Mechanical control of the mammalian circadian clock via YAP/TAZ and TEAD

  1. Juan F. Abenza
  2. Leone Rossetti
  3. Malèke Mouelhi
  4. Javier Burgués
  5. Ion Andreu
  6. Keith Kennedy
  7. Pere Roca-Cusachs
  8. Santiago Marco
  9. Jordi García-Ojalvo
  10. Xavier Trepat

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

Autonomous circadian clocks exist in nearly every mammalian cell type. These cellular clocks are subjected to a multilayered regulation sensitive to the mechanochemical cell microenvironment. Whereas the biochemical signaling that controls the cellular circadian clock is increasingly well understood, mechanisms underlying regulation by mechanical cues are largely unknown. Here we show that the fibroblast circadian clock is mechanically regulated through YAP/TAZ nuclear levels. We use high-throughput analysis of single-cell circadian rhythms and apply controlled mechanical, biochemical, and genetic perturbations to study the expression of the clock gene Rev-erbα. We observe that Rev-erbα circadian oscillations are disrupted with YAP/TAZ nuclear translocation. By targeted mutations and overexpression of YAP/TAZ, we show that this mechanobiological regulation, which also impacts core components of the clock such as Bmal1 and Cry1, depends on the binding of YAP/TAZ to the transcriptional effector TEAD. This mechanism could explain the impairment of circadian rhythms observed when YAP/TAZ activity is upregulated, as in cancer and aging.

Article activity feed

  1. Version published to 10.1083/jcb.202209120
  2. Review Commons

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

    Learn more at Review Commons

    Reply to the reviewers

    In this section we list all the comments done by the three referees and our corresponding action.

    Regarding Reviewer #1:

    1. On "mechanical control": The authors show changes in circadian power fraction with changes in YAP and with cytoskeletal inhibitors, but there are no properly-controlled experiments that directly perturb mechanics. The authors show a correlation between YAP nuclear/cytoplasmic ratio and circadian power, but YAP N/C alone is not a readout of mechanotrasndcution, per se. The authors have shown two different experiments where cells are cultured on a stiff (30kPa) substrate and soft substrate (300Pa), but they do not shown a direct comparison of YAP nuclear localization and circadian power under these two conditions in the same experiment. Direct, controlled perturbation of mechanical cues is necessary to support the title's use of the phrase "mechanical control."

    We agree with the referee that further mechanical perturbations could strengthen our conclusions. In our original manuscript we directly controlled the mechanical environment by culturing cells on substrates of 300Pa and 30kPa in stiffness. These differences in stiffness were not sufficient to drive changes in circadian power fraction and YAP localisation, as depicted in Fig. 3C (we note that the direct comparison requested by the referee is shown in that figure). We hypothesise that this negative result is due to a very low “rigidity threshold” or to secretion of extracellular matrix that stiffens the initially soft substrate. In any case, we plan to strengthen the “mechanical control” message of our paper with one or more of the below experiments:

    A) We will measure circadian power fraction and YAP localisation in even extremer stiffness/adhesion conditions, using 300 Pa and 30 kPa polyacrylamide gels with a different fibronectin coating protocol, as described in Elósegui-Artola et al., 2017. This allows a much finer control of the concentration of fibronectin coated, so we can reach low enough levels to compromise the cell adhesion to the substrate and cross down the threshold that would lead to cytosolic localisation of YAP. We will perform this experiment in presence of the FUD peptide, which inhibits matrix deposition (Tomasini-Johansson et al., 2001; this peptide has already been tested in our lab).

    B) We will use the approach described in Fig. 2E to compare the circadian power fraction in cells spread in stadium-shaped islands of 2400 um2 and 1200 um2. Oakes et al., 2014 already showed that traction forces exerted by 3T3 fibroblasts depend on the size of the spread area of the cells, so we expect differences in mechanotransduction that should affect YAP localisation and, if our hypothesis is correct, the RevVNP circadian oscillations.

    C) We will abolish the physical connection between the actin cytoskeleton and the nucleus by disrupting the LINC complex via the overexpression of a dominant negative (DN) nesprin-1 KASH domain (Lombardi et al., 2011). The plasmid designed for the inducible overexpression of the DN KASH domain, originally tested in NIH3T3 cells (Mayer et al., 2019), is available in our lab and has been used to prove that uncoupling cytoskeleton and nucleus leads to nuclear YAP decrease in single cells (Kechagia at al., 2022). We will aim to increase the circadian power fraction in low density cells upon the overexpression of the DN KASH domain.

