Continuous purification of Candida antarctica lipase B using 3‐membrane adsorber periodic counter‐current chromatography

Abstract

Batch chromatography has several disadvantages, such as insufficient utilization of the capacity of the resin, high buffer consumption and discontinuity. Considering the high costs for downstream processing, a continuously working chromatographic system with three membrane adsorber units was designed, tested and put into operation. The basic principle of the setup is periodic counter‐current chromatography (PCCC). The PCCC system was used for capturing and purifying Candida antarctica lipase B (CalB) directly from cell lysate in one single unit operation. The best purification result was achieved by means of anion‐exchange chromatography. The dynamic binding capacity with Sartobind® Q 75 amounted to 4.2 mg (56 g/cm²). After transferring the method to the 3MA‐PCCC, 0.22 g CalB (73 U/mg) were obtained from 0.9 L E. coli lysate within 6 h and a recovery of 80%. Compared to the batch process, the productivity could be increased by 36% and the buffer consumption could be reduced by about 20%. Although the purification of CalB from lysate by means of anion‐exchange chromatography was not selective and quantitative using the 3MA‐PCCC device, it could be shown that the concept of the system was successfully implemented and led to a significant improvement of CalB purification.

Abbreviations

A280

absorption at 280 nm

AEX

anion‐exchange chromatography

CalB

Candida antarctica lipase B

CV

column volume

CWW

cell wet weight

HIC

hydrophobic interaction chromatography

MA

membrane adsorber

MW

molecular weight

PCCC

periodic counter‐current chromatography

1. Introduction

Downstream processing plays a key role in providing pure and safe biopharmaceuticals. Therefore, about 80% of the whole process costs have to be invested in the purification of the product 1, 2. In this context, the interest in continuous purification techniques increases because they can improve the productivity. As chromatography can be used for almost every separation task and is often the method of choice, it offers great potential for optimization by continuous operation 3.

Considering the high costs for downstream processing, continuous chromatographic methods, such as periodic counter‐current chromatography (PCCC), could replace conventional batch chromatography 4. Periodic counter‐current chromatography is a pseudo‐continuous chromatographic method that uses a column switching approach.

In PCCC, three or more columns are used to carry out the chromatography continuously 5. Two columns are connected in series during the loading steps, so that the product breakthrough of the first column is guided directly to a second column 5, 6. This approach allows for a loading of the column up to nearly its static binding capacity 5 and results in a higher capacity utilization of the resin 4, 6. In contrast to batch chromatography, the loading can be carried out beyond the dynamic binding capacity 5, 6.

The introduction of continuous chromatography has few disadvantages concerning mostly acquisition and setup, but as shown in Table 1, its advantages compared to batch chromatography prevail. Furthermore, the disadvantageous aspects mainly refer to the non‐recurring costs while significantly reducing the running costs.

Table 1.

Advantages and disadvantages of the PCCC

Advantages	Disadvantages
More effective use of resin 31, 32, less buffer consumption, higher productivity Processing costs savings about 5–30% 26 FDA advocates continuous techniques 33 Smaller facilities, higher flexibility 34	Higher residence times Four times higher costs for acquisition 26 Complex constructions Planning/Handling more complex 35

As an alternative to conventional bed columns, disposable membrane adsorbers (MA) can be used in continuous chromatographic systems to further improve the process. Ligands are attached to the membrane surface so that mass transport is mainly based on convection and not on pore diffusion 7. In addition to high mechanical stability, membrane adsorbers also show a strong salt‐ and pH‐tolerance 8. The capacity of membrane adsorbers is independent of the flow rate 9 and allows for shorter cycle times. Nevertheless, the breakthrough capacities of columns and membrane adsorbers are comparable 10. In addition, up‐ and downscaling the membrane adsorber is much easier, as the capacity only depends on the membrane surface area 11.

PCCC with membrane adsorbers can be used to purify enzymes that are of high biotechnological interest. The enzyme class of hydrolases includes the lipases that have attracted profound scientific interest, as no co‐factor is needed for their enzymatic reaction. Lipases can cleave ester bonds in the presence of water, synthesize ester in the absence of water and are stereospecific. Therefore, lipases are used in the detergent, cleaning, and food industries 12, 13, 14.

