Polarisome scaffolder Spa2-mediated macromolecular condensation of Aip5 for actin polymerization

A multiprotein complex polarisome nucleates actin cables for polarized cell growth in budding yeast and filamentous fungi. However, the dynamic regulations of polarisome proteins in polymerizing actin under physiological and stress conditions remains unknown. We identify a previously functionally unknown polarisome member, actin-interacting-protein 5 (Aip5), which promotes actin assembly synergistically with formin Bni1. Aip5-C terminus is responsible for its activities by interacting with G-actin and Bni1. Through N-terminal intrinsically disordered region, Aip5 forms high-order oligomers and generate cytoplasmic condensates under the stresses conditions. The molecular dynamics and reversibility of Aip5 condensates are regulated by scaffolding protein Spa2 via liquid-liquid phase separation both in vitro and in vivo. In the absence of Spa2, Aip5 condensates hamper cell growth and actin cable structures under stress treatment. The present study reveals the mechanisms of actin assembly for polarity establishment and the adaptation in stress conditions to protect actin assembly by protein phase separation.

Overall, the in vitro work is pretty impressive and complete, and demonstrates a very interesting property of Aip5 to form glass-like aggregates in absence of Spa2 but liquid droplets in presence of Spa2. The main unaddressed question is what the relevance of this behaviour is in vivo. Indeed, the cell biology work supports a role for Aip5 in promoting actin assembly, but the function of the aggregation and/or de-mixing is not directly addressed. A second issue is that the text needs clarification -there are many grammatical errors, which is several instances lead to possible misunderstanding or lack of clarity of the text. I address these issues in more detail below.

Link between in vitro observations and in vivo:
In general, the possible role in vivo of the phase separation observed in vitro is not addressed. In vitro, Spa2 is shown to recruit Aip5 to liquid droplets and it is proposed that the same happens in vivo upon stress (loosely defined), with Spa2 helping prevent Aip5 aggregates. The implication of this view is that Aip5 aggregates would be detrimental to the cell. However, this is not tested. There are at least two testable predictions: 1. That the disorganization of actin observed in spa2∆ (fig 5g-h) depends on the formation of Aip5 aggregates: actin patches in spa2∆ upon energy depletion would be predicted to disappear upon deletion of Aip5; 2. That these changes have effect on cell viability in stressed conditions. Fig S6f shows that Aip5 C-terminus has a dominant negative effect at 37°C, which is not immediately concordant with the idea that Aip5 aggregates are detrimental. The authors should test whether spa2∆ has stress phenotypes, and whether deleting aip5 alleviates these phenotypes. More generally, the manuscript does not address at all what the function of the phase separation may be during normal growth. Addressing this would require the generation of Aip5 and/or Spa2 mutant with altered aggregation/droplet properties. Even if not addressed experimentally, it should be discussed. As is, the discussion is very superficial and vague and the model figure 6 does not propose any specific function for phase separation / aggregation by Spa2 and Aip5.
Role of Aip5: I am not entirely convinced about the claimed synergy between Aip5 and Bni1 in actin assembly. The data clearly shows that Aip5 accelerates actin assembly on its own and does so also in presence of Bni1. The magnitude of the increase does not appear to be more with than without Bni1. The plots shown in fig 1d-e seem to indicate more additivity than synergy. In vitro, the only pieces of data truly convincing of synergy is the barbed end uncapping. Similarly, the genetic interaction between aip5∆ and bni1∆ (Fig 1h) suggest additive (independent) functions rather than Aip1 feeding into the Bni1 assembly function. This genetic interaction is however not entirely reproduced in Fig S6f. Thus, while the effect of Aip5 C-terminus on actin assembly is clear, the Editorial Note: Parts of this Peer Review File have been redacted as indicated to maintain the confidentiality of unpublished data.
possibly synergy with Bni1 should be phrased more carefully.
