Mitochondrial dynamics quantitatively revealed by STED nanoscopy with an enhanced squaraine variant probe

Mitochondria play a critical role in generating energy to support the entire lifecycle of biological cells, yet it is still unclear how their morphological structures evolve to regulate their functionality. Conventional fluorescence microscopy can only provide ~300 nm resolution, which is insufficient to visualize mitochondrial cristae. Here, we developed an enhanced squaraine variant dye (MitoESq-635) to study the dynamic structures of mitochondrial cristae in live cells with a superresolution technique. The low saturation intensity and high photostability of MitoESq-635 make it ideal for long-term, high-resolution (stimulated emission depletion) STED nanoscopy. We performed time-lapse imaging of the mitochondrial inner membrane over 50 min (3.9 s per frame, with 71.5 s dark recovery) in living HeLa cells with a resolution of 35.2 nm. The forms of the cristae during mitochondrial fusion and fission can be clearly observed. Our study demonstrates the emerging capability of optical STED nanoscopy to investigate intracellular physiological processes with nanoscale resolution for an extended period of time.

The authors state 'the fluorescent molecule could melt', with high laser power. Not sure that 'melt' it the correct term here.
While the images in Sup Fig 2 certainly look like mitochondrial networks, it would be nice to see co-localization with another mitochondrial marker for these other cell lines.
Generally speaking, many of the images are too small to adequately see at 100% magnification. The manuscript by Yang and coworkers addresses a timely need in super-resolution imaging, namely the generation of live-cell imaging compatible probes to follow mitochondrial dynamics. The topic per se is thus relevant and interesting to a wide readership. A very similar paper appeared in Proceedings of the National Academy of Sciences earlier this year (Wang et al., PNAS 116, p15817 (2019)). In that paper, also a mitochondria-selective dye was presented for STED imaging, allowing observation of cristae dynamics. Therefore, claims of novelty such as "for the first time" in the abstract and elsewhere are problematic and that prior work should be acknowledged. Imaging in the far red as in the present manuscript (STED laser at 775 nm) is desirable to reduce interactions with cellular constituents. In terms of imaging performance, I would rate the PNAS paper higher (albeit at 660 nm STED wavelength) and I would also rate a recent paper by the Jakobs group higher in terms of imaging quality, using SNAP tag and silicon rhodamine for depletion at 775 nm (Stephan et al., Sci Rep 640920 (2019)). Therefore, the specific merit of the manuscript is to provide a label for STED microscopy at 775 nm which has affinity for mitochondria. However, the manuscript has several shortcomings, such that I cannot recommend it for publication in its present state. Several of the claims are not sufficiently supported by the quality of the data. In other instances the information provided is insufficient to truly assess the validity of the measurements.
Please find my detailed assessment below.
Major points: 1) Absorption coefficient, quantum yield, fluorescence lifetime etc. for the squaraine dye should be given.
2) Reporting of imaging parameters and timing of recordings is insufficient, making it either impossible to gauge the true merits or rendering statements misleading. E.g. the section on STED imaging in the Methods section states that each frame was obtained at 1 s and the image series were obtained at a 1-minute interval but it is not clear to which measurements this refers. Precise timing information is warranted in each image caption and detailed imaging parameters should be stated for each measurement. Similarly, all other experiments should be described in sufficient detail to reproduce them. 3) In the abstract the authors state that live imaging over 50 minutess is possible. This is a meaningless statement if the actual time it takes for one frame and the intermittent (resting) times are not stated. As far as I understand, this statement refers to Suppl. Video 4, where each frame took about 4 seconds and then a recovery period of about 70 seconds was used, yielding 40 images in total. The authors should rather state the number of images that were taken. 4) The authors claim a resolution of 35.2 nm. This statement rests on a single line profile shown and a histogram without any detail on how data was analysed. From the supplementary videos and the image material provided, I am very doubtful that signal-to-noise ratio is sufficient to claim 35 nm resolution or even 35.2 nm. The authors are encouraged to refer to the relationship of resolution and signal-to-noise ratio in STED microscopy, discussed e.g. by the Vicidomini group (Tortarolo et al., Evaluating image resolution in stimulated emission depletion microscopy. Optica, 5, p32 (2018)). Also the resolution determination in Fig. 2c is problematic due to insufficient signal-to-noise ratio. From most of the images provided, I find it hard to clearly delineate individual cristae. In comparison, cristae stand out very clearly in the Scientific Reports paper from the Jakobs group. 5) I am somewhat concerned with the specificity of labeling for mitochondria. In Suppl. Fig. 1, there seems to be strong correlation with ER, which is, however, mentioned in the main text. 6) p5: The description of Suppl. Fig. 6 in the main text implies that no toxicity arises during 1 h of STED imaging. However, in Suppl. Fig. 6 it is not clear whether STED imaging has been applied in addition to the incubation. It is not described what the viability assay is. 7) p 5: The statement on saturation intensity is contradictory. This measurement is not described properly. From the data provided it is not clear that this measurement was done in a way that excludes e.g. the possibility that bleaching might influence the measurement outcome. While superior imaging performance over Atto647N would be an achievement, this statement would need to be supported by further data. 8) p5: I find the wording on 3D imaging somewhat misleading. The authors apply only resolution enhancement in the focal plane, not along the optical axis. Taking a stack of 5 STED images with xy resolution enhancement and confocal resolution axially is not particularly challenging. Therefore, also the statement in the conclusion "STED imaging of 3D stacks can reveal the ultrastructure of mitochondria in live cells" is not justified for the measurements presented in the manuscript. 9) p7: The authors observe formation of bubbles and fission of mitochondria in long-term STED imaging. It should be checked that these dynamics are not related to the photo-burden inflicted by the imaging. 10) Suppl. Note 3: No data is shown to back up the claims that the squaraine dye has wide applicability. 11) The Supplements, in particular Suppl. note 4 seem to be an unedited draft version with a series of problematic statements. For example, the statement on acquisition speed is not tenable in this unspecified way in view of the work by Schneider et al., Nature Methods 12, p827 (2015) where 125 frames per second super-resolution imaging rate was achieved. 12) Suppl. Fig. 5: If MitoESq-635 is covalently linked to inner membrane proteins, it is no surprise that it stays bound also with prolonged imaging. This does, however, not imply that the mitochondrial membrane potential is till intact.
Minor points: 1) p2: Wording is unfortunate: "ATP, which is the only energy form that cells can absorb". ATP is certainly the most prevalent energy carrier, but it's not absorbed by the cells but rather synthetized from ADP. Also there are other high-energy bonds in cells that can be used to drive biochemical processes. 2) p 2: I am not sure that mitochondrial dynamics "regulate" cancer cell activity. 3) p2: The statement on the sampling theorem is unfortunate. Resolution is defined as the capability to discern two nearby structures. 4) p2: There is an inconsistency with the author name in the citation Dannie et al.. 5) Several of the citations are incomplete. 6) p3: It would be helpful to include a statement on the absorption wavelength range of OPA1. 7) p3: "Stimulated emission" is not a laser process. "Laser" means "light amplification by stimulated emission of radiation". Stimulated emission is an important component of the laser process, but not all stimulated emission is equivalent to lasing. 8) p3: "…the efficiency is generally too low to generate effective STED processing." This sentence needs rewording. 9) p4: I am not sure whether the authors want to state that they label the "plasma membrane". 10) Fig 1c: The authors plot power, not "intensity", as stated in the figure caption. 11) In Suppl. Fig. 1, the last column is not adequately described. 12) p 8: What do the authors precisely mean by "melting" of dye molecules? 13) Fig. 4: What is a "skeletal image"? 14) The data in Suppl. Fig. 4 is consistent with covalent labeling, but not a "proof". 15) Suppl. note 1, line 66: "literature procedure" would warrant a citation.
Reviewer #3 (Remarks to the Author): The paper titled "Mitochondrial dynamics quantitatively revealed by STED nanoscopy with an enhanced squaraine variant probe" reported a new fluorescence living cell probe，MitoESq 635. It is an excellent candidate for live cell STED nanoscopy, which will make real impacts on pushing super resolution microscopy towards practical live cell STED imaging. The manuscript is well written and timely. With much better performance over the existing mitochondrial probes, the authors obtained beautiful super resolution images of mitochondrial ultrastructure in 3D and long term (about 1 hour) using STED nanoscopy. They got mitochondrial cristae dynamical videos with a decent resolution of sub-50 nm for 1 hour for the first time in super resolution microscopy and mitochondrial research field. I have several minor questions for the authors to address before it is published on Nature C ommunications. 1. In figure 1b, I suggest the authors to add the labels of excitation and STED wavelengths. 2. In figure 2c, figure 3c and figure 4d, the details of how the data points were fitted should be described. 3. In Supplementary Figure 9, all the scale bars on fluorescence images were missing. 4. Although this probe worked pretty well with STED nanoscopy, the performance of this probe in live cell STED could be further improved if it can be combined with recent new STED technologies such as RESC UE STED, MINFIELD, DyMIN, STEDD, et al. C an the authors discuss about this？ Specific Issues: (1) The section describing how the Mito-ES1-635 dye labels mitochondria is confusing.
The authors state that it 'labels the plasma membrane with high density'. Then the authors say that the signal is 'concentrated in the eccentric perinuclear regions', but I'm not sure what they mean by these statements and how they are relevant to mitochondrial staining. Clearly the dye does co-localize to mitochondria based on the images shown, but the description in the text is not clear.

