Journal of Chromatography B, 969 (2014) 241–248
Contents lists available at ScienceDirect
Journal of Chromatography B journal homepage: www.elsevier.com/locate/chromb
Use of mep HyperCel for polishing of human serum albumin Karl B. McCann ∗ , Yvonne Vucica, John Wu, Joseph Bertolini Research and Development Department, CSL Behring (Australia), 189-209 Camp Rd, Broadmeadows, Victoria 3047, Australia
a r t i c l e
i n f o
Article history: Received 1 May 2014 Received in revised form 21 August 2014 Accepted 22 August 2014 Available online 28 August 2014 Keywords: MEP HyperCel Human serum albumin Hydrophobic charge induction chromatography Plasma
a b s t r a c t The manufacture of human serum albumin by chromatographic procedures involves gel filtration chromatography as a final polishing step. Despite this step being essential to remove high molecular weight impurity proteins and thus ensure a stable and safe final product, it is relatively inefficient. This paper explores the use of hydrophobic charge induction chromatographic media, MEP HyperCel as an alternative to Sephacryl S200HR gel filtration for the polishing of human serum albumin derived by ion exchange chromatographic purification of Cohn Supernatant I. The use of MEP HyperCel results in a product with a higher purity than achieved with gel filtration and in a less time consuming manner and with potential resource savings. MEP HyperCel appears to have great potential for incorporation into downstream processes in the plasma fractionation industry as an efficient means of achieving polishing of intermediates or capture of proteins of interest. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Commercial human serum albumin is predominantly manufactured using ethanol precipitation methods based on those originally developed by Cohn et al. (1946) [1] and the subsequent modifications of Kistler & Nitschmann (1962) [2]. The continued purification of albumin by ethanol precipitation can be attributed to the fact that these methods allow the efficient processing the large volumes encountered in a modern plasma fractionation process. However, smaller but increasing amounts of albumin are fractionated using chromatographic processes, introduced either as a final polishing step following initial ethanol fractionation [3] or as fully integrated chromatographic processes [4–6]. Chromatographic purification of albumin has a number of benefits including higher process yields, higher purity [3,5,7] and improved tolerability and clinical safety of the product [8]. CSL Behring (Australia) have manufactured human serum albumin (Albumex) using a chromatography method based on that developed by Berglof et al. (1983) for almost two decades [9]. This chromatographic process, which includes two ion exchange and a size exclusion columns, was initially applied to delipidated Supernatant II + III, but was subsequently adapted to the processing of a
Abbreviations: ID, Internal diameter; ELISA, Enzyme linked immunosorbent assay; DEAE, Diethylaminoethyl; CM, Carboxymethyl; pI, Isoelectric point; HCIC, Hydrophobic charge induction chromatography. ∗ Corresponding author. Tel.: +61 3 9246 5493. E-mail address: [email protected] (K.B. McCann). http://dx.doi.org/10.1016/j.jchromb.2014.08.029 1570-0232/© 2014 Elsevier B.V. All rights reserved.
delipidated Supernatant I feedstream to allow the complete chromatographic purification of IgG for the Intragam P process [5,10]. Despite the above mentioned benefits, the widespread adoption of chromatographic processes for fractionation of human albumin has been limited due to cost and inefficiencies associated with the processes at industrial scale, with respect to the large number of chromatographic cycles required to process a batch. This is particularly the case in the fully chromatographic processes, with the Sephacryl S200HR gel filtration step for final polishing to remove high molecular weight impurity proteins. The efficiency of a chromatographic process for the purification of albumin could be greatly enhanced if an alternative mode to Sephacryl S200HR was developed as a final polishing step. There have been a number of reports that hydrophobic interaction chromatography (HIC) offers sufficient selectivity to allow separation of plasma proteins based on differences in the surface hydrophobicity of the proteins [11,12]. This has allowed HIC to be used successfully for the purification of albumin from human plasma [13], human placenta [14] and of recombinant human serum albumin from Pichia pastoris [15]. Despite HIC offering an orthogonal mode of separation to the ion exchange steps used for chromatographic purification of albumin, these resins have not been used by any of the commercial plasma fractionators. Their use at large-scale has been limited by the fact that HIC-based chromatography requires use of significant amounts of lyotropic salts in order to achieve the binding and fractionation of proteins [16]. An alternative to conventional HIC is hydrophobic charge induction chromatography (HCIC). One such resin for HCIC is 4-mercaptoethylpyridine (MEP) HyperCel [16]. The
242
K.B. McCann et al. / J. Chromatogr. B 969 (2014) 241–248
MEP ligand permits separation based on the surface hydrophobicity of proteins but does not require the addition of lyotropic salts [17]. MEP HyperCel has been shown to have a high selectivity for immunoglobulins and for this reason there has been a major focus on the use of this media for the isolation of monoclonal antibodies [17–20]. In addition, MEP HyperCel has been proven to be effective for the separation of penicillin acylase [21] and Fc-fusion proteins [22]. In this paper an alternative method for the polishing of human serum albumin using MEP HyperCel is investigated. IgG is one of the major residual proteins present in chromatographically purified albumin. Therefore the optimal conditions for MEP HyperCel chromatography for the partitioning of albumin and IgG were initially evaluated using purified samples of albumin and IgG. The defined conditions were then verified by examining the capability of MEP HyperCel to polish the ion exchange purified albumin process intermediate. The impact of flow rate on the achieved separation and optimal protein sample loading was also determined. After successful demonstration of the new process at small-scale, the process was evaluated at Pilot-scale and the final product was compared to the albumin produced at industrial-scale. 2. Materials and methods 2.1. Protein solutions Process intermediates including Cohn Supernatant I (SNI) depleted of lipids and euglobulins, albumin eluate from CM Sepharose-FF chromatography, purified human serum albumin (HSA) and purified human immunoglobulin G (IgG) and purified albumin from Sephacryl S200HR chromatography were obtained from CSL Behring (Broadmeadows, Victoria, Australia). All protein solutions were filtered through a 0.22 m membrane (Durapore® , Millipore Corporation, Bedford, MA, USA) prior to chromatography. 2.2. Antibodies and ELISA reagents Antibodies against the human plasma proteins of transferrin, ␣2 -macroglobulin, ␣1 -glycoprotein, ␣1 -antitrypsin, haptoglobin and ceruloplasmin were obtained from DakoCytomation Denmark A/S (Glostrup, Denmark), inter ␣ trypsin inhibitor antibodies were obtained from The Binding Site (Birmingham, England), and apolipoprotein A1 and apolipoprotein B antibodies were obtained from Siemens Healthcare Diagnostics Inc. (Tarrytown, NY, USA). ELISA kits for IgG and IgM were obtained from Bethyl Laboratories Inc. (Montgomery, TX, USA) and the ELISA kit for ␣2 -macroglobulin was obtained from GenWay Biotech. Inc. (San Diego, CA, USA). Antibodies for the IgA ELISA were obtained from DakoCytomation Denmark A/S (Glostrup, Denmark) and the IgA standard was obtained from Calbiochem-Novabiochem Corporation (San Diego, CA, USA). Antibodies against the human plasma proteins of albumin, IgG, IgA and IgM, which were used for nephelometry analysis, were obtained from Beckman-Coulter Inc. (Fullerton, CA, USA). 2.3. Chromatographic resins MEP HyperCel resin was obtained from Pall Life Sciences (East Hills, NY, USA). DEAE Sepharose-FF, Capto DEAE, CM Sepharose-FF and Sephacryl S200HR resins were obtained from GE Healthcare BioSciences AB (Uppsala, Sweden). 2.4. Characterisation of Sephacryl S200HR process The albumin intermediates before and after Sephacryl S200HR chromatography were characterised using nephelometry
(IMMAGE Immunochemistry system, Beckman-Coulter Inc., Fullerton, CA, USA) to determine the levels of albumin, IgG, IgA and IgM, and immunoelectrophoresis to determine the levels of transferrin, ␣2 -macroglobulin, ␣1 -glycoprotein, ␣1 -antitrypsin, inter ␣ trypsin inhibitor, haptoglobin, apolipoprotein A1, apolipoprotein B and ceruloplasmin.
2.5. Impact of heating on turbidity of albumin solutions The excellent viral safety record of commercial HSA solutions is attributed largely to the pasteurisation step (60 ◦ C for 10 h) [3]. The albumin is stabilised by addition of sodium caprylate, however, contaminant proteins present not protected by the caprylate are susceptible to heat induced denaturation. Thus evaluation of the turbidity of an albumin solution following pasteurisation is a good indication of the presence of residual proteins. A control sample of albumin eluate from CM Sepharose-FF chromatography and samples of purified albumin derived from Sephacryl S200HR chromatography were concentrated to 15–20% w/v protein by ultrafiltration (10 kDa OMEGA membrane; Pall Corporation, Hauppauge, NY, USA) and formulated with 32 mM sodium caprylate. The samples were pasteurised by heating at 60 ◦ C for 10 h. At the completion of pasteurisation the turbidity of the samples was assessed using a Hach 2100AN Turbidimeter (Hach Company, Loveland, CO, USA).