Similarly, the authors should be careful with their claimed in vivo Aip5 concentrations. Though GFP fluorescence can be used as a proxy to measure concentrations in situ, the uncertainty is significant. Claiming precise concentrations without standard deviation (which should be added) is misleading, especially as the difference in concentration measured at the cell tip and in the cytosol is less than 2-fold. As the authors point out, the actin polymerization activity of Aip5 is very modest at the measured in vivo concentrations. Given the formation of aggregates/liquid droplets, this measured concentration may be vastly off at the very local level. This would be worth discussing.
Text issues: In general, the text needs attending to by a fluent English-speaker. I do not like to write this when it is purely esthetics, but there are instances here where the clarity of the text suffers (for instance, "Aip5 promotes actin assembly solely and synergistically with formin Bni1" is probably meant to state that Aip5 promotes actin assembly "on its own", as well as with Bni1, rather than "only with Bni1"; the sentence "In the absence or presence of crowded environments, Aip5 Nterminal variants showed amorphous assemblies in different size in the N-terminal lengthdependent manner" is difficult to understand...). It would also help the reader to provide a little more interpretation to some of the results, as the text is rather dry.
The authors should also revise some of their description of the polarisome. For instance, the first sentence of the abstract sounds like the polarisome is both a well-established complex and an accepted form of non-membranous compartment, which has not been shown previously. The polarisome is a loosely-defined entity. To my knowledge, while individual components have been shown to interact with one another, there is no evidence that it forms a structural complex that can be purified as a whole (like the proteasome, the ribosome, …). It is thus misleading to present it as such. I would also refrain from including Epo1 in it. At this rate any protein interacting with Bni1, Bud6 or Spa2 would be part of the polarisome, and this makes up dozens of proteins.
The discussion states "Aip5 that further recruits G-actin and Bni1", but I do not think the paper shows that Aip5 recruits Bni1.
Other comments: The text in Fig S1A is so small that it is very difficult to read.
In fig 1f-g, the concentration of proteins used is not indicated.
In Fig 2g-h, the Aip5C alone controls are missing. Fig 5h legend does not state what condition is quantified. In either case, it is missing either the energy depletion of the control to test whether the effect is specific to energy depletion or whether spa2∆ has an effect also in steady-state growth.
I do not understand the statement that "full length Aip5 eluted earlier than the predicted size on gel filtration column chromatography (Fig S4A)". What is the predicted size, and at what size does it elute?
The effect of L-arginine on the elution of Aip5 is not very clear.

Reviewer #2 (Remarks to the Author):
Xie et al discovered a new member in polarisome, a multiprotein center for actin nucleation in budding yeast and filamentous fungi. The authors named the new member Aip5 and carried out detailed characterization for Aip5 with various assays. Two major conclusions are made: (I) Cterminus folded domain of Aip5 promotes actin nucleation by interacting with G-actin and Bni1; (II) N-terminus disordered domain of Aip5 forms functional high-order oligomer, undergoes condensation when crowded, and demixes with Spa2 for stress adaption. I found conclusion (I) is well supported with data, but not so with conclusion (II). I would not recommend for publication unless the authors address my following concerns. 1. the data does not support the notion that Aip5 itself undergo liquid condensation (Figure 4d). 2. The authors show Aip5 is a client molecule that can be recruited to Spa2 condensates. They looked at the effect on Aip5 dynamics, they should also look at how this affects actin nucleation by Aip5. Without this data, the title is not meaningful, then the authors should change the title. 3. The authors argue that interacting with Spa2 helps Aip5 stress adaption by looking at Aip5 'aggregates' under stress conditions and Sap2 deletion background. Authors did no characterization with Aip5 puncta in stressed cells and yet implied that they are in similar aggregating state within that in vitro ( Figure 4d which the authors called condensates instead of aggregates, without any justification). Based on the reversibility and sensitivity to 1,6 hexanediol, Aip5 assembly in stressed cells represent liquid condensates. The authors should further characterize the material properties for Aip5 assembly in vitro and in stressed cells before implying they are the same. One alternative explanation for the observations under stress condition is Aip5 goes to liquid stress bodies and Spa2 has an effect on the formation of such stress body. To rule out this possibility, the authors should also look at how Spa2 condensation and the recruitment of Aip5 to Spa2 condensates are affected by pH in vitro, and also characterize Spa2 localization in stressed cells, its relation to Aip5 foci and the material properties. 4. The authors made a lot of assumptions when using different terms to describe material properties of protein assemblies. For example, Line 194: coalescence. No data is given to shown the growth is via coalescence. Line 229: glassy, the FRAP data only show it's not dynamic, no data is given to show the material is a glass.
