Journal of Chromatography A, 1368 (2014) 155–162
Contents lists available at ScienceDirect
Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma
Two-stage chromatographic separation of aggregates for monoclonal antibody therapeutics Vijesh Kumar, Anurag S. Rathore ∗ Department of Chemical Engineering, Indian Institute of Technology Delhi, Hauz Khas, New Delhi 110016, India
a r t i c l e
i n f o
Article history: Received 5 June 2014 Received in revised form 17 September 2014 Accepted 27 September 2014 Available online 6 October 2014 Keywords: Monoclonal antibodies Purification Aggregates Cation exchange chromatography Hydrophobic interaction chromatography
a b s t r a c t Aggregates of monoclonal antibody (mAb) therapeutics, due to their perceived impact on immunogenicity, are recognized as a critical quality attribute by the regulatory authorities as well as the industry. Hence, removal of aggregates is a key objective of bioprocessing. At present, this is achieved by a combination of two or more orthogonal chromatographic steps with possible modalities of ion exchange, hydrophobic interaction and mixed mode. A two-stage chromatographic purification process consisting of ion-exchange and hydrophobic interaction modes is proposed in this paper for effective and efficient control of aggregates for a mAb therapeutic. The proposed scheme does not require any intermediate processing of the process stream. Further, baseline separation is achieved for monomer and aggregates resulting in robust performance. This was possible because the proposed operational scheme allowed for an addition of selectivities of the two chromatography modes vs. the traditional two column scheme where part of the separation of aggregates achieved by the first column is lost upon pooling. The proposed process scheme yielded improved separation of aggregates (0% vs. 1–2%) at >95% recovery and reduced overall process time (6 h vs. 14 h) for a typical application. Further, clearance of host cell proteins was also shown to have improved with the suggested process scheme. Successful implementation of the proposed scheme has been demonstrated for two different monoclonal antibody therapeutic products. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Monoclonal antibodies (mAbs) constitute an increasingly larger share of biotech therapeutics for diseases like cancer, rheumatoid arthritis, psoriasis and multiple sclerosis [1,2]. Due to the complexity of these products and the large amounts that need to be produced, last two decades have witnessed major efforts toward improving the quality of these products and reducing the cost of manufacturing [3,4]. These products are typically expressed in mammalian cell culture followed by a multistep purification process such that the different steps work on different separation mechanisms and thus together are able to provide the product which meets the stringent purity levels expected of therapeutic products. Downstream processing of mAbs is generally based on a platform process which includes Protein A affinity chromatography as a capture step, followed by one or more polishing steps [5]. The
∗ Corresponding author. Tel.: +91 11 26591098; fax: +91 1126581120; mob: +91 9650770650. E-mail address: [email protected] (A.S. Rathore). URL: http://www.biotechCMZ.com (A.S. Rathore). http://dx.doi.org/10.1016/j.chroma.2014.09.077 0021-9673/© 2014 Elsevier B.V. All rights reserved.
latter could be an ion exchange or hydrophobic interaction or mixed mode chromatography. The primary impurities that require clearance during bioprocessing are the host cell proteins (HCP), host cell DNA, and aggregates [6]. Protein aggregation is an undesirable event that occurs during manufacture of therapeutic mAbs [7,8]. As mentioned above, presence of even trace amounts of aggregates is considered to be undesirable due to the concern of adverse immunological response [9]. Cation exchange chromatography (CEX) is one of the polishing steps used for achieving effective removal of protein aggregates for mAbs. It involves binding of a protein under pre-equilibrated low salt condition followed by a buffer wash and elution of product either by step or linear addition of salt or by altering the pH [10]. Product related impurities such as mAb aggregates can be more effectively separated by linear salt gradient. Aggregate species in the CEX elution pool generally elute out in the later fractions due to their high avidity compared to the monomer species [11,13]. To get the desired purity, elution can be fractionated and pooled according to the target aggregate content. Alternatively, elution of aggregate species can be controlled by changing from a linear gradient to constant salt at the concentration where elution of the aggregate species is expected. Key factors that impact aggregate clearance on CEX include, resin ligand chemistry,
156
V. Kumar, A.S. Rathore / J. Chromatogr. A 1368 (2014) 155–162
sample loading, binding and elution conditions, and pooling criteria [11–15]. Hydrophobic interaction chromatography (HIC) is another major mode of chromatography for achieving effective clearance of aggregates [16–18]. More recently, HIC membrane adsorbers have also been successfully used for clearance of these aggregates [19]. Protein binding in HIC is achieved under high salt condition and elution is performed by step or linear gradient reduction in salt concentration. Removal of protein aggregates in HIC is also achieved in flow-through mode in addition to bind elute mode, where the salt condition is just enough to bind the more hydrophobic variant while the product flows through. In this mode, high protein load capacity can be achieved. However, addition of high concentrations of salt to the product to promote protein binding in HIC can be detrimental to the product itself [20]. Process parameters and raw material attributes that have been known to impact the clearance in HIC include resin ligand, lyotropic salts used in binding and pH [17]. Almost always, both CEX and HIC steps are required to get the desired clearance of aggregates. The two steps are generally performed one after the other, thus requiring more process time. In this paper, we propose use of a two-stage chromatography process for highly efficient and high throughput separation of aggregates for mAb therapeutic products. This has been achieved by selecting operating conditions such that the elution from the CEX step, operated in a bind and elute mode, can be directly loaded on the HIC column with the HIC column operated in the flow through mode connected in series. The proposed configuration results in optimal and robust removal of the aggregate species. 2. Materials and methods 2.1. Instruments and materials 2.1.1. mAb samples Two human IgG1 mAbs (mAb A and mAb B) were donated from a major producer of biotherapeutics. These were produced from CHO cell lines. The pI was 8.2–8.5 for mAb A and 7.7–8.2 for mAb B. Both the mAbs has a molecular weight of 150 kDa. The feed material consisted of neutralized Protein A elute from different batches pooled together. Feed concentrations were in the range of 10–20 mg/mL. 2.1.2. Preparative chromatography Äkta Explorer (GE Healthcare Life Sciences, Uppsala, Sweden) with UV and conductometric detectors with a fraction collector attached was used for performing process chromatography. Samples were injected using sample loops and the data was collected using Unicorn software (GE Healthcare Life Sciences, Uppsala, Sweden). 2.1.3. Columns and membrane Glass columns (Tricorn, GE Healthcare Life Sciences, Uppsala, Sweden) with I.D. = 0.5 cm, L = 20 cm and 10 cm were packed with different stationary phases. Poros XS (PXS), 50 m and Poros HS (PHS), 50 m, from Applied Biosystems, Capto Impress SP(CI), 39–42 m from GE Healthcare, Ceramic HyperD S (CS), 50 m from Pall Life Sciences, and Fractogel S (FS) and Fractogel COO− (FC) from Millipore were used for CEX. Phenyl Sepharose (90 m), Capto Butyl (75 m), and Capto Phenyl (75 m) from GE Healthcare (Uppsala, Sweden) were used for HIC. Sartobind Phenyl (3 mL) membrane adsorber was obtained from Sartorius AG, Germany, for performing HIC membrane chromatography. 2.1.4. Chemicals Acetic acid, sodium acetate anhydrous, sodium chloride, sodium dihydrogenphosphate, disodium hydrogen phosphate, sodium
sulfate, potassium di-hydrogenphosphate, dipotassium hydrogen phosphate, citric acid, sodium citrate, HEPES buffer were purchased from Merck, Germany, and Merck, India. For preparative chromatography, chemicals were of analytical grade and for analytical chromatography, of HPLC grade. 2.1.5. Analysis Ultra performance liquid chromatography (UPLC) (ACQUITY UPLC Waters corporation, Milford USA) consisting of an autosampler, a column compartment, and a PDA detector was used. Data collection and analysis was performed using Empower 2 software. Super SW3000 TSKGel (Tosoh Bioscience LLC, Tokyo, Japan) of I.D. = 0.4 cm, L = 30 cm, and particle size = 5 m was used for analysis of HMW content by size exclusion chromatography (SEC). UV 280 absorbance was used to measure sample concentration via Spectra Max (Molecular Devices, Sunnyvale, USA). Samples were analyzed for HCP content using microtiter plate immunoenzymatic assay (ELISA – catalog no. F015) purchased from Cygnus Technologies, Southport, NC, USA. 2.2. Methods 2.2.1. Cation exchange chromatography (CEX) CEX was performed using a manually packed column of 20 cm bed height and a column volume (CV) of 4 mL. Prior to sample loading, the column was equilibrated with the buffer to obtain the desired pH and conductivity. For equilibration, 15 mM sodium phosphate buffer (NaPO4 ) was used for pH range of 6.0–7.5 and 50 mM sodium acetate buffer (NaOAc) was used for pH range of 5.0–5.5. The column was loaded with mAb after adjusting the sample to the desired pH using phosphoric acid (for pH 6.0–7.0) or acetic acid (for pH 5.0–5.5). Conductivity was adjusted to that of the equilibration buffer either by diluting with buffer or by adding salt. Sample loading was done for a maximum of 30 mg/mL of the resin. The wash step was performed with the equilibration buffer if the elution and binding pH were same. If the elution and binding pH were different then wash was performed at pH same as of the elution buffer but without any salt. Elution was performed by forming a gradient from equilibration buffer by linear addition of salt. Eluted intermediate was collected in 2 mL fractions (maximum–minimum: 50–50 mAu as per UV 280) and analyzed for concentration using UV 280 and for aggregate content using SEC. The column was regenerated and sanitized with 1 M NaCl and 0.5 M NaOH, respectively, after every run. Experimental parameters that were examined in this study included elution pH (5.0–7.5), linear gradients (3.6, 5.5 and 11.2 mM/CV), elution salts (NaCl, KCl and Na2 SO4 ) at three pH values and CEX resin media listed above. The flow rate was kept at 300 cm/h for all runs. All the experiments were performed at ambient temperature. 2.2.2. Hydrophobic interaction chromatography (HIC) 2.2.2.1. Column chromatography. A manually packed column of 20 cm bed height and CV of 4 mL was used for initial optimization. Prior to sample loading, the column was equilibrated with buffer to obtain the desired pH and conductivity. For equilibration, 15 mM NaPO4 buffer was used for pH range of 6–7.5 and 50 mM NaOAc buffer was used for pH range of 5.0–5.5. The equilibration buffer and sample also contained lyotropic salts (phosphate, citrate and sulphates) of concentration greater than 0.2 M. Sample loading was kept at a maximum of 25 mg/mL of the resin and the wash step was performed with the equilibration buffer. Elution was performed by forming a gradient of equilibration buffer by linear removal of salt. The elution was collected in 2 mL fractions and analyzed for concentration using UV 280 and for aggregate content using SEC. The