2.6. Determination of MEP HyperCel operating conditions for partitioning of albumin and IgG The operating parameters for MEP HyperCel chromatography were explored using purified samples of albumin and IgG. The binding and elution of albumin and IgG was determined separately using a range of equilibration conditions including; 50 mM sodium phosphate (pH 7.0); 50 mM sodium acetate (pH 5.5); 50 mM sodium acetate (pH 5.5) + 150 mM NaCl; 50 mM sodium acetate (pH 5.5) + 300 mM NaCl; 50 mM sodium acetate (pH 7.0); and 110 mM sodium acetate (pH 7.0), shown in Table 3. The pure IgG and pure albumin samples were prepared in the appropriate equilibration buffer at a concentration of approximately 5-10 mg/mL and filtered through a 0.22 m membrane (Durapore® , Millipore Corporation, Bedford, MA, USA) prior to chromatography. The samples were loaded on to an equilibrated MEP HyperCel column (17.5 cm × 1.6 cm ID; GE Healthcare BioSciences AB, Uppsala, Sweden) at 20 g per litre of resin. The unbound protein was recovered during the sample loading period and during the subsequent post-sample loading wash with equilibration buffer. Any bound protein was eluted using sodium acetate buffer at pH 3.0. All chromatography steps were conducted at a flow rate of 100 cm/h and monitored at 280 nm using an AKTA Explorer 100 equipped with Unicorn software (GE Healthcare BioSciences AB, Uppsala, Sweden). The amount of protein recovered in the flow through and eluate fractions was determined using nephelometry. The most suitable equilibration buffer conditions defined from the above experiment was subsequently verified using a sample containing a 9:1 mixture of albumin and IgG. The pure albumin and pure IgG was mixed in 110 mM sodium acetate (pH 7.0) at a ratio of 9:1 to give a final protein concentration of approximately 5–10 mg/mL. The sample was loaded on to the equilibrated MEP HyperCel column at protein loadings of 100, 200 and 300 g per litre of resin. The unbound protein was recovered using 110 mM Sodium acetate (pH 7.0) equilibration buffer and the bound protein was eluted using sodium acetate buffer at pH 3.0. The amount of albumin and IgG in the flow through and eluate fractions was determined using nephelometry.
K.B. McCann et al. / J. Chromatogr. B 969 (2014) 241–248
2.7. Polishing of ion exchange purified albumin using MEP HyperCel After demonstrating that the MEP HyperCel could successfully separate an albumin and IgG mixture, the ability of MEP HyperCel to purify the ion exchange purified albumin was assessed. The albumin eluate from CM Sepharose-FF chromatography was adjusted to pH 7.0 ± 0.1 and clarified by passage through a 0.22 m filter (Durapore® , Millipore Corporation, Bedford, MA, USA) prior to loading on to the MEP HyperCel column. The albumin sample (1000 mL) was loaded on to the MEP HyperCel column (17.5 cm × 1.6 cm ID), which had been equilibrated with 110 mM sodium acetate (pH 7.0), followed by a 110 mM sodium acetate (pH 7.0) post-sample wash to recover unbound protein. Bound protein was eluted with 100 mM sodium acetate (pH 3.0) and the resin was cleaned in place with 1 M NaOH. The flow through fraction was collected at 10 mL intervals and assessed for the levels of the major impurity proteins of IgG, IgA, IgM and ␣2 -macroglobulin by ELISA. The initial analysis of the impurity protein break through curves showed that sample loadings of up to 1000 mL on to a 35 mL MEP HyperCel column were excessive. The albumin sample was loaded on to the equilibrated MEP HyperCel column at 100, 200, 300, 400 and 500 mL. The flow through and post-sample wash was collected as a bulk sample along with the pH 3.0 eluate. These bulk fractions were assessed for the levels of albumin by nephelometry and the major impurity proteins of IgG, IgA, IgM and ␣2 -macroglobulin by ELISA. In attempt to improve the selectivity of the MEP HyperCel for binding of residual proteins, the above experiment was repeated with 50 mM NaCl added to the equilibration buffer and albumin sample. Under both buffer conditions the effect of sample loading was examined. The albumin sample was loaded on to the equilibrated MEP HyperCel column at 100, 200, 300, 400 and 500 mL. The flow through and post-sample wash was collected as a bulk sample along with the pH 3.0 eluate. These bulk fractions were assessed for the levels of albumin by nephelometry and the major impurity proteins of IgG, IgA, IgM and ␣2 -macroglobulin by ELISA. 2.8. Assessment of flow rate on performance of MEP HyperCel The four major impurity proteins appeared to have differing affinities to the MEP HyperCel resin. Therefore studies were conducted to investigate the impact of the flow rate on binding. Using the chromatography conditions from above and with buffers without the addition of 50 mM NaCl, the albumin sample (100 mL) was loaded on to the MEP HyperCel column at flow rates of 50, 100 and 150 cm/h. The flow through and eluate fractions were collected as bulk samples and assessed for the levels of albumin by nephelometry and the major impurity proteins of IgG, IgA, IgM and ␣2 -macroglobulin by ELISA. 2.9. Manufacture of albumin using MEP HyperCel chromatography at pilot-scale As the small-scale studies showed that the MEP HyperCel could be successfully used for the polishing of chromatographically purified albumin, the reproducibility of the new process scheme was evaluated at pilot-scale over three independent batches. For each batch a 15 kg sample of Cohn SNI, which was delipidated and depleted of euglobulins, was purified on the ion exchange columns (17.5 cm × 10 cm ID; GE Healthcare BioSciences AB, Uppsala, Sweden). The albumin eluate from CM Sepharose-FF chromatography was adjusted to pH 7.0 and clarified by passage through a 0.22 m membrane (Durapore® , Millipore Corporation, Bedford, MA, USA). The albumin sample was loaded on to an equilibrated MEP HyperCel column (17.5 cm × 10 cm ID; GE Healthcare
243
Table 1 Characterisation of albumin sample before and after Sephacryl S200HR chromatography.
Component Albumin IgG IgA IgM ␣2 -macroglobulin ␣1 -antitrypsin Apolipoprotein A1 Total
Albumin before Sephacryl S200HRa
Albumin after Sephacryl S200HR
Clearance across S200HR
% of total 98.58 0.37 0.26 0.08 0.62 0.05 0.02 1.42%
% of total 99.81 0.096 0.008 0.004 0.017 0.029 0.034 0.19%
% NA 74.0 97.0 94.0 97.3 42.0 −46.7
a CM Sepharose-FF eluate. NA = Not Applicable.
Table 2 Effect of pasteurisation on turbidity of albumin derived from ion exchange chromatography and Sephacryl S200HR chromatography. Turbidity (NTU) Sample
Before pasteurisation
After pasteurisation
Sephacryl S200HR eluate Ion exchange eluate
2.23 2.75
2.36 9.36
BioSciences AB, Uppsala, Sweden) at a protein loading of 80 g/L and a flow rate of 100 cm/h. The purified albumin flow through was further processed according to the current Albumex process. The final Albumex product for the three independent batches was characterised for protein composition, turbidity levels and a range of impurity proteins including; IgG, IgA, transferrin, ␣2 macroglobulin, ␣1 -glycoprotein, ␣1 -antitrypsin, inter ␣ trypsin inhibitor, haptoglobin, apolipoprotein A1, apolipoprotein B and ceruloplasmin. The results were compared to three control batches of Albumex, which were manufactured at full-scale using the DEAE Sepharose-FF, CM Sepharose-FF and Sephacryl S200HR chromatography steps. 3. Results Before commencing MEP HyperCel studies the effectiveness of the current Sephacryl S200HR purification was characterised in order to identify the major contaminant proteins in the albumin intermediate. The concentrated ion exchange albumin intermediate which was loaded onto the Sephacryl S200HR contained 1.42% impurity proteins, with ␣2 - macroglobulin, IgG and IgA predominant (Table 1). These major impurity proteins all have molecular weights significantly higher than that of albumin, thus making the Sephacryl S200HR gel filtration suitable for their removal. Clearance rates were greater than 94% for ␣2 - macroglobulin, IgA and IgM, and greater than 74.0% for IgG. Following Sephacryl S200HR chromatography the total impurity content had been reduced to 0.19%, with IgG being the most abundant impurity protein present. Despite the albumin eluate derived from ion exchange chromatography having a purity of over 98% and thus complying with the albumin Pharmacopoeial requirements for protein composition and purity, the removal of high molecular weight impurities is critical for the quality of the final albumin product. This is clearly demonstrated by the fact that when the albumin derived from ion exchange chromatography is concentrated, formulated with sodium caprylate, and subjected to pasteurisation at 60 ◦ C for 10 h a significant increase in turbidity is observed (Table 2). No turbidity increase was encountered when the purified albumin derived from Sephacryl S200HR is pasteurised.
