Curious what the ciliation levels are in MG132 treated serum starved cells in these experiments. There is a difference between ciliation percentage in the RPGRIP1l mutants vs wild type as in Figure 1C which also show a different percentage of CC3+ PCD in figure 1D but perhaps no difference in CC3+ PCD between the mutant and wild-type (RPGRIP1l +/+ w MG132 and RPGRIP1l -/- w MG132) in the presence of MG132 (though these are not shown on the same graph and perhaps not done in the same experiment?)

For this and subsequent figures, some quantification or multiple cells rather than images of a single cell would be more convincing (even just a line scan of fluorescence intensity along the length of the flagella in each channel, for spatial coincidence). Single channel grayscale would also make it visually more clear. It’s a bit difficult to discern as shown.

There is also an IFT20-mCherry strain available in the chlamy center. It’s expressed in the ift20 mutant background. Curious also what the chr1 localization is in ift20 mutants.

Since you have pre-immune serum, I’m guessing you generated this chlamy-specific ift-20 antibody yourself? Wild type and IFT-20 mutant (either ift20-1 or a CLiP mutant) lysate by side on the same blot would be a convincing demonstration of antibody specificity.

What is the source of these antibodies?

Curious if you’ve tried expressing any hyperactive mutants of HSP104 to see if increased activity has similar effects to overexpression or ATPase dead mutants to distinguish between enzyme activity vs other interaction effects of overexpression.

Hi, I was interested in the full dataset but it looks like the data either have not yet been deposited or the placeholders have accidentally not been replaced here. I was also unable to also find them via an admittedly cursory search directly in the appropriate repos (recognizing that the US government shutdown may be delaying updating of NCBI).

1. This is super interesting work! Wonder if you’re considering the flip side: to use either phylogenetic comparative methods to hypothesize candidates that could give rise to the extremophilic properties of pacifica/other extremophiles vs reinhardtii/others such that they can be tested/targeted in reinhardtii or screening of reinhardtii mutants that share growth conditions with pacifica so that you can leverage all the fine tuning properties for biotech applications that are more mature/optimized in reinhardtii. I suspect the later may have been tried but maybe not genome wide or conditions this extreme.

Would be interesting to see if the source of phenotypic heterogeneity in isogenic lines can be teased apart by high dimensional phenotyping. For example one might expect isogenic lines that vary by phenotype in phototaxis to also maintain segregation among correlated phenotypes like sensitivity to high light stress. If subpopulations continue to segregate in these circumstances, the cause may be stochastic factors but if segregated populations maintain separate phenotypes for assays where the phenotype is correlated, it may be epigenetic or other heritable but primarily genetic sequence-independent factors.

In short, high dimensional phenotyping incorporating assays with varying degrees of mutual information may have promise in understanding micro-environmental, epigenetic, and stochastic factors of variation.

Rather than separating by positive and negative interactions, the “innovativeness” of genetic interactions might better be described by how unexpected the fitness of the double mutants is relative to the multiplicative outcome of the singles…so the strength of their interactions in either direction. Information theoretic metrics like mutual information or KL divergence could be informative here. I’d be interested to see if the more surprising/unexpected interactions that actually confer greater information gain are perceived as interesting or impactful by the scientific community based on citations in the literature.

I found this result a bit puzzling with respect to the maintenance of a strong cortical network in the presence of CK-666 (which could be due to timing of the experiment if there was insufficient time for branched network turnover and inhibition of newly nucleated actin branches) but then began to wonder if part of the issue is due to this being done in trypsinized cells shortly after replating -- I'm sure I'm not thinking of something obvious but why not just grow and image directly on chamber slides to preserve the cortical actin network in a more native state?

Regardless, I also wanted to mention that I found this to be an very elegant study using an approach that indeed can illuminate some important properties of actin binding protein behavior.

Do you see excess membrane budding/blebbing/ectocytosis in CFAP ubiquitin-binding mutants that can't retrieve membrane-associated cargo like in BBSome mutants?