V. Kumar, A.S. Rathore / J. Chromatogr. A 1368 (2014) 155–162
157
due to the large number of experiments. For the two mAbs under consideration, it was found that the selectivity was independent of the feed quality in the range of HMW % examined. To compare the selectivity toward separation of HMW across different resins and operating conditions, the total cumulative content of aggregates percentage for each fraction was plotted against the protein yield percentage. Plots with a shallower slope indicated higher selectivity. Mass balance was also performed and it was noted that in some cases aggregate species eluted in the regeneration peak along with other impurities. 3.1. Optimization of cation exchange (CEX) chromatography Operation parameters that were examined in this study included pH, resin, elution gradient, elution salts and buffer species. Sample loading was kept constant throughout this study at 15 mg/mL, since higher loading is known to decrease resolution. Resins were evaluated for their capability to withstand the high back pressure generated at flow rate of 300 cm/h. The bed height was kept constant at 20 cm to achieve a residence time of 4 min for all the runs.
Fig. 1. Setup for the two-stage chromatography process for removal of aggregates from monoclonal antibody therapeutic products (A) CEX + HIC column, (B) CEX column + HIC membrane adsorber.
column was regenerated and sanitized with water and 0.5 M NaOH, respectively, after each run. Experimental parameters that were examined in this study included elution pH (5.0–7.5), binding salts (sodium sulfate, sodium phosphate, sodium citrate), and HIC resin media listed above. The flow rate was maintained at 300 cm/h for all runs. 2.2.2.2. Membrane chromatography. Sartobind Phenyl membrane adsorber of 3 mL volume was used to perform HIC. The method used was same as for column chromatography except that sodium sulphate was the only salt evaluated in flow-through mode at three different pH values (5.0, 6.0 and 7.0). 2.2.3. Two stage chromatography A 0.5 × 10 cm glass column was packed with HIC media and was connected in series at the outlet of 0.5 × 20 cm CEX glass column as shown in Fig. 1A. Similarly, HIC membrane was also connected in series to the outlet of CEX column (Fig. 1B). The CEX method mentioned above was used except that a water wash was included after the regeneration step of 1 M NaCl. The eluted fractions were analyzed for concentration using UV 280 and HMW content using SEC. 3. Results and discussion The strategy that was undertaken in this study involved optimization of the CEX and HIC steps individually followed by their integration. However, there were several considerations that needed to be kept in mind during the optimization studies so as to ensure a smooth integration. The resin media had to be able to sustain high back pressure and exhibit sufficient selectivity toward clearance of aggregates with a bed height of less than or equal to 20 cm. In view of this, the resins examined were those with a bead size of 50 m (CS, PHX, PXS and) or less (CI) so as to achieve high selectivity. The feed material used in the CEX and HIC optimization studies consisted of Protein A elute with an aggregate content of 2.5–4.0%. For any particular set of experiments, the feed material was kept same. However, the feed material quality varied from study to study
3.1.1. Effect of pH The upper end of the pH operating range was chosen based on the pI of the mAbs and their binding capacity at that pH. The lower end of the pH range was chosen such that product recovery of at least 90% could be achieved. Since the pI of mAb B was lower than that of mAb A, the pH range for mAb B was 5.0–7.0 while that for mAb A was 5.5–7.5. Two sets of experiments were performed to check for the effect of binding pH on separation of aggregates for the CEX resins under consideration (CI, CS, FS and FC) at a fixed linear gradient of 11.2 mM/CV. In the first set, binding and elution pH were kept different while in the second set, binding and elution pH were kept the same. For the binding pH of 5.5 and elution pH of 7.5, the separation was identical to the runs where binding and elution pH were kept same at 7.5. Similarly, when the binding pH was 7.5 and elution pH was 5.5, the results were identical to the runs where binding and elution pH were 5.5. Thus, it was concluded that in the operational window under consideration, binding pH did not have a significant effect. As the elution pH decreased from 7.5 to 5.5, separation of HMW improved. Plots for resin CI are shown in Fig. 2A and B and it is seen that the behavior is quite similar for both mAb A and mAb B. Similar trends were also observed for other resins included in this study (data for PHS is show in Fig. 2C and D). 3.1.2. Effect of elution gradient slope Various elution gradients (3.6, 5.5 and 11.2 mM/CV) with NaCl at pH 7.5 were examined and the results are shown in Fig. 3A and B for the CS and CI resin for separation of HMW from mAb A, respectively. It is seen that the gradient slope does not have a significant impact on the separation of HMW in either case in the chosen range. However, the elution volume and retention increased as the slope of gradient became shallower. It can be concluded from the data that separation of HMW is occurring via linear chromatography and as a result it is likely that any increase in the product loading (15–30 mg/mL) will not impact resolution. This was confirmed by experiments performed at higher loading of 30 mg/mL without any loss in resolution (data presented later). 3.1.3. Effect of salts and buffer species A variety of salts (KCl, NaCl NaPO4 and Na2 SO4 ) were examined for their effect on separation of HMW. The gradient was kept constant at 11.2 mM/CV. The resulting data for mAb A is shown in Figure 4A. It is seen that the choice of salt does not have a significant effect on the quality of the separation. In addition, phosphate buffer
158
V. Kumar, A.S. Rathore / J. Chromatogr. A 1368 (2014) 155–162
Fig. 2. Effect of elution pH in CEX chromatography for removal of aggregates at a linear gradient of 11.2 mM/CV. (A) Effect of elution pH on separation of HMW species for mAb A with Capto Impress (CI) resin. Separation improved at lower elution pH. (B) Effect of elution pH on separation of HMW species for mAb B with Capto Impress SP (CI) resin. Separation improved at lower elution pH. (C) Effect of elution pH on separation of HMW species for mAb A with Poros HS (PHS) resin. Separation improved at lower elution pH. (D) Effect of elution pH on separation of HMW species for mAb B with Poros HS (PHS) resin. Separation improved at lower elution pH.
was replaced with HEPES buffer at pH 7.5. Similarly, any change in buffer also did not have a significant effect on separation. Since our proposed scheme requires that CEX elution be directly loaded on the HIC column, the observations above led us to the conclusion that the choice of the buffer and the salt should be based on the HIC column optimization studies. 3.1.4. Effect of resin Different resins were evaluated for their capability toward separation of HMW (CS, FC, FS PHX, PXS, and CI). The results are shown in Fig. 4B–D for mAb A at an elution and binding pH of 5.5–7.5. In general, it was observed that the separation improved as pH was lowered (as mentioned earlier) and also that separation improved with retention. The only exceptions were CI and FC, for both of which the separation was better than the observed retention. This
could be attributed to the lower elution volume and sharper peak observed with CI and the fact that FC is a weak cation exchanger (ligand COO− ), while all the remaining resins are strong cation exchangers. Similar results were obtained for separation of HMW for mAb B. Another key observation is that complete clearance of HMW cannot be achieved without peak cutting as the HMW species co-elute with the monomer in the later portion of the peak. Table 1 summarizes the conductivity at which mAb A elutes in CEX chromatography for a variety of resins under different binding and elution pH for NaCl. This data can be used to choose the appropriate CEX resin for designing the two-stage chromatographic process. Since the effect of type of salt was not present, we expect this result to be equally applicable for other salts. Although, the behavior for HMW clearance was similar, it must be noted that the dynamic binding capacity (DBC) will be different
Fig. 3. Effect of gradient slope in CEX chromatography for removal of aggregates with NaCl at pH 7.5 sodium phosphate buffer. (A) Effect of gradient slope on separation of HMW species for mAb A with Ceramic HyperD S (CS) resin. Slope does not seem to have a significant effect on separation. (B) Effect of gradient slope on separation of HMW species for mAb A with Capto Impress SP (CI) resin. Slope does not seem to have a significant effect on separation.
