What shape is a milk jug?
Milk jugs have an interesting 3D shape – symmetric from many angles and then, boom, a handle. The handle helps the shopper to transport the jug from the fridge to the checkout and eventually home. It also plays a supporting role in milk storage, being hollow and integrated into the larger body of the milk jug.
Now, imagine your very first view of a milk jug was at a distance with all the jugs in the store oriented such that the handles were all pointing toward the back of the refrigerator. Without peering around to the back of the jug (changing your viewpoint) or picking up and rotating a jug, you’d never know that this marvelous handle existed. Or course, if the jugs were placed on the shelf with more varying orientations, this would not be an issue, though you’d still need to find a way to see the bottom of those jugs!
Imaging proteins by cryo-EM has similar requirements when 3-dimensional information is needed. To “see” the overall shape of a protein, or any asymmetric moiety, it’s necessary to capture structural information all around the particle. Without access to all the necessary views (the handles and the bottoms), critical information is missing and cannot be inferred.
How do we see all the way around?
First, it’s important to explain what’s really happening with “single particle analysis” workflows. Yes, we image the 2D project of a single particle within the vitreous layer, which we’ll refer to as “ice.” We do not, however, “zoom in” on a single instance of a protein in ice. Rather, we image many copies of the protein, sometimes millions, and then sort all images both for averaging (bolstering signal to noise) and to find all the needed angles.
Being a 2D projection, each image of a single particle contains only some of the information needed to reconstruct the 3D structure. By taking images of lots of particles, which we hope are randomly oriented in the ice, we can find all the views we need to complete the picture. This is like having the milk jugs placed on the shelf in different ways. Each milk jug position tells us something about the overall 3D structure of the container.
But proteins don't always distribute themselves randomly...
Proteins, like milk jugs, rarely orient themselves in with perfect randomness when vitrified on a cryo-EM grid. Some proteins, by virtue of their morphology (e.g., rods and discs) tend to orient with their long axis parallel to the grid. Furthermore, surface properties of both the protein and the air-water interface (AWI) create “stickiness.” For a detailed study of protein distributions within the vitreous layer visit: https://elifesciences.org/articles/34257.
Hydrophobic patches on the protein often associate with the likewise hydrophobic AWI as they tumble in soluble during sample preparation. This leads to instances of what cryo-EM specialists call “preferred orientation.” With highly preferred samples, important features like the handle of a milk jug may be completely missing.
Can preferred orientation be addressed?
There are a variety of strategies which may mitigate or reduce the challenges of preferred orientation. These include:
- Extended data collection: By collecting larger sets of data, it may be possible to obtain additional views missing or underrepresented in shorter imaging sessions. Recent camera upgrades (Gatan K3 and the TFS Falcon 4) now support high throughput image acquisition, thus making larger particles sets available for processing.
- Tilting the grid during data acquisition: While limited in range, increase views obtained in this way may be sufficient for some samples.
- Faster sample prep: Strategies that minimize the duration of opportunity for the protein to stick at the air-water interface. This includes using next generation sample preparation devices like chameleon (SPT Labtech)
- Alternative grid supports
- Modulation of protein-AW interactions: This may include modification of protein surface characteristics, selection of different protein/complex constructs, or the addition of surfactant or other additives just prior to vitrification.
Preferred orientation is a complex phenomenon requiring a multi-pronged approach to sample characterization, optimization, and exploration of vitrification space. We'll note here that AWI issues can be much more severe than preferred orientation - particle deformation, denaturation, complex dissociation, etc, all complicate vitrification efforts, even for samples that are very well behaved in solution.
This is especially so when robust and reproducible sample prep methods are desired for ongoing medicinal chemistry support. Investment in upfront effort to optimize both sample prep and data acquisition strategies can have significant returns in minimizing the cost and time to structure and, most importantly, in ensuring return of complete structural information for high value drug targets.