Response:
Thanks for the reviewer's recognition on the importance of this research and our results. We are sorry for the unclear descriptions in the main text. We have removed the sentence "The fluorescence signals are concentrated in eccentric perinuclear regions, whereas little signal is presented in cytoplasm and nucleus regions." As shown in the new Supplementary Figure 1  (2) The authors claim that the Mito-ESq-635 dye is covalently binding with mitochondrial VDPs. However, in Sup Fig. 3, they show that MitoESq-635 binds to TRX-1, a cytosolic protein. Based on the evidence provided, it is unclear to me that the dye is in fact binding specifically mitochondrial membrane proteins, which I think is what the authors are trying to say.

Response:
Thanks for your comments here. In Supplementary Fig. 3, we performed SDS-PAGE experiment to prove the probe Mito-ESq-635 can covalently bind to the proteins with two vicinal thiol residues (VDPs), as there is a phenylarsenate moiety in the probe. Thioredoxin-2 (Trx-2) is known as one of typical VDPs mainly distributed in mitochondria . [1.2] As shown in Sup. Fig 3, we identified that the probe can selectively bind to the VDPs, indicating that it is readily to label the VDPs inside cells. The small molecule structure makes Mito-ESq-635 very easy to penetrate the cell membrane by transmembrane interaction and reach to mitochondria in short time (less than 5 minute) because of the plus charge and suitable hydrophilicity-hydrophobicity itself. Our image-based colocalization analysis (Supplementary Figure 1) also indicates that the MitoESQ-635 labels the mitochondria for a variety of cell species, with high R ratio with the commonly used Mito Tracker. We have modified the description in the main text and supporting information accordingly.
(3) I'm not sure that I agree with the authors that there is no phototoxicity. They describe changes in the mitochondria, 'bubble structures', with cristae growing inside, which I would interpret as swelling of the mitochondria. In particular, the changes in fig 3 between the first frame and the final frame look like swelling and fragmentation of the mitochondria, which is you would expect to see with phototoxicity. If this was the 'normal' state of mitochondria, it should be evident from the start of the imaging, not just later on.

Response:
We are sorry for our previous overstatement on non-phototoxicity. After careful investigation, we agree with the reviewer that there exists some phototoxicity during long time imaging process in Fig. 3. The case in Fig. 3 is a demonstration of light induced cell unhealthy, including swelling and fragmentation of the mitochondria, but with less significant fluorescence signal degradation.
During STED fluorescence nanoscopy, because of the power requirement for the donut beam, the balance between the phototoxicity (cell viability and organelle healthy) and the photodamage to the fluorescence dye is much more challenging. Lots of efforts have been spent to meet those issues, including improved STED super resolution techniques and novel dye developments [1][2][3][4][5][6][7]. Especially, it lacks the dyes of better performance for living cell STED imaging. During the review process of our manuscript, another group reported a photostable fluorescent marker for super resolution STED live imaging, termed MitoPB Yellow (λex = 488 nm; λSTED = 660 nm), in which the swelling phenomenon is also observed [8]. For box plot of width of mitochondria, every frame has 6 data points, which means 6 mitochondria in the frames are analyzed. Their widths are detected by using intensity profile in ImageJ. The figure is plot using Origin software, Plot-Statistics-Box Chart. For fluorescence intensity, using the same original images data with width mitochondria, choose 6 squares (10×10 pixels) in the frames. In ImageJ, Images -Stacks -Measure Stacks, then the total intensity of every square (10×10 pixels) can be obtained in all frames. In every frame, we have 6 intensity data (6 squares), calculated the mean value intensity of every frame. For resolution, the intensity profile is obtained by using ImageJ, and Gaussian Cumulative Fit Peak is processed with Origin software. (4) This swelling also appears evident in Fig 4, as the width of the mitochondria appears to increase significantly over time, and not just due to fusion. While it appears that they can in fact catch fusion events, due to the swelling, I'm not sure how well the cristae dynamics they observe reflect what normally happens during fusion.

