Comparative analysis of actin visualization by genetically encoded probes in cultured neurons

This article has been Reviewed by the following groups

Read the full article See related articles

Listed in

Log in to save this article


Actin cytoskeleton predominantly regulates the formation and maintenance of synapses by controlling dendritic spine morphology and motility. To visualize actin dynamics, actin molecules can be labelled by genetically fusing fluorescent proteins to actin monomers or using fluorescently tagged actin-binding proteins or single-chain anti-actin antibodies. However, the effects of these labels on the morphology of neurons have not been quantitatively compared yet. In the present study, we analysed Actin-Chromobody-GFP, LifeAct-GFP and EGFP-actin with respect to their effects on actin-related features in mouse cultured hippocampal neurons.

The actin-binding probes LifeAct and Actin-Chromobody showed similar affinity to F-actin, and along with EGFP-actin, were enriched in dendritic protrusions. In contrast to EGFP-actin, neither of these constructs was able to detect subtle changes in actin remodelling between mature mushroom shaped spine and less developed filopodia. None of the compared probes altered filopodial motility compared to control EGFP expression, however, within 24 hours expression, minor changes in dendritic spine morphology and density were visible. Furthermore, while EGFP-actin and LifeAct-GFP expression did not alter dendritic arborization, AC-GFP expressing neurons displayed a reduced dendritic arborization. We therefore conclude that careful consideration of cellular consequences is required before performing experiments with a particular actin labelling probe in primary neurons.

Article activity feed

  1. Review coordinated via ASAPbio’s crowd preprint review

    This review reflects comments and contributions by Ashley Albright, Luciana Gallo, Sam Lord, Dipika Mishra, Claudia Molina, Arthur Molines, Sónia Gomes Pereira, Parijat Sil, Rinalda Proko. The comments were synthesized by Richa Arya.

    The reviewers like the motivation behind the study as a lot is still unknown about the impact of fluorescent tags on various mechanisms in biology. The work is impactful. However we outline below some major questions and several minor points:

    1. Related to the data analysis:

    The findings are valuable however the analyses may not be sufficiently sensitive to pick up morphological changes. Maybe other more sensitive approaches for measuring interference in the biology of these neurons could also be tested, like bulk growth rates, a stimulus added to the culture medium or other?

    Some of the phenotypes (see Figure 1D and Figure 3D) are relatively subtle and the manuscript relies heavily on statistics to support the claims. Independent of the statistics, the differences are not striking by eye examination. Perhaps more data is necessary to bolster some of the reported claims.

    As continuation to the expression analysis, it is important to estimate the expression levels of the actin binding probes used, in order to rule out the fact that some of the observed differences between LifeACT-GFP and AC-GFP may be due to discrepancies in the extent of overexpression of these probes. It would greatly add to the study to include, at least for some of the phenotypes, whether the measured parameters respond to the low versus high expression levels of the same probe.

    1. Related to the transient expression:

    Figure 1: The transient expression method used in the manuscript shows a lot of variability in expression levels, between cells, and between replicates.

    Expression levels could confound the interpretation. One of the constructs could be expressed more or less than the other, resulting in stronger or weaker phenotypes, not because it is more or less toxic than the other per se but because its expression level is different. It would be relevant to "normalize" for the expression level of each construct. Another way to circumvent this, at least partially, would be to substantially increase the number of cells analyzed, which would allow for a range of expression values to be represented in the data.

    1. Related to FRAP analysis Last paragraph result 1: 'and depends only on their affinity to F-actin, that is similar in AC-GFP and LifeAct-GFP (Figure 1A, Figure 1B, Supplementary Video 1). In addition, our results suggest that even without photomanipulation, minor changes in actin reorganization cannot be revealed when using actin-binding probes..’- Based on the images reported, it is not possible to establish how much of the signal is due to the population of probes being bound to actin versus the population that is free floating in the cytoplasm. The recovery could be due to the diffusion of free-floating probes and therefore give no information about affinity for actin. EGFP alone was used as a baseline for cytoplasmic diffusion, the slower recovery from the EGFP-actin implies that some portion of the EGFP-actin is incorporated in filaments. Recommend replacing "Affinity" with "relative ability to incorporate into filaments." A possibility to address the issue of size-based diffusion in cytoplasm is to complete FRAP measurements in latrunculin-treated cells that depolymerize most of the actin filaments. This will enable to set a baseline for each of the probes here (which will now probably be either free or G-actin bound) and provide a complement to the Jasplakinolide treatment.

    ‘In addition, our results suggest that even without photomanipulation, minor changes in actin reorganization cannot be revealed when using actin-binding probes.’- There has been only two actin binding probes tested, both with similar turnover as measured in FRAP in their own assay. It might be worth making a comparison in this experiment with a very strong actin binding probe as control, such as Utrophin.

    1. Figure 2: Theat measurement shown is not a very good proxy for filopodia motility.

    The study used an intensity-weighted center of mass. This means that the center of mass moves, not only because the shape of the filopodia changes but also because the signal intensity changes. In other words, the shape of a filopodia could be constant (no motility) and yet have a center of mass that moves because the mCherry signal fluctuates inside it. This could be avoided if the center of mass of the shape is used, not weighed by intensity. This is especially a concern because the signal from the cytoplasmic mCherry is used for the analysis. If a folipodia locally thickens in the Z-direction, the cytoplasmic signal will increase locally, displacing the intensity weighted center of mass even if the 2D contour has not changed. Using a membrane signal would provide a better alternative. It would also be possible to make use of the resource Filotracker, that tracks the length of the filopodia as a measure of filopodia dynamics. Find the paper and the resource here:,

    1. Result 2 last para, ‘We found no significant difference in center of mass displacement between actin probe expressing cells and EGFP expressing control filopodia (Figure 2B)…’. This section needs more clarity and evidence to conclude that the probes do not alter filopodia dynamics. Maybe filopodia growth rate or some additional measurements? Failing to find significance does not equate to finding evidence of absence. It may be that this one parameter is not sufficiently sensitive. Maybe this possible uncertainty should be discussed in the last sentence of the paragraph, to note that the data highlights the possibility that the tested actin labeling proteins do not interfere.

