- Email: [email protected]
Preparation of filamentous proteins for electron microscopy visualization and reconstruction
CHAPTER ELEVEN
Preparation of filamentous proteins for electron microscopy visualization and reconstruction Mariusz Matyszewski, Jungsan Sohn* Department of Biophysics and Biophysical Chemistry, Johns Hopkins University School of Medicine, Baltimore, MD, United States *Corresponding author: e-mail address: [email protected]
Contents 1. Introduction 2. Construct design 2.1 Controlling oligomerization 2.2 Filament separation (bundling) 2.3 Buffer components 3. Negative stain electron microscopy 3.1 Staining protocol 4. Cryo-EM sample preparation 4.1 Grid type 4.2 Sample concentration and distribution 4.3 Other parameters and tips References
168 168 169 169 170 171 171 173 173 174 175 175
Abstract Cryo electron microscopy (cryo-EM) has become a mainstream tool for determining the structures of macromolecular complexes at the atomic resolution. It has many advantages over other techniques such as X-ray crystallography and nuclear magnetic resonance (NMR). However, it also entails several challenges, a major one being preparation of an ideal sample. Recent studies have identified that DNA sensors and inflammasomes often assemble into filamentous oligomers, which poses a unique set of challenges in preparing ideal samples for high-resolution reconstruction using cryo-EM. This chapter will discuss how to overcome several major issues in cryo-EM sample preparation including construct design, screening using negative stain (ns) EM, and tips on working with filamentous proteins.
Methods in Enzymology, Volume 625 ISSN 0076-6879 https://doi.org/10.1016/bs.mie.2019.06.007
#
2019 Elsevier Inc. All rights reserved.
167
168
Mariusz Matyszewski and Jungsan Sohn
1. Introduction Thanks to the “resolution revolution,” the use of cryo-EM in structural biology has skyrocketed (K€ uhlbrandt, 2014). Currently, there are almost 3000 high-resolution cryo-EM structures deposited in the PDB, 700 of which were deposited in 2018 alone (www.rcsb.org). While the majority of the structures were solved using the Single Particle Reconstruction (SPR) approach, there are many higher order protein complexes whose structures were determined via helical reconstructions. This is noteworthy as several DNA sensors and inflammasomes (at least their signaling ends) assemble into helical filaments (Lu et al., 2014; Lu et al., 2016; Li et al., 2018; Matyszewski et al., 2018; Morrone et al., 2015). The general principles of working with helical assemblies are similar to those of SPR projects (see elsewhere in this volume), but there are some key differences. By definition, helical samples have helical symmetry which can be used to improve final resolution, often requiring a smaller data set to achieve comparable results. Additionally, some SPR problems like preferential orientation do not apply to filaments as each filament contains all the views necessary for reconstruction (Egelman, 2007). Nevertheless, getting filamentous samples to the collection-ready stage can still be as challenging as in any SPR project. Here we share techniques and tips for preparing an ideal sample for cryo-EM reconstruction of filamentous proteins.
2. Construct design The first step is to design a construct suitable for cryo-EM. In many instances of filamentous complexes found in inflammasomes and DNA sensors, only one domain of the protein is responsible for filamentation (Morrone et al., 2015; Shen et al., 2019; Tenthorey et al., 2017). Isolating such a domain can significantly simplify the challenge of getting a highresolution structure of the minimal filament. On the other hand, working with the full-length protein can provide additional insight into the workings of the filament, but comes at a cost of a much more difficult system to work with (i.e., expression and purification). Moreover, the rest of the protein usually has regulatory roles or act as a scaffold for filament assembly, often resulting in disordered arrays surrounding the ordered filament core (Shen et al., 2019). Nevertheless, we found that even working with isolated polymerization domains poses many challenges.
Preparation of filamentous proteins for electron microscopy
169
2.1 Controlling oligomerization One of the biggest problems working with filamentous protein is controlling their oligomeric state. While the final goal is to capture the protein as a filament, it is often advantageous to prepare it as a monomer and to later induce polymerization in a controlled manner. Monomeric constructs tend to be significantly more soluble than their filamented counterparts, allowing higher yields and better concentration control. The most straightforward solution for the oligomerization problem is to use cleavable solubility tags. In many cases, large protein tags such as Maltose Binding Protein (MBP) can sterically interfere with filament assembly ( Jin, Perry, Smith, Jiang, & Xiao, 2013), greatly reducing the intrinsic oligomerization activity. Combined with a protease cut site, such as Tabacco Etch Virus (TEV) protease, this not only prevents oligomerization, but also allows for temporal control over filamentation (Morrone et al., 2015). For the vast majority of inflammasome signaling components, the polymerization domain is located at the N-terminus (e.g., PYrin Domain, PYD or CAspase-Recruiting Domain, CARD). For best prospect of inhibiting polymerization, the tag should be placed at the N-terminus of either PYD or CARD with minimal linker (Fig. 1A; MBP-CARD/PYD construct).