Response:
We agree with the reviewer's comment. The swelling may influence the cristae dynamics together with other environmental conditions (such as temperature, pressure, pH, and so on). Here we tried to reduce this influence by using short frame time (3.9s per frame) and long dark recovery time (71.5 sec) to avoid light induced swelling.
In order to investigate the influence on the mitochondrial swelling induced by light exposure, we have added a new experiment to quantitatively analyse the width of the mitochondria over time, under different STED beam powers, as shown in the revised Figure 3. It shows the width and fluorescence intensity change faster when STED intensity is higher. The 775 nm STED power of 7-9 mW should be an optimized intensity in our case with relatively little photobleaching and photodamage. To further improve the resolution, STED intensity should be increased, in which case a longer dark recovery can help to reduce photobleaching and photodamage.
The long-term STED imaging of mitochondria can separate the phototoxicity and photodamage, enabling us to visualize photo-induced fusion events of mitochondria at high spatial resolution.
(5) In order to accurately compare the Mito-ESq-635 dye to other dyes used to image mitochondrial cristae, they should show their own data using their scope and settings, rather than pulling numbers from other publications, as settings and laser power are often unique to each microscope. Response: We agree with the reviewer. In figure 1, we compared our results with the results by EM from other publications which can be considered as ground truth of the shape of mitochondria cristae. In order to accurately compare the Mito-ESq-635 dye to other dyes used to image mitochondrial cristae in live cells at molecule level resolution, we compared our results  General comments: (1) I would completely remove the first paragraph, which I feel is unnecessary. The evolutionary history of mitochondria and their role in cancer are not really relevant to the current work.
Response: Thanks for the suggestion. We have revised the first paragraph in the revision.
(2) Though mutations in OPA1 cause an optic atrophy phenotype, OPA1 does not encode a photoreceptor, and this has nothing to do with sensitivity to light or phototoxicity.
Response: Thanks for your kind suggestion. As also suggested by Reviewer 2, we have deleted the description of OPA1 in the revision.
(3) The authors need to include a bit more information regarding the deconvolution analysis they are applying to their images.
Response: Thank you for your suggestion. We use the Huygens Software for deconvolution analysis directly. Images are taken using Leica TCS SP8 STED 3X system, then uploaded to Huygens Software. Deconvolution process is completed by using auto setting in Huygens Software embedded in the system. (4) Regarding the average widths of cristae, they use 42 nm and 43 nm for the width following fission. I assume this should be the same number.
Response: We apologize for our typo. It should be 42 nm and we have revised it in the revision.
(5) The authors state 'the fluorescent molecule could melt', with high laser power. Not sure that 'melt' it the correct term here.
Response: Thank you for your suggestion. Here we wish to express the phenomenon of burning-like signal bursting, as shown in Supplementary Figure 8. This is typically caused by excessive heat deposit due to the high STED laser power. We have hence changed "melting" into "bursting". Response: We agree that co-localization is helpful to illustrate correct labelling. Figure  R1 is co-localization in U2OS, and figure R2 is co-localization in HeLa, figure R3 is co-localization in SH-SY5Y and SKOV cells.

Supplementary
(7) Generally speaking, many of the images are too small to adequately see at 100% magnification.
Response: Thanks for the suggestion. We have enlarged all the small images, such as Figure 4.

[REDACTED]
Response: Thanks for the suggestion. We added the white dashed boxed in supplementary video 8 (video 6 in old version) in the revision as you can see below.
The Supplementary video 5 is deleted because we revised Figure 3 and Figure 4 in main text.
SI Video 8 Raw data of Figure 4h: skeleton image of mitochondrial revealed cristae dynamics during fusion quantitatively. Scale bar, 5 μm.

Reviewer #2 (Remarks to the Author):
The manuscript by Yang and coworkers addresses a timely need in super-resolution imaging, namely the generation of live-cell imaging compatible probes to follow mitochondrial dynamics. The topic per se is thus relevant and interesting to a wide readership. A very similar paper appeared in Proceedings of the National Academy of Sciences earlier this year (Wang et al., PNAS 116, p15817 (2019)). In that paper, also a mitochondria-selective dye was presented for STED imaging, allowing observation of cristae dynamics. Therefore, claims of novelty such as "for the first time" in the abstract and elsewhere are problematic and that prior work should be acknowledged. Imaging in the far red as in the present manuscript (STED laser at 775 nm) is desirable to reduce interactions with cellular constituents. In terms of imaging performance, I would rate the PNAS paper higher (albeit at 660 nm STED wavelength) and I would also rate a recent paper by the Jakobs group higher in terms of imaging quality, using SNAP tag and silicon rhodamine for depletion at 775 nm (Stephan et al., Sci Rep 640920 (2019)). Therefore, the specific merit of the manuscript is to provide a label for STED microscopy at 775 nm which has affinity for mitochondria.
However, the manuscript has several shortcomings, such that I cannot recommend it for publication in its present state. Several of the claims are not sufficiently supported by the quality of the data. In other instances, the information provided is insufficient to truly assess the validity of the measurements.

Response:
Thanks for the reviewer's recognition on the importance of this research and the specific merit of our new live cell Mitochondrial probe for STED nanoscopy. Indeed, a struggle against photodamage as well as photobleaching, blinking and saturation is always there if using fluorescence as contrast agents. Because of the power requirement for donut beam, the struggle against those flaws in STED fluorescence nanoscopy is much more challenging. Lots of efforts have been spent to meet those issues, including improved STED super resolution instruments and novel probes developments [1][2][3][4][5][6][7]. Especially, the dyes of better performance for living cell STED imaging are lacking. During the review process of our manuscript, as the reviewer 2 mentioned, Shigehiro Yamaguchi group and Stefan Jakobs group released their works with quite nice results. We have added the two references accordingly. In our revision, we have performed several new experiments, and the images/ videos and quantitative analyses are updated.
For the MitoPB Yellow (λex = 488 nm; λSTED = 660 nm) is real a photostable fluorescent marker for super resolution STED live imaging [8] . , the STED power (108 mW-270 mW) in during their video recording is real high which is not so nice for cell imaging. Also, most of the images/videos shown in their paper are after deconvolution which will let the images looks much sharper and clearer but with artifacts. However, most of the images and all the videos in Scientific report paper and our update manuscript are raw data.
For the COX8A-SNAP fusion proteins, real beautiful images and video shown in their paper but limited frame number/ imaging periods. Also, note that the OX8A-SNAP fusion proteins need quite complex operating steps. However, our probe offers the option of imaging mitochondria in living cells using STED nanoscopy without the necessity of an additional tagging step. Taking higher photostability and a lower saturation intensity, our MitoESq-635 probe revealed quite long term (200 frames, 600 s, see figure 3 and Supplementary Video 5) and high resolution (down to ~35 nm) live mitochondrial imaging under low level STED power. The supplementary table 2 compared our probes with the two side by side.
(2)MitoESq-635 is compatible with the live cell, with low photo-toxicity.
(3)MitoESq-635 with a phenylarsenicate moiety is able to covalently bind to VDPs, make sure the simple cell incubation.
(4)Fluorescent enhancement upon binding to mitochondria, reducing background signals.
Please find my detailed assessment below.
Major points: (1) Absorption coefficient, quantum yield, fluorescence lifetime etc. for the squaraine dye should be given. Response: Thanks for the reviewer's suggestions. We have investigated the fluorescence quantum yields of MitoEsq-635 and fluorescence lifetime in solution and live ells, respectively. Firstly, we measured the relative fluorescence quantum yield and molar extinction coefficient of Mito-ESq-635 in different solvent (see Table 1 and Table 2).
Absorption and Fluorescence spectra were measured with a UV/Vis absorption spectrometer (GBC Cintra 2020, Australia) and a fluorescence spectrometer (Horiba iHR320, American), respectively. The relative fluorescence quantum yields were determined with Rhodamine B as a standard and calculated using the following equation: where A is absorbance at the excitation wavelength; I is the integration of the emission spectra; Φs represents the fluorescence quantum yield of the reference standard (Rhodamine B Φs = 0.97 in ethanol solvents) [2] ; λex is the excitation wavelength; n is the refractive index of the solution (because of the low concentrations of the solutions (10 -7 -10 -8 mol/L), the refractive indices of the solutions were replaced with those of the solvents); and the subscripts x and s refer to the unknown and the standard, respectively.