    Minor Comments

    • Introduction: ‘Actin is a key cytoskeletal element in mammalian cells involved in many cellular mechanisms’. mammalian cells can be replaced with eukaryotic cells. It would also be nice to mention some of the cellular mechanisms involved such as cell division, and migration, among others.
    • Introduction: it would be good to describe the various phenotypes observed in previous studies when actin was labeled or when actin-binding proteins were used. It would give readers context about the level of toxicity and what phenotypes to expect.
    • Introduction last paragraph: ‘…and to exclude certain actin structures from labeling (Munsie et al., 2009)’: one more reference could be added for this statement: Sanders et al., 2013
    • Figure1: It would be nice to have grayscale images of the actin channel in addition to the overlay.
    • Figure 1 B, C: For all of the FRAP recovery curves, recommend providing insets, zooming in on the first 30 to 60 sec of the recovery, as that's when most of the recovery happens. The last 120 sec of the plots show a "flat" plateau.
    • Figure 1D, in the fluorescence recovery plateau %."In addition, our results suggest that even without photo manipulation, minor changes in actin reorganization cannot be revealed when using actin-binding probes." This claim relies on the p-values. In looking at Figure 1D, left panel, EGFP-actin (orange) dots there appears to be an outlier. Independent of the outlier, the collection of dots does not appear that different by eye, recommend providing additional data to support this claim.
    • Result 1: ‘In Jasplakinolide-treated neurons, as expected, we observed an almost immediate recovery of fluorescence in the EGFP expressing group, whereas the EGFP-actin signal did not recover…’. The fact that the EGFP-actin signal did not recover is surprising. Normally not all of the actin present in the protrusion is incorporated into filaments, some of it is floating around freely. Hence, some of the signal should be recovering, even after stabilization of the actin filament, simply due to diffusion. For example, the EGFP signal recovers in presence of Jasp. due to diffusion of the free-floating probe. Recommend some discussion about the absence of recovery for the EGFP-actin.
    • Figure 2A: ‘Red lines show the movement track of intensity weighted center of mass..’. The red dots for the center of mass cluster and overlap, recommend color coding the dots so that it is clear visually what the displacement of the center of mass was and showing an overlay of the contours used for the analysis. Additionally, in the Supplementary Video 2 it looks like EGFP and EGFP-actin centers of mass are more displaced than AC-GFP or LifeAct-GFP. It would be good to clarify if this is exactly the same example as shown in the figure.
    • Figure 2B: ‘Average intensity-weighted center of mass displacement over 60s time periods…’ Why was only a 60 sec interval considered when there are images up to 120 sec and the video goes until 180 sec? Additionally, please specify if these are the first 60 sec of imaging.
    • Result 3: It is known that expression levels of actin binding probes can alter actin structures and their dynamics. It would have been great to do the following: (a) estimate the levels of expressed lifeact-GFP/AC-GFP and see how they compare with each other, (b) note or look for phenotypic differences as a function of the expression levels of these probes. It might be worth plotting the spine morphometric data in categories of low, medium and high expression levels of the two actin binding probes as well as EGFP-Actin (since this can affect nucleation/treadmilling etc at very high expression levels). Just as the identity of the actin binding probe being used is an important consideration in studies of actin dynamics, so is the expression levels of these probes.
    • Result 3: use p-values to compare different cell lines, the n used in the statistics should be the number of samples, not the number of spines.
    • Result 3: ‘This is like due to the known high background fluorescence level of LifeAct, originating from its affinity to G-actin (Melak, Plessner and Grosse, 2017)…’. Actin chromobody is also known to bind G actin. Is there a significant difference in G Actin binding affinity for LifeACT versus AC that can account for this explanation?
    • Figure 3C ‘Expression of EGFP-actin or LifeAct-GFP for 24 hours did not influence total protrusion density’ - Please indicate whether these morphological analyses were done blinded as to what the cells were expressing, or any steps taken to reduce bias.
    • Figure 3D: There is a similar concern here as for Figure 1D. Here the number of cells is higher, but the density of the points is not shown. By eye the box plots do not look very different, violin plots may be better for these data so that the distribution of data points is more apparent.
    • Figure 3F: it would be useful to have a representative image of each (stubby, thin, and mushroom) class, to help non-experts better visualize what's being analyzed .
    • Result 4, paragraph 1, ‘..whether dendritic arborization is altered within 24 h after the transfection of the tested probes…’ All the experiments were performed 24h after transfection, would it be worth testing different time intervals (e.g. 12-16h and/or 48h)?
    • Result 4C,D E: suggest adding quantification to enhance the data.
    • Materials and Methods section, ‘Live cell imaging and FRAP experiments, post-bleach in every case (Supplementary Video 2)…’. Should this read video 1.
    • Materials and Methods section, ‘Live cell imaging and FRAP experiments,Then, cumulative displacement curves were calculated, and the 60 sec points were compared and statistically analysed (Supplementary Video 1)…’. Should this read video 2.
    • Materials and Methods section: There are several custom-made plugins used in this work. It is good practice to make these available to the community by depositing them in a repository (e.g. GitHub, zenodo).