Alpha-fetoprotein is a prospective biopharmaceutical candidate currently undergoing advanced-stage clinical trials for autoimmune indications. The high insoluble AFP expression yields in E. coli renders the inclusion body route potentially advantageous for process-scale commercial manufacture, if high throughput refolding can be achieved. This study reports the successful development of an 'anion exchange chromatography'-based refolding process for recombinant human AFP (rhAFP), which carries the challenges of contaminant spectrum and molecule complexity. rhAFP was readily refolded on-column at rhAFP concentrations unachievable with dilution refolding due to viscosity and solubility constraints. DEAE-FF functioned as a refolding enhancer to achieve rhAFP refolding yield of 20% and product purity of 95% in 3 h, at 1 mg/ml protein refolding concentration. Optimization of both refolding and chromatography column operation parameters (i.e. resin chemistry, column geometry, redox potential and feed conditioning) significantly improved rhAFP refolding efficiency. Compared to dilution refolding, on-column rhAFP refolding productivity was 6-fold higher, while that of off-column refolding was an order of magnitude higher. Successful demonstration that a simple anion exchange column can, in a single step, readily refold and purify semi-crude rhAFP comprising 16 disulfide bonds, will certainly extend the application of column refolding to a myriad of complex industrial IB proteins.
Keywords: refolding; disulfide bond; alpha-fetoprotein; inclusion body; ion-exchange; chromatography
The remarkable growth in the global market for biopharmaceuticals, currently estimated to be worth around USD 50 billion , reflects an increasing demand for first- and second-generation protein pharmaceuticals. Their effective treatment of many diseases and their role as the only approved therapy for some diseases, make these protein-based drugs indispensable healthcare essentials today. However, the widespread use of biopharmaceuticals across all patient segments continues to be hindered by the high cost of producing these molecules, compared to small molecule drug or traditional medicine. Most protein pharmaceuticals are produced recombinantly, and a huge portion of the production cost is attributed to product recovery costs. In addition, recombinant production of proteins using Escherichia coli (E. coli) often results in insoluble expression of the target protein [2,3], thus necessitating an additional in vitro refolding step to recover the protein in a soluble and biologically active form. The frequently low refolding efficiency coupled with poor product recovery inevitably translates into a high cost-of-good metric. Therefore, biopharmaceutical companies today are seeking more streamlined manufacturing routes to enhance product recovery and reduce the costs of protein drugs.
In the decade to come, the bacterium, E. coli is expected to remain the favored expression host for recombinant protein production due to cost advantages and ease of host manipulation. Production of recombinant proteins in the form of insoluble proteins, or 'inclusion bodies' (IBs) is advantageous with respect to upstream yield and protection from product degradation , but a high refolding yield to convert the insoluble proteins to biologically active proteins is critical to maintain a high overall product recovery . Although simple and easy to perform, refolding by batch dilution necessitates refolding to be conducted at extremely low protein concentrations to minimize competing reactions between first order refolding and higher order aggregation. Furthermore, as protein refolding yields often vary in a contaminant-dependent fashion, the need for pre- and post-purification steps increases the number of unit operations in a process and is prohibitively expensive to scale-up for industrial applications. In recent years, refolding on chromatographic matrices has attracted industrial interest due to the ability to perform refolding at much higher protein concentrations than dilution [6-8]. Chromatography column scalability from lab to industrial stages has also been well-documented, rendering it attractive as an enabling technology to intensify protein pharmaceutical manufacturing. Survey of recent literature on column refolding, however, shows that column chromatography refolding systems have been widely tested on pure protein systems, where high refolding yields (i.e. >50%) have been reported (Table 1). The use of commercially purchased pure proteins eliminates 'cellular contaminant'-related bottlenecks that could affect protein refolding behavior on-column. To truly verify the potential of packed chromatography columns for protein refolding and demonstrate its relevance in bioprocessing of biopharmaceuticals, the use of 'host cell'-derived protein extracts as model proteins is necessary. To date, only a very small number of IB proteins have been successfully renatured on-column at good yields and at high protein loads.
