From Microscopy to Nanoscopy: Contemporary Physical Methods in Mitochondrial Structural Biology

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The review by Nesterov takes on the challenge of reviewing contemporary physical methods in the study of mitochondrial structural biology. The structural biology field is rather broadly defined, as is appropriate in my view, and the topics range from Raman spectroscopy at the level of the chemical bond to the connectivity of mitochondrial networks in cells and tissues. As stated, "the review indicates the mitochondrial processes that each method can analyze and outlines general directions for further development of these studies applied to mitochondria." I found it quite easy to read and useful as a compendium of the literature, and especially the output of innovative Russian labs that may receive less attention in general. My comments are intended as constructive criticism and I recommend publication after minor revision.

The first issue is mostly esthetic. Most of the figures illustrate the methods rather than the results. Mitochondrial research produces some of the most beautiful images in cell biology. The reader would be drawn more effectively in with representative examples. There are a few toward the end, but there is room for more. These would also illustrate the difference between achievable results. The remainder of my comments address specific sub-chapters.

Fluorescence microscopy is correctly emphasized and a number of dye chemistries are described. The major pitfall of fluorescence should also be mentioned. This is not photo-bleaching but photo-toxicity, whose influence is perhaps unappreciated in light of the fact that most everyone follows similar protocols. The illumination intensity in a fluorescence microscope can be orders of magnitude greater that the solar radiation on a clear day, even three or four orders for super-resolution methods. Combined with the well-known (but less well characterized) free radical and triplet oxygen generation, precisely in the tiny volume sensitive to redox potential, this is cause for serious concern. There is also a possible influence of the dyes on the very potential they report. The common JC-1 ratiometric dye, for example, may cause a loss of cristae organization (doi:10.7554/eLife.29929).

In the section on Mathematical post-processing, a distinction should be made between intensity analysis and morphological analysis. This would clarify the discussion and also motivate the deep learning for label-free organelle detection.

In electron microscopy, the classical methods based on thin sections are amenable to high throughput analysis, allowing a classification of physiological conditions than would be possible by tomography or scanning probe methods. (doi:10.1083/jcb.201812081).

The opening statement in the section on Scanning Transmission Electron Microscopy describes the integration of scanning and transmission electron microscopy within a single instrument, but there is no reference. Perhaps one of these is intended (doi:10.1038/s41598-021-00979-z, doi:10.1371/journal.pone.0059573)? Cryo-STEM tomography is touted for contextualizing, and is useful to study the relation to surrounding cytoskeleton and endoplasmic reticulum or other organelles. The criticism in comparison with volume EM is only fair to the extent that cryogenic fixation is considered irrelevant. Direct comparisons indicate that this is not the case, i.e., that fixation can have dramatic effects on ultrastructure (doi:10.1016/j.yexcr.2017.06.022). As a vEM method, cryo-FIB-SEM is indeed feasible and in my own view a major development (doi:10.1016/j.jsb.2013.09.024). The 3D resolution is often marginal due to distortions between successive images, however. An example that nicely highlights mitochondria appears in an article with an unrelated title (doi:10.1016/j.jsb.2020.107528).

The major missing technique, as I can see, is that of patch clamp (for example, a review in doi:10.1016/j.bbabio.2020.148357). I believe the technique deserves mention as it can address membrane permeability most directly.

Finally, some of the references are missing journal names or complete doi. I noticed 98, 108, 109, 150, 171. 

Author Response

Answer to Reviewer 1

Comment 1. The first issue is mostly esthetic. Most of the figures illustrate the methods rather than the results. Mitochondrial research produces some of the most beautiful images in cell biology. The reader would be drawn more effectively in with representative examples. There are a few toward the end, but there is room for more. These would also illustrate the difference between achievable results.

Response 1.

Thank you for your helpful advice. We have added Figure 11 to the additional section, which takes into account the suggestions from both reviews regarding the need for additional illustrations and general guidance on the application of the methods described.