    Elosegui-Artola A, Andreu I, Beedle AEM, Lezamiz A, Uroz M, Kosmalska AJ, Oria R, Kechagia JZ, Rico-Lastres P, Le Roux AL, et al (2017) Force Triggers YAP Nuclear Entry by Regulating Transport across Nuclear Pores. Cell 171: 1397-1410.e14

    Kechagia Z, Sáez P, Gómez-González M, Zamarbide M, Andreu I, Koorman T, Beedle AEM, Derksen PWB, Trepat X, Arroyo M, et al (2022) The laminin-keratin link shields the nucleus from mechanical deformation and signalling Cell Biology

    Lombardi ML, Jaalouk DE, Shanahan CM, Burke B, Roux KJ & Lammerding J (2011) The Interaction between Nesprins and Sun Proteins at the Nuclear Envelope Is Critical for Force Transmission between the Nucleus and Cytoskeleton*. Journal of Biological Chemistry 286: 26743–26753

    Mayer CR, Arsenovic PT, Bathula K, Denis KB & Conway DE (2019) Characterization of 3D Printed Stretching Devices for Imaging Force Transmission in Live-Cells. Cel Mol Bioeng 12: 289–300

    Oakes PW, Banerjee S, Marchetti MC & Gardel ML (2014) Geometry regulates traction stresses in adherent cells. Biophysical Journal 107: 825–833

    Tomasini-Johansson BR, Kaufman NR, Ensenberger MG, Ozeri V, Hanski E & Mosher DF (2001) A 49-Residue Peptide from Adhesin F1 of Streptococcus pyogenes Inhibits Fibronectin Matrix Assembly*. Journal of Biological Chemistry 276: 23430–23439

    *2. *On "via YAP/TAZ": In addition to above, it is necessary to show that the changes in Circadian power fraction induced by mechanical cues in fact require YAP/TAZ signaling. Thus, an experiment comparing soft (300Pa) substrate with Stiff (30kPa) substrate in the presence or absence of YAP/TAZ is necessary to state that YAP and TAZ are the mechanistic mediators of mechanical cues on the clock.

    We are currently generating via CRISPR-KO and shRNA silencing a YAP1/TAZ double mutant. We plan to use this cell line in those conditions where YAP is prominently nuclear (low density in stiff substrates) with the purpose of rescuing the RevVNP circadian power fraction.

    1. While the TEAD-binding domain mutant experiment is elegant, to claim that TEAD is the transcriptional mediator, it must be demonstrated that this mutant indeed fails to induce TEAD-mediated transcription. This could be simply executed by demonstrating that the CCD mutant expresses reduced CTGF and Cyr61 (for example), compared to the 5SA, under these conditions. Further, endogenous YAP is still active and available to bind to TEAD in this system, which should be discussed.

    We plan to carry out quantitative real-time PCR of CTGF and Cyr61 in all the YAP mutants and the control. Regarding the presence of endogenous YAP, we will clarify in the text that a) the overexpression of the different YAP mutants was done in high-density conditions, where endogenous YAP is significantly less localised in the nucleus, and that b) the levels of the exogenous YAP are much higher (we already have western blots showing this).

    1. In Figure 3a: The cell perimeter needs to be shown either by actin staining or by brightfield images. The manually marking of cell boundaries is insufficient, specifically because the drugs used in this experiment affect the cytoskeleton. It would be very helpful to see this via actin staining or in the least with brightfield images.

    The cell perimeter was drawn based on the cytosolic YAP immunostaining, whose levels are high enough to infer the cell shape (higher resolution images can be attached if necessary). As stated in the manuscript, the YAP nuclear-to-cytosolic ratio is calculated using two adjacent areas of identical size, one inside the nucleus and the other one just outside (see Materials and Methods/Immunostainings), so the exact cell shape is irrelevant for this particular quantification.

    Regarding Reviewer #2:

    Effects on the circadian clock

    1. The authors use the fluorescent reporter created by Nagoshi from sections of the Rev-erbα gene. This reporter is widely used to estimate relative circadian timing in individual cells but it does not provide direct information on the circadian clock activity. In other words, while Reverb rhythmic expression is driven by the clock, it is not known whether less-rhythmic or non-rhythmic expression or change in expression level of Rev-erbα is affecting the core clock. For example, it has been shown that Rev-erbα knock-down cells are rhythmic as long as Rev-erb-beta is present. Thus, one major shortcoming of the current version of the manuscript is the missing dissection between Rev-erbα rhythmicity/expression and the circadian clock. More concretely, it remains unclear whether the change in Rev-erbα expression is a direct effect or caused by a defect clock. Since the authors presume a direct effect of YAP/TAP on Rev-erb expression, the former is likely. If that is the case, the data could be interpreted as that (missing) mechanic stimuli can lead to nuclear YAP/TAZ, which rises the level of Rev-erbα (and maybe interfere with its rhythmic accumulation). Beyond Rev-erbα expression, there may or may not be an effect on the circadian clock (core clock, CCGs). With the current version we do not know since the authors do not look beyond Rev-erbα expression. Thus, the claims on circadian clock or circadian rhythms in their cells is not studied in this version of the manuscript. The current version is still very interesting and provides insights into the Rev-erbα modulation, but additional work would be needed to show links with the core clock machinery. For this the authors could show influence (or at least correlation) of the YAP/TAZ/REVERBA phenotype on the oscillations of core clock genes or clock-controlled genes. Either through the use of alternative (ideally constitutive) reporters (e.g. PER2, BMAL1, fluorescent or LUC), or/and by analyzing RNA/Protein of core clock genes or output genes. This would not be necessary for all experiments, but at least for some were its possible (e.g. experiments with drugs perturbations). Otherwise, any claim like "YAP/TAZ perturbs the circadian clock ..." or "the circadian clock deregulation in nuclear YAP-enriched cells" is potentially flawed and has to be removed/reformulated.