This manuscript by Xie and colleagues addresses a very interesting question: the possible role of lowcomplexity domains in proteins associated with actin assembly. They focus their attention on Aip5 protein, which they identified as a formin Bni1 interactor. They crystalize the only folded region of Aip5, its C-terminus which they show binds Bni1 and G-actin and promotes F-actin assembly in vitro. Aip5 Nterminus is of low-complexity and the authors show it forms aggregates in vitro, whose size is modulated by concentration and PEG in the buffer. Interestingly, this N-terminal part of Aip5 binds Spa2, which itself forms de-mixed liquid droplets in vitro and recruits Aip5 to these droplets. In vivo, Aip5 localizes to the bud tip dependent on Spa2, and aip5 deletion shows synthetic sickness with bni1∆. Upon energy depletion or lowering the pH, Aip5 forms aggregates in vivo, which are exacerbated in absence of Spa2.
Overall, the in vitro work is pretty impressive and complete, and demonstrates a very interesting property of Aip5 to form glass-like aggregates in absence of Spa2 but liquid droplets in presence of Spa2. The main unaddressed question is what the relevance of this behaviour is in vivo. Indeed, the cell biology work supports a role for Aip5 in promoting actin assembly, but the function of the aggregation and/or de-mixing is not directly addressed. A second issue is that the text needs clarification -there are many grammatical errors, which is several instances lead to possible mis-understanding or lack of clarity of the text. I address these issues in more detail below.

Link between in vitro observations and in vivo:
In general, the possible role in vivo of the phase separation observed in vitro is not addressed. In vitro, Spa2 is shown to recruit Aip5 to liquid droplets and it is proposed that the same happens in vivo upon stress (loosely defined), with Spa2 helping prevent Aip5 aggregates. The implication of this view is that Aip5 aggregates would be detrimental to the cell. However, this is not tested. There are at least two testable predictions: 1. That the disorganization of actin observed in spa2∆ (fig 5g-h) depends on the formation of Aip5 aggregates: actin patches in spa2∆ upon energy depletion would be predicted to disappear upon deletion of Aip5; We thank Reviewer #1 for this great suggestion to elucidate the function of Aip5 aggregates in vivo better. We have now added the following experiments to discuss this point.
First, upon energy depletion, the disorganization of actin cable starts as early as 15 min in wild type and around 5 min in the aip5∆ and spa2∆ mutant cells ( Figure S9a). This data suggests that both Spa2 and Aip5 involve in the cellular protection of actin cables under stress conditions. Consistently, deletion of AIP5 in the WT or in the background of spa2∆ did not reduce actin patches upon energy depletion ( Figure S9a). Second, the Aip5 aggregates appeared starting from 30 min ( Figure S9b), which is later than the actin cable disorganization starting from ~ 15min ( Figure S9b). It indicates that the actin remodeling within the initial 30 min upon energy depletion is less likely due to Aip5 aggregation.
We discussed these new results in the last paragraph of the Discussion section. We think the formation of the aggregates of Aip5 is a cellular response to energy depletion from the nature of its intermolecular multivalency that was used as a temporary and rapid solution in response to stress conditions ( Figure 6a).
2. That these changes have effect on cell viability in stressed conditions. Fig S6f shows that Aip5 Cterminus has a dominant negative effect at 37°C, which is not immediately concordant with the idea that Aip5 aggregates are detrimental. The authors should test whether spa2∆ has stress phenotypes, and whether deleting aip5 alleviates these phenotypes.
