Affinity Membranes and Monoliths for Protein Purification

Affinity capture represents an important step in downstream processing of proteins and it is conventionally performed through a chromatographic process. The performance of this step highly depends on the type of matrix employed. In particular, resin beads and convective materials, such as membranes and monoliths, are the commonly available supports. The present work deals with non-competitive binding of bovine serum albumin (BSA) on different chromatographic media functionalized with Cibacron Blue F3GA (CB). The aim is to set up the development of the purification process starting from the lab-scale characterization of a commercially available CB resin, regenerated cellulose membranes and polymeric monoliths, functionalized with CB to identify the best option. The performance of the three different chromatographic media is evaluated in terms of BSA binding capacity and productivity. The experimental investigation shows promising results for regenerated cellulose membranes and monoliths, whose performance are comparable with those of the packed column tested. It was demonstrated that the capacity of convective stationary phases does not depend on flow rate, in the range investigated, and that the productivity that can be achieved with membranes is 10 to 20 times higher depending on the initial BSA concentration value, and with monoliths it is approximately twice that of beads, at the same superficial velocity.


Introduction
Selectivity, versatility and efficiency make chromatography a widely used technique for protein purification and separation, due to the great number of possible chemical interactions that intervene between the target biomolecule and the stationary phase and to the availability of many adsorbent materials. Among the different kind of chromatographic separations usually needed for recovery and purification of a particular molecule, the main step in downstream processing is represented by the affinity capture. It allows the selective and efficient recovery of the target molecule thanks to a specific and reversible interaction between a ligand, immobilized on the support material, and the product itself.
Affinity chromatography is the most largely used separation technique for the purification and recovery of highly valuable biomolecules, such as enzymes, hormones, vaccines, DNA and RNA fragments and monoclonal antibodies [1,2]. Columns packed with functionalized resins or polymeric matrices, such as agarose, in the shape of spherical beads, represent the equipment traditionally used. Affinity particles are typically 50 to 100 µm in diameter, to minimize pressure drops [3] and the main mass transport mechanism toward the binding sites is diffusion [4]. Columns packed with chromatographic resins exhibit a good performance in terms of binding capacity, thanks to their high surface area, but the functionalized materials are usually very expensive and several limitations, Pure protein concentration was measured by UV absorption at 280 nm using a UV-vis spectrophotometer (UV-1601 Shimadzu Italia, Milan, Italy).

Membrane Functionalization
Affinity membranes were prepared by chemical immobilization of CB dye ligand on the surface of the convective supports, that were cut into discs of 2.6 cm diameter. Before modification, the membranes were equilibrated overnight in Phosphate Buffer Saline (PBS) at pH 7.0. A schematic representation of the CB immobilization reaction is shown in Figure 1a.
Ligand immobilization was performed incubating the membranes with a solution of 10 mg/mL of CB in water at 60 • C for 1 h, under stirring. This reaction was followed by the addition of 20% w/v NaCl aqueous solution; after 1 h, a solution of 25% w/v Na 2 CO 3 was added to catalyze the reaction, carried out at 80 • C for 4 h [54,57,61].
The affinity membranes obtained were washed several times with hot water, 20% v/v methanol, 2 M NaCl aqueous solution, adsorption buffer, elution buffer, water, 20% v/v methanol and 2 M NaCl, in this order, until the unbound dye was completely removed. Membranes were stored at 4 • C in 0.05 M phosphate aqueous solution at pH 7.0 containing 0.02% sodium azide to prevent microbial growth [62].
CB-affinity ligand density was experimentally determined for a different number of modified membranes, following the procedure developed by Ruckenstein and Zeng [63]. CB membranes were hydrolyzed with 2 mL of 12 N hydrochloric acid for 30 min at 80 • C. The obtained solution was diluted to 6 N using distilled water and finally neutralized with 4 mL of 6 N NaOH aqueous solution [47]. Dye concentration in the final solution was determined by absorbance reading at 610 nm.
Membranes 2020, 10, x 4 of 12 The affinity membranes obtained were washed several times with hot water, 20% v/v methanol, 2 M NaCl aqueous solution, adsorption buffer, elution buffer, water, 20% v/v methanol and 2 M NaCl, in this order, until the unbound dye was completely removed. Membranes were stored at 4 °C in 0.05 M phosphate aqueous solution at pH 7.0 containing 0.02% sodium azide to prevent microbial growth [62].
CB-affinity ligand density was experimentally determined for a different number of modified membranes, following the procedure developed by Ruckenstein and Zeng [63]. CB membranes were hydrolyzed with 2 mL of 12 N hydrochloric acid for 30 min at 80 °C. The obtained solution was diluted to 6 N using distilled water and finally neutralized with 4 mL of 6 N NaOH aqueous solution [47]. Dye concentration in the final solution was determined by absorbance reading at 610 nm.

