Protein Crystallization

Protein crystallization is the act and method of creating structured, ordered lattices for complex macromolecules. Proteins are typically amorphous as solids and prone to denaturation, whereas proteins existing in crystalline lattices resist denaturation and possess greater stability. Proteins can crystallize if placed in suitable, hospitable environments. Protein crystallization is viewed as much as an art as a science, due to the highly complex interactions of variables that influence the protein crystallization process.

The most common reason for creating protein crystals is to support structural biology investigations, typically via x-ray diffraction crystallography. Investigators can visualize areas in the framework that permit interaction with other molecules by understanding the 3-dimensional structures of proteins (for example, to better understand how enzymes, substrates and ligands interact). Furthermore, crystallization is an effective means of creating pure proteins that are free of contamination from other proteins or extraneous biological matter, thus offering an alternative means of purification and separation to preparative chromatography. 

Proteins are increasingly important as therapeutic medicines. However, producing proteins with the necessary stability poses key challenges, including product variation, loss of structure that would reduce efficacy and the formation of potentially hazardous by-products. Protein crystallization is a method to produce pure, stable solid dosage forms and a number of injected and infused therapeutics that are associated with protein crystals. In the past, protein crystallization has been a lab activity to support x-ray crystallography, but due to the exponential growth of protein-based therapeutics, scaling-up protein crystallizations to production levels is increasingly important and an area of intensive focus.

Scale-up of protein crystallization is fundamentally analogous to small molecule crystallization in its requirements to optimize Critical Process Parameters (CPPs) at lab scale to achieve Critical Quality Attributes (CQAs) of a crystallized drug substance at commercial scales. A holistic platform for protein crystallization process development and scale-up typically encompasses:

• In-situ particle system characterization
• Spectroscopic monitoring and polymorph characterization
• Optimization of procedure and feedback control strategies through integrated modules and sensors

There are major pharmaceutical applications associated with protein crystallization: medicinal chemistry via structural biology, protein drug development, scale-up, separation and purification, formulation and drug delivery.

• Medicinal chemistry via structural biology. Advances in protein crystallization techniques and improved x-ray crystallography have led to deduction of tens of thousands of 3-dimensional protein structures. Understanding drug-receptor interactions in great detail is possible by determining the structure of these complex macromolecules and visualizing binding sites. This information enables greater understanding on a molecular level of protein inhibitors, allosteric activators, co-factors and even enzyme catalytic sites or 'active sites' – resulting in better drugs with improved efficacy. A classic example of the importance of understanding protein-inhibitor structure via crystallography is demonstrated by the great advances in the development of HIV proteases.
• Development, scale-up, separation and purification for protein-based therapeutics. In addition to growing single protein crystals for x-ray crystallography, the pharmaceutical industry has increased efforts in the development of protein crystal-based therapeutics. The goal of these applications is to develop and scale up to large quantities of pure, stable protein crystals which have the correct particle size and well-controlled particle size distribution. Relative to preparative chromatography, protein crystallization has the potential to be a cost-effective alternative means of separation and purification of macromolecules and for isolation of chiral molecules. The tools for scaling up protein crystallizations are analogous to those used for scaling up chemical reactions with the need for understanding CPPs and their effect on CQAs.
• Formulation and drug delivery. Protein therapeutics are administered as solutions and may be delivered via injection or infusion. Because high amounts of protein need to be delivered, the infused protein solutions are diluted to avoid viscosity and aggregation issues. This is less convenient for the patient and the hospital due to the time required to infuse the whole treatment, but may also contribute to the phenomenon lower patient compliance with any form of parenteral administration. Using the crystalline form of proteins in suspension is of great interest, since these formulations have less viscosity issues and can be delivered at higher concentrations. Another key point is that the size of protein crystals must be carefully controlled to ensure the suspension can easily pass though an injection or infusion needle. Also of importance, any residual crystals which do not dissolve during reconstitution must also not exceed critical particle sizes which could cause adverse side-effects due to capillary accumulation in vasculature, generally minimizing significant particle quantities >15 µm. 