    We agree with the reviewer. In order to understand if the core clock is affected, beyond REV-ERBA, by YAP/TAZ expression and localisation, we plan to perform the two experimental approaches explained below. For both of them we will use high-density cells with and without YAP-5SA overexpression since the other conditions (drugs, micropatterned cells, low density) may not render enough cells for analytical approaches that are not based on fluorescent microscopy (real-time qPCR or luminescence recordings). Also, the potential results obtained with YAP-5SA overexpression will be more informative regarding causality YAP-circadian clock than those using the other conditions described in the manuscript.

    1. We will use NIH3T3 bmal1::luc cells (already generated in our lab with the pABpuro-BluF plasmid; https://www.addgene.org/46824/) and an adapted microscopy-based system to track bioluminescence. We will need to give our cells a synchronisation shock since the single-cell signal with this reporter is too low and noisy to perform single-cell tracking.
    2. We will check during 48 hours, every 4 hours, the mRNA levels of Bmal1, Clock, Cry1, Per2, Yap1 and Rev-erbα via quantitative real-time PCR. As in A), we will need to synchronise our cells prior RNA collection. In case the expression of the other components of the clock are not affected by YAP-5SA overexpression, we will modify the message of our manuscript to emphasize the role of REV-ERBA. As the referee mentions (and we thank them for that comment), finding that the modulation of Rev-erbα is mechano-sensitive and dependent on YAP/TAZ signalling would be still very relevant, given the role of this factor in metabolism, inflammation, mitochondrial activity, or Alzheimer’s disease, as discussed in lines 231-235 in the manuscript.
    1. The authors aim to discard the possibility of paracrine signals by showing no increase in circadian power fraction of cells growing in low density with conditioned medium (Figure 2D). A paracrine signal coming from an oscillatory system is likely to oscillate and in that case, I do not see how growing cells in constant conditional medium can discard the effects of an oscillatory paracrine signal. I believe the elegant experiment shown in Figure 2E more precisely address this issue.

    The reviewer is right in the sense that paracrine coupling of circadian oscillators would require a circadian paracrine signal, like shown in Finger et al., 2021, and that we provide sufficient experimental evidence of a mechanics- rather than paracrine-driven control of the RevVNP circadian oscillations. Specifically, by using micropatterning (Fig. 2E) and gap closure (Fig. 2A) we show that cells under the same paracrine medium are able to display acute differences in RevVNP expression. The experiment with conditioned medium, which is a traditional technique used in some papers in the field like in Noguchi et al., 2013, was performed to rule out the possibility that secreted factors, even if not circadian, could ultimately impact the low-density cells’ circadian clock. We will rephrase the manuscript to stress out this reasoning.

    Finger AM, Jäschke S, del Olmo M, Hurwitz R, Granada AE, Herzel H & Kramer A (2021) Intercellular coupling between peripheral circadian oscillators by TGF-β signaling. Science Advances 7

    Noguchi T, Wang LL & Welsh DK (2013) Fibroblast PER2 circadian rhythmicity depends on cell density. Journal of Biological Rhythms 28: 183–192

    Data analysis methodology:

    1. Single-cell circadian recordings like the ones analyzed here are characterized by noisy amplitude and non-sinusoidal waveforms with fluctuating period (Bieler et al., 2014; Feillet et al., 2014). The authors interpolate, smooth, detrend and normalize their data; operations that are known to introduce spectral artifacts that can mislead the interpretation of the power spectrum. Moreover, the time-series pre-processing operations described by the authors in the methods sections is incomplete and the authors should more explicitly describe all their operations with exact methods applied, filter parameters and time-windows sizes (if applicable). To validate their pre-processing steps the authors could provide their time-series analysis pipeline code and/or provide a few examples of raw versus pre-processed data together with their respective spectrums before and after pre-processing. In addition, the authors could provide their raw trace signal data together with the corresponding post-processed signal data as plain text files.

    In our response to the reviewers, we will address this point exactly as requested by the reviewer. We will rewrite our methods section to explain better our analysis pipeline, clarifying that we do not apply detrending, that we resort rarely to interpolation of missing points, and stating the specifics of the standard low-pass filter we apply. We will then strengthen Supplementary Figure 1 with more examples of raw-data and processed data, and will provide raw trace signal data and the corresponding processed data to illustrate our approach.

    1. The authors rely on Fourier analysis and a reasonable self-made definition of circadian strength named as "circadian power fraction". Using a stationary-based method for noisy non-stationary data can lead to inaccurate spectrum power estimations. As the current version of the manuscript does not provide any alternative/complementary analysis method nor we have any available raw signal data it is unclear if their analysis appropriately represents the circadian power. The authors could consider implementing complementary data-analysis strategies to validate their conclusions. Fortunately, there are multiple suitable data analysis strategies already available that are exactly designed for this kind of data (eg. (Price et al., 2008; Leise et al., 2012; Leise, 2013; Bieler et al., 2014; Mönke et al., 2020). This time-series analysis methods is a crucial step as all main results on this manuscript rely on the authors self-made definition of circadian power. This is particularly important as there is no standardized method in the circadian field to estimate circadian rhythmicity and/or circadian power of single-cell traces.