Curious if you're actually in TIRF with this much cell body fluorescence and if you have WT CFAP36 data from this same experiment?

Since deubiquitinases can be regulated, can function non-catalytically, and have specificity for ubiquitinated substrates, perhaps USP5 could still be a viable candidate

Interesting study with evidence of functional rod-dominant vision on the basis of phototransduction genes present (and comparable optic tectum). I see in figures 1 and 2 there is an intact RPE and wonder if you have higher resolution of the morphology there? Also, I guess I would expect for there to be less disc shedding and RPE phagocytosis if there is a lower demand for chromophore regeneration. Curious about molecular evolution of retinal cycle and phagocytosis genes in the RPE that might show evidence of adaptation in a circumstance without light-dependent circadian cycling but retained functional phototransduction.

Curious if CYFIP2 can be tagged while maintaining integrity of the WRC and if so, if there would be any difference in the efficiency of proper localization/targeting of the K vs E variants in your WT knockdown/overexpression context (either in differentiated SH-SY5Y cells or primary neurons).

There does seem to loss of global polarity in CYFIP2 KO cells -- do you know if there are differences in the microtubule cytoskeleton/MTOC positioning in CYFIP2 KOs?

Both these questions are kind of getting at whether there are other novel/unexpected functions of CYFIP2 or effects of their variants than what we would expect solely related to WRC integrity/activity.

Curious if you’ve done any motility or phototaxis experiments in the RAB18 knockdown cells. Since you have such a clean result for impaired ciliary entry of the BBsome independent of IFT, this and other analyses of ciliary signaling in your knockdowns would help tease apart which phenotypes from BBsome mutants are due to cell body roles vs ciliary roles of the BBsome.

One thing I think is not explicitly stated here but which I think is really great is that the cells were grown in constant light in TAP media, meaning that they won't be synchronized/as homogenous, and therefore will likely capture a wider range of biology in the dataset than might otherwise be observed.

Another note is the use of the mat3-4 mutant strain, which as stated has the experimental benefit of a smaller size for vitrification (almost half the diameter of wild-type strains). If I recall correctly when we were looking for actin filaments in mat3-4 tomograms, they seemed much more infrequent than we would have expected based on how abundant filaments are when phalloidin staining CC-125 or CC-124 wild-type strains (https://doi.org/10.1091/mbc.E19-03-0141). And I recall we tried phalloidin staining the mat3-4s and had trouble doing so. So my suspicion is that there might be a few key differences in macromolecular composition in these mutant strains that might prevent some structures from resolving or being abundant enough to detect readily. It's not a criticism, but just worth noting I think for those trying to extrapolate from this extremely helpful resource.

Did you notice any acentriolar microtubule organizing centers in your dataset? Even if rarely occurring?

Some very interesting and useful results! Now that you've shown more robust expression and secretion in incerta, curious if due to the relatively small difference in gene number (a few hundred fewer in incerta per Craig et al., 2021, https://doi.org/10.1093/plcell/koab026), if there are any patterns of gene family evolution that might rationally point to reinhardtii mutants that could exhibit more robust transgene expression and secretion while preserving other desirable properties like growth.

A quantification would be helpful

I see no meaningful difference between the CSE-treated and untreated cells for MSS4-GFP. How are you determining this difference? Is there a quantification?

There does not appear to be a methods section. How were the images taken? For example, was this a slice of a confocal z-stack such that signal in the middle of the cell really is in the cytosol or a wide field image such that the "cytosolic" signal may just be signal from accumulation of Ede1 at the cell surface (aka endocytosis is stuck at Ede1 recruitment step so a lot more enrichment is visible at the top and bottom of the cell). In other words have these images been taken in a way that can distinguish these possibilities?

I might be missing it somewhere but the two time points do not appear to be in the figure or figure legend -- from the text it appears maybe the bottom row is 2 hours?

Curious what the distribution of centrosomal linker proteins (rootletin, C-Nap1) looks like in stretched or fractured centrioles —ie whether they diffuse, or remain centriole associated under these conditions of mechanical stress. Related, do you think mitotic centriole disengagement mechanisms downstream of Nek2 are distinct from or might they be involved in Dyrk3-mediated mechanisms of regulating mechanical integrity of centrosomes in interphase?

