Continuous Crystallization of Proteins in a Stirred Classified Product

Jun 26, 2017 - Dariusch Hekmat† , Max Huber†, Christian Lohse†, Nikolas von den ... of Munich, Ernst-Otto-Fischer-Strasse 3, 85748 Garching, Ger...
0 downloads 0 Views 2MB Size
Subscriber access provided by NEW YORK UNIV

Article

Continuous crystallization of proteins in a stirred classified product removal tank with a tubular reactor in bypass Dariusch Hekmat, Max Huber, Christian Lohse, Nikolas von den Eichen, and Dirk Weuster-Botz Cryst. Growth Des., Just Accepted Manuscript • DOI: 10.1021/acs.cgd.7b00436 • Publication Date (Web): 26 Jun 2017 Downloaded from http://pubs.acs.org on June 26, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Crystal Growth & Design is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 35

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Crystal Growth & Design

Continuous crystallization of proteins in a stirred classified product removal tank with a tubular reactor in bypass Dariusch Hekmat1*, Max Huber1, Christian Lohse1, Nikolas von den Eichen1, Dirk Weuster-Botz1,2 1 Institute of Biochemical Engineering, Technical University of Munich, Boltzmannstr. 15, 85748 Garching, Germany 2 Research Center for Industrial Biotechnology, Technical University of Munich, Ernst-Otto-Fischer-Str. 3, 85748 Garching, Germany Abstract: A lab-scale stirred tank with a cooled tubular reactor in bypass was applied for continuous crystallization of lysozyme and a full-length therapeutic monoclonal antibody. The stirred tank was operated as a mixed suspension classified product removal crystallizer. Lysozyme was crystallized by a combination of cooling crystallization and salting-out. The antibody was crystallized by a combination of cooling crystallization and isoelectric crystallization. It was deduced that nucleation rates were enhanced when the protein solutions passed through the cooled tubular bypass. It was further deduced that crystal growth rates were enhanced in the stirred tank which was operated at a higher temperature compared to the tubular reactor. Classified product removal was possible by controlled sedimentation of protein crystals. The continuous crystallization system allowed a targeted control of crystal morphology and size (Figure 1). No sedimentation occurred in the tubular reactor and precipitation was avoided at all times. High crystallization yields of more than 90% were obtained. Crystals of the monoclonal antibody were continuously produced for the first time with a space-time-yield of up to 12 g L-1 h-1.

Figure 1. Time course of protein concentration in solution for two independent continuous crystallization runs using pre-treated harvest of a therapeutic monoclonal antibody. Run #1; Run #2. Orthorhombic-like crystals were obtained.

*Corresponding author: Dariusch Hekmat Institute of Biochemical Engineering, Technical University of Munich Boltzmannstr. 15, 85748 Garching, Germany Tel: +49-89-289-15770, Fax: +49-89-289-15714. E-mail: [email protected], Web-address: http://www.biovt.mw.tum.de 1 ACS Paragon Plus Environment

Crystal Growth & Design

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 35

Continuous crystallization of proteins in a stirred classified product removal tank with a tubular reactor in bypass

Dariusch Hekmat1*, Max Huber1, Christian Lohse1, Nikolas von den Eichen1, Dirk Weuster-Botz1,2 1

Institute of Biochemical Engineering, Technical University of Munich, Boltzmannstr.

15, 85748 Garching, Germany 2

Research Center for Industrial Biotechnology, Technical University of Munich, Ernst-

Otto-Fischer-Str. 3, 85748 Garching, Germany

Abstract: A lab-scale stirred tank with a cooled tubular reactor in bypass was applied for continuous crystallization of lysozyme and a full-length therapeutic monoclonal antibody. The stirred tank was operated as a mixed suspension classified product removal crystallizer. Lysozyme was crystallized by a combination of cooling crystallization and salting-out. The antibody was crystallized by a combination of cooling crystallization and isoelectric crystallization. It was deduced that nucleation rates were enhanced when the protein solutions passed through the cooled tubular bypass. It was further deduced that crystal growth rates were enhanced in the stirred tank which was operated at a higher temperature compared to the tubular reactor. Classified product removal was possible by controlled sedimentation of protein crystals. The continuous crystallization system allowed a targeted control of crystal morphology and size. No sedimentation occurred in the tubular reactor and precipitation was avoided at all times. High crystallization yields of more than 90% were obtained. Crystals of the monoclonal antibody were continuously produced for the first time with a space-time-yield of up to 12 g L-1 h-1. 2 ACS Paragon Plus Environment