    We will take this point into consideration by running a complementary analysis of our data with one of the methods recommended by the reviewer. Our choice is pyBOAT, as presented in Mönke et al. (2020), because on first inspection its implementation of the wavelet method appears to be the most suitable for our dataset type. If we find that our time-series are too short for these methods we will use the RAIN algorithm (Thaben and Westermark, 2014) instead.

    Mönke G, Sorgenfrei FA, Schmal C & Granada AE (2020) Optimal time frequency analysis for biological data - pyBOAT Systems Biology

    Thaben PF & Westermark PO (2014) Detecting Rhythms in Time Series with RAIN. J Biol Rhythms 29: 391–400

    1. The authors mainly show circadian power fraction and analyze rhythmicity scores/powers. Is there the a chance that a rise in the basal expression level of Rev-erbα is reducing the rhythmicity score? Or to phrase it otherwise, the absolute amplitude may remain the same, but the relative amplitude may be reduced? Would that affect the FT analysis power scored? To clarify this the authors could provide an analysis of the relative amplitude in addition to the circadian intensity (as in Fig.1C).

    Our analysis pipeline subtracts the mean signal from each cell’s intensity-time trace, and then divides each trace by its standard deviation. This procedure eliminates any bias due to basal expression of Rev-erbα. We will address this point by clarifying the methods section and providing examples in Supplementary Figure 1 of raw data with high-basal levels and low basal levels, showing their pre- and post-processed spectra.

    Minor points by text-line:

    YAP and TAZ should be introduced to the reader during introduction. by set a of proteins. Here the authors probably meant that cells were not reset nor entrained during the experiment. "..expression depends on..". This is a correlation, not proof of causation is shown until this point. This is an overstatement. Using the term "provoked" suggests a causal relationship not shown. Similarly last sentence "This result established.... is caused..". Again, this is an overstatement as only correlation is shown. According to their description the authors are not using any image-preprocessing steps, eg background subtraction or other filters. Is this correct? It is not clear what image metric for the single-cell signals are the authors using, eg. integrated nuclear intensity or mean/median nuclear intensity. I am not familiar with TrackMate but it might be possible to export and share with the readers the image-analysis pipeline used which would clarify any questions about image processing and signal extraction.

    We thank the reviewer for pointing out all these minor points. We will address each one of them to make the paper clearer.

    Regarding Reviewer #3:

    The authors state in lines 163-165: 'This striking anticorrelation reveals that the robustness of the Rev-erbα circadian expression depends on the nucleocytoplasmic transport of YAP and its mechanosensitive regulation'. Although interesting, the data in figure 3 to which this statement refers is, as the authors identify, correlative, rather than causative. I would strongly suggest altering this statement to better reflect the data.

    We will modify the text to eliminate this overstatement.

    It looks to me as though all experiments were carried out in the same clonal reporter 3T3 line. To avoid possible issues with founder effects, I would ask that the authors repeat the initial experiment in figure 1B, and the associated analysis as in 1C-E with a different clonal 3T3 line. Hopefully this will not be very arduous, as the methods suggest that multiple clonal 3T3 reporter lines were made initially. With time to defrost, plate, record and analyse the data, I would hope that this would not take more than six weeks maximum.

    We will perform the experiments regarding the cell density effect on the RevVNP oscillations (Fig. 1) in another clonal 3T3 line as the reviewer suggests. We have already initiated the experimental repeats with the alternative clone.

    I would note that the custom software used for analysis does not appear to be generally available. I would assume that the authors would make this available upon request.

    We will extend the explanation of our method as suggested by Reviewer #2 and make the code available to the community.

    Experiments appear to have been adequately replicated in terms of n. However, the robustness of these findings would be supported though use of a different clonal reporter line, as discussed above.

    We will solve this problem as stated above.

    Statistical analysis is generally appropriate. I would suggest including statistical analysis in figures 3B and S4B to demonstrate that the pharmacological treatments are indeed having a statistically significant effect on the MAL and YAP nuclear/cytoplasmic ratio.

    We will perform the corresponding statistical analysis on those data.

    For Figure 4, it is not stated which statistical tests have been used, with only P values given in table S1. Please state which test has been used.

    We will specify the statistical test used in the figure legend.

    Furthermore, it would be valuable to see if it is possible to perform statistical analysis looking at the populations should in Figure 4A, to either support or refute the statement made in Line 189-90 that 'we overexpressed 5SA-S94A-YAP, a mutant version of YAP unable to interact with TEAD and observed that the cells recovered, to a large extent, both the RevVNP circadian power fraction and the REV-ERBα basal levels displayed by the wild-type high-density population'

    The p-values corresponding to that dataset are represented in Table S1, but we will move them to the figure legend so the extent of the differences between the YAP mutants and the control becomes more noticeable. This applies too to the next comment of the reviewer.

    Additionally, it is a little unclear to me why exact p values are reported in table S1. It seems that they might be better placed in the relevant figure legend.

    Minor comments:

    Although the authors took good care to try to ensure that there was minimal phase synchrony between cells, it would be good to see some analysis to confirm that these efforts were successful. This is of particular concern, given that many things that commonly happen during cell handling, such as temperature change and media change, even with conditioned media, can act to synchronise cells. Hopefully, this information should be available from your existing analysis.

