Accepted Manuscript Title: Large Scale monoclonal antibody purification by continuous chromatography, from process design to scale-up Author: Val´erie Girard Nicolas-Julian Hilbold Candy K.S. Ng Laurence Pegon Wael Chahim Fabien Rousset Vincent MonchoisTel.: +32 71 347900.Tel.: +33 383 49 70 00.Tel.: +1 610 494 0447. PII: DOI: Reference:
S0168-1656(15)00202-3 http://dx.doi.org/doi:10.1016/j.jbiotec.2015.04.026 BIOTEC 7100
To appear in:
Journal of Biotechnology
Received date: Revised date: Accepted date:
21-10-2014 13-4-2015 30-4-2015
Please cite this article as: Girard, V., Hilbold, N.-J., Ng, C.K.S., Pegon, L., Chahim, W., Rousset, F., Monchois, V.,Large Scale monoclonal antibody purification by continuous chromatography, from process design to scale-up, Journal of Biotechnology (2015), http://dx.doi.org/10.1016/j.jbiotec.2015.04.026 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Highlights A chained process including 2 SMCC steps and a polishing step was implemented for mAb purification Better process performance was achieved operating in SMCC mode compared with that in batch mode
ip t
The SMCC development was proved to be not as complicated as expected compared with that for batch
Ac ce p
te
d
M
an
us
cr
A better process understanding is expected through the implementation of SMCC
1
Page 1 of 37
[TITLE]
Large Scale monoclonal antibody purification by continuous chromatography, from process design to scale-up
Valérie GIRARD
ip t
[AUTHORS] [1]
[2]
us
Nicolas-Julian HILBOLD*
cr
[email protected]
[email protected]
[3]
an
Candy K. S. NG
[email protected]
[2]
M
Laurence PEGON
[1]
Ac ce p
[email protected]
te
Wael CHAHIM
d
[email protected]
Fabien ROUSSET
[2]
[email protected]
Vincent MONCHOIS
[2]
[email protected]
2
Page 2 of 37
[CONTACT INFORMATION] [1] Novasep Belgium 16 rue Clément Ader 6041 Gosselies
ip t
Belgium
cr
+32 71 347900
[2]
us
Novasep Process SAS 82 Boulevard de la Moselle Site Eiffel – BP50
an
54340 Pompey France
M
+33 383 49 70 00
23 Creek Circle – Boothwyn PA 19061
Ac ce p
USA
te
Novasep Inc.
d
[3]
+1 610 494 0447
[KEY WORDS]
Continuous biomanufacturing, Monoclonal antibody, Continuous chromatography, Antibody purification, Continuous bioprocessing, Downstream processing
3
Page 3 of 37
[ARTICLE CONTENT]
Abstract The development and optimization of a purification process of monoclonal antibodies based on two continuous chromatography steps for capture and intermediate purification is presented. The two chromatography steps were
ip t
individually optimized using either batch chromatography or Sequential Multi-Column Chromatography (SMCC) only. A proprietary simulation software was used to optimize SMCC and to evaluate the potential gains compared with batch chromatography. The SMCC recipes provided by the simulation software were evaluated
cr
experimentally. A good correlation was found between the simulated results and experimental observations.
Significant gains were observed on the productivity, buffer consumption and on the volume of resin required for
us
SMCC over batch chromatography. Based on these results, a chained process from the capture to polishing steps was implemented. This chained process demonstrated significantly better performance compared with the batch
an
equivalent whilst satisfying the specifications. The expected positive impact provided by implementing continuous chromatography is also discussed.
M
1. Introduction
The market of biotechnologies and biopharmaceuticals is continuously growing with a value of $20 billion in 2001
d
and is estimated to reach $216 billion in 2015 (1), which includes $79 billion solely for monoclonal antibodies (mAb) that have proven to be interesting therapeutic solutions for numerous diseases (2).