We appreciate Reviwer#1 for raising the excellent point. We have carried out a new experiment and added it in Figure 6g now. Under the prolonged adverse condition, we found that Aip5 aggregates are indeed detrimental to the cell as Reviewer #1 predicted. Once we deleted AIP5 in the background of spa2∆, it alleviated the cell growth sickness comparing to spa2∆ mutant cells. It indicates Aip5 in vivo aggregates are toxic under long-term adverse conditions at the later stage of stress. We have now added the new data in Figure 6g and discussed the function Aip5 aggregates carefully in the last paragraph of the Discussion.
In addition, to better define Spa2 condensates recruits Aip5 in vivo, we added new experiments that showed Aip5 demixes with Spa2 that controls the size and structure of in vivo Spa2-Aip5 macromolecular condensates (Figs. 5d-h and s8a, d-i). Aip5 condensates undergo rapid recovery once favourable nutrients were supplemented. In contrast, without Spa2 demixing effect, Aip5 condensates has slow recovery in vivo, and also detrimental to cell survival if under long-term stress. Together, these observations agreed with in vitro data where Spa2-Aip5 droplet reversed Aip5 amorphous assemblies by increasing dynamics of Aip5.
More generally, the manuscript does not address at all what the function of the phase separation may be during normal growth. Addressing this would require the generation of Aip5 and/or Spa2 mutant with altered aggregation/droplet properties. Even if not addressed experimentally, it should be discussed. As is, the discussion is very superficial and vague and the model figure 6 does not propose any specific function for phase separation/aggregation by Spa2 and Aip5.
We thank Reviwer#1 great suggestion and apologize for the lack of discussion about the physiological function of Aip5/Spa2 phase separation.
First, as Reviewer #1 suggested, we have performed new experiments and discussed more in the first paragraph of the Discussion section.
Second, to investigate more the mechanisms by which the Aip5 multivalency, including the dimerization and oligomerization states, directly regulate its activities as nucleator and NPF of Bni1, we performed the following additional experiments. 1, we compared Aip5 activities between the Aip5 dimer and Aip5 monomer. We generated a monomeric version of Aip5-C by truncating away 35 amino acids from its N terminal region (Figures s4a-c). Compared to Aip5-C dimer, the monomeric version of Aip5-C (Aip5-mC) showed an apparent reduction in biochemical activities for assembling actin and promoting Bni1-mediated nucleation (Figures s4d-f), suggesting the importance of the dimeric interaction of Aip5 for its biochemical activity. 2, to determine the changes of Aip5 activities by a highordered oligomerization state through Spa2 recruitment, we examined Aip5-mediated actin polymerization rate, in the absence or presence of Spa2-LC, with or without crowding agent ( Figure R1). Unfortunately, in such in vitro condition, we did not observe the significant change of the Aip5 activities from the enhanced multivalent interaction with Spa2. At this moment, without a full reconstitution of polarisome complex proteins, including Bni1, Spa2, Aip5, and Bud6, which comprise all necessary interactions domains for orchestrating a multivalent complex, we would not explicitly claim the regulatory mechanism of Aip5 activities from its demixing into Spa2 droplet. This comment is in the same line as Reviewer #2. To prevent any potential misleading at this stage, as Reviewer #2 suggested, we changed the title to "Macromolecular condensation of actin assembly factor Aip5 in polarisome complex". We wish Reviewer #1 endorse our opinion that such title change does not affect the quality, the conclusion, and the significance of our report.
Furthermore, as suggested, we also amended our model in Figure 7 to indicate the different functions and states of Aip5 assemblies under normal growth conditions and stress conditions. Role of Aip5: I am not entirely convinced about the claimed synergy between Aip5 and Bni1 in actin assembly. The data clearly shows that Aip5 accelerates actin assembly on its own and does so also in presence of Bni1. The magnitude of the increase does not appear to be more with than without Bni1. The plots shown in fig 1d-e seem to indicate more additivity than synergy. In vitro, the only pieces of data truly convincing of synergy is the barbed end uncapping. Similarly, the genetic interaction between aip5∆ and bni1∆ ( Fig   1h) suggest additive (independent) functions rather than Aip1 feeding into the Bni1 assembly function. This genetic interaction is however not entirely reproduced in Fig S6f. Thus, while the effect of Aip5 Cterminus on actin assembly is clear, the possibly synergy with Bni1 should be phrased more carefully.