This study therefore aims to investigate the effectiveness of column refolding to refold an IB-derived protein, alpha-fetoprotein (AFP), which was directly extracted from the E. coli host. AFP makes an interesting model from a molecular perspective because it has 32 cysteines, which will oxidize to form 16 disulfide bonds during refolding. The anticipated challenges relating to competitive contaminant binding coupled with product molecular complexity (i.e. immobilization of a highly disulfide bonded protein on a chromatography matrix may constrain free disulfide shuffling and hence impede refolding) merits further research to overcome these limitations and increase column refolding yields, as will be researched in this study. Successful refolding of the matrix-immobilized high-cysteine AFP will extend the application of column-refolding to other complex proteins which suffer from poor refolding efficiency. Importantly, AFP is a potential biopharmaceutical candidate for the treatment of autoimmune indications, making it a commercial valuable protein [9-14]. Transgenically-derived recombinant AFP has recently successfully completed a Phase Two clinical trial study for rheumatoid arthritis indications . The launch of this protein on market shelves in the near future will subsequently demand cheaper second-generation product when product patent expires, thus necessitating new processes that can reduce product cost. The use of an inclusion body route to produce AFP has been investigated in earlier studies [16,17]. The outcome of those studies demonstrated evidence of a potentially feasible E. coli process for preparative to large-scale production of recombinant human AFP (rhAFP), if overall process productivity can be improved. Our early study showed that the best rhAFP refolding yield achieved by batch dilution was 40% at 0.04 mg/ml protein concentration . Refolding at low protein concentration was necessary to keep aggregation low. However, the low protein concentration post-refolding necessitates an additional chromatography purification step to concentrate and purify refolded rhAFP from soluble aggregates, thus decreasing AFP recovery and overall process productivity . These preliminary results indicate the potential for further process improvement to favor higher refolding and overall process productivities. Column refolding is therefore investigated to increase refolding throughput in this study, the outcome of which would provide an indication of the feasibility of a 'column refolding'-based process for rhAFP production.
In this study, the development of an ion-exchange based matrix refolding strategy for rhAFP was investigated. The dependence of rhAFP refolding efficiency on important chromatography operating parameters such as feed conditioning, column geometry, resin chemistry and refolding redox potential was studied. With adequate optimization of the aforementioned parameters, rhAFP was successfully refolded on both Q-FF and DEAE-FF chromatography column matrices at higher protein concentrations than those allowable by dilution refolding (i.e. > 0.5 mg/ml rhAFP). Immobilisation of rhAFP on a solid matrix enhanced productivity by reducing the time required to attain steady state refolding equilibrium compared to dilution refolding. This behavior was especially evident in DEAE-FF, which appeared to assume the role of a refolding enhancer to improve rhAFP refolding efficiency. Our results clearly demonstrate that adsorptive refolding is not limited to pure and/or simple proteins, but is also readily applicable for high-throughput refolding of IB-derived proteins with complex molecular structures.
Recombinant E. coli strain BL21(DE3)RIL containing plasmid pET24D engineered to express human AFP was constructed by the Protein Expression Facility, University of Queensland, Australia. Standard AFP (std-AFP) (i.e. glycosylated AFP derived from human amniotic fluid) was purchased in a lyophilized form (purity >96%) from Fitzgerald Industries International (Concord, MA, USA). Kanamycin, isopropylthio-β-D-galactoside (IPTG), Laemmli sample buffer (2-concentrate), acetonitrile (HPLC grade), polyethyleneimine (PEI), urea, sodium chloride (NaCl), L-arginine, tris(hydroxymethyl)aminomethane (Tris), ethylenediaminetetraacetic acid (EDTA), dithiothreitol (DTT), and reduced (GSH) and oxidized glutathione (GSSG) were purchased from Sigma-Aldrich. Amberlite, HiTrap Q Sepharose Fast-Flow ion exchange columns (Q-FF) (1 ml and 5 ml), HiTrap DEAE Sepharose Fast-Flow ion exchange columns (DEAE-FF) (1 ml and 5 ml), and PD-10 desalting columns containing Sephadex G-25 medium were purchased from GE Healthcare.
To reduce the occurrence of protein carbamylation, all urea solutions used in this study were treated with Amberlite. Amberlite (1% (w/v)) was added into 9 M stock urea solutions in distilled H2O and incubated for 1 h at room temperature on the shaker, after which the resin was removed by filtration (0.45 µm, Millipore, Singapore). Tris, EDTA, GSH, GSSG, NaCl, L-arginine and DTT stock solutions were added to the 9 M stock urea solution to prepare all buffers used for refolding experiments.