Page 3 of 35

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Crystal Growth & Design

Introduction

Many proteins, in particular therapeutic proteins like monoclonal antibodies, need to be available in a highly purified form in large amounts and at reduced costs compared to the present state-of-the-art. In order to achieve the required high level of purity, several sequential preparative batch chromatography steps are usually applied. However, batch chromatography procedures are known to have low purification capacities, to be time-consuming, and to require expensive consumables. As a result, chromatography may represent an unfavorable downstream processing bottleneck, especially at large scale.1 In order to implement overall process intensification, conversion from batch to continuous operation has been proposed.2-4 Continuous chromatography may be carried out using multiple column systems such as simulated moving bed (SMB) or periodic counter-current chromatography (PCC).5 However, such systems require sophisticated and expensive equipment like accurate and stable process analytical technology, in particular for monitoring protein concentrations via UV detectors as well as complex real-time control technology.6 Therefore, non-chromatographic alternatives have been addressed in the past like precipitation or crystallization.7-9 Successful protein batch precipitation processes of therapeutic proteins in stirred tanks have been described in literature.10-12 Recently, continuous precipitation of recombinant monoclonal antibodies in tubular reactors has been reported.13,14 It was shown that after resolubilizing the precipitated antibodies, the secondary structure and the isoform distribution did not change compared with material purified by protein A chromatography. Successful applications of batch crystallization of proteins, including recombinant proteins, in stirred tank reactors have also been described in literature.15-21 However, crystallization of full-length monoclonal antibodies has mainly 3 ACS Paragon Plus Environment

Crystal Growth & Design

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 35

been reported only on the non-agitated µL-scale (i.e. vapor diffusion and/or microbatch experiments).22-31 Large-scale batch crystallization of a full-length therapeutic monoclonal antibody mAb01 was achieved for the first time by Smejkal et al. in 2013.32 In this work, batch crystallization of mAb01 from pure solutions as well as from pre-treated clarified CHO cell culture harvest was performed. It was shown that the antibody crystals had a high bioactivity and could be resolubilized easily for recrystallization and formulation. The purification efficiency of crystallization was similar to protein A chromatography as reported earlier for another monoclonal antibody.30 The antibody crystals had an increased mechanical stability and enhanced resolubilization kinetics compared to precipitated antibodies. Batch crystallization of the antibody was successfully scaled-up from the non- agitated µLscale to the L-scale in stirred tanks. It was further shown that large-scale protein batch crystallization in a stirred tank (e.g. 1,000 L) is conceivable by applying the maximum local energy dissipation as a suitable scale-up criterion.33 To the knowledge of the authors, continuous crystallization of proteins has so far only been published once.34 In this work, lysozyme was crystallized in a tubular reactor. The tubular reactor was fed with a buffered protein solution containing 100 g L-1 lysozyme and 16 g L-1 of the crystallization agent sodium chloride and a solution containing 64 g L-1 sodium chloride being mixed prior to entry at a ratio of 1:1 (v/v). Crystallization took place via single pass flow through a 13 m long tubular reactor with an inner diameter of 2 mm at a low linear flow rate of 1.9 mm s-1 resulting in a residence time of 113 min. A narrow residence time distribution (close to plug flow) was achieved by segmenting the flowing liquid into well-mixed slugs by adding air bubbles pulswise at the entry. The tubular reactor was split into three sections each being immersed in water baths with slightly varying temperatures hereby adjusting three different supersaturation levels. Thus, nucleation rates and crystal growth rates 4 ACS Paragon Plus Environment