We have now performed additional biochemistry experiments using total internal reflection fluorescence microscopy (TIRFM) assay to strengthen our results and rephrased our statement carefully regarding the synergy effect between Aip5 and Bni1 in vivo.
First, we performed more in vitro biochemical experiments to quantitatively analyze Aip5 activities. The bulk actin assembly assay pyrene assay (Figures 1d, e) on its own could not distinguish Aip5's function in different events of actin assembly, such as actin nucleation, elongation or even actin filament annealing. To provide better and quantitative biochemical analysis for in vitro synergy activities between Aip5 and Bni1, we have now performed additional TIRF experiment to monitor actin nucleation and elongation in real-time (Figures 1f-i), which can better dissect the actin assembly activities, for both nucleation and elongation. From the new TIRF experiment, we quantified the actin nucleation seeds number at early time point 5 min as well as actin filament barbed end elongation speed. A significant increase of actin seeds number was observed when introducing both Aip5 and Bni1 in the same reaction, an average of 99 actin seeds were formed in the defined areas for quantification, which is greater than the sum of actin seeds (42 actin seeds) that are generated by Aip5 and Bni1 on their own. Furthermore, neither Aip5 nor Bni1 affected actin filament barbed end elongation speed. We hope Reviewer #1 finds our new date supports the synergistic effect in actin nucleation better now.
Second, we agree with Reviewer #1's comment on the growth results of the double mutant. Since the growth assay is not as quantitative as the biochemical assay and may reflect a mixed effect, the more growth defects of double mutant than the single mutant could not distinguish the additive (independent) and synergistic function of BNI1 and AIP5 easily. We think our new TIRF assay offers more validation now that Aip5 has both nucleation and NPF activities, though Aip5 could also have additional functions in the polarisome complex. The conclusion about the NPF activity of Aip5 in an exclusive manner by using the growth assay results of double mutant aip5∆ and bni1∆ is not appropriated. We have now rephrased the sentence and described the result more carefully.
Similarly, the authors should be careful with their claimed in vivo Aip5 concentrations. Though GFP fluorescence can be used as a proxy to measure concentrations in situ, the uncertainty is significant. Claiming precise concentrations without standard deviation (which should be added) is misleading, especially as the difference in concentration measured at the cell tip and in the cytosol is less than 2-fold. As the authors point out, the actin polymerization activity of Aip5 is very modest at the measured in vivo concentrations. Given the formation of aggregates/liquid droplets, this measured concentration may be vastly off at the very local level. This would be worth discussing.
We thank Reviwer#1 for the concern of concentration measurement and suggestions. We did the following new experiments to address these concerns.
First, we apologize for the missing SD value, which is corrected now in Figure S5g.