2.2 Filament separation (bundling) In many cases, despite controlling their polymerization using the methods described above, resulting filaments aggregate and “bundle up” (Fig. 1B). If simply lowering the concentration does not solve this issue, the next best option is to introduce another tag which would repulse other filaments, while still allowing polymerization. The identity of such repulsion tags can vary, but enhanced Green Fluorescence Protein (eGFP) has been used with great success (Li et al., 2018; Matyszewski et al., 2018). However, care should be taken not to affect the native form of the filament (Lu et al., 2015). To avoid this issue, the tag should ideally be placed as to mimic the rest of the native protein. If working with the full-length construct, the tag should be placed at the opposite terminus from the solubility tag (Fig. 1A; MBPCARD/PYD-eGFP construct). The end result should be a double tagged protein, with filaments only forming after solubility tag is cleaved (Figs. 1B-b and c). Note: the repulsion-tags should not be visible in the final reconstruction, as they should be flexibly attached to the filament core without affecting its
170
Mariusz Matyszewski and Jungsan Sohn
A
CARD/PYD MBP
Controlled polymerization
TEVp cleavage site MBP
Controlled polymerization/
eGFP reduced “bundling”
B
a.
b.
c.
Fig. 1 Construct variants for filamentous protein. (A) Typical inflammasome receptors contain either PYD or CARD at the N-terminus, which is capable of auto-polymerization. Adding MPB with TEV protease cut-site to the N-terminus of CARD/PYD allows controlled polymerization. Adding eGFP at the C-terminus minimizes bundling. (B) nsEM micrographs of (a) auto-assembled NLRC4CARD, (b) MBP-CARD of NRLC4 (NLRC4CARD)-eGFP, and (c) NLRC4CARD-eGFP after removing MBP via TEVp.
native architecture. Otherwise, the presence of the tag in the final reconstruction could suggest that the native symmetry has been tempered with, especially for truncated constructs (Lu et al., 2015). An even less intrusive way of preventing bundling is temporal control. Since bundling requires pre-formed filaments, decreasing the incubation time before freezing or staining samples might reduce aggregation. However, the efficacy of this approach will vary by samples and is usually best when combined with other approaches.
2.3 Buffer components While most buffer components do not affect the quality of the sample in cryo-EM, there are a few things to avoid, or be cautious about, in the final sample buffer. 1. Salt: High salt concentrations have been shown to reduce contrast, and thus Signal-to-Noise Ratio (SNR) of cryo-EM micrographs. Typically, concentrations below 300 mM should be used (Drulyte et al., 2018).
Preparation of filamentous proteins for electron microscopy
171
2. Glycerol: Reduces the SNR and should be kept at or below 5% concentration (Drulyte et al., 2018) for cryo-EM data collection. 3. Detergent: While detergents above critical micelle concentration will add to the background, use of detergent can also be beneficial in changing particle distribution, especially when it comes to air-water interface interactions. Since it can affect how the sample freezes, only introduce it if the particle distribution problem occurs. It will often require screening many detergents before the best one is found. 4. Phosphate: Interferes with nsEM. Uranyl will react with phosphate to form crystals, making it impossible to screen with phosphate buffers present. It can be safely reintroduced if only doing cryo-EM.
3. Negative stain electron microscopy Imaging the purified protein sample by Negative Stain Electron Microscopy (nsEM) is a critical step even if the final goal is a cryo-EM reconstruction. While nsEM is unlikely to provide anything more than a symmetry estimate, it can provide plenty of qualitative information on how the sample filaments behave. The most crucial piece of information is if the protein actually polymerizes into a filament vs. disordered aggregates (this cannot be distinguished by size-exclusion chromatography). This is especially critical when testing new constructs and buffer conditions, as nsEM is a much higher throughput method than cryo-EM. However, one has to keep in mind that conditions might not transfer to cryo-EM; freezing is much different than staining a sample. Once filamentation has been confirmed, one could potentially try to reconstruct the nsEM filaments. Usually it is at least worth to check for symmetry by collecting a power spectrum. However, in some cases the helical rise might be too small to obtain a power spectrum by nsEM. In most cases, fully optimizing a sample in nsEM is not necessary, as conditions (primarily protein concentrations and some buffer components) will vary between nsEM and cryo-EM, unless one wants to perform nsEM reconstruction first. It is worth stopping once a suitable range of conditions has been found where protein filaments form without excessive bundling.