Page 5 of 35

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Crystal Growth & Design

varied along the tubular crystallizer. Tetragonal lysozyme crystals with a narrow size distribution and a maximum crystal size of 20-25 µm were obtained. The continuous crystallization rate was 0.72 g h-1 at a relatively low yield of 68%. In order to increase the yield, longer residence times would have been required which, however, would lead to excessively long tubular reactors. On the other hand, longer residence times can be achieved in continuous stirred tanks. Continuous stirred tank crystallizers have been used in the past for crystallization of small molecules. Frequently, cooling crystallization has been applied where efficient cooling is provided by an external bypass. This bypass can be a tube or tube bundle heat exchanger. Such a configuration can be scaled-up without difficulty since the external bypass provides flexibility regarding the extent of required heat transfer surface area as well as improved heat flux rates compared to a stirred tank without bypass. Therefore, the aim of the present study was to apply a combination of a stirred tank with a cooled tubular reactor in bypass for efficient and easily up-scalable continuous crystallization of proteins. The stirred tank will be fed continuously with a buffered protein solution and the crystallization agent. High nucleation rates will be achieved constantly when the protein solution passes through the cooled tubular bypass due to increased supersaturation. The stirred tank will be operated as a mixed suspension classified product removal crystallizer. The protein slurry with the settled larger protein crystals will be withdrawn at the bottom of the stirred tank in a quasi-continuous manner. In such a configuration, desirable long mean residence times can be achieved in the stirred tank in order to obtain a high yield. Continuous crystallization of lysozyme will be studied and compared to the data published so far.34 Continuous crystallization of the above mentioned full-length therapeutic monoclonal antibody mAb01 will be

5 ACS Paragon Plus Environment

Crystal Growth & Design

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

studied for the first time. Both proteins exhibit a decreasing solubility with decreasing temperature (like many proteins). The continuous experiments to be performed require large amounts of protein which is particularly challenging in the case of the antibody. Furthermore, the conditioning of the protein feed solutions and the continuous crystallization experiments themselves are quite laborious and time-consuming. Therefore, it was not the aim of the present work to perform a systematic study. Instead, three exemplary continuous runs with cooling of the tubular bypass, one for lysozyme and two for the antibody will be performed and analyzed quantitatively. The continuous crystallization processes will be evaluated in terms of attainable crystallization rate, yield, and space-time-yield as well as the quality of obtained crystal morphology and reproducibility.

Experimental Section Proteins and chemicals. Purified lysozyme from chicken egg white was obtained from Sigma-Aldrich Chemie GmbH, Taufkirchen, Germany (No. 62971). Clarified harvest from CHO cell cultivation containing about 2.3 g L-1 of the humanized fulllength therapeutic monoclonal IgG1 type antibody mAb01 was provided by Novartis Pharma AG, Basel, Switzerland. The harvest contained about 267,000 ppm of host cell proteins (HCP) (equal to ~21% w/v total protein) and about 77,700 ppb DNA. All chemicals were analytical grade purchased from Sigma-Aldrich Chemie GmbH, Taufkirchen, Germany, Carl Roth GmbH & Co. KG, Karlsruhe, Germany, and VWR International GmbH, Bruchsal, Germany. Pre-treatment of mAb01 harvest. Pre-treatment of the clarified harvest of mAb01 was necessary in order to increase protein concentration via ultrafiltration (UF) and decrease the ionic strength of the solution via diafiltration (DF).32 At first, a pH-shift was carried out from 7.2 to 5.0 of 9 L of clarified harvest in a stirred tank using 10% 6 ACS Paragon Plus Environment

