Protein adsorption by a high-capacity cation-exchange membrane prepared via surface-initiated atom transfer radical polymerization

Abstract

In this study, a weak cation-exchange (WCX) membrane was prepared via a “post-polymerization modification” method, which involved the surface-initiated atom transfer radical polymerization of glycidyl methacrylate (GMA) and subsequent two-step derivation. By varying the graft time of poly-GMA, a series of WCX membranes with different densities of carboxyl groups were fabricated. For the membrane with a graft time of 12 h, a high adsorption capacity of 125.0 mg mL⁻¹ was obtained by using lysozme (Lys) as a model protein, which is higher than that reported. The new parameters, the utilization percentage of carboxyl (UP) and the stoichiometric displacement parameter (Z) were introduced to theoretically investigate how the ligand density affects the adsorption behavior of Lys for the first time. The UP revealed an “increase first and then decrease” trend with the prolonging of graft time, which may result from the mutual effect of the flexibility of the polymer chain, the steric hindrance and the “adsorption-caused hindrance” effect. Remarkably, the Z value was found to increase with the prolonging of graft time, suggesting that more effective binding sites were interacting with a protein molecule when the density of carboxyl was increased. Finally, the WCX membranes were applied to purify Lys from egg white with a high recovery of 95.7%, which depends significantly on the adsorption capacity of the membranes.

---

1. Introduction

In the last decade, ion-exchange membranes have been extensively studied for the effective separation of protein, DNA and monoclonal antibodies from complex bio-samples.^1–4 In contrast to column chromatography, membrane chromatography brings the solute to the ion-exchange sites primarily by convection rather than diffusion, thus reducing the resistance to mass transfer and allowing low pressure drops, fast binding rate and high productivity. These properties make membrane chromatography a promising alternative approach for the high-throughput separation and purification of bio-macromolecules.^5–8 However, membrane chromatography suffers from lower binding capacity for bio-molecules due to the lower surface to bed volume ratio.^9–11 Recent research has shown that creating three-dimensional coatings by grafting polymer tentacles from the surface of membranes is an effective method to improve the binding capacity of membranes.^12–15

As a living controlled polymerization technique, surface-initiated atom transfer radical polymerization (SI-ATRP) stands out from various polymerization methods for the grafting of polymer because of its fine controllability,^16–20 and the direct SI-ATRP of charged monomers was often used to prepare ion-exchange membranes with high adsorption capacity.^21–24 For example, 2-dimethylaminoethyl methacrylate was directly grafted from regenerated cellulose (RC) membrane to prepare anion exchange membrane.²¹ The weak cation-exchange (WCX) membrane was also prepared by direct grafting of acrylic acid via SI-ATRP.²³ However, the acidic monomers were found to cause the catalyst deactivation during the polymerization process due to the coordination complex between the deprotonated acidic monomers and the copper catalysts, leading to the ineffective polymerization.²⁵ Alternatively, the “post-polymerization modification” method, which prepares the poly-acidic chains via the conversion of the pre-grafted neutral polymer chains, can be used to avoid the catalyst deactivation.^26,27 Moreover, when a monomer is expensive or commercially unavailable, the “post-polymerization modification” is a good alternative approach for SI-ATRP.^28,29 Therefore, the “post-polymerization modification” strategy could easily produce a variety of functional groups owing to the flexibility of the side chain, and the modified membranes often possess high adsorption capacity.

On the other hand, since SI-ATRP was introduced in the preparation of the high-capacity adsorbents,³⁰ most of the studies do have only focused on the relationship between the adsorption capacity and the polymerization conditions, such as the graft time and initiator density,^{21–24,28,29} whereas few works were conducted to theoretically investigate the adsorption behavior of proteins. As a result, the adsorption of protein is not well understood yet on the three-dimensional polymer grafted membranes. Among the present adsorption models,^31,32 stoichiometric displacement model for adsorption (SDM-A) is predicted to elucidate the protein adsorption on the three-dimensional polymer grafted membranes. SDM-A was originally introduced to describe the retention behavior of proteins on ion-exchange surfaces by Regnier and coworkers.³³ It implies a fundamental stoichiometric relationship between solute, solvent, and stationary phase. The called stoichiometric displacement parameter in SDM-A, Z, is the apparent number of solvent molecule releasing from the contact interface between the protein and the absorbent as a protein molecule is adsorbed.³⁴ In the previous reports, Z was used to investigate the effects of temperature and pH on the retention behavior of protein on traditional particle-type adsorbents.^35–38 However, the functional groups on the used adsorbents are monolayers rather than three-dimensional polymers, and their densities are often constant due to the difficulty in the controllability of the functionalization. By contrast, SI-ATRP technology can easily produce a series of polymers grafted surfaces with controllable density of functional groups,²⁴ therefore, Z is anticipated to theoretically investigate the effect of ligand density on the adsorption behavior of protein. This investigation would be helpful to understand the adsorption phenomenon on the polymers grafted surfaces constructed by SI-ATRP.

