(279e) In-Situ Monitoring during Mixing for Protein Crystallisation

Bio-active peptides are the next generation of pharmaceuticals which lead to a significant increase in demand over the last years [1]. Remarkable improvements in the upstream processing of peptides has been achieved, leading to a shift in the bottleneck to downstream purification [2]. Recently, crystallisation has emerged as a novel technology to replace expensive downstream purification techniques in the production line of bio-active peptides in order to increase economic efficiency [3, 4]. Due to their high molecular weight, low diffusivity, and soft and fragile nature, peptides require a gentle, but intensive mixing in order to achieve the same level of crystallisation reproducibility at larger scales than microlitres. The lack of suitable mixing strategies for large-scale peptide crystallisation, while guaranteeing high crystallisation throughput and reproducibility of the crystal qualities, has been identified as one of the main limitations that has to be overcome in order to apply crystallisation as a sustainable downstream purification technique

In this study, the importance of mixing on the controllability and reproducibility of the crystallisation process in large-scale batch crystallisers will be discussed. Instead of commonly used mixing strategies, such as over-head stirring, that usually leads to a more uncontrolled crystallisation with a poor crystallisation reproducibility, the impact of a gentler mixing strategy by using mechanical, oscillatory shaking will be reported. Furthermore, a newly developed image-analysis tool specific to peptide crystals will be presented. It is used to determine crystal sizes and distribution in-situ by processing images of crystals to evaluate the impact of mixing. A simple model peptide with a low solubility and molecular weight, which is known to crystallise, is used to investigate the importance of mechanical mixing on achieving a reproducible peptide crystallisation process. A tubular shaped batch crystalliser with volumes from the millilitres to the tens of millilitres along with different geometries was utilised. The crystallisation condition, such as pH, concentration, shaker speed, templating additives, and the orientation of the crystalliser to the shaking plane have been optimised. The impact on the crystallisation process is evaluated by means of measuring induction time, crystal yield, crystal sizes, and size distribution.

The induction time was determined from peptide concentration by UV-vis spectroscopy every 20-30 minutes and compared to optical microscopy images. These images were analysed using a novel automated evaluation-routine we developed to evaluate the crystallisation behaviour and crystal properties. This new computer-based image evaluation-routine enables the autonomous processing of hundreds of peptide crystal images. Thereby, commonly applied image processing steps, such as background noise filtering and image-gradient based edge determination for obtaining the crystal contours, have been utilised to determine the macro-scale crystal properties, such as crystal size distribution. This novel evaluation-routine is less expensive, able to calculate the macro-scale crystal properties within a couple of minutes, and only requires tens of microlitres of the crystal suspension instead of tens of milligrams of solid crystals, which is required in commonly used analytical techniques such as dynamic light scattering. Furthermore, it enables the evaluation of hundreds of crystals with sizes from the micrometre to the tens of centimetres scale. This makes the evaluation-routine suitable for studying macro-scale crystal properties for expensive and poorly soluble peptides in-situ, allowing to optimise crystallisation parameters during the crystallisation process. Comparison of crystal size distribution obtained with the new routine and manually obtained crystal sizes was used to optimise and identify problems within the image processing steps.

Monitoring the peptide’s concentration gradient in crystallisers with volumes of up to 15 millilitres found that an increase in oscillatory shaking speed from 50 to 150 rpm leads to a reproduceable higher homogeneity of the peptide concentration within the crystalliser resulting in a higher consistency of the crystallisation processes. Additionally, higher shaking speed lead to a faster decrease in peptide concentration implying a reduce in induction time and increase in crystal growth rate. These results can be supported by investigation of the crystal size distribution, obtained with the evaluation-routine, showing that a shaking speed of 150 rpm results in a more uniform crystal size distribution with a smaller span whereas a shaking speed of 50 rpm results in a wider spread size distribution. Furthermore, the reproducibility of controlling nucleation with templating additives, such as amino acids, can be enhanced significantly when a faster shaking speed is utilised, leading to a more homogeneous distribution of the additives. Mounting the tubular crystalliser in different orientations to the mixing plane found that if the axial direction of the crystalliser is aligned to the mixing plane, higher mixing length scales can be achieved resulting in an increase of mixing efficiency while keeping the energy input constant.

This study demonstrates the importance of mixing to achieve a high reproducibility of the crystallisation process in crystalliser volumes up to the tens of millilitres for peptides with low solubilities and molecular weights. Higher shaker speed enhances the mass transfer leading to a faster crystallisation process and a narrower crystal size distribution. To study the impact of mixing on the macro-scale crystal properties in-situ, hundreds of crystal images in combination with a novel developed evaluation-routine were evaluated.

Bibliography

1. Kastin, A.J., Handbook of Biologically Active Peptides. 2 ed. 2013: Academic Press. 2032.
2. Guillemot-Potelle, C., et al., Cost of Goods Modeling and Quality by Design for Developing Cost-Effective Processes. BioPharm International, 2010. 23(6): p. 26-35.
3. Yang, H., et al., Development and Workflow of a Continuous Protein Crystallization Process: A Case of Lysozyme. Crystal Growth & Design, 2019. 19(2): p. 983-991.
4. Roque, A.C.A., et al., Anything but Conventional Chromatography Approaches in Bioseparation.Biotechnology Journal, 2020.