Absorption coefficient:
, where A is absorbance at the excitation wavelength; [c] is the molar concentration. Note that the fluorescence quantum yield of MitoESq-635 is lower in very polar solvents (ethanol) than those in less polar solvents (DCM, DMSO), in particular, in water it is difficult to calculate the integral of the fluorescence (Fx), probably due to the high polarity of water. We tried to measure fluorescence lifetime of the probe in PBS solution instead.  The relative fluorescence quantum yields were measured on a Horiba fluorescent spectrometer with he method reported by H.H. Tønnesen [4] . And the fluorescence lifetime was measured on a commercially available equipment (DCS-120 timecorrelated single photon counting equipment (2) Reporting of imaging parameters and timing of recordings is insufficient, making it either impossible to gauge the true merits or rendering statements misleading. E.g. the section on STED imaging in the Methods section states that each frame was obtained at 1 s and the image series were obtained at a 1-minute interval but it is not clear to which measurements this refers. Precise timing information is warranted in each image caption and detailed imaging parameters should be stated for each measurement.
Similarly, all other experiments should be described in sufficient detail to reproduce them. Response: We agree with the reviewer the sufficient detail is needed to reproduce them.
We have put all the parameters of images and videos in Supplementary table 1.  (2018)). Also the resolution determination in Fig. 2c is problematic due to insufficient signal-to-noise ratio. From most of the images provided, I find it hard to clearly delineate individual cristae. In comparison, cristae stand out very clearly in the Scientific Reports paper from the Jakobs group.
Response: We thank the reviewer for bringing up the important literature, and the FRC method. Here we have used FRC-based resolution metrics to evaluate the resolution, as shown in Supplementary Figure 10. The FRC solution is ~43nm which is close to intensity profile resolution 35.2nm. (5) I am somewhat concerned with the specificity of labeling for mitochondria. In Suppl. Fig. 1, there seems to be strong correlation with ER, which is, however, mentioned in the main text. Response: We have performed a new set of experiments for the correlation between Mito-ESQ-635 and Mito Tracker. As can be seen from the Supplementary Figure 1, a Pearson Coefficient of R > 0.8 can be obtained for all the four different cell lines, indicating an excellent colocalization between our dye and mitochondria.

Supplementary
The previous result may be due to the overdosed concentration of the dye, which also targets the ER network. (6) p5: The description of Suppl. Fig. 6 in the main text implies that no toxicity arises during 1 h of STED imaging. However, in Suppl. Fig. 6 it is not clear whether STED imaging has been applied in addition to the incubation. It is not described what the viability assay is. Response: In Suppl. Fig. 6, we mainly checked the cytotoxicity of the probe itself to live cells, chemical toxicity of fluorescent probe. From the result, it can be found that the fluorescent probe (Mito-ESq635) is safe to live cells (almost no chemical toxicity to live cells) under out experimental conditions. We did not use the STED laser to illuminate the cell sample in this case.
(7) p 5: The statement on saturation intensity is contradictory. This measurement is not described properly. From the data provided it is not clear that this measurement was done in a way that excludes e.g. the possibility that bleaching might influence the measurement outcome. While superior imaging performance over Atto647N would be an achievement, this statement would need to be supported by further data. Response: We apologize for our clerical error. Here we changed the original text 'higher' into 'lower'.
Original: 'Yet, the squaraine-STED dye presents a lower saturation intensity, 0.893 mW, which is ~3.4-fold higher than that of ATTO 647N (Figure 1).' Revised: 'Yet, the squaraine-STED dye presents a lower saturation intensity, 0.893 mW, which is ~3.4-fold lower than that of ATTO 647N (Figure 1).' We agree that the measurement may be influenced by bleaching. In Figure 1, when STED power is small, the effect of bleaching is small. When STED power becomes bigger, bleaching may influence fluorescence intensity. To be honest we didn't exclude the possibility that bleaching might influence the measurement outcome. But we believe these data can reflect the relationship between the saturation intensity of the squaraine-STED dye and ATTO 647N that squaraine-STED dye's saturation intensity is lower than ATTO 647N's.
(8) p5: I find the wording on 3D imaging somewhat misleading. The authors apply only resolution enhancement in the focal plane, not along the optical axis. Taking a stack of 5 STED images with xy resolution enhancement and confocal resolution axially is not particularly challenging. Therefore, also the statement in the conclusion "STED imaging of 3D stacks can reveal the ultrastructure of mitochondria in live cells" is not justified for the measurements presented in the manuscript.
Response: We agree that we apply only resolution enhancement in the focal plane, not along the optical axis. This is because we need to balance between xy resolution and optical resolution when fixing the power of STED beam. Comparing with the recent publications of STED imaging of mitochondria in live cell, we are currently the only group that can demonstrate STED in a 3D z-stack, benefitted from the photostability of our Mito-ESQ-635 dye. We have made the following revision: Original text: "STED imaging of 3D stacks can reveal the ultrastructure of mitochondria in live cells".
Revised text: "STED imaging of 3D z-stacks can reveal the ultrastructure and spatial organization of mitochondria in live cells in different depths." (9) p7: The authors observe formation of bubbles and fission of mitochondria in longterm STED imaging. It should be checked that these dynamics are not related to the photo-burden inflicted by the imaging.
Response: We agree with the reviewer that there exists some phototoxicity during long time imaging process in Fig. 3. The case in Fig. 3 is a demonstration of light induced cell unhealthy, including swelling and fragmentation of the mitochondria, but with less significant fluorescence signal degradation.
During STED fluorescence nanoscopy, because of the power requirement for the donut beam, the balance between the phototoxicity (cell viability and organelle healthy) and the photodamage to the fluorescence dye is much more challenging. Lots of efforts have been spent to meet those issues, including improved STED super resolution techniques and novel dye developments [1][2][3][4][5][6][7]. Especially, it lacks the dyes of better performance for living cell STED imaging. During the review process of our manuscript, another group reported a photostable fluorescent marker for super resolution STED live imaging, termed MitoPB Yellow (λex = 488 nm; λSTED = 660 nm), in which the swelling phenomenon is also observed [8]. In the meantime, we developed an enhanced squaraine variant dye (MitoESq-635), to study the dynamic structures of mitochondrial cristae in live cells. The low saturation intensity and high photostability make MitoESq-635 ideal for long-term, high-resolution STED nanoscopy for live cell imaging.
In order to investigate the parameters affecting the resolution, brightness, the width of the mitochondria in living cells, we performed quantitative analysis on long term live STED imaging of mitochondria (See figure 3, and Supplementary Video 5) For box plot of width of mitochondria, every frame has 6 data points, which means 6 mitochondria in the frames are analyzed. Their widths are detected by using intensity profile in ImageJ. The figure is plot using Origin software, Plot-Statistics-Box Chart. For fluorescence intensity, using the same original images data with width mitochondria,   Frame # choose 6 squares (10×10 pixels) in the frames. In ImageJ, Images -Stacks -Measure Stacks, then the total intensity of every square (10×10 pixels) can be obtained in all frames. In every frame, we have 6 intensity data (6 squares), calculated the mean value intensity of every frame. For resolution, the intensity profile is obtained by using ImageJ, and Gaussian Cumulative Fit Peak is processed with Origin software. (10) Suppl. Note 3: No data is shown to back up the claims that the squaraine dye has wide applicability. Response: We agree that this statement should be supported by further data. As the fluorescent MitoESq-635 reported in this work is a kind of new fluorophores derived from Squaraine dyes, rare work has been reported on the same fluorophore, especially used for super-resolution imaging. However, there are several works on traditional Squaraine dyes and its derivatives applied in biological imaging listed as follows: We have added the references in our revision.