Herein, considering the problem caused by acidic monomers in the preparation of WCX membrane, a facile “post-polymerization modification” method, which involved the SI-ATRP of glycidyl methacrylate (GMA) and subsequent two-step derivations, was designed to prepare WCX membranes with different surface structures from the reported ones.^39,40 By varying the graft time of poly-GMA, a series of WCX membranes with different density of carboxyl were fabricated to obtain high-capacity membranes. By using lysozme (Lys) as the model protein, the Z value in SDM-A was utilized to theoretically investigate how the ligands density affects the adsorption behavior of protein for the first time. In addition, the utilization percentage of carboxyl (UP) was presented to characterize the overall accessibility of ion exchange sites. At last, the WCX membranes were applied to purify Lys from egg white to indicate the significant dependence of the productivity on the adsorption capacity of membranes.

2. Experimental

2.1. Materials

RC membranes with a diameter of 47 mm, average pore size of 0.45 μm and thickness of 160 μm were purchased from Sartorius (Göttingen, Niedersachsen, Germany). Lys (WOLSEN, 99.0%) was obtained from Sigma-Aldrich. 2-Bromoisobutyryl bromide (Aladdin Inc., 98.0%), CuBr (Sinopharm, 98%), 2,2′-bipyridyl (Sinopharm, 99.5%), GMA (Aladdin Inc., 98.0%), ethanediamine (Tianli Chemical Reagent Co., Ltd., AR) and chloroacetic acid (Fuchen Chemical Reagent Factory, AR) as well as other chemicals were analytical grade.

2.2. Preparation and characterization of the WCX membrane

2.2.1 Preparation of the initiator functionalized membranes. RC membranes were immersed in methanol for 15 min to eliminate glycerine and then dried prior to modification. The dried RC membranes were swelled in anhydrous tetrahydrofuran for 20 min, then the initiator 2-bromoisobutyryl bromide (0.1 mol, 0.5 mL) and triethylamine (0.1 mol, 0.5 mL) were added, and the reaction system was incubated in an ice bath for 3 h and then at 35 °C for 12 h. The obtained initiator functionalized membranes were rinsed extensively with methanol and distilled water in sequence, then stored in a vacuum oven until next use.

2.2.2 Grafting of the poly-GMA. The poly-GMA chains were grafted from the RC membranes via SI-ATRP according to our previous report.²⁸ Briefly, 2-propanol (30 mL), the initiator functionalized membranes (0.15 g) and GMA (2.0 mL) were added into a two-necked flask. This mixture was purged with N₂ for 30 min to remove the dissolved oxygen, then 2,2′-bipyridine (0.30 g) and copper(I) chloride (0.10 g) were added under a nitrogen atmosphere. The reaction was performed at 40 °C and the graft time was controlled to be 1, 3, 6 and 12 h. After the reaction, the poly-GMA grafted membranes were washed with acetone to remove the homo-polymer on the surface. Then the poly-GMA grafted membranes were immersed into 10% EDTA solution, stirred at room temperature to remove Cu²⁺, and then sequentially washed with water and methanol. The poly-GMA grafted membranes were dried under vacuum at 40 °C for 24 h until a constant weight was obtained.

The density of epoxy groups (EPO, mmol cm⁻²) of the poly-GMA grafted membranes was calculated according to the eqn (1) using 10 pieces of the membranes with the area of 4.34 cm² per piece.

Where, Δw is the weight increase after polymerization (mg), S is the total area of membranes (cm²) and M is the mole mass of GMA unit in polymer chain.

2.2.3 Amination of the poly-GMA grafted membranes. The poly-GMA grafted membranes were ammoniated via the “ring-opening” reaction between epoxy groups and amine. Briefly, the poly-GMA grafted membranes (0.4 g) were immersed in 10% ethanediamine aqueous solution (40 mL) and the reaction mixture was allowed to react at 80 °C for 12 h under stirring. After reaction, the ammoniated membranes were washed thoroughly with de-ionized water and subjected to the next reaction directly.