3.1 Staining protocol 3.1.1 Equipment required • Carbon coated copper grids (glow discharged) • EM tweezers • Parafilm
172
• •
Mariusz Matyszewski and Jungsan Sohn
Blotting paper Uranyl based stains (Either formate or acetate recommended) at 0.75% (keep on ice and away from light) Aspirator (or fume hood if slowly drying the samples instead)
•
3.1.2 Protocol (Fig. 2) 1. Prepare (at least) 5–10 μL samples, giving them enough time to polymerize if using cleavable tags (e.g., 30 min upon TEV addition 1:10 M ratio against target proteins). 2. Lay out a piece of parafilm on the benchtop. Approximately 3 5 in should have a plenty of room for 6–8 samples. Use a razor blade on the corners of the film to adhere it to the benchtop. B e pl Sa m
Sa m
pl
e
#2
#1
A
1. Sample (~10 µL)
3. Stain (20 µL)
C
Applying sample to grid
Blotting sample
Aspirating sample
Fig. 2 Negative Staining Equipment and Setup. (A) Equipment setup. (B) Suggested layout of sample and stain droplets. (C) Sample handling examples. When blotting, always try to blot from the side of the grid first. If slow drying is preferred, blot most of the dye before moving the sample to the fumehood.
Preparation of filamentous proteins for electron microscopy
173
3. Place sample droplets on the parafilm. If excessive bubbles are forming, try using 1 μL less than the total volume. 4. Place grids carbon side down into the sample. Let them sit for approximately 2 min each. 5. While the samples are incubating, place two 20 μL drops of stain per sample. 6. After 2 min, blot the excess sample from the grid and transfer to the first drop. Let this grid float for about 30 s, or until all the other samples have been transferred. The grid might sink if the buffer contained detergent, but this is normal. 7. Move the grid into the second drop for another 30 s. 8. If using the aspirator, remove the stain from the sample and dry it out before transferring to a storage box. If using the drying method, blot away the excess stain and let the grid dry out in a fume hood. This might take 30 min to 1 h. 3.1.3 Variations • Instead of using two drops of stain, the first one can be replaced by buffer as a wash step • Aspiration produces lighter stain. If darker stain is preferred, use the fume hood to dry out the sample instead. This will also create a bigger stain gradient, and often leaves more protein on the grid.
4. Cryo-EM sample preparation Finding the right conditions for freezing cryo-EM grids is a significant part of the project. While the basic idea is the same as nsEM, freezing in ice entails many more variables and challenges. The key challenges are ice thickness and sample distribution. Since cryo-EM sample preparation has been covered many times, and in much more detail in other chapters in this volume. We will only focus on differences relevant to filamentous samples.
4.1 Grid type Many labs choose C-Flat or Quantifoil grids as they are a very popular choice in SPR. Their advantages are still useful for filaments but are not critical. While lacey grids might reduce data collection speed due to their ran˚ dom nature, final reconstructions are rarely affected by grid type (e.g., 3.4 A ˚ lacey Matyszewski et al., 2018 vs. 3.6 A Quantifoil Li et al., 2018).
174
Mariusz Matyszewski and Jungsan Sohn
A
B
Fig. 3 C-Flat vs. Lacey Grids. (A) C-Flat grid with filaments preferentially adhering to carbon over ice-filled holes. (B) Lacey carbon grid has much less carbon to adhere to, thus forcing filaments into ice. Some filament bundling is present in both samples.
Additionally, C-Flat grids tend to have much more carbon which can lead to filaments adhering to the carbon rather than ice-filled holes (Fig. 3).