Candida antarctica lipase B (CalB) (EC 3.1.1.3) is one of the lipases most widely used in biocatalysis 15. CalB consists of 317 amino acids and has a size of 33 kDa 16. Advantageous characteristics of CalB are e.g. the broad substrate spectrum, the high pH‐ and temperature stability (until 60°C) as well as the activity in aqueous and organic solvents 17. Figure 1 shows the net charge of CalB depending on the pH value. At the isoelectric point (pI), a protein is uncharged. Since CalB has no titratable side chains in the range of pH 5–8 18, the total charge does not change in this pH range, and the net charge is 0 17. As a result, CalB is stable in a pH range of pH 3–10. The theoretical pI value of CalB is pH 6.0 18.

Figure 1.

Net charge of CalB as a function of pH, modified according to 18. CalB is stable from pH 3 to 10 due to the pI range from pH 5 to 8.

CalB is used in the detergent industry 15, in the production of fruit esters or L‐menthol, the synthesis of geranyl acetate (odorous component in food and perfume), as well as the removal of malodor of sheep and goat milk products 19, 20, 21, 22.

Enzymes like CalB are produced recombinantly in microorganisms such as E. coli, Aspergillus niger or Pichia pastoris (new: Komagataella phaffii). In order to use CalB in different industries for various purposes, it must be partially purified. Different chromatographic purification methods are described in the literature 18, 23, 24, 25. In particular, ion‐exchange chromatography is applied to the purification of CalB, which is rather difficult due to the broad range of the pI value. Apart from ion‐exchange chromatography, hydrophobic interaction chromatography (HIC) and biomimetic affinity chromatography (B‐AC) are used. Even though the methods are described in the literature, each method has disadvantages, such as a low capacity of the resin, low recovery or loss of activity.

In this study, the purification of CalB should be improved by using membrane adsorbers combined with a continuous chromatography device. HIC and AEX (anion‐exchange chromatography) membrane adsorbers are tested under different buffer conditions in batch mode and compared to bed column methods 18, 23, 24, 25. To increase the productivity of the process, the established downstream process is further transferred to the continuously operated three membrane adsorber periodic counter‐current chromatography (3MA‐PCCC) device. The batch process is compared with the continuous process with regard to the quality and quantity of the purification, the usage of consumables and the process time.

2. Materials and methods

2.1. Materials

Commercial CalB was purchased from c‐LEcta, Germany, and pre‐filtered before usage with a 0.22 μm filter. Para‐nitrophenyl acetate (p‐NPA) and BSA were provided by Sigma Aldrich, USA, whereas lysozyme was provided by Fluka (Sigma Aldrich), USA.

2.2. Methods

2.2.1. CalB production

CalB was expressed in E. coli Rosetta 2 (DE3) pLysS::pET26B(+). Therefore, the sequence of an adapted Candida antarctica (LF 058) gene was used for lipase B (GenBank Z30645.1). For cloning restriction enzymes and DNA, ligase was chosen. The plasmid was introduced into the bacteria by heat transformation. For the expression, the cryo culture was grown on lysogeny broth(LB)‐agar and incubated for 24 h at 37°C. One colony was picked and the pre‐preculture in terrific broth(TB)‐media (12 g/L tryptone, 24 g/L yeast extract, 4 mL/L glycerol, 100 mL/L of solution of 0.17 M KH₂PO₄ and 0.72 M K₂HPO₄) was inoculated (20 mL in 100 mL shake flask) and cultured overnight at 37°C and 180 rpm. The preculture was inoculated using a pre‐preculture to a starting OD₆₀₀ = 0.2 AU (100 mL in 500 mL shake flask). The preculture was incubated for 2 h at 37°C and 180 rpm in TB media. The main culture was started with OD₆₀₀ = 0.1 AU (400 mL in 2 L shake flask) in TB‐media. The culture was induced after 2 h of cultivation at 37°C and 180 rpm using 0.8 mM IPTG and a temperature shift to 20°C.

Cell lysis was performed four times by sonification for 45 s on ice (amplitude 100%, 100 W, at time intervals of 0.6 s) in equilibration buffer. A centrifugation and sterile filtration step was carried out to yield cell lysate for the purification by chromatography.