Comment 2. Fluorescence microscopy is correctly emphasized and a number of dye chemistries are described. The major pitfall of fluorescence should also be mentioned. This is not photo-bleaching but photo-toxicity, whose influence is perhaps unappreciated in light of the fact that most everyone follows similar protocols. The illumination intensity in a fluorescence microscope can be orders of magnitude greater that the solar radiation on a clear day, even three or four orders for super-resolution methods. Combined with the well-known (but less well characterized) free radical and triplet oxygen generation, precisely in the tiny volume sensitive to redox potential, this is cause for serious concern. There is also a possible influence of the dyes on the very potential they report. The common JC-1 ratiometric dye, for example, may cause a loss of cristae organization (doi:10.7554/eLife.29929).

Response 2. We fully agree with those statements about the problems of fluorescent microscopy techniques. We have already mentioned for some dyes that they are known to induce photoinduction or phototoxicity. Also we added additional part to “limitations” section:

“It is important to remember that using fluorescent dyes is an invasive technique that may alter certain properties of the studied living system. This can result from specific interactions of the dye with proteins or lipids, as well as phototoxicity effects inherent to nearly all dyes. Additionally, high-intensity illumination in the microscope directly affects the sample, potentially causing local heating or free radical generation. Depending on the instrument settings, these effects can vary, complicating both the reproducibility and interpretation of results.”

Comment 3. In the section on Mathematical post-processing, a distinction should be made between intensity analysis and morphological analysis. This would clarify the discussion and also motivate the deep learning for label-free organelle detection.

Response 3. In the context of this work, although we mention the use of fluorescent dyes to study various functional aspects in mitochondria, the focus of the review is on structural studies. In this regard, the section on mathematical post-processing discusses the main aspects of this particular part of the research - the problem of objectively assessing the formation of mitochondrial networks. We have added a clearer indication of this in the text:

“For an objective assessment of mitochondrial clustering and the formation of mitochondrial networks, an important step is the mathematical processing of the obtained images. The key stages of such processing are briefly described below.”

Comment 4. In electron microscopy, the classical methods based on thin sections are amenable to high throughput analysis, allowing a classification of physiological conditions than would be possible by tomography or scanning probe methods. (doi:10.1083/jcb.201812081).

Response 4.

Thank you for the useful addition. We have added an indication of this optimized EM usage methodology:
“An interesting enhancement of EM for improved convenience and performance is MultiCLEM, a high-throughput correlative light and electron microscopy technique that uses fluorescence barcodes to identify cell types in mixed populations and analyze the structure of organelles in them, including mitochondria [131].”

Comment 5. The opening statement in the section on Scanning Transmission Electron Microscopy describes the integration of scanning and transmission electron microscopy within a single instrument, but there is no reference. Perhaps one of these is intended (doi:10.1038/s41598-021-00979-z, doi:10.1371/journal.pone.0059573)? Cryo-STEM tomography is touted for contextualizing, and is useful to study the relation to surrounding cytoskeleton and endoplasmic reticulum or other organelles. The criticism in comparison with volume EM is only fair to the extent that cryogenic fixation is considered irrelevant. Direct comparisons indicate that this is not the case, i.e., that fixation can have dramatic effects on ultrastructure (doi:10.1016/j.yexcr.2017.06.022). As a vEM method, cryo-FIB-SEM is indeed feasible and in my own view a major development (doi:10.1016/j.jsb.2013.09.024). The 3D resolution is often marginal due to distortions between successive images, however. An example that nicely highlights mitochondria appears in an article with an unrelated title (doi:10.1016/j.jsb.2020.107528).

Response 5.

Thank you for the useful clarifications, the STEM methodology was really poorly disclosed, and references to useful literature for readers are not enough for a deeper acquaintance with the topic. We have expanded this section a bit, without diving into excessive subtleties, but outlining a little more aspects regarding this method:
“An intriguing approach involves the integration of scanning and transmission electron microscopy within a single instrument, enabling the capture of nearly all electrons scattered by the specimen and thereby enhancing analytical capabilities [143]. A key advantage of this setup for biological samples is the elimination of the need for heavy metal staining.  The possibility of usage in cryogenic mode [144] gives this method an advantage against volume microscopy methods relying on chemical fixation, which can disturb the native structure [145]. The complexity of this method is the need for preliminary unroofing of cells, that is, the removal of part of the membrane to expose cytoplasmic structures. But after that, opening of the sample deep layers can be carried out by sublimation under controlled temperature cycling in the cryo-STEM device allowing to reduce artifacts associated with the deformation of native structures [146].”