Monolith Functionalization
Before modification, monoliths were thoroughly washed with distilled water. The immobilization of the affinity ligand on the surface of the polymeric monoliths took place in recirculation mode, placing each monolith disc into a Plexiglas module, connected to a peristaltic pump (Miniplus 3, Gilson, Milan, Italy). A schematic representation of the CB immobilization reaction is shown in Figure 1b.
A total of 50 mL of 5 mg/mL CB solution containing 1 M NaOH was fed to the column at 80 °C for 3 h. After modification, the affinity monoliths were washed using distilled water and an aqueous solution of 20% methanol, to remove the unbound CB, and were stored at 4 °C in water containing 0.02% sodium azide to prevent microbial growth [62].
CB-ligand density on the surface of monoliths was not measured, since the technique adopted for membranes was disruptive. However, monoliths after modification were homogeneously colored in blue and the effectiveness of modification was proved by BSA adsorption tests.

Monolith Functionalization
Before modification, monoliths were thoroughly washed with distilled water. The immobilization of the affinity ligand on the surface of the polymeric monoliths took place in recirculation mode, placing each monolith disc into a Plexiglas module, connected to a peristaltic pump (Miniplus 3, Gilson, Milan, Italy). A schematic representation of the CB immobilization reaction is shown in Figure 1b.
A total of 50 mL of 5 mg/mL CB solution containing 1 M NaOH was fed to the column at 80 • C for 3 h. After modification, the affinity monoliths were washed using distilled water and an aqueous solution of 20% methanol, to remove the unbound CB, and were stored at 4 • C in water containing 0.02% sodium azide to prevent microbial growth [62].
CB-ligand density on the surface of monoliths was not measured, since the technique adopted for membranes was disruptive. However, monoliths after modification were homogeneously colored in blue and the effectiveness of modification was proved by BSA adsorption tests.

Chromatographic Materials Characterization
Membranes, monoliths and the packed column were characterized in dynamic adsorption experiments, using pure BSA solutions. Dynamic binding capacity at saturation and at 10% breakthrough were calculated from the relevant breakthrough curves while productivity was derived from the mass of BSA eluted. A Fast Protein Liquid Chromatography (FPLC) and an AKTA Purifier 100 (GE Healthcare Life Sciences, Milan, Italy) were used to perform the dynamic characterization of all supports.
A summary of the dynamic experiments parameters is reported in Table 2. Table 2. Flow rate (F) and corresponding superficial velocity (v) and residence time (τ) for each chromatographic support. Pure BSA solutions at different initial concentrations, from 0.25 to 1.4 mg/mL, were fed to the chromatographic materials that were then washed and eluted using the appropriate buffers (see Section 2.1). Membranes and monoliths were placed inside specific holders. In particular, a layered stack of 5 membrane discs of 2.6 cm diameter (due to the presence of o-rings inside the membrane holder, the effective diameter reduces to 2.2 cm) was placed into a stainless steel membrane holder, designed by our research group to ensure the correct flow distribution and the appropriate degree of membrane compression. For the case of monoliths, each CIM disc was placed inside a column holder, designed and provided by BIA Separations.

