Non-Invasive Mechanical-Functional Analysis of Individual Liver Mitochondria by Atomic Force Microscopy

  1. Ekaterina O. Zorikova
  2. Sabita Chourasia
  3. Irit Rosenhek-Goldian
  4. Sidney R. Cohen
  5. Semen V. Nesterov
  6. Atan Gross

Reviewed by

Read the full article See related articles

Discuss this preprint

Start a discussion What are Sciety discussions?

Listed in

Log in to save this article

Abstract

Mitochondria play a pivotal role in energy production, signaling, and apoptosis. Yet, probing their functional state at the single-organelle level without invasive labels remains a major challenge. Here, we introduce a novel, label-free approach that leverages Atomic Force Microscopy (AFM) beyond its traditional imaging role, transforming it into a powerful tool for functional analysis of individual, isolated mitochondria. By immobilizing mouse liver mitochondria on polylysine-coated mica, we achieved nanoscale resolution of mitochondrial mechanical properties including height, height fluctuation power spectra, and Young’s modulus, under different respiratory states. Strikingly, fluctuations in mitochondrial height fluctuations below 20 Hz showed robust correlation with the mitochondria membrane potential (ΔΨ m ), a cornerstone of mitochondrial function. This relationship allows AFM to sensitively detect changes in the mitochondria bioenergetic status. Applying this method to mitochondria from liver-specific MTCH2 liver-conditional knockout mice, a model of mitochondrial malfunction, we confirmed AFM’s diagnostic potential. The technique reliably distinguished malfunctional mitochondria, mirroring and adding new insights beyond conventional fluorescence assays. By bridging nanomechanics and mitochondrial bioenergetics, this approach paves the way for non-invasive, high-resolution diagnostics at the single-organelle level, holding promise to monitor the actual functional state of mitochondria in clinical settings.

Article activity feed

  1. Review Commons

    Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

    Learn more at Review Commons

    Reply to the reviewers

    please find separate uploaded file entitled "Response to Reviewer"

  2. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #3

    Evidence, reproducibility and clarity

    The article titled 'Non-Invasive Mechanical-Functional Analysis of Individual Liver Mitochondria by Atomic Force Microscopy' discusses how the mechanical properties of mitochondria as response to various drugs, like CCCP and ADP or rotenon and antimycin A.

    Key findings:

    The Authors correlated the thermal noise power spectrum (PSD) measured in contact on the top of mitochondria using atomic force microscope (AFM) with membrane activity of the organelles measured using fluorescence markers.

    They identified correlation trends between PSD, height, elasticity and fluorecence marker intensities for various cases, where the organelle activity was modified using drugs or genetic changes. The work is a very interesting approach, an excellent application of mechanobiology to gain further understanding of the properties of the energy producing organelles of eukaryotes. However, the overall results the Authors present have some serious flaws.

    I would recommend for publication after significant changes were made.

    Major comments:

    Upon measuring the power spectrum density (PSD) of thermal fluctuations in contact of an organelle, there are several factors influencing the measurements, such as: spatial inhomogeneity of the mitochondrion, the loading force applied, the feedback system of the AFM, hydrodynamic drag of the media on the cantilever.

    None of the above points are addressed in the manuscript. That is:

    • what was the spatial variability of the signal on the top of the organelle? (Using a tip with 30 nm apex radius has a relatively high variability even in microscopically homogeneous systems)
    • what was the loading force applied, and how did the PSD vary with the loading force?
    • according to the text on the bottom of page 5 the feedback was ON. How did this influence the recorded PSD? Significance of differences between organelles can be only properly estimated in relation to the spatial and load dependence of the same information.