1. Choose an appropriate solvent. Common considerations include how much solute can be dissolved (solubility), how practical the solvent is to handle (safety) and how the solvent will effect stability and free energy states of the peptide or protein.
2. Achieve supersaturation. Heat the soluble peptide/protein drug substance in solvent or buffer to a maximum allowed tolerance (usually limited by denaturation). The solution will become supersaturated.
3. Crystallize the product by controlling CPPs. Reduce solubility by cooling, anti-solvent addition, evaporation or reaction. As solubility is reduced, a point is reached where crystals will nucleate and then grow. Highly pure product crystals should grow, although motifs with high free energy may form alternative or non-crystal structures; preventing this requires careful CPP control if it is even practical.
4. Allow the system to reach equilibrium. This is done to maximize the yield of drug product solid.
5. Filter and dry the purified product.

There are several methods that are employed to grow protein crystals; among the most common are vapor diffusion and microbatch. The former method has two variations, described as hanging drop and sitting drop, and both fundamentally work the same way. In a sealed system, a drop of solution that has the protein, buffer and precipitant equilibrates with a reservoir containing higher solute concentrations of the same solution, but without the protein. Water from the drop containing the protein vaporizes and is attracted to the higher concentration solution in the reservoir. As the concentration of solutes in the drop increases, protein begins to crystallize. In the microbatch method, the proteins and other reagents are introduced and contained under a thin layer of mineral oil that eliminates diffusion of water. The microbatch is essentially a small reactor containing all the necessary ingredients at the correct, stable concentrations to allow crystallization to occur.

Because finding the correct set of reagents, parameters and conditions needed to crystallize a protein is often empirical, the most practical means for successful protein crystallization is to use a varying range of experimental conditions that are based on previous experience. The easiest way to do this is to use commercially available protein crystal screening kits that contain pre-packaged solutions with buffers, precipitants, etc. contained in multi-well plates. The hanging drop vapor diffusion method is applied to grow crystals with these kits. When large-scale screening is required, there are high throughput systems available that automate the microbatch method, saving time and tedium while concurrently minimizing the amount of purified protein required. For large scale protein crystallizations required for pharmaceutical or industrial applications, the information obtained in screening experiments form the basis for application in larger scale automated reactor systems characterized by in-situ, real-time PAT. 

In support of process development and manufacturing, one of the most important initial steps in successfully crystallizing proteins is to ensure that an adequate supply of highly purified and well characterized protein is available. In addition to the protein being pure (i.e. free of contaminants, such as impurities from fermentation matrices), the protein should be homogeneous and stable (or stabilized) in solution with:

• Minimal aggregation (Figure A)
• Correct secondary, tertiary and quaternary structure (Figure B)
• Any conjugate or recombinant modifications (Figure C)

Additionally, all reagents and solvents used in the protein crystallization process need to be free of contaminants, as these can have significant effect on nucleation.

Similar to small molecule crystallization, proteins crystallize from supersaturated solutions. For small molecules that form robust crystal lattices, manipulating temperature is a primary means to achieve supersaturation; this is also useful for proteins. However, a reduced temperature range is permissible since protein crystals typically have solvent as part of the crystal structure, are frequently temperature sensitive, and prone to denaturing and aggregation.

Small molecules have significantly fewer degrees of stereoscopic freedom and crystallize much easier as a result. Large molecule peptides have many more internal acidic, basic, polar, non-polar interactions and repulsive forces that resist ordered crystallization, a phenomenon that is exacerbated for even larger proteins. Most crystallization-compatible peptides also contain significant numbers of disulfide bonds which are crucial to immobilizing structural motifs. The extent of intramolecular disorder or possible variation in structure is an important attribute determining whether a peptide or protein could ever crystallize rather than precipitate. Even if the molecule is predisposed with attributes which are amenable to crystallization, there are a number of influencing process parameters that effect process commercial viability and need to be characterized.

Beyond temperature, a number of other factors that affect/aid protein crystallization via supersaturation including pH adjustment and the ionic strength of a solution, adding salts and precipitants such as PEG to help reduce protein solubility, type of ligand present, etc. The key is to supersaturate the solution within an environment that does not disturb the normal character of the protein. This implies the importance of finding a set of variables and conditions that result in crystals with the correct morphology, purity and acceptable growth rates.