Page 6 of 35

Page 7 of 35

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Crystal Growth & Design

acetic acid. Precipitated HCP was removed by microfiltration using a 0.5 µm multilayered pleated filter cartridge type WFMLP0.5 (Wolftechnik Filtersysteme GmbH & Co. KG, Weil der Stadt, Germany). Then, cross-flow ultrafiltration was applied using a benchtop UF/DF equipment type Sartorius Slice (Sartorius Stedim Biotech GmbH, Göttingen, Germany) equipped with a polyethersulfone membrane (0.1 m2, 10 kDa MWCO) in a Sartocon Slice holder. The transmembrane pressure drop was 2 bar using a SartoJet membrane pump. The harvest was concentrated about 4-fold. A diafiltration step using 10 mM histidine/acetate buffer at pH 5.0 followed lasting seven diafiltration volumes. Precipitate was removed again using the 0.5 µm filter cartridge. Afterwards, a second ultrafiltration step was applied to concentrate the solution again about 4-fold. Finally, a dead-end microfiltration was performed using a tandem filter type Opticap XL 300 SHC 0.5/0.2 µm (Merck Millipore, Darmstadt, Germany). Protein assays. Concentrations of protein solutions were measured by standard UV absorption at 280 nm using a quartz cuvette with a path length of 10 mm in a UV/VIS spectrophotometer type Genesis 10S (Thermo Fisher Scientific GmbH, Dreieich, Germany). The extinction coefficients were 2.50 L g-1 cm-1 for lysozyme and 1.62 L g1

cm-1 for mAb01. A linear correlation with protein concentration was obtained at

absorption values within the range of 0.2-0.9. Standard sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and standard micro-agglutination IgG assays were performed to analyze the mAb01 harvest, the mAb01 harvest after pre-treatment, and the supernatant of the mAb01 crystal suspension obtained from the stationary phase of the continuous crystallizations. The assays for HCP content and DNA content are described elsewhere.32 The amount of crystal water of mAb01 crystals was measured via thermogravimetric analysis using a thermo-microbalance type STA 449 C Jupiter (Netzsch-Gerätebau GmbH, Selb, Germany). A dynamic temperature profile from 30-120 °C with a heating rate of 0.2 K min-1 was applied. 7 ACS Paragon Plus Environment

Crystal Growth & Design

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 35

This corresponded to a drying time of 7.5 h. An inert condition in the sample chamber was maintained by a constant nitrogen flow rate of 30 mL min-1. Protein crystallization experiments. The experiments were performed using a continuously operated stirred tank with a cooled tubular reactor in bypass.

A

schematic diagram of the experimental set-up is given in Figure 1.

Protein solution

Crystallization agent solution

T1

T2

Crystal sediment

Micro-vibrating motors Product discharge

Figure 1. Schematic diagram of the continuous crystallization system consisting of a stirred classified product removal tank with a tubular reactor in bypass. The stirred tank had no temperature control and was exposed to room temperature. The tubular reactor was immersed in a refrigerated batch thermostat such that T1 < T2.

Similar combinations have been described in literature for crystallization of small molecules.35,36 The stirred tank was made of an unbaffled glass tube with round bottom and an outlet at the bottom (Figure 2). The inner diameter of the stirred tank was D = 49 mm. The working volume was approximately 150 mL (including the volume of the crystal sediment). A three-bladed segment impeller with a diameter of 8 ACS Paragon Plus Environment

Page 9 of 35

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Crystal Growth & Design

24.8 mm and a height of 18 mm was used. The material of impeller and impeller axis was stainless steel. The ratio of impeller diameter d and D was approximately 0.5. The stirrer was positioned nearly equidistant between fluid fill level and sediment level. The ratio of height to diameter of the mixed crystal slurry column above the crystal sediment was about 1.5. The large flat blades had an inclination angle of 45° which produced a simultaneous axial and radial low flow with low shear forces at low stirrer speed. This impeller was derived from mammalian cell culture bioreactors to ensure gentle mixing of the crystal slurry.37 The stirring speed was controllable. This configuration of the stirred tank crystallizer ensured well adjustable hydrodynamic conditions with low shear forces as described previously.20,21

20 mm

Figure 2. Photograph of continuous stirred tank crystallizer equipped with a threebladed segment impeller. Upper dashed line indicates the fluid fill level (V ≈ 150 mL). Lower dashed line indicates the level of the protein crystal sediment (h ≈ 20 mm).