    Minor comments:

    • Numerical Fourier transform generating the PSD is very noise prone, thus many curves need to be averaged for a good result. Please provide statistical information on this aspect of the obtained curves.
    • In the text it is mentioned that characteristic changes of the PSD were observerd. What are the characteristic changes between unperturbed and drog affected mitochondria? Please highlight them on the graphs of PSD.
    • How is the distribution of the results e.g. in Figure 1.E? Histogram and box-plots are more informative than bar plots.
    • How many curves were recorded for the individual mitochrondia? (30 mitochrondia were measured)
    • Figure 2.A and Figure S1.C indicate nicely how heterogeneous the mitochondria are. How did you eliminate the corresponding error from the PSD measurements?
    • To highlight correlations, simple plots of the parameters as the function of each-other can be very informative.
    • On Figure 1, the correlation between the fluorescence intensities and the PSD integrals are only qualitative.
    • On Figure 3 the inverste correlation between the height and Young's modulus is not clear. Can it be plot such a way that the intended information becomes clear?
    • While the Authors are claiming that the PSD is charactersitic to the mechanical properties of the organelles, its direct connection remains elusive and is not discussed in the paper. Again, loading force dependence is expected to be present and influence whether the probe is detecting changes in membrane properties or sense something deeper, structures under the membrane.
    • While the Authors correlate various measures derived from AFM data, these are only ensemble comparisons, since imaging and PSD measurements were done using different AFMs, thus different sample points. This should be clearly stated in the text.
    • QI mode is very robust for imaging, but its Young's moduli are difficult to compare to any real situation, since the measurement si performed typically at the 500 - 2000 Hz frequency range. Not mentioning that the individual force curves are usually rather noisy for biological samples.
    • In Figure S1.B, nothing is visible for the CCCP sample.
    • In Figure S2, what does the value of 300 means for alpha in the first sentence?
    • While the frequency dependence of the PSD makes sense, the data indicated in figure S2 also indicates very high noise, making the fits unreliable. What would be the exponent value in the 5% - 95% confidence interval?
    • It may be also informative to see a common plot of individual PSDs for the various cases, and in the representative plot see mean +/- SE plots for each frequency points.
    • In the experiment description stands: 'Bruker Multimode AFM was used for overall imaging and power spectra in tapping mode.' This is misleading, because in tapping mode the end of the cantilever is driven by a constant frequency, which would interfere with the thermal PSD measurement. If it was done so, this is a driven state which should be discussed, and which is also dependent on the driving frequency.
    • When preparing the PLL surfaces, how were the mica substrates washed before adding the organelles?
    • The topography images are most probably measured Z-piezo sensor outputs. However, this is not mentioned.
    • Imaging conditions of QI mode are incomplete the point measurement frequency, parameter to the apparent Young's modulus is not mentioned.

    Referee cross-commenting

    Reading the review of Reviewer 1 highlights the flaws in the organelle biology part of the work I was not aware of. (I am expert in mechanical characterization in the molecular - cellular level.) Putting the reviews together highlights that this study is in a very early state of investigation. It would be really interesting to see its results, but claiming it to be a novel diagnosis tool may be far fetched. (I agree with Referee 1.)

    Significance

    In general, the idea of estimating the mechanical properties of mitochondria and correlate them to the activity of the organelles is a very interesting idea in the field of mechanobiology. The Authors have done a relatively large amount of experiments to identify correlation between activity followed by more traditional fluorescence labels and the AFM data they generated. They performed many experiments spanning also three AFM devices and other experimental methods in their work.

    Limitations:

    I believe however, they missed some key points influencing their results, most importantly the dependence of the data on the:

    • normal loading force
    • spatial inhomogeneity (their own images prove the presence of this)

    I am afraid some of the effects they detect are not only qualitative, but also biased, but with the current figures and data I cannot substantiate.

    Audience: specific to microbiology, especially the audience interested in mechanobiology

    I believe this is an interesting work, and contributes to our understanding of micromechanics at the organelle level. Thus I would really like to see it published in a more complete form.

    Advance: Mitochondria is known to respond to environmental clues and can remodel its internal structure in response to stresses. However, it is difficult to find studies on the individual mechanical properties of these organelles, even in ex-situ environments.

  3. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #2

    Evidence, reproducibility and clarity

    In the article "Non-Invasive Mechanical-Functional Analysis of Individual Liver Mitochondria by Atomic Force Microscopy", O. Zorikova and colleagues propose the use of Atomic Force Microscopy (AFM) as a tool for characterizing the biophysical properties of individual mitochondria. By analyzing parameters such as height, membrane fluctuation power spectra, and Young's modulus under various drug treatments and genetic mutations, the authors aim to provide a novel, label-free method for assessing mitochondrial functionality.