2.1. rhAFP expression
Recombinant E. coli strain BL21(DE3)RIL harbouring the pET24D-rhAFP plasmid was grown in a 2 L flask containing 500 ml 2xYT broth with 50 mg/ml kanamycin. The cells were grown at 37 °C, and incubated under 200 rpm shaking conditions for 1 to 2 h until an optical density at 600 nm (OD600) of 1.0±0.1, measured using a spectrophotometer (Eppendorf BioPhotometer), was achieved. The cell culture was then induced with 0.4 mM IPTG, and was further incubated for 2 h until an OD600 2.0±0.1 was obtained. Cells were harvested by centrifugation (4000g, 10 min) and rhAFP expression was analyzed by SDS-PAGE. The cell pellets post-centrifugation were washed in phosphate-buffered saline (PBS, pH 7.2) and centrifuged again (4000g, 4 °C, 20 min) before sonication or chemical extraction for product release.
2.2. Product release and solubilization
2.2.1 Mechanical disruption (MD)
Cell pellet from a 250 ml cell culture was re-suspended in 50 ml Tris-HCl buffer (50 mM, pH 8.0). Mechanical disruption to release intracellular proteins was conducted using a sonicator (Branson® digital sonifier, Danbury, CT, USA). Sonication of the re-suspended cell pellets was conducted on ice bath for 180 s. The disrupted cells were centrifuged (10,000g, 4 °C, 20 min), and the pellet were recovered and washed in Tris-HCl buffer (50 mM, pH 8.0). The washed pellet was solublized in 7.5 to 50 ml Denaturation Buffer (8 M urea, 20 mM Tris, 3 mM EDTA, 20 mM DTT, pH 8.5) to achieve different final rhAFP protein concentrations (0.2 to 2 mg/ml, respectively), and incubated under gentle shaking conditions for 5 h, at room temperature (21 °C). A subsequent centrifugation step (10,000 g, 4 °C, 20 min) was conducted to remove insoluble contaminants. The supernatant fraction following centrifugation was analyzed by reversed phase high performance liquid chromatography (RP-HPLC) and a chip-based electrophoresis method performed on an Agilent 2100 Bioanalyzer® (Agilent) to determine protein concentration and purity for protein refolding experiments.
2.2.2. Chemical extraction (CE)
A cell pellet from a 250 ml cell culture was re-suspended in 25 ml Denaturation Buffer. Extraction was conducted for 2 h at room temperature under gentle shaking conditions. This extract mixture was treated with 0.03% (w/v) PEI for DNA removal. The PEI-treated mixture was centrifuged (10,000g, 4 °C, 20 min) and the supernatant was analyzed on the Bioanalyzer to determine protein concentration and purity.
2.3. Determination of DNA concentration
Concentration of DNA in the protein samples were determined by absorbance measurement at 260 nm and 280 nm after MD or CE. 50 µl protein samples were added into a 50 µl cuvette with a 1 cm path length, and absorbance measurement of the protein samples was conducted using a spectrophotometer (Eppendorf BioPhotometer). An absorbance value of 1.0 at 260 nm corresponds to 50 μg/ml of double stranded DNA and 33 μg/ml single stranded DNA, with the use of a 1 cm path length cuvette .
2.4. Refolding by batch dilution
DTT was removed from the re-solubilized protein mixture after MD using a PD-10 desalting column. A PD-10 column was first equilibrated in 25 ml Equilibration Buffer (8 M urea, 20 mM Tris, 3 mM EDTA, pH 8.5). 2.5 ml of the protein mixture was loaded onto the PD-10 column and 3.5 ml Equilibration Buffer was used to elute protein mixture. The DTT-free protein mixture was immediately diluted into the Refolding Buffer and incubated under gentle shaking conditions at 4 °C for 6 h. The final composition of Refolding Buffer was 1.6 to 3 M urea, 20 mM Tris, 3 mM EDTA, 2.7 mM GSH and 2.7 mM GSSG, pH 8.5.
2.5. Refolding by anion-exchange chromatography
All ion exchange (IEX) chromatography experiments were performed on an ÄKTA explorer workstation (GE Healthcare) at room temperature.