2.2.4 Carboxylation of the ammoniated membranes. The ammoniated membranes were then carboxylated successively by reacting with 40 mL of 0.1 mol L⁻¹ chloroacetic acid (pH 11.0) prepared in 0.12 mol L⁻¹ Na₂CO₃ solution at 80 °C for 18 h. The resultant carboxylated membranes were washed thoroughly with deionized water and vacuum drying.

Chemical composition of the membranes was characterized by the attenuated total reflectance fourier transform infrared spectroscopy (ATR-FTIR, Thermo-Nicolet 5700, America) and X-ray photoelectron spectroscopy (XPS, K-Alpha, VG Instruments, UK). Scanning electron microscope (SEM) (EDAX s-2700, USA) was used to observe the changes in the surface topography of the membranes.

2.3. Determination of the ion exchange capacity

Ion-exchange capacity of the WCX membrane was determined using back titration according to the references with minor modification.⁴¹ After immersed in distilled water, two pieces of membranes with the area of 4.34 cm² per piece were soaked in 10 mL 1.0 mol L⁻¹ HCl to change them into H⁺ form, washed with distilled water to remove excess HCl, then equilibrated in 0.5 mol L⁻¹ KCl for 12 h. Finally, several aliquots of 5 μL 0.05 mol L⁻¹ KOH were gradually added and the corresponding pH values were recorded until it reached 7.0. Ion-exchange capacity of the membrane was calculated by the consumed volume of KOH and the area of membrane.

2.4. Water flux measurement

For the water flux, the whole membrane disc with the diameter of 47 mm was used. The water flux was measured by a solvent filtration/degassing system with an air valve connected between the filtration system and the pump to adjust the pressure.²⁴ Before the measurement, each membrane was initially pre-compacted until a constant flux was observed. Then, the water flux was measured under 0.03, 0.06, 0.1 MPa and calculated in accordance with eqn (2)Where, J is the water flux (L m⁻² h⁻¹), V_w is the volume of the trans-membrane water (L), A is the area of the membrane (cm²) and t is the ultrafiltration time (h).

2.5. Determination of the adsorption capacity of Lys

Static adsorption capacity of the membrane was determined with batch adsorption by using Lys as the model protein. Two pieces of membranes with the area of 4.34 cm² per piece were inculcated in 5.0 mL of protein solution with different concentrations. The equilibrium concentration of Lys was measured by UV-Vis spectrophotometer (UV-2550, SHIMADZU) at 280 nm. The adsorption amount, Q (mg mL⁻¹), was calculated in accordance with eqn (3)Where, C₀ (mg mL⁻¹) is the initial concentration of protein, C_e (mg mL⁻¹) is the equilibrium concentration of protein, V₀ (mL) is the volume of protein solution and V (mL) is the volume of membrane.

Dynamic adsorption capacity of the membrane was determined by frontal analysis. A stack of 8 membrane discs with the diameter of 15 mm was densely packed into a membrane module fabricated according to the reference without any modification except that the dimensions were reduced to half.⁸ The column was equilibrated with loading buffer (20 mmol L⁻¹ phosphate buffer, pH 7.0) for 30 min. Next, 3.0 mg mL⁻¹ Lys was pumped through the column at a flow-rate of 1.0 mL min⁻¹ and the effluent was collected and measured by UV spectrum at 280 nm. The breakthrough curves, absorbance at 280 nm vs. effluent volume, were manually plotted. The volume at the point of 10% breakthrough (V_break) was used to estimate dynamic binding capacity according to eqn (4):⁸

Where, Q_10% is the dynamic binding capacity (mg mL⁻¹), C₀ is the concentration of protein in solution (mg mL⁻¹), V_dead is the system dead volume (mL) which was measured by 1.0 mg mL⁻¹ of NaNO₂, and V_c is the column volume (mL).

2.6. Purification of Lys from egg white by frontal analysis

The equipment was the same as that in the dynamic adsorption capacity measurement. The loading buffer was 20 mmol L⁻¹ phosphate (pH 7.0) and the eluting buffer was 20 mmol L⁻¹ phosphate containing 1.0 mol L⁻¹ NaCl (pH 7.0). Egg white was obtained from fresh egg and diluted in loading buffer at 1 : 12 dilution followed by gently stirring for 30 min at room temperature. The homogeneous was centrifuged at 10 000 rpm under 4 °C, and the suspension was collected and stored at 4 °C prior to use. The protein concentration was measured by Bradford assay.⁴²

