Protein Crystallization in Microfluidics - Alex Brosseau, Cory Thomas, and Adam Selsman
Introduction
Protein crystallization is the process of forming protein crystals. Screening for crystallization conditions is an art. There is no way to predict ahead of time the conditions that produce crystals.3 Since each protein structure is very unique the best combination of protein concentration, pH, temperature, salt concentration, and numerous other variables, varies with each protein. For example, some proteins, like many membrane proteins, do not crystallize well due to their irregular structure. Membrane proteins barely interact with one another, so an unstable crystal lattice is typically formed unless it is crystallized at highly specific conditions unique to each membrane protein. The history of protein crystallization dates back to over 150 years ago. In 1840, the first observations of protein crystallization were recorded, using evaporated worm blood to form hemoglobin crystals.3 With that discovery, protein crystallization became a popular purification method for proteins. Many scientists utilized solely protein precipitation methods, such as salting out, to separate proteins from solution rather than forming crystals. Ammonium sulfate and sodium chloride became popular salts for the protein separation technique of salting out, but other ways of precipitating protein from solution were used as well, such as temperature and pH variation. The use of these precipitation methods were used commonly during that time but were often harsh on the proteins and only the most stable ones survived the treatment. Later on, precipitating reagents like polyethylene glycol (PEG) (shown in Figure 1) became common in protein precipitation because it was found to have little effect on the structure of the protein. However, in 1935 Bernal and Crowfoot created a hydrated protein crystal and observed that the crystal gave a diffraction pattern. This led to the first successful use of protein crystallization to determine the molecular structure of myoglobin in 1958.13 With the advent of microfluidics in 1990's, protein crystallization and precipitation became a common topic to study in microfluidic devices. Originally protein crystallization and precipitation were performed in large batches, but as time went along, liquid handling robots were developed allowing for crystallization in well plates to become common. Both the development of microfluidics and liquid handling robots allowed for crystallization screening at many conditions, without using a lot of protein.
Theory of Protein Crystallization
Phase Diagrams
Protein crystallization largely depends on a variety of factors such as protein structure, protein concentration, precipitating reagent concentration (like PEG), salt concentration, temperature, pH, size of the solution (affects kinetics of crystallization), method of protein crystallization (microbatch, vapour diffusion) and so on. In order to crystallize a protein one must achieve a supersaturated state so that the protein molecules attract to other protein molecules rather than dissolve in the solution. There are two types of protein solids that can form from a supersaturated state: crystals (Figure 2) or precipitates (Figure 3). Similar to how solid water forms, the nature of the protein solid depends on the thermodynamics and the kinetics of solution. Thus the size of the solution can have an effect on crystallization of a protein3,9.
Typically when scientists are looking for crystal hits for a particular protein, they vary only the concentration of protein and one other parameter. A phase diagram for a protein, like in Figure 2, can be developed by varying one parameter and protein concentration independently. The phase diagram consists of 4 distinct regions: undersaturated region, metastable region, labile (nucleation) region, and precipitation region. In the undersaturated region, the adjustable parameter or protein concentration is too low to induce any protein solid formation. The metastable region is a region in which crystals can exist and grow but cannot form. Even when only two variables are varied, like in the phase diagram in Figure 2, there exists many conditions where the protein will be undersaturated. This means there is a great need for microfluidic devices that can screen at many conditions using very little protein. Many scientists put protein solution in the metastable region then add preformed microcrystals of protein to further grow the crystal. The labile, or nucleation, region is where nuclei can form that eventually lead to crystals. When solution is in the nucleation region, but too close to the precipitation region, high amounts of small, low quality crystals form. As a result many scientists limit the amount of nucleation events by shifting solution into the metastable region, where nucleation events cannot occur, by adding protein free buffer to solution.3,9
Over the years many proteins have been identified as model crystallization proteins. These proteins are often used to study crystallization or other crystallization related subjects. These model proteins include hen egg white lysozyme, hemoglobin, xylanase, and many others. Some things these proteins all have in common is that they are relatively small so they form distinct geometric crystals, are globular proteins, and are easily purified from biological sources in massive amounts (one hen egg white yields about 6 g of lysozyme). Another thing to note is that the crystals of a protein are unlike the crystals of molecules, like sodium chloride. A protein crystal is not rigid in structure rather it has a jelly like consistency since it actually contains some water, even though water exclusion is the main driving force behind protein crystal formation.