V. Kumar, A.S. Rathore / J. Chromatogr. A 1368 (2014) 155–162
159
Fig. 4. Effect of buffers, salts and choice of resin medial in CEX chromatography for removal of aggregates at a linear gradient of 11.2 mM/CV. (A) Effect of buffers and salts on separation of HMW species for mAb A with Capto Impress SP (CI) resin. Neither seem to have a significant effect on separation. (B) Effect of resin media on separation of HMW species for mAb A at binding and elution pH of 5.5. Choice of resin media does show an effect on the separation. (C) Effect of resin media on separation of HMW species for mAb A at binding and elution pH of 6.5. Choice of resin media does show an effect on the separation. (D) Effect of resin media on separation of HMW species for mAb A at binding and elution pH of 7.5. Choice of resin media does show an effect on the separation.
for different resins. For all the resins in general, it would be higher at lower pH. Since the loading capacities were kept well below the maximum in this study, there was not much influence of this factor from resin to resin. 3.2. HIC optimization Three HIC resins, namely Capto Phenyl, Capto Butyl and Phenyl Sepharose, were evaluated for their capability toward separation of HMW clearance. Operational parameters that were examined in this study included the type and concentration of salt. The objective of this study was to identify an elution window of salt concentration such that only the HMW binds to the HIC column and monomer flows through. Since the elution of CEX was directly loaded on the HIC column, the minimum salt concentration where HMW could bind was determined. As seen from Table 1, a salt conductivity range of >11 mS/cm could be used for elution in the CEX column and hence this range was examined for the HIC column as well and a pH of 6.5 (for mAb A) was chosen for optimization for each salt. First, the HIC column was operated in the bind and elute mode. Once the salt concentration at which the product elutes had been determined, this information was used to design column
operation in a flow-through mode. As per the Hofmeister series, the order of salts with respect to their hydrophobicity is sodium citrate > sodium sulfate > sodium phosphate. Similarly, the order of resin hydrophobicity is Capto Phenyl > Phenyl Sepharose > Capto Butyl. Table 2 presents the results from the HIC column studies. It is seen that due to the low concentration of salt in the feed material, most of the conditions resulted in the product flowing through the column. As mentioned above, our desired process scheme involves use of CEX column in the bind-elute mode and HIC column in flow-through mode. As seen in Table 2, in the case of Capto Phenyl with sodium citrate as the binding salt, the affinity of mAb A was found to be very strong and as a result the recovery was poor. When sodium phosphate was used as binding salt, HMW content in the eluted pool was 0.52%. Further, sodium sulphate as a binding salt gave maximum clearance of HMW (HMW 0.15%). In contrast, the binding with Capto Butyl was suboptimal even with 0.5 M salt concentration irrespective of the salt chosen and HMW in the flow-through was 0.6–2.07%. Further, HMW content for Phenyl Sepharose with sodium sulphate was 1.4%. Based on these results, Capto Phenyl with sodium sulphate at pH 5.5 was further examined in this study.
Table 1 Conductivity at which the mAb A elutes at the maximum peak height in CEX chromatography for a variety of resins and at different binding and elution pH for NaCl. Resin pH
CI (mS/cm)
CS (mS/cm)
FS (mS/cm)
FC (mS/cm)
PXS (mS/cm)
7.5 6.5 5.5
10.2 11.2 13.9
4.4 5.5 8.6
11.2 12.4 17.4
9.3 10.4 14.1
17.1 18.8 19.14
160
V. Kumar, A.S. Rathore / J. Chromatogr. A 1368 (2014) 155–162
Table 2 Aggregate levels in the eluted pool for the different HIC resins examined in this study for mAb A. The load capacity was fixed to 20 mg/mL of resin. The peaks were collected from 50 mAu to 50 mAu. Bold indicates the optimal conditions that result in acceptable yield as well as aggregate clearance. Resin
pH
Salt
Salt conc. (M)
Mode
Yield %
HMW %
Capto Phenyl
6.5 6.5 7.5 6.5 5.5 6.5 6.5 5.5 7.5 7.5 6.5 6.5 6.5
Sodium phosphate Sodium citrate Sodium sulfate Sodium sulfate Sodium sulfate Sodium sulfate Sodium sulfate Sodium sulfate Sodium sulfate Sodium sulfate Sodium sulfate Sodium sulfate Sodium sulfate
0.2 0.2 0.5 0.5 0.2 0.5 0.15 0.5 0.5 0.1 0.05 0.15 0.5
Flow Through Bind Bind Bind Flow Through Bind Flow Through Bind Bind Flow Through Flow Through Flow Through Flow Through
90.76 82.76 40.2 78.05 96.0 96.5 97.5 91.78 97.4 94.8 94.5 92.1 90.26
0.52 0.4 0.2 0.15 0.1 1.4 1.9 1.7 2.1 2.38 2.07 2.41 NAa
Phenyl Sepharose
Capto Butyl
a
Not analyzed.
3.3. Integration of CEX and HIC into a two-stage chromatography process 3.3.1. Combination of CEX and HIC column chromatography Since the HIC step is the second stage of the proposed scheme and because a column height of 10 cm would be sufficient, pressure drop was not expected to be a major concern for this step. For mAb A, CI and PXS were examined as the CEX resins based on the desired conductivity window for elution (11–20 mS/cm, Table 1) and satisfactory clearance of HMW. Also, Capto Phenyl was examined as possible HIC resin. Table 3 shows the results of the various combinations that were examined. HMW% in the elution pool for the combination of CI (for CEX) and Capto Phenyl (for HIC) at pH values of 6.5 and 5.5 was 1.02% and 0.4%, respectively, for recovery >95% without need of peak cutting. With the combination PXS and Capto Phenyl, HMW content was 0% with recovery >96% at pH 5.5 without any peak cutting. Both of these two options offer robust solutions. Fig. 5 illustrates the chromatograms obtained from CEX chromatography and from the two-stage CEX-HIC chromatography. While the former shows partial separation of the HMW from mAb A monomer, the latter achieves baseline separation of the two with the HMW species eluting as part of the regeneration peak. Recovery of 97% was achieved. Similar optimization was performed for mAb B and PHS-Capto Butyl emerged as the most optimal combination and yielded 0% HMW content in the elution pool with 96% recovery. It can be concluded that for different mAbs the optimal combination of resins is likely to be different and would depend on the pI and hydrophobicity of the mAb.
Fig. 5. Chromatogram overlays of CEX Poros XS (PXS) column and CEX-HIC (PXSCapto Phenyl) column chromatography. Gradient: Sodium sulphate at pH 5.5. A clear hump is seen in the first case which was abundant in HMW %. In combined run no such peak was seen, indicating that the HMW species have moved out to regeneration peak.
3.3.2. Combination of CEX column and HIC membrane chromatography Since we are using the HIC column in a flow-through mode, we wanted to examine the feasibility of substituting the HIC column with a HIC membrane adsorber. This option will reduce the overall pressure drop in the process and also offer other key advantages associated to membrane chromatography [21,22]. Since the HIC membrane adsorber offers negligible back pressure, it would broaden the pool of CEX resins which otherwise may not be able to withstand very high pressure drops. A combination of CI (for CEX)
Fig. 6. Comparison of HMW clearance for CEX alone and for CEX column + HIC membrane Sartobind Phenyl (at a linear gradient of 11.2 mM/CV with sodium sulphate) for (A) mAb A and (B) mAb B. It is evident that the combination gives better separation of HMW in both cases though a complete separation cannot be achieved without peak cutting in the case of HIC membrane chromatography. This is in contrast to resin chromatography where baseline separation was achieved and thus no peak cutting was required.