2.2.2. CalB batch purification

The purification was developed using a commercial FPLC system. In general, a chromatography run consisted of the following phases: 15 CV (column volume) equilibration, 5 mL sample application, 10 CV wash, 25 CV elution and regeneration. Depending on the experiment, the elution phase was adjusted. For AEX, three buffer conditions were tested: 0.02 M TRIS‐HCl) was inoculated (20 mL in 100 mL shake flask) and cultured overnight at 37°C and pH 8.5 and elution with 1 M NaCl, 0.02 M TRIS‐HCl pH 8.5 and elution at pH 7.0, 0.05 M ammonium acetate pH 8.5 and elution using 1 M NaCl according to Llerena‐Suster et al., 2014 24. For HIC, one buffer was tested according to Ujiie et al., 2015 25: 0.02 M TRIS‐HCl, pH 8.0 + 0.8 M ammonium acetate and elution with decreasing salt concentration. The chromatography was run at room temperature at a flow rate of 5 mL/min. The developed batch method was further transferred to a continuous chromatographic system.

2.2.3. CalB continuous purification

The PCCC device was tested and placed into operation using the model proteins BSA (as the product) and lysozyme (as the impurity), thus providing a simple two component purification task to evaluate the system. The overall loading and elution cycle was developed with an Äkta pure system for the membrane adsorber capsules. The optimal cycle procedure (buffer system, flow rate, duration of binding and elution) was transferred to the PCCC system and controlled by a UV signal.

Both the limit of detection (LOD) and quantification (LOQ) for the PCCC were determined employing the data of the calibration curve and the following equations:

$$\mathrm{LOD}=\frac{3·\mathrm{Standard}\mathrm{deviation}\mathrm{of}\mathrm{the}\mathrm{lowest}\mathrm{concentration}}{\mathrm{Slope}\mathrm{of}\mathrm{calibration}\mathrm{curve}}$$ (1)

$$\mathrm{LOQ}=\frac{10·\mathrm{Standard}\mathrm{deviation}\mathrm{of}\mathrm{the}\mathrm{lowest}\mathrm{concentration}}{\mathrm{Slope}\mathrm{of}\mathrm{calibration}\mathrm{curve}}$$ (2)

The basic PCCC cycle is presented in Fig. 3. In the first step, membrane adsorber 1 is loaded until a certain amount of the product breaks through. The process switches to step two, and membrane adsorber 2 is added to the loading. The breakthrough of membrane adsorber 1 is bound to membrane adsorber 2. Step two stops when membrane adsorber 1 is saturated with product 4, 5. The membrane adsorber is decoupled from the circuit, the product is eluted separately and the adsorber is regenerated and coupled back into the circuit. The described procedure for membrane adsorber 1 is continued using membrane adsorber 2 and 3 6. By parallel washing, elution and regeneration, after step six, the process can start again at step one. To fulfill this cycle, the decoupling time for regeneration and elution of one membrane adsorber must be less than or equal to the loading time 4.

Figure 3.

Schematic diagram of a continuous 3MA‐PCCC cycle. The three membrane adsorbers are loaded, washed, eluted and regenerated following the principle of periodic counter‐current chromatography. The first membrane adsorber is loaded until a certain amount of product breaks through. Then the second membrane adsorber is added to the loading (loading zone, light blue) until saturation of membrane adsorber 1. Membrane adsorber 2 is loaded according to the same principle. Meanwhile membrane adsorber 1 is regenerated (regeneration zone, gray).

To switch from one step to another, two automated switching events are required for the PCCC 5:

• The first switching condition (SC1) is the product breakthrough with a predefined loss (defined as the dynamic binding capacity).

• The second switching condition (SC2) indicates the defined saturation of the membrane adsorber 4, 5.

Figure 4 shows the double breakthrough curve with the switching events of the simple two‐component purification task using the PCCC device. This switching approach is based upon the UV absorption difference (ΔUV) of the solution to be purified and the membrane adsorber outlet (measured UV signal) 4, 5. UV_max is the UV signal of the unpurified solution of BSA and lysozyme, and UV_Lysozyme is the UV signal of the non‐binding lysozyme in the solution. The double breakthrough curve was recorded by feeding the solution onto one membrane adsorber. The non‐binding lysozyme immediately breaks through, and the characteristic plateau of the impurity is visible (Fig. 4). Gradually, the membrane adsorber is saturated with product so that a product breakthrough can be measured at the outlet of the membrane adsorber.

Figure 4.

Double breakthrough curve for the determination of the switching conditions with the 3MA‐PCCC. The switching conditions 1 and 2 can be determined using the UV signal of the impurity (UV_Lysozyme) and the UV signal of the solution to be purified (UV_max). The difference of UV_max and UV_Lysozyme is ΔUV. Here SC1 is 10% of ΔUV and SC2 is 70% of ΔUV.