Common precipitation reagents
Since a lot proteins are unstable outside standard pH (pH 7) and room temperature (27 degrees Celsius), proteins cannot be crystallized by extreme pH variance or temperature variance like other smaller molecules. Thus proteins are largely crystallized by the addition of a precipitating reagent like salt or polyethylene glycol. However, some proteins can be precipitated out of solution by methods such as isoelectric focusing, which involves changing the pH of solution until it is near its pI (isoelectric point). The isoelectric point is the pH at which the amino acid residues of the protein possess no net charge. Since the protein at pI possesses no net charge, it is more likely to interact with other protein molecules via hydrophobic interactions rather than hydrophilic interactions with water. Bound ligands and other chemical modifications, like methylation, acetylation, and phosphorylation, have an effect on the proteins crystallization since they can shift conformation of the oligomer or other levels of structure of proteins.
Polyethylene glycol
The mechanism by which polyethylene glycol leads to formation of protein solids is not fully understood. The current theory suggests that since PEG has little effect on the protein stability, it must work by excluding water from the protein solvation layer (layer of water that surrounds a protein).1 The hydrophilic attraction of PEG pulls water away from this layer leading to protein aggregation. PEG is commonly used in industry to precipitate proteins out of solution. Various molecular weights of PEG have been made because it has been found that higher molecular weights of PEG lead to increased protein precipitation at lower concentrations of the precipitation reagent. This indicates PEG may also work by size exclusion to precipitate protein.1
Salts
Salts are very commonly used in industry to precipitate proteins out of solution. Salting out is the mechanism by which salts cause proteins to come out of solution (salting in also exists and leads to protein go into solution). Sodium chloride and ammonium sulfate are the most common of the salts for protein precipitation. Ammonium sulfate can be rather harsh in its precipitation so it is reserved for the most stable proteins. High amounts of salt shield the proteins charged residues which prevent it from interacting with water, much like isoelectric focusing. Often salts are combined with PEG or other reagents to precipitate a protein out of solution. Particularly, protein solutions at a pH far away from pI of a protein require salt to reduce the protein's charge. Studies have been done that determined the most favorable anions and cations for salting out. The order of the anion and cations are known as the Hofmeister series (shown by Figure 5). Certain salts, like sodium bromide, have to be avoided since they lead to denaturing of the protein.
Methods of Protein Crystallization
There are many setups to make protein crystals. Microbatch and vapor diffusion are discussed in this section.
Batch Crystallization
Batch crystallization is very simple in concept. A solution of the protein, precipitating reagent, salt, or other additives are mixed together to form a cocktail. A 100% silicon oil or paraffin oil is placed on top of the solution to prevent evaporation. Depending on the kinetics of the crystallization the crystals can form in matter of 15 minutes, or days. Since oil is being added, the oil and solution interface can have an effect on the crystallization, so the use of oil can change whether a crystallization hit is achieved.
Vapor Diffusion Crystallization
Vapor diffusion crystallization consists of hanging drop and sitting drop methods (Figure 6). In both methods, a drop of protein solution is put onto a surface. As the names imply the drop is either hanging or sitting on the surface. The drop and surface is put into an air tight container with a reservoir solution, typically water. The solution and the drop reach a dynamic equilibrium of vapor pressure so that the drop dehydrates. This technique leads to higher levels of supersaturation than batch crystallization but can also lead to the formation of salt crystals, if salt is used.3
Microdialysis Crystallization
In microdialysis, a protein solution is put in a container. A lid with a semi-permeable dialysis membrane that does not allow protein to escape is put on top. The container is put in a reservoir solution that is comprised of precipitant solution. The precipitant molecules are small enough to diffuse through the membrane to the protein solution, raising the protein solution to multiple supersaturation levels. This method is not commonly done for protein crystallization screening.3
Free Interface and Counter Diffusion Crystallization
In both free interface and counter diffusion crystallization, the protein and precipitant solution are brought into contact in such a manner that no convective mixing occurs but only diffusion (in a wave like manner). Free interface crystallization involves formation of one type of precipitate. Counter diffusion, on the other hand, has multiple precipitation events at different points in the phase diagram. Both methods rely upon the fact proteins do not diffuse quickly (due to their size), so they stay relatively in the same place the precipitant solution diffuses to the protein. These techniques are often done by utilizing a small capillary tube to allow for concentration gradients. Other experiments have been developed using zero gravity conditions. Free interface and counter diffusion crystallization are not as common as microbatch or vapor diffusion, but microfluidic devices for protein crystallization are well suited for them due to diffusion being main driving force of mixing on microscale.3
Techniques to Analyze Protein Crystallization