244
K.B. McCann et al. / J. Chromatogr. B 969 (2014) 241–248
Table 3 Percentage recovery of albumin and IgG in flow through and eluate fractions collected from MEP Hypercel chromatography using different equilibration buffers. Recovery (%) Albumin Buffer conditions 50 mM sodium phosphate (pH 7.0) 50 mM sodium acetate (pH 5.5) 50 mM sodium acetate (pH 5.5) + 150 mM NaCl 50 mM sodium acetate (pH 5.5) + 300 mM NaCl 50 mM sodium acetate (pH 7.0) 110 mM sodium acetate (pH 7.0)
Flow through 86.9
Table 4 Percentage recovery of albumin and IgG in flow through and eluate fractions collected from MEP Hypercel chromatography at different protein loadings of a 9:1 albumin:IgG mixture. Recovery (%)
IgG Eluate 5.3
Flow through
Eluate
ND
63.9
44.7
39.6
42.1
68.1
79.9
17.3
ND
97.3
81.1
14.6
ND
90.9
104.5
5.2
ND
96.0
89.2
5.1
ND
90.5
ND = Not detectable.
IgG as one of the major impurities in the ion-exchange albumin intermediate and also post-Sephacryl S200HR chromatography. Therefore the initial development work on the MEP HyperCel chromatography for polishing of the albumin intermediate focussed on optimising the ability to separate albumin from IgG. Initially the binding and elution characteristics of pure human albumin and human IgG to MEP HyperCel was examined using sodium phosphate (pH 7.0) and sodium acetate (pH 5.5) equilibration buffers. Previous studies by Schwartz et al. (2001) showed that MEP HyperCel has a maximum binding capacity when the sample pH is adjusted to approximately pH 7.0 [17]. The separation of albumin and IgG using MEP HyperCel was also evaluated at pH 5.5 as the albumin is currently eluted from the CM Sepharose-FF column at pH 5.5 using a sodium acetate buffer. With equilibration of the MEP HyperCel with sodium phosphate buffer, pH 7.0 effective binding of IgG occurred, with undetectable levels of IgG being recovered in the flow through fraction (Table 3). Only minor amounts of albumin bound to the MEP HyperCel, and greater than 85% of the albumin recovered in the flow through fraction. Equilibration with the sodium acetate buffer, pH 5.5 resulted in half of both the IgG and albumin binding to the resin. The addition of sodium chloride to the sodium acetate buffer, pH 5.5 improved IgG binding, with undetectable IgG levels being recovered in the flow through fraction (Table 3). There was also improved albumin recovery in the drop through fraction. However, approximately 15–20% of the albumin was bound and recovered in the eluate fraction. This was significantly higher binding than when the MEP HyperCel was equilibrated with sodium phosphate buffer, pH 7.0. Given that sodium acetate buffers are currently used in the chromatographic albumin process during the ion exchange chromatography steps, the binding and elution characteristics of purified albumin and IgG to the MEP HyperCel resin was assessed using sodium acetate buffers at pH 7.0. Using 50 mM and 110 mM sodium acetate buffers at pH 7.0 resulted in similar IgG and albumin recoveries to that observed with the 50 mM sodium phosphate buffer, pH 7.0. There were undetectable levels of IgG in the flow through fraction and only minor amounts of albumin being bound with approximately 5% of total being detected in the eluate (Table 3). The results suggest that the pH of the equilibration buffer and sample is the critical parameter that influences binding to the MEP HyperCel resin, with buffer type and conductivity of the equilibration buffer and sample having little impact.
Albumin
IgG
Loading (g Protein/L resin)
Flow through
Eluate
Flow through
Eluate
100 g/L 200 g/L 300 g/L
92.0 90.0 98.2
1.8 1.3 0.9
ND ND 1.4
92.1 86.9 86.2
ND = Not detectable.
As the MEP HyperCel showed good selectivity between albumin and IgG if equilibrated with 110 mM sodium acetate buffer, pH 7.0, this buffer was selected for further evaluation with an IgG and albumin mixture and ion exchange albumin intermediate. In the current chromatographic albumin process, albumin is eluted from the CM Sepharose-FF column with 110 mM sodium acetate buffer, pH 5.5. Thus albumin eluted from the CM Sepharose-FF column would only require a pH adjustment step prior to loading on to the MEP HyperCel column. The separation characteristics of MEP HyperCel were evaluated using a 9:1 mixture of albumin and IgG. The MEP HyperCel showed excellent separation of the albumin and IgG mixture. At protein loadings up to 200 g/L, the flow through fraction contained undetectable levels of IgG and over 90% of the total albumin loaded (Table 4). Most of the IgG was recovered in the eluate fraction and only minor amounts of albumin were lost in this fraction. The results suggest that the MEP HyperCel can remove significant amounts of IgG from an albumin solution that has a similar composition to the ion exchange eluate in the chromatographic albumin process. The MEP HyperCel was further evaluated by assessing the ability to purify the albumin eluate from the CM Sepharose-FF column. As shown previously in Table 1, the CM Sepharose-FF elute contains various protein impurities including ␣2 -macroglobulin, IgA, IgM and IgG. These proteins were used as purity markers to evaluate the capability of MEP HyperCel to further purify the CM Sepharose-FF eluate albumin intermediate to final product. CM Eluate (1000 mL) was adjusted to pH 7.0 and loaded on to the MEP HyperCel column. The flow through was collected in 10 mL fractions and tested for the marker proteins. The impurity break through profiles varied significantly for the four impurity proteins (Fig. 1). During the initial stages of sample loading only minor amounts of IgG were detected in the flow through (Fig. 1a). Despite a significant increase in IgG break through during the latter stages of the loading, the maximum IgG concentration detected only represented 1.1% of the IgG concentration in the sample load, suggesting that the MEP HyperCel has a high selectivity for IgG. The break through profile for IgA followed an almost linear increase throughout the sample loading (Fig. 1b). The maximum IgA concentration detected in the flow through represented 9.3% of the IgA concentration in the sample load. Also during the initial stages of the sample loading, only minor amounts of IgM were detected in the flow through (Fig. 1c). Towards the latter stages of the sample loading a significant increase in IgM was detected in the flow through. The maximum IgM concentration detected at the later stages of sample loading represented 53.1% of the IgM concentration of the sample load, suggesting that the capacity or affinity of MEP HyperCel for IgM is significantly lower than for IgG and IgA. Significant amounts of ␣2 -macroglobulin were detected in the flow through immediately after commencement of sample loading on to the MEP HyperCel column (Fig. 1d). The rate of increase in ␣2 -macroglobulin breakthrough was faster than any of the
K.B. McCann et al. / J. Chromatogr. B 969 (2014) 241–248
245
Fig. 1. Impurity break through profiles during MEP Hypercel purification of pH adjusted CM eluate (a) IgG, (b) IgA and (c) IgM and (d) ␣2 - macroglobulin.