9 ACS Paragon Plus Environment

Crystal Growth & Design

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 35

The temperature of the stirred tank was determined by the preset temperature of the cooled tubular bypass and stayed constant at a level which resulted from the difference of room temperature and temperature of the tubular bypass. The stirred tank crystallizer was fed with protein solution and a crystallization agent solution using a multi-channel peristaltic pump type Ismatec Reglo ICC4 (tube inner diameter 0.25 mm) (Cole Palmer GmbH, Wertheim, Germany). Fast mixing of the solutions was achieved by locating the outlets of both feeding tubes next to the outer diameter of the stirred tank impeller. Protein crystal sediment slurry was discharged from the stirred tank crystallizer using a low-shear peristaltic pump (Masterflex drive No. 752135 with standard pump head type L/S 14 (tube inner diameter 1.6 mm) (Cole Palmer GmbH, Wertheim, Germany). Due to lack of an adequate sensor to monitor the upper level of the protein crystal sediment, the discharge pump was operated intermittently using a clock timer and manual supervision to avoid larger deviations of the level. The tubular reactor consisted of silicone tubing with a length of 10 m, an inner diameter of 2 mm and a working volume of approximately 30 mL. Recirculation of the protein crystal slurry was maintained by a low-shear peristaltic pump with a larger pump head operating at lower speed in order to minimize crystal attrition (Masterflex drive No. 7521-35 with standard pump head type L/S 16 (tube inner diameter 3.1 mm) (Cole Palmer GmbH, Wertheim, Germany). The linear flow velocity in the tubular reactor was adjusted within the range of 8.5-17 cm s-1. This resulted in a laminar flow regime (Reynolds number Re = 170-340 assuming a viscosity of 1 mPa s for the crystal fines slurry). The axial dispersion coefficient was not measured. The tubular reactor was immersed in a programmable refrigerated bath thermostat type 1157P (PolyScience, Niles, USA) for efficient cooling of the recirculated protein crystal slurry. The pressure drop of the tubular reactor was measured by a digital

10 ACS Paragon Plus Environment

Page 11 of 35

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Crystal Growth & Design

manometer type Greisinger GMH3181 (GHM Messtechnik GmbH, Regenstauf, Germany). Resolubilization of mAb01 crystals. Crystal suspension of mAb01 was centrifuged at 3260 g for 5 min using a floor-standing centrifuge type Rotixa 50 RS (Andreas Hettich GmbH & Co. KG, Tuttlingen, Germany). The supernatant was discarded and the crystal pellets were resolubilized in 10 mM histidine/acetate buffer at pH 5.0 while being stirred. Light microscopy and image analysis. Microphotographs of the protein crystals were obtained using a light microscope type Axioplan equipped with two objectives (Plan-Neofluar 40x/1.30, Plan-Neofluar 100x/1.30) and a digital camera type AxioCam ICc3 (Carl Zeiss AG, Oberkochen, Germany). The image processing software was AxioVision SE64, version 4.9 (Carl Zeiss AG, Oberkochen, Germany).

Results and Discussion Continuous crystallization of purified lysozyme. Each continuous crystallization run in the combined crystallization system was preceded by a batch phase. The batch process was started by mixing a solution containing 96 g L-1 lysozyme and a crystallization agent solution containing 80 g L-1 sodium chloride with a volume of 90 mL each at a ratio of 1:1 v/v and a stirrer speed of 200 min-1. Both solutions contained a 50 mM sodium acetate buffer at pH 4.6. The total reaction volume was 180 mL. The recirculation pump of the tubular reactor was adjusted to a linear flow velocity of 17 cm s-1. The concentration of lysozyme in solution was monitored by UV-assay. The process was switched to continuous operation after intense turbidity of the solution was observed indicating the start of crystal growth and the concentration of lysozyme in solution had showed a steep decline. At this point, 11 ACS Paragon Plus Environment

