–O–C bonds in CS, and the peak located at 288.6 eV is due to the C=O bonds in PVP.⁴² In Fig. 5(b) and Fig. 6(B), the C–C bond content reaches up to 58.6%, which might be attributed to the high-carbon structure of the multi-walled carbon nanotubes. Fig. 6(C) illustrates that the peak at 286.0 eV is attributed to the C=N bond, which results from the intermolecular cross-linking polymeric structure of chitosan, glutaraldehyde and dopamine.⁴³ The peak at 288.2 eV originates from the C–OH bonds in chitosan or dopamine, and the content of C–OH bonds reaches up to 19.1%. In general, the abundant functional groups and well-defined network are favorable for efficient protein imprinting and rebinding.

Fig. 5 XPS spectra of CP-MIM (a), CP/CNT-MIM (b) and CP/CNT/DA-MIM (c).

Fig. 6 C 1s XPS spectra for (A) CP-MIM, (B) CP/CNT-MIM and (C) CP/CNT/DA-MIM.

3.2.4 DSC. To achieve greater insight into the thermal behavior of the membrane, DSC measurements were performed, and the resulting thermograms are shown in Fig. 7 and Table 3. Any changes to the physical properties of a polymer matrix from crosslinking can be reflected in the glass transition temperature (T_g), melting temperature (T_m) or degradation temperature (T_d).⁴⁴ The DSC thermogram of pure chitosan shows a T_g, T_m and T_d of 79.5, 192.9 and 312.7 °C, respectively (Table 3). The results correspond with the results of Rachipudi et al.⁴⁴ and He et al.,⁴⁵ but the T_g is quite different from the T_g of 203 °C reported by Sakurai et al.³³ These results are likely attributable to the different molecular weights and deacetylation degrees of the CS or the different procedures for the DSC measurements. Fig. 7(b, d and f) show the increasing trend of the endothermic peaks at approximately 187.6, 198.5 and 241.2 °C for the T_m values of CP-MIM, CP/CNT-MIM and CP/CNT/DA-MIM, respectively. Table 3 also shows that the T_d increased in the order of CP-MIM, CP/CNT-MIM and CP/CNT/DA-MIM; the thermal stabilities of the membranes increased with the addition of MWCNTs and DA. However, it was difficult to clearly observe the T_g values of these three membranes from the DSC curves. These results agree with previous research.⁴⁵

Fig. 7 The DSC curves of CS (a), CP-MIM (b), PVP (c), CP/CNT-MIM (d), DA (e) and CP/CNT/DA-MIM (f).

Table 3 The values of T_g, T_m and T_d for chitosan, polyvinyl pyrrolidone, dopamine and the membranes with different compositions

Sample	Tg	Tm	Td
CS	79.5	192.9	312.7
PVP	52.4	192.4	>400
DA	—	249.8	347.6
CP-MIM	—	187.6	300.9
CP/CNT-MIM	—	198.5	316.5
CP/CNT/DA-MIM	—	241.2	320.0

---

3.2.5 XRD. The crystalline properties of the membranes were analyzed with an X-ray diffractometer, as shown in Fig. 8. The two main peaks observed at 10.5° and 20.0° correspond to the 020 and 110 planes, respectively, which are the characteristic peaks of chitosan.³⁵ They were also found in all of the membranes, confirming the presence of chitosan (Fig. 8(a, b and c)). The diffraction peaks at 2θ = 26.4° and 43.4°, corresponding to the 111 and 100 planes, respectively, are the typical Bragg peaks of pristine MWCNTs (Fig. 8(d)).⁴⁶ However, there were no strong adsorption peaks observed for the MWCNTs in all of the membranes, suggesting that the incorporation of MWCNTs did not significantly affect the crystalline structure of CS. The results show agreement with previous studies,³⁵ possibly attributed to the lower amount of MWCNTs compared to chitosan. The results also suggest that MWCNTs were well-wrapped and dispersed into the membrane matrix.