According to Warikoo et al. 5, SC2 was set to 70% in the three‐module system. An SC1 at 10% is regarded as a standard value for dynamic binding capacity‐controlled processing. To determine these two switching events, the solution of BSA and lysozyme was directly injected into the cuvette and UV_max was measured. UV_Lysozyme was determined by recording the double breakthrough curve of the solution to be purified. Subsequently, SC1 and SC2 could be calculated using the following equations.

$$\mathrm{SC}1=0.1·\mathrm{\Delta }\mathrm{UV}+\mathrm{U}{\mathrm{V}}_{\mathrm{Lysozyme}}$$ (3)

$$\mathrm{SC}2=0.7·\mathrm{\Delta }\mathrm{UV}+\mathrm{U}{\mathrm{V}}_{\mathrm{Lysozyme}}$$ (4)

Depending on the product, SC1 and SC2 can be customized. To fulfill the principle of the PCCC, SC2 should not lead to a product breakthrough on the second membrane adsorber (see Fig. 3 step 2).

2.2.4. Analytical methods

2.2.4.1. Determination of protein content using the Bradford assay

To determine the total protein concentration in an aqueous solution, the colorimetric Bradford method was applied. For concentration determination, 20 μL of the sample were mixed with 300 μL of the Bradford reagent (Quick Start™ Bradford 1× DyeReagent, Bio‐Rad) in a microtiter plate, shaken for 30 s and incubated for 5 min. At 595 nm, the samples were measured. BSA and CalB standards of 0.0125–0.2 g/L were used for calibration.

2.2.4.2. Qualitative analysis using SDS‐PAGE with silver staining

SDS‐PAGE was used for qualitative evaluation. The gel consisted of a collecting gel (6%) and a separating gel (12%). The samples were mixed with Lämmli buffer in the ratio 1:1 and boiled for 10 min at 95°C. 3 to 10 μL of the sample were used for the gel compared to 5 μL of the marker (Unstained Protein Molecular Weight Marker, Thermo Fisher Scientific). The SDS‐PAGE was run for 15 min at 100 V and then for 45–60 min at 150 V in TGS buffer. To visualize the protein bands, silver staining was used.

2.2.4.3. Determination of CalB activity

The activity assay with para‐nitrophenyl acetate (p‐NPA) was used to determine the activity of Candida antarctica lipase B. p‐NPA is hydrolyzed to para‐nitrophenol and acetate. The reaction can be measured photometrically at 405 nm (para‐nitrophenol). The assay was performed in 50 mM phosphate buffer at pH 7.4 and room temperature. The molar extinction coefficient of 18,450 M⁻¹cm⁻¹ for para‐nitrophenol was applied.

2.3. Equipment

2.3.1. Membrane adsorbers

Two types of membrane adsorber units, Sartobind® Q 75 (AEX) and Sartobind® Phenyl nano 3 mL (HIC) from Sartorius Stedim Biotech, Germany, were employed for the experiments in this study.

2.3.2. Äkta pure

For method development, the chromatographic system Äkta pure (GE Healthcare) was used in combination with membrane adsorbers.

2.3.3. 3MA‐PCCC setup

For the continuous purification of CalB, a three‐membrane adsorber periodic counter‐current chromatographic device was used. This PCCC device was designed and built at the Institute of Technical Chemistry, Hanover. The circuit diagram can be seen in Fig. 2.

Figure 2.

Circuit diagram of the PCCC device with three membrane adsorbers. The system consists of a peristaltic pump, 21 3/2‐way valves, 3 membrane adsorbers and 3 UV measurement units. The valves E1 ‐ E6 are used for buffer A and B, the valves F1 ‐ F8 for the feed solution and M1 ‐ M7 to connect the membrane adsorbers according to the principle of the PCCC.

The PCCC device consists of a peristaltic pump (Type Pump Reglo ICC, Ismatec, Germany) with a pump rate of up to 8.6 mL/min, depending on the pump tubing. The channels are used for equilibration buffer/buffer A (A1), elution buffer/buffer B (B1) and feed (L). The mixing chamber (Type Dynamic Mixer GT0387, Hitachi, Japan) can be applied to realize different elution strategies, such as step elution or linear gradient elution. The interconnection of 21 magnetic 3‐2‐way valves (E1–E6, F1–F8, M1–M7) and three membrane adsorbers (MA1, MA2, MA3) is required to realize the principle of a periodic counter‐current chromatography. Valves (The Lee company, USA) designated with E are exclusively used for buffer A or B, valves with F for feed solution and valves with M are necessary to connect the membrane adsorbers. Further components are three UV flow through cuvettes with a 10 mm path length (Hellma Analytics, Germany), three UV spectrometers (Ocean Optics, USA) (UV1–UV3, 190–650 nm), a fraction collector (Type Model 2110, BioRad, USA) and a light source (Type DH‐2000‐BAL, Ocean Optics, USA) for wavelengths between 230 and 2500 nm. For the experiments, a wavelength of 280 nm was applied for protein detection.