4.2 Sample concentration and distribution Sample distribution between nsEM and cryo-EM will differ significantly. Often, higher concentration of sample is required to create an even distribution of particles in cryo-EM. This is mainly due to a larger amount of sample being blotted away, as well as a lack of carbon to adhere to. Filaments are big particles, and their presence can be checked for even before the sample ice thickness has been optimized simply by imaging at higher defocus levels (usually about 5 μm). In many cases, changing concentration is fairly straightforward, but be aware that it does not necessarily have a linear correlation with sample distribution. Additionally, distribution of filaments over the grid might not be uniform. Fortunately, in cryo-EM the whole grid does not have to be perfect; a portion of a grid can provide more than enough data in most cases. If increasing protein concentration causes filament bundling, it is possible to modify grid properties instead. One common approach is to change glow-discharging conditions, as it will affect how the sample adheres to grid (Passmore & Russo, 2016). A more drastic approach is to introduce a Thin Carbon Layer (TCL), which will cause the grid to behave more like a standard nsEM sample. However, it comes at a cost of lower signal-to-noise ratio. Either option will allow for working with lower concentration of protein by causing more filaments to adhere to the grid. Additionally, if repulsion tags were not already being used, it can be beneficial to introduce them now, as well as trying to freeze earlier.
Preparation of filamentous proteins for electron microscopy
175
4.3 Other parameters and tips One thing to keep in mind is that buffer composition can still be changed. If the ice is not forming well, lowering glycerol or detergent could help. If filaments are bundling, changing the buffer can help to separate them again. When changing the buffer, it is usually worthwhile to go back to nsEM and test how different components affect filament formation. Be sure to keep the overall components within working limits for cryo-EM. As for humidity and temperature settings of the blotting machine, in most cases setting humidity to 100% and temperature to 4 °C is most advantageous. Both will prevent sample evaporation after blotting. Additionally, drain time setting should ideally be kept to a minimum, if not eliminated, to reduce the exposure to the air-water interface. Since the blotting chamber will usually be at high humidity, blotting paper condition should be monitored. New paper will absorb differently than paper that has been in the machine for a few minutes. This can be a source of ice thickness variability, so make sure to keep a record of the blotting paper “age” when making grids.
References Drulyte, I., et al. (2018). Approaches to altering particle distributions in cryo-electron microscopy sample preparation. Acta Crystallographica. Section D, Structural Biology, 74(Pt 6), 560–571. Egelman, E. H. (2007). The iterative helical real space reconstruction method: Surmounting the problems posed by real polymers. Journal of Structural Biology, 157(1), 83–94. Jin, T., Perry, A., Smith, P., Jiang, J., & Xiao, T. S. (2013). Structure of the absent in melanoma 2 (AIM2) pyrin domain provides insights into the mechanisms of AIM2 autoinhibition and inflammasome assembly. The Journal of Biological Chemistry, 288(19), 13225–13235. K€ uhlbrandt, W. (2014). The resolution revolution. Science, 343(6178), 1443–1444. Li, Y., et al. (2018). Cryo-EM structures of ASC and NLRC4 CARD filaments reveal a unified mechanism of nucleation and activation of caspase-1. Proceedings of the National Academy of Sciences, 115(43), 10845–10852. Lu, A., et al. (2014). Unified polymerization mechanism for the assembly of ASC-dependent inflammasomes. Cell, 156(6), 1193–1206. Lu, A., et al. (2015). Plasticity in PYD assembly revealed by cryo-EM structure of the PYD filament of AIM2. Cell Discovery, 1, 15013. Lu, A., et al. (2016). Molecular basis of caspase-1 polymerization and its inhibition by a new capping mechanism. Nature Structural & Molecular Biology, 23(5), 416–425. Matyszewski, M., Zheng, W., Lueck, J., Antiochos, B., Egelman, E. H., & Sohn, J. (2018). Cryo-EM structure of the NLRC4CARD filament provides insights into how symmetric and asymmetric supramolecular structures drive inflammasome assembly. The Journal of Biological Chemistry, 293(52), 20240–20248. Morrone, S. R., Matyszewski, M., Yu, X., Delannoy, M., Egelman, E. H., & Sohn, J. (2015). Assembly-driven activation of the AIM2 foreign-dsDNA sensor provides a polymerization template for downstream ASC. Nature Communications, 6, 7827.
176
Mariusz Matyszewski and Jungsan Sohn
Passmore, L. A., & Russo, C. J. (2016). Specimen preparation for high-resolution Cryo-EM. Methods in Enzymology, 579, 51–86. Shen, C., et al. (2019). Molecular mechanism for NLRP6 inflammasome assembly and activation. Proceedings of the National Academy of Sciences, 116(6), 2052–2057. Tenthorey, J. L., et al. (2017). The structural basis of flagellin detection by NAIP5: A strategy to limit pathogen immune evasion. Science, 358(6365), 888–893.