    All our experiments, except for the gap closure ones (which imply an unavoidable medium shock after the removal of the gasket where the cells are cultured to achieve high density) are carried out in a similar way (see Materials and Methods). This approach does not involve the typical shock of serum, dexamethasone, or other hormones, because we want to avoid biochemical signalling that could mask the “pure” effect of mechanics on the pathways that affect the circadian clock. In any case, a certain level of synchrony should not affect the analysis we perform, since this is single cell-based and does not consider the phase but the strength of its circadian frequency. But as requested by the reviewer we will analyze the phase signal and report the results if relevant to the project.

    It would be informative to see both phase and period analysis for the data shown in figure 2C. Do cells at the edge show differences in relative synchrony following the removal of the PDMS barrier and Rev-erba induction? Is there a period difference between cells at the edge and those that remain confluent?

    We agree with the referee that the “shock” received by the cells at the edge should work as a reset of their circadian phase and we have tried to analyse this effect. However, there are technical limitations that make this analysis difficult, mainly the short duration of the experiment and the fact that these cells transition very fast, upon gap closure, from a non-circadian to a circadian behaviour. We will attempt to better report this interesting effect by using the WAVECLOCK (Price et al., 2008) or the pyBOAT method (Mönke et al., 2020), suggested by Reviewer #2, which are designed to analyse non-stationary data.

    Mönke G, Sorgenfrei FA, Schmal C & Granada AE (2020) Optimal time frequency analysis for biological data - pyBOAT Systems Biology

    Price TS, Baggs JE, Curtis AM, Fitzgerald GA & Hogenesch JB (2008) WAVECLOCK: wavelet analysis of circadian oscillation. Bioinformatics 24: 2794–2795

    Figure 2B - the text states that those cells far from the edge oscillate robustly thoughout the experiment, but this is not easy to see from this kymograph due to the dynamic range. Is there another way of presenting this that might make it easier to confirm?

    We will calculate the circadian power fraction of the “bulk” cells as we do for the other conditions described in the manuscript. We can also show examples of individual traces if the average shown in Fig. 2C or the kymograph in Fig. 2B are not clear enough.

    Figure 1D-E - the text provides periodicity for the high-density cells, but not the low density ones. Could you provide periodicity for both populations - do they differ?

    We will represent in more detail the results of the frequency analysis on the low-density cells so the diversity of periods (frequencies) at this condition gets more evident.

    Figure S3 - it is interesting to note the difference in population rhythmicity between the bulk and edge data here, which is not seen so clearly in cells without thymidine. Could the authors comment on this?

    We agree with the referee that there is an obvious difference regarding RevVNP expression (mainly on the edge cells but also in the bulk) between the experiments with and without thymidine. We hypothesise this is due to the pronounced decrease in cell divisions in the presence of thymidine, which considerably slows down the gap closure and impacts the density of the entire cell population. We will comment this effect in the manuscript.

    Line 148 - it is unclear here what is meant by 'the onset of circadian oscillations'. Could you rephrase this for clarity?

    We will change that sentence.

    Line 173 - a few words to highlight that Lats is a kinase and the function of YAP phosphorylation by Lats would aid clarity here. Similarly, explanation of the functional difference between the protein with 4 Serine to alanine mutations and 5 mutations and why both of these mutants were used would be helpful.

    We will clarify this point following the reviewer’s suggestion.

    Line 174 - for accuracy, this should perhaps read 'fibroblast circadian clock', as this work is only in 3T3 cells, and therefore may not apply more generally.

    We will implement this change.

    Line 202 - could you expand to explain the existing limitations of studying cell signalling cascades in synchronised cells? This is not clear to me. Thanks.

    We will discuss the signalling effects caused by 50% serum shocks and other traditional ways to synchronise the cells as requested by the reviewer.

    Figures 1D and 4B - the choice of colour range used in these kymographs is skewed towards the warmer colours, making it quite hard to discern differences between the groups. I would suggest using the cooler colour range for a greater proportion of the data set, to make rhythmicity, or lack of it, clearer to see.

    We will invest further efforts to finding the optimal colour map and range for our datasets.

  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 #3

    Evidence, reproducibility and clarity

    Summary:

    In this study, the authors employ the NIH 3T3 fibroblast cell line to study the effect of cell density and the associated mechanical cues on cellular circadian rhythmicity. For this, they generate a Rev-erbα:VENUS line and, combined with constitutive nuclear mCherry expression, are able to track Rev-erbα: expression in single cells within populations of differing densities. Using overexpression of the transcriptional co-regulator YAP and specific mutants thereof, they suggest a role for YAP and its associated transcription factor family, TEAD, in the regulation of Rev-erbα expression under conditions of differing cell density.

    Major comments:

    • Are the key conclusions convincing? Yes, the major conclusions are supported by the data shown.

    • Should the authors qualify some of their claims as preliminary or speculative, or remove them altogether?

    The authors state in lines 163-165: 'This striking anticorrelation reveals that the robustness of the Rev-erbα circadian expression depends on the nucleocytoplasmic transport of YAP and its mechanosensitive regulation'. Although interesting, the data in figure 3 to which this statement refers is, as the authors identify, correlative, rather than causative. I would strongly suggest altering this statement to better reflect the data.

