Practical limitations of superresolution imaging due to conventional sample preparation revealed by a direct comparison of CLSM, SIM and dSTORM.

We evaluate the suitability of conventional sample preparation and labelling methods for two superresolution techniques, structured illumination microscopy and direct stochastic optical reconstruction microscopy, by a comparison to established confocal laser scanning microscopy. We show that SIM is compatible with standard fixation procedures and immunofluorescence labelling protocols and improves resolution by a factor of two compared to confocal laser scanning microscopy. With direct stochastic optical reconstruction microscopy, fluorophores can theoretically be localized with much higher precision. However, in practice, with indirect immunofluorescence labelling density can be insufficient due to the bulky probes to reveal biological structures with high resolution. Fine structures like single actin fibres are in fact resolved with direct stochastic optical reconstruction microscopy when using small affinity probes, but require proper adjustment of the fixation protocol. Finally, by a direct comparison of immunofluorescent and genetic labelling with fluorescent proteins, we show that target morphology in direct stochastic optical reconstruction microscopy data sets can differ significantly depending on the labelling method and the molecular environment of the target.