Preparation of filamentous proteins for electron microscopy visualization and reconstruction Mariusz Matyszewski, Jungsan Sohn* Department of Biophysics and Biophysical Chemistry, Johns Hopkins University School of Medicine, Baltimore, MD, United States *Corresponding author: e-mail address: [email protected]
Contents 1. Introduction 2. Construct design 2.1 Controlling oligomerization 2.2 Filament separation (bundling) 2.3 Buffer components 3. Negative stain electron microscopy 3.1 Staining protocol 4. Cryo-EM sample preparation 4.1 Grid type 4.2 Sample concentration and distribution 4.3 Other parameters and tips References
168 168 169 169 170 171 171 173 173 174 175 175
Abstract Cryo electron microscopy (cryo-EM) has become a mainstream tool for determining the structures of macromolecular complexes at the atomic resolution. It has many advantages over other techniques such as X-ray crystallography and nuclear magnetic resonance (NMR). However, it also entails several challenges, a major one being preparation of an ideal sample. Recent studies have identified that DNA sensors and inflammasomes often assemble into filamentous oligomers, which poses a unique set of challenges in preparing ideal samples for high-resolution reconstruction using cryo-EM. This chapter will discuss how to overcome several major issues in cryo-EM sample preparation including construct design, screening using negative stain (ns) EM, and tips on working with filamentous proteins.
Methods in Enzymology, Volume 625 ISSN 0076-6879 https://doi.org/10.1016/bs.mie.2019.06.007
#
2019 Elsevier Inc. All rights reserved.
167
168
Mariusz Matyszewski and Jungsan Sohn
1. Introduction Thanks to the “resolution revolution,” the use of cryo-EM in structural biology has skyrocketed (K€ uhlbrandt, 2014). Currently, there are almost 3000 high-resolution cryo-EM structures deposited in the PDB, 700 of which were deposited in 2018 alone (www.rcsb.org). While the majority of the structures were solved using the Single Particle Reconstruction (SPR) approach, there are many higher order protein complexes whose structures were determined via helical reconstructions. This is noteworthy as several DNA sensors and inflammasomes (at least their signaling ends) assemble into helical filaments (Lu et al., 2014; Lu et al., 2016; Li et al., 2018; Matyszewski et al., 2018; Morrone et al., 2015). The general principles of working with helical assemblies are similar to those of SPR projects (see elsewhere in this volume), but there are some key differences. By definition, helical samples have helical symmetry which can be used to improve final resolution, often requiring a smaller data set to achieve comparable results. Additionally, some SPR problems like preferential orientation do not apply to filaments as each filament contains all the views necessary for reconstruction (Egelman, 2007). Nevertheless, getting filamentous samples to the collection-ready stage can still be as challenging as in any SPR project. Here we share techniques and tips for preparing an ideal sample for cryo-EM reconstruction of filamentous proteins.
2. Construct design The first step is to design a construct suitable for cryo-EM. In many instances of filamentous complexes found in inflammasomes and DNA sensors, only one domain of the protein is responsible for filamentation (Morrone et al., 2015; Shen et al., 2019; Tenthorey et al., 2017). Isolating such a domain can significantly simplify the challenge of getting a highresolution structure of the minimal filament. On the other hand, working with the full-length protein can provide additional insight into the workings of the filament, but comes at a cost of a much more difficult system to work with (i.e., expression and purification). Moreover, the rest of the protein usually has regulatory roles or act as a scaffold for filament assembly, often resulting in disordered arrays surrounding the ordered filament core (Shen et al., 2019). Nevertheless, we found that even working with isolated polymerization domains poses many challenges.
Preparation of filamentous proteins for electron microscopy
169
2.1 Controlling oligomerization One of the biggest problems working with filamentous protein is controlling their oligomeric state. While the final goal is to capture the protein as a filament, it is often advantageous to prepare it as a monomer and to later induce polymerization in a controlled manner. Monomeric constructs tend to be significantly more soluble than their filamented counterparts, allowing higher yields and better concentration control. The most straightforward solution for the oligomerization problem is to use cleavable solubility tags. In many cases, large protein tags such as Maltose Binding Protein (MBP) can sterically interfere with filament assembly ( Jin, Perry, Smith, Jiang, & Xiao, 2013), greatly reducing the intrinsic oligomerization activity. Combined with a protease cut site, such as Tabacco Etch Virus (TEV) protease, this not only prevents oligomerization, but also allows for temporal control over filamentation (Morrone et al., 2015). For the vast majority of inflammasome signaling components, the polymerization domain is located at the N-terminus (e.g., PYrin Domain, PYD or CAspase-Recruiting Domain, CARD). For best prospect of inhibiting polymerization, the tag should be placed at the N-terminus of either PYD or CARD with minimal linker (Fig. 1A; MBP-CARD/PYD construct).