V. Kumar, A.S. Rathore / J. Chromatogr. A 1368 (2014) 155–162
161
Table 3 Aggregate levels in elution pool obtained from the various options examined for the proposed two-stage chromatography process. The protein load density on the columns was 20 mg/mL. Resin media
pH
Peak cuttinga
Elution volume (mL)
Protein conc. (mg/mL)
HCP levelb (ppm)
Aggregate %
Yield (%)
CI only
5.5 6.5 5.5 6.5 5.5 6.5 5.5 5.5 6.5 5.5
Yes Yes No No Yes Yes No No No No
14 12 20 16 12 10 26 26 26 26
5.26 6.07 3.84 4.75 6.13 7.20 2.95 2.98 2.95 2.62
NAc NAc NAc NAc 9.1 NAc 5.9 BDLd NAc BQLe
0.5 1.1 0.4 1.02 0.5 0.7 0.15 0 0.3 0.3
92 91 96 95 92 90 96 97 96 85
Two-stage CI + Capto Phenyl PXS only Capto Phenyl only Two-stage PXS + Capto Phenyl PXS followed by Capto Phenyl separately a b c d e
Theoretical peak cutting with yield set to >90%. HCP level in Load sample 65 ppm. Not analyzed. Below detection limit (LOD 1.7 ppm). Below quantification limit (LOQ 4 ppm).
and Sartobind Phenyl membrane adsorber was evaluated for mAb A at pH values of 5.5 and 6.5 and pH 5.0 and 6.0 for mAb B. The results are shown in Fig. 6A and B for mAb A and mAb B, respectively. It is seen that in all cases significant improvement in performance is seen when CEX-HIC combination is used but not as good as resin chromatography. Fig. 7 illustrates the chromatograms obtained from CI and from CI-Sartobind Phenyl membrane chromatography. While the partial separation of the aggregates was achieved in the case of CI only, a significantly better clearance is achieved with the CI and HIC membrane combination. The fact that the regeneration peak increases considerably in size for the case of CEX-HIC combination indicates the possibility of the aggregate eluting in the regeneration peak. This combination for mAb A yielded a HMW % of 0.32 in the eluted pool and a recovery of 93%. The optimal combination for mAb B was that of PHS-Sartobind Phenyl Membrane which yielded a HMW% of 0.1 in the eluted pool and a recovery of 94%. 3.4. Discussion The proposed two-stage chromatography process offers several key advantages over the traditional two column process. Table 3 shows a comparison of the proposed process against the traditional process. First, it is clear from the results presented above that the proposed scheme is able to better harness the selectivities of the two columns and thus, results in lower HMW% in the final pool. When the CEX process sample was loaded on HIC column
with same parameters of operation as proposed in the present scheme, the HMW content in the final elution pool was 0.3% against 0% in the proposed two-stage process. This is likely because when the fractions in CEX are pooled together, part of the separation achieved by the CEX is lost. Since the proposed two-stage process involves loading of the eluate coming from the CEX column directly on the HIC column, the separation achieved in CEX is fully capitalized and significantly better selectivity is achieved. Second, the recoveries are also higher with the CEX-HIC combination as peak cutting is not required and thus, nearly all the monomer can be pooled. Third, the proposed two-step process requires significantly less bed height for the HIC column (10 cm vs 20 cm) and thus would substantially lower the cost of goods. Fourth, the proposed two-stage process provides adequate flexibility to the user in picking the pH and the salt concentration required for the separation. Thus, one can use conditions of high salt concentration or low pH that are known to result in mAb aggregation [20]. Fifth, the proposed two-stage process is likely to result in a significant reduction in the processing time of the two steps. This is due to the continuous operation that occurs in the proposed process. Finally, as peak cutting is not required in the proposed process, the process is significantly more robust than the traditional process. The proposed process also provides better clearance of other key impurities such as HCP (Table 3). Further, the overall elution volume remains unchanged in the proposed process when compared to the traditional two-step process. Most monoclonal antibody platforms today utilize the combination of IEX and HIC for purification. Hence, the proposed two-stage, single step process is quite relevant for purification of monoclonal antibody products. 4. Conclusions A robust and effective process capable of complete removal of HMW species of mAb with >95% recovery of the product has been proposed. Successful demonstration of the proposed twostage process has been shown for two different mAb therapeutic products. We think that the approach followed in this study can be extended to any mAb product and that the proposed process would have a great appeal to those interested in implementing continuous bioprocessing.
Fig. 7. Chromatogram overlays of CEX Capto Impress SP (CI) column and CEX-HIC (Capto Impress SP (CI)-Sartobind Phenyl membrane) column-membrane chromatography. Gradient: Sodium sulphate at pH 5.5. The hump in the case of Capto Impress SP (CI) column only is not as prominent as in Fig. 5. However, similar to Fig. 5, a larger regeneration peak is observed for the two-stage process indicating the possibility of the aggregate eluting in the regeneration peak.
References [1] A.L. Nelson, E. Dhimolea, J.M. Reichert, Development trends for human monoclonal antibody therapeutics, Nat. Rev. Drug Discov. 9 (2010) 767–774. [2] G. Walsh, Biopharmaceutical benchmarks 2010, Nat. Biotechnol. 28 (9) (2010) 917–926.
162
V. Kumar, A.S. Rathore / J. Chromatogr. A 1368 (2014) 155–162
[3] G. Guiochon, L.A. Beaver, Separation science is the key to successful biopharmaceuticals, J. Chromatogr. A 1218 (2011) 8836–8858. [4] A.T. Hanke, M. Ottens, Purifying biopharmaceuticals: knowledge-based chromatographic process development, Trends Biotechnol. 32 (2014) 210–220. [5] A.A. Shukla, B. Hubbard, T. Tressel, S. Guhan, D. Low, Downstream processing of monoclonal antibodies—application of platform approaches, J. Chromatogr. B 848 (2007) 28–39. [6] W.S. Putnam, S. Prabhu, Y. Zheng, M. Subramanyam, Y.M.C. Wang, Pharmacokinetic, pharmacodynamic and immunogenicity comparability assessment strategies for monoclonal antibodies, Trends Biotechnol. 28 (2010) 509–516. [7] A.S. Rathore, Follow-on protein products: scientific issues, developments and challenges, Trends Biotechnol. 27 (2009) 698–705. [8] P. Arosio, G. Barolo, T. Müller-Späth, H. Wu, M. Morbidelli, Aggregation stability of a monoclonal antibody during downstream processing, Pharm. Res. 28 (2011) 1884–1894. [9] A.S. Rosenberg, Effects of protein aggregates: an immunologic perspective, AAPS J. 8 (2006) E501–E507. [10] J.C. Rea, G.T. Moreno, Y. Lou, D. Farnan, Validation of a pH gradient-based ionexchange chromatography method for high-resolution monoclonal antibody charge variant separations, J. Pharm. Biomed. Anal. 54 (2011) 317–323. ˛ [11] W. Marek, R. Muca, S. Wo´s, W. Piatkowski, D. Antos, Isolation of monoclonal antibody from a Chinese hamster ovary supernatant. I: Assessment of different separation concepts, J. Chromatogr. A 1305 (2013) 55–63. ˛ [12] W. Marek, R. Muca, S. Wo´s, W. Piatkowski, D. Antos, Isolation of monoclonal antibody from a Chinese hamster ovary supernatant. II: Dynamics of the integrated separation on ion exchange and hydrophobic interaction chromatography media, J. Chromatogr. A 1305 (2013) 64–75.
[13] Z. Xu, J. Li, J.X. Zhou, Process development for robust removal of aggregates using cation exchange chromatography in monoclonal antibody purification with implementation of quality by design, Prep. Biochem. Biotechnol. 42 (2012) 183–202. [14] J. Fogle, N. Mohan, E. Cheung, J. Persson, Effects of resin ligand density on yield and impurity clearance in preparative cation exchange chromatography. I. Mechanistic evaluation, J. Chromatogr. A 1225 (2012) 62–69. [15] J. Fogle, J. Persson, Effects of resin ligand density on yield and impurity clearance in preparative cation exchange chromatography. II. Process characterization, J. Chromatogr. A 1225 (2012) 70–78. [16] S. Ghose, Y. Tao, L. Conley, D. Cecchini, Purification of monoclonal antibodies by hydrophobic interaction chromatography under no-salt conditions, MAbs 5 (5) (2013) 795–800. [17] J. Chen, J. Tetrault, A. Ley, Comparison of standard and new generation hydrophobic interaction chromatography resins in the monoclonal antibody purification process, J. Chromatogr. A 1177 (2008) 272–281. [18] Y. Lu, B. Williamson, R. Gillespie, Recent advancement in application of hydrophobic interaction chromatography for aggregate removal in industrial purification process, Curr. Pharm. Biotechnol. 10 (2009) 427–433. [19] N. Frau, M. Kuczewski, G.Z. Papastoitsis, M. Hirai, Hydrophobic membrane adsorbers for large-scale downstream processing, BioPharm Int. Suppl. (2009). [20] V. Joshi, T. Shivach, V. Kumar, N. Yadav, A. Rathore, Avoiding antibody aggregation during processing: establishing hold times, Biotechnol. J. (2014), http://dx.doi.org/10.1002/biot.201400052. [21] A.S. Rathore, A. Shirke, Recent developments in membrane-based separations in biotechnology processes: review, Prep. Biochem. Biotechnol. 41 (4) (2011) 398–421. [22] S. Muthukumar, A.S. Rathore, High throughput process development (HTPD) platform for membrane chromatography, J. Membr. Sci. 442 (2013) 245–253.