Supplementary
(11) The Supplements, in particular Suppl. note 4 seem to be an unedited draft version with a series of problematic statements. For example, the statement on acquisition speed is not tenable in this unspecified way in view of the work by Schneider et al., Nature Methods 12, p827 (2015) where 125 frames per second super-resolution imaging rate was achieved. Response: Thanks for your reminding and apologize for the problematic statements. We revised the Suppl. Note 4 by deleting the statement "Also, the acquisition speed of these methods (fourth column) is far slower than the biological dynamics in living cells, so it is difficult to get motion-induced artifacts, which means no single parameter can be optimized without compromising the others to technical consideration." (12) Suppl. Fig. 5: If MitoESq-635 is covalently linked to inner membrane proteins, it is no surprise that it stays bound also with prolonged imaging. This does, however, not imply that the mitochondrial membrane potential is still intact.
Response: We thank the reviewer for the constructive question. To address this, we performed a new colocalization experiment with the probe and Mitotracker green. The cell was fixed with 4% PFA after incubation of SKOV cells with the probe. As showed in Figure R1, the probe showed well colocatization imaging with the Mitotracker Green, which means the probe covalently bound to the mitochondrial membrane VDPs. In the previous experiment, we used membrane potential dependent Mitotracker (Rho123) to colocalize with the probe in live cells, the mitochondrial membrane potential did not change a lot as the Rho123 showed well mitochondrial imaging. With scanning of strong laser, it can be clearly found that Rho123 slowly diffused out of mitochondria due to the change of the membrane potential, whereas MitoESQ-635 is still retained on mitochondria without marked diffusion. These results proved that the covalent binding of the probe with mitochondria, and the change in the mitochondrial membrane potential by the probe is much smaller than that of Rho 123 ( Figure R2).  Minor points: 1) p2: Wording is unfortunate: "ATP, which is the only energy form that cells can absorb". ATP is certainly the most prevalent energy carrier, but it's not absorbed by the cells but rather synthetized from ADP. Also there are other high-energy bonds in cells that can be used to drive biochemical processes. Response: We apologize for the wrong statements here. We agree with that ATP is not absorbed by the cells and other high-energy bonds in cells can be used to drive biochemical process. As the first reviewer point out that first paragraph is relevant to the current work, we have deleted it in the revised txt.
2) p 2: I am not sure that mitochondrial dynamics "regulate" cancer cell activity.
Response: We thank the reviewer for pointing this out. We have revised the text.
Original text: "Its plasticity and structural dynamics can regulate a series of cancerous cell functions such as proliferation, migration, and resistance to therapy." Revised text: "Their various structural characteristics can reflect a number of cell activities, such as proliferation, migration, and resistance to therapy." 3) p2: The statement on the sampling theorem is unfortunate. Resolution is defined as the capability to discern two nearby structures. Response: We agree that resolution is defined as the capability to discern two nearby structures. To be honest, according to the sampling theorem, the pixel step should smaller than 50 nm to achieve 100nm resolution. We have revised the statement on the sampling theorem in the revised version.
Origin: "Considering the Nyquist-Shannon sampling theorem, one needs ~50 nm resolution to visualize the gaps between mitochondria cristae." Revised: ", which means imaging with resolution far below 100 nm is necessary to visualize the gaps between mitochondria cristae." 4) p2: There is an inconsistency with the author name in the citation Dannie et al.
Response: Thanks for reminding. We corrected the author name in the revised version. 6) p3: It would be helpful to include a statement on the absorption wavelength range of OPA1. Response: Thanks for reminding. Based on the first reviewer's comment, we have deleted the description of OPA1. 7) p3: "Stimulated emission" is not a laser process. "Laser" means "light amplification by stimulated emission of radiation". Stimulated emission is an important component of the laser process, but not all stimulated emission is equivalent to lasing. Response: We agree with the reviewer. We have deleted this description "(laser process)" for clarity.
Original: "the fluorescence in the "donut" area should be converted to stimulated emission (laser process) through high power laser illumination" Revised: "the fluorescence in the "donut" area should be converted to stimulated emission through high power laser illumination" 8) p3: "…the efficiency is generally too low to generate effective STED processing." This sentence needs rewording. Response: We thank the reviewer for pointing this out. We have revised it in the revised text.
Original: "they are not photostable enough to endure long-term STED imaging, and the efficiency is generally too low to generate effective STED process" Revised: "they are not photostable enough to endure long-term STED imaging at high resolution". 9) p4: I am not sure whether the authors want to state that they label the "plasma membrane". Response: Yes, it is exactly what we mean. We agree with that there is some correlation with ER when the dye is in high density. We hope to use this one dye to see both mitochondria and ER at the same time in living cell. We want to distinguish them by using a program. To be honest we are still research on it. One idea is distinguishing them by different intensity information between them. In other words, mitochondria are brighter than ER, we want to use this information to distinguish them. Another idea we are thinking about is using machine learning to distinguish them when we have enough training data.
10) Fig 1c: The authors plot power, not "intensity", as stated in the figure caption. Response: Sorry for our wrong words. We have revised the word 'power' in the figure into 'intensity' in the revised version. 11) In Suppl. Fig. 1, the last column is not adequately described. Response: Thank you for your reminder. We added the description about Suppl. Fig 1 in the new version: "the last column is 2D intensity histogram, and R is Pearson's R value. This analysis is performed with Fiji (ImageJ) plugin Coloc 2. The results show that for all the cell lines, the Pearson colocalization coefficients are greater than 0.8." 12) p 8: What do the authors precisely mean by "melting" of dye molecules? Response: Thank you for your suggestion. Here we wish to express the phenomenon of burning-like signal bursting, as shown in Supplementary Figure 8. This is typically caused by excessive heat deposit due to the high STED laser power. We have hence changed "melting" into "bursting". Confocal STED

[REDACTED]
Response: Thanks for suggestion. We added explain for skeletal image in Methods and materials. You can also find it as below.
Skeleton connectivity quantitative analysis is similar to our previous work [32] Skeletonize3D (download from http://imagej.net/Skeletonize3D), a Fiji plugin, was utilized to perform the skeletonization of mitochondria, and their geometrical and topological features of the original structure were obtained [33] .Thereby the lengths and numbers of mitochondrial structural skeletons were counted for the morphological binary images.
14) The data in Suppl. Fig. 4 is consistent with covalent labeling, but not a "proof". Response: Thanks for reminding. We agree that Suppl. Fig. 4 is consistent with covalent labeling, but not a "proof". We have revised it in the revised text.