Packed Column Membrane Monolith
By combining different values of initial BSA concentration with different values of flow rate, a significant number of experiments were performed, using more than one set of modified membranes and monoliths. However, due to the limited amount of monoliths and packed columns available, the experiments were performed as single sets of measurements. For this reason a statistical analysis of the data obtained was not possible.

Membrane Ligand Density
The functionalization of regenerated cellulose membranes resulted in an average CB density of 120 mg/mL. This value is much higher than that characterizing the commercially available packed column used in this work, since the technical sheet of this stationary phase reports a ligand density of 4 mg of CB per mL of resin. However, the higher CB density on membranes does not reflect a higher capacity, probably because the ligands are not completely available. As far as the monolith is concerned, the ligand density was indirectly measured performing BSA binding experiments, to check the effectiveness of the modification procedure.

Dynamic Binging Capacity
A parameter expressing the performance of chromatographic materials is the dynamic binding capacity at saturation (DBC 100% ), defined by the following equation: Membranes 2020, 10, 1 6 of 12 where m ads,100% is mass of product adsorbed performing the adsorption step until saturation and V support is the volume of the chromatographic support. However, due to the high product value and the high downstream processing costs, in the biopharmaceutical industry the adsorption step is usually stopped before column saturation, when the solute concentration in the stream exiting the chromatographic column reaches a particular value that usually corresponds to 10% breakthrough. In this case the parameter of interest to be calculated is the dynamic binding capacity at 10% breakthrough (DBC 10% ), that represents the amount of protein adsorbed per column volume, expressed by the following equation: where m ads,10% is mass of product adsorbed performing the adsorption step until 10% breakthrough and V support is the volume of the chromatographic media.
The effect of flow rate on the values of dynamic binding capacity at 10% breakthrough is shown in Figure 2 for membranes and monoliths, as well as for packed columns, for two different values of BSA concentration in the feed. The packed column is the one whose performance is mostly and negatively influenced by the feed flow rate, since DBC 10% decreases as the flow rate increases, contrary to what happens for the convective materials. This means that in the case of a packed column, the capture process has to be performed at a low velocity, increasing process time but maintaining the capacity high.
Membranes 2020, 10, x 6 of 12 where ,100% ads m is mass of product adsorbed performing the adsorption step until saturation and support V is the volume of the chromatographic support.
However, due to the high product value and the high downstream processing costs, in the biopharmaceutical industry the adsorption step is usually stopped before column saturation, when the solute concentration in the stream exiting the chromatographic column reaches a particular value that usually corresponds to 10% breakthrough. In this case the parameter of interest to be calculated is the dynamic binding capacity at 10% breakthrough (DBC10%), that represents the amount of protein adsorbed per column volume, expressed by the following equation: where ,10% ads m is mass of product adsorbed performing the adsorption step until 10% breakthrough and support V is the volume of the chromatographic media.