3. Results and discussions

3.1. Preparation and characterization of the membranes

To avoid the catalyst deactivation in the case of the direct grafting of acrylic acid from RC membrane,²³ we designed a “post-polymerization modification” route to prepare the WCX membrane. A series of poly-GMA grafted membranes were prepared by manipulating the graft time, and then reacted with ethanediamine and followed by chloroacetic acid to create high-density of –COOH as cationic exchange sites (Fig. 1). Both derivative reactions are easily carried out under mild condition, and also the productivities are often quite high. The chemical composition of membrane was characterized by ATR-FTIR and XPS (Fig. 2). In comparison with the RC membrane, there appears an obvious peak at 1740 cm⁻¹ for poly-GMA grafted membrane (Fig. 2A, curve c), which is ascribed to the characteristic peak of ester carbonyl. For the ethanediamine derived membrane (Fig. 2A, curve d), the strong peak around 1255 cm⁻¹ is ascribed to the in-plane bending of N–H of primary amino, and the peak at 3600–3200 cm⁻¹ can be ascribed to –OH. After reacted with chloroacetic acid, the membrane indicated an enhanced adsorption for carboxyl around 1740 cm⁻¹ (Fig. 2A, curve e). All these results indicated the successful preparation of the WCX membranes according to the expected routine in Fig. 1.

Fig. 1 Preparation routine of the WCX membrane.

Fig. 2 (A) ATR-FTIR spectrums of different membranes (a: original RC membrane, b: 2-BIB functionalized, c: poly-GMA grafted, d: EA derived, e: chloroacetic acid derived RC membrane); (B) XPS spectrums of the WCX membranes with SI-ATRP time of 1 h (a), 3 h (b), 6 h (c) and 12 h (d).

The elemental compositions of the WCX membranes were characterized by XPS (Fig. 2B). N 1s peak at 400.08 eV was observed in the XPS spectrums, and the N contents were calculated to be 1.1%, 1.6%, 4.7% and 5.8% corresponding to the graft time of 1, 3, 6 and 12 h, respectively. Table 1 summarizes the density of epoxy groups, amine moieties calculated from the N contents and the carboxyl group capacity measured by titration. It can be seen that the densities of all the functional groups were increasing with the prolonging of graft time and the density of carboxyl groups on the membrane was as high as 9.8 μmol cm⁻², which was much higher than that of WCX membranes prepared by direct bonding.^10,40

Table 1 Density of functional groups on the membranes for each modification step

ATRP time (h)	Density of functional groups (μmol cm^−2)
	Epoxy groups	Amine moieties	Carboxyl groups
1	2.1	3.5	1.8
3	4.7	6.2	2.8
6	9.1	16.2	6.7
12	12.4	20.4	9.8

---

3.2. Effect of graft time on the permeability

The water flux was measured to investigate the penetrability of membranes, which was closely related to the structure of membrane pores. As shown in Fig. S1,† the water flux of the membranes decreases substantially with the increase of graft time. The changes in the topography of membranes were observed by SEM (Fig. S2†) and Image-Pro PLUS software was used to quantitatively display the changes in the pore diameter.^28,43 The statistic results indicated that the average surface pore diameter decreased from 1.16 to 0.54 μm as the graft time was varied from 1 to 16 h. The decreased pore diameter may come from the clog of membrane pores by the growing polymer chains when the polymerization time was extended. As a result, the permeability of membranes decreased. Therefore, in order to ensure lower pressure drop in membrane chromatography, it is very important to the control the graft time when the SI-ATRP was proceeded to yield high adsorption capacity.

3.3. Effect of graft time on adsorption capacity

Generally, the adsorption capacity of ion-exchange membrane is closely related to the number of ion exchange sites. From this point of view, the graft time would have great impacts on the adsorption capacity of protein because more –COOH groups were introduced when the graft time was prolonged. As shown in Fig. 3A, the membranes have strong adsorptions towards Lys and the saturation adsorption occurs when the concentration of Lys is 3.0 mg mL⁻¹. Langmuir, BET and Frendlich models were used to fit the adsorption isotherm to evaluate the adsorption behavior of Lys, among which Langmuir model give the best fitting. The parameters obtained from Langmuir equation are listed in Table S1.† From the fitting equations, the static adsorption capacities (Q_m) are calculated to be 23.3, 55.6, 100.0 and 125.0 mg mL⁻¹ corresponding to the membranes with graft time of 1, 3, 6 and 12 h, respectively, indicating a positive relationship between the adsorption capacity and graft time.