Crystal Growth & Design

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 35

continuous feeding of the stirred tank of a solution containing 192 g L-1 lysozyme and a solution containing 80 g L-1 sodium chloride at a ratio of 1:1 v/v commenced. Both solutions contained 50 mM sodium acetate buffer at pH 4.6. At the same time, the stirrer speed was reduced to 50 min-1. The height of the protein crystal sediment was controlled within a range of 15-25 mm by intermittent operation of the discharge pump via a clock timer in order to achieve quasi-continuous product removal. The first crystallization run was performed at room temperature in order to identify experimental key parameters. An appropriate flow velocity of the tubular reactor was set in order to prevent sedimentation of crystals in the tubular reactor. An adequate stirring rate was set in order to achieve sedimentation of larger crystals and maintain a well but gently mixed crystal fines slurry column in the stirred tank. The protein feed flow rate was adjusted to a maximum value in order to achieve stationary operation with a maximized crystallization rate. An adequate level of sedimented crystals in the stirred tank was achieved by adjusting the operating time of the discharge pump. These settings were adapted to the conditions of the subsequent experiments with cooling using lysozyme and the antibody. The induction time of the first crystallization run at room temperature took about 4 h. Continuous operation was started after about 6 h. Sedimentation of lysozyme crystals in the stirred tank was observed a few minutes after start of the continuous operation. The crystal fines slurry above the sediment had a uniform turbidity during continuous operation. Hence, the slurry was well mixed at all times. A photograph of the continuous stirred tank crystallizer during stationary operation is given in Figure 3.

12 ACS Paragon Plus Environment

Page 13 of 35

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Crystal Growth & Design

20 mm

Figure 3. Left: Photograph of stirred tank crystallizer during continuous stationary operation at room temperature using lysozyme. Upper right: Lysozyme crystals of recirculated crystal fines suspension. Lower right: Lysozyme crystals of discharged sediment slurry.

Tetragonal-like crystals were obtained. The size distributions of the recirculated crystal fines suspension and the discharged crystal sediment slurry are presented in Figure 4.

13 ACS Paragon Plus Environment

Crystal Growth & Design

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 35

Figure 4. Comparison of lysozyme crystal size distributions and cumulative distributions of recirculated crystal suspension ( slurry (

) and discharged crystal sediment

) during continuous crystallization at room temperature after 21 h of

continuous operation.

The mean crystal size of the sediment was approximately 17 µm and was about a factor of 2-3 larger than the recirculated crystals. A second continuous crystallization run using lysozyme was performed with reduced operating temperatures. The tubular reactor was cooled down to T1 = 10 °C resulting in a temperature of the stirred tank crystallizer T2 = 13 °C. The time course of lysozyme concentration in solution is presented in Figure 5.

14 ACS Paragon Plus Environment

Page 15 of 35

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Crystal Growth & Design

Figure 5. Time course of lysozyme concentration in solution for a continuous crystallization run. Start of continuous operation after 2 h. Cooling of the tubular crystallizer to T1 = 10 °C. Temperature of stirred tank crystallizer T2 = 13 °C. Lysozyme feed concentration = 96 g L-1, NaCl feed concentration = 40 g L-1, V = 6 mL h-1. Tetragonal-like crystals were obtained.

As can be seen, crystal growth started almost instantly. Hence, the induction time was significantly reduced compared to the crystallization run at room temperature. Solubility of lysozyme decreases with decreasing temperature.38,39 Furthermore, solubility of lysozyme decreases with increasing salt concentration. For low salt concentrations, lysozyme solubility has a pronounced dependency on temperature. For higher salt concentrations, the dependency of lysozyme solubility on temperature 15 ACS Paragon Plus Environment

Crystal Growth & Design

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 35

is lower40 (see Supporting Information, Figure S1). The supersaturation is defined as c/c* where c is the actual protein concentration and c* is the equilibrium protein concentration. The supersaturation of lysozyme during the continuous crystallization at room temperature was 2.0. The supersaturation of lysozyme during the continuous crystallization run in the stirred tank at 13 °C was 4.27.39 Supersaturation was increased to 5.26 in the tubular reactor at 10 °C. The higher overall supersaturation of the continuous crystallization run with cooling of the tubular reactor compared to the continuous crystallization run at room temperature led to increased nucleation resulting in a pronounced reduction of induction time. A similar observation was described in literature where fine-tuning of the supersaturation level allowed control of nucleation.34 As a result of reduced induction time, continuous operation was already started after 2 h. Stationary operation was reached fast at a feed rate of 6 mL h-1.The recirculation ratio

(defined

as

V 

 /V )

was

320.