other impurities and as a result the maximum ␣2 -macroglobulin concentration detected in the flow through was 76.0% of the ␣2 macroglobulin in the sample loading. As the ␣2 -macroglobulin is the most abundant impurity protein in the CM eluate and the MEP HyperCel has the weakest affinity for this protein, it indicates that ␣2 -macroglobulin will be the key impurity marker for subsequent determination of allowed sample loading onto the MEP HyperCel column. Loading onto MEP HyperCel column of pH adjusted CM eluate was examined over the range of 100–500 mL. Partitioning of the impurity proteins were consistent with the impurity break through profiles shown in Fig. 1. For IgG, recovery in the flow through fraction was less than 1% of the total IgG loaded at the highest sample loading (Table 5). At the highest sample loadings the recovery of IgA and IgM was also relatively low at 1.8% and 3.6%, respectively. However, even at the lowest sample loading of 100 mL, 5.2% of the total ␣2 -macroglobulin loaded on to the MEP HyperCel was recovered in the flow through fraction. ␣2 -macroglobulin break through increased proportionally with sample load, with 35.4% of the total ␣2 -macroglobulin being recovered in the flow through at the highest loading. As ␣2 -macroglobulin is the predominant impurity protein in the CM eluate and the MEP HyperCel exhibits the weakest affinity to ␣2 -macroglobulin, further studies were made to optimise the ␣2 -macroglobulin clearance by MEP HyperCel. Earlier data (Table 3) showed that addition of NaCl to 50 mM sodium acetate, pH 5.5 buffer improved albumin and IgG partitioning by MEP HyperCel. Therefore the impact of 50 mM sodium chloride addition to the pH adjusted CM eluate and equilibration buffer at pH 7.0 on the removal of impurities from albumin at various loadings was assessed. The recovery of IgG, IgA, IgM and ␣2 -macroglobulin in the flow through fractions was similar for the MEP HyperCel separations conducted with and without addition of 50 mM sodium chloride (Table 5). This indicated that the addition of sodium chloride did not have any impact on the selectivity of MEP HyperCel for the impurity proteins found in the CM eluate albumin
intermediate. The albumin recovered in the eluate fraction of these runs ranged between 0.3–2.3% indicating minimal albumin loss (Table 5). The albumin product recovery obtained across the MEP HyperCel compared well with those achieved across the Sephacryl S200HR column, where 3-5% losses are typically observed. The impact of flow rate on the separation of impurity proteins by MEP HyperCel was assessed between 50 and 150 cm/h using a sample loading of 100 mL of pH adjusted CM eluate. At 50 cm/h
Table 5 Percentage recovery of proteins in flow through fraction derived from MEP Hypercel chromatography conducted using pH adjusted CM eluate sample loadings ranging from 100 to 500 mL with and without addition of 50 mM NaCl to the sample and equilibration buffer. Recovery (%) Sample loading volume (without NaCl) Component
100 mL
IgG IgA IgM Albumin ␣2 -macroglobulin
0.01 0.3 1.2 84.9 5.2
200 mL 0.03 0.5 1.5 71.7 8.8
300 mL
400 mL
500 mL
0.04 1.2 1.8 99.6 19.8
0.05 1.6 2.6 97.7 22.2
0.06 1.8 3.6 98.9 35.4
0.4
0.8
0.3
pH 3.0 eluate fraction Albumin
1.4
0.9
Sample loading volume (with 50 mM NaCl) Component
100 mL
IgG IgA IgM Albumin ␣2 -macroglobulin
0.01 0.3 1.7 99.2 1.6
200 mL 0.03 1.2 1.6 103.3 8.3
300 mL 0.05 2.0 2.2 113.1 12.8
400 mL 0.09 1.6 2.9 106.8 21.0
500 mL 0.1 1.7 4.3 119.7 27.7
pH 3.0 eluate fraction Albumin
2.3
1.4
1.1
1.0
0.8
246
K.B. McCann et al. / J. Chromatogr. B 969 (2014) 241–248
Table 6 Percentage recovery of proteins in flow through fraction derived from MEP Hypercel conducted at flow rates ranging from 50 to 150 cm/h. Recovery (%) Flow rate (cm/h) Component IgG IgA IgM Albumin ␣2 -macroglobulin
50 0.01 0.8 0.1 91.3 0
100
150
0.02 2.4 2.3 95.8 5.3
0.04 12.4 1.6 96.2 11.3
Table 7 Characterisation testing of Albumex manufactured at Pilot-scale using MEP HyperCel. Control batches were obtained from manufacturing-scale process, which used Sephacryl S200HR chromatography for polishing. Data shown is the mean and standard deviation (in parentheses) from three independent batches. Test type
MEP Hypercel-purified batches
Sephacryl S200HR control batches
Aggregates (%) Dimer (%) Monomer (%) Fragments (%) IgA (g/mL) IgG (mg/mL) Transferrin (mg/mL) ␣2 -macroglobulin (mg/mL) Apolipoprotein A1 (mg/mL) ␣1 -glycoprotein (mg/mL) ␣1 -proteinase inhibitor (mg/mL) Haptoglobin (mg/mL) Inter-␣-trypsin inhibitor (IU/mL) Turbidity (NTU)
0.6 (0.35) 1.4 (0.40) 98.0 (0.75) 0 1.69 (0.71) < 0.009 < 0.020 < 0.024 < 0.003 < 0.012 0.039 (0.014) < 0.034 < 0.063 2.0 (0.32)
1.7 (0.45) 1.3 (0.21) 97.0 (0.59) 0 4.1 (2.13) 0.162 (0.107) < 0.020 < 0.024 0.039 (0.016) < 0.012 0.079 (0.014) < 0.034 < 0.063 5.6 (2.72)
less than 1% of each impurity protein was recovered in the flow through fraction (Table 6). Increasing the flow rate caused increased breakthrough of all impurities, but this was more pronounced for IgA and ␣2 -macroglobulin. Given the improved characteristics achieved for the albumin processed through the MEP HyperCel this process was evaluated at Pilot-scale. The Albumex derived from the pilot-scale batches incorporating MEP HyperCel was compared to control batches of Albumex manufactured using Sephacryl S200HR. The purity of the MEP HyperCel purified Albumex was higher than the control batches, containing significantly lower levels of IgG, IgA, apolipoprotein A1 and ␣1 -proteinase inhibitor (Table 7). The higher purity of the MEP HyperCel purified Albumex 20 was also reflected in the aggregate content and turbidity, both of which were lower than the control batches. 4. Discussion The ion-exchange chromatographic steps in the chromatographic albumin–Albumex process consisting of DEAE SepharoseFF and CM Sepharose-FF are an effective means of purifying human serum albumin. The feedstock (delipidated and euglobulin depleted Cohn SNI), which is loaded on to the initial DEAE SepharoseFF column, has an albumin purity of approximately 70%. This is increased to over 98% after processing through the two ion exchange columns. These ion exchange processes select proteins with pI values between 4.5 and 5.2. Prin et al. (1995) [23] showed that the isoelectric point (pI) ranges of IgG, IgA and IgM are broad and overlap the pI window selected for by the conditions used for the DEAE Sepharose-FF and CM Sepharose-FF columns in the Albumex process. Given this broad pI range of the immunoglobulins, ion-exchange chromatography cannot be expected to remove
all immunoglobulin contaminants from the albumin and thus highlights the need for an additional orthogonal mode of purification. Fortunately, the majority of the protein impurities in the ionexchange purified albumin have molecular weights significantly higher than albumin, thus gel filtration chromatography has served as an effective polishing step for albumin. It is known that these impurities can also be effectively removed from albumin by the traditional ethanol fractionation process. Albumin derived from the Cohn process has only minor levels of immunoglobulin and ␣2 macroglobulin contaminants. The majority of the gamma globulins (IgG, IgA and IgM) and some ␣2 - macroglobulin are recovered in the Fraction II + III, which is formed when the ethanol content of plasma is adjusted to approximately 25% [1]. As the precipitation of albumin requires an ethanol concentration of 40% [1] it suggests that there is a distinct difference in the solubility between albumin and these high molecular weight impurities. Although the precipitation of plasma proteins in ethanol solutions and the salting-out of plasma proteins occur by different mechanisms, similar trends are observed. With the Cohn method, fibrinogen is precipitated at 8% ethanol, the gamma globulins at 25% and albumin at 40% [1]. This follows the same order as the precipitation of plasma proteins during salting-out. Campbell & Hanna (1937) showed that when sodium sulphite was used for the precipitation of plasma proteins, a concentration of 12–12.5% was required for the complete precipitation of fibrinogen compared to the globulin fraction which required 21% [24]. Additionally, Cullen & Van Slyke (1920) demonstrated that if ammonium sulfate was used, the globulin fraction precipitated at half saturation, while the albumin fraction required a solution fully saturated with ammonium sulfate [25]. With Melander & Horvath (1977) subsequently noting that empirical observations typically show that the elution order from HIC media parallels the solubility of proteins, one would expect HIC media to offer an effective method to separate albumin from immunoglobulins [26]. The method scouting experiment for MEP HyperCel showed that the critical parameter for the separation of albumin and IgG on the MEP HyperCel was the equilibration and sample pH. The molarity and type of buffer did not appear to have any noticeable impact on separation at pH 7.0. These results align well with previous reports by Schwartz et al. (2001) who showed that a pH range of 7–8.5 resulted in the highest IgG binding capacity for MEP HyperCel [17]. As the albumin is eluted from the CM Sepharose-FF column at pH 5.5 in a 110 mM sodium acetate buffer, the albumin intermediate would only require pH adjustment to 7.0 prior to loading onto the MEP HyperCel column. The addition of sodium chloride to the sodium acetate buffer at pH 5.5 improved the selectivity of binding between IgG and albumin. Previous studies have shown that the IgG binding capacity of MEP HyperCel is not enhanced [17] or can even be reduced [27] by the addition of salts. However these previous studies were conducted using samples at neutral pH values, as compared to a pH 5.5 used in the current study. This is relatively close to the pKa of the MEP ligand at 4.8 thus the effect of salt addition must be dependent on the operating pH [16]. It was shown that the addition of 50 mM sodium chloride to the sodium acetate buffer at pH 7.0 did not improve the removal of impurity proteins from the CM Eluate sample. These results are more consistent with the results of Schwartz et al. (2001) who showed that the IgG binding capacity of MEP HyperCel to be independent of salt concentration at neutral pH [17]. Break through analysis performed with loading of pH adjusted CM Eluate onto MEP HyperCel showed significant differences in the binding of IgG, IgA, IgM and ␣2 -macroglobulin. The MEP HyperCel showed highest capacity for both IgG and IgA, and lower capacity for IgM and ␣2 - macroglobulin. These observations are consistent with those of Schwartz et al. (2001) who showed that the IgM