    • Would additional experiments be essential to support the claims of the paper? Request additional experiments only where necessary for the paper as it is, and do not ask authors to open new lines of experimentation.

    • Are the suggested experiments realistic in terms of time and resources? It would help if you could add an estimated cost and time investment for substantial experiments.

    It looks to me as though all experiments were carried out in the same clonal reporter 3T3 line. To avoid possible issues with founder effects, I would ask that the authors repeat the initial experiment in figure 1B, and the associated analysis as in 1C-E with a different clonal 3T3 line. Hopefully this will not be very arduous, as the methods suggest that multiple clonal 3T3 reporter lines were made initially. With time to defrost, plate, record and analyse the data, I would hope that this would not take more than six weeks maximum.

    • Are the data and the methods presented in such a way that they can be reproduced?

    Yes. I would note that the custom software used for analysis does not appear to be generally available. I would assume that the authors would make this available upon request.

    • Are the experiments adequately replicated and statistical analysis adequate?

    Experiments appear to have been adequately replicated in terms of n. However, the robustness of these findings would be supported though use of a different clonal reporter line, as discussed above.

    Statistical analysis is generally appropriate. I would suggest including statistical analysis in figures 3B and S4B to demonstrate that the pharmacological treatments are indeed having a statistically significant effect on the MAL and YAP nuclear/cytoplasmic ratio.

    For Figure 4, it is not stated which statistical tests have been used, with only P values given in table S1. Please state which test has been used.

    Furthermore, it would be valuable to see if it is possible to perform statistical analysis looking at the populations should in Figure 4A, to either support or refute the statement made in Line 189-90 that 'we overexpressed 5SA-S94A-YAP, a mutant version of YAP unable to interact with TEAD and observed that the cells recovered, to a large extent, both the RevVNP circadian power fraction and the REV-ERBα basal levels displayed by the wild-type high-density population'

    Additionally, it is a little unclear to me why exact p values are reported in table S1. It seems that they might be better placed in the relevant figure legend.

    Minor comments:

    • Specific experimental issues that are easily addressable. Although the authors took good care to try to ensure that there was minimal phase synchrony between cells, it would be good to see some analysis to confirm that these efforts were successful. This is of particular concern, given that many things that commonly happen during cell handling, such as temperature change and media change, even with conditioned media, can act to synchronise cells. Hopefully, this information should be available from your existing analysis.

    It would be informative to see both phase and period analysis for the data shown in figure 2C. Do cells at the edge show differences in relative synchrony following the removal of the PDMS barrier and Rev-erba induction? Is there a period difference between cells at the edge and those that remain confluent?

    • Are prior studies referenced appropriately?

    To the best of my knowledge, yes.

    • Are the text and figures clear and accurate?

    Figure 2B - the text states that those cells far from the edge oscillate robustly thoughout the experiment, but this is not easy to see from this kymograph due to the dynamic range. Is there another way of presenting this that might make it easier to confirm?

    Figure 1D-E - the text provides periodicity for the high-density cells, but not the low density ones. Could you provide periodicity for both populations - do they differ?

    Figure S3 - it is interesting to note the difference in population rhythmicity between the bulk and edge data here, which is not seen so clearly in cells without thymidine. Could the authors comment on this?

    Line 148 - it is unclear here what is meant by 'the onset of circadian oscillations'. Could you rephrase this for clarity?

    Line 173 - a few words to highlight that Lats is a kinase and the function of YAP phosphorylation by Lats would aid clarity here. Similarly, explanation of the functional difference between the protein with 4 Serine to alanine mutations and 5 mutations and why both of these mutants were used would be helpful.

    Line 174 - for accuracy, this should perhaps read 'fibroblast circadian clock', as this work is only in 3T3 cells, and therefore may not apply more generally.

    Line 202 - could you expand to explain the existing limitations of studying cell signalling cascades in synchronised cells? This is not clear to me. Thanks.

    • Do you have suggestions that would help the authors improve the presentation of their data and conclusions?

    Figures 1D and 4B - the choice of colour range used in these kymographs is skewed towards the warmer colours, making it quite hard to discern differences between the groups. I would suggest using the cooler colour range for a greater proportion of the data set, to make rhythmicity, or lack of it, clearer to see.

    Significance

    • Describe the nature and significance of the advance (e.g. conceptual, technical, clinical) for the field.

    This work provides a potential mechanism for the modulation of cellular rhythmicity under conditions of varying cell density, a currently relatively understudied area to which the contribution made here will be valuable. However, the work is limited by its use of only one cell type (NIH 3T3) and one reporter (Rev-erb:VENUS), which makes the work difficult to generalise in the context of all cell types and environments that exist in a mammalian context.

    • Place the work in the context of the existing literature (provide references, where appropriate).

    Previous work has identified YAP activity as a mechanism of signalling growth substrate stiffness (Halder et al. 2012, Panciera et al. 2017). It has also been speculated, but not demonstrated, that YAP might influence circadian rhythmicity (Streuli and Meng, 2019). This work provides some initial evidence to support this speculation.

    • State what audience might be interested in and influenced by the reported findings.

    This work would be of genera; interest to those working on mammalian cellular circadian rhythmicity. Additionally, given YAP's status as an oncogene, this work would also be relevant to those considering circadian disruption in cancer.