    While the manuscript presents an interesting approach, the introduction would benefit from a clearer and more cohesive narrative. The authors highlight the need to monitor the function of individual mitochondria, which is indeed an important challenge, but the rationale for doing so should be more explicitly stated. A stronger emphasis on the biological importance of mitochondrial biophysical parameters and the added value of using AFM would enhance the motivation for the study. Additionally, the symbol Δψ, referring to mitochondrial membrane potential, should be defined and briefly explained in the introduction for clarity.

    In the results section, a schematic diagram of the experiment would aid comprehension, especially for readers less familiar with this technique. In general, in the figures it would be good to find the individual data points. The integration of the results into the main text could also be improved. Currently, several findings are presented in a descriptive manner, but the biological interpretation or relevance is not always clear. For example, the sentence "Figure 2 presents a comprehensive analysis of the height and elastic properties of mitochondria" could be expanded to explain what those findings actually mean and how they help support the main goal of the study. Similarly, the statement that "the integrated power of mitochondrial membrane fluctuations decreased significantly upon valinomycin treatment" is presented without explanation of what this metric represents or why valinomycin was chosen. When discussing MTH2, the authors refer to "mechanical alterations in mitochondria lacking this protein" without explaining what MTH2 is, where it is localized, or why it is biologically relevant.

    Finally, in the discussion, the interpretation of results could be expanded. For example, the statement "MKO/MLM exhibited increased integrated power/potential, increased modulus/stiffness, and decreased height" would benefit from more biological context - what do these changes imply about mitochondrial function or physiology? Adding this kind of interpretation would help the reader better understand the broader significance of the findings.

    Methods: The authors say they record the piezo movement but it is not clear to the reviewer if the authors perform a closed-loop force-feedback experiment. If so, this will introduce noise into the measurement which can be avoid by performing an open loop measurement. Why did the authors not record the cantilever fluctuation at a constant piezo height? This gives enough bandwidth and low noise to record Angstrom deflections. Likewise, it is unclear to this reviewer why the power spectrum is given in V and not in nm, as it is typical in AFM measurements. I assume the authors calibrated the deflection sensitivity and spring constant of the cantilever, hence, if possible, the authors should convert the PSD into nm/Hz.

    During the elasticity measurements, did the authors correct for the finite thickness of the mitochondria? What was the contact force and indentation depth, and how thick were the mitochondria to begin with? If the indentation is larger than 20%, I suggest to perform a correction to account for the infinite stiffness of the substrate. Given that the mitochondrial stiffness is in the tens of kPa, this seems to be important (perhaps not for relative values but for absolute stiffness measurements).

    Figures. The figures are well constructed and aid the reader through the important messages of the paper. The authors however, should not excessively overuse bar charts without explicitly mentioning number of measurements for each condition. In essence, I strongly recommend plotting individual data points to see the distribution and replace the stars with actual p-values.

    Significance

    The premise of the study is compelling and could have important clinical implications for distinguishing dysfunctional mitochondria in pathological contexts. However, the manuscript in its current should be improved. First of all, non-invasive is more than an euphemism, as the mitochondria need to be taken out of the cell, which is highly invasive. The authors should delete non-invase from the title.

    As the work presents an orthogonal and non-standard approach, the authors introduced a novel assay that can guide future investigations into the biophysics of mitochondrial physiology. Thus the paper is of high interest, timely and cutting edge.

    In summary, the study presents a promising approach with potentially high relevance for mitochondrial research.

  4. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #1

    Evidence, reproducibility and clarity

    Summary:

    The authors use atomic force microscopy (AFM) to study mitochondria isolated from primary mouse livers, and they attempt to correlate these measurements with mitochondrial membrane potential and oxygen consumption under different bioenergetic conditions. They argue that AFM could be used diagnostically to assess mitochondrial function. While there is some novelty in potentially using AFM to assess mitochondrial function in the clinic, it is not clear how this would be more efficient or meaningful that assessing mitochondrial parameters by more standard methods, such as respirometry, confocal microscopy, etc. Considerably more work would need to be performed, particularly on relevant patient samples, to show that AFM holds potential as a diagnostic tool. It is important to note that the authors of this study have not taken sufficient care to quantify the mitochondrial membrane potential in a manner that could be considered reliable, which casts further doubt upon the merits of this method for diagnosing mitochondrial function. These concerns, laid out in detail below, should be thoroughly addressed before publication.