Comment 6. The major missing technique, as I can see, is that of patch clamp (for example, a review in doi:10.1016/j.bbabio.2020.148357). I believe the technique deserves mention as it can address membrane permeability most directly.

Response 6.

Patch-clamp was not included in the review as it is primarily an electrophysiological technique focused on measuring ion channel currents and membrane potentials, rather than providing structural information about mitochondria (e.g., morphology, cristae architecture, or protein distributions). Structural insights typically require imaging methods like EM, confocal, or super-resolution microscopy. We also did not include many other techniques that do not provide direct structural information. In part, the review indeed contains methods that provide functional information, in particular, an entire chapter is devoted to fluorescent probes. But this is because, in combination with the appropriate microscope, they also allow visualization of the spatial distribution of the fluorescence.

Comment 7. Finally, some of the references are missing journal names or complete doi. I noticed 98, 108, 109, 150, 171. 

Response 7.

Thank you for your attention when reviewing the links. For some reason doi of bioRxiv preprints were cut. However, some of these articles to this time have been already published in journals. Citations were replaced or corrected.

 

Reviewer 2 Report

Comments and Suggestions for Authors

This review addresses an important and timely topic: modern physical/biophysical methods for studying mitochondrial structure and function across spatial scales. The manuscript’s scope—spanning advanced fluorescence and super-resolution microscopy, electron/volume EM, scanning probe methods, and cryo-EM/cryo-ET—fits the journal and will be useful to readers entering the field. The emphasis on quantitative imaging, modern probes, and machine learning-based analysis is also appropriate.

That said, the current version would benefit from clearer structuring, more explicit guidance on method selection (what to use when), and a more critical discussion of limitations, artifacts, and comparability across techniques. Certain parts read more like lists than integrated discussions.

 

Major

1. The manuscript describes a range of measurement methods in a mostly parallel, descriptive way, but the connections between them are unclear. Consider integrating the sections by comparing which biological questions each method addresses best, along with the appropriate sample types, resolution, and/or throughput. This would substantially increase the practical value.
2. Table 1 includes only a small subset of the dyes discussed in the main text. It would be more practically useful if, for each dye, you also summarized key considerations such as phototoxicity, probe accumulation, dependence on membrane potential, pH sensitivity, and off-target localization.
3. In the super-resolution section, (1) The authors state that STORM is suitable for specific live-cell applications, but the issue of long acquisition times needs to be addressed, and references should be added to support this claim. (2)It would also be helpful to discuss the limitations of each method, such as reconstruction artifacts, labeling-density constraints, and fixation effects.
4. The manuscript notes that cryo-EM has enabled full structures of large mitochondrial complexes (e.g., ATP synthase, respirasomes). This is a strong point, but the section would be improved by: (1) defining cryo-EM single-particle analysis, (2) clearly contrasting what single-particle cryo-EM resolves (high-resolution structures of isolated/purified complexes) versus what cryo-ET resolves (in situ supramolecular organization in a membrane context), and (3) using consistent terminology (e.g., “cryo-TEM” vs “cryo-EM”) and clarifying when “TEM” refers to an imaging mode versus the broader cryo-EM umbrella.
5. The review mentions machine learning–based image analysis; please also discuss: (1) ground-truth requirements, annotation bias, overfitting, and domain shift across labs/instruments; (2) reproducibility practices (reporting training data, code availability, and metrics beyond accuracy); and (3) situations where classic feature-based analysis may be preferable.

