High-quality ultrastructural preservation using cryofixation for 3D electron microscopy of genetically labeled tissues

Thank you for submitting your article "High-quality ultrastructural preservation using cryofixation for 3D electron microscopy of genetically labeled tissues" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor, who served as one of the reviewers, and Eve Marder as the Senior Editor. The following individual involved in review of your submission has agreed to reveal his identity: Richard Leapman (Reviewer #3).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

The manuscript reports a modified approach for sample fixation and staining that opens the possibility to combine cryofixation with 3D electron microscopic imaging and correlated light-electron microscopic analysis.

While the reviewers see the merits of this work, they raised a set of key concerns, and after discussion decided to request the following points for the revision of this manuscript:

- Proper citation and discussion of the literature as indicated by reviewer 2; clarification of which additional advances were made compared to the published protocols.

- More extensive discussion of the limitations of the approach (see reviewers 1 and 3): tissue volume effects, potentially problematic effects of the rehydration step; clarification for which examples the "improved temporal resolution" apply.

- Better description of the reported applications (see comments by reviewers 1 and 3 below): which resolution of correlated imaging was achieved; how does this depend on the specimen and procedure; how does it combine with circuit reconstruction (if this is envisioned to be possible in this approach).

All other points raised by the reviewers can be treated as recommendations:

Reviewer #1:

The manuscript "High-quality ultrastructural preservation using cryofixation for 3D electron microscopy of genetically labeled tissues" by Tsang et al. describes a novel fixation protocol providing the possibility to cryofix tissue followed by high-contrast heavy metal staining amenable to 3D electron microscopy and offering the possibility to perform correlated light EM imaging.

The challenges addressed by this manuscript are substantial and the solutions presented appear convincing. This is an important methodological contribution that in principle would merit a publication in eLife. However, I have a few concerns that in my view need to be addressed:

1) The concrete applications of the improved protocol remain a little unclear. Especially it is not clear at what resolution and for which volumes correlated LM-EM imaging becomes possible with this approach. The authors show data that speaks for cellular level correlation, i.e. the identification of stained cell bodies which has been possible before (Bock et al.; Briggman et al.) and would not be a major improvement. A correlation at the level of subcellular structures while still being able to acquire large 3D EM volumes would be highly commendable but it is not clear whether that is in fact achieved. Moreover, numbers on dataset size, resolution, etc. are missing from the Results section. They can be found in the Materials and methods but for such a methodological paper it would be absolutely critical in this reviewer’s view to have the numbers on acquired 3D EM volumes, resolution etc. in the main text. It appears that the SBEM volume acquired has a z extent of about 50 micrometers, however at a section thickness of 70 nanometers. 70 nanometers of section thickness would not routinely allow for neuronal circuitry construction – neither in fly nor in mammalian brains. This should be addressed or clarified. Again, all of these points would profit from a clearer description of the concrete application examples.

2) The description is not extremely clear, for instance the usage of the word "temporal" in the Abstract comes as a surprise and is only understandable after realizing that the authors are referring to the time scale of fixations. This is just one example of several.

3) Along these lines the figures are not optimal. While they show some images that look plausible, more and better labeled panels may be needed to explain the application examples, potentially in a protocol sketch for each figure, making it clearer which possible application (which attempted imaging resolution, scope, LM-EM correlation at subcellular or cellular scale, cell bodies or dendrites or axons etc.) is shown.

Reviewer #2:

Tsang et al. describe a hybrid chemical/freezing fixation method to results in specimens in an aqueous environment. The authors show this method allows subsequent enzymatic reactions, high-contrast en bloc staining, and correlative fluorescence microscopy. They also demonstrate the applicability in cultured cells, fly antennae, and acute mouse brain slices.

The methodological advance claimed by the authors is a rehydration step following freeze substitution. They compare their method to (van Donselaar et al., 2007), but there are uncited examples of hybrid protocols involving rehydration to achieve an aqueous environment for further tissue processing: Ripper, Schwarz and Stierhof, 2008; Stierhof and Kasmi, 2010.

Indeed, these studies demonstrate advances that Tsang et al. seem to be claiming as novel. For example, (Stierhof et al., 2010) explicitly show the ability to image GFP, YFP and RFP expression following rehydration.

I suppose the demonstration of high-contrast heavy metal staining and enzyme chemistry in an aqueous environment following freeze substitution is new, but not particularly surprising. Given the prior use of rehydration following freeze substitution for similar purposes, I doubt the presented method requires a new acronym (CCM).

Finally, one of the benefits claimed in the text and Figure 1 is the high temporal resolution of the method. For cell culture and fly antenna this is true but, in practice, this does not apply to the data presented from mouse brain in which chemical fixation by perfusion was first carried out.

Reviewer #3:

This paper highlights a new approach – CryoChem Method (CCM), which researchers can use to perform 3D electron microscopy on large specimens (up to a scale of ~500 µm in each dimension); these samples are first frozen rapidly at high pressure in their native state, and then freeze-substituted with mild fixation, before being rehydrated. This rehydration step is the key aspect of the work: it improves ultrastructural quality, while also importantly maintaining the viability of proteins that can be subsequently imaged by fluorescent tagging, e.g., DAB-based labeling schemes, such as miniSOGs and DAB polymerization before staining and embedding for SBEM, or other 3D imaging techniques. Although the CCM might seem like an obvious extension of existing approaches, it is not yet widely known by the scientific community, and is therefore likely to be of considerable value to structural and cell biologists.

Many valuable new techniques have some limitations in addition to their important advantages. Perhaps the authors could discuss some potential limitations (if they exist). For example, presumably some molecules that are retained in standard freeze-substitution protocols might be lost in the rehydration step. Is this a concern? One of the most important applications of the CCM approach is its ability to provide correlative LM/EM on large specimens. However, the LM is performed in the liquid state after hydration before embedding, whereas the 3D SBEM is performed after embedding. What is the precision of the correlative imaging? Does the rehydration introduce swelling or the embedding introduce shrinkage? The only figure that illustrates the correlative imaging is Figure 6, but it is difficult to assess the precision of the registration between the SBEM and the confocal LM images in this example. Could the authors provide some estimate of this?