te
The advances in upstream processing over the last decades led to important improvements in the production yields and the process bottleneck has been displaced towards downstream processing (DSP). Depending on the
Ac ce p
process, purification steps can constitute up to 80% of the total manufacturing costs (3). Therapeutic doses of most antibodies are high, with 1 gram or more of active protein per dose being quite common, compared with the doses of other type of therapeutic proteins (4). With the current manufacturing modes and strategies, this high dosage implies very large scale production facilities for products that have successfully passed clinical trials, which usually implies high Capital Expenditures (CAPEX) and important Operational Expenses (OPEX). The product and process knowledge accumulated from the early 1990s, as well as the technological environment, led pharmaceutical companies to develop similar purification processes based on a common sequence of unit operations (5–9). Although some process adaptations and development have to be made given the variations from one mAb to another, this combination of common unit operations is nowadays considered as a typical mAb manufacturing process (Figure 1).This convergence of the manufacturing process is expected to accelerate and facilitate process development in order to generate better defined and cost-efficient processes. This definition of a general manufacturing process is also expected to facilitate process transfer (10) from a department to another (from R&D to pilot for example), from one facility to another within the company or to a custom manufacturing
4
Page 4 of 37
organization when outsourcing is considered. The purification of a target protein – consisting of capture, intermediate and polishing – mostly involves chromatography (11, 12) and is conventionally operated in batch as discontinuous steps, where the operation is expensive and slow (13). In batch mode chromatography, operating conditions do not allow a full utilization of the media’s total capacity (Figure 2, left), leading to a sub-utilization of the media (14, 15). Recent developments in continuous chromatography have demonstrated the potential of the
ip t
technology to overcome these problems (16) and the development of fully continuous downstream processes.
Sequential Multi-Column Chromatography (SMCC) is an open-loop process for the separation of multi-component
cr
mixtures using multiple buffers (17, 18). The concept behind SMCC is to perform the different steps in smaller
columns (from 2 to 8 columns; compared with batch chromatography) connected in series, to maximize media
us
utilization (Figure 2, right) without any product loss. The BioSC® is an automated system based on the SMCC process, designed and provided by Novasep. Depending on process conditions, it is operated with 2 to 6 columns. The feed is injected into the first column (Figure 3, t0). The product breakthrough is automatically
an
monitored by appropriate probes. In batch design, the operating conditions are set to stop loading before observing the product breakthrough. This is unlike the operation of SMCC, for which the unbound product is
M
adsorbed by the second column as the media in the first column becomes more and more saturated (t0+∆t). When the first column is saturated, a “Wash* step” (t0+2∆t) pushes the product in the column void volume into the next column to ensure all product will be recovered from the interstitial liquid. This feature is a major difference
d
compared with fully continuous open-loop SMB process, for which this fraction of target protein is lost. The
te
saturated column is automatically disconnected from the feed stream and undergoes successively the wash, elution, regeneration and equilibration steps, while the other columns remain connected to the feed stream to
Ac ce p
continue the capture process (t0+3∆t). Once equilibrated, the first column returns to loading as the last column in series (or at the end of the loading zone), allowing another saturated column to undergo elution (t0+4∆t). The repetition of these steps enables a cyclic steady state to be established and constitutes a cyclic, optimized chromatography process.
The SMCC concept enables the productivity (kg of mAb per volume of resin per day) to be maximized by disconnecting the kinetics from the thermodynamics of capture step. In batch mode, there is a well-known correlation between the dynamic binding capacity (DBC; thermodynamics) and flow rate (kinetics): when the flow rate increases the amount of protein that can be captured before product breakthrough reduces (19), and thus requires compromises of the operating conditions. By separating this dependence, the SMCC mode unlocks the potential of a chromatography step and enables the use of a widened set of process conditions, as driven by the process objectives. For example, since the media is used more efficiently, its volume can be significantly reduced (up to 75%) compared with the batch equivalent given the same volume of feed to be purified. This would also imply a reduction in the column and system size, and buffer consumption.
5
Page 5 of 37
Based on the experience gained in the industry of small therapeutic molecules, the Food and Drug Administration (FDA), as well as the European Medicines Agency (EMEA), encourages increasing process understanding and control by the implementation of steady states in future manufacturing processes. Other potential advantages include a more rapid treatment of labile and unstable components (e.g., aggregation for mAbs and the deactivation of blood factors) by reducing the global process time and ultimately by removing holding steps for
ip t
fully continuous processes.