    Major comments:

    The authors used azide to inhibit complex V, but azide is also a potent inhibitor of complex IV (Bowler et al., 2006). Why did the authors not use oligomycin, which is more specific, to inhibit complex V? In Fig. 1 H - K, the y axes are labelled in a confusing or ambiguous way. The legend says that all data represent the mean {plus minus} SEM; however, panels D, F, H, and K have no error bars. For example, the data in H and K are shown as violin plots. Typically, the y axis would say what the name of the quantity is (e.g., mean TMRM fluorescence intensity) followed by the units (e.g., a.u.) in parentheses. However, the authors write, for example, in panel K "Mean pixel (TMRM)." The authors seem to follow the correct convention in panels D - G, so it is not clear why H - K are written incorrectly. In any event, the authors need to specify how these data were obtained, as there are virtually no details as to the methods of how these measurements of mitochondrial membrane potential were acquired. For example, JC-1 is a ratiometric probe. In its monomeric form, it emits a green signal, but, as the dye aggregates into so-called J-aggregates, the emission is red. The correct way of analyzing JC-1 signal is to compute the ratio of red over green fluorescence intensity. However, in the authors' quantifications, they simply say "Fluorescence (JC-1)." The units of the y axes go from zero to 20,000, which means that the authors likely did not assess the ratio of these emissions, so the data are not informative as to the actual mitochondrial membrane potential. Moreover, the authors indicate that they use 5 µM JC-1. This seems quite a high concentration, particularly for staining isolated mitochondria, which means that the dye has direct access to the organelle without having to cross the plasma membrane. There is no information about how long the dye was allowed to load and whether it was washed off prior to obtaining the measurements with the plate reader. Likewise, the authors used TMRM to also try to assess the mitochondrial membrane potential. In this case, they used 0.5 µM, but they did not indicate for what duration the mitochondria were exposed to the dye before going through the FACS. It should be noted, too, that TMRM is a Nernstian probe, which effectively stains mitochondria at concentrations as low is 1 nM. Accordingly, it is known that TMRM (and other mitochondrial dyes) can be toxic at higher concentrations, inhibiting essential processes such as OXPHOS. The very low dynamic range of the TMRM signal in panels H and K suggest that the signal was saturated, because there was too much dye loaded into the mitochondria. Moreover, the values, ranging merely from zero to 80 suggest a very insensitive method for quantifying the mitochondrial membrane potential. In Fig. S1 A-B, the authors used confocal microscopy to assess the isolated mitochondria. It would be wise to continue to use this technique for the other experiments, as plate readers and FACS offer no direct visual cues to validate that the numbers reflect bona fide biological measurements. Especially in the case of FACS, where there is an exceedingly large number of events, the statistics become essentially meaningless, as it is possible to show that almost anything is statistically significantly different if there is a sufficiently high number of samples or events. The authors should bear in mind that measuring the mitochondrial membrane potential is not trivial. One needs to understand the properties of the probes that are being employed as well as the instruments that are used to make the measurements. Care must be taken to ascertain that the quantifications reflect true biological processes. The authors claim, for Fig. 1, that there is an "excellent correlation" between height fluctuations and mitochondrial membrane potential. Given that the mitochondrial membrane potential measurements were associated with various errors (see above), it is premature to assert that there is any correlation, at all. Furthermore, if the authors want to argue that there is indeed a correlation between these variables, then they should perform an appropriate statistical analysis, e.g., a pearson correlation coefficient test.

    For the reasons explained above, the JC-1 and TMRM measurements in Figs. 3 and 4 are not convincing. The authors must demonstrate, unambiguously, that they understand the use of these probes and that they are making accurate measurements.