3. Results

3.1. Purification of Candida antarctica lipase B

The purification of CalB is problematic due to the wide pI range, so that no optimal strategy has yet been developed with conventional bed columns. Mainly, ion‐exchange chromatography has been used, which requires the CalB to be charged. Due to the pI range of CalB, it must be operated under pH 5 (positively charged) or above pH 8 (negatively charged). Additionally, CalB should not be affected in its activity in these pH ranges.

In this study, two membrane adsorbers were tested: one HIC MA (Sartobind® Phenyl) and one AEX MA (Sartobind® Q). AEX was run under three different buffer conditions: an ammonium acetate buffer, such as published by Llerena‐Suster et al., 2014 24 and two TRIS buffers with pH elution and salt elution that have not as yet been published. HIC was performed using a TRIS buffer, as described in Ujiie et al., 2015 25. The method development, including the binding experiments (using linear elution) and optimization (with elution steps) of the method were carried out by means of commercial CalB. This allowed for a statement regarding recovery and binding capacity. The commercial CalB was not pure and had a high viscosity, so that the dosage did vary during the experiments for method development.

3.2. Downstream processing for commercial CalB

3.2.1. Comparison of methods

Comparing the methods using HIC‐ and AEX membrane adsorbers and different buffer systems (data not shown), the best result could be achieved using the AEX method and TRIS buffer salt elution. This method had a recovery rate of more than 99% when employing commercial CalB, where the specific activity of CalB in the buffer system was 73 U/mg. Under these conditions, CalB was charged negatively.

The AEX method with pH elution showed comparable results but the activity decreased during elution. The binding experiment for AEX with ammonium acetate buffer did not indicate any binding of CalB, while HIC produced a recovery of only 56%.

3.3. Optimization and binding capacity using commercial CalB

The filtered 0.3 g/L CalB solution (254 U/mg) was placed onto the membrane adsorber, and an elution step of 20% buffer B was added to separate CalB from impurities. The method included further elution steps at 50 and 100% buffer B. This modification (Fig. 5A) resulted in a high peak of CalB at 20% buffer B due to a weak binding to the adsorber. The peak was very pure (Fig. 5B) and showed a high specific activity of 217 U/mg. Impurities were eluted at 50% buffer B. The recovery of this method was higher than 99%.

Figure 5.

(A) Chromatogram of the binding experiment using AEX‐MA–step elution using 20, 50, and 100% buffer B. (B) SDS‐PAGE of the binding experiment using AEX‐MA.

To determine the capacity, a breakthrough curve with 0.11 g/L CalB solution was recorded. The dynamic binding capacity at 10% (DB₁₀) was 4.2 mg (56 μg/cm²) and when compared to the manufacturer's information, with 60 mg BSA per unit, it was quite low. The static binding capacity (DB₅₀) was 6.3 mg (84 μg/cm²) and was 0.5 times higher than the dynamic binding capacity. Compared to the published capacity of CalB in an affinity column 23, the dynamic binding capacity is 5‐fold and the static binding capacity is 7.5‐fold higher, even with this far less specific interaction.

3.4. Purification of CalB from E. coli lysate

The procedure described in 3.3 for the purification of the CalB was applied to E. coli lysate containing 1.7 g/L total protein and 55 U/mg specific activity of CalB. This lysate included host cell proteins and possibly DNA (not tested). The result of the chromatography and the SDS‐PAGE are presented in Fig. 6. The breakthrough was not analyzed as the applied amount of CalB was below the dynamic binding capacity at 10% and no product breakthrough was expected. At 20% buffer B, CalB eluted in a sharp peak (Peak 1). On the gel it is evident that CalB fractions are highly pure when compared to cell lysate. Only within the first fraction other bands are visible. In addition, a high specific activity (223 U/mg) could be measured in the fractions of peak 1. The recovery of CalB was 97%. Peaks 2 and 3 did not contain any CalB, however impurities (host cell proteins) are visible in the cell lysate. As the peaks of the elution steps of 50% buffer B and 100% buffer B contained impurities, but no CalB, the method was simplified for the PCCC. The elution step at 50% buffer B was eliminated, so that the method consists of two steps: a 1^st elution step of CalB at 20% buffer B and a 2^nd step for the elution of all impurities at 100% buffer B. This promising method for capture and purification of CalB had a purification factor between 1.3 and 4.0, and was transferred to the 3MA‐PCCC device.