Contents lists available at ScienceDirect
Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma
Two-stage chromatographic separation of aggregates for monoclonal antibody therapeutics Vijesh Kumar, Anurag S. Rathore ∗ Department of Chemical Engineering, Indian Institute of Technology Delhi, Hauz Khas, New Delhi 110016, India
a r t i c l e
i n f o
Article history: Received 5 June 2014 Received in revised form 17 September 2014 Accepted 27 September 2014 Available online 6 October 2014 Keywords: Monoclonal antibodies Purification Aggregates Cation exchange chromatography Hydrophobic interaction chromatography
a b s t r a c t Aggregates of monoclonal antibody (mAb) therapeutics, due to their perceived impact on immunogenicity, are recognized as a critical quality attribute by the regulatory authorities as well as the industry. Hence, removal of aggregates is a key objective of bioprocessing. At present, this is achieved by a combination of two or more orthogonal chromatographic steps with possible modalities of ion exchange, hydrophobic interaction and mixed mode. A two-stage chromatographic purification process consisting of ion-exchange and hydrophobic interaction modes is proposed in this paper for effective and efficient control of aggregates for a mAb therapeutic. The proposed scheme does not require any intermediate processing of the process stream. Further, baseline separation is achieved for monomer and aggregates resulting in robust performance. This was possible because the proposed operational scheme allowed for an addition of selectivities of the two chromatography modes vs. the traditional two column scheme where part of the separation of aggregates achieved by the first column is lost upon pooling. The proposed process scheme yielded improved separation of aggregates (0% vs. 1–2%) at >95% recovery and reduced overall process time (6 h vs. 14 h) for a typical application. Further, clearance of host cell proteins was also shown to have improved with the suggested process scheme. Successful implementation of the proposed scheme has been demonstrated for two different monoclonal antibody therapeutic products. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Monoclonal antibodies (mAbs) constitute an increasingly larger share of biotech therapeutics for diseases like cancer, rheumatoid arthritis, psoriasis and multiple sclerosis [1,2]. Due to the complexity of these products and the large amounts that need to be produced, last two decades have witnessed major efforts toward improving the quality of these products and reducing the cost of manufacturing [3,4]. These products are typically expressed in mammalian cell culture followed by a multistep purification process such that the different steps work on different separation mechanisms and thus together are able to provide the product which meets the stringent purity levels expected of therapeutic products. Downstream processing of mAbs is generally based on a platform process which includes Protein A affinity chromatography as a capture step, followed by one or more polishing steps [5]. The
∗ Corresponding author. Tel.: +91 11 26591098; fax: +91 1126581120; mob: +91 9650770650. E-mail address: [email protected] (A.S. Rathore). URL: http://www.biotechCMZ.com (A.S. Rathore). http://dx.doi.org/10.1016/j.chroma.2014.09.077 0021-9673/© 2014 Elsevier B.V. All rights reserved.
latter could be an ion exchange or hydrophobic interaction or mixed mode chromatography. The primary impurities that require clearance during bioprocessing are the host cell proteins (HCP), host cell DNA, and aggregates [6]. Protein aggregation is an undesirable event that occurs during manufacture of therapeutic mAbs [7,8]. As mentioned above, presence of even trace amounts of aggregates is considered to be undesirable due to the concern of adverse immunological response [9]. Cation exchange chromatography (CEX) is one of the polishing steps used for achieving effective removal of protein aggregates for mAbs. It involves binding of a protein under pre-equilibrated low salt condition followed by a buffer wash and elution of product either by step or linear addition of salt or by altering the pH [10]. Product related impurities such as mAb aggregates can be more effectively separated by linear salt gradient. Aggregate species in the CEX elution pool generally elute out in the later fractions due to their high avidity compared to the monomer species [11,13]. To get the desired purity, elution can be fractionated and pooled according to the target aggregate content. Alternatively, elution of aggregate species can be controlled by changing from a linear gradient to constant salt at the concentration where elution of the aggregate species is expected. Key factors that impact aggregate clearance on CEX include, resin ligand chemistry,
156
V. Kumar, A.S. Rathore / J. Chromatogr. A 1368 (2014) 155–162
sample loading, binding and elution conditions, and pooling criteria [11–15]. Hydrophobic interaction chromatography (HIC) is another major mode of chromatography for achieving effective clearance of aggregates [16–18]. More recently, HIC membrane adsorbers have also been successfully used for clearance of these aggregates [19]. Protein binding in HIC is achieved under high salt condition and elution is performed by step or linear gradient reduction in salt concentration. Removal of protein aggregates in HIC is also achieved in flow-through mode in addition to bind elute mode, where the salt condition is just enough to bind the more hydrophobic variant while the product flows through. In this mode, high protein load capacity can be achieved. However, addition of high concentrations of salt to the product to promote protein binding in HIC can be detrimental to the product itself [20]. Process parameters and raw material attributes that have been known to impact the clearance in HIC include resin ligand, lyotropic salts used in binding and pH [17]. Almost always, both CEX and HIC steps are required to get the desired clearance of aggregates. The two steps are generally performed one after the other, thus requiring more process time. In this paper, we propose use of a two-stage chromatography process for highly efficient and high throughput separation of aggregates for mAb therapeutic products. This has been achieved by selecting operating conditions such that the elution from the CEX step, operated in a bind and elute mode, can be directly loaded on the HIC column with the HIC column operated in the flow through mode connected in series. The proposed configuration results in optimal and robust removal of the aggregate species. 2. Materials and methods 2.1. Instruments and materials 2.1.1. mAb samples Two human IgG1 mAbs (mAb A and mAb B) were donated from a major producer of biotherapeutics. These were produced from CHO cell lines. The pI was 8.2–8.5 for mAb A and 7.7–8.2 for mAb B. Both the mAbs has a molecular weight of 150 kDa. The feed material consisted of neutralized Protein A elute from different batches pooled together. Feed concentrations were in the range of 10–20 mg/mL. 2.1.2. Preparative chromatography Äkta Explorer (GE Healthcare Life Sciences, Uppsala, Sweden) with UV and conductometric detectors with a fraction collector attached was used for performing process chromatography. Samples were injected using sample loops and the data was collected using Unicorn software (GE Healthcare Life Sciences, Uppsala, Sweden). 2.1.3. Columns and membrane Glass columns (Tricorn, GE Healthcare Life Sciences, Uppsala, Sweden) with I.D. = 0.5 cm, L = 20 cm and 10 cm were packed with different stationary phases. Poros XS (PXS), 50 m and Poros HS (PHS), 50 m, from Applied Biosystems, Capto Impress SP(CI), 39–42 m from GE Healthcare, Ceramic HyperD S (CS), 50 m from Pall Life Sciences, and Fractogel S (FS) and Fractogel COO− (FC) from Millipore were used for CEX. Phenyl Sepharose (90 m), Capto Butyl (75 m), and Capto Phenyl (75 m) from GE Healthcare (Uppsala, Sweden) were used for HIC. Sartobind Phenyl (3 mL) membrane adsorber was obtained from Sartorius AG, Germany, for performing HIC membrane chromatography. 2.1.4. Chemicals Acetic acid, sodium acetate anhydrous, sodium chloride, sodium dihydrogenphosphate, disodium hydrogen phosphate, sodium
sulfate, potassium di-hydrogenphosphate, dipotassium hydrogen phosphate, citric acid, sodium citrate, HEPES buffer were purchased from Merck, Germany, and Merck, India. For preparative chromatography, chemicals were of analytical grade and for analytical chromatography, of HPLC grade. 2.1.5. Analysis Ultra performance liquid chromatography (UPLC) (ACQUITY UPLC Waters corporation, Milford USA) consisting of an autosampler, a column compartment, and a PDA detector was used. Data collection and analysis was performed using Empower 2 software. Super SW3000 TSKGel (Tosoh Bioscience LLC, Tokyo, Japan) of I.D. = 0.4 cm, L = 30 cm, and particle size = 5 m was used for analysis of HMW content by size exclusion chromatography (SEC). UV 280 absorbance was used to measure sample concentration via Spectra Max (Molecular Devices, Sunnyvale, USA). Samples were analyzed for HCP content using microtiter plate immunoenzymatic assay (ELISA – catalog no. F015) purchased from Cygnus Technologies, Southport, NC, USA. 2.2. Methods 2.2.1. Cation exchange chromatography (CEX) CEX was performed using a manually packed column of 20 cm bed height and a column volume (CV) of 4 mL. Prior to sample loading, the column was equilibrated with the buffer to obtain the desired pH and conductivity. For equilibration, 15 mM sodium phosphate buffer (NaPO4 ) was used for pH range of 6.0–7.5 and 50 mM sodium acetate buffer (NaOAc) was used for pH range of 5.0–5.5. The column was loaded with mAb after adjusting the sample to the desired pH using phosphoric acid (for pH 6.0–7.0) or acetic acid (for pH 5.0–5.5). Conductivity was adjusted to that of the equilibration buffer either by diluting with buffer or by adding salt. Sample loading was done for a maximum of 30 mg/mL of the resin. The wash step was performed with the equilibration buffer if the elution and binding pH were same. If the elution and binding pH were different then wash was performed at pH same as of the elution buffer but without any salt. Elution was performed by forming a gradient from equilibration buffer by linear addition of salt. Eluted intermediate was collected in 2 mL fractions (maximum–minimum: 50–50 mAu as per UV 280) and analyzed for concentration using UV 280 and for aggregate content using SEC. The column was regenerated and sanitized with 1 M NaCl and 0.5 M NaOH, respectively, after every run. Experimental parameters that were examined in this study included elution pH (5.0–7.5), linear gradients (3.6, 5.5 and 11.2 mM/CV), elution salts (NaCl, KCl and Na2 SO4 ) at three pH values and CEX resin media listed above. The flow rate was kept at 300 cm/h for all runs. All the experiments were performed at ambient temperature. 2.2.2. Hydrophobic interaction chromatography (HIC) 2.2.2.1. Column chromatography. A manually packed column of 20 cm bed height and CV of 4 mL was used for initial optimization. Prior to sample loading, the column was equilibrated with buffer to obtain the desired pH and conductivity. For equilibration, 15 mM NaPO4 buffer was used for pH range of 6–7.5 and 50 mM NaOAc buffer was used for pH range of 5.0–5.5. The equilibration buffer and sample also contained lyotropic salts (phosphate, citrate and sulphates) of concentration greater than 0.2 M. Sample loading was kept at a maximum of 25 mg/mL of the resin and the wash step was performed with the equilibration buffer. Elution was performed by forming a gradient of equilibration buffer by linear removal of salt. The elution was collected in 2 mL fractions and analyzed for concentration using UV 280 and for aggregate content using SEC. The