Industrial application of continuous chromatography in commercial processes have existed for decades in the
cr
food and fermentation industries (20) and the pharmaceutical industry with closed-loop Multi-Columns
Chromatography (MCC) (21) that is mostly adapted to the separation of binary mixtures, including chiral
us
molecules (such as UCB’s Keppra, Lundbeck’s Lexapro and Pfizer’s Zoloft). In the biopharmaceutical industry, there are several examples of hybrid designs integrating the concept of continuous operations (10). Today, a common design combines a continuous upstream step, for instance by using a perfusion bioreactor, with
an
discontinuous downstream steps based on batch operation. The association of a batch or fed-batch cell culture and a continuous chromatography step has been recently investigated as well and has proven to be economically
M
beneficial (22). More recently, a patent application describing an approach for a manufacturing process operated in continuous mode has been released (23), as a first step toward an integrated approach of continuous processing.
d
This work describes a downstream process combining two BioSC® for the protein A capture and intermediate ion
te
exchange steps. The objectives were a) to implement a 2-step continuous chromatography process supplemented by a final polishing step, and b) to design a platform for easy scale-up and industrialization by employing only
Ac ce p
industrial-compatible conditions (e.g., resins that are commercially available in large amount, common and simple buffers, sustainable operational conditions). This work outlines the method used to develop a DSP mAb platform with two continuous chromatography steps and discloses the qualitative and quantitative results of both individual and chained steps.
2. Materials and methods 2.1. Analytics
The purity (including aggregates contents) and concentration were determined by HPLC-SEC (TSK Gel G3000 SWXL, Tosoh Bioscience, 5µm on a 7.8 mm I.D. x 300 mm column). IgG purity was measured by SDS-PAGE. The leaching of protein-A ligands and the level of HCP were evaluated by commercial ELISA kits. DNA content was measured by picogreen method.
2.2. Antibody
6
Page 6 of 37
An IgG1 subtype monoclonal antibody was produced from batch cultures of a stable CHO cell line in a 50L singleuse bioreactor at Novasep Belgium, using a third-party process. The crudes were clarified by depth filtration. The characteristics of the clarified crudes are summarized in Table 1. The different crudes were used as-is for the
ip t
downstream processing development.
2.3. Screening
Only media that are commercially available in large amount for industrial preparative chromatography were
cr
selected for the screening: 2 protein-A media for the capture step (media A1 and A2)
11 ion exchange resins for the intermediate step (media IEX-1 to IEX-11)
2 chromatographic membranes for the polishing step (P1 and P2)
us
an
For all the steps (capture, intermediate and polishing), Design of Experiments (DoE) was used to define the screening of parameters including the media, salt concentration, buffers type and pH. To minimize crude and time usage and to optimize the tested matrix, screenings were performed in microplates. The responses followed for
M
the selection of the best parameters were the best recovery and purity of the IgG. For each step, the conditions chosen from the screening were tested and optimized using a batch chromatography column (0.5 mm I.D. x 5 cm
d
column).
te
2.4. Optimization on a batch column
For each step, parameters optimization was performed on lab-scale batch column (11 mm I.D. x 5 cm bed height)
Ac ce p
by injecting 80% of DBC10. The parameters that were optimized included the following:
Volume and mAb concentration of the load
Flow rate of the loading step
Buffer concentration, volumes and pH
Elution buffer conductivity (for ion exchange)
2.5. Simulation and SMCC recipe optimization An integrated experimental and modeling approach for the design of sequential multi-column chromatography recipes was developed by Novasep and has been detailed by Ng et al. (24, 25). The approach is summarized as followed: the relevant data (see Table 2) are collected during batch chromatography optimization and are incorporated into ColHelp, a software that enables the modeling-simulation of the chromatographic system and the in-silico optimization of the SMCC recipe. The details on the models used are found in (24) and (25). Several optimization strategies are available for selection by the user as a key driver of the process development and
7
Page 7 of 37
include the optimization (minimization) of resin volume, the minimization of buffer consumption and the reduction of process time. The simulated results based on the SMCC recipes provided by the software were experimentally verified and further optimized by adjusting the parameters, such as the loading flow rates and the column volumes
ip t
for the wash steps.
3. Results
The purification process developed in this work aimed to purify a monoclonal IgG1 produced in CHO that satisfies
cr
the pre-determined specifications (Table 3). A protein A capture step and ion exchange step both using SMCC,
us
and a polishing step using membrane chromatography in flow through mode were chained.