Second, to validate the fluorescent image-based approach in measuring Aip5 in vivo concentrations, we performed additional western blot assay. The new results are added in Figures S5h, i. We generated a standard curve using recombinant Aip5-GFP protein by western blot detection via anti-GFP, where the standard curve was plotted from Aip5-GFP loading mass against the quantified protein signal intensity from western blot. In vivo Aip5-GFP concentration were estimated by detect endogenous Aip5-GFP expressing from the total cell lysate. In order to calculate the Aip5 protein concentration in vivo, the protein mass (m: deducted from standard curve), molecular weight (Mw=143 kDa), cell number (n, calculated based on loaded O.D.600) and average haploid cell volume (V=40 µm 3 ) were taken into account based on the formula C=m/Mw/n/V. From the western blot approach, we found that the calculated Aip5 protein concentration through western blot (~80 nM) was similar to that was derived from fluorescence method (~83 nM in the cytoplasm). By considering the results from both experiments, we think our results showed a general protein concentration of Aip5 in vivo. However, we do acknowledge Reviewer #1's concern about the potential inaccuracy of the exact protein concentration because both western-blot and microscopic approaches have their technical limitations. Fluorescence (cross)-correlation spectroscopy used by Kaksonen and Knop group (Boeke et al., 2014) could be a more accurate approach for in vivo protein measurement, especially for low abundant proteins. Here, we added new western-blot result, and also described the potential limitation for current protein concentration measurement in the Discussion section. We found that our presented protein concentration is useful in addressing Reviewer #1's other concerns. a) In the initially submitted manuscript, we used pyrene actin polymerization assay, which is not sensitive enough to demonstrate the Aip5 functions as nucleator and NPF of Bni1. Now, in the revised Figures 1f-g using TIRF actin assay, compared to the reaction without Aip5, 20 nM of Aip5 is sufficient to generate more actin seeds, during both the spontaneous actin polymerization and synergistic nucleation in Bni1-mediated actin assembly. Therefore, such in vivo low Aip5 concentration (<100nM) is sufficient to provide biochemical activities in actin assembly. b) Such protein concentration measurement could provide a proximate stoichiometry of Aip5 and Spa2 that guided us to demonstrate in vitro Aip5 participation in Spa2 droplet at a concentration and stoichiometry close to in vivo situation ( Figure S7f). We demixed Spa2-LC and Aip5 at a stoichiometry of ~5:1 with a concentration of 500 nM and 100 nM, respectively, which is close to their physiological concentration and the relative stoichiometry at bud tip. In results, we could still observe the demixing of Aip5 into the Spa2-LC droplet in the presence of crowding reagent. Now we added this new result to Figures S7f and more discussion.
Text issues: In general, the text needs attending to by a fluent English-speaker. I do not like to write this when it is purely esthetics, but there are instances here where the clarity of the text suffers (for instance, "Aip5 promotes actin assembly solely and synergistically with formin Bni1" is probably meant to state that Aip5 promotes actin assembly "on its own", as well as with Bni1, rather than "only with Bni1"; the sentence "In the absence or presence of crowded environments, Aip5 N-terminal variants showed amorphous assemblies in different size in the N-terminal length-dependent manner" is difficult to understand...). It would also help the reader to provide a little more interpretation to some of the results, as the text is rather dry.
We apologize for the confusion, we have now revised it and provided more interpretation for the results.
The authors should also revise some of their description of the polarisome. For instance, the first sentence of the abstract sounds like the polarisome is both a well-established complex and an accepted form of non-membranous compartment, which has not been shown previously. The polarisome is a loosely-defined entity. To my knowledge, while individual components have been shown to interact with one another, there is no evidence that it forms a structural complex that can be purified as a whole (like the proteasome, the ribosome, …). It is thus misleading to present it as such. I would also refrain from including Epo1 in it. At this rate any protein interacting with Bni1, Bud6 or Spa2 would be part of the polarisome, and this makes up dozens of proteins.
We thank Reviwer#1 for these excellent comments. We have now revised our description in the Abstract.
The discussion states "Aip5 that further recruits G-actin and Bni1", but I do not think the paper shows that Aip5 recruits Bni1.
We have now corrected it to a more precise way "Aip5 that directly interacts with G-actin and Bni1".
Other comments: The text in Fig S1A is so small that it is very difficult to read.
We have now enlarged the font size for better visualization.
In fig 1f-g, the concentration of proteins used is not indicated.
Thanks for pointing this out. We added the protein concentration now in the figure legend.
In Fig 2g-h, the Aip5C alone controls are missing.
To improve the quality and clarity, we have now performed additional actin polymerization experiments by monitoring actin nucleation using total internal reflection fluorescence (TIRF) microscopy. All the control and quantification are added now in the revised Figures 2f and g .   Fig 5h legend does not state what condition is quantified. In either case, it is missing either the energy depletion of the control to test whether the effect is specific to energy depletion or whether spa2∆ has an effect also in steady-state growth. We apologize for the missing text and quantification graph. We have now specified the conditions for quantification in Figures 6c-e. We have also