2.5.1. On-column refolding
On-column refolding was conducted on 1 ml and 5 ml Q-FF columns and DEAE-FF columns. A constant flow rate of 0.5 column volumes (CVs)/min was used for sample loading. The column was first equilibrated with 12 ml Denaturation Buffer and then loaded with up to 12 ml protein extract containing 0.23 mg/ml denatured-reduced rhAFP. Refolding was initiated by switching buffer from Denaturation Buffer to Refolding Buffer (3 M urea, 20 mM Tris, 1 mM EDTA, 0 to 5 mM GSH, and 0 to 3.6 mM GSSG, pH 8.5) over 10 CVs and 4 CVs for 1 ml and 5 ml columns, respectively, at 0.5 CVs /min. The protein was then incubated on-column for 0 to 24 h. Elution was initiated by a salt gradient over 10 CVs and 4 CVs for 1 ml and 5 ml columns, respectively, from Refolding Buffer to Elution Buffer (20 mM Tris, 1 mM EDTA, 1 M NaCl, pH 8.5). The column was washed and regenerated with 5 to 10 CVs of Stripping Buffer (8 M urea, 20 mM Tris, 3 mM EDTA, 20 mM DTT, 1 M NaCl, pH 8.5) after each refolding cycle. Samples from the flow through, buffer switching and elution fractions were collected and analyzed by SDS-PAGE, RP-HPLC and Bioanalyzer to determine product recovery, refolding yield and product purity.
2.5.2. Off-column refolding
Off-column refolding was conducted on a 1 ml Q-FF column. Column equilibration and sample loading procedures were the same as on-column refolding. After washing away DTT and unbound proteins with 10 CVs of Equilibration Buffer, the mobile phase flow rate was reduced from 0.5 ml/min to 0.25 ml/min, and elution was initiated immediately by buffer change from Equilibration Buffer to Refolding Buffer (20 mM Tris, 1 mM EDTA, 4.5 mM GSH, 0.9 mM GSSG, 0.15 M NaCl, pH 8.5) with or without 0.5 M L-arginine over 10 CVs. The column was washed and regenerated with 10 CVs of Stripping Buffer. The eluate were incubated at 4°C for 3 h and subsequently analyzed by RP-HPLC and Bioanalyzer.
2.6. Analytical methods
RP-HPLC analysis was performed on a Shimadzu LC-10AVP high performance liquid chromatography (HPLC) system using a C5 Jupiter reversed phase (RP) column (5 µm particle size, 300 Å pore size, 150 - 4.6 mm, Phenomenex). The column was first equilibrated with 10 ml 40% (v/v) acetonitrile, followed by a 47-57% (v/v) acetonitrile-water gradient over 30 min. 0.05% (v/v) TFA was added to all RP-HPLC buffers. Absorbance was measured at 214 nm at room temperature. Protein mass was determined by peak integration, based on a standard curve attained by calibration using native and denatured-reduced std-AFP. Peak tailing appearing in the eluted protein traces was excluded from peak integration, which was not present in RP-HPLC traces of std-AFP. Refolding yield was calculated as the mass ratio of final refolded rhAFP to total denatured-reduced rhAFP. Refolding productivity was determined according to Equation (1):
where P is refolding productivity (mg ml-1 h-1), Y is refolding yield (%), M0 is the mass of total denatured rhAFP in the refolding system (mg), Vr is the volume of the refolding reactor (ml) and t is refolding incubation time (h).
Reducing SDS-PAGE was performed using precast 4-12% gradient NuPAGE Bis-Tris polyacrylamide gels (Invitrogen). Protein samples were mixed with Laemmli sample buffer 2- concentrate at a ratio of 1:1 (v/v) and heated for 10 min at 100 °C. 10 μl of the protein sample was loaded into each well and electrophoresis was conducted for 50 min at 200 V. Protein bands were detected using SimpleBlueTM SafeStain (Invitrogen) staining and the gel was destained using distilled water.
Total protein concentration and rhAFP purity were determined using Bioanalyzer in combination with the Protein 230 Chip kit (Agilent). Chips were prepared according to the protocol provided by the supplier.