Quantifying protein precipitation and crystallization can be done by using methods such as UV-Vis (light scattering) or differential scanning fluorimetry (DSF). UV-Vis techniques involve getting an optical density (OD) measurement or turbidity reading for the solution, by having a device scatter light off solution at a wavelength of 562 or 600 nm.3 One could do this measurement for various precipitating reagent concentrations and see what critical concentration leads to a spike in turbidity. The concentrations below this turbidity spike often are best for protein crystallization. DSF is important if one believes their protein solution is causing the protein to denature. In this technique a hydrophobic sensitive dye is put in the protein solution, so that it binds in the core of the protein. The protein is then heating until it unfolds (or melts). Since the dye is fluorescent, as the protein unfolds the fluorescence can be seen and melting temperature can be recorded. A decrease in melting temperature between the protein in the precipitating solution and the protein by itself indicates that the precipitating reagent may affect protein stability. Visual inspection can often be a way to tell if precipitation is occurring as well. Many reagents, like PEG, cause phase separation when mixed with the protein solution (shown by Figure 7). Phase separation is an indication of protein precipitation. Formation of a solid precipitate or a cloudiness to the solution also indicates precipitation. Often the solid precipitate will lie at bottom of the solution and supernatant will become clearer when precipitation occurs.3
Microfluidics and Protein Crystallization
Microfluidic designs for protein crystallization is done often in microfluidics. The nanoliter scale of microfluidics allows little amounts of protein to be tested in a combinatoric fashion for crystallization hits. Microfluidic devices also allow for a lot of crystallization screens to be done on a single chip, simultaneously (high thoroughput screening). However, due to the small scale, a crystallization hit on a nanoliter scale may give a false positive result. This means that when crystallization is scaled up to yield high amount of crystals under the same conditions as the microfluidic device, it will not necessarily yield crystals on the larger scale. This can be avoided by making the scale of microfluidic device large enough to eliminate false positive results. Many designs have been made to crystallize proteins but a lot of them have problems that need to be overcome. One such design was done by Quake et al.(2002) that involved an elastomer chip. In the device crystallization agents are pipetted into 48 wells and an aluminum carrier with a top and bottom plate holds the device. The carrier then uses pneumatic pressure on the wells to pressurize the fluid, forcing them into the chip.4 Protein solution is then added to each well in the device.4 In order to observe the crystallization, windows are created in the device. With this type of device, 6,048 crystallization experiments can be conducted at once using 150 microliters of protein solution. The main problem with this device is that it requires high pressures up to 7 psi, as well as a custom designed aluminum carrier device. However, using a device such as the one created by Quake et al. allows for protein crystallization screening that can not be done on a larger scale very successfully.
Quick Protein Solubility Chip
Progress in Academia to Date
A centrifugal PDMS-glass derived microfluidic device has been developed by Gang Li et al. for high output protein crystallization. The circular device has 24 wells for precipitants around the outside of the device and a protein inlet in the middle, as seen in Figure 9. The precipitant and protein solutions are metered by the capillary effect of the glass through the channels, and the centrifugal force allows for the mixing of precipitant and protein in the reaction well.12
The device has been experimentally proven and provides an economical way to test different solution's effect on a desired protein. To use the device, a lab would need a centrifugal device to effectively spin the device and a X-ray diffraction machine for analysis.12
Project Progress
The goal of the project is to develop an easy to use microfluidic device that can test for protein precipitation (not crystallization) with 4 precrystallization test solutions (PCT) from Hampton Research.5 The motivation for why this device should be made is that there seems to be no design that can easily test for whether protein precipitates at various conditions. This quick solubility protein test can be used by anyone and would allow for testing of many precipitating solutions in a short amount of time. Although this device will not be designed for testing for crystallization it will be designed for protein precipitation, which gives scientists a lead on what optimal conditions could be close to those necessary for crystallization. The current design has a total of 5 wells: 4 for the PCT solutions and 1 for protein solution. The PCT and protein solutions are driven out from their wells by centrifugal force due to rotation of the chip.10,11 The wells are placed so that fluid flow is in the direction of the centrifugal force. Metering of PCT and protein solutions are done by varying rotation speeds and the distances that the wells are from the center mixing well, as well as using capillary valves. The center well will be mixed by shaking the device. There are some challenges associated with the device design. PDMS, the material being used for the device design, is inherently hydrophobic which prevents PCT solutions from flowing in the channels. The use of a glass base on the device allows for solution to flow, due to glass being hydrophilic. Determining the right dimensions for the capillary valves is also a challenge. The constriction of the valve needs to be small enough to prevent flow of solution through, but large enough to allow for centrifugal force of rotation to push the liquid through. This issue can be described using the Capillary Number, which is a dimensionless number that refers to the ratio of viscous forces to surface tension and described by the equation, [math]\displaystyle{ Ca=\mu V/ \sigma }[/math]. In order for successful use of the capillary valves, when standing alone, the surface tension must dominate. This means that the liquid will not be able to overcome the surface tension and move through the valve. This surface tension is created due to the capillary pressure created by the valve, which keeps the fluid from flowing through the valve without external force. In order to utilize the capillary valves, centrifugal force is used to overcome the capillary pressure and break the surface tension, allowing for the solution to flow through the valves. A precisely calculated diameter for the tube is key to successfully utilizing these capillary valves.