Curious if you have data from other markers ...are these MTs acetylated? Are there motor proteins in this compartment?

You tested the ASD and CHD and ASD-CHD genes to identify those that impact survival or proliferation of NPCs in either direction. I was curious about the breakdown of genes based on direction of phenotype (increased vs decreased proliferation, increased vs decreased survival). For example, knock down of cilia-promoting genes (or different ones) having a positive impact on proliferation given the need for centrosomal networks being required for both cell division and ciliogenesis (so these being somewhat mutually exclusive).

One interesting use case would be to be able to detect cells on the pathway to growth arrest or cell death before a dramatic cellular phenotype as a way to evaluate phenotypes of only the "healthy" cells (if that is even a thing given we don't traditionally assay for more than a few dimensions). That said, even with the dramatic cellular phenotype, the shift is minor so perhaps early signals of poor cell health would not be detectable?

I wonder if you’ve tried this in your hands or done so recently with phalloidin conjugated to brighter and more photostable fluorophores like Atto dyes. Even if filaments are short and dynamic, they may be concentrated here sufficiently to detect with phalloidin under appropriate conditions. I mention it because we were ultimately able to get this to work in alga for detecting an actin with even less homology to metazoan actin than trypanosomes when sufficiently concentrated (figure 7 of https://doi.org/10.1091%2Fmbc.E19-03-0141)

Super interesting result, particularly the clear demonstration that the site of mechanical deciliation is proximal to the transition zone.

One thing I’d be careful about is the claim that delay in regeneration is due to the need for protein synthesis of B9d1 on the basis of antibody staining with chx treatment. While it’s clear from figure 4 that this TZ component does increase its expression upon deciliation, antibody staining is likely insufficiently sensitive to make a claim of this protein being limiting for ciliary assembly in the precursor pool.

Seriously this is such an incredibly cool result to be able to drive actin polymerization by a random protein that forms condensates when fused to an actin binding peptide! The implications for what might happen inside cells is wild!

Have you done any pyrene or seeded pyrene assays with your eps15-lifeact to see effects on actin nucleation or elongation?

Super interesting results/model! Years ago, we saw nascent proteins near basal bodies (using a puromycin analog to label as well) in Chlamy, which doesn’t have a clear PCM-1 ortholog/centiolar satellites. So according to this, where there is protein synthesis at satellites whether they are (vertebrates) or are not (flies) near centrosomes/cilia, maybe there is different, satellite-free local protein synthesis at cilia/centrioles of a different pool of mRNAs. Or maybe your waystation hypothesis is correct rather than protein synthesis actually happening at satellites directly.

What is the percentage of ciliated cells and cilia length with PTX treatment in the iso-osmotic conditions? If hyperosmotic cilium changes are accompanied by excess MT formation, prevented by MT depolymerization, and both hyperosmotic and PTX treated cells promote cytoplasmic MT bundling/polymerization, what is the effect of PTX-associated MT polymerization (without hyperosmotic shock) on cilium frequency and length, and ODF2-positive cells under these experimental conditions? This would tell you if the excessive microtubules alone vs other factors caused by hyperosmotic shock were drivers of ciliary shortening/loss.

Curious if the number of cells between pre- and post- heat shock is also decreased? It's clear that the the percentage of ciliated cells is decreased with or without TNF treatment but if there is a lot of concomitant cell death, the decrease in ciliated cell percentage could be tied to overall decreased cell health or other factors that are not cilia specific.

It's less critical in this case since you see no appreciable effect, but just worth noting as you use this drug in the future that the concentration used is quite high, above the level known to inhibit myosin 2, not just formins (ref: PMC8121067)

This filamentous network of centric polymers is also seen in Chlamydomonas and forms a cage like structure around the nucleus. Curious if you ever see centrin filaments around the nucleus dinoflagellates or changes in the filamentous organization in response to calcium. While there have been many hypotheses about the nature and function of these central extending filaments and the function of calcium-mediated centrin filament contraction, to my knowledge, this hasn't been really settled.