V. Kumar, A.S. Rathore / J. Chromatogr. A 1368 (2014) 155–162
157
due to the large number of experiments. For the two mAbs under consideration, it was found that the selectivity was independent of the feed quality in the range of HMW % examined. To compare the selectivity toward separation of HMW across different resins and operating conditions, the total cumulative content of aggregates percentage for each fraction was plotted against the protein yield percentage. Plots with a shallower slope indicated higher selectivity. Mass balance was also performed and it was noted that in some cases aggregate species eluted in the regeneration peak along with other impurities. 3.1. Optimization of cation exchange (CEX) chromatography Operation parameters that were examined in this study included pH, resin, elution gradient, elution salts and buffer species. Sample loading was kept constant throughout this study at 15 mg/mL, since higher loading is known to decrease resolution. Resins were evaluated for their capability to withstand the high back pressure generated at flow rate of 300 cm/h. The bed height was kept constant at 20 cm to achieve a residence time of 4 min for all the runs.
Fig. 1. Setup for the two-stage chromatography process for removal of aggregates from monoclonal antibody therapeutic products (A) CEX + HIC column, (B) CEX column + HIC membrane adsorber.
column was regenerated and sanitized with water and 0.5 M NaOH, respectively, after each run. Experimental parameters that were examined in this study included elution pH (5.0–7.5), binding salts (sodium sulfate, sodium phosphate, sodium citrate), and HIC resin media listed above. The flow rate was maintained at 300 cm/h for all runs. 2.2.2.2. Membrane chromatography. Sartobind Phenyl membrane adsorber of 3 mL volume was used to perform HIC. The method used was same as for column chromatography except that sodium sulphate was the only salt evaluated in flow-through mode at three different pH values (5.0, 6.0 and 7.0). 2.2.3. Two stage chromatography A 0.5 × 10 cm glass column was packed with HIC media and was connected in series at the outlet of 0.5 × 20 cm CEX glass column as shown in Fig. 1A. Similarly, HIC membrane was also connected in series to the outlet of CEX column (Fig. 1B). The CEX method mentioned above was used except that a water wash was included after the regeneration step of 1 M NaCl. The eluted fractions were analyzed for concentration using UV 280 and HMW content using SEC. 3. Results and discussion The strategy that was undertaken in this study involved optimization of the CEX and HIC steps individually followed by their integration. However, there were several considerations that needed to be kept in mind during the optimization studies so as to ensure a smooth integration. The resin media had to be able to sustain high back pressure and exhibit sufficient selectivity toward clearance of aggregates with a bed height of less than or equal to 20 cm. In view of this, the resins examined were those with a bead size of 50 m (CS, PHX, PXS and) or less (CI) so as to achieve high selectivity. The feed material used in the CEX and HIC optimization studies consisted of Protein A elute with an aggregate content of 2.5–4.0%. For any particular set of experiments, the feed material was kept same. However, the feed material quality varied from study to study
3.1.1. Effect of pH The upper end of the pH operating range was chosen based on the pI of the mAbs and their binding capacity at that pH. The lower end of the pH range was chosen such that product recovery of at least 90% could be achieved. Since the pI of mAb B was lower than that of mAb A, the pH range for mAb B was 5.0–7.0 while that for mAb A was 5.5–7.5. Two sets of experiments were performed to check for the effect of binding pH on separation of aggregates for the CEX resins under consideration (CI, CS, FS and FC) at a fixed linear gradient of 11.2 mM/CV. In the first set, binding and elution pH were kept different while in the second set, binding and elution pH were kept the same. For the binding pH of 5.5 and elution pH of 7.5, the separation was identical to the runs where binding and elution pH were kept same at 7.5. Similarly, when the binding pH was 7.5 and elution pH was 5.5, the results were identical to the runs where binding and elution pH were 5.5. Thus, it was concluded that in the operational window under consideration, binding pH did not have a significant effect. As the elution pH decreased from 7.5 to 5.5, separation of HMW improved. Plots for resin CI are shown in Fig. 2A and B and it is seen that the behavior is quite similar for both mAb A and mAb B. Similar trends were also observed for other resins included in this study (data for PHS is show in Fig. 2C and D). 3.1.2. Effect of elution gradient slope Various elution gradients (3.6, 5.5 and 11.2 mM/CV) with NaCl at pH 7.5 were examined and the results are shown in Fig. 3A and B for the CS and CI resin for separation of HMW from mAb A, respectively. It is seen that the gradient slope does not have a significant impact on the separation of HMW in either case in the chosen range. However, the elution volume and retention increased as the slope of gradient became shallower. It can be concluded from the data that separation of HMW is occurring via linear chromatography and as a result it is likely that any increase in the product loading (15–30 mg/mL) will not impact resolution. This was confirmed by experiments performed at higher loading of 30 mg/mL without any loss in resolution (data presented later). 3.1.3. Effect of salts and buffer species A variety of salts (KCl, NaCl NaPO4 and Na2 SO4 ) were examined for their effect on separation of HMW. The gradient was kept constant at 11.2 mM/CV. The resulting data for mAb A is shown in Figure 4A. It is seen that the choice of salt does not have a significant effect on the quality of the separation. In addition, phosphate buffer
158
V. Kumar, A.S. Rathore / J. Chromatogr. A 1368 (2014) 155–162
Fig. 2. Effect of elution pH in CEX chromatography for removal of aggregates at a linear gradient of 11.2 mM/CV. (A) Effect of elution pH on separation of HMW species for mAb A with Capto Impress (CI) resin. Separation improved at lower elution pH. (B) Effect of elution pH on separation of HMW species for mAb B with Capto Impress SP (CI) resin. Separation improved at lower elution pH. (C) Effect of elution pH on separation of HMW species for mAb A with Poros HS (PHS) resin. Separation improved at lower elution pH. (D) Effect of elution pH on separation of HMW species for mAb B with Poros HS (PHS) resin. Separation improved at lower elution pH.
was replaced with HEPES buffer at pH 7.5. Similarly, any change in buffer also did not have a significant effect on separation. Since our proposed scheme requires that CEX elution be directly loaded on the HIC column, the observations above led us to the conclusion that the choice of the buffer and the salt should be based on the HIC column optimization studies. 3.1.4. Effect of resin Different resins were evaluated for their capability toward separation of HMW (CS, FC, FS PHX, PXS, and CI). The results are shown in Fig. 4B–D for mAb A at an elution and binding pH of 5.5–7.5. In general, it was observed that the separation improved as pH was lowered (as mentioned earlier) and also that separation improved with retention. The only exceptions were CI and FC, for both of which the separation was better than the observed retention. This
could be attributed to the lower elution volume and sharper peak observed with CI and the fact that FC is a weak cation exchanger (ligand COO− ), while all the remaining resins are strong cation exchangers. Similar results were obtained for separation of HMW for mAb B. Another key observation is that complete clearance of HMW cannot be achieved without peak cutting as the HMW species co-elute with the monomer in the later portion of the peak. Table 1 summarizes the conductivity at which mAb A elutes in CEX chromatography for a variety of resins under different binding and elution pH for NaCl. This data can be used to choose the appropriate CEX resin for designing the two-stage chromatographic process. Since the effect of type of salt was not present, we expect this result to be equally applicable for other salts. Although, the behavior for HMW clearance was similar, it must be noted that the dynamic binding capacity (DBC) will be different
Fig. 3. Effect of gradient slope in CEX chromatography for removal of aggregates with NaCl at pH 7.5 sodium phosphate buffer. (A) Effect of gradient slope on separation of HMW species for mAb A with Ceramic HyperD S (CS) resin. Slope does not seem to have a significant effect on separation. (B) Effect of gradient slope on separation of HMW species for mAb A with Capto Impress SP (CI) resin. Slope does not seem to have a significant effect on separation.