Figure 6.

(A) Chromatogram of the method for the purification of CalB in lysate with elution at 20, 50, and 100% buffer B: Sartobind® Q with TRIS‐HCl pH 8.5 at 5 mL/min. (B) SDS‐PAGE of the method development for the purification of CalB from E. coli lysate.

3.4.1. Preparations for PCCC

The developed optimal downstream procedure with step elution at 20% buffer B for CalB and 100% buffer B for impurities was transferred to the PCCC device. The Äkta pure system and the PCCC device have a major difference in the detection system: the Äkta system has a 5‐fold smaller path length (2 mm) compared to the PCCC system (10 mm). Since the PCCC system can measure up to 1.5 AU (linear absorption range with a 10 mm cuvette), the lysate was diluted.

To measure CalB in the linear range, a calibration curve was recorded in the range of 0.0625–1.5 g/L commercial CalB. All measuring points were in the linear measuring range below 1.5 AU. As commercial CalB is not highly purified, deviations occurred between the theoretical and practical values. In addition, the buffer system (TRIS buffer) may have disturbed the measurement. The following limits of the measurement were determined from the calibration curve: The LOD for CalB was 0.04 g/L, and the LOQ was 0.13 g/L.

The feed solution for continuous purification consisted of E. coli lysate of 0.15 g cell wet weight (CWW)/L containing 0.37 g/L total protein in 20 mM TRIS‐HCl buffer of pH 8.5. To determine the switching conditions for the PCCC, UV_max was determined in the cuvette and the double breakthrough curve for the determination of UV_Impurity was recorded (see Fig. 7). UV_max was 0.620 AU and UV_Impurity was 0.070 AU. As determined in the PCCC setup, SC1 was 10% (0.116 AU) and SC2 was 70% (0.452 AU). SC1 was above the detection limit (data not shown) and large enough to be measured.

Figure 7.

Double breakthrough curve for the determination of UV_Impurity for CalB purification. UV_max was measured in the cuvette.

3.4.2. PCCC using E. coli lysate containing CalB

The PCCC cycle (Fig. 3) was adjusted by setting the switching conditions from 3.4.1. Due to the low binding capacity of CalB to the membrane adsorber, the loading pump rate was reduced to 2.5 mL/min, whereas the regeneration pump rate was increased to 8.6 mL/min. Furthermore, this adaptation guaranteed that the loading lasted longer than the regeneration of the membrane adsorber to enable continuous processing. The capacity of the membrane adsorber did not change as a result of the pump rate. The purification of 0.9 L feed of E. coli lysate using 0.15 g CWW/L containing 0.37 g/L total protein took 6 h and needed 3 L buffer (A and B). 0.22 g CalB was achieved over 5 PCCC‐cycles with three membrane adsorbers before a visible loss of capacity occurred which might be due to unspecific binding of host cell proteins. The elution was divided into two phases during the purification of CalB: product elution at 20% buffer B, elution of the impurities at 100% buffer B. The result of the PCCC run is shown in Fig. 8A.

Figure 8.

(A) PCCC for the purification of CalB with five cycles (impurity breakthrough, breakthrough of product and other binding components, elution of membrane adsorber 1–3 in two steps: 20% buffer B, 100% buffer B). (B) SDS‐PAGE of cycle 1 and 5 of the PCCC run for the purification of CalB (W = Wash; E20 = Elution of CalB at 20% buffer B; E100 = Elution of impurities at 100% buffer B).