Reviewer #3 (Remarks to the Author):
The paper titled "Mitochondrial dynamics quantitatively revealed by STED nanoscopy with an enhanced squaraine variant probe" reported a new fluorescence living cell probe，MitoESq-635. It is an excellent candidate for live cell STED nanoscopy, which will make real impacts on pushing super resolution microscopy towards practical live cell STED imaging. The manuscript is well written and timely. With much better performance over the existing mitochondrial probes, the authors obtained beautiful super resolution images of mitochondrial ultrastructure in 3D and long term (about 1 hour) using STED nanoscopy. They got mitochondrial cristae dynamical videos with a decent resolution of sub-50 nm for 1 hour for the first time in super resolution microscopy and mitochondrial research field. I have several minor questions for the authors to address before it is published on Nature Communications.
1. In figure 1b, I suggest the authors to add the labels of excitation and STED wavelengths. Response: Thanks for your reminding. We added them in figure 1b in the revised version. 2. In figure 2c, figure 3c and figure 4d, the details of how the data points were fitted should be described.
Response: We have added descriptions of the details data processing in Figure 3 and Figure 4 in Methods and materials: Data processing. [REDACTED] In Figure 3, for box plot of width of mitochondria, every frame has 6 data points, which means 6 mitochondria in the frames are analyzed. Their widths are detected by using intensity profile in Fiji. The figure is plot using Origin software, Plot-Statistics-Box Chart. For fluorescence intensity, using the same original images data with width mitochondria, choose 6 squares (10 pixels x 10 pixels) in the frames. In Fiji, Images -Stacks -Measure Stacks, then the total intensity of every square (10 pixels x 10 pixels) can be obtained in all frames. In every frame, we have 6 intensity data (6 squares), calculated the mean value intensity of every frame. To normalize the intensity, all intensities are divided by the mean value of the intensity first frame. Then calculated the mean value and variance of intensity in every frame by using Statistics on Rows in Origin software. At last plot Y errors. For resolution, the intensity is obtained by using Fiji, and Gaussian Cumulative Fit Peak is processed with Origin software. In Figure 4, data fitting is finished by using Gaussian Cumulative Fit Peak tool in Origin software.
3. In Supplementary Figure 9, all the scale bars on fluorescence images were missing. Response: Sorry for the missing scale bar. We have added them in the revision. You can also find it as below.
The comparison of ultra-fine structure of mitochondria were resolved by STED and EM [1,2]. Scale bar, black 500 nm, white, 1 µm. [REDACTED] 4. Although this probe worked pretty well with STED nanoscopy, the performance of this probe in live cell STED could be further improved if it can be combined with recent new STED technologies such as RESCUE STED, MINFIELD, DyMIN, STEDD, et al.
Can the authors discuss about this？ Response: We agree that combined with new STED technologies such as RESCUE STED, MINFIELD, DyMIN, STEDD is helpful in living cell image.
For RESCUE STED, excitation and STED are shut off at positions where no signal arises. In conventional STED these places are pre-stressed and pre-recorded. Thus, RESCUE STED can reduce photobleaching in STED. Combined with RESCUE STED and this probe, should be beneficial to living cell image. [R1] MINFIELD reduces bleaching in STED/RESOLFT nanoscopy through restricting the scanning to subdiffraction-sized regions. By safeguarding the molecules from the intensity of the maxima and exposing them only to the lower intensities (around the minima) needed for the off-switching (on-switching), MINFIELD largely avoids detrimental transitions to higher molecular states. A bleaching reduction by up to 100fold is demonstrated. Combined with MINFIELD STED and this probe, bleaching can be reduced even more. [R2, R3] The principle of DyMIN is that it only uses as much on/off-switching light as needed to image at the desired resolution. Fluorescence can be recorded at those positions where fluorophores are found within a subresolution neighborhood. DyMIN is shown to lower the dose of STED light on the scanned region up to ∼20-fold under common biological imaging conditions, and >100-fold for sparser 2D and 3D samples. Combined with MINFIELD STED and this probe, bleaching can be reduced even more as well. [R4] Stimulated emission double depletion (STEDD) as a method to selectively remove artificial background intensity. This method is beneficial when considering lowerpower, less redshifted depletion pulses. By combining STEED and this probe, we can keep balance in both better resolution and less bleaching for longer imaging. This probe is also beneficial to STEED.
To sum up, combined with those new STED technologies above with this probe is helpful in living cell image.
Overall, the data and images in the manuscript show that the MitoESq-635 dye is a valuable tool for imaging mitochondria via STED microscopy. Yang et al. have made several key changes that help improve the manuscript. However, while the response the reviews was quite thorough, I did not feel the revised version completely addressed all of the key points discussed. Moreover, in my opinion some of the claims in the paper are still overstated. For example, there is clearly photobleaching which occurs over time, and there is clearly phototoxicity with the mitochondrial morphology and cristae structure changing over time. This will likely be true for any dye and does not detract from the paper, but to claim otherwise is misleading. What is valuable is the direct side by side comparison between the new dye and existing options, which clearly demonstrate the superiority of MitoEsq-635 (e.g. Sup fig 7). Perhaps some of this data could be moved to data in the main paper rather than supplemental.
Specific Issues: Additional details for the imaging over 50 minutes are required in the abstract.
Be careful about the specificity of various statements. The following is not true: "conventional optical microscopy techniques are insufficient to visualize the structure and dynamics of mitochondria" Generally speaking, researchers have been using conventional microscopy for nearly two decades to study mitochondrial structure and dynamics. However, if you are referring to sub-mitochondrial structures (cristae) being imaged live, the sentence would be more appropriate.
Not sure how this sentence fits in the context of a paragraph on microscopy "The mitochondrial intermembrane space can be measured with indirect approaches such as proteomic mapping" The following statement is still false: "Mitochondria are more sensitive to light than other cellular organelles due to the existence of photoreceptors" Mitochondria do not have photoreceptors. Photoreceptors are a cell type, which have mitochondria. Furthermore, this is irrelevant to mitochondrial sensitivity to light.
The text is still unclear with reference to the dye labeling the plasma membrane. The plasma membrane is the membrane at the exterior of the cell and is distinct from the mitochondria and mitochondrial membranes. Thus, it is unclear how the dye is binding to the plasma membrane and labelling mitochondria. Perhaps the authors mean the dye can bind membranes generally, rather than the plasma membrane specifically?
The following sentence is poorly constructed: "Benefitted from its low saturation intensity, the MitoESq635 probe exhibited no marked toxicity in HeLa cells at a concentration of 1 μM after 1 hour of incubation during STED superresolution imaging microscopy". However, the bigger issue is that details on the imaging conditions are still absent, making the statement useless.
While the following statement does have some key information, it is still incomplete: "With MitoESq-635, rare photobleaching and mitochondrial shape variations are observed upon exposure to STED scanning for over 10 minutes (1.2 s per frame, 700 × 700 pixels, 12.6 µm × 12.6 µm, STED beam of 30.2 mW at 775 nm), as shown in Supplementary Figure 7". The reader needs to know how many images were captured over the 10 minutes. Is this one image every minute (e.g. a 1 minute recovery), or one image every 1.2 seconds (e.g. constant imaging).
In There are still several small typos/formatting issues throughout that required correction. For example, "standard gold live cell" should be "gold standard live cell" Reviewer #2 (Remarks to the Author): The authors have made improvements to the manuscript. However, they have glossed over several of the reviewers' comments and have given explanations to reviewers without fully addressing the actual issues in the text. The paper is still not exact enough in many aspects, especially concerning imaging. I am still reluctant to recommend publication.
Please find some specific points below.
1) The authors have not removed some of the statements that I find problematic from the abstract and elsewhere, such as 35.2 nm resolution and "for long periods of time with minimal phototoxicity." In fact, phototoxicity seems to be a problem, as mitochondria do display marked swelling. The authors now quantify swelling of mitochondria. Statements such as "showed healthy behavior with few changed to spheroidicity" might still be perceived as downplaying potential phototoxicity, especially as in the discussion, the other super-resolution imaging options for mitochondria are called out as being too phototoxic. Also in the present paper there seems to be a substantial light effect on the mitochondria and it should be discussed more fully that much of the observed alterations/dynamics could simply be due to phototoxicity. It should also be noted that phototoxicity is not just dependent on laser power. A large portion of it is mediated by fluorophore bleaching, such that the bleaching rate and dye concentration are also important parameters for phototoxic responses.
2) The paper by Wang et al. in PNAS should be acknowledged in the main text and the specific merits of the two approaches discussed. It is not sufficient to cite it as a supplementary reference.
3) I pointed out in my previous comments that signal-to-noise ratio is an important factor when determining resolution. E.g. the resolution measurement in Frame 60 of Fig. 3 is not meaningful. Signa-to-noise ratio in Fig. 2c is too low to perform a reliable resolution measurement on it. 4) Fig. 1c: On what sample is the saturation intensity measured? How is such a measurement taken precisely? The authors state in the rebuttal letter that they cannot exclude that bleaching contributes to the apparently reduced saturation intensity of their probe vs. Atto 647N. Without a proper description of the procedures, it is not clear what the merit of such a measurement is and whether the conclusions drawn are valid. 5) Reviewer #1 asked for a description of deconvolution. It is not helpful if this is in the response to the reviewer and does not enter the methods section. 6) "Plasma membrane" is the membrane that separates the cell interior from the outside. From the authors' explanation in the rebuttal letter I am still not sure they want to state that they label the plasma membrane rather than a mitochondrial membrane. 7) Power and intensity in Fig.1 are still mixed up. Milliwatts are not a unit for intensity. This also applies to the saturation intensities stated. 8) C an the authors please state how they assess viability (Suppl. Fig. 8, see my comment in the original referee report)? The data in Suppl. Fig. 8 is meaningless if the authors don't give the details on how they perform such an analysis. 9) The new section on data processing is not understandable.
10) The caption in suppl. fig. 9 and 10 does not explain how cells were labelled. It would be good to have such information directly in each figure caption. 11) Suppl. Fig. 14: I cannot read the axes in panels d and e. 12) Whenever a nonlinear lookup table is given, a colour bar for the lookup table should be provided. This also applies to the 2D histograms in suppl. figs. 1-3. 13) Suppl. Fig. 2: It appears that slightly different focus positions are imaged in the two colour channels, lowering overlap. 14) Hemozoin is a crystalline substance, such that "melting" is more appropriate there than for dye molecules.
Reviewer #3 (Remarks to the Author): After revision, I recommend this manuscript to be published.  . Sup fig 7). Perhaps some of this data could be moved to data in the main paper rather than supplemental.