The effect of flow rate on the values of dynamic binding capacity at 10% breakthrough is shown in Figure 2 for membranes and monoliths, as well as for packed columns, for two different values of BSA concentration in the feed. The packed column is the one whose performance is mostly and negatively influenced by the feed flow rate, since DBC10% decreases as the flow rate increases, contrary to what happens for the convective materials. This means that in the case of a packed column, the capture process has to be performed at a low velocity, increasing process time but maintaining the capacity high. The performance of the chromatographic materials studied can be also evaluated by plotting the dynamic binding capacity at 10% breakthrough against the residence time. From the plots shown in Figure 3 it can be concluded that membranes and monoliths give the same capacities as the resin but at lower residence time, demonstrating once again that the capture process is faster and more efficient using convective media. The performance of the chromatographic materials studied can be also evaluated by plotting the dynamic binding capacity at 10% breakthrough against the residence time. From the plots shown in Figure 3 it can be concluded that membranes and monoliths give the same capacities as the resin but at lower residence time, demonstrating once again that the capture process is faster and more efficient using convective media.  The influence of flow rate on the shape of the breakthrough curves for all stationary phases studied can be observed in Figure 4. The most visible effect is noticeable for the packed column, Figure 4e,f: this result indicates that the resin is affected by non-negligible diffusional limitations, that influences the performance of the chromatographic material.
The data reported are the result of a single set of experiments. However, many chromatographic cycles were performed for each support, changing the initial BSA concentration and the flow rate; more data can be found in the Supplementary Material, Figures S1 and S2. The results reported here and in the Supplementary Material confirm the characteristic behavior of convective chromatographic materials, in that the performance of convective stationary phases is much less affected by flow rate than packed columns as indicated in the literature [5,6,[31][32][33][34][35][36][37][38][39].  The influence of flow rate on the shape of the breakthrough curves for all stationary phases studied can be observed in Figure 4. The most visible effect is noticeable for the packed column, Figure 4e,f: this result indicates that the resin is affected by non-negligible diffusional limitations, that influences the performance of the chromatographic material.
The data reported are the result of a single set of experiments. However, many chromatographic cycles were performed for each support, changing the initial BSA concentration and the flow rate; more data can be found in the Supplementary Material, Figures S1 and S2. The results reported here and in the Supplementary Material confirm the characteristic behavior of convective chromatographic materials, in that the performance of convective stationary phases is much less affected by flow rate than packed columns as indicated in the literature [5,6,[31][32][33][34][35][36][37][38][39]. The influence of flow rate on the shape of the breakthrough curves for all stationary phases studied can be observed in Figure 4. The most visible effect is noticeable for the packed column, Figure 4e,f: this result indicates that the resin is affected by non-negligible diffusional limitations, that influences the performance of the chromatographic material.
The data reported are the result of a single set of experiments. However, many chromatographic cycles were performed for each support, changing the initial BSA concentration and the flow rate; more data can be found in the Supplementary Material, Figures S1 and S2. The results reported here and in the Supplementary Material confirm the characteristic behavior of convective chromatographic materials, in that the performance of convective stationary phases is much less affected by flow rate than packed columns as indicated in the literature [5,6,[31][32][33][34][35][36][37][38][39].