Looks like your PYP ortholog detection started with a sequence identity-based search. Have you also done a protein structure based search, eg via foldseek, to identify potential orthologs that may have low sequence similarly but high structural conservation to uncover additional candidates with beneficial properties?

should this be 2F and 2G?

should this read Fig. 2D, or 2E where this is quantified?

This is a super interesting set of results, particularly the localization inside mitos! Do you think that the mitochondrial phenotype in PFN1 knockouts is due to spontaneous nucleation of actin filaments within mitos? Also I may have missed if you checked this but do you think the decreased f-actin in PFN1 KOs as shown in figure 4H is from degradation of actin as seen in this other system (https://doi.org/10.1073/pnas.1721935115) or is that already known for these PFN1 KOs in these cells?

Would be interesting to also have a control that lacks a different domain (the kinase, LID, or NMP domains) to see if the DPY30 domain containing deletion mutants can still localize to cilia to exclude the possibility of global misfolding, or other regions being responsible for ciliary localization.

I think this conclusion is the most difficult to support from the data shown in figure 1 a-c and B. The images seem of insufficient resolution in a-c to claim unanchored microtubules along with subtle differences in satellite distribution based on mean distance or percentage within a window.

Seems like the radial pattern of control MTs may just be exacerbated due to physical/constraint proximity to the nucleus rather than a particular directional growth. It would also be be helpful to see cell morphology via brightfield/other in these images to see what the other membrane or cellular features might contribute to this polarization.

Thanks for sharing this very interesting and useful study! The ability to simultaneously visualize different actin isoforms with reduced effects on endogenous dynamics is fantastic and will no doubt lead to future discovery of differential functions.

The pitfalls of N-term actin tagging are well documented as you note, so strategies that allow for faithful binding to endogenous nucleators would indeed be beneficial. However, the preferred internal 229/230 tag still shows no greater co-localization (and perhaps reduced co-localization, as I am unsure of the statistical difference in figure 1C) with total f-actin/phalloidin staining relative to N-terminal tagging. This suggests that there are indeed additional effects of the internal tag on dynamics (likely driven by affected ABP binding) despite largely not identifying those defects in your assays. I would have also therefore have been interested to see the N-term tagged control for figure 3 alongside the internal tags. This control wouldn’t be quantitatively comparable of course but I can’t remember if formin binding is affected for n-term tagged actins or just getting through the formin ring.

Regardless, I’ll emphasize the importance of additional tools and information such as what you present here. The extensive interactions of actin with hundreds of binding proteins with myriad functions throughout cells highlights extensive combinatorial complexity that benefits from the availability of a full suite of actin labels so that the right labeling strategy can be selected based on application. This is therefore a very welcome addition to that suite of strategies!

I think the cytoplasm/extracellular facing portion of receptors depicted in Figure 7 don't seem to preserve topology? For example if you follow the purple receptor, the terminal yellow tail is sometimes cytoplasmic facing, sometimes extracellular/lumen facing. I only mention it because made the trafficking dynamics in the diagram a bit unclear.

I think this should say Extended data Fig. 9

The control image here looks quite a bit different than typical actin localization in a wild-type cell (with a more prominent localization to what I am assuming is the cell apex relative to mid-cell actin). I'm wondering if this is specific to this/a cw strain or whether this is a non-representative image. Also curious what the larger number of cells looks like in NaCl as skewness is measured but the clear phenotype from this image appears to be an increase in actin bundling, just as in Arabidopsis. If this is representative, is there an increase in actin bundler expression or expression of formins in your Chlamy dataset? Lastly, for visualization, fixed cell phalloidin staining of these cells might give you better resolution to identify the specific phenotypic consequences of osmotic stress with respect to the cytoskeletal rearrangement.

Using lifeact in this particular Chlamy strain, do you typically see apical actin puncta in untreated cells (which we think are endocytic pits, see https://doi.org/10.1101/2020.11.24.396002) and if so, are these quantifiably different in NaCl?