V. Kumar, A.S. Rathore / J. Chromatogr. A 1368 (2014) 155–162
159
Fig. 4. Effect of buffers, salts and choice of resin medial in CEX chromatography for removal of aggregates at a linear gradient of 11.2 mM/CV. (A) Effect of buffers and salts on separation of HMW species for mAb A with Capto Impress SP (CI) resin. Neither seem to have a significant effect on separation. (B) Effect of resin media on separation of HMW species for mAb A at binding and elution pH of 5.5. Choice of resin media does show an effect on the separation. (C) Effect of resin media on separation of HMW species for mAb A at binding and elution pH of 6.5. Choice of resin media does show an effect on the separation. (D) Effect of resin media on separation of HMW species for mAb A at binding and elution pH of 7.5. Choice of resin media does show an effect on the separation.
for different resins. For all the resins in general, it would be higher at lower pH. Since the loading capacities were kept well below the maximum in this study, there was not much influence of this factor from resin to resin. 3.2. HIC optimization Three HIC resins, namely Capto Phenyl, Capto Butyl and Phenyl Sepharose, were evaluated for their capability toward separation of HMW clearance. Operational parameters that were examined in this study included the type and concentration of salt. The objective of this study was to identify an elution window of salt concentration such that only the HMW binds to the HIC column and monomer flows through. Since the elution of CEX was directly loaded on the HIC column, the minimum salt concentration where HMW could bind was determined. As seen from Table 1, a salt conductivity range of >11 mS/cm could be used for elution in the CEX column and hence this range was examined for the HIC column as well and a pH of 6.5 (for mAb A) was chosen for optimization for each salt. First, the HIC column was operated in the bind and elute mode. Once the salt concentration at which the product elutes had been determined, this information was used to design column
operation in a flow-through mode. As per the Hofmeister series, the order of salts with respect to their hydrophobicity is sodium citrate > sodium sulfate > sodium phosphate. Similarly, the order of resin hydrophobicity is Capto Phenyl > Phenyl Sepharose > Capto Butyl. Table 2 presents the results from the HIC column studies. It is seen that due to the low concentration of salt in the feed material, most of the conditions resulted in the product flowing through the column. As mentioned above, our desired process scheme involves use of CEX column in the bind-elute mode and HIC column in flow-through mode. As seen in Table 2, in the case of Capto Phenyl with sodium citrate as the binding salt, the affinity of mAb A was found to be very strong and as a result the recovery was poor. When sodium phosphate was used as binding salt, HMW content in the eluted pool was 0.52%. Further, sodium sulphate as a binding salt gave maximum clearance of HMW (HMW 0.15%). In contrast, the binding with Capto Butyl was suboptimal even with 0.5 M salt concentration irrespective of the salt chosen and HMW in the flow-through was 0.6–2.07%. Further, HMW content for Phenyl Sepharose with sodium sulphate was 1.4%. Based on these results, Capto Phenyl with sodium sulphate at pH 5.5 was further examined in this study.
Table 1 Conductivity at which the mAb A elutes at the maximum peak height in CEX chromatography for a variety of resins and at different binding and elution pH for NaCl. Resin pH
CI (mS/cm)
CS (mS/cm)
FS (mS/cm)
FC (mS/cm)
PXS (mS/cm)
7.5 6.5 5.5
10.2 11.2 13.9
4.4 5.5 8.6
11.2 12.4 17.4
9.3 10.4 14.1
17.1 18.8 19.14
160
V. Kumar, A.S. Rathore / J. Chromatogr. A 1368 (2014) 155–162
Table 2 Aggregate levels in the eluted pool for the different HIC resins examined in this study for mAb A. The load capacity was fixed to 20 mg/mL of resin. The peaks were collected from 50 mAu to 50 mAu. Bold indicates the optimal conditions that result in acceptable yield as well as aggregate clearance. Resin
pH
Salt
Salt conc. (M)
Mode
Yield %
HMW %
Capto Phenyl
6.5 6.5 7.5 6.5 5.5 6.5 6.5 5.5 7.5 7.5 6.5 6.5 6.5
Sodium phosphate Sodium citrate Sodium sulfate Sodium sulfate Sodium sulfate Sodium sulfate Sodium sulfate Sodium sulfate Sodium sulfate Sodium sulfate Sodium sulfate Sodium sulfate Sodium sulfate
0.2 0.2 0.5 0.5 0.2 0.5 0.15 0.5 0.5 0.1 0.05 0.15 0.5
Flow Through Bind Bind Bind Flow Through Bind Flow Through Bind Bind Flow Through Flow Through Flow Through Flow Through
90.76 82.76 40.2 78.05 96.0 96.5 97.5 91.78 97.4 94.8 94.5 92.1 90.26
0.52 0.4 0.2 0.15 0.1 1.4 1.9 1.7 2.1 2.38 2.07 2.41 NAa
Phenyl Sepharose
Capto Butyl
a
Not analyzed.
3.3. Integration of CEX and HIC into a two-stage chromatography process 3.3.1. Combination of CEX and HIC column chromatography Since the HIC step is the second stage of the proposed scheme and because a column height of 10 cm would be sufficient, pressure drop was not expected to be a major concern for this step. For mAb A, CI and PXS were examined as the CEX resins based on the desired conductivity window for elution (11–20 mS/cm, Table 1) and satisfactory clearance of HMW. Also, Capto Phenyl was examined as possible HIC resin. Table 3 shows the results of the various combinations that were examined. HMW% in the elution pool for the combination of CI (for CEX) and Capto Phenyl (for HIC) at pH values of 6.5 and 5.5 was 1.02% and 0.4%, respectively, for recovery >95% without need of peak cutting. With the combination PXS and Capto Phenyl, HMW content was 0% with recovery >96% at pH 5.5 without any peak cutting. Both of these two options offer robust solutions. Fig. 5 illustrates the chromatograms obtained from CEX chromatography and from the two-stage CEX-HIC chromatography. While the former shows partial separation of the HMW from mAb A monomer, the latter achieves baseline separation of the two with the HMW species eluting as part of the regeneration peak. Recovery of 97% was achieved. Similar optimization was performed for mAb B and PHS-Capto Butyl emerged as the most optimal combination and yielded 0% HMW content in the elution pool with 96% recovery. It can be concluded that for different mAbs the optimal combination of resins is likely to be different and would depend on the pI and hydrophobicity of the mAb.
Fig. 5. Chromatogram overlays of CEX Poros XS (PXS) column and CEX-HIC (PXSCapto Phenyl) column chromatography. Gradient: Sodium sulphate at pH 5.5. A clear hump is seen in the first case which was abundant in HMW %. In combined run no such peak was seen, indicating that the HMW species have moved out to regeneration peak.
3.3.2. Combination of CEX column and HIC membrane chromatography Since we are using the HIC column in a flow-through mode, we wanted to examine the feasibility of substituting the HIC column with a HIC membrane adsorber. This option will reduce the overall pressure drop in the process and also offer other key advantages associated to membrane chromatography [21,22]. Since the HIC membrane adsorber offers negligible back pressure, it would broaden the pool of CEX resins which otherwise may not be able to withstand very high pressure drops. A combination of CI (for CEX)
Fig. 6. Comparison of HMW clearance for CEX alone and for CEX column + HIC membrane Sartobind Phenyl (at a linear gradient of 11.2 mM/CV with sodium sulphate) for (A) mAb A and (B) mAb B. It is evident that the combination gives better separation of HMW in both cases though a complete separation cannot be achieved without peak cutting in the case of HIC membrane chromatography. This is in contrast to resin chromatography where baseline separation was achieved and thus no peak cutting was required.