3.1. Development of the capture step
an
The SMCC parameters for the capture of the mAb were identified by following a 3-step screening approach with support by simulation. First, the washing and elution buffers were screened on microplates and selected using an experimental design approach (results not shown) with two different protein A media (A1 & A2). A similar
M
approach was adopted for both media, and thus only experimental details for one of the media (A2 as this was the selected protein A media) are provided.
d
At the end of the first screening step, the washing buffer (50mM Tris pH8.0) and elution buffer (100mM Glycine pH3.5) were both selected (for media A2) based on the best IgG recovery and purity observed under static
te
conditions using the microplates. These selected buffers were tested on a single chromatography column under
Ac ce p
dynamic conditions. The characterization and mass balance of the fractions collected (Table 4) revealed the formation of aggregates with media A2 during the run under the tested conditions. This formation of aggregates was further confirmed by centrifugation and SDS-PAGE. In order to reduce aggregation during the run, the concentration of Tris in the washing buffer (Figure 4) and different elution buffers (Figure 5) were further tested. Another possible solution is to reduce the IgG load but this option was not considered to maximize the productivity. As seen in Figure 4, the Tris concentration has a significant impact on IgG recovery and aggregate contents. Based on these results, 10mM Tris, pH8.0, was selected as the washing buffer. The data presented in Figure 5 demonstrate the impact of the pH and buffer type on the IgG recovery and the aggregate contents. Both the citrate buffer at pH3.8 and glycine buffer at pH3.7 are potential elution buffers as they presented the best IgG recovery with the lowest aggregate contents. Eventually the latter was selected as the elution buffer. Batch chromatography designs were optimized for both media A1 and A2. Parameters required for the design of SMCC were determined as described in Section 2.5. For media A2, the static binding capacity (SBC) was estimated at 41 g/L and the dynamic binding capacities (DBC) were determined for three velocities: 300 cm/h (39 g/L), 600 cm/h (34 g/L) and 900 cm/h (33 g/L). These parameters were used to model the chromatographic
8
Page 8 of 37
systems. The modeled systems were used to optimize chromatography recipes with the aid of simulation to achieve the best productivity while reducing the volume of resin used (Table 5). BioSC runs were performed in order to confirm the predictions of the model and to fine-tune the SMCC recipes. During the SMCC process finetuning, a loss of product was observed in the flow through when loading at 900 cm/h. The number of column volumes (CV) allocated to the equilibration step was reduced from 15 to 10 CV. This enabled the loading flow rate
ip t
to decrease from 900 to 830 cm/h without any product loss, instead of decreasing it to 750 cm/h without modifying equilibration step CVs. By maintaining a high flowrate for the loading zone, the productivity was maintained as
cr
well. A good agreement was observed between the simulated data (Table 5) and experimental data (Table 6).
The difference in the number of columns between SMCC recipes for media A1 and media A2 is a direct
us
consequence of the difference of performance between the two media, both in terms of the binding capacity and the maximum operating flow rate (media A2 has a higher capacity and could be used at a higher flow rate). In both cases, the productivity of the SMCC mode was around 4.5 folds higher than their batch equivalents, whereas
an
buffer consumption was around 1.1–1.4 fold lower. Productivity is defined as the amount of monoclonal antibody that is purified per liter of media over 24 hours operation, and it was calculated from the injected amount per cycle
M
multiplied by the number of cycles that can be performed in 24 hours. In the identified conditions, resin volume was reduced by 4-fold in SMCC mode versus batch mode when considering the same volume of feed to be treated in the same time. The quality attributes were similar between batch chromatography and SMCC, which
d
were in line with the expected results. The media A2 was the preferred protein A media for the capture step,
te
because of the higher productivity compared with that for media A1. This capture step provides a good example of the important role of batch chromatography for the development
Ac ce p
and optimization of continuous chromatography. A good understanding and the optimization of batch chromatography, prior to the development of continuous chromatography, is a prerequisite for SMCC process development.
3.2. Development of the intermediary step A similar approach was applied to develop the intermediate step based on an anion exchange media, using the eluate from the capture step:
The appropriate separation conditions were screened for 11 different media using microplates following an experimental design plan. The media IEX-2 was identified as the most promising and the loading and elution conditions were then optimized.
Simulations with ColHelp were used to determine SMCC parameters starting from breakthrough analysis. The first computational optimization performed, was driven by the optimization of media volume
9
Page 9 of 37
to limit resin-associated costs (Table 7, BioSC-1). An additional simulation was performed to assess the impact of a flow rate increase on the productivity.
BioSC runs were performed to verify the performance and to fine-tune the recipes (Table 8).
