Label-Free Detection of Cellular Senescence in Fibroblasts via Third Harmonic Generation

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The article under review is devoted to a relevant topic - research into cell aging. The article is well written and easy to read. I have 2 small comments:

- in lines 288 and 308, the subscripts need to be corrected;

- the purity of H₂O₂ is not indicated in the text;

Author Response

Comments 1 : in lines 288 and 308, the subscripts need to be corrected;

Subscripts in lines 256 and 277 are now corrected.

Comments 2 : the purity of H₂O₂ is not indicated in the text;

This is a 30% (w/w in H₂O) Sigma‒Aldrich, USA, it is now indicated in line 136

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Summary

The paper uses label-free THG microscopy to detect senescence in H₂O₂-treated NIH 3T3 fibroblasts by quantifying lipid-droplet signal as a “lipid index” (z-summed lipid area / projected cell area). TPEF of Ki-67 serves as a proliferation control. After 800 µM H₂O₂ for 7–10 days, cells show enlarged cell/nuclear size, reduced Ki-67, and markedly higher lipid index and droplet area; THG also separates treated vs. control in unfixed live cells. The authors propose THG lipid readouts as a practical senescence marker and suggest extending to other cell types and pairing with spectral/FLIM imaging for chemical specificity.

 

Comments

1. Please standardize all figure legends and Methods to explicitly tag each dataset/panel as label-free or stained.
2. For live, unfixed cells, can you provide the same quantitative lipid-index analysis as for the fixed samples.
3. Since THG is sensitive to refractive-index interfaces, could you validate that the binarized THG area truly corresponds to lipid droplets either by co-localization with a lipid dye and reporting quantitative overlap metrics, or by providing spectral/FLIM or polarization-contrast evidence characteristic of lipids? Please include representative images and per-cell statistics. If such experiments are not feasible, explicitly discuss limitations and potential confounders and how they might bias the lipid index.
4. Given that the lipid index is computed using a fixed threshold and manual delineation of the cell outline, both are susceptible to operator bias. Did you apply z-attenuation compensation and/or flat-field shading correction?
5. Could you report effect sizes for the primary endpoints, for instance, lipid index, mean droplet area, and total lipid area, and show per-replicate data points in the figures? It is suggested to fit a mixed-effects model across biological replicates.
6. Given your acquisition settings (~1 s per frame, 20× frame averaging per plane, 7–10 z-planes), each sample requires 140–200 seconds. Could you quantify imaging efficiency and provide pre/post-scan viability/proliferation or morphology-stability readouts?

 

 

Author Response

Comments 1: Please standardize all figure legends and Methods to explicitly tag each dataset/panel as label-free or stained.

In Figure 1 nuclei are stained with DAPI to enable the population measurement of nuclei, and ki-67 (2^nd antibody CF488) in Figure 1 to show the proliferation activity. The cells of Figure 2 contain β-galactosidase staining. In Figure 3 the nuclei are labelled with Hoechst 33342. Figure 4 that presents the non-linear results and Supplementary Fig1, display ki-67 (2nd antibody: CF568 for red) to reveal the proliferation activity, while lipids are label-free. Supplementary Figure 2 is label-free.  All above information is mentioned in the legends and Material and Methods section and is now highlighted.

Comments 2: For live, unfixed cells, can you provide the same quantitative lipid-index analysis as for the fixed samples.

 The scope of this study is to show the utility of THG to serve as an additional tool to identify cellular senescence on fixed fibroblasts initially. We tried to perform similar measurements on live cells and the results are quite promising. This is a project that requires further experiments and more replicates in order to be complete. At this moment, we do not have enough results to present a quantitative report on the lipids of the live cells.

Comments 3 : Since THG is sensitive to refractive-index interfaces, could you validate that the binarized THG area truly corresponds to lipid droplets either by co-localization with a lipid dye and reporting quantitative overlap metrics, or by providing spectral/FLIM or polarization-contrast evidence characteristic of lipids? Please include representative images and per-cell statistics. If such experiments are not feasible, explicitly discuss limitations and potential confounders and how they might bias the lipid index.

There have been performed a number of colocalization experiments previously not only on various types of cells [1-3] but also on organisms [4] by our group and others, confirming that the main and more effective source of THG signals are the lipid domains. Pearson´s correlation method has been widely used to quantify the colocalization between THG and TPEF arising from stained lipids. Watanabe et al. [1] showed on mouse embryos that labelled Endoplasmic Reticulum (ER) and mitochondria both show relatively poor co-localisation with THG signal, having Pearson's Correlation coefficients between 0.29-0.39. The visual overlap of THG and LipidTox signal resulted in higher Pearson's Correlation of 0.65, proving that the high contrast in cytoplasmic bodies detected by THG appear predominantly to be the lipids [1]. In drosophila haemocytes, colocalization analysis revealed a strong correlation between the two signals [Pearson's correlation coefficient (PCC) > 0.9]. Our studies have proved that THG signals colocalise well with lipids stained with BODIPY [4,5] and Nile Red [2,4], while lipofuscin [4], mitochondria or endosomes [3] and lysosomes [5] show poor or no colocalisation.