2.2 Filament separation (bundling) In many cases, despite controlling their polymerization using the methods described above, resulting filaments aggregate and “bundle up” (Fig. 1B). If simply lowering the concentration does not solve this issue, the next best option is to introduce another tag which would repulse other filaments, while still allowing polymerization. The identity of such repulsion tags can vary, but enhanced Green Fluorescence Protein (eGFP) has been used with great success (Li et al., 2018; Matyszewski et al., 2018). However, care should be taken not to affect the native form of the filament (Lu et al., 2015). To avoid this issue, the tag should ideally be placed as to mimic the rest of the native protein. If working with the full-length construct, the tag should be placed at the opposite terminus from the solubility tag (Fig. 1A; MBPCARD/PYD-eGFP construct). The end result should be a double tagged protein, with filaments only forming after solubility tag is cleaved (Figs. 1B-b and c). Note: the repulsion-tags should not be visible in the final reconstruction, as they should be flexibly attached to the filament core without affecting its
170
Mariusz Matyszewski and Jungsan Sohn
A
CARD/PYD MBP
Controlled polymerization
TEVp cleavage site MBP
Controlled polymerization/
eGFP reduced “bundling”
B
a.
b.
c.
Fig. 1 Construct variants for filamentous protein. (A) Typical inflammasome receptors contain either PYD or CARD at the N-terminus, which is capable of auto-polymerization. Adding MPB with TEV protease cut-site to the N-terminus of CARD/PYD allows controlled polymerization. Adding eGFP at the C-terminus minimizes bundling. (B) nsEM micrographs of (a) auto-assembled NLRC4CARD, (b) MBP-CARD of NRLC4 (NLRC4CARD)-eGFP, and (c) NLRC4CARD-eGFP after removing MBP via TEVp.
native architecture. Otherwise, the presence of the tag in the final reconstruction could suggest that the native symmetry has been tempered with, especially for truncated constructs (Lu et al., 2015). An even less intrusive way of preventing bundling is temporal control. Since bundling requires pre-formed filaments, decreasing the incubation time before freezing or staining samples might reduce aggregation. However, the efficacy of this approach will vary by samples and is usually best when combined with other approaches.
2.3 Buffer components While most buffer components do not affect the quality of the sample in cryo-EM, there are a few things to avoid, or be cautious about, in the final sample buffer. 1. Salt: High salt concentrations have been shown to reduce contrast, and thus Signal-to-Noise Ratio (SNR) of cryo-EM micrographs. Typically, concentrations below 300 mM should be used (Drulyte et al., 2018).
Preparation of filamentous proteins for electron microscopy
171
2. Glycerol: Reduces the SNR and should be kept at or below 5% concentration (Drulyte et al., 2018) for cryo-EM data collection. 3. Detergent: While detergents above critical micelle concentration will add to the background, use of detergent can also be beneficial in changing particle distribution, especially when it comes to air-water interface interactions. Since it can affect how the sample freezes, only introduce it if the particle distribution problem occurs. It will often require screening many detergents before the best one is found. 4. Phosphate: Interferes with nsEM. Uranyl will react with phosphate to form crystals, making it impossible to screen with phosphate buffers present. It can be safely reintroduced if only doing cryo-EM.
3. Negative stain electron microscopy Imaging the purified protein sample by Negative Stain Electron Microscopy (nsEM) is a critical step even if the final goal is a cryo-EM reconstruction. While nsEM is unlikely to provide anything more than a symmetry estimate, it can provide plenty of qualitative information on how the sample filaments behave. The most crucial piece of information is if the protein actually polymerizes into a filament vs. disordered aggregates (this cannot be distinguished by size-exclusion chromatography). This is especially critical when testing new constructs and buffer conditions, as nsEM is a much higher throughput method than cryo-EM. However, one has to keep in mind that conditions might not transfer to cryo-EM; freezing is much different than staining a sample. Once filamentation has been confirmed, one could potentially try to reconstruct the nsEM filaments. Usually it is at least worth to check for symmetry by collecting a power spectrum. However, in some cases the helical rise might be too small to obtain a power spectrum by nsEM. In most cases, fully optimizing a sample in nsEM is not necessary, as conditions (primarily protein concentrations and some buffer components) will vary between nsEM and cryo-EM, unless one wants to perform nsEM reconstruction first. It is worth stopping once a suitable range of conditions has been found where protein filaments form without excessive bundling.