Fig. 3 (A) Adsorption isotherms of Lys on the WCX membranes; (B) the typical breakthrough curves at the flow rate of 1.0, 3.0 and 5.0 mL min⁻¹ for the WCX membrane with graft time of 12 h.

The dynamic adsorption capacity at 10% breakthrough (Q_10%) for the column was determined by frontal analysis using 3.0 mg mL⁻¹ Lys as the feed solution and the typical breakthrough curves are shown in Fig. 3B. When the flow rate was 1.0 mL min⁻¹, the Q_10% was calculated to be 19.6, 43.8, 79.4 and 114.9 mg mL⁻¹ corresponding to the membranes prepared at SI-ATRP time of 1, 3, 6 and 12 h, respectively. The measured Q_10% was about 78–92% of the Q_m, demonstrating that the charged cation-exchange groups are highly accessible.²¹ High accessibility of protein to binding sites was further supported by measuring the Q_10% at higher volumetric flow rates. The Q_10% at the flow rate of 3.0 and 5.0 mL min⁻¹ were calculated to be 102.6 and 100.9 mg mL⁻¹, respectively, which were very close to the Q_10% at the flow rate of 1.0 mL min⁻¹. Therefore, the mass transfer of protein molecules to the binding sites of the macroporous membrane is limited primarily by convection.⁶

The adsorption capacity and productivity of the WCX membrane obtained at the grafted time of 12 h were compared with the commercial Sartobind C membrane and the previously reported ones (Table 2). The Q_10% of the commonly used carboxylic acid C Sartobind C membrane was reported to be 0.6 mg cm⁻² (equivalent to 21.8 mg mL⁻¹). In contrast, the Q_10% of the WCX membrane in our work is approximately 6 times higher than that of commercial Sartobind C membrane. Moreover, the WCX membrane in our work also exhibits higher adsorption capacities for Lys in comparison with the previously reported weak or strong cation-exchange membranes. When comes to the productivity, our WCX membrane achieved a productivity of 17.7 mg mL⁻¹ min⁻¹ at the flow rate of 1.0 mL min⁻¹, also being much higher than the previously reported ones. Hence, the WCX membrane in our study shows quite high adsorption capacity and productivity.

Table 2 Comparison of the lysozyme adsorption capacity and productivity

Membranes or resins	Q10% (mg mL^−1)	Qm (mg mL^−1)	Productivity (mg mL^−1 min^−1)	Reference
a Data from the manufacturer.b Resin.c Data calculated at the flow rate of 1.0 mL min^−1. crAAAm: Co-grafting of the “cross-linker” and monomer acrylamide (AAm) on the cellulose-based macroporous membrane. SPMAK: 3-sulfopropyl methacrylate, potassium salt NASS: sodium 4-styrenesulfonate AA: acrylic acid.
Carboxylic acid C Sartobind C	21.8^a
Fractogel EMD COO–^b	47.0	50.0	0.2^c	44
MacroPrep CM^b	13.0	50.0	1.3^c	44
crAAAm	16.2	89.5	2.1^c	45
Poly-SPMAK grafted RC membrane	70.0	97.0	14.2^c	29
Poly-NASS grafted RC membrane	61.2	129.5	5.5^c	24
Particle-loaded hollow fiber membrane	60.0	60.0		46
Poly-AA grafted RC membrane	71.2	98.5		23
Polysulfone-based cation-exchange membrane	15.6			39
Poly-COO^− RC membrane	114.9	125.0	17.7^c	This study

---

3.4. Effect of graft time on the accessibility of carboxyl groups

The protein adsorption is usually dominated by the accessibility of the binding sites, which was affected by three factors. Firstly, the lengthening of polymer chain would lead to higher flexibility of the polymer chain and thus, better accessibility of the binding sites. Secondly, the steric hindrance resulting from the clog of membrane pores makes tough contact of protein with the binding sites when the polymer chains become longer. Thirdly, the “adsorption-caused hindrance” effect, which becomes severer with the increase of adsorption capacity, makes negative contribution to the adsorption.^35,47

Herein, the overall utilization efficiency of carboxyl groups, UP, was used to evaluate the comprehensive accessibility of the ion-exchange sites and UP was defined as the following:

As can be seen from Fig. 4, the UP increased sharply when the graft time was increased from 1 to 6 h but decreased after 6 h. This variation trend can be explained by three factors stated above. Generally, the contribution of each factor on UP increases with the graft time. However, during the initial stage of polymerization, the steric hindrance and “adsorption-caused hindrance” would produce relatively small contributions to UP because the pore diameter did hardly change²¹ and the adsorption capacity was also lower. Therefore, the change in UP mainly came from the higher flexibility of polymer chain, and thus trended to increase with the graft time. After 6 h, the flexibility increased with the longer polymer chain, whereas longer polymer chain made the steric hindrance and “adsorption-caused hindrance” increase so fast that more carboxyl became un-accessible to protein, resulting in the decrease in UP. Hence, for convenience and timesaving consideration as well as the adsorption capacity of membranes, it is advised that the length of grafted polymer chain should not to be too long.