The typical double breakthrough curve of the PCCC is more difficult to see in this experiment since the impurity plateau is substantially flatter (0.07 AU). Although the concentration of the lysate was adapted to the PCCC, proteins which bind to the membrane adsorber were present in the lysate. This nonselective purification of CalB is not optimal for the application of the PCCC, but could nevertheless be operated over five cycles. Up to cycle 4, the double breakthrough can be seen with all membrane adsorbers. In cycle 5, the UV signal of membrane adsorber 3 is widened and irregular. The breakthrough curves were broadened during the cycles of the PCCC and the loading time until SC2 increased, while the subsequent membrane adsorber was loaded with the breakthrough. As a result, SC2 led to a product breakthrough on the following membrane adsorber, and the time window for the elution was reduced, which only led to the 100% buffer B elution in the last cycle. Cycles 1 and 5 were sampled. The samples from the phases of washing, elution at 20% buffer B and elution at 100% buffer B were each pooled in a uniform proportion. The pooled samples of the elution which peaked at 20% buffer B had protein concentration values of 0.24–0.31 g/L and showed an increased specific activity compared to the applied cell lysate (Table 2).

Table 2.

Bradford analytics and activity of the cell lysate and product peaks at 20% buffer B in cycle 1 (C1) and 5 (C5)

Sample	Protein concentration [g/L]	Activity [U/mL]	Specific activity [U/mg]
Cell lysate	0.37	15.9	43
MA1 C1	n.d.a	n.d.	–
MA2 C1	0.27	19.7	73
MA3 C1	0.30	18.5	62
MA1 C5	0.24	14.5	60
MA2 C5	0.31	17.7	57
MA3 C5	n.d.	n.d.	–

^an.d.: not determined (technical problems).

To evaluate the quality of the purification, an SDS‐PAGE was performed (Fig. 8B). The fractions of the wash steps (W) and the elution steps (E20 or E100) can be seen in the gels. For membrane adsorber 1, the samples of the wash step and the elution are missing at 20% buffer B due to technical problems of the fraction collector. In cycles 1 and 5, thick bands at the size of CalB can be seen during the elution at 20% buffer B. In cycle 1, only impurities and no product eluted at 100% buffer B. In cycle 5, the fractions of the step elution at 100% buffer B contained higher amounts of impurities and had bands at the level of CalB. During the washing step, a product loss occurred, as the protein solution in the tubes, pipes and valves was pumped into the waste. To prevent this loss, the wash step could be applied to a regenerated membrane adsorber (white box in Fig. 3). This complement is called an interconnected wash step 26, 27 and requires a modification of the PCCC device with more valves.

Both gels show that CalB could be eluted highly purified. The CalB concentrations in the eluate were above 1 g/L (calibration curve not shown). The yield was 0.22 g CalB and corresponds to a recovery of 80%. The quality of chromatography decreased during the PCCC, which is clearly evident in the chromatogram and in the SDS‐PAGE.

4. Discussion

In this study, a method for the purification of Candida antarctica lipase B (CalB) was developed using the AEX membrane adsorber Sartobind® Q 75. The buffer of choice was 20 mM TRIS‐HCl, pH 8.5 and the elution of both CalB and bound impurities was carried out using 0.2 M NaCl (20% buffer B) and 1 M NaCl (100% buffer B), respectively. The method was developed by applying commercial CalB. The dynamic binding capacity reached 56 μg/cm² commercial CalB and the static binding capacity 84 μg/cm² commercial CalB when the membrane adsorber Sartobind® Q 75 was used. Finally, the purification was performed using E. coli lysate. The purification of CalB from lysate led to a high specific activity of 223 U/mg (where lysate with host cell proteins were 55 U/mg) and a high recovery of 97%. The CalB fractions thus showed a high purity. As shown in Table 3, the developed method for CalB had very good characteristics compared to the published methods with conventional bed columns: high recovery, high activity and high purification factor. The purity could not be determined as a numerical value due to the silver staining, but was visually evaluated to be high.

Table 3.

Comparison of methods for the purification of CalB from the literature and the own method

Method	Recovery	Activity	Purity	Capacity	Purification factor
CEX 18	‐	‐	‐	‐	1.5
B‐AC 23	73%	‐	91%	0.4 mg/mL CV	‐
AEX 24	47%	18.8 U	‐	‐	1.38
HIC 25	23%	90 U	‐	‐	1.29
Own method	97%	223 U/mg or 91 U/mL, activity of E. coli lysate: 55 U/mg or 93 U/mL	high	2 mg/mL CV	Up to 4.0

This promising method for capture and purification of CalB was transferred to the 3MA‐PCCC device. Within the scope of the PCCC, 0.9 L feed of E. coli lysate containing 0.15 g CWW/L and 0.37 g/L total protein was purified within 6 h. CalB could be eluted actively, in high purity (visually evaluated) and with a concentration of up to 1 g/L. The yield was 0.22 g with a recovery of 80%. Using the interconnected washing step, another 10% CalB could be recovered.