Protein crystallization proceeds through a series of interdependent mechanisms:

1. Nucleation
2. Growth
3. Oiling Out in Crystallization
4. Agglomeration
5. Breakage
6. Seeding
7. Polymorphism Chemistry

These mechanisms, which are often hidden form scientists, play a dominant role in defining the outcome of a crystallization process.

Development and scale-up of crystallizations for protein-based therapeutics requires an in-depth understanding of the variables and conditions that affect the crystallization process. As in other crystallizations, achieving desired crystal form, shape, particle size, and size distribution requires careful control to achieve consistent product quality. Any variation in the CQAs of protein crystals suspensions will directly affect therapeutic efficacy. Because protein crystals are highly sensitive to their growth environment and can be easily damaged by excessive temperature, loss of solvent and other external stresses or even the process of sampling, it is desirable to measure CPPs and product quality in situ, i.e., without removal of the crystals from their solution environment.

• EasyMax and Optimax automated chemical reactors enable precise control, adjustment and recording of parameters, including temperature, pH, conductivity, agitation rate and dosing of buffers and excipients, which are critically important for successful development and scale-up of protein crystallizations
• In-situ analytical technology tools that aid in monitoring, troubleshooting and correcting process deviations are also useful
  • In-situ Raman spectroscopy and FTIR spectroscopy track reagent concentrations, monitor crystal structure growth and also measure concentration of supersaturated solutions
  • ParticleTrack (with FBRM) technology monitors the changing crystal size and size distribution in real time as key segments of the crystallization process proceed
  • EasyViewer enables visualization and image analysis of the crystallization process as it progresses

When automated lab reactors are used in association with real-time in-situ analyzers, the impact of process parameters on crystallization mechanisms such as aggregation, growth, agglomeration, breakage and shape change can be identified. Precise control over the crystallization process in association with continuous, in-situ monitoring of the crystal formation progress, enables optimization and scale-up to be fact-based and data-driven.

The EasyMax automated lab reactor is a complete workstation for parameter optimization in protein crystallization to support development and scale-up design and execution. This fully automated system enables precise control over process parameters.

• Manage CPPs including dosing rates, temperature, agitation, pH and conductivity
• Readily integrates with in-situ analytics for crystallization including ReactIR (FTIR), ReactRaman (Raman), ParticleTrack (FBRM), and EasyViewer (in-situ image analysis)
• iC software enables automated execution and control of all experiments and peripheral equipment as well as full data capture and record keeping

ReactIR uses mid-infrared spectroscopy to measure solution phase in crystallization studies. ReactIR measurements are not impacted by suspended solids, bubbles or other particulate mass.

• Track individual crystallization solutes in situ and in real time.
• Measure solution concentrations in support of supersaturation studies
• Provide actionable data analytics for real-time decision making and control

ReactRaman uses Raman spectroscopy to provide molecular analysis of particles in solution

• Track crystal growth, structure and morphology non-destructively, in-situ and in real time
• Determine parameter effects and optimize crystallization conditions and variables to achieve quality objectives

ParticleTrack uses Focused Beam Reflectance Measurement (FBRM) technology and is an optical, probe-based instrument that is inserted directly into a vessel to track changing particle size and count in real time. ParticleTrack uses Chord Length Distribution measurements to obtain Crystal Size Distribution (CSD).

• Protein crystals are monitored continuously, as experimental conditions are varied
• Track batch or continuous growth during crystallization in real time with no sample removal required
• Characterize conditions which affect nucleation, aggregation, polymorphism and other system attributes

EasyViewer is a probe-based instrument that is inserted directly into a vessel to capture images of particle systems. EasyViewer performs image analysis and measures particle count, size distribution and morphology from time-resolved, inline images of the process allowing populations to be trended over time.