Response Letter
Response: Thank you for your helpful review and your recognition on several key improvements in the revised manuscript. We agree that some of claims in the paper are still overstated such as phototoxicity.
We revised the overstated claims in the revised manuscript to make it more rigorous (you can find the changes as follow). We also move the comparison figure between the new dye and existing options (old SI figure 7) to main text to highlight the superiority of MitoEsq-635. We are appreciated for your kind suggestions again.
Changes about overstated claims are as follow:

Changes:
Abstract, Paragraph 1, (Page 1) "Our study demonstrates the emerging capability of optical STED nanoscopy to investigate intracellular physiological processes at nanoscale resolution for long periods of time with minimal phototoxicity." Changed to: "Our study demonstrates the emerging capability of optical STED nanoscopy to investigate intracellular physiological processes at nanoscale resolution for long period of time." Changes:

Introduction, Paragraph 4, (Page 3)
"To address the challenges in long term, high resolution STED live cell imaging, in this work, we have developed a new squaraine dye derivative  that is compatible with live cells and produces low phototoxicity."

Changed to:
"To address the challenges in long term STED live cell nanoscopic imaging, in this work, we have developed a new squaraine dye derivative (MitoESq-635) that is compatible with live cells."

Results, Photostability and stimulated emission saturation intensity of the dye, Paragraph 3 (Page 7)
"Benefitted from its low saturation intensity, the MitoESq-635 probe exhibited no marked toxicity in HeLa cells at a concentration of 1 μM after 1 hour of incubation during STED superresolution imaging microscopy (Supplementary Figure 8). The low saturation intensity, extended photostability, and low toxicity make the dye very suitable for long-term STED imaging in live cells."

Changed to:
"The MitoESq-635 probe exhibited low toxicity in HeLa cells at a concentration of 1 μM after 1 hour of incubation ( Supplementary Figure 9), which suggests that the morphological changes of mitochondria are primarily from the STED imaging light dose. But with the concentration and incubation time increase, cytotoxicity of the probe will happen as well. The low saturation intensity, and extended photostability make the dye very suitable for long-term STED imaging in live cells."

Changes:
Results, Subcellular dynamic nanoscopic imaging with MitoESq-635, (Page 9) " Here is the response to specific issues one by one.

Response to comment 5
Comment: The text is still unclear with reference to the dye labeling the plasma membrane. The plasma membrane is the membrane at the exterior of the cell and is distinct from the mitochondria and mitochondrial membranes. Thus, it is unclear how the dye is binding to the plasma membrane and labelling mitochondria. Perhaps the authors mean the dye can bind membranes generally, rather than the plasma membrane specifically?
Response: Thank you for your comment. Sorry for our misleading by using "plasma membrane".
Originally, we would like to mention the possibility for wide applications in fluorescent labeling of enhanced squaraine dyes derivatives (Supplementary Note 2). MitoESq-635 is one of enhanced squaraine dyes derivatives we designed for mitochondria. According to the experiment of colocalization (Supplementary Figure 1, 2, 3), MitoESq-635 is specially binding to the membrane proteins in the mitochondria. Thus, to make it clear, we revised these phrases in the revised manuscript. Response: Thank you for your comments. Here, we present 3D z-stack STED imaging (multiple layer 2D STED) to give more spatial information about the mitochondria, comparing with 2D STED imaging in other figures (one layer). Previously, it was challenging to perform mitochondria STED imaging in 3D zstacks because of photobleaching during imaging. We admit that it is better to improve the resolution in z axis with smaller z-steps; and even with 3D-STED (the donut also applies on z axis). Restricted to the STED microscopy system, we cannot achieve this at the moment. We believe that a better z-resolution and a smaller z-step can be achieved with our MitoESQ-635 in future. Response: Thank you for your comments and we agree with your comments. We changed the color to combine them. We are sorry for the different details between the two comparison. We added all Response: Thank you for your comments. We agree that signal-to-noise ratio is an important factor when determining resolution. Figure 3b also confirm this, when frame number is bigger, we can see the SNR is slower and resolution becomes worse. For Fig. 2c, here we have used FRC-based resolution metrics to evaluate the resolution, as shown in Supplementary Figure 10. The FRC solution is 59.1 nm which is close to intensity profile resolution 51 nm. We also added it into Supplementary Figure 10 as below.