K.B. McCann et al. / J. Chromatogr. B 969 (2014) 241–248
binding capacity of MEP HyperCel was
Contents lists available at ScienceDirect
Journal of Chromatography B journal homepage: www.elsevier.com/locate/chromb
Use of mep HyperCel for polishing of human serum albumin Karl B. McCann ∗ , Yvonne Vucica, John Wu, Joseph Bertolini Research and Development Department, CSL Behring (Australia), 189-209 Camp Rd, Broadmeadows, Victoria 3047, Australia
a r t i c l e
i n f o
Article history: Received 1 May 2014 Received in revised form 21 August 2014 Accepted 22 August 2014 Available online 28 August 2014 Keywords: MEP HyperCel Human serum albumin Hydrophobic charge induction chromatography Plasma
a b s t r a c t The manufacture of human serum albumin by chromatographic procedures involves gel filtration chromatography as a final polishing step. Despite this step being essential to remove high molecular weight impurity proteins and thus ensure a stable and safe final product, it is relatively inefficient. This paper explores the use of hydrophobic charge induction chromatographic media, MEP HyperCel as an alternative to Sephacryl S200HR gel filtration for the polishing of human serum albumin derived by ion exchange chromatographic purification of Cohn Supernatant I. The use of MEP HyperCel results in a product with a higher purity than achieved with gel filtration and in a less time consuming manner and with potential resource savings. MEP HyperCel appears to have great potential for incorporation into downstream processes in the plasma fractionation industry as an efficient means of achieving polishing of intermediates or capture of proteins of interest. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Commercial human serum albumin is predominantly manufactured using ethanol precipitation methods based on those originally developed by Cohn et al. (1946) [1] and the subsequent modifications of Kistler & Nitschmann (1962) [2]. The continued purification of albumin by ethanol precipitation can be attributed to the fact that these methods allow the efficient processing the large volumes encountered in a modern plasma fractionation process. However, smaller but increasing amounts of albumin are fractionated using chromatographic processes, introduced either as a final polishing step following initial ethanol fractionation [3] or as fully integrated chromatographic processes [4–6]. Chromatographic purification of albumin has a number of benefits including higher process yields, higher purity [3,5,7] and improved tolerability and clinical safety of the product [8]. CSL Behring (Australia) have manufactured human serum albumin (Albumex) using a chromatography method based on that developed by Berglof et al. (1983) for almost two decades [9]. This chromatographic process, which includes two ion exchange and a size exclusion columns, was initially applied to delipidated Supernatant II + III, but was subsequently adapted to the processing of a
Abbreviations: ID, Internal diameter; ELISA, Enzyme linked immunosorbent assay; DEAE, Diethylaminoethyl; CM, Carboxymethyl; pI, Isoelectric point; HCIC, Hydrophobic charge induction chromatography. ∗ Corresponding author. Tel.: +61 3 9246 5493. E-mail address: [email protected] (K.B. McCann). http://dx.doi.org/10.1016/j.jchromb.2014.08.029 1570-0232/© 2014 Elsevier B.V. All rights reserved.
delipidated Supernatant I feedstream to allow the complete chromatographic purification of IgG for the Intragam P process [5,10]. Despite the above mentioned benefits, the widespread adoption of chromatographic processes for fractionation of human albumin has been limited due to cost and inefficiencies associated with the processes at industrial scale, with respect to the large number of chromatographic cycles required to process a batch. This is particularly the case in the fully chromatographic processes, with the Sephacryl S200HR gel filtration step for final polishing to remove high molecular weight impurity proteins. The efficiency of a chromatographic process for the purification of albumin could be greatly enhanced if an alternative mode to Sephacryl S200HR was developed as a final polishing step. There have been a number of reports that hydrophobic interaction chromatography (HIC) offers sufficient selectivity to allow separation of plasma proteins based on differences in the surface hydrophobicity of the proteins [11,12]. This has allowed HIC to be used successfully for the purification of albumin from human plasma [13], human placenta [14] and of recombinant human serum albumin from Pichia pastoris [15]. Despite HIC offering an orthogonal mode of separation to the ion exchange steps used for chromatographic purification of albumin, these resins have not been used by any of the commercial plasma fractionators. Their use at large-scale has been limited by the fact that HIC-based chromatography requires use of significant amounts of lyotropic salts in order to achieve the binding and fractionation of proteins [16]. An alternative to conventional HIC is hydrophobic charge induction chromatography (HCIC). One such resin for HCIC is 4-mercaptoethylpyridine (MEP) HyperCel [16]. The
242
K.B. McCann et al. / J. Chromatogr. B 969 (2014) 241–248
MEP ligand permits separation based on the surface hydrophobicity of proteins but does not require the addition of lyotropic salts [17]. MEP HyperCel has been shown to have a high selectivity for immunoglobulins and for this reason there has been a major focus on the use of this media for the isolation of monoclonal antibodies [17–20]. In addition, MEP HyperCel has been proven to be effective for the separation of penicillin acylase [21] and Fc-fusion proteins [22]. In this paper an alternative method for the polishing of human serum albumin using MEP HyperCel is investigated. IgG is one of the major residual proteins present in chromatographically purified albumin. Therefore the optimal conditions for MEP HyperCel chromatography for the partitioning of albumin and IgG were initially evaluated using purified samples of albumin and IgG. The defined conditions were then verified by examining the capability of MEP HyperCel to polish the ion exchange purified albumin process intermediate. The impact of flow rate on the achieved separation and optimal protein sample loading was also determined. After successful demonstration of the new process at small-scale, the process was evaluated at Pilot-scale and the final product was compared to the albumin produced at industrial-scale. 2. Materials and methods 2.1. Protein solutions Process intermediates including Cohn Supernatant I (SNI) depleted of lipids and euglobulins, albumin eluate from CM Sepharose-FF chromatography, purified human serum albumin (HSA) and purified human immunoglobulin G (IgG) and purified albumin from Sephacryl S200HR chromatography were obtained from CSL Behring (Broadmeadows, Victoria, Australia). All protein solutions were filtered through a 0.22 m membrane (Durapore® , Millipore Corporation, Bedford, MA, USA) prior to chromatography. 2.2. Antibodies and ELISA reagents Antibodies against the human plasma proteins of transferrin, ␣2 -macroglobulin, ␣1 -glycoprotein, ␣1 -antitrypsin, haptoglobin and ceruloplasmin were obtained from DakoCytomation Denmark A/S (Glostrup, Denmark), inter ␣ trypsin inhibitor antibodies were obtained from The Binding Site (Birmingham, England), and apolipoprotein A1 and apolipoprotein B antibodies were obtained from Siemens Healthcare Diagnostics Inc. (Tarrytown, NY, USA). ELISA kits for IgG and IgM were obtained from Bethyl Laboratories Inc. (Montgomery, TX, USA) and the ELISA kit for ␣2 -macroglobulin was obtained from GenWay Biotech. Inc. (San Diego, CA, USA). Antibodies for the IgA ELISA were obtained from DakoCytomation Denmark A/S (Glostrup, Denmark) and the IgA standard was obtained from Calbiochem-Novabiochem Corporation (San Diego, CA, USA). Antibodies against the human plasma proteins of albumin, IgG, IgA and IgM, which were used for nephelometry analysis, were obtained from Beckman-Coulter Inc. (Fullerton, CA, USA). 2.3. Chromatographic resins MEP HyperCel resin was obtained from Pall Life Sciences (East Hills, NY, USA). DEAE Sepharose-FF, Capto DEAE, CM Sepharose-FF and Sephacryl S200HR resins were obtained from GE Healthcare BioSciences AB (Uppsala, Sweden). 2.4. Characterisation of Sephacryl S200HR process The albumin intermediates before and after Sephacryl S200HR chromatography were characterised using nephelometry
(IMMAGE Immunochemistry system, Beckman-Coulter Inc., Fullerton, CA, USA) to determine the levels of albumin, IgG, IgA and IgM, and immunoelectrophoresis to determine the levels of transferrin, ␣2 -macroglobulin, ␣1 -glycoprotein, ␣1 -antitrypsin, inter ␣ trypsin inhibitor, haptoglobin, apolipoprotein A1, apolipoprotein B and ceruloplasmin.
2.5. Impact of heating on turbidity of albumin solutions The excellent viral safety record of commercial HSA solutions is attributed largely to the pasteurisation step (60 ◦ C for 10 h) [3]. The albumin is stabilised by addition of sodium caprylate, however, contaminant proteins present not protected by the caprylate are susceptible to heat induced denaturation. Thus evaluation of the turbidity of an albumin solution following pasteurisation is a good indication of the presence of residual proteins. A control sample of albumin eluate from CM Sepharose-FF chromatography and samples of purified albumin derived from Sephacryl S200HR chromatography were concentrated to 15–20% w/v protein by ultrafiltration (10 kDa OMEGA membrane; Pall Corporation, Hauppauge, NY, USA) and formulated with 32 mM sodium caprylate. The samples were pasteurised by heating at 60 ◦ C for 10 h. At the completion of pasteurisation the turbidity of the samples was assessed using a Hach 2100AN Turbidimeter (Hach Company, Loveland, CO, USA).