Fig. 4 Effect of graft time on the UP.

3.5. Effect of graft time on stoichiometric displacement parameter

According to the SDM-A, the stoichiometric displacement parameter, Z, is the apparent number of solvent molecule releasing from the contact interface between the protein and the absorbent as a protein molecule is adsorbed.³⁴ On the ion exchange surfaces, Z was used to characterize the apparent number of ion exchange sites interacting with the protein molecule in the adsorption process^32,35,36 and the Z value can be determined by measuring the retention of protein over a range of ionic strengths according to eqn (6):³³

log K_d = β − Z log a_D (6)

where, K_d is the distribution coefficient of protein and a_D is the ionic strength. Then, a linear relationship between log K_d and log a_D can be correlated according to eqn (6) and Z can be calculated from the slope of equation.

Until now, Z was found to relate with the molecular conformation of protein that would be affected by the temperature and pH of the solution.^35–38 However, there is no theoretical research about the effect of surface properties on the Z value. Here, a series of WCX membranes with controllable density of carboxyl groups were constructed via SI-ATRP by varying the graft time, and then Z was obtained for the different WCX membranes according to SDM-A for the first time. As is shown in Table 3, the Z value was found to increase with the prolonging of graft time, suggesting that more carboxyl groups were interacting with each protein molecule when the graft time was increased. Actually, the increased Z value was ascribed to the increased density of the carboxyl groups caused by the prolonging of graft time. Therefore, the findings provide the theoretical evidences that the ligand density has great effect on the adsorption behavior of protein and is helpful to understand the protein adsorption on the three-dimensional coating membrane surfaces constructed by SI-ATRP.

Table 3 Linear regression parameters obtained from SDM

ATRP time (h)	β	Z	R^2
1	−2.36	1.04	0.988
3	−2.14	1.30	0.956
6	−2.18	1.57	0.970
12	−2.11	1.86	0.953

---

3.6. Purification of Lys from egg white by frontier analysis

Purification of Lys from chicken egg white is the most widely used model to evaluate the performance of cation-exchange absorbent since it is composed of a mixture of competitive proteins. Among the proteins in egg white, ovalbumin (45 kDa, pI = 4.6), ovomucoid (30–35 kDa, pI = 4.0), conalbumin (77 kDa, pI = 6.6) are the major components constituting 54%, 11% and 13%, respectively, with Lys (14.4 kDa, pI = 11.1) as the minor component about 3.5%.⁴⁸ At a working pH of 7.0, Lys is positively charged, whereas the other proteins are negatively charged. Then, the reversible interaction of Lys with –COO⁻ groups on the membrane can occur by electrostatic attraction, while the negative proteins are excluded. Frontier analysis of egg white is shown in Fig. 5. On the loading stage (0–15 min), the diluted egg white was pumped through the membrane column. The positive Lys was absorbed on the membrane while the other proteins were excluded and broke through from the column. As is shown in Fig. 6, fractions of the loading stage (lane 3) consists most of the egg white proteins except Lys compared with the egg white extract (Fig. 6, lane 2). On the washing stage (15–35 min), the physically absorbed protein was washed by the phosphate buffer (Fig. 6, lane 4). Then, on the elution stage (35–50 min), the elution buffer containing 1.0 mol L⁻¹ NaCl was pumped to elute the absorbed protein via electrostatic attraction. As is shown in Fig. 6 lane 5, the protein band corresponding to Lys is observed in elution fraction with high purity. The results indicated that the prepared WCX membranes can be successfully applied in membrane chromatography to effectively purity Lys from egg white.

Fig. 5 Purification of Lys from egg white by frontier analysis.

Fig. 6 SDS-PAGE analysis of effluent fractions in Fig. 5 (lane 1: protein molecular weight marker, lane 2: egg white extract, lane 3: fraction in 0–15 min, lane 4: fraction in 15–35 min, lane 5: fraction in 35–50 min, lane 6: standard Lys).