Once again, a good agreement between simulated and experimental data was observed. Compared to batch chromatography, the two recipes BioSC-1 and BioSC-2 demonstrated significantly improved performance in terms
ip t
of productivity, resin volume reduction and buffer consumption. After this step, the specifications targeted in Table 3 were reached for the protein A (
S0168-1656(15)00202-3 http://dx.doi.org/doi:10.1016/j.jbiotec.2015.04.026 BIOTEC 7100
To appear in:
Journal of Biotechnology
Received date: Revised date: Accepted date:
21-10-2014 13-4-2015 30-4-2015
Please cite this article as: Girard, V., Hilbold, N.-J., Ng, C.K.S., Pegon, L., Chahim, W., Rousset, F., Monchois, V.,Large Scale monoclonal antibody purification by continuous chromatography, from process design to scale-up, Journal of Biotechnology (2015), http://dx.doi.org/10.1016/j.jbiotec.2015.04.026 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Highlights A chained process including 2 SMCC steps and a polishing step was implemented for mAb purification Better process performance was achieved operating in SMCC mode compared with that in batch mode
ip t
The SMCC development was proved to be not as complicated as expected compared with that for batch
Ac ce p
te
d
M
an
us
cr
A better process understanding is expected through the implementation of SMCC
1
Page 1 of 37
[TITLE]
Large Scale monoclonal antibody purification by continuous chromatography, from process design to scale-up
Valérie GIRARD
ip t
[AUTHORS] [1]
[2]
us
Nicolas-Julian HILBOLD*
cr
[email protected]
[email protected]
[3]
an
Candy K. S. NG
[email protected]
[2]
M
Laurence PEGON
[1]
Ac ce p
[email protected]
te
Wael CHAHIM
d
[email protected]
Fabien ROUSSET
[2]
[email protected]
Vincent MONCHOIS
[2]
[email protected]
2
Page 2 of 37
[CONTACT INFORMATION] [1] Novasep Belgium 16 rue Clément Ader 6041 Gosselies
ip t
Belgium
cr
+32 71 347900
[2]
us
Novasep Process SAS 82 Boulevard de la Moselle Site Eiffel – BP50
an
54340 Pompey France
M
+33 383 49 70 00
23 Creek Circle – Boothwyn PA 19061
Ac ce p
USA
te
Novasep Inc.
d
[3]
+1 610 494 0447
[KEY WORDS]
Continuous biomanufacturing, Monoclonal antibody, Continuous chromatography, Antibody purification, Continuous bioprocessing, Downstream processing
3
Page 3 of 37
[ARTICLE CONTENT]
Abstract The development and optimization of a purification process of monoclonal antibodies based on two continuous chromatography steps for capture and intermediate purification is presented. The two chromatography steps were
ip t
individually optimized using either batch chromatography or Sequential Multi-Column Chromatography (SMCC) only. A proprietary simulation software was used to optimize SMCC and to evaluate the potential gains compared with batch chromatography. The SMCC recipes provided by the simulation software were evaluated
cr
experimentally. A good correlation was found between the simulated results and experimental observations.
Significant gains were observed on the productivity, buffer consumption and on the volume of resin required for
us
SMCC over batch chromatography. Based on these results, a chained process from the capture to polishing steps was implemented. This chained process demonstrated significantly better performance compared with the batch
an
equivalent whilst satisfying the specifications. The expected positive impact provided by implementing continuous chromatography is also discussed.
M
1. Introduction
The market of biotechnologies and biopharmaceuticals is continuously growing with a value of $20 billion in 2001
d
and is estimated to reach $216 billion in 2015 (1), which includes $79 billion solely for monoclonal antibodies (mAb) that have proven to be interesting therapeutic solutions for numerous diseases (2).