2.6. Determination of MEP HyperCel operating conditions for partitioning of albumin and IgG The operating parameters for MEP HyperCel chromatography were explored using purified samples of albumin and IgG. The binding and elution of albumin and IgG was determined separately using a range of equilibration conditions including; 50 mM sodium phosphate (pH 7.0); 50 mM sodium acetate (pH 5.5); 50 mM sodium acetate (pH 5.5) + 150 mM NaCl; 50 mM sodium acetate (pH 5.5) + 300 mM NaCl; 50 mM sodium acetate (pH 7.0); and 110 mM sodium acetate (pH 7.0), shown in Table 3. The pure IgG and pure albumin samples were prepared in the appropriate equilibration buffer at a concentration of approximately 5-10 mg/mL and filtered through a 0.22 m membrane (Durapore® , Millipore Corporation, Bedford, MA, USA) prior to chromatography. The samples were loaded on to an equilibrated MEP HyperCel column (17.5 cm × 1.6 cm ID; GE Healthcare BioSciences AB, Uppsala, Sweden) at 20 g per litre of resin. The unbound protein was recovered during the sample loading period and during the subsequent post-sample loading wash with equilibration buffer. Any bound protein was eluted using sodium acetate buffer at pH 3.0. All chromatography steps were conducted at a flow rate of 100 cm/h and monitored at 280 nm using an AKTA Explorer 100 equipped with Unicorn software (GE Healthcare BioSciences AB, Uppsala, Sweden). The amount of protein recovered in the flow through and eluate fractions was determined using nephelometry. The most suitable equilibration buffer conditions defined from the above experiment was subsequently verified using a sample containing a 9:1 mixture of albumin and IgG. The pure albumin and pure IgG was mixed in 110 mM sodium acetate (pH 7.0) at a ratio of 9:1 to give a final protein concentration of approximately 5–10 mg/mL. The sample was loaded on to the equilibrated MEP HyperCel column at protein loadings of 100, 200 and 300 g per litre of resin. The unbound protein was recovered using 110 mM Sodium acetate (pH 7.0) equilibration buffer and the bound protein was eluted using sodium acetate buffer at pH 3.0. The amount of albumin and IgG in the flow through and eluate fractions was determined using nephelometry.
K.B. McCann et al. / J. Chromatogr. B 969 (2014) 241–248
2.7. Polishing of ion exchange purified albumin using MEP HyperCel After demonstrating that the MEP HyperCel could successfully separate an albumin and IgG mixture, the ability of MEP HyperCel to purify the ion exchange purified albumin was assessed. The albumin eluate from CM Sepharose-FF chromatography was adjusted to pH 7.0 ± 0.1 and clarified by passage through a 0.22 m filter (Durapore® , Millipore Corporation, Bedford, MA, USA) prior to loading on to the MEP HyperCel column. The albumin sample (1000 mL) was loaded on to the MEP HyperCel column (17.5 cm × 1.6 cm ID), which had been equilibrated with 110 mM sodium acetate (pH 7.0), followed by a 110 mM sodium acetate (pH 7.0) post-sample wash to recover unbound protein. Bound protein was eluted with 100 mM sodium acetate (pH 3.0) and the resin was cleaned in place with 1 M NaOH. The flow through fraction was collected at 10 mL intervals and assessed for the levels of the major impurity proteins of IgG, IgA, IgM and ␣2 -macroglobulin by ELISA. The initial analysis of the impurity protein break through curves showed that sample loadings of up to 1000 mL on to a 35 mL MEP HyperCel column were excessive. The albumin sample was loaded on to the equilibrated MEP HyperCel column at 100, 200, 300, 400 and 500 mL. The flow through and post-sample wash was collected as a bulk sample along with the pH 3.0 eluate. These bulk fractions were assessed for the levels of albumin by nephelometry and the major impurity proteins of IgG, IgA, IgM and ␣2 -macroglobulin by ELISA. In attempt to improve the selectivity of the MEP HyperCel for binding of residual proteins, the above experiment was repeated with 50 mM NaCl added to the equilibration buffer and albumin sample. Under both buffer conditions the effect of sample loading was examined. The albumin sample was loaded on to the equilibrated MEP HyperCel column at 100, 200, 300, 400 and 500 mL. The flow through and post-sample wash was collected as a bulk sample along with the pH 3.0 eluate. These bulk fractions were assessed for the levels of albumin by nephelometry and the major impurity proteins of IgG, IgA, IgM and ␣2 -macroglobulin by ELISA. 2.8. Assessment of flow rate on performance of MEP HyperCel The four major impurity proteins appeared to have differing affinities to the MEP HyperCel resin. Therefore studies were conducted to investigate the impact of the flow rate on binding. Using the chromatography conditions from above and with buffers without the addition of 50 mM NaCl, the albumin sample (100 mL) was loaded on to the MEP HyperCel column at flow rates of 50, 100 and 150 cm/h. The flow through and eluate fractions were collected as bulk samples and assessed for the levels of albumin by nephelometry and the major impurity proteins of IgG, IgA, IgM and ␣2 -macroglobulin by ELISA. 2.9. Manufacture of albumin using MEP HyperCel chromatography at pilot-scale As the small-scale studies showed that the MEP HyperCel could be successfully used for the polishing of chromatographically purified albumin, the reproducibility of the new process scheme was evaluated at pilot-scale over three independent batches. For each batch a 15 kg sample of Cohn SNI, which was delipidated and depleted of euglobulins, was purified on the ion exchange columns (17.5 cm × 10 cm ID; GE Healthcare BioSciences AB, Uppsala, Sweden). The albumin eluate from CM Sepharose-FF chromatography was adjusted to pH 7.0 and clarified by passage through a 0.22 m membrane (Durapore® , Millipore Corporation, Bedford, MA, USA). The albumin sample was loaded on to an equilibrated MEP HyperCel column (17.5 cm × 10 cm ID; GE Healthcare
243
Table 1 Characterisation of albumin sample before and after Sephacryl S200HR chromatography.
Component Albumin IgG IgA IgM ␣2 -macroglobulin ␣1 -antitrypsin Apolipoprotein A1 Total
Albumin before Sephacryl S200HRa
Albumin after Sephacryl S200HR
Clearance across S200HR
% of total 98.58 0.37 0.26 0.08 0.62 0.05 0.02 1.42%
% of total 99.81 0.096 0.008 0.004 0.017 0.029 0.034 0.19%
% NA 74.0 97.0 94.0 97.3 42.0 −46.7
a CM Sepharose-FF eluate. NA = Not Applicable.
Table 2 Effect of pasteurisation on turbidity of albumin derived from ion exchange chromatography and Sephacryl S200HR chromatography. Turbidity (NTU) Sample
Before pasteurisation
After pasteurisation
Sephacryl S200HR eluate Ion exchange eluate
2.23 2.75
2.36 9.36
BioSciences AB, Uppsala, Sweden) at a protein loading of 80 g/L and a flow rate of 100 cm/h. The purified albumin flow through was further processed according to the current Albumex process. The final Albumex product for the three independent batches was characterised for protein composition, turbidity levels and a range of impurity proteins including; IgG, IgA, transferrin, ␣2 macroglobulin, ␣1 -glycoprotein, ␣1 -antitrypsin, inter ␣ trypsin inhibitor, haptoglobin, apolipoprotein A1, apolipoprotein B and ceruloplasmin. The results were compared to three control batches of Albumex, which were manufactured at full-scale using the DEAE Sepharose-FF, CM Sepharose-FF and Sephacryl S200HR chromatography steps. 3. Results Before commencing MEP HyperCel studies the effectiveness of the current Sephacryl S200HR purification was characterised in order to identify the major contaminant proteins in the albumin intermediate. The concentrated ion exchange albumin intermediate which was loaded onto the Sephacryl S200HR contained 1.42% impurity proteins, with ␣2 - macroglobulin, IgG and IgA predominant (Table 1). These major impurity proteins all have molecular weights significantly higher than that of albumin, thus making the Sephacryl S200HR gel filtration suitable for their removal. Clearance rates were greater than 94% for ␣2 - macroglobulin, IgA and IgM, and greater than 74.0% for IgG. Following Sephacryl S200HR chromatography the total impurity content had been reduced to 0.19%, with IgG being the most abundant impurity protein present. Despite the albumin eluate derived from ion exchange chromatography having a purity of over 98% and thus complying with the albumin Pharmacopoeial requirements for protein composition and purity, the removal of high molecular weight impurities is critical for the quality of the final albumin product. This is clearly demonstrated by the fact that when the albumin derived from ion exchange chromatography is concentrated, formulated with sodium caprylate, and subjected to pasteurisation at 60 ◦ C for 10 h a significant increase in turbidity is observed (Table 2). No turbidity increase was encountered when the purified albumin derived from Sephacryl S200HR is pasteurised.