• Protein crystals are monitored continuously, as experimental conditions are varied
• Track batch or continuous growth during crystallization in real time with no sample removal required
• Visualize crystallization parameters that affect nucleation, aggregation, polymorphism and other system attributes in real time

Pandit, A., Katkar, V., Ranade, V., & Bhambure, R. (2018). Real-Time Monitoring of Biopharmaceutical Crystallization: Chord Length Distribution to Crystal Size Distribution for Lysozyme, rHu Insulin, and Vitamin B12. Industrial & Engineering Chemistry Research, 58(18), 7607–7619.

The authors reported success in developing an efficient approach to real-time monitoring of biopharmaceutical crystallizations. This procedure uses ParticleTrack to measure the CLD for a series of biopharmaceutical molecules under both static and dynamic conditions, and then apply mathematical modeling to convert the CLD to CSD. The biomolecules that were investigated have different crystal forms, which are dependent on the experimental conditions of the crystallization. Lysozyme can produce tetragonal and needle shaped crystals, rHu insulin forms rhombohedral crystals and B12 forms polyhedral crystals.

The authors developed models to determine CSD for the low aspect ratio tetragonal, rhombohedral, polyhedral crystals and high aspect ratio needle crystals. The authors used three different models for determining the crystal size distribution:

• A model developed by another research group that assumed that crystals could be represented as spheres (spherical equivalent diameter, SED) was effective for the low aspect ratio crystals
• For high aspect ratio crystals, the authors in this study developed a model that is width based
• Lastly, they developed a model that is length-based model and required the aspect ratio to be specified

The results from the models developed for calculating CSD from the FBRM measurements were then compared to CSD measurements obtained by image analysis.

Attempting to monitor an opaque, crystal slurry presents challenges, especially when manufacturing specifications call for a narrow particle size distribution (PSD). For injectable protein crystal formulations, the PSD can greatly influence viscosity as well as serving as an early warning for aggregation or other degradation events that can potentially compromise the formulation. In-situ monitoring was used in a scale-up batch crystallization study to characterize human growth hormone (hGH) crystal formation and growth. Real-time analysis detected a deviation from the usual PSD profile caused by differences in mixing between the vessels. This was confirmed by subsequent offline Image Analysis (IA).

• Orehek, J., Teslić, D., & Likozar, B. (2020). Continuous Crystallization Processes in Pharmaceutical Manufacturing: A Review. Organic Process Research & Development, 25(1), 16–42. https://doi.org/10.1021/acs.oprd.0c00398
• Unno, J., & Hirasawa, I. (2020). Parameter Estimation of the Stochastic Primary Nucleation Kinetics by Stochastic Integrals Using Focused-Beam Reflectance Measurements. Crystals, 10(5), 380. https://doi.org/10.3390/cryst10050380
• Pandit, A., Katkar, V., Ranade, V., & Bhambure, R. (2018). Real-Time Monitoring of Biopharmaceutical Crystallization: Chord Length Distribution to Crystal Size Distribution for Lysozyme, rHu Insulin, and Vitamin B12. Industrial & Engineering Chemistry Research, 58(18), 7607–7619. https://doi.org/10.1021/acs.iecr.8b04613
• Wang, Z., Liu, T., Lu, C., & Dang, L. (2018). Manipulating crystallization in lysozyme and supramolecular self-arrangement in solution using ionic liquids. CrystEngComm, 20(16), 2284–2291. https://doi.org/10.1039/c8ce00107c
• Kwon, J. S. I., Nayhouse, M., Christofides, P. D., & Orkoulas, G. (2014). Modeling and control of crystal shape in continuous protein crystallization. Chemical Engineering Science, 107, 47–57. https://doi.org/10.1016/j.ces.2013.12.005
• Kwon, J. S. I., Nayhouse, M., Orkoulas, G., & Christofides, P. D. (2014). Enhancing the Crystal Production Rate and Reducing Polydispersity in Continuous Protein Crystallization. Industrial & Engineering Chemistry Research, 53(40), 15538–15548. https://doi.org/10.1021/ie5008163
• Basu, S. K., Govardhan, C. P., Jung, C. W., & Margolin, A. L. (2004). Protein crystals for the delivery of biopharmaceuticals. Expert Opinion on Biological Therapy, 4(3), 301–317. https://doi.org/10.1517/14712598.4.3.301