te
The advances in upstream processing over the last decades led to important improvements in the production yields and the process bottleneck has been displaced towards downstream processing (DSP). Depending on the
Ac ce p
process, purification steps can constitute up to 80% of the total manufacturing costs (3). Therapeutic doses of most antibodies are high, with 1 gram or more of active protein per dose being quite common, compared with the doses of other type of therapeutic proteins (4). With the current manufacturing modes and strategies, this high dosage implies very large scale production facilities for products that have successfully passed clinical trials, which usually implies high Capital Expenditures (CAPEX) and important Operational Expenses (OPEX). The product and process knowledge accumulated from the early 1990s, as well as the technological environment, led pharmaceutical companies to develop similar purification processes based on a common sequence of unit operations (5–9). Although some process adaptations and development have to be made given the variations from one mAb to another, this combination of common unit operations is nowadays considered as a typical mAb manufacturing process (Figure 1).This convergence of the manufacturing process is expected to accelerate and facilitate process development in order to generate better defined and cost-efficient processes. This definition of a general manufacturing process is also expected to facilitate process transfer (10) from a department to another (from R&D to pilot for example), from one facility to another within the company or to a custom manufacturing
4
Page 4 of 37
organization when outsourcing is considered. The purification of a target protein – consisting of capture, intermediate and polishing – mostly involves chromatography (11, 12) and is conventionally operated in batch as discontinuous steps, where the operation is expensive and slow (13). In batch mode chromatography, operating conditions do not allow a full utilization of the media’s total capacity (Figure 2, left), leading to a sub-utilization of the media (14, 15). Recent developments in continuous chromatography have demonstrated the potential of the
ip t
technology to overcome these problems (16) and the development of fully continuous downstream processes.
Sequential Multi-Column Chromatography (SMCC) is an open-loop process for the separation of multi-component
cr
mixtures using multiple buffers (17, 18). The concept behind SMCC is to perform the different steps in smaller
columns (from 2 to 8 columns; compared with batch chromatography) connected in series, to maximize media
us
utilization (Figure 2, right) without any product loss. The BioSC® is an automated system based on the SMCC process, designed and provided by Novasep. Depending on process conditions, it is operated with 2 to 6 columns. The feed is injected into the first column (Figure 3, t0). The product breakthrough is automatically
an
monitored by appropriate probes. In batch design, the operating conditions are set to stop loading before observing the product breakthrough. This is unlike the operation of SMCC, for which the unbound product is
M
adsorbed by the second column as the media in the first column becomes more and more saturated (t0+∆t). When the first column is saturated, a “Wash* step” (t0+2∆t) pushes the product in the column void volume into the next column to ensure all product will be recovered from the interstitial liquid. This feature is a major difference
d
compared with fully continuous open-loop SMB process, for which this fraction of target protein is lost. The
te
saturated column is automatically disconnected from the feed stream and undergoes successively the wash, elution, regeneration and equilibration steps, while the other columns remain connected to the feed stream to
Ac ce p
continue the capture process (t0+3∆t). Once equilibrated, the first column returns to loading as the last column in series (or at the end of the loading zone), allowing another saturated column to undergo elution (t0+4∆t). The repetition of these steps enables a cyclic steady state to be established and constitutes a cyclic, optimized chromatography process.
The SMCC concept enables the productivity (kg of mAb per volume of resin per day) to be maximized by disconnecting the kinetics from the thermodynamics of capture step. In batch mode, there is a well-known correlation between the dynamic binding capacity (DBC; thermodynamics) and flow rate (kinetics): when the flow rate increases the amount of protein that can be captured before product breakthrough reduces (19), and thus requires compromises of the operating conditions. By separating this dependence, the SMCC mode unlocks the potential of a chromatography step and enables the use of a widened set of process conditions, as driven by the process objectives. For example, since the media is used more efficiently, its volume can be significantly reduced (up to 75%) compared with the batch equivalent given the same volume of feed to be purified. This would also imply a reduction in the column and system size, and buffer consumption.
5
Page 5 of 37
Based on the experience gained in the industry of small therapeutic molecules, the Food and Drug Administration (FDA), as well as the European Medicines Agency (EMEA), encourages increasing process understanding and control by the implementation of steady states in future manufacturing processes. Other potential advantages include a more rapid treatment of labile and unstable components (e.g., aggregation for mAbs and the deactivation of blood factors) by reducing the global process time and ultimately by removing holding steps for
ip t
fully continuous processes.