The

stationary

lysozyme

concentration was 7 g L-1, and the crystallization rate was 0.58 g h-1. This value was in the same order of magnitude as reported in literature.34 The yield of the combined crystallization system was 93%. As expected, the yield was higher compared to literature.34 A space-time-yield of 3.2 g L-1 h-1 was achieved. Again, tetragonal-like lysozyme crystals were obtained. However, the mean size of the discharged crystals was slightly lower at 13 µm compared to the crystallization run at room temperature. No crystal sedimentation occurred in the tubular reactor. The pressure drop in the tubular reactor during the stationary phase was constant at approximately 200 mbar. Precipitation of lysozyme was avoided at all times.

Continuous crystallization of antibody mAb01 from pre-treated harvest. Previous batch crystallization experiments of mAb01 revealed that this protein 16 ACS Paragon Plus Environment

Page 17 of 35

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Crystal Growth & Design

crystallizes best in solutions with low ionic strength.41 Contrary to lysozyme, crystallization of mAb01 via addition of a crystallization agent like NaCl did not yield satisfactory results. However, mAb01 solubility exhibits a typical U-shaped dependency on solution pH (see Supporting Information, Figure S2). Furthermore, the solubility of mAb01 decreases with decreasing temperature like lysozyme. Therefore,

crystallization

of

mAb01

was

performed

at

reduced

operating

temperatures near the isoelectric point (pI = 6.8) by adding TRIS base as the crystallization agent. Again, the continuous crystallization run was preceded by a batch phase. The batch process without operation of the bypass was started with 180 mL of pre-treated harvest which was titrated to a pH of 6.8 by 1 M TRIS base at a stirrer speed of 200 min-1. The start solution contained approximately 16 g L-1 protein. The crystallization temperature of the stirred tank was controlled at 10 °C via immersion in a refrigerating bath. The obtained mAb01 crystals were orthorhombiclike (see Supporting Information, Figure S3). During the initial crystal growth phase (after 30 min), long needle-like crystal rods were formed with a mean length of about 32 µm and a mean width of about 2.3 µm. Hence, the average length-to-width ratio was about 14. These crystals grew predominantly in width and after 1 h, robust crystal rods with a mean length of about 32 µm and a mean width of about 5.3 µm were formed. Hence, the average length-to-width ratio was now about 6. The process was switched to continuous operation by feeding of a solution with about 16 g L-1 protein in 10 mM histidine buffer and a 80 mM TRIS base solution at a ratio of 10:1. This ratio resulted in a pH of 6.8 - 6.9. At the same time, the stirrer speed was reduced to 50 min-1. The height of the crystal sediment during continuous operation was again controlled by intermittent operation of the discharge pump as mentioned above.

17 ACS Paragon Plus Environment

Crystal Growth & Design

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 35

Two continuous crystallization runs were performed using pre-treated mAb01 harvest. The tubular reactor was cooled down to T1 = 0 °C resulting in a temperature of the stirred tank crystallizer T2 = 10 °C. The time courses of total protein concentration in solution for the crystallization runs are presented in Figure 6.

Figure 6. Time courses of total protein concentration in solution for two independent continuous crystallization runs using pre-treated mAb01 harvest containing 21% (w/v) HCP. Start of continuous operation after 2 h. Cooling of the tubular crystallizer to T1 = 0 °C. Temperature of stirred tank crystallizer T2 = 10 °C. Protein feed concentration = 14.5 g L-1,

Run #1: V = 77 mL h-1.

Run #2: V = 85 mL h-1. Orthorhombic-

like crystals were obtained. The measurement error was approximately 8 %.

18 ACS Paragon Plus Environment

Page 19 of 35

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Crystal Growth & Design

The concentration of total protein in solution was monitored by UV-assay. For reasons of simplification, the UV extinction coefficients of mAb01 and HCP were assumed to be identical. As can be seen, induction time was low and continuous operation was started after 2 h. It was observed during continuous crystallization that agglomerates of native mAb01 crystals were formed in the sediment. Most of these agglomerates were disintegrated by applying controlled vibration to the lower part of the stirred tank during operation of the discharge pump using circumferentially attached micro-vibrating motors (see Figure1 and Supporting Information). As a result, singular orthorhombic-like crystals and some crystal fragments were obtained as shown in Figure 6. The mean crystal size was 39 µm and the amount of crystals