3.1 Staining protocol 3.1.1 Equipment required • Carbon coated copper grids (glow discharged) • EM tweezers • Parafilm
172
• •
Mariusz Matyszewski and Jungsan Sohn
Blotting paper Uranyl based stains (Either formate or acetate recommended) at 0.75% (keep on ice and away from light) Aspirator (or fume hood if slowly drying the samples instead)
•
3.1.2 Protocol (Fig. 2) 1. Prepare (at least) 5–10 μL samples, giving them enough time to polymerize if using cleavable tags (e.g., 30 min upon TEV addition 1:10 M ratio against target proteins). 2. Lay out a piece of parafilm on the benchtop. Approximately 3 5 in should have a plenty of room for 6–8 samples. Use a razor blade on the corners of the film to adhere it to the benchtop. B e pl Sa m
Sa m
pl
e
#2
#1
A
1. Sample (~10 µL)
3. Stain (20 µL)
C
Applying sample to grid
Blotting sample
Aspirating sample
Fig. 2 Negative Staining Equipment and Setup. (A) Equipment setup. (B) Suggested layout of sample and stain droplets. (C) Sample handling examples. When blotting, always try to blot from the side of the grid first. If slow drying is preferred, blot most of the dye before moving the sample to the fumehood.
Preparation of filamentous proteins for electron microscopy
173
3. Place sample droplets on the parafilm. If excessive bubbles are forming, try using 1 μL less than the total volume. 4. Place grids carbon side down into the sample. Let them sit for approximately 2 min each. 5. While the samples are incubating, place two 20 μL drops of stain per sample. 6. After 2 min, blot the excess sample from the grid and transfer to the first drop. Let this grid float for about 30 s, or until all the other samples have been transferred. The grid might sink if the buffer contained detergent, but this is normal. 7. Move the grid into the second drop for another 30 s. 8. If using the aspirator, remove the stain from the sample and dry it out before transferring to a storage box. If using the drying method, blot away the excess stain and let the grid dry out in a fume hood. This might take 30 min to 1 h. 3.1.3 Variations • Instead of using two drops of stain, the first one can be replaced by buffer as a wash step • Aspiration produces lighter stain. If darker stain is preferred, use the fume hood to dry out the sample instead. This will also create a bigger stain gradient, and often leaves more protein on the grid.
4. Cryo-EM sample preparation Finding the right conditions for freezing cryo-EM grids is a significant part of the project. While the basic idea is the same as nsEM, freezing in ice entails many more variables and challenges. The key challenges are ice thickness and sample distribution. Since cryo-EM sample preparation has been covered many times, and in much more detail in other chapters in this volume. We will only focus on differences relevant to filamentous samples.
4.1 Grid type Many labs choose C-Flat or Quantifoil grids as they are a very popular choice in SPR. Their advantages are still useful for filaments but are not critical. While lacey grids might reduce data collection speed due to their ran˚ dom nature, final reconstructions are rarely affected by grid type (e.g., 3.4 A ˚ lacey Matyszewski et al., 2018 vs. 3.6 A Quantifoil Li et al., 2018).
174
Mariusz Matyszewski and Jungsan Sohn
A
B
Fig. 3 C-Flat vs. Lacey Grids. (A) C-Flat grid with filaments preferentially adhering to carbon over ice-filled holes. (B) Lacey carbon grid has much less carbon to adhere to, thus forcing filaments into ice. Some filament bundling is present in both samples.
Additionally, C-Flat grids tend to have much more carbon which can lead to filaments adhering to the carbon rather than ice-filled holes (Fig. 3).