244
K.B. McCann et al. / J. Chromatogr. B 969 (2014) 241–248
Table 3 Percentage recovery of albumin and IgG in flow through and eluate fractions collected from MEP Hypercel chromatography using different equilibration buffers. Recovery (%) Albumin Buffer conditions 50 mM sodium phosphate (pH 7.0) 50 mM sodium acetate (pH 5.5) 50 mM sodium acetate (pH 5.5) + 150 mM NaCl 50 mM sodium acetate (pH 5.5) + 300 mM NaCl 50 mM sodium acetate (pH 7.0) 110 mM sodium acetate (pH 7.0)
Flow through 86.9
Table 4 Percentage recovery of albumin and IgG in flow through and eluate fractions collected from MEP Hypercel chromatography at different protein loadings of a 9:1 albumin:IgG mixture. Recovery (%)
IgG Eluate 5.3
Flow through
Eluate
ND
63.9
44.7
39.6
42.1
68.1
79.9
17.3
ND
97.3
81.1
14.6
ND
90.9
104.5
5.2
ND
96.0
89.2
5.1
ND
90.5
ND = Not detectable.
IgG as one of the major impurities in the ion-exchange albumin intermediate and also post-Sephacryl S200HR chromatography. Therefore the initial development work on the MEP HyperCel chromatography for polishing of the albumin intermediate focussed on optimising the ability to separate albumin from IgG. Initially the binding and elution characteristics of pure human albumin and human IgG to MEP HyperCel was examined using sodium phosphate (pH 7.0) and sodium acetate (pH 5.5) equilibration buffers. Previous studies by Schwartz et al. (2001) showed that MEP HyperCel has a maximum binding capacity when the sample pH is adjusted to approximately pH 7.0 [17]. The separation of albumin and IgG using MEP HyperCel was also evaluated at pH 5.5 as the albumin is currently eluted from the CM Sepharose-FF column at pH 5.5 using a sodium acetate buffer. With equilibration of the MEP HyperCel with sodium phosphate buffer, pH 7.0 effective binding of IgG occurred, with undetectable levels of IgG being recovered in the flow through fraction (Table 3). Only minor amounts of albumin bound to the MEP HyperCel, and greater than 85% of the albumin recovered in the flow through fraction. Equilibration with the sodium acetate buffer, pH 5.5 resulted in half of both the IgG and albumin binding to the resin. The addition of sodium chloride to the sodium acetate buffer, pH 5.5 improved IgG binding, with undetectable IgG levels being recovered in the flow through fraction (Table 3). There was also improved albumin recovery in the drop through fraction. However, approximately 15–20% of the albumin was bound and recovered in the eluate fraction. This was significantly higher binding than when the MEP HyperCel was equilibrated with sodium phosphate buffer, pH 7.0. Given that sodium acetate buffers are currently used in the chromatographic albumin process during the ion exchange chromatography steps, the binding and elution characteristics of purified albumin and IgG to the MEP HyperCel resin was assessed using sodium acetate buffers at pH 7.0. Using 50 mM and 110 mM sodium acetate buffers at pH 7.0 resulted in similar IgG and albumin recoveries to that observed with the 50 mM sodium phosphate buffer, pH 7.0. There were undetectable levels of IgG in the flow through fraction and only minor amounts of albumin being bound with approximately 5% of total being detected in the eluate (Table 3). The results suggest that the pH of the equilibration buffer and sample is the critical parameter that influences binding to the MEP HyperCel resin, with buffer type and conductivity of the equilibration buffer and sample having little impact.
Albumin
IgG
Loading (g Protein/L resin)
Flow through
Eluate
Flow through
Eluate
100 g/L 200 g/L 300 g/L
92.0 90.0 98.2
1.8 1.3 0.9
ND ND 1.4
92.1 86.9 86.2
ND = Not detectable.
As the MEP HyperCel showed good selectivity between albumin and IgG if equilibrated with 110 mM sodium acetate buffer, pH 7.0, this buffer was selected for further evaluation with an IgG and albumin mixture and ion exchange albumin intermediate. In the current chromatographic albumin process, albumin is eluted from the CM Sepharose-FF column with 110 mM sodium acetate buffer, pH 5.5. Thus albumin eluted from the CM Sepharose-FF column would only require a pH adjustment step prior to loading on to the MEP HyperCel column. The separation characteristics of MEP HyperCel were evaluated using a 9:1 mixture of albumin and IgG. The MEP HyperCel showed excellent separation of the albumin and IgG mixture. At protein loadings up to 200 g/L, the flow through fraction contained undetectable levels of IgG and over 90% of the total albumin loaded (Table 4). Most of the IgG was recovered in the eluate fraction and only minor amounts of albumin were lost in this fraction. The results suggest that the MEP HyperCel can remove significant amounts of IgG from an albumin solution that has a similar composition to the ion exchange eluate in the chromatographic albumin process. The MEP HyperCel was further evaluated by assessing the ability to purify the albumin eluate from the CM Sepharose-FF column. As shown previously in Table 1, the CM Sepharose-FF elute contains various protein impurities including ␣2 -macroglobulin, IgA, IgM and IgG. These proteins were used as purity markers to evaluate the capability of MEP HyperCel to further purify the CM Sepharose-FF eluate albumin intermediate to final product. CM Eluate (1000 mL) was adjusted to pH 7.0 and loaded on to the MEP HyperCel column. The flow through was collected in 10 mL fractions and tested for the marker proteins. The impurity break through profiles varied significantly for the four impurity proteins (Fig. 1). During the initial stages of sample loading only minor amounts of IgG were detected in the flow through (Fig. 1a). Despite a significant increase in IgG break through during the latter stages of the loading, the maximum IgG concentration detected only represented 1.1% of the IgG concentration in the sample load, suggesting that the MEP HyperCel has a high selectivity for IgG. The break through profile for IgA followed an almost linear increase throughout the sample loading (Fig. 1b). The maximum IgA concentration detected in the flow through represented 9.3% of the IgA concentration in the sample load. Also during the initial stages of the sample loading, only minor amounts of IgM were detected in the flow through (Fig. 1c). Towards the latter stages of the sample loading a significant increase in IgM was detected in the flow through. The maximum IgM concentration detected at the later stages of sample loading represented 53.1% of the IgM concentration of the sample load, suggesting that the capacity or affinity of MEP HyperCel for IgM is significantly lower than for IgG and IgA. Significant amounts of ␣2 -macroglobulin were detected in the flow through immediately after commencement of sample loading on to the MEP HyperCel column (Fig. 1d). The rate of increase in ␣2 -macroglobulin breakthrough was faster than any of the
K.B. McCann et al. / J. Chromatogr. B 969 (2014) 241–248
245
Fig. 1. Impurity break through profiles during MEP Hypercel purification of pH adjusted CM eluate (a) IgG, (b) IgA and (c) IgM and (d) ␣2 - macroglobulin.