 

1. Watanabe et al., BMC Cell Biol. 2010, 11:
2. Mari et al., J. Biophotonics 2023, 16 (12).
3. Gavgiotaki et al., Struct. Biol. 2015, 189 (2) :105-113
4. Tserevelakis et al., PLoS ONE 2014, 9 (1): e84431.
5. Palikaras et al., Journal of Lipid Research 2017, 58 (1), 72 - 80

Comments 4 : Given that the lipid index is computed using a fixed threshold and manual delineation of the cell outline, both are susceptible to operator bias. Did you apply z-attenuation compensation and/or flat-field shading correction?

For the lipid calculation, a fixed threshold was applied on the raw THG data, consequently there was not any other attenuation or correction applied. For the measurement of the cell outline, an intensity attenuation was applied on the z-projection of all slices of THG until the contour was clear to measure. Another way to measure the contour would be to stain the cytoskeleton of the cell, but this may have affected the lipids too. Due to the THG signals of the membrane lipids, the applied intensity attenuation was low as the contour was well-contrasted to the dark background. In addition, we display the comparison graphs of the total lipid content as well as the average sizes of the lipid droplets, which show a significant increase and are in accordance with the lipid index comparison (updated Figure 4).

Comments 5 : Could you report effect sizes for the primary endpoints, for instance, lipid index, mean droplet area, and total lipid area, and show per-replicate data points in the figures? It is suggested to fit a mixed-effects model across biological replicates.

We have three replicates per condition and the variances between the data of the replicates were small. We chose the one-way ANOVA with a post hoc Tukey HSD test for multiple comparisons between the groups. Figure 4 is now updated presenting the data points per replicate for every condition. The figure legend and text has now been changed accordingly.

Comments 6 : Given your acquisition settings (~1 s per frame, 20× frame averaging per plane, 7–10 z-planes), each sample requires 140–200 seconds. Could you quantify imaging efficiency and provide pre/post-scan viability/proliferation or morphology-stability readouts?

All our experiments were operated at safe energy limits without affecting the cell morphology. All the quantification experiments were performed on fixed samples. The laser energy at focal spot is less than 0.5 nJ. This is the advantage of nonlinear imaging that all the laser energy is deposited at the focal spot for a few microseconds, in our case 4μs for every pixel and the image is constructed using raster scanning of 500 x 500 pixels.

 

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

This research article by Mari et al. explores the use of a nonlinear optical technique known as third harmonic generation (THG) to probe lipid content and its correlation with cellular senescence. The authors propose that THG can serve as a useful tool due to its ability to generate an enhanced signal output when lipid content changes. As a proof-of-concept, they designed an experiment in which cells were treated with hydrogen peroxide (H₂O₂) to induce oxidative stress, which in turn led to increased lipid content. In the following sections, THG and multiphoton spectroscopy were used to distinguish between senescent and non-senescent cells based on this correlation.

There is a growing need for optical techniques that allow deep-tissue imaging beyond biological windows, with tunability in excitation and emission responses without needing to change optical components. Nonlinear optical phenomena such as THG and second harmonic generation (SHG) do offer a versatile platform to study various biological processes and have been applied in fields like biomedical sensing and imaging. However, this report does not introduce any new advancements or insights to the scientific community. Furthermore, several control experiments and justifications are missing, which could have improved the study’s overall impact.

Here are my suggestions to improve the study:

1. The Introduction is relatively long and scattered. I recommend the authors condense and restructure it to present a clear narrative. For example:

~ What is the problem?

~Why is it important to solve this problem?

~ How is the problem currently being addressed and what are the limitations?

~What do the authors propose and why might it be a better approach?

2. Nonlinear optical processes like THG require a femtosecond pulsed laser, which can cause phototoxicity and damage biological samples and makes it less than ideal for bioimaging. The authors suggest that stress can induce senescence in cells, and high-energy light photons are also a form of stress. I could not find, however, any control experiment showing that light exposure alone (i.e., without H₂O₂) does not contribute to senescence. A control experiment is essential to clarify whether light photons themselves might be enhancing senescence.

3. THG is a third-order nonlinear process, and the authors used a 1064 nm laser, which would emit ~354 nm photons in the UV range. These high-energy UV photons can damage cells and potentially induce senescence or other changes. Were any control experiments performed to evaluate the impact of this UV exposure on the cells?

4. How did the authors optimize the photon energy dosage for THG experiments? What parameters were considered, and were any control studies conducted to evaluate the role of photon energy on cells?