    • Define your field of expertise with a few keywords to help the authors contextualize your point of view. Indicate if there are any parts of the paper that you do not have sufficient expertise to evaluate.

    Mammalian circadian cell biology and biochemistry.

  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 #2

    Evidence, reproducibility and clarity

    Summary:

    Abenza et al. investigate an important question of how the physical environment affects the properties of the individual circadian clocks. The authors utilize a set of clever experiments, pharmacological manipulations and data analysis techniques to unveil a potential role of YAP/TAZ in the circadian clock.

    Major comments:

    Effects on the circadian clock

    1. The authors use the fluorescent reporter created by Nagoshi from sections of the Rev-erbα gene. This reporter is widely used to estimate relative circadian timing in individual cells but it does not provide direct information on the circadian clock activity. In other words, while Reverb rhythmic expression is driven by the clock, it is not known whether less-rhythmic or non-rhythmic expression or change in expression level of Rev-erbα is affecting the core clock. For example, it has been shown that Rev-erbα knock-down cells are rhythmic as long as Rev-erb-beta is present. Thus, one major shortcoming of the current version of the manuscript is the missing dissection between Rev-erbα rhythmicity/expression and the circadian clock. More concretely, it remains unclear whether the change in Rev-erbα expression is a direct effect or caused by a defect clock. Since the authors presume a direct effect of YAP/TAP on Rev-erb expression, the former is likely. If that is the case, the data could be interpreted as that (missing) mechanic stimuli can lead to nuclear YAP/TAZ, which rises the level of Rev-erbα (and maybe interfere with its rhythmic accumulation). Beyond Rev-erbα expression, there may or may not be an effect on the circadian clock (core clock, CCGs). With the current version we do not know since the authors do not look beyond Rev-erbα expression. Thus, the claims on circadian clock or circadian rhythms in their cells is not studied in this version of the manuscript. The current version is still very interesting and provides insights into the Rev-erbα modulation, but additional work would be needed to show links with the core clock machinery. For this the authors could show influence (or at least correlation) of the YAP/TAZ/REVERBA phenotype on the oscillations of core clock genes or clock-controlled genes. Either through the use of alternative (ideally constitutive) reporters (e.g. PER2, BMAL1, fluorescent or LUC), or/and by analyzing RNA/Protein of core clock genes or output genes. This would not be necessary for all experiments, but at least for some were its possible (e.g. experiments with drugs perturbations). Otherwise, any claim like "YAP/TAZ perturbs the circadian clock ..." or "the circadian clock deregulation in nuclear YAP-enriched cells" is potentially flawed and has to be removed/reformulated.

    2. The authors aim to discard the possibility of paracrine signals by showing no increase in circadian power fraction of cells growing in low density with conditioned medium (Figure 2D). A paracrine signal coming from an oscillatory system is likely to oscillate and in that case, I do not see how growing cells in constant conditional medium can discard the effects of an oscillatory paracrine signal. I believe the elegant experiment shown in Figure 2E more precisely address this issue.

    Data analysis methodology:

    1. Single-cell circadian recordings like the ones analyzed here are characterized by noisy amplitude and non-sinusoidal waveforms with fluctuating period (Bieler et al., 2014; Feillet et al., 2014). The authors interpolate, smooth, detrend and normalize their data; operations that are known to introduce spectral artifacts that can mislead the interpretation of the power spectrum. Moreover, the time-series pre-processing operations described by the authors in the methods sections is incomplete and the authors should more explicitly describe all their operations with exact methods applied, filter parameters and time-windows sizes (if applicable). To validate their pre-processing steps the authors could provide their time-series analysis pipeline code and/or provide a few examples of raw versus pre-processed data together with their respective spectrums before and after pre-processing. In addition, the authors could provide their raw trace signal data together with the corresponding post-processed signal data as plain text files.

    2. The authors rely on Fourier analysis and a reasonable self-made definition of circadian strength named as "circadian power fraction". Using a stationary-based method for noisy non-stationary data can lead to inaccurate spectrum power estimations. As the current version of the manuscript does not provide any alternative/complementary analysis method nor we have any available raw signal data it is unclear if their analysis appropriately represents the circadian power. The authors could consider implementing complementary data-analysis strategies to validate their conclusions. Fortunately, there are multiple suitable data analysis strategies already available that are exactly designed for this kind of data (eg. (Price et al., 2008; Leise et al., 2012; Leise, 2013; Bieler et al., 2014; Mönke et al., 2020). This time-series analysis methods is a crucial step as all main results on this manuscript rely on the authors self-made definition of circadian power. This is particularly important as there is no standardized method in the circadian field to estimate circadian rhythmicity and/or circadian power of single-cell traces.

    3. The authors mainly show circadian power fraction and analyze rhythmicity scores/powers. Is there the a chance that a rise in the basal expression level of Rev-erbα is reducing the rhythmicity score? Or to phrase it otherwise, the absolute amplitude may remain the same, but the relative amplitude may be reduced? Would that affect the FT analysis power scored? To clarify this the authors could provide an analysis of the relative amplitude in addition to the circadian intensity (as in Fig.1C).