other impurities and as a result the maximum ␣2 -macroglobulin concentration detected in the flow through was 76.0% of the ␣2 macroglobulin in the sample loading. As the ␣2 -macroglobulin is the most abundant impurity protein in the CM eluate and the MEP HyperCel has the weakest affinity for this protein, it indicates that ␣2 -macroglobulin will be the key impurity marker for subsequent determination of allowed sample loading onto the MEP HyperCel column. Loading onto MEP HyperCel column of pH adjusted CM eluate was examined over the range of 100–500 mL. Partitioning of the impurity proteins were consistent with the impurity break through profiles shown in Fig. 1. For IgG, recovery in the flow through fraction was less than 1% of the total IgG loaded at the highest sample loading (Table 5). At the highest sample loadings the recovery of IgA and IgM was also relatively low at 1.8% and 3.6%, respectively. However, even at the lowest sample loading of 100 mL, 5.2% of the total ␣2 -macroglobulin loaded on to the MEP HyperCel was recovered in the flow through fraction. ␣2 -macroglobulin break through increased proportionally with sample load, with 35.4% of the total ␣2 -macroglobulin being recovered in the flow through at the highest loading. As ␣2 -macroglobulin is the predominant impurity protein in the CM eluate and the MEP HyperCel exhibits the weakest affinity to ␣2 -macroglobulin, further studies were made to optimise the ␣2 -macroglobulin clearance by MEP HyperCel. Earlier data (Table 3) showed that addition of NaCl to 50 mM sodium acetate, pH 5.5 buffer improved albumin and IgG partitioning by MEP HyperCel. Therefore the impact of 50 mM sodium chloride addition to the pH adjusted CM eluate and equilibration buffer at pH 7.0 on the removal of impurities from albumin at various loadings was assessed. The recovery of IgG, IgA, IgM and ␣2 -macroglobulin in the flow through fractions was similar for the MEP HyperCel separations conducted with and without addition of 50 mM sodium chloride (Table 5). This indicated that the addition of sodium chloride did not have any impact on the selectivity of MEP HyperCel for the impurity proteins found in the CM eluate albumin
intermediate. The albumin recovered in the eluate fraction of these runs ranged between 0.3–2.3% indicating minimal albumin loss (Table 5). The albumin product recovery obtained across the MEP HyperCel compared well with those achieved across the Sephacryl S200HR column, where 3-5% losses are typically observed. The impact of flow rate on the separation of impurity proteins by MEP HyperCel was assessed between 50 and 150 cm/h using a sample loading of 100 mL of pH adjusted CM eluate. At 50 cm/h
Table 5 Percentage recovery of proteins in flow through fraction derived from MEP Hypercel chromatography conducted using pH adjusted CM eluate sample loadings ranging from 100 to 500 mL with and without addition of 50 mM NaCl to the sample and equilibration buffer. Recovery (%) Sample loading volume (without NaCl) Component
100 mL
IgG IgA IgM Albumin ␣2 -macroglobulin
0.01 0.3 1.2 84.9 5.2
200 mL 0.03 0.5 1.5 71.7 8.8
300 mL
400 mL
500 mL
0.04 1.2 1.8 99.6 19.8
0.05 1.6 2.6 97.7 22.2
0.06 1.8 3.6 98.9 35.4
0.4
0.8
0.3
pH 3.0 eluate fraction Albumin
1.4
0.9
Sample loading volume (with 50 mM NaCl) Component
100 mL
IgG IgA IgM Albumin ␣2 -macroglobulin
0.01 0.3 1.7 99.2 1.6
200 mL 0.03 1.2 1.6 103.3 8.3
300 mL 0.05 2.0 2.2 113.1 12.8
400 mL 0.09 1.6 2.9 106.8 21.0
500 mL 0.1 1.7 4.3 119.7 27.7
pH 3.0 eluate fraction Albumin
2.3
1.4
1.1
1.0
0.8
246
K.B. McCann et al. / J. Chromatogr. B 969 (2014) 241–248
Table 6 Percentage recovery of proteins in flow through fraction derived from MEP Hypercel conducted at flow rates ranging from 50 to 150 cm/h. Recovery (%) Flow rate (cm/h) Component IgG IgA IgM Albumin ␣2 -macroglobulin
50 0.01 0.8 0.1 91.3 0
100
150
0.02 2.4 2.3 95.8 5.3
0.04 12.4 1.6 96.2 11.3
Table 7 Characterisation testing of Albumex manufactured at Pilot-scale using MEP HyperCel. Control batches were obtained from manufacturing-scale process, which used Sephacryl S200HR chromatography for polishing. Data shown is the mean and standard deviation (in parentheses) from three independent batches. Test type
MEP Hypercel-purified batches
Sephacryl S200HR control batches
Aggregates (%) Dimer (%) Monomer (%) Fragments (%) IgA (g/mL) IgG (mg/mL) Transferrin (mg/mL) ␣2 -macroglobulin (mg/mL) Apolipoprotein A1 (mg/mL) ␣1 -glycoprotein (mg/mL) ␣1 -proteinase inhibitor (mg/mL) Haptoglobin (mg/mL) Inter-␣-trypsin inhibitor (IU/mL) Turbidity (NTU)
0.6 (0.35) 1.4 (0.40) 98.0 (0.75) 0 1.69 (0.71) < 0.009 < 0.020 < 0.024 < 0.003 < 0.012 0.039 (0.014) < 0.034 < 0.063 2.0 (0.32)
1.7 (0.45) 1.3 (0.21) 97.0 (0.59) 0 4.1 (2.13) 0.162 (0.107) < 0.020 < 0.024 0.039 (0.016) < 0.012 0.079 (0.014) < 0.034 < 0.063 5.6 (2.72)
less than 1% of each impurity protein was recovered in the flow through fraction (Table 6). Increasing the flow rate caused increased breakthrough of all impurities, but this was more pronounced for IgA and ␣2 -macroglobulin. Given the improved characteristics achieved for the albumin processed through the MEP HyperCel this process was evaluated at Pilot-scale. The Albumex derived from the pilot-scale batches incorporating MEP HyperCel was compared to control batches of Albumex manufactured using Sephacryl S200HR. The purity of the MEP HyperCel purified Albumex was higher than the control batches, containing significantly lower levels of IgG, IgA, apolipoprotein A1 and ␣1 -proteinase inhibitor (Table 7). The higher purity of the MEP HyperCel purified Albumex 20 was also reflected in the aggregate content and turbidity, both of which were lower than the control batches. 4. Discussion The ion-exchange chromatographic steps in the chromatographic albumin–Albumex process consisting of DEAE SepharoseFF and CM Sepharose-FF are an effective means of purifying human serum albumin. The feedstock (delipidated and euglobulin depleted Cohn SNI), which is loaded on to the initial DEAE SepharoseFF column, has an albumin purity of approximately 70%. This is increased to over 98% after processing through the two ion exchange columns. These ion exchange processes select proteins with pI values between 4.5 and 5.2. Prin et al. (1995) [23] showed that the isoelectric point (pI) ranges of IgG, IgA and IgM are broad and overlap the pI window selected for by the conditions used for the DEAE Sepharose-FF and CM Sepharose-FF columns in the Albumex process. Given this broad pI range of the immunoglobulins, ion-exchange chromatography cannot be expected to remove
all immunoglobulin contaminants from the albumin and thus highlights the need for an additional orthogonal mode of purification. Fortunately, the majority of the protein impurities in the ionexchange purified albumin have molecular weights significantly higher than albumin, thus gel filtration chromatography has served as an effective polishing step for albumin. It is known that these impurities can also be effectively removed from albumin by the traditional ethanol fractionation process. Albumin derived from the Cohn process has only minor levels of immunoglobulin and ␣2 macroglobulin contaminants. The majority of the gamma globulins (IgG, IgA and IgM) and some ␣2 - macroglobulin are recovered in the Fraction II + III, which is formed when the ethanol content of plasma is adjusted to approximately 25% [1]. As the precipitation of albumin requires an ethanol concentration of 40% [1] it suggests that there is a distinct difference in the solubility between albumin and these high molecular weight impurities. Although the precipitation of plasma proteins in ethanol solutions and the salting-out of plasma proteins occur by different mechanisms, similar trends are observed. With the Cohn method, fibrinogen is precipitated at 8% ethanol, the gamma globulins at 25% and albumin at 40% [1]. This follows the same order as the precipitation of plasma proteins during salting-out. Campbell & Hanna (1937) showed that when sodium sulphite was used for the precipitation of plasma proteins, a concentration of 12–12.5% was required for the complete precipitation of fibrinogen compared to the globulin fraction which required 21% [24]. Additionally, Cullen & Van Slyke (1920) demonstrated that if ammonium sulfate was used, the globulin fraction precipitated at half saturation, while the albumin fraction required a solution fully saturated with ammonium sulfate [25]. With Melander & Horvath (1977) subsequently noting that empirical observations typically show that the elution order from HIC media parallels the solubility of proteins, one would expect HIC media to offer an effective method to separate albumin from immunoglobulins [26]. The method scouting experiment for MEP HyperCel showed that the critical parameter for the separation of albumin and IgG on the MEP HyperCel was the equilibration and sample pH. The molarity and type of buffer did not appear to have any noticeable impact on separation at pH 7.0. These results align well with previous reports by Schwartz et al. (2001) who showed that a pH range of 7–8.5 resulted in the highest IgG binding capacity for MEP HyperCel [17]. As the albumin is eluted from the CM Sepharose-FF column at pH 5.5 in a 110 mM sodium acetate buffer, the albumin intermediate would only require pH adjustment to 7.0 prior to loading onto the MEP HyperCel column. The addition of sodium chloride to the sodium acetate buffer at pH 5.5 improved the selectivity of binding between IgG and albumin. Previous studies have shown that the IgG binding capacity of MEP HyperCel is not enhanced [17] or can even be reduced [27] by the addition of salts. However these previous studies were conducted using samples at neutral pH values, as compared to a pH 5.5 used in the current study. This is relatively close to the pKa of the MEP ligand at 4.8 thus the effect of salt addition must be dependent on the operating pH [16]. It was shown that the addition of 50 mM sodium chloride to the sodium acetate buffer at pH 7.0 did not improve the removal of impurity proteins from the CM Eluate sample. These results are more consistent with the results of Schwartz et al. (2001) who showed that the IgG binding capacity of MEP HyperCel to be independent of salt concentration at neutral pH [17]. Break through analysis performed with loading of pH adjusted CM Eluate onto MEP HyperCel showed significant differences in the binding of IgG, IgA, IgM and ␣2 -macroglobulin. The MEP HyperCel showed highest capacity for both IgG and IgA, and lower capacity for IgM and ␣2 - macroglobulin. These observations are consistent with those of Schwartz et al. (2001) who showed that the IgM
K.B. McCann et al. / J. Chromatogr. B 969 (2014) 241–248
binding capacity of MEP HyperCel was