5. Why was SHG not used to probe lipids? SHG is widely applied to study lipid bilayer membranes and is known to provide a stronger signal than THG due to the higher probability of two-photon interactions. Did the authors compare SHG with THG in this context? Was autofluorescence observed in their samples?

6. The authors claim that THG allows label-free imaging based on refractive index changes. In dynamic systems like cells, however, the refractive index can fluctuate due to factors like the secretion of electrolytes, Brownian motion, H₂O₂ presence, and other components in the medium. Were any control experiments conducted to measure refractive index changes in both treated and untreated cells?

7. What was the local temperature change during THG experiments? The combination of femtosecond laser exposure and presence of lipids may lead to localized heating. Was this effect quantified or controlled?

Some minor suggestions to improve the content and grammar in the manuscript…

~ please avoid the use of “novel” for THG

~ some sentences are very long, for instance

“To assess the broader applicability of THG microscopy in detecting senescence-associated lipid changes, future studies should include a wider range of cell types and tissue specimens—such as sections from young versus aged animals—to further validate this approach in physiologically relevant contexts.”

~ use of “—"!!

Author Response

There is a growing need for optical techniques that allow deep-tissue imaging beyond biological windows, with tunability in excitation and emission responses without needing to change optical components. Nonlinear optical phenomena such as THG and second harmonic generation (SHG) do offer a versatile platform to study various biological processes and have been applied in fields like biomedical sensing and imaging. However, this report does not introduce any new advancements or insights to the scientific community. Furthermore, several control experiments and justifications are missing, which could have improved the study’s overall impact.

Our study demonstrates that THG microscopy offers a versatile alternative to fluorescence and dye-based approaches for investigating lipid biology in the context of cellular senescence.

Our contribution lies in presenting a label-free, noninvasive THG imaging method to detect this phenotype in situ. By demonstrating significant differences in lipid accumulation between senescent and nonsenescent fibroblasts, we provide an innovative technical approach to detect cellular senescence. This method underscores the relevance of lipid accumulation as a detectable phenotype of senescence, supports further investigation into the metabolic remodeling that accompanies this state and opens new avenues for detecting and studying age-related diseases, where senescence plays a crucial role. (page 3)

Comments 1: The Introduction is relatively long and scattered. I recommend the authors condense and restructure it to present a clear narrative. For example:

~ What is the problem?

~Why is it important to solve this problem?

~ How is the problem currently being addressed and what are the limitations?

~What do the authors propose and why might it be a better approach?

Introduction is now more concise responding to the questions and the recommendations of the reviewer. All the changes are now highlighted in red.

Comments 2: Nonlinear optical processes like THG require a femtosecond pulsed laser, which can cause phototoxicity and damage biological samples and makes it less than ideal for bioimaging. The authors suggest that stress can induce senescence in cells, and high-energy light photons are also a form of stress. I could not find, however, any control experiment showing that light exposure alone (i.e., without H₂O₂) does not contribute to senescence. A control experiment is essential to clarify whether light photons themselves might be enhancing senescence.

A plethora of studies over the past three decades from different research groups (see references of the manuscript) have employed nonlinear imaging and especially THG technique as a non-invasive diagnostic tool that provides new insights to fundamental biological problems. This technology enables label-free, improved resolution, high contrast images with increased penetration depth into the biological samples. The use of low pulse energies at the sample plane (less than 0,5nJ) minimize phototoxicity and photodamage. In addition, since THG is a nonlinear scattering phenomenon that does not involve energy absorption from the sample, it can be used for prolonged monitoring of biological activity.

We anticipate that the laser photons themselves have minimal influence on enhancing senescence, particularly considering the very short exposure time ~4μsec per pixel. Moreover, in this study we employed irradiation parameters comparable to those previously reported by our group and others under which limited side effects have been observed to the specimens.

A total of 48 untreated cells are presented in this manuscript and if these light photons could enhance senescence, then this would be reflected on our results. The data of the replicates of the graphs of Figure 4 (e-g) show the lipid content of 48 untreated cells which fluctuates reasonably.

Comments 3: THG is a third-order nonlinear process, and the authors used a 1064 nm laser, which would emit ~354 nm photons in the UV range. These high-energy UV photons can damage cells and potentially induce senescence or other changes. Were any control experiments performed to evaluate the impact of this UV exposure on the cells?

We agree with the reviewer that the optimal solution is the employment of a tunable ultrafast laser oscillator as the excitation source (e g Chameleon Discovery from Coherent, 680 to 1300nm). Such sources are able to generate THG signals in the visible part of the spectrum. However, this option includes a significantly higher cost. In practice, the majority of the scientific studies have employed affordable, compact, single wavelength fs laser oscillators emitting in the region of ~1000nm. This configuration allows the detection of THG photons in the near UV region of the spectrum and is preferable to the use of conventional Ti:Sapphire lasers operating at 800nm. As the three-photon cross section is quite low, the minimal number of the emitted near UV photons is not expected to induce photodamage or other adverse effects in the biological specimen under investigation.