Competition between actin binding proteins for stabilizing different actin populations is a super interesting phenomenon. I always imagined this being due to some actin binding proteins acting as a sink for monomers dropping local actin concentration available for other ABPs or sequestering monomers to other spatial domains (something we’ve observed in other systems). However, I found it really interesting that an actin bundler forming presumably linear actin tunneling nanotubes that seems to be antagonistic to branched actin formation by ARP2/3 directly binds components of the ARP2/3 complex. Do you imagine that the higher branching ARP2/3 complex isoforms are also sequestered from the rest of the complex where lamellipodia would form? Are you able to localize those Eps8-interacting ARP2/3 components to nanotubes?

It looks like each iteration involves permeabilization. How necessary is this step in each iteration rather than simply rounds of primary and secondary antibodies with washes in between?

Also we’ve found detection of actin with fluorescent phalloidin to require improved signal and reduced background, particularly in photosynthetic cells with a great deal of chlorophyll autofluorescence. This requires 1) low concentrations, 2) reduced incubation time, and 3) brighter fluorophore such as Atto compared to Alexa dyes. For your antibodies, you mention low concentrations also result in low signal, so this is suboptimal so I’m still wondering if we could improve phalloidin staining for difficult samples with iteration. I know you don’t find similar improvements in phalloidin staining with iteration but I’m wondering if your phalloidin staining protocol had been optimized as much as you’ve optimized the antibody concentrations for your 1X designations to minimize signal to noise before performing phalloidin iterations to increase signal. Further, it could be that the absolute target abundance matters for the changes in signal to noise ratio with iteration. So by comparing iteration for low abundance antibody targets with iteration for high abundance non-antibody labeling, you may not see comparable improvements for a non-antibody label. If you were to use this protocol on cells where the f-actin signal is harder to detect than in these cells, might you see improvements with iterative labeling? Or alternatively if you compared antibody iteration with a non-antibody label against a target of lower abundance in the same cells, perhaps the signal to noise difference would change more with iteration? It would be of general interest if iterative staining could also improve signal to noise for low signal or low abundance non-antibody labels.

Thanks for publishing this interesting study!

Re: figure 4 and the associated claim, is acetylated alpha-tubulin normalized by the total alpha tubulin intensity less? Or is it proportional to reduced microtubes generally in vim-/- MEFs? Given PCM defects and impaired re-assembly of mictotubules after nocodazole treatment, basal turnover and the total mictotubule network may also be affected in vim-/- MEFs, but if I’m not mistaken, that’s never quantified here. Depending on the normalized outcome, the reduced acetylation rates may be independently reliant upon vimentin or be a consequence of globally reduced microtubules.

This is an incredibly interesting and novel result identifying a role for the CEP290 ciliary transition zone protein in cytoplasmic microtubule dynamics and adhesion. One immediate question given the work is done in IMCD3 cells that are competent to form cilia is whether there is expression and a role for CEP290 in cells that are not, such as Jurkat T cells that express many ciliary proteins for use in the immunological synapse but never form a cilium. Those are suspension cells so the adhesion phenotypes may not be evident, but given the more generalized role in microtubule dynamics shown here, I wonder whether CEP290 is expressed in this cell type, when during the cell cycle it might be expressed in a cilium-incompetent cell type such as this, and if there are phenotypes associated in its knockdown in such a context. The last certainly is out of scope of the current work but perhaps existing publicly available datasets would contain the information about expression and cell cycle-associated (and therefore microtubule reorganization associated) timing of CEP290 expression in truly cilium-independent context.