The correct appearance of AFP epitopes was verified by the AFP ELISA diagnostic kit (Leinco Technologies, USA). Two unique antibodies (goat polyclonal and mouse monoclonal) were used to recognize distinct antigenic determinants on the rhAFP molecule. The plastic wells were supplied pre-coated with murine monoclonal anti-AFP. Refolded rhAFP (20 μl) diluted with bovine serum to concentrations ranging from 10 to 350 ng/ml was added to the wells prior to incubation (5 min). Goat polyclonal anti-AFP horseradish peroxidase conjugate (200 μl) was added to each well. After 60 min incubation, wells were washed to remove unbound labeled-antibody. Enzyme substrate-chromogen (100 μl) (hydrogen peroxide, H2O2, and tetramethylbenzidine, TMB) was added to each well and incubated for 30 min at room temperature. H2SO4 (1.0 N) was then added to stop the reaction, and product concentration was read at 450 nm using a microplate reader (Bio-Rad xMarkTM Absorbance Microplate Spectrophotometer). The absorbance of Refolding Buffer, enzyme-labeled goat polyclonal anti-AFP and denatured rhAFP at 450 nm were tested as controls.
3. Results and discussion
3.1. Recovery of rhAFP by mechanical disruption and chemical extraction
rhAFP was expressed after the bacterial cells were grown in 500 ml of 2xYT medium to an OD600 of 2.0±0.1. The total cell lysate was analyzed by SDS-PAGE for AFP expression (Fig. 1). The estimated expression yield of rhAFP was 35±3%, where 13±2% was soluble and 87±2% was insoluble (Fig. 2a).
Like most IB proteins, the refolding yield of denatured rhAFP has been reported to vary in a contaminant-dependent manner . Therefore, maintaining high product purity pre-refolding was important to maintain good refolding yields. Two different IB isolation bioprocesses were investigated and compared in this study to determine a superior means for upstream production of rhAFP with respect to product concentration and purity. Our results showed that mechanical disruption by sonication coupled with a centrifugation step yielded higher rhAFP purity (i.e. 68±3%) compared to that obtained via chemical extraction (i.e. 21±3%). Additionally, the ability to tune the re-suspension volumes of MD pellets allows the attainment of higher rhAFP concentration in the MD-processed extract (i.e. up to 2 mg/ml rhAFP) compared to CE (0.15 mg/ml rhAFP) (Fig. 2). The presence of DNA contaminants was also examined using these two bioprocesses by measuring protein sample absorbance at 260 nm and 280 nm, and comparing the 260 nm to 280 nm absorbance ratio. Co-extraction of cellular DNA in chemical extraction necessitated the addition of a DNA chelating agent such as PEI, to substantially remove cellular DNA. Our results showed that sonication coupled with centrifugation, was able to substantially remove cellular DNA from the IB pellet, to levels comparable to DNA removal by PEI in the chemical extraction mixture (Table 2). Considering that an anion exchange matrix will be used for rhAFP refolding in this work, the concentration of DNA contaminants in the sample load must be significantly reduced to minimize binding of competing contaminants on the positively-charged matrix.
Although the CE bioprocess was simpler to conduct (Fig. 3), the purity of rhAFP was considerably lower than that obtained from MD processing (Fig. 2a and 2b). rhAFP capture on the Q-FF 1 ml column was higher for MD-processed rhAFP compared to PEI-treated CE extract under identical binding conditions, where a 4-fold increase in rhAFP binding concentration could be achieved with MD-derived rhAFP compared with CE-derived protein (data not shown). It is likely that other more strongly negatively-charged cellular contaminants in the CE extract dominated adsorption on the matrix. Therefore, another pre-refolding purification step would be required to render the CE protein extract suitable for an adsorption-based refolding step. For the MD-processed protein extract, 80±5% of the denatured-reduced rhAFP loaded was adsorbed (data not shown). Based on these preliminary adsorption studies, MD was subsequently used to prepare all protein samples for refolding by batch dilution and column.
3.2. Refolding by batch dilution
Batch dilution refolding of denatured-reduced rhAFP was conducted as a control study to compare refolding yield and productivity against those obtained in column refolding. Different rhAFP starting concentration was diluted between 2.6- to 5-fold into Refolding Buffer to achieve different rhAFP refolding concentrations ranging from 0.03 to 0.45 mg/ml. Refolding yields > 40% were readily achievable at rhAFP refolding concentrations lower than 0.1 mg/ml, after 6 h incubation. Due to upstream constraints, the highest protein refolding concentration allowable for dilution refolding was 0.45 mg/ml, which gave a refolding yield of 14% (Fig. 4).