To see the difference in the purification efficiency among the WCX membranes with different graft time, three membranes (SI-ATRP time of 3, 6 and 12 h) were subjected to purity Lys from egg white by frontier analysis. The same volume of sample was loaded on the membrane column and the eluted fraction was collected to measure the concentration of Lys. The results are summarized in Table 4. It can be found that the yield of Lys depended significantly on the graft time, increasing from 47.1% to 95.7% when the graft time was increased from 1 h to 12 h. At the short graft time, the obtained WCX membranes had low adsorption capacity for Lys, and thus adsorbed fewer Lys from the solution. When the adsorption capacity was enhanced by the longer graft time, more and more Lys can be adsorbed from the solution, thus giving a higher yield of Lys. Therefore, SI-ATRP is an effective method to prepare high capacity adsorbents to improve the purification efficiency of protein.

Table 4 Lys yields from egg white by frontier analysis

ATRP time (h)	Sample volume (mL)	Total protein (mg)	Production of Lys (mg)	Yields^a (%)
a Yields = [production of Lys/(total protein × 3.5%)] × 100%.
3	15	80.4	1.32	47.1
6			2.04	72.9
12			2.70	95.7

---

4. Conclusions

In this study, the “post-polymerization modification” method was designed to effectively prepare WCX membranes via the conversion of the neutral poly-GMA chains that were pre-grafted from RC membranes by SI-ATRP. The procedures are easily carried out and capable of avoiding the catalyst deactivation caused by acidic monomers during the SI-ATRP. By varying the graft time of poly-GMA, a series of WCX membranes with different density of carboxyl were fabricated, among which the one at polymerization time of 12 h was found to produce a capacity of 125.0 mg mL⁻¹ for Lys, being higher than the reported values. Two new parameters, UP and Z, were proposed to theoretically investigate the adsorption behavior of Lys on the three-dimensional coating WCX membrane for the first time and the findings can be helpful to understand the protein adsorption behavior. At last, the purification efficiency of the WCX membranes was found to depend significantly on the graft time. Overall, the SI-ATRP can be a promising method to improve the adsorption capacity of adsorbents and thus enhance the productivity of proteins in the real applications.

Conflict of interest

The authors have declared no conflict of interest.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (No. 21275115, 21475104 and 21575114) and program for Changjiang Scholars and Innovative Research Team in University (IRT-15R55).