Fig. 8 The XRD curves of CP/CNT/DA-MIM (a), CP/CNT-MIM (b), CP/CNT-NIM (c) and pristine MWCNTs (d).

3.3 Adsorption properties

The adsorption curves of the different membranes for BSA are shown in Fig. 9(A). The results intuitively illustrate that the adsorption capacities of CP/DA-MIM and CP/CNT/DA-MIM are relatively close and that both capacities are greater than that of CP-MIM. This indicates that dopamine played an important role in improving the binding capacity of the membrane, and the addition of MWCNTs showed less of an impact on the adsorption capacity of the membrane. The adsorption isotherm curves of CP/CNT/DA-MIM and CP/CNT/DA-NIM are shown in Fig. 9(B). The saturated adsorption capacity and imprinted factor of CP/CNT/DA-MIM for BSA were 0.726 μmol cm⁻³ and 2.8, respectively. The imprinted factor of CP/CNT/DA-MIM for Lyz was 1.2, which was much lower than that for BSA (Fig. 9(C)), indicating the specific selectivity of CP/CNT/DA-MIM for the template protein. The similarities between the sizes and compositions of the amino acids of BSA and HSA caused the CP/CNT/DA-MIM to show a certain selectivity for HSA (IF = 1.9), which was still lower than that for BSA. The results fully confirm that CP/CNT/DA-MIM shows a high specificity for the target protein.

Fig. 9 (A) Fluorescence spectra of BSA before and after adsorption by the different membranes; (B) adsorption isotherm curves of CP/CNT/DA-MIM and CP/CNT/DA-NIM; (C) adsorption capacities and imprinted factors of the membranes for different substrates.

3.4 Sample analysis

To investigate its adsorption, CP/CNT/DA-MIM was used to separate BSA from bovine blood samples, and the results are shown in Fig. 10. After treatment with CP/CNT/DA-MIM, the intensity of the BSA band in the blood sample (100-fold) was significantly weaker (lane 2). Then, the BSA band appeared again in the eluted solution after treatment with SDS–acetic acid (10%) (lane 3). The results indicate that CP/CNT/DA-MIM could specifically anchor BSA in the bovine blood sample. Thus, CP/CNT/DA-MIM has potential value in practical applications.

Fig. 10 SDS-PAGE analysis. Lane 1, bovine blood diluted 100-fold; lane 2, bovine blood sample (diluted 100-fold) after treatment with CP/CNT/DA-MIM; lane 3, the eluted solution; lane 4, 0.25 mg·mL⁻¹ of standard BSA solution; lane 5, standard weight molecular protein markers.

4. Conclusions

CP/CNT/DA-MIM was prepared for the selective anchoring of target proteins in a complex matrix, and its preparation conditions were optimized to achieve the best adsorption capacity, imprinted factor and mechanical properties. FTIR, XRD, DSC, XPS, SEM and FESEM were used to study the morphologies and physical/chemical properties of CP/CNT/DA-MIM. The results of this study show that the different molecules contained within CP/CNT/DA-MIM were uniformly dispersed and cross-linked together. The addition of appropriate amounts of PVP, MWCNTs and DA reduced the adhesion and nonspecific adsorption, enhanced the mechanical properties and improved the binding selectivity of the membrane, respectively. The applicability and separation effectiveness of CP/CNT/DA-MIM were also evaluated, and it was successfully used for the separation of BSA from a bovine blood sample. The generated membrane is an economical, stable and biocompatible material for the selective separation and purification of target proteins, which can overcome the disadvantages of conventional separation methods (such as the use of antibodies), which can be time-consuming, expensive and difficult to implement. Additionally, CP/CNT/DA-MIM was prepared and applied completely in the aqueous phase or under mild circumstances, which perfectly overcame the deficiencies of conventional preparation methods for the molecular imprinting of polymers. Such a nontoxic, biocompatible, hydrophilic and low cost membrane system could potentially be an outstanding separation material for large-scale continuous selective separations of target proteins from complex matrices in applications such as industrial protein purification, basic biomedical research and clinical diagnostics.