    Given that MTCH2 was recently reported to function as an insertase of the OMM (Guna et al., 2022), understanding the KO phenotype is extremely challenging, since it implicates the downstream loss of function of numerous other proteins. It would be valuable to examine other KO models with more specific mitochondrial defects, which can simplify the interpretation of the data. For example, suppression of any of the large Dynamin GTPases that control mitochondrial shape, i.e., MFN1/2, OPA1, or DRP1. Conversely, modulation of mitochondrial membrane composition by suppression of specific phospholipid biosynthetic enzymes would be valuable. It is important to note that the authors are attempting to highlight AFM as a novel way to assess patient samples, but they do not provide any data as to whether mitochondria, derived from a patient with a known mitochondrial defect, could be meaningfully assessed by this method. It is worth pointing out, too, that isolating mitochondria from primary tissues involves a significant amount of stress to the organelle. To understand mitochondrial function in a manner that reflects an in vivo state as much as possible, it would be essential to show that the isolated mitochondria from the liver are largely the same as those in intact liver cells. The authors should be aware that isolating live hepatocytes is far from a trivial thing to do (Charni-Natan & Goldstein, 2020). Simply mincing the liver and subjecting it to mechanical and enzymatic dissociation likely involves significant mitochondrial stress, which implies that the values derived from isolated mitochondria represent a highly non-physiological, even dysfunctional, condition. These are fundamental concerns which should be considered and discussed in any report that is lauding the potential diagnostic benefits of quantifying isolated mitochondria from primary tissues.

    The authors say, in the discussion, "Accordingly, the AFM method employed here measured several characteristics such as morphology and elastic modulus of the structures, as well as fully exploiting the rich information available from the noise spectra." There was no measurement of "morphology" in this study. Differences in height are not what is generally considered in discussions of mitochondrial morphology, which reflects the dynamic changes in organelle shape and connectivity, typically in the x-y (rather than z) axes.

    The authors performed experiments on fixed and dried mitochondria; however, there is no systematic comparison of the integrated power and other parameters compared to the live mitochondria isolates. This is a key comparison that should have been performed, as it would offer a basic frame of reference for the values of the live organelles. Another key experiment that is lacking in this study is measurement of the same organelle over time to understand the variance in individual organelles from moment to moment.

    Minor comments:

    Generally, the authors should moderate their claims that AFM could be used diagnostically until the above concerns are addressed.

    There needs to be considerably more detail as to the methods that were used here. This is essential insofar as the authors wish to convince potential readers that the experiments were carefully conducted and that the data is reliable. Putting numbers on the margin of the manuscript would be helpful for the referee to specifically address certain points.

    References:

    Bowler MW, Montgomery MG, Leslie AG, Walker JE. How azide inhibits ATP hydrolysis by the F-ATPases. Proc Natl Acad Sci U S A. 2006 Jun 6;103(23):8646-9. doi: 10.1073/pnas.0602915103. Epub 2006 May 25. PMID: 16728506; PMCID: PMC1469772.

    Guna A, Stevens TA, Inglis AJ, Replogle JM, Esantsi TK, Muthukumar G, Shaffer KCL, Wang ML, Pogson AN, Jones JJ, Lomenick B, Chou TF, Weissman JS, Voorhees RM. MTCH2 is a mitochondrial outer membrane protein insertase. Science. 2022 Oct 21;378(6617):317-322. doi: 10.1126/science.add1856. Epub 2022 Oct 20. PMID: 36264797; PMCID: PMC9674023.

    Charni-Natan M, Goldstein I. Protocol for Primary Mouse Hepatocyte Isolation. STAR Protoc. 2020 Aug 13;1(2):100086. doi: 10.1016/j.xpro.2020.100086. PMID: 33111119; PMCID: PMC7580103.

    Significance

    I am an expert in imaging of mitochondria, with considerable direct knowledge of various super-resolution and advanced imaging systems. I have also studied mitochondrial function, using standard biochemical and molecular approaches. I have great familiarity with mitochondrial behavior and dynamics, as understood from live-cell imaging approaches and morphological analysis.

    This study is potentially interesting due to its relatively novel use of AFM to examine mitochondria. However, there is a lot of uncertainty in the measurements due to technical oversights and lack of relevant controls. Whether AFM could be useful in the clinic remains an open question. If the authors could address the comments above, it would go a long way to finding out one way or the other.

  5. Version published to 10.1101/2025.06.01.657242 on bioRxiv