Industrial application of continuous chromatography in commercial processes have existed for decades in the
cr
food and fermentation industries (20) and the pharmaceutical industry with closed-loop Multi-Columns
Chromatography (MCC) (21) that is mostly adapted to the separation of binary mixtures, including chiral
us
molecules (such as UCB’s Keppra, Lundbeck’s Lexapro and Pfizer’s Zoloft). In the biopharmaceutical industry, there are several examples of hybrid designs integrating the concept of continuous operations (10). Today, a common design combines a continuous upstream step, for instance by using a perfusion bioreactor, with
an
discontinuous downstream steps based on batch operation. The association of a batch or fed-batch cell culture and a continuous chromatography step has been recently investigated as well and has proven to be economically
M
beneficial (22). More recently, a patent application describing an approach for a manufacturing process operated in continuous mode has been released (23), as a first step toward an integrated approach of continuous processing.
d
This work describes a downstream process combining two BioSC® for the protein A capture and intermediate ion
te
exchange steps. The objectives were a) to implement a 2-step continuous chromatography process supplemented by a final polishing step, and b) to design a platform for easy scale-up and industrialization by employing only
Ac ce p
industrial-compatible conditions (e.g., resins that are commercially available in large amount, common and simple buffers, sustainable operational conditions). This work outlines the method used to develop a DSP mAb platform with two continuous chromatography steps and discloses the qualitative and quantitative results of both individual and chained steps.
2. Materials and methods 2.1. Analytics
The purity (including aggregates contents) and concentration were determined by HPLC-SEC (TSK Gel G3000 SWXL, Tosoh Bioscience, 5µm on a 7.8 mm I.D. x 300 mm column). IgG purity was measured by SDS-PAGE. The leaching of protein-A ligands and the level of HCP were evaluated by commercial ELISA kits. DNA content was measured by picogreen method.
2.2. Antibody
6
Page 6 of 37
An IgG1 subtype monoclonal antibody was produced from batch cultures of a stable CHO cell line in a 50L singleuse bioreactor at Novasep Belgium, using a third-party process. The crudes were clarified by depth filtration. The characteristics of the clarified crudes are summarized in Table 1. The different crudes were used as-is for the
ip t
downstream processing development.
2.3. Screening
Only media that are commercially available in large amount for industrial preparative chromatography were
cr
selected for the screening: 2 protein-A media for the capture step (media A1 and A2)
11 ion exchange resins for the intermediate step (media IEX-1 to IEX-11)
2 chromatographic membranes for the polishing step (P1 and P2)
us
an
For all the steps (capture, intermediate and polishing), Design of Experiments (DoE) was used to define the screening of parameters including the media, salt concentration, buffers type and pH. To minimize crude and time usage and to optimize the tested matrix, screenings were performed in microplates. The responses followed for
M
the selection of the best parameters were the best recovery and purity of the IgG. For each step, the conditions chosen from the screening were tested and optimized using a batch chromatography column (0.5 mm I.D. x 5 cm
d
column).
te
2.4. Optimization on a batch column
For each step, parameters optimization was performed on lab-scale batch column (11 mm I.D. x 5 cm bed height)
Ac ce p
by injecting 80% of DBC10. The parameters that were optimized included the following:
Volume and mAb concentration of the load
Flow rate of the loading step
Buffer concentration, volumes and pH
Elution buffer conductivity (for ion exchange)
2.5. Simulation and SMCC recipe optimization An integrated experimental and modeling approach for the design of sequential multi-column chromatography recipes was developed by Novasep and has been detailed by Ng et al. (24, 25). The approach is summarized as followed: the relevant data (see Table 2) are collected during batch chromatography optimization and are incorporated into ColHelp, a software that enables the modeling-simulation of the chromatographic system and the in-silico optimization of the SMCC recipe. The details on the models used are found in (24) and (25). Several optimization strategies are available for selection by the user as a key driver of the process development and
7
Page 7 of 37
include the optimization (minimization) of resin volume, the minimization of buffer consumption and the reduction of process time. The simulated results based on the SMCC recipes provided by the software were experimentally verified and further optimized by adjusting the parameters, such as the loading flow rates and the column volumes
ip t
for the wash steps.
3. Results
The purification process developed in this work aimed to purify a monoclonal IgG1 produced in CHO that satisfies
cr
the pre-determined specifications (Table 3). A protein A capture step and ion exchange step both using SMCC,
us
and a polishing step using membrane chromatography in flow through mode were chained.
