An assumption in microscopy and other methods is that chemical fixation of the cells is fast enough not to affect the biological process we study. A new paper in eLife shows that fixation affects the appearance of phase-separated proteins.
In many experiments, we fix the cells – usually by paraformaldehyde (PFA) – in order to take a “snapshot” of the condition in the cell. PFA chemically cross-links biomolecules (mostly proteins & nucleic acids), thus spatially “fixing” them in place. Fixed cell imaging is used widely – e.g. for immunofluorescence, for DNA FISH and single molecule FISH and more. Usually our assumption is correct. In this new paper from ShaSha Chong’s lab at CalTech, they show that in some cases, for example when trying to image proteins that go liquid-liquid phase separation, fixation can alter the appearance of the protein, compared to live cell imaging.
They compared GFP-tagged protein or protein domains that go LLPS in live and PFA-fixed cells and show that there are more, brighter and differnt shaped puncta after fixation. This effect was also seen with different concentrations of PFA from 1% to 8%, or addition of glutaraldehyde (0.2%). Live imaging during fixation shows the changes are happening withing the first ~100 seconds. Interestingly, replacing the GFP with DsRed or Halotag affected the outcome: Halotag showed diminshed granules. Some proteins did not show any differences between live & fixed cells.
To play with the system, they added glycine to the cells pre-fixation. Glycine eacts strongly with PFA and is used to quench the fixation reaction. Addition of glycine affected the fixation artefacts and created new ones – donut shaped “granules”. But glycine also affected the granule size in live imaging.
They developed a computational kinetic model which suggests that the observed artefacts of LLPS systems are driven by 3 factors: protein–protein interaction dynamics, the absolute overall fixation rate, and different fixation rates in and out of granules. In other words, if fixation dynamics are faster that protein dynamics, there will be less artefacts. The authors show this experimentally and also cite in the introduction examples for these cases – for example transcription factors with fast dissociation kinetics.
The authors stop here with no real solution to the problem. They suggest that fixation should be compared to live imaging whenever possible, and also sompare differnt fixatives (e.g. methanol – which fixes cells by dehydration). They refer specifically to LLPS where proteins aggregate and tend to have more contacts suitable for cross-linking, but any case where protein dynamics are faster than fixation rate can be affected.
Their paper is based on imaging using fluorescent proteins or protein-fluorophore conjugates, but they did not test the effect with immunofluorescence. You may recall my own story about fixation artefacts: how addition of glutaraldehyde to the PFA fixation for smFISH caused my mRNA to disapear. It is possible that artefacts seen with fixed fluorescent proteins will look differnet when doing immunofluorescence, and there might also be differences between antibodies and nanobodies due to their size in their ability to penetrate through the cross-linked proteins.
Overall an important paper that highlights how a key step in a microscopy method can affect the image and its interpertation.
Fixation can change the appearance of phase separation in living cells
Shawn Irgen-Gioro, Shawn Yoshida, Victoria Walling, Shasha Chong (2022) eLife 11:e79903