This is an incredibly interesting and novel result identifying a role for the CEP290 ciliary transition zone protein in cytoplasmic microtubule dynamics and adhesion. One immediate question given the work is done in IMCD3 cells that are competent to form cilia is whether there is expression and a role for CEP290 in cells that are not, such as Jurkat T cells that express many ciliary proteins for use in the immunological synapse but never form a cilium. Those are suspension cells so the adhesion phenotypes may not be evident, but given the more generalized role in microtubule dynamics shown here, I wonder whether CEP290 is expressed in this cell type, when during the cell cycle it might be expressed in a cilium-incompetent cell type such as this, and if there are phenotypes associated in its knockdown in such a context. The last certainly is out of scope of the current work but perhaps existing publicly available datasets would contain the information about expression and cell cycle-associated (and therefore microtubule reorganization associated) timing of CEP290 expression in truly cilium-independent context.

Re: figure 4 and the associated claim, is acetylated alpha-tubulin normalized by the total alpha tubulin intensity less? Or is it proportional to reduced microtubes generally in vim-/- MEFs? Given PCM defects and impaired re-assembly of mictotubules after nocodazole treatment, basal turnover and the total mictotubule network may also be affected in vim-/- MEFs, but if I’m not mistaken, that’s never quantified here. Depending on the normalized outcome, the reduced acetylation rates may be independently reliant upon vimentin or be a consequence of globally reduced microtubules.

Using lifeact in this particular Chlamy strain, do you typically see apical actin puncta in untreated cells (which we think are endocytic pits, see https://doi.org/10.1101/2020.11.24.396002) and if so, are these quantifiably different in NaCl?

The control image here looks quite a bit different than typical actin localization in a wild-type cell (with a more prominent localization to what I am assuming is the cell apex relative to mid-cell actin). I'm wondering if this is specific to this/a cw strain or whether this is a non-representative image. Also curious what the larger number of cells looks like in NaCl as skewness is measured but the clear phenotype from this image appears to be an increase in actin bundling, just as in Arabidopsis. If this is representative, is there an increase in actin bundler expression or expression of formins in your Chlamy dataset? Lastly, for visualization, fixed cell phalloidin staining of these cells might give you better resolution to identify the specific phenotypic consequences of osmotic stress with respect to the cytoskeletal rearrangement.

I think this should say Extended data Fig. 9

It looks like each iteration involves permeabilization. How necessary is this step in each iteration rather than simply rounds of primary and secondary antibodies with washes in between?

Also we’ve found detection of actin with fluorescent phalloidin to require improved signal and reduced background, particularly in photosynthetic cells with a great deal of chlorophyll autofluorescence. This requires 1) low concentrations, 2) reduced incubation time, and 3) brighter fluorophore such as Atto compared to Alexa dyes. For your antibodies, you mention low concentrations also result in low signal, so this is suboptimal so I’m still wondering if we could improve phalloidin staining for difficult samples with iteration. I know you don’t find similar improvements in phalloidin staining with iteration but I’m wondering if your phalloidin staining protocol had been optimized as much as you’ve optimized the antibody concentrations for your 1X designations to minimize signal to noise before performing phalloidin iterations to increase signal. Further, it could be that the absolute target abundance matters for the changes in signal to noise ratio with iteration. So by comparing iteration for low abundance antibody targets with iteration for high abundance non-antibody labeling, you may not see comparable improvements for a non-antibody label. If you were to use this protocol on cells where the f-actin signal is harder to detect than in these cells, might you see improvements with iterative labeling? Or alternatively if you compared antibody iteration with a non-antibody label against a target of lower abundance in the same cells, perhaps the signal to noise difference would change more with iteration? It would be of general interest if iterative staining could also improve signal to noise for low signal or low abundance non-antibody labels.

Thanks for publishing this interesting study!

Competition between actin binding proteins for stabilizing different actin populations is a super interesting phenomenon. I always imagined this being due to some actin binding proteins acting as a sink for monomers dropping local actin concentration available for other ABPs or sequestering monomers to other spatial domains (something we’ve observed in other systems). However, I found it really interesting that an actin bundler forming presumably linear actin tunneling nanotubes that seems to be antagonistic to branched actin formation by ARP2/3 directly binds components of the ARP2/3 complex. Do you imagine that the higher branching ARP2/3 complex isoforms are also sequestered from the rest of the complex where lamellipodia would form? Are you able to localize those Eps8-interacting ARP2/3 components to nanotubes?