Minor

1. The abbreviation list is extensive and helpful. Please ensure every abbreviation is also defined at first use in the main text and check for consistency (e.g., ExM described as “expansile microscopy”—confirm the correct term and use it consistently).
2. A careful edit to improve concision and remove repetitive phrasing would improve readability, especially in the introduction and “technology overview” sections.

In summary, the manuscript is promising and likely publishable after revisions that improve structure, add method-selection guidance, and strengthen the critical appraisal of limitations, artifacts, and reproducibility.

Author Response

Answers to Reviewer 2

Comment 1. That said, the current version would benefit from clearer structuring, more explicit guidance on method selection (what to use when), and a more critical discussion of limitations, artifacts, and comparability across techniques. Certain parts read more like lists than integrated discussions. The manuscript describes a range of measurement methods in a mostly parallel, descriptive way, but the connections between them are unclear. Consider integrating the sections by comparing which biological questions each method addresses best, along with the appropriate sample types, resolution, and/or throughput. This would substantially increase the practical value.

Response 1. We agree that more general view on the problem, presented in a single image for guidance, will be useful for better readability and convenience. We have added Figure 11 and table 2 with an additional section before the conclusion. It takes into account the suggestions from both reviews regarding the need for additional illustrations and general guidance on the application of the methods described.

 

 

Comment 2. Table 1 includes only a small subset of the dyes discussed in the main text. It would be more practically useful if, for each dye, you also summarized key considerations such as phototoxicity, probe accumulation, dependence on membrane potential, pH sensitivity, and off-target localization.

 

Response 2.

The table was enlarged and other dyes were included.

 

Comment 3. In the super-resolution section, (1) The authors state that STORM is suitable for specific live-cell applications, but the issue of long acquisition times needs to be addressed, and references should be added to support this claim. (2)It would also be helpful to discuss the limitations of each method, such as reconstruction artifacts, labeling-density constraints, and fixation effects.

 

Response 3.

This part was redone and citation added. Also more super resolution methods were discussed now.

 

Comment 4. The manuscript notes that cryo-EM has enabled full structures of large mitochondrial complexes (e.g., ATP synthase, respirasomes). This is a strong point, but the section would be improved by: (1) defining cryo-EM single-particle analysis, (2) clearly contrasting what single-particle cryo-EM resolves (high-resolution structures of isolated/purified complexes) versus what cryo-ET resolves (in situ supramolecular organization in a membrane context), and (3) using consistent terminology (e.g., “cryo-TEM” vs “cryo-EM”) and clarifying when “TEM” refers to an imaging mode versus the broader cryo-EM umbrella.

 

Response 4.

The changes have been done to this section. The abbreviations changed and some additional information for better understanding of single particle analysis and subtomographic averaging was added.

 

Comment 5. The review mentions machine learning–based image analysis; please also discuss: (1) ground-truth requirements, annotation bias, overfitting, and domain shift across labs/instruments; (2) reproducibility practices (reporting training data, code availability, and metrics beyond accuracy); and (3) situations where classic feature-based analysis may be preferable.

 

Response 5.

We add such additional warning in the new section summarizing all discussed techniques:
“For several methods the machine learning is an integral part. It has two sides now – first it allows analyzing much larger data sets and decreasing the person-based discrepancies. On the other hand it critically depends on initial training data set (annotation quality, instrument in which it was obtained). That may impede the free transfer of pre-trained models between laboratories, reduce the reproducibility of results, and lead to unintentional distortions. For small samples or those with significant peculiarities, it is still more reasonable to use manual processing by specialists. At the same time, many machine learning-based approaches have already proven themselves to be indispensable components of data processing (for example, for cryo-TEM). Development in this area, optimized for standard instrumentation, will likely mitigate most of these shortcomings in the future.”

 

 

Comments 6 (minor). The abbreviation list is extensive and helpful. Please ensure every abbreviation is also defined at first use in the main text and check for consistency (e.g., ExM described as “expansile microscopy”—confirm the correct term and use it consistently).
A careful edit to improve concision and remove repetitive phrasing would improve readability, especially in the introduction and “technology overview” sections.

Response 6.
The abbreviations were revised.