Acknowledgements

This work was financially supported by the National Natural Science Foundation of China (No. 81173024 and No. 81227802), the China Postdoctoral Science Foundation Funded Project (No. 2014M562428) and the Shaanxi Health Department of Scientific Research Project (No. 2014D73). We are grateful to Dr. Min Zhang for revising the paper.

References

1. R. Gao, X. Mu, J. Zhang and Y. Tang, J. Mater. Chem. B, 2014, 2, 783–792.
2. A. Nematollahzadeh, W. Sun, C. S. A. Aureliano, D. Lutkemeyer, J. Stute, M. Abdekhodaie, A. Shojaei and B. Sellergren, Angew. Chem., Int. Ed., 2011, 50, 495–498.
3. Y. Yang, X. He, Y. Wang, W. Li and Y. Zhang, Biosens. Bioelectron., 2014, 54, 266–272.
4. Z. Ding, S. W. Annie Bligh, L. Tao, J. Quan, H. Nie, L. Zhu and X. Gong, Mater. Sci. Eng., C, 2015, 48, 469–479.
5. B. T. S. Bui and K. Haupt, Anal. Bioanal. Chem., 2010, 398, 2481–2492.
6. M. J. Whitcombe, I. Chianella, L. Larcombe, S. A. Piletsky, J. Noble, R. Porter and A. Horgan, Chem. Soc. Rev., 2011, 40, 1547–1571.
7. G. Vasapollo, R. D. Sole, L. Mergola, M. R. Lazzoi, A. Scardino, S. Scorrano and G. Mele, Int. J. Mol. Sci., 2011, 12, 5908–5945.
8. M. Yang, L. Liao, G. Zhang, X. Xiao, Y. Lin and C. Nie, Anal. Bioanal. Chem., 2013, 405, 7545–7551.
9. R. Gao, L. Zhang, Y. Hao, X. Cui and Y. Tang, RSC Adv., 2014, 4, 64514–64524.
10. W. Du, Q. Fu, G. Zhao, P. Huang, Y. Jiao, H. Wu, Z. Luo and C. Chang, Food Chem., 2013, 139, 24–30.
11. J. Yin, Y. Cui, G. Yang and H. Wang, Chem. Commun., 2010, 46, 7688–7690.
12. Y. Fuchs, O. Soppera and K. Haupt, Anal. Chim. Acta, 2012, 717, 7–20.
13. J. Liu and G. Wulff, Angew. Chem., Int. Ed., 2004, 43, 1287.
14. S. K. Tsermentseli, P. Manesiotis, A. N. Assimopoulou and V. P. Papageorgiou, J. Chromatogr. A, 2013, 1315, 15–20.
15. Y. Hoshino, H. Koide, T. Urakami, H. Kanazawa, T. Kodama, N. Oku and K. J. Shea, J. Am. Chem. Soc., 2010, 132, 6644–6645.
16. Z. Luo, A. Zeng, P. Zheng, P. Guo, W. Du, K. Du and Q. Fu, Anal. Methods, 2014, 6, 7865–7874.
17. L. Meng, X. Qiao, J. Song, Z. Xu, J. Xin and Y. Zhang, J. Agric. Food Chem., 2011, 59, 12745–12751.
18. M. Shamsipur, H. R. Rajabi, S. M. Pourmortazavi and M. Roushani, Spectrochim. Acta, Part A, 2014, 117, 24–33.
19. L. Longo and G. Vasapollo, Mini-Rev. Org. Chem., 2008, 5, 163–170.
20. J. Fan, L. Li, Z. Tian, C. Xie, F. Song, X. Zhang and J. Zhu, J. Membr. Sci., 2014, 467, 13–22.
21. E. Verheyen, J. P. Schillemans, M. van Wijk, M. Demeniex, W. E. Hennink and C. F. van Nostrum, Biomaterials, 2011, 32, 3008–3020.