Previous studies on human dermal fibroblasts (HDF) showed that cellular senescence was induced after 6hr UV irradiation exposure on 100 mJ/cm² [1].

[1] Moon KC et al., Effects of Ultraviolet Irradiation on Cellular Senescence in Keratinocytes Versus Fibroblasts. J Craniofac Surg. 2019, (1):270-275

Comments 4 : How did the authors optimize the photon energy dosage for THG experiments? What parameters were considered, and were any control studies conducted to evaluate the role of photon energy on cells?

The selected photon energy dosage corresponds to the minimal energy per pulse at the sample plane that is required to generate detectable non-linear phenomena (THG) from the specimen. For our experiments, the following parameters were employed: less than 0.5 nJ energy per pulse, 2.5KW peak power, power density ~1 MW/cm², and fluence ~30mJ/cm². These values consistent with those reported in well-established studies (Refererences : 43, 49, 55 of the manuscript ) are known to induce minimal phototoxicity and photodamage effects to the samples.

Comments 5: Why was SHG not used to probe lipids? SHG is widely applied to study lipid bilayer membranes and is known to provide a stronger signal than THG due to the higher probability of two-photon interactions. Did the authors compare SHG with THG in this context? Was autofluorescence observed in their samples?

THG imaging has been widely used over the past two decades as a non-invasive label free modality for lipid identification from biological samples. This technology has also the potential to provide the quantitative analysis of the lipid content in specimens (References [43-47] and [52-56] in the manuscript). While SHG can provide complementary information from the biological sample, such measurements were beyond the scope of the present study. Minimal autofluorescence signals were detected from the samples, as the THG signals were quite strong. Any other signals than THG were eliminated by applying a constant threshold for lipid quantification during the image processing stage.

Comments 6: The authors claim that THG allows label-free imaging based on refractive index changes. In dynamic systems like cells, however, the refractive index can fluctuate due to factors like the secretion of electrolytes, Brownian motion, H₂O₂ presence, and other components in the medium. Were any control experiments conducted to measure refractive index changes in both treated and untreated cells?

As mentioned in the Materials and Methods section (page 3), the cells were treated at various concentrations of H₂O₂ and incubated for 1 h at 37°C.  Then the media was replaced with fresh standard DMEM–HG, and the cultures were left for 1, 3, 7, 10 and 14 days. The medium was renewed every 2–3 days. All quantified cells were fixed during imaging. The aim of the present study was to quantify lipid content in fibroblast by using THG imaging. A constant threshold was applied during image processing to solely detect and quantify the high THG signals arising from the lipids. All other potential THG sources were eliminated and excluded from further processing. The measurement of the refractive index changes in both treated and untreated cells falls beyond the scope of this work.

Comments 7: What was the local temperature change during THG experiments? The combination of femtosecond laser exposure and presence of lipids may lead to localized heating. Was this effect quantified or controlled?

Previous studies [1] have shown that when using an excitation wavelength at ~1000nm, with a pulse duration of ~200fs, a repetition rate of 50 to 80 MHz, and an objective lens with a NA of ~0.8 a significant temperature increase can occur in the biological sample when the pulse energy at the sample plane exceeds 2 nJ. No temperature measurements were performed in the present study to monitor potential local heating of the samples induced by fs laser irradiation. 1. G.J Tserevelakis et al., J. of Biophotonics 5, 200-207 (2012)

 

Some minor suggestions to improve the content and grammar in the manuscript…

~ please avoid the use of “novel” for THG

~ some sentences are very long, for instance

“To assess the broader applicability of THG microscopy in detecting senescence-associated lipid changes, future studies should include a wider range of cell types and tissue specimens—such as sections from young versus aged animals—to further validate this approach in physiologically relevant contexts.”

~ use of “—"!!

We have now responded to the minor suggestions of the reviewer (lines 431-434).

Round 2

Reviewer 3 Report

Comments and Suggestions for Authors

The font size and scale bars in Figures 1 to 3 are very small and difficult to see.

Author Response

Comments and Suggestions for Authors

The font size and scale bars in Figures 1 to 3 are very small and difficult to see.

Dear Referee,

We would like to thank you for your time and effort in reviewing our manuscript. We tried to consider in detail each of the points in the reviews and made every possible effort to address them in detail during this challenging period of time. In doing so, we strived to include substantial additional information and data, both within the main text/figures, as well as in the Supplementary Information section. We believe that with your encouraging and constructive input, we have been able to resubmit a significantly stronger and more comprehensive report.

We have now changed the fonts and the scale bars to larger sizes.