References

1. M. A. Teeters, S. E. Conrardy, B. L. Thomas, T. W. Root and E. N. Lightfoot, J. Chromatogr. A, 2003, 989, 165–173.
2. Y. Qiang, D. M. Michael, J. C. Hall and R. Ghosh, J. Chromatogr. A, 2008, 1187, 128–137.
3. C. S. Rao, Process Biochem., 2001, 37, 247–256.
4. R. D. Sloboda and L. M. Belfi, Protein Expression Purif., 1998, 13, 205–209.
5. E. Klein, J. Membr. Sci., 2000, 179, 1–27.
6. J. Thömmes and M. Etzel, Biotechnol. Prog., 2007, 23, 42–45.
7. R. van Reis and A. Zydney, J. Membr. Sci., 2007, 297, 16–50.
8. V. B. Bharat, K. A. Christensen and S. M. Husson, J. Chromatogr. A, 2010, 1217, 4495–4946.
9. H. Y. Gan, Z. H. Shang and J. D. Wang, J. Chromatogr. A, 2000, 867, 161–168.
10. C. Y. Wu, S. Y. Suen, S. C. Chen and J. H. Tzeng, J. Chromatogr. A, 2003, 996, 53–70.
11. R. Ghosh, J. Chromatogr. A, 2002, 952, 13–27.
12. X. Meng, Y. Fang, L. Wan, X. Huang and Z. Xu, Langmuir, 2012, 28, 13616–13623.
13. H. Dong, Y. Xu, Z. Yi and J. Shi, Appl. Surf. Sci., 2009, 255, 8860–8866.
14. Z. Wu, H. Chen, X. Liu, Y. Zhang, D. Li and H. Huang, Langmuir, 2009, 25, 2900–2906.
15. F. J. Xu, J. P. Zhao, E. T. Kang, K. G. Neoh and J. Li, Langmuir, 2007, 23, 8585–8592.
16. Z. Huang, X. Zhang, S. Wang, B. Yang, K. Wang, J. Yuan, L. Tao and Y. Wei, RSC Adv., 2015, 5, 65884–65889.
17. J. Jin, X. Li, P. Sun and H. Zhao, RSC Adv., 2013, 3, 7023–7029.
18. Q. Feng, D. Tang, H. Lv, W. Zhang and W. Li, RSC Adv., 2015, 5, 62024–62032.
19. W. Qin, W. Zhang, L. Song, Y. Zhang and X. Qian, Anal. Chem., 2010, 82, 9461–9468.
20. H. Huang and L. He, RSC Adv., 2014, 4, 13108–13118.
21. V. B. Bharat and S. M. Husson, J. Membr. Sci., 2009, 337, 215–223.
22. B. V. Bhut, S. R. Wickramasinghe and S. M. Husson, J. Membr. Sci., 2008, 325, 176–183.
23. N. Singh, J. Wang, M. Ulbricht, S. R. Wickramasinghe and S. M. Husson, J. Membr. Sci., 2008, 309, 64–72.
24. Y. Wei, J. Ma and C. Wang, J. Membr. Sci., 2013, 427, 197–206.
25. A. Y. Sankhe, S. M. Husson and S. M. Kilbey, Macromolecules, 2006, 39, 1376–1383.
26. S. G. Boyes, B. Akgun, W. J. Brittain and M. D. Foster, Macromolecules, 2003, 36, 9539–9548.
27. M. Biesalski and J. Rühe, Macromolecules, 2003, 36, 1222–1227.
28. X. L. Qian, H. Fan, C. Z. Wang and Y. M. Wei, Appl. Surf. Sci., 2013, 271, 240–247.
29. H. C. S. Chenette, J. R. Robinson, E. Hobley and S. M. Husson, J. Membr. Sci., 2012, 423, 43–52.
30. N. Li, R. Bai and C. Liu, Langmuir, 2005, 21, 11780–11787.
31. F. Yan, Y. Chu, K. Zhang, F. Zhang, N. Bhandari, G. Ruan, Z. Dai, Y. Liu, Z. Zhang, A. T. Kan and M. B. Tomson, Chem. Eng. J., 2015, 281, 921–930.
32. J. Xie, M. I. Aguilar and M. T. W. Hearn, J. Chromatogr. A, 1995, 711, 43–52.
33. W. Kopaciewicz, M. A. Rounds, J. Fausnaugh and F. E. Regnier, J. Chromatogr. A, 1983, 266, 3–21.
34. Y. M. Wei, Q. Z. Luo, T. Liu and X. D. Geng, Chin. J. Chem., 2000, 18, 60–65.
35. R. D. Whitley, R. Wachter, F. Liu and N. H. L. Wang, J. Chromatogr., 1989, 465, 137–156.
36. D. J. Roush, D. S. Gill and R. C. Willson, J. Chromatogr. A, 1993, 653, 207–218.
37. F. S. Marques, G. L. Silva, M. E. Thrash Jr and A. C. Dias-Cabral, Colloids Surf., B, 2014, 122, 801–807.
38. S. Bernardi, D. Gétazb, N. Forrerc and M. Morbidelli, J. Chromatogr. A, 2013, 1283, 46–52.
39. J. K. Fang, H. C. Chiu, J. Y. Wu and S. Y. Suen, React. Funct. Polym., 2004, 59, 171–183.
40. Y. Kacar and M. Y. Arica, Colloids Surf., B, 2001, 22, 227–236.
41. M. N. Mohamed and H. Saidi, J. Membr. Sci., 2003, 216, 27–38.
42. C. D. Georgiou, K. Grintzalis, G. Zervoudakis and I. Papapostolou, Anal. Bioanal. Chem., 2008, 391, 391–403.
43. N. Singh, S. M. Husson, B. Zdyrko and I. Luzinov, J. Membr. Sci., 2005, 262, 81–90.
44. A. Staby, J. H. Jacobsen, R. G. Hansen, U. K. Bruus and I. H. Jensen, J. Chromatogr. A, 2006, 1118, 168–179.
45. J. Wang, R. Faber and M. Ulbricht, J. Chromatogr. A, 2009, 1216, 6490–6501.
46. M. E. Avramescu, Z. Borneman and M. Wessling, J. Membr. Sci., 2008, 322, 306–313.
47. Y. L. Ling, T. S. Peng, D. X. Yan and Y. Sun, J. Chromatogr. A, 2013, 1305, 76–84.
48. Q. Luo, H. Zou, X. Xiao, Z. Guo, L. Kong and X. Mao, J. Chromatogr. A, 2001, 926, 255–264.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra24678d

---