The comparison to the batch chromatography is shown in Table 4. In contrast to batch operation, the use of the PCCC led to a productivity increase of 36%¹. In addition, about 20% of the buffers were saved. This productivity increase cannot be assumed for every purification task. But for these experiments, the productivity increase of 36% is attributed to the loading principle of the PCCC (see Fig. 3) and the usage of membrane adsorbers. As the membrane adsorber is loaded to 70% instead of 10% of the product breakthrough, the capacity of the membrane adsorber is more efficiently utilized. In addition, the loading of the membrane adsorber is combined, so that the product breakthrough is captured on another membrane adsorber. The binding capacity of membrane adsorbers is independent of the flow rate 9. Hence the loading pump rate could easily be adjusted to fulfill the PCCC principle. This is a major advantage compared to conventional bed columns as here the flow rate strongly affects the capacity due to pore diffusion.

Table 4.

Comparison of PCCC and batch for purification of CalB

Batch	PCCC
Yield: 11.25 mg CalB in 0.44 h Consumption of 150 mL buffer Recovery of CalB 87% (3% loss of method + 10% dynamic binding capacity) Specific activity of cell lysate: 55 U/mg Specific activity of purified CalB: 223 U/mg	Yield: 0.22 g CalB in 6 h Consumption of 3 L buffer Recovery of CalB 80% (90% with interconnected wash step) Specific activity of cell lysate: 43 U/mg Highest specific activity of purified CalB: 73 U/mg

Since the method for the purification of CalB from lysate also binds impurities, elution was carried out in two steps using 20 and 100% buffer B, respectively. This is not optimal for the use of the PCCC, as the binding impurities lead to a gradual decline of the process performance. The quality of the chromatography was influenced because the phases of the method overlapped after 5 cycles. Considering that the lysate contained binding host cell proteins and possibly DNA, a decrease in the performance was predictable. An additional CIP (cleaning in place) step and an intense back‐flushing of the whole system would be useful.

The method for the purification of CalB was proved to be feasible in the PCCC, but under the restriction that the number of cycles was limited due to a deterioration in the performance. The nonselective binding of CalB was not optimal for the application of the PCCC. Therefore, the removal of binding impurities should be considered in a previous step, or a chromatographic method which exclusively binds CalB should be taken into account.

In the PCCC washing steps, 10–15% of the product was lost. This problem has already been described in the literature 26, 27 and can be solved by changing the system. By passing the wash step onto a further membrane adsorber, the product is not lost. For this modification, additional valves and tubings are necessary and are highly recommended.

The membrane adsorbers were loaded up to five times in the experiments. The manufacturer claims that 100 loadings are possible, before a regeneration step should be carried out 28. This statement is most likely related to the batch mode (about 10% product breakthrough). In the conducted experiments, the membrane adsorbers were loaded up to a value of 70%. Therefore, it is questionable to what extent the manufacturer's data apply to the use in a PCCC system.

The PCCC device should be operated with specifically binding products such as antibodies or His‐tagged proteins (affinity chromatography). The purification of monoclonal antibodies is almost exclusively reported in the literature for PCCC application 4, 26, 29, 30. Secreted products could be purified more easily using the PCCC, as extracellular products in a supernatant do not contain host cell proteins or DNA, which influence the performance of the chromatography. In any case, the product has to be detected with sufficient accuracy to fulfill the principle of the PCCC. The long‐term goal is to integrate the system into a continuous manufacturing process.

Even though small additions to the system are necessary to achieve an optimal result, the periodic counter‐current chromatography device was successfully put into operation and further used for the continuous purification of Candida antarctica lipase B. It could be confirmed that the productivity was increased compared to batch chromatography for this application.

Practical application

This study presents the establishment of a self‐built continuous chromatographic device. The system is operated using three membrane adsorbers according to the principle of periodic counter‐current chromatography. The application of this continuous chromatographic method increases the productivity of a purification process by saving consumables, time, and by better utilizing the capacity of membrane adsorbers. The productivity of the purification of Candida antarctica lipase B could be enhanced by as much as 36%. In order to evaluate the potential of this continuous chromatographic device, it was operated using affinity chromatography interactions (His‐tagged proteins, antibodies). The application of periodic counter‐current chromatography is particularly reasonable for very expensive products and resins, such as antibody purification, as its usage may result in significant cost savings for the purification process.

The authors have declared no conflict of interest.