V. Kumar, A.S. Rathore / J. Chromatogr. A 1368 (2014) 155–162
161
Table 3 Aggregate levels in elution pool obtained from the various options examined for the proposed two-stage chromatography process. The protein load density on the columns was 20 mg/mL. Resin media
pH
Peak cuttinga
Elution volume (mL)
Protein conc. (mg/mL)
HCP levelb (ppm)
Aggregate %
Yield (%)
CI only
5.5 6.5 5.5 6.5 5.5 6.5 5.5 5.5 6.5 5.5
Yes Yes No No Yes Yes No No No No
14 12 20 16 12 10 26 26 26 26
5.26 6.07 3.84 4.75 6.13 7.20 2.95 2.98 2.95 2.62
NAc NAc NAc NAc 9.1 NAc 5.9 BDLd NAc BQLe
0.5 1.1 0.4 1.02 0.5 0.7 0.15 0 0.3 0.3
92 91 96 95 92 90 96 97 96 85
Two-stage CI + Capto Phenyl PXS only Capto Phenyl only Two-stage PXS + Capto Phenyl PXS followed by Capto Phenyl separately a b c d e
Theoretical peak cutting with yield set to >90%. HCP level in Load sample 65 ppm. Not analyzed. Below detection limit (LOD 1.7 ppm). Below quantification limit (LOQ 4 ppm).
and Sartobind Phenyl membrane adsorber was evaluated for mAb A at pH values of 5.5 and 6.5 and pH 5.0 and 6.0 for mAb B. The results are shown in Fig. 6A and B for mAb A and mAb B, respectively. It is seen that in all cases significant improvement in performance is seen when CEX-HIC combination is used but not as good as resin chromatography. Fig. 7 illustrates the chromatograms obtained from CI and from CI-Sartobind Phenyl membrane chromatography. While the partial separation of the aggregates was achieved in the case of CI only, a significantly better clearance is achieved with the CI and HIC membrane combination. The fact that the regeneration peak increases considerably in size for the case of CEX-HIC combination indicates the possibility of the aggregate eluting in the regeneration peak. This combination for mAb A yielded a HMW % of 0.32 in the eluted pool and a recovery of 93%. The optimal combination for mAb B was that of PHS-Sartobind Phenyl Membrane which yielded a HMW% of 0.1 in the eluted pool and a recovery of 94%. 3.4. Discussion The proposed two-stage chromatography process offers several key advantages over the traditional two column process. Table 3 shows a comparison of the proposed process against the traditional process. First, it is clear from the results presented above that the proposed scheme is able to better harness the selectivities of the two columns and thus, results in lower HMW% in the final pool. When the CEX process sample was loaded on HIC column
with same parameters of operation as proposed in the present scheme, the HMW content in the final elution pool was 0.3% against 0% in the proposed two-stage process. This is likely because when the fractions in CEX are pooled together, part of the separation achieved by the CEX is lost. Since the proposed two-stage process involves loading of the eluate coming from the CEX column directly on the HIC column, the separation achieved in CEX is fully capitalized and significantly better selectivity is achieved. Second, the recoveries are also higher with the CEX-HIC combination as peak cutting is not required and thus, nearly all the monomer can be pooled. Third, the proposed two-step process requires significantly less bed height for the HIC column (10 cm vs 20 cm) and thus would substantially lower the cost of goods. Fourth, the proposed two-stage process provides adequate flexibility to the user in picking the pH and the salt concentration required for the separation. Thus, one can use conditions of high salt concentration or low pH that are known to result in mAb aggregation [20]. Fifth, the proposed two-stage process is likely to result in a significant reduction in the processing time of the two steps. This is due to the continuous operation that occurs in the proposed process. Finally, as peak cutting is not required in the proposed process, the process is significantly more robust than the traditional process. The proposed process also provides better clearance of other key impurities such as HCP (Table 3). Further, the overall elution volume remains unchanged in the proposed process when compared to the traditional two-step process. Most monoclonal antibody platforms today utilize the combination of IEX and HIC for purification. Hence, the proposed two-stage, single step process is quite relevant for purification of monoclonal antibody products. 4. Conclusions A robust and effective process capable of complete removal of HMW species of mAb with >95% recovery of the product has been proposed. Successful demonstration of the proposed twostage process has been shown for two different mAb therapeutic products. We think that the approach followed in this study can be extended to any mAb product and that the proposed process would have a great appeal to those interested in implementing continuous bioprocessing.
Fig. 7. Chromatogram overlays of CEX Capto Impress SP (CI) column and CEX-HIC (Capto Impress SP (CI)-Sartobind Phenyl membrane) column-membrane chromatography. Gradient: Sodium sulphate at pH 5.5. The hump in the case of Capto Impress SP (CI) column only is not as prominent as in Fig. 5. However, similar to Fig. 5, a larger regeneration peak is observed for the two-stage process indicating the possibility of the aggregate eluting in the regeneration peak.
References [1] A.L. Nelson, E. Dhimolea, J.M. Reichert, Development trends for human monoclonal antibody therapeutics, Nat. Rev. Drug Discov. 9 (2010) 767–774. [2] G. Walsh, Biopharmaceutical benchmarks 2010, Nat. Biotechnol. 28 (9) (2010) 917–926.
162
V. Kumar, A.S. Rathore / J. Chromatogr. A 1368 (2014) 155–162
[3] G. Guiochon, L.A. Beaver, Separation science is the key to successful biopharmaceuticals, J. Chromatogr. A 1218 (2011) 8836–8858. [4] A.T. Hanke, M. Ottens, Purifying biopharmaceuticals: knowledge-based chromatographic process development, Trends Biotechnol. 32 (2014) 210–220. [5] A.A. Shukla, B. Hubbard, T. Tressel, S. Guhan, D. Low, Downstream processing of monoclonal antibodies—application of platform approaches, J. Chromatogr. B 848 (2007) 28–39. [6] W.S. Putnam, S. Prabhu, Y. Zheng, M. Subramanyam, Y.M.C. Wang, Pharmacokinetic, pharmacodynamic and immunogenicity comparability assessment strategies for monoclonal antibodies, Trends Biotechnol. 28 (2010) 509–516. [7] A.S. Rathore, Follow-on protein products: scientific issues, developments and challenges, Trends Biotechnol. 27 (2009) 698–705. [8] P. Arosio, G. Barolo, T. Müller-Späth, H. Wu, M. Morbidelli, Aggregation stability of a monoclonal antibody during downstream processing, Pharm. Res. 28 (2011) 1884–1894. [9] A.S. Rosenberg, Effects of protein aggregates: an immunologic perspective, AAPS J. 8 (2006) E501–E507. [10] J.C. Rea, G.T. Moreno, Y. Lou, D. Farnan, Validation of a pH gradient-based ionexchange chromatography method for high-resolution monoclonal antibody charge variant separations, J. Pharm. Biomed. Anal. 54 (2011) 317–323. ˛ [11] W. Marek, R. Muca, S. Wo´s, W. Piatkowski, D. Antos, Isolation of monoclonal antibody from a Chinese hamster ovary supernatant. I: Assessment of different separation concepts, J. Chromatogr. A 1305 (2013) 55–63. ˛ [12] W. Marek, R. Muca, S. Wo´s, W. Piatkowski, D. Antos, Isolation of monoclonal antibody from a Chinese hamster ovary supernatant. II: Dynamics of the integrated separation on ion exchange and hydrophobic interaction chromatography media, J. Chromatogr. A 1305 (2013) 64–75.
[13] Z. Xu, J. Li, J.X. Zhou, Process development for robust removal of aggregates using cation exchange chromatography in monoclonal antibody purification with implementation of quality by design, Prep. Biochem. Biotechnol. 42 (2012) 183–202. [14] J. Fogle, N. Mohan, E. Cheung, J. Persson, Effects of resin ligand density on yield and impurity clearance in preparative cation exchange chromatography. I. Mechanistic evaluation, J. Chromatogr. A 1225 (2012) 62–69. [15] J. Fogle, J. Persson, Effects of resin ligand density on yield and impurity clearance in preparative cation exchange chromatography. II. Process characterization, J. Chromatogr. A 1225 (2012) 70–78. [16] S. Ghose, Y. Tao, L. Conley, D. Cecchini, Purification of monoclonal antibodies by hydrophobic interaction chromatography under no-salt conditions, MAbs 5 (5) (2013) 795–800. [17] J. Chen, J. Tetrault, A. Ley, Comparison of standard and new generation hydrophobic interaction chromatography resins in the monoclonal antibody purification process, J. Chromatogr. A 1177 (2008) 272–281. [18] Y. Lu, B. Williamson, R. Gillespie, Recent advancement in application of hydrophobic interaction chromatography for aggregate removal in industrial purification process, Curr. Pharm. Biotechnol. 10 (2009) 427–433. [19] N. Frau, M. Kuczewski, G.Z. Papastoitsis, M. Hirai, Hydrophobic membrane adsorbers for large-scale downstream processing, BioPharm Int. Suppl. (2009). [20] V. Joshi, T. Shivach, V. Kumar, N. Yadav, A. Rathore, Avoiding antibody aggregation during processing: establishing hold times, Biotechnol. J. (2014), http://dx.doi.org/10.1002/biot.201400052. [21] A.S. Rathore, A. Shirke, Recent developments in membrane-based separations in biotechnology processes: review, Prep. Biochem. Biotechnol. 41 (4) (2011) 398–421. [22] S. Muthukumar, A.S. Rathore, High throughput process development (HTPD) platform for membrane chromatography, J. Membr. Sci. 442 (2013) 245–253.