The 4 PCT solutions provided by Hampton research are made of 0.1 M Tris hydrochloride (pH 8.5), 2.0 M Ammonium sulfate; 0.1 M Tris hydrochloride (pH 8.5), 0.2 M Magnesium chloride hexahydrate, 30% Polyethylene glycol 4,000; 0.1 M Tris hydrochloride (pH 8.5), 1.0 M Ammonium sulfate; and 0.1 M Tris hydrochloride (pH 8.5), 0.2 M Magnesium chloride hexahydrate, 15% Polyethylene glycol 4,000.5 The PCT solutions are designed to evaluate whether a certain protein concentration leads to undersaturation or precipitate formation. The formation of precipitate in at least half of reagents provides a lead for protein concentration that is suitable for crystallization. These PCT solutions are designed for efficient screening which makes them well suited for a microfluidic screening device. Challenges such as non uniform mixing are inherent to larger scale crystallization. Also, larger scale designs do not easily allow for screening at many supersaturation points on the phase diagram without utilizing immense amounts of protein and precipitant materials. Using a microfluidic device would allow for consistent uniform mixing of precipitant solution and protein. Overall, a quick protein solubility chip would easily allow some to screen for protein crystallization in an easy manner while also utilizing small amounts of protein.
References
1-Sim, S.-L.; He, T.; Tscheliessnig, A.; Mueller, M.; Tan, R. B.; Jungbauer, A. Protein precipitation by polyethylene glycol: A generalized model based on hydrodynamic radius. Journal of Biotechnology 2012, 157 (2), 315–319 DOI: 10.1016/j.jbiotec.2011.09.028.
2-Paterová, J.; Rembert, K. B.; Heyda, J.; Kurra, Y.; Okur, H. I.; Liu, W. R.; Hilty, C.; Cremer, P. S.; Jungwirth, P. Reversal of the Hofmeister Series: Specific Ion Effects on Peptides. The Journal of Physical Chemistry B 2013, 117 (27), 8150–8158 DOI: 10.1021/jp405683s.
3-Luft, J. R.; Newman, J.; Snell, E. H. Crystallization screening: the influence of history on current practice. Acta Crystallographica Section F Structural Biology Communications 2014, 70 (7), 835–853 DOI: 10.1107/s2053230x1401262x.
4-Hansen, C. L.; Skordalakes, E.; Berger, J. M.; Quake, S. R. A robust and scalable microfluidic metering method that allows protein crystal growth by free interface diffusion. Proceedings of the National Academy of Sciences 2002, 99 (26), 16531–16536 DOI: 10.1073/pnas.262485199.
5-Hampton Research http://www.hamptonresearch.com/ (accessed Apr 7, 2017).
6-NF Monographs: Polyethylene Glycol http://www.pharmacopeia.cn/v29240/usp29nf24s0_m66430.html (accessed Apr 7, 2017).
7-Darryl, E. Hofmeister Series http://www.chemistry.sjsu.edu/deggers/.
8-Adachi, H.; Takano, K.; Morikawa, M.; Kanaya, S.; Yoshimura, M.; Mori, Y.; Sasaki, T. Application of a two-liquid system to sitting-drop vapour-diffusion protein crystallization http://journals.iucr.org/d/issues/2003/01/00/ts0154/ts0154fig1.html
9-Chayen, N. E.; Saridakis, E. Protein crystallization: from purified protein to diffraction-quality crystal. Nature Methods 2008, 5 (2), 147–153 DOI: 10.1038/nmeth.f.203.
10-Zoval, J.; Jia, G.; Kido, H.; Kim, J.; Kim, N.; Madou, M. Centrifuge-Based Fluidic Platforms. Springer Handbook of Nanotechnology 2007, 549–570 DOI: 10.1007/978-3-540-29857-1_20.
11-Schwemmer, F.; Blanchet, C. E.; Spilotros, A.; Kosse, D.; Zehnle, S.; Mertens, H. D. T.; Graewert, M. A.; Rössle, M.; Paust, N.; Svergun, D. I.; et al. LabDisk for SAXS: a centrifugal microfluidic sample preparation platform for small-angle X-ray scattering. Lab Chip 2016, 16 (7), 1161–1170 DOI: 10.1039/c5lc01580d.
12-Li, G.; Chen, Q.; Li, J.; Hu, X.; Zhao, J. Analytical Chemistry 2010, 82 (11), 4362–4369.
13-Kendrew, J. C.; Bodo, G.; Dintzis, H. M.; Parrish, R. G.; Wyckoff, H.; Phillips, D. C. A Three-Dimensional Model of the Myoglobin Molecule Obtained by X-Ray Analysis. Nature 1958, 181 (4610), 662–666 DOI: 10.1038/181662a0.