3.3. Refolding by IEX chromatography
3.3.1. On-column refolding
Two types of anion exchangers having cross-linked agarose matrices, Q-FF and DEAE-FF, were investigated for on-column refolding of rhAFP. Early studies have indicated that agarose gels showed higher binding capacities for pure proteins compared to synthetic polymer matrices such as Source Q, Toyopearl DEAE and Fractrogel® EMD DEAE . It is possible that the low-charge agarose beads minimize the occurrence of ionic repulsion between the resin and protein, thus improving protein adsorption. The use of both strong (Q-FF) and weak (DEAE-FF) anion exchangers was aimed to study if differing extents of protein-matrix interaction due to different ionization states of the functional groups, would affect rhAFP refolding. From our preliminary adsorption studies, it was found that the rhAFP refolding concentration possible on both Q-FF and DEAE-FF matrices readily exceeded the maximum allowable for batch dilution. Both columns showed comparable dynamic binding capacities for rhAFP for protein load between 0.15 to 2.4 mg, at a load flow rate of 0.5 CVs/min.
A rhAFP refolding concentration that was two times higher than that allowable by dilution refolding (i.e. 1 mg/ml) was chosen to form the basis of all refolding studies reported, except for the investigation of the effect of protein load on refolding yield. Previous refolding studies showed that refolding yields of cysteine-containing proteins including AFP can be improved by controlling the redox environment within which refolding takes place [25,26] because refolding buffer redox potential directly influences the oxido-shuffling of disulfide bonds, and hence the rate of correct refolding. When column refolding was conducted in the absence of redox agents, rhAFP refolding yields were extremely low (i.e. < 5%). Therefore, an optimum redox environment to facilitate on-column refolding of highly disulfide-bonded rhAFP was first investigated, with the aim to achieve an optimum redox-imposed physicochemical environment for refolding. It was interesting to observe that DEAE-FF and Q-FF required significantly different reduced to oxidized glutathione (GSH:GSSG) ratio to achieve maximum refolding yields (Fig. 5). This result suggests that the protein-matrix interaction behavior on the two columns is different. Therefore different redox environments are required to facilitate disulfide shuffling to attain a thermodynamically stable protein conformation. This finding indicates the importance of selecting a matrix that interacts favorably with the protein under an optimum physicochemical environment to allow free cysteine shuffling. A 1:1 and 5:1 GSH:GSSG ratio was used for subsequent on-column refolding studies for DEAE-FF and Q-FF, respectively.
Different amounts of rhAFP (0.15 to 2.4 mg) was loaded on Q-FF 1 ml and DEAE-FF 1 ml, and refolded under identical conditions (refer to Materials and Methods, Section 2.5.1) to determine the effect of protein load on refolding yield. Refolding yields decreased from 20% to 7% for Q-FF 1 ml column, and 22% to 8% for DEAE-FF 1 ml column, when the amount of rhAFP loaded increased from 0.15 mg to 2.2 mg, and 0.4 mg to 2.4 mg, respectively (Fig. 6). It is clear that aggregation increased with increasing protein loads. The decrease in refolding yield with increasing protein load is likely due to a non-uniform distribution of proteins across the column during loading; a high local protein concentration at any section of the column may enhance aggregation and hence decrease refolding yields. A mass balance of the on-column refolding process was conducted to determine protein recovery (Table 3). It is clear that protein was predominantly lost as a result of irreversible binding on the column during refolding and elution, as indicated by the amount of protein recovered during the stripping step. The protein aggregates were found to be strongly adsorbed on the column matrix and could only be eluted from the column using Stripping Buffer which contained 8 M urea, 20 mM DTT and 1 M NaCl. The refolded protein fraction that was eluted from the column consisted predominantly of correctly refolded rhAFP. RP-HPLC and Bioanalyzer analyses of the refolded protein revealed that > 95% of the protein recovered in the refolded protein fraction eluted were correctly refolded, thus ruling out the loss of rhAFP yield to soluble aggregates.
An FPLC chromatogram showing protein elution following on-column incubation in Refolding Buffer is shown in Fig. 7. Almost no rhAFP was lost in the flow through fractions during sample loading or buffer changing steps, as determined by SDS-PAGE analysis (Fig. 8). The elution profile showed two peaks when a linear salt gradient elution was conducted. Analysis of the first peak revealed no protein content, and the absorbance at 280 nm was attributed to absorbance by GSH and GSSG in Refolding Buffer. RP-HPLC and Bioanalyzer analyses of fractions from the second peak showed that it contained predominantly rhAFP (0.12 mg/ml at 95% purity). A higher rhAFP concentration would be readily obtained by substituting gradient elution with step elution.
Since process productivity is directly dependent on refolding time, the effect of column incubation time on refolding yield was also studied for both columns. When elution was initiated immediately after buffer exchange from Denaturation Buffer to Refolding Buffer, no refolded rhAFP was detected, as determined by RP-HPLC analysis. Refolding yields of rhAFP increased from 0% to 15±1% for Q-FF, and 0% to 19±1% for DEAE-FF, as on-column incubation was increased from 0 h to 6 h and 0 h to 3h, respectively (Fig. 9). These results indicate that DEAE-FF had a more positive influence on rhAFP refolding kinetics compared to Q-FF, suggesting that the DEAE ion exchanger surface may have assumed the role of a refolding aid. Additionally, Q-FF is a strong anion exchanger with a stronger charge ionization state compared to DEAE-FF. Therefore, a strong rhAFP-matrix interaction which impacts the free movement of protein molecule to attain the correctly refolded conformation, upon introduction of Refolding Buffer, is likely. A weaker protein-matrix interaction on DEAE-FF, on the other hand, may allow more flexible movement of the protein molecule, thus possibly, allowing more efficient disulfide-shuffling. In parallel to correct refolding, irreversible protein aggregation had occurred instantaneously upon contact with Refolding Buffer, resulting in no further increase in refolding yield after 6 h and 3 h, respectively. Formation of protein aggregate precipitate layers, especially at the top section of the column, may impede free movement of protein molecules across the matrix and halt further increase in refolding yield. To verify this hypothesis, the effect varying column geometry on refolding yield was subsequently investigated.
On-column refolding was performed on 5 ml Q-FF and DEAE-FF columns using the same refolding method and conditions as the 1 ml columns. 1 mg of denatured-reduced rhAFP was loaded on the columns at a mobile phase flow rate of 0.5 ml/min and 2.5 ml/min for the 1 ml and 5 ml columns, respectively, and incubated for 6 h in Refolding Buffer. These volumetric mobile phase flowrates translate to a fixed linear flow rate of 1.3 cm/min when normalized against column diameter. Refolding yields increased to 25% and 32% for both Q-FF and DEAE-FF columns, respectively (Table 4). It is likely that at the same protein load, the 5 ml columns which offer a broader inner surface area facilitated better dispersion of the protein across the column, and hence reduced the occurrence of a high local protein concentration at a given region on a matrix. Consequently, inter-molecular protein interaction is minimized, and higher refolding yields could be achieved.
3.3.2. Off-column refolding
Off-column refolding was also investigated in parallel to compare rhAFP refolding yields without subjecting the protein molecules to column incubation with Refolding Buffer. Instead, following the binding and wash steps, the protein was eluted immediately with Refolding Buffer containing 0.15 M NaCl and incubated for 3 h. Off-column refolding yields were comparable to on-column refolding (for 3 h incubation) without the addition of L-arginine to the Refolding Buffer (Fig. 10). However, refolding yields reached 42% (for 1 mg rhAFP load) when 0.5 M L-arginine was added into Refolding Buffer. It was not possible to add L-arginine in the Refolding Buffer for on-column refolding because the high solution conductivity contributed by L-arginine interferes with protein adsorption on the matrix. Furthermore, the use of L-arginine in Refolding Buffer will also add considerably to process cost, which is undesirable for scale-up applications. The significant difference between off-column and on-column refolding yields confirms a significant extent of rhAFP aggregation was column-induced, and this was reduced by increasing column diameter.
3.4. Comparison of batch dilution and IEX column refolding
0.45 mg/ml was the highest rhAFP refolding concentration attainable with batch dilution in this study due to solubility and viscosity constraints in the upstream sample preparation step. In a direct comparison between batch dilution and column refolding, refolding at a rhAFP concentration of 0.45 mg/ml on-column (i.e. the lowest concentration studied for on-column refolding) resulted in a refolding yield of 22% on DEAE-FF after 3 h incubation, while a 14% refolding yield was achieved by batch dilution following 6 h incubation. By comparing dilution and off-column refolding yields, the latter was also superior where 23% to 42% refolding yields were readily achievable at high rhAFP refolding concentrations (i.e. 1 to 2 mg/ml). These results clearly indicate that upon adequate optimization of chromatography and refolding parameters, both on- and off-column refolding readily outperformed dilution refolding in terms of refolding yields as a function of refolding concentration. Fig. 11 shows the RP-HPLC chromatogram profiles of refolded rhAFP fractions following dilution and column refolding. The significant 'peak shouldering' observed in peak D compared to peak E suggests a lower refolding efficiency and product purity by batch dilution compared to on-column refolding, after the same refolding incubation time. Both batch- and column-refolded rhAFP gave positive responses in the ELISA test, with controls consisting of denatured-reduced rhAFP, yielding negative responses (data not shown). This result confirms that the column-refolded rhAFP maintains a spatial conformation that exhibits bioactivity to the same extent as human-derived AFP.
The superiority of column refolding in terms of rhAFP refolding productivity was also demonstrated in this study. Tables 5a and 5b provide the experimental parameters employed in both dilution and on-column required for determination of refolding productivity. Even with the use of a refolding concentration that was two times higher than dilution refolding, refolding productivity in column-assisted refolding increased by 6-fold compared to dilution refolding. The increased productivity in column refolding studies was attributed mainly to the higher protein refolding concentrations possible, while requiring less refolding incubation time compared to batch dilution refolding. Off-column refolding further improved refolding productivity by 13-fold compared to dilution refolding, by eliminating column-induced aggregation. It is also clear that total solvent consumption required for both dilution and column refolding was comparable despite the higher rhAFP concentration achieved for column refolding. For process scale applications, the refolding buffer requirement for column refolding will be lesser than dilution refolding, where a proportional increase in buffer consumption is required for the latter but not the former.
Protein concentration post-refolding was also higher in column-refolding compared to batch dilution, which is expected considering that adsorption-based chromatography often function as a protein concentration platform. At 1 mg/ml rhAFP refolding concentration, the concentration of the refolded rhAFP fraction was easily 3-fold higher than that attainable by dilution refolding at half the protein refolding concentration. For column refolding, a higher rhAFP protein concentration is expected to be readily achieved with a step elution upon process scale-up, thereby eliminating the need for a protein concentration step post-refolding. Another clear advantage for column refolding over dilution refolding is the simultaneous purification achieved concurrent to refolding, where rhAFP purity after column refolding processing was enhanced from 68% (achieved in batch dilution) to 90-95% (Table 5b). This result indicates that column refolding was able to purify the correctly refolded protein from misfolded, incompletely folded proteins and other contaminant proteins. Based on this study on IB-derived AFP, contribution of column refolding toward bioprocess intensification is clear, where its capability to simultaneously purify and refold proteins in a simple and readily automated manner shows suitability for large-scale IB refolding applications.
This study demonstrates the effectiveness of adsorption-based refolding technology to refold a commercially valuable IB protein, having 16 disulfide bonds, at refolding concentrations which were unattainable with dilution refolding (i.e. > 0.5 mg/ml). The influence of resin chemistry, column geometry, redox potential and feed conditioning on rhAFP refolding yield is significant, thus emphasizing the importance of adequate optimization of both refolding and column chromatography operations to maximize refolding yields. The simultaneous separation of correctly refolded rhAFP from refolding aggregates and contaminant proteins further simplifies bioprocessing and enhances overall product recovery. Increase in rhAFP refolding productivity by on-column refolding was most evident with the use of a DEAE matrix. A 20 % refolding yield was readily attained at 1 mg/ml rhAFP refolding concentration, after only 3 h incubation. By substituting dilution refolding with column refolding, an increase in rhAFP refolding purity and productivity by up to 1.5- and 6-fold, respectively, was readily achieved. Off-column refolding yielded an even higher rhAFP refolding productivity (i.e. 13-fold increase from dilution refolding), when refolding is conducted in the eluate instead of on the column matrix.
From this study, a potentially feasible process route based on on-column refolding for large-scale manufacture of rhAFP is identified and demonstrated. The high refolding productivity achieved by column refolding compared to dilution refolding will translate into more rapid product delivery to market and a cheaper product, upon adequate optimization of the overall bioprocess. The ready availability of the chromatographic resin used, coupled with the simple and readily scalable on-column refolding process employed, indicate the widespread applicability of this technology for simultaneous refolding and purification of other complex IB molecules.