3.1. Development of the capture step
an
The SMCC parameters for the capture of the mAb were identified by following a 3-step screening approach with support by simulation. First, the washing and elution buffers were screened on microplates and selected using an experimental design approach (results not shown) with two different protein A media (A1 & A2). A similar
M
approach was adopted for both media, and thus only experimental details for one of the media (A2 as this was the selected protein A media) are provided.
d
At the end of the first screening step, the washing buffer (50mM Tris pH8.0) and elution buffer (100mM Glycine pH3.5) were both selected (for media A2) based on the best IgG recovery and purity observed under static
te
conditions using the microplates. These selected buffers were tested on a single chromatography column under
Ac ce p
dynamic conditions. The characterization and mass balance of the fractions collected (Table 4) revealed the formation of aggregates with media A2 during the run under the tested conditions. This formation of aggregates was further confirmed by centrifugation and SDS-PAGE. In order to reduce aggregation during the run, the concentration of Tris in the washing buffer (Figure 4) and different elution buffers (Figure 5) were further tested. Another possible solution is to reduce the IgG load but this option was not considered to maximize the productivity. As seen in Figure 4, the Tris concentration has a significant impact on IgG recovery and aggregate contents. Based on these results, 10mM Tris, pH8.0, was selected as the washing buffer. The data presented in Figure 5 demonstrate the impact of the pH and buffer type on the IgG recovery and the aggregate contents. Both the citrate buffer at pH3.8 and glycine buffer at pH3.7 are potential elution buffers as they presented the best IgG recovery with the lowest aggregate contents. Eventually the latter was selected as the elution buffer. Batch chromatography designs were optimized for both media A1 and A2. Parameters required for the design of SMCC were determined as described in Section 2.5. For media A2, the static binding capacity (SBC) was estimated at 41 g/L and the dynamic binding capacities (DBC) were determined for three velocities: 300 cm/h (39 g/L), 600 cm/h (34 g/L) and 900 cm/h (33 g/L). These parameters were used to model the chromatographic
8
Page 8 of 37
systems. The modeled systems were used to optimize chromatography recipes with the aid of simulation to achieve the best productivity while reducing the volume of resin used (Table 5). BioSC runs were performed in order to confirm the predictions of the model and to fine-tune the SMCC recipes. During the SMCC process finetuning, a loss of product was observed in the flow through when loading at 900 cm/h. The number of column volumes (CV) allocated to the equilibration step was reduced from 15 to 10 CV. This enabled the loading flow rate
ip t
to decrease from 900 to 830 cm/h without any product loss, instead of decreasing it to 750 cm/h without modifying equilibration step CVs. By maintaining a high flowrate for the loading zone, the productivity was maintained as
cr
well. A good agreement was observed between the simulated data (Table 5) and experimental data (Table 6).
The difference in the number of columns between SMCC recipes for media A1 and media A2 is a direct
us
consequence of the difference of performance between the two media, both in terms of the binding capacity and the maximum operating flow rate (media A2 has a higher capacity and could be used at a higher flow rate). In both cases, the productivity of the SMCC mode was around 4.5 folds higher than their batch equivalents, whereas
an
buffer consumption was around 1.1–1.4 fold lower. Productivity is defined as the amount of monoclonal antibody that is purified per liter of media over 24 hours operation, and it was calculated from the injected amount per cycle
M
multiplied by the number of cycles that can be performed in 24 hours. In the identified conditions, resin volume was reduced by 4-fold in SMCC mode versus batch mode when considering the same volume of feed to be treated in the same time. The quality attributes were similar between batch chromatography and SMCC, which
d
were in line with the expected results. The media A2 was the preferred protein A media for the capture step,
te
because of the higher productivity compared with that for media A1. This capture step provides a good example of the important role of batch chromatography for the development
Ac ce p
and optimization of continuous chromatography. A good understanding and the optimization of batch chromatography, prior to the development of continuous chromatography, is a prerequisite for SMCC process development.
3.2. Development of the intermediary step A similar approach was applied to develop the intermediate step based on an anion exchange media, using the eluate from the capture step:
The appropriate separation conditions were screened for 11 different media using microplates following an experimental design plan. The media IEX-2 was identified as the most promising and the loading and elution conditions were then optimized.
Simulations with ColHelp were used to determine SMCC parameters starting from breakthrough analysis. The first computational optimization performed, was driven by the optimization of media volume
9
Page 9 of 37
to limit resin-associated costs (Table 7, BioSC-1). An additional simulation was performed to assess the impact of a flow rate increase on the productivity.
BioSC runs were performed to verify the performance and to fine-tune the recipes (Table 8).
Once again, a good agreement between simulated and experimental data was observed. Compared to batch chromatography, the two recipes BioSC-1 and BioSC-2 demonstrated significantly improved performance in terms
ip t
of productivity, resin volume reduction and buffer consumption. After this step, the specifications targeted in Table 3 were reached for the protein A (