4.2 Sample concentration and distribution Sample distribution between nsEM and cryo-EM will differ significantly. Often, higher concentration of sample is required to create an even distribution of particles in cryo-EM. This is mainly due to a larger amount of sample being blotted away, as well as a lack of carbon to adhere to. Filaments are big particles, and their presence can be checked for even before the sample ice thickness has been optimized simply by imaging at higher defocus levels (usually about 5 μm). In many cases, changing concentration is fairly straightforward, but be aware that it does not necessarily have a linear correlation with sample distribution. Additionally, distribution of filaments over the grid might not be uniform. Fortunately, in cryo-EM the whole grid does not have to be perfect; a portion of a grid can provide more than enough data in most cases. If increasing protein concentration causes filament bundling, it is possible to modify grid properties instead. One common approach is to change glow-discharging conditions, as it will affect how the sample adheres to grid (Passmore & Russo, 2016). A more drastic approach is to introduce a Thin Carbon Layer (TCL), which will cause the grid to behave more like a standard nsEM sample. However, it comes at a cost of lower signal-to-noise ratio. Either option will allow for working with lower concentration of protein by causing more filaments to adhere to the grid. Additionally, if repulsion tags were not already being used, it can be beneficial to introduce them now, as well as trying to freeze earlier.
Preparation of filamentous proteins for electron microscopy
175
4.3 Other parameters and tips One thing to keep in mind is that buffer composition can still be changed. If the ice is not forming well, lowering glycerol or detergent could help. If filaments are bundling, changing the buffer can help to separate them again. When changing the buffer, it is usually worthwhile to go back to nsEM and test how different components affect filament formation. Be sure to keep the overall components within working limits for cryo-EM. As for humidity and temperature settings of the blotting machine, in most cases setting humidity to 100% and temperature to 4 °C is most advantageous. Both will prevent sample evaporation after blotting. Additionally, drain time setting should ideally be kept to a minimum, if not eliminated, to reduce the exposure to the air-water interface. Since the blotting chamber will usually be at high humidity, blotting paper condition should be monitored. New paper will absorb differently than paper that has been in the machine for a few minutes. This can be a source of ice thickness variability, so make sure to keep a record of the blotting paper “age” when making grids.
References Drulyte, I., et al. (2018). Approaches to altering particle distributions in cryo-electron microscopy sample preparation. Acta Crystallographica. Section D, Structural Biology, 74(Pt 6), 560–571. Egelman, E. H. (2007). The iterative helical real space reconstruction method: Surmounting the problems posed by real polymers. Journal of Structural Biology, 157(1), 83–94. Jin, T., Perry, A., Smith, P., Jiang, J., & Xiao, T. S. (2013). Structure of the absent in melanoma 2 (AIM2) pyrin domain provides insights into the mechanisms of AIM2 autoinhibition and inflammasome assembly. The Journal of Biological Chemistry, 288(19), 13225–13235. K€ uhlbrandt, W. (2014). The resolution revolution. Science, 343(6178), 1443–1444. Li, Y., et al. (2018). Cryo-EM structures of ASC and NLRC4 CARD filaments reveal a unified mechanism of nucleation and activation of caspase-1. Proceedings of the National Academy of Sciences, 115(43), 10845–10852. Lu, A., et al. (2014). Unified polymerization mechanism for the assembly of ASC-dependent inflammasomes. Cell, 156(6), 1193–1206. Lu, A., et al. (2015). Plasticity in PYD assembly revealed by cryo-EM structure of the PYD filament of AIM2. Cell Discovery, 1, 15013. Lu, A., et al. (2016). Molecular basis of caspase-1 polymerization and its inhibition by a new capping mechanism. Nature Structural & Molecular Biology, 23(5), 416–425. Matyszewski, M., Zheng, W., Lueck, J., Antiochos, B., Egelman, E. H., & Sohn, J. (2018). Cryo-EM structure of the NLRC4CARD filament provides insights into how symmetric and asymmetric supramolecular structures drive inflammasome assembly. The Journal of Biological Chemistry, 293(52), 20240–20248. Morrone, S. R., Matyszewski, M., Yu, X., Delannoy, M., Egelman, E. H., & Sohn, J. (2015). Assembly-driven activation of the AIM2 foreign-dsDNA sensor provides a polymerization template for downstream ASC. Nature Communications, 6, 7827.
176
Mariusz Matyszewski and Jungsan Sohn
Passmore, L. A., & Russo, C. J. (2016). Specimen preparation for high-resolution Cryo-EM. Methods in Enzymology, 579, 51–86. Shen, C., et al. (2019). Molecular mechanism for NLRP6 inflammasome assembly and activation. Proceedings of the National Academy of Sciences, 116(6), 2052–2057. Tenthorey, J. L., et al. (2017). The structural basis of flagellin detection by NAIP5: A strategy to limit pathogen immune evasion. Science, 358(6365), 888–893.