    Minor points by text-line:

    1. YAP and TAZ should be introduced to the reader during introduction.
    2. by set a of proteins.
    3. Here the authors probably meant that cells were not reset nor entrained during the experiment.
    4. "..expression depends on..". This is a correlation, not proof of causation is shown until this point. This is an overstatement.
    5. Using the term "provoked" suggests a causal relationship not shown.
    6. Similarly last sentence "This result established.... is caused..". Again, this is an overstatement as only correlation is shown.
    7. According to their description the authors are not using any image-preprocessing steps, eg background subtraction or other filters. Is this correct?
    8. It is not clear what image metric for the single-cell signals are the authors using, eg. integrated nuclear intensity or mean/median nuclear intensity. I am not familiar with TrackMate but it might be possible to export and share with the readers the image-analysis pipeline used which would clarify any questions about image processing and signal extraction.

    References:

    1. Bieler, J, Cannavo, R, Gustafson, K, Gobet, C, Gatfield, D, and Naef, F (2014). Robust synchronization of coupled circadian and cell cycle oscillators in single mammalian cells. Mol Syst Biol 10, 739.

    2. Leise, TL (2013). Wavelet analysis of circadian and ultradian behavioral rhythms. J Circadian Rhythms 11, 5.

    3. Leise, TL, Wang, CW, Gitis, PJ, and Welsh, DK (2012). Persistent Cell-Autonomous Circadian Oscillations in Fibroblasts Revealed by Six-Week Single-Cell Imaging of PER2::LUC Bioluminescence. PLoS One 7, 1-10.

    4. Mönke, G, Sorgenfrei, F, Schmal, C, and Granada, A (2020). Optimal time frequency analysis for biological data - pyBOAT. BioRxiv 179, 985-986.

    5. Price, TS, Baggs, JE, Curtis, AM, FitzGerald, GA, and Hogenesch, JB (2008). WAVECLOCK: wavelet analysis of circadian oscillation. Bioinformatics 24, 2794-2795.

    Significance

    I believe this manuscript is of high significant both for the circadian as well as the mechanobiology fields. Readers from single-cell signalling studies will also be very interested in this work.

    To my knowledge the discussed link has not been studied before at single cell level, which as the authors show can provide multiple new insights.

    I do work with similar single-cell signals, have broad expertise in microscopy, image analysis methods, time series analysis, and the circadian clock mechanisms but very little experience in mechanobiology.

  5. 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

    Here Abenza & Rossetti et al. show that in 3T3 fibroblasts, the circadian clock depends on cell density, correlates with YAP activity, and further demonstrate that circadian power fraction is suppressed by genetic YAP activation (5SA), but is rescued by expression of 5SA YAP without the tead-binding domain. This is a striking study on an important question; however, the data do not directly support the conclusions and the title of the paper. These conclusions/title should be altered, or supported with additional experiments, as detailed below:

    Major Critiques:

    1. On "mechanical control": The authors show changes in circadian power fraction with changes in YAP and with cytoskeletal inhibitors, but there are no properly-controlled experiments that directly perturb mechanics. The authors show a correlation between YAP nuclear/cytoplasmic ratio and circadian power, but YAP N/C alone is not a readout of mechanotrasndcution, per se. The authors have shown two different experiments where cells are cultured on a stiff (30kPa) substrate and soft substrate (300Pa), but they do not shown a direct comparison of YAP nuclear localization and circadian power under these two conditions in the same experiment. Direct, controlled perturbation of mechanical cues is necessary to support the title's use of the phrase "mechanical control."

    2. On "via YAP/TAZ": In addition to above, it is necessary to show that the changes in Circadian power fraction induced by mechanical cues in fact require YAP/TAZ signaling. Thus, an experiment comparing soft (300Pa) substrate with Stiff (30kPa) substrate in the presence or absence of YAP/TAZ is necessary to state that YAP and TAZ are the mechanistic mediators of mechanical cues on the clock.

    3. While the TEAD-binding domain mutant experiment is elegant, to claim that TEAD is the transcriptional mediator, it must be demonstrated that this mutant indeed fails to induce TEAD-mediated transcription. This could be simply executed by demonstrating that the CCD mutant expresses reduced CTGF and Cyr61 (for example), compared to the 5SA, under these conditions. Further, endogenous YAP is still active and available to bind to TEAD in this system, which should be discussed.

    4. In Figure 3a: The cell perimeter needs to be shown either by actin staining or by brightfield images. The manually marking of cell boundaries is insufficient, specifically because the drugs used in this experiment affect the cytoskeleton. It would be very helpful to see this via actin staining or in the least with brightfield images.

    Significance

    This is an exciting paper that potentially links mechanotransduction to the circadian clock. While my group is not focused on circadian rhythms, and I don't have the background to comment on the measurements or robustness of circadian power, the idea is striking and significant.

    I strongly recommend inclusion of loss-of-function approaches (either genetic or pharmacologic) in addition to the gain-of-function methods employed here to support the necessity of YAP/TAZ signaling. Also, appropriately controlled experiments to show true mechanical effects on the circadian clock are necessary.

  6. Version published to 10.1101/2022.02.04.478830v2 on bioRxiv
  7. Version published to 10.1101/2022.02.04.478830v1 on bioRxiv