22. B. Krajewska, Enzyme Microb. Technol., 2004, 35, 126–139.
23. X. Zheng, Q. Lian and H. Yang, RSC Adv., 2014, 4, 42478–42485.
24. S. Chen, Y. Wu, F. Mi, Y. Lin, L. Yu and H. Sung, J. Controlled Release, 2004, 96, 285–300.
25. S. Wang, L. Shen, W. Zhang and Y. Tong, Biomacromolecules, 2005, 6, 3067–3072.
26. Y. Xia, T. Guo, M. Song, B. Zhang and B. Zhang, Biomacromolecules, 2005, 6, 2601–2606.
27. Y. Xia, T. Guo, H. Zhao, M. Song, B. Zhang and B. Zhang, J. Biomed. Mater. Res., Part A, 2009, 90, 326–332.
28. B. Haimovich, L. Difazio, D. Katz, L. Zhang, R. S. Greco, Y. Dror and A. Freeman, J. Appl. Polym. Sci., 1997, 63, 1393–1400.
29. A. Lejardi, R. Hernández, M. Criado, J. I. Santos, A. Etxeberria, J. R. Sarasua and C. Mijangos, Carbohydr. Polym., 2014, 103, 267–273.
30. A. R. Sarasam, R. K. Krishnaswamy and S. V. Madihally, Biomacromolecules, 2006, 7, 1131–1138.
31. T. H. M. Abou-Aiad, K. N. Abd-El-Nour, I. K. Hakim and M. Z. Elsabee, Polymer, 2006, 47, 379–389.
32. S. Sun, Y. Tang, Q. Fu, X. Liu, L. Guo, Y. Zhao and C. Chang, Int. J. Biol. Macromol., 2012, 50, 1002–1007.
33. K. Sakurai, T. Maegawa and T. Takahashi, Polymer, 2000, 41, 7051–7056.
34. B. Wang, X. Liu, Y. Ji, K. Ren and J. Ji, Carbohydr. Polym., 2012, 90, 8–15.
35. J. Venkatesan, B. Ryu, P. N. Sudha and S. K. Kim, Int. J. Biol. Macromol., 2012, 50, 393–402.
36. J. I. Lim, H. Im and W. K. Lee, J. Biomater. Sci., Polym. Ed., 2015, 26, 32–41.
37. H. Lee, S. M. Dellatore, W. M. Miller and P. B. Messersmith, Science, 2007, 318, 426–430.
38. Z. Xia, Z. Lin, Y. Xiao, L. Wang, J. Zheng, H. Yang and G. Chen, Biosens. Bioelectron., 2013, 47, 120–126.
39. P. I. Morgado, P. F. Lisboa, M. P. Ribeiro, S. P. Miguel, P. C. Simoes, I. J. Correia and A. Aguiar-Ricardo, J. Membr. Sci., 2014, 469, 262–271.
40. Y. T. Shieh and Y. F. Yang, Eur. Polym. J., 2006, 42, 3162–3170.
41. P. N. Venkatesan and S. Dharmalingam, J. Membr. Sci., 2013, 435, 92–98.
42. Y. Pan, H. Bao and L. Li, ACS Appl. Mater. Interfaces, 2011, 3, 4819–4830.
43. B. Ge, Y. Tan, Q. Xie, M. Ma and S. Yao, Sens. Actuators, B, 2009, 137, 547–554.
44. P. S. Rachipudi, A. A. Kittur, A. M. Sajjan and M. Y. Kariduraganavar, J. Membr. Sci., 2013, 441, 83–92.
45. L. He, L. Yao, D. Yang, Q. Cheng, J. Sun, R. Song and Y. Hao, J. Macromol. Sci., Part B: Phys., 2011, 50, 2454–2463.
46. Y. Hu, J. Li, Z. Zhang, H. Zhang, L. Luo and S. Yao, Anal. Chim. Acta, 2011, 698, 61–68.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra11966a

---