I am working with amyloid oligomers and am particularly interested in identifying annular (ring-like) and globular oligomeric forms. I recently collected some low-magnification cryo-TEM data, but I am having difficulty interpreting the micrographs.
Specifically, I am unsure whether the features I observe correspond to actual oligomers or are artifacts from ice or ethane contamination. I do see some ring-like structures with outer diameters in the range of ~5–10 nm, as well as globular features of varying sizes. However, these appearances seem quite similar to what I might expect from contamination (e.g., ethane or ice), making it difficult to confidently distinguish real particles from background artifacts.
Does anyone have experience distinguishing amyloid oligomers from ice/ethane contamination in cryo-EM micrographs? Are there specific visual cues or validation steps (e.g., defocus dependence, consistency across micrographs, 2D classification, or power spectrum features) that could help confirm whether these features are genuine oligomers?
For reference, the micrograph is taken at -10 defocus due to ice thickness.
I’d agree with @olibclarke, but would also add, it can be helpful to have a buffer control to compare to when you are just starting out (or if you do something weird with a new buffer, etc). It’s worth it to make one extra grid and take some screening images at the beginning of a session to confirm if you can see your protein in particular versus ice / ethane contamination. Not much to add specifically on amyloids, but if you have a lot of different sizes and shapes it’s going to be an uphill battle to get much from 2Ds onwards anyways, so you definitely want to be sure you see a difference between protein and buffer conditions.
Agree with both; no protein visible and while a buffer-only control isn’t always possible, but it’s certainly desirable!
Also remember that just because you use the same conditions twice, doesn’t mean the grids will be the same. Reproducibility of grids can be a real headache. It improves with experience, but even then you can do the same thing five times and get different results in terms of ice quality, thickness, protein dispersion, etc. Machines help, but the first volume of sample you take out of the Eppendorf is not the same as the second, third and so on…
Thank you for the responses. As suggested by @tlevitz I managed to collect a buffer control grid, sharing the comparison here. Those ring-like structures I was excited about also show up in the buffer control, which was a bit heartbreaking to see. Even if my peptide grid does contain real annular oligomers, distinguishing them from ice artifacts of similar size is going to be a real challenge.
Would running 2D classification on buffer control particles and peptide grid particles and directly comparing class averages be a reasonable validation step? Looks like it’s going to be a long journey with this type of sample.
Given how much data you would need to collect, it might not be worth it. But for one shot, if you’ve got a few hundred micrographs it might help settle your mind about the sample, so… why not?
I’m glad that helped (even if it’s not the situation you were hoping for). Generally speaking, my experience is that once you have sufficient protein concentrations such that a protein condition and a buffer condition look clearly different, then differentiation comes out pretty easily in the 2D classes, and if your concentration is so low that there isn’t much of a difference between sample and buffer conditions, processing with the goal of getting something isn’t worth it. but as @rbs_sci says, if you’ve already got the data, you might as well go for it – it’s good workflow practice if nothing else! It does look like your protein vs buffer conditions are at different mags though, so be careful when comparing as any differences may be artifacts of that.