Response to comment
The illumination and donut beams induce photobleaching as the imaging time increase. The time lapse STED images of living cell will have both fluorescence signal and resolution loss. What we want to show here is that the resolution will drop due to SNR and other complex factors in (figure 3b and c). As the reviewer pointed out here, the resolution measurement in Frame 60 is not so reliable due to its low SNR.
We also put the FRC result panels here and add a discussion about SNR issues.

Added:
Results, Subcellular dynamic nanoscopic imaging with MitoESq-635, Paragraph 1, (Page 9) "The time lapse STED images of living cell will have both fluorescence signal and resolution loss with frame number increasing. Note that due to its low SNR, the resolution measurement in Frame 60 is difficult to have a reliable cross section profile and determine the resolution relying on its FWHM, so we also put the FRC analysis results of frame #1 and #60 here."

Response:
Thank you for your suggestions. For materials with less photobleaching, such as quantum dots, nanoparticles are used to measure saturation intensity [R1]. For materials which can be influenced by photobleaching and dipole orientation, saturation measured by single molecule and thin film are used and compared by S. W. Hell, samples in solution are discussed as well [R2]. In our experiment, we first prepared for the dye solution (1 mM) by dissolving the sample in DMSO solvent. Then the obtained solution was diluted with ethanol solvent to make the experimental solution (<0.1 μM). Small amount of the diluted dye solution in ethanol (<0.1 μM) were dispersed onto a clean glass slide for the measurement of the STED saturation intensity using a commercial Leica-SP8 microscope. After the evaporation of the ethanol solvent, the residue of dye molecules on the glass slide was observed to be sparsely fluorescent spots, subsequently sealed with a coverslip. Some fluorescent particles were selected to measure the saturated intensity. The bleaching effect is reduced, although we can't avoid photobleaching completely. And with this method, we calculated the saturation intensity of ATTO 647N in our experiment to be 15 MW/cm 2 , which is close to the data of 10 MW/cm 2 in reference [R3]. Therefore, we believed the saturation intensity of MitoESq-635 is valid as well.

Results, Photostability and stimulated emission saturation intensity of the dye, Paragraph 3, (Page 6)
However, the squaraine-STED dye has a lower saturation intensity at 0.893 mW, which is ~3.4fold lower than that of ATTO 647N (3.069 mW, Figure 1c).

Changed to:
However, the squaraine-STED dye has a saturation intensity of 4.37 MW/cm 2 , which is ~3.4-fold lower than that of ATTO 647N (15.0 MW/cm 2 , Figure 1c,which is close to the data 10 MW/cm 2 in reference 24 ).

Response to comment 5
Comment: Reviewer #1 asked for a description of deconvolution. It is not helpful if this is in the response to the reviewer and does not enter the methods section.
Response: Thank you for your suggestions. We added it to the methods section in revised manuscript. Response to comment 6 Comment: "Plasma membrane" is the membrane that separates the cell interior from the outside. From the authors' explanation in the rebuttal letter I am still not sure they want to state that they label the plasma membrane rather than a mitochondrial membrane.
Response: Thank you for your comment. Sorry for our misleading by using "plasma membrane".
Originally, we would like to mention the possibility for wide applications in fluorescent labeling of

[REDACTED]
Response to comment 8 Comment: Can the authors please state how they assess viability (Suppl. Fig. 8, see my comment in the original referee report)? The data in Suppl. Fig. 8 is meaningless if the authors don't give the details on how they perform such an analysis.
Response: Thank you for your suggestions. We have added the detailed explanation on the cell viability assessment and experimental procedures in the revised Supplementary Figure 9. The experimental details are listed as follows.

Added:
Detail for Supplementary Figure 9 In this experiment, a commercial chemical reagent of CCK-8, being nonradioactive, allows sensitive colorimetric assays for the determination of the number of viable cells in cell proliferation and cytotoxicity assays. Herein, we use the products from Dojindo (CCK-8) to carry out the cytotoxic assay. The detailed experimental procedures are listed as follows: 100 μL of HeLa cell suspension (about 5000 cells/well) was firstly dispensed in a 96-well plate, and was pre-incubated for 24 hours in a humidified incubator (37°C, 95% humidity, 5% CO2). Then 10 μL of various concentrations of MitoESq-635 (final concentrations are 0, 0.5, 1, 2, 5 μM) were added to the 96-well plate to be tested, which the 96-well plate was incubated for different time (1, 3, 6 hours) in the incubator. And then 10 μL of CCK-8 solution was carefully added to each well of the above plate to avoid introducing bubbles into the wells. The obtained 96-well plate was incubated for another 3 hours in the incubator. And finally, the absorbance of the samples at 450 nm were measured using a microplate reader (Rayto RT-6100, Shenzhen, China).

Response to comment 9
Comment: The new section on data processing is not understandable.
Response: Thank you for your comments and sorry for our inaccurate statement. We revised it in the revised manuscript. You can find the revised section as fellow.

Changes:
Methods and materials, Data processing, Page 16 "Data processing -In Figure 3, for box plot of width of mitochondria, each frame has 6 data points, which means 6 mitochondria in that frame are analyzed. Their widths are detected by using intensity profile in Fiji. The figure is plotted using Origin software, Plot-Statistics-Box Chart. For fluorescence intensity, using the same original images data with width mitochondria, choose 6 areas (10 pixels x 10 pixels) in the frames.
In Fiji, Images -Stacks -Measure Stacks, then the total intensity of every square (10 pixels x 10 pixels) can be obtained in all frames. In every frame, we have 6 intensity data (6 squares), calculated the mean value intensity of every frame. To normalize the intensity, all intensities are divided by the mean value of the intensity of first frame. Then calculated the mean value and variance of intensity in each frame by using

Statistics on Rows in Origin software. For resolution, intensity is obtained using Fiji, and Gaussian
Cumulative Fit Peak is processed with Origin software."

Changed to:
"Statistics analysis -Width of mitochondria, and fluorescence intensity are detected by using Fiji software.
Both Box plot chart of width of mitochondria and error bar chart of fluorescence intensity are obtained by using Origin software. Gaussian Cumulative Fit Peak is processed with Origin 2018 software."

Response to comment 10
Comment: The caption in suppl. fig. 9 and 10 does not explain how cells were labelled. It would be good to have such information directly in each figure caption.
Response: Thank you for your suggestions. We added how cells were labelled in Suppl. Fig. 12 (was Suppl. Fig. 10). We delete the original Suppl. Fig. 9, because we are not interested in the influence of excitation beam. In Suppl. Fig. 12, we use 0.5 μM MitoESq-635 here, and the probe was subject to be incubated with