Productivity
Productivity was used as an alternative and more appropriate way to express the performance of a chromatographic support. In this paper, productivity, at 10% breakthrough, was calculated according to the following equation: For each material studied, productivity was plotted against the superficial velocity, at fixed concentration. The results are reported in Figure 5.

Productivity
Productivity was used as an alternative and more appropriate way to express the performance of a chromatographic support. In this paper, productivity, at 10% breakthrough, was calculated according to the following equation: where m elu is the mass of product eluted from the support, V support is the volume of the chromatographic media and t cycle is the duration of a complete chromatographic cycle. For each material studied, productivity was plotted against the superficial velocity, at fixed concentration. The results are reported in Figure 5. Membranes 2020, 10, x 9 of 12 (a) (b) Figure 5. Comparison of the productivity at 10% breakthrough as a function of superficial velocity for an initial BSA concentration of (a) 0.5 mg/mL and (b) 1.4 mg/mL.
The plots in Figure 5a,b clearly show that a higher productivity can be obtained with membranes and monoliths with respect to the packed column, when working at the same superficial velocity. Remarkably, the increase of productivity with increasing velocity is very limited for the resin if compared to membranes and monoliths.

Conclusions
An experimental study regarding BSA capture was performed through affinity chromatography using different chromatographic media functionalized with Cibacron Blue F3GA ligand. Resins, membranes and monoliths were characterized and their performance compared in terms of binding capacity and productivity. The preliminary results obtained demonstrate that the dynamic binding capacity at 10% breakthrough is independent on flow rate in the case of membranes and monoliths. Thus, convective stationary phases were not affected by kinetic limitations in the range of superficial velocities investigated. Additionally, a strong effect of flow rate was observed for the packed bed column. According to this, the shape of the breakthrough curve was greatly affected by the increase of flow rate in the case of the packed column. At the industrial level, the capture step is usually performed until the concentration of the protein to recover reaches 10% breakthrough in the stream exiting the column. Therefore, according to the preliminary set of experiments performed, membranes and monoliths perform better than the packed column, since they can achieve higher capacities at lower residence times, speeding up the purification process. Similar conclusions are drawn from the results obtained in terms of productivity: at fixed initial BSA concentration and at fixed superficial velocity a productivity from 10 to 20 times higher than the packed column can be achieved using membranes.
The results shown demonstrate that the use of convective stationary phases can reduce the purification process time significantly, thus reducing buffer consumption and avoiding protein degradation. Moreover, the use of convective stationary phases can overcome packing requirements, as well as Clean In Place (CIP) and re-validation procedures.
The interesting flow behavior of convective materials is confirmed by the experimental investigation carried out and the obtained results can be used as a solid base for process scale-up and development. Indeed, these properties explain the enormous interest that the bioprocess industry has for membrane chromatography which enables a fast and efficient purification of large biomolecules such as viral vectors used in gene and cell therapies.
Supplementary Materials: The following are available online at www.mdpi.com/xxx/s1, Figure S1: DBC10% as a function of flow rate at fixed initial BSA concentration for (a) resin, (b) membrane and (c) monolith. Each point in the plots represent a chromatographic experiment. All the data presented were obtained without considering the dispersion contributions, that take into account for the system dead volume; for this reason, the values of  Figure 5. Comparison of the productivity at 10% breakthrough as a function of superficial velocity for an initial BSA concentration of (a) 0.5 mg/mL and (b) 1.4 mg/mL.
The plots in Figure 5a,b clearly show that a higher productivity can be obtained with membranes and monoliths with respect to the packed column, when working at the same superficial velocity. Remarkably, the increase of productivity with increasing velocity is very limited for the resin if compared to membranes and monoliths.

Conclusions
An experimental study regarding BSA capture was performed through affinity chromatography using different chromatographic media functionalized with Cibacron Blue F3GA ligand. Resins, membranes and monoliths were characterized and their performance compared in terms of binding capacity and productivity. The preliminary results obtained demonstrate that the dynamic binding capacity at 10% breakthrough is independent on flow rate in the case of membranes and monoliths. Thus, convective stationary phases were not affected by kinetic limitations in the range of superficial velocities investigated. Additionally, a strong effect of flow rate was observed for the packed bed column. According to this, the shape of the breakthrough curve was greatly affected by the increase of flow rate in the case of the packed column. At the industrial level, the capture step is usually performed until the concentration of the protein to recover reaches 10% breakthrough in the stream exiting the column. Therefore, according to the preliminary set of experiments performed, membranes and monoliths perform better than the packed column, since they can achieve higher capacities at lower residence times, speeding up the purification process. Similar conclusions are drawn from the results obtained in terms of productivity: at fixed initial BSA concentration and at fixed superficial velocity a productivity from 10 to 20 times higher than the packed column can be achieved using membranes.
The results shown demonstrate that the use of convective stationary phases can reduce the purification process time significantly, thus reducing buffer consumption and avoiding protein degradation. Moreover, the use of convective stationary phases can overcome packing requirements, as well as Clean In Place (CIP) and re-validation procedures.
The interesting flow behavior of convective materials is confirmed by the experimental investigation carried out and the obtained results can be used as a solid base for process scale-up and development. Indeed, these properties explain the enormous interest that the bioprocess industry has for membrane chromatography which enables a fast and efficient purification of large biomolecules such as viral vectors used in gene and cell therapies.
Supplementary Materials: The following are available online at http://www.mdpi.com/2077-0375/10/1/1/s1, Figure S1: DBC 10% as a function of flow rate at fixed initial BSA concentration for (a) resin, (b) membrane and (c) monolith. Each point in the plots represent a chromatographic experiment. All the data presented were obtained without considering the dispersion contributions, that take into account for the system dead volume; for this reason, the values of DBC 10% are higher than those reported in the paper (please, refer to Figure

Conflicts of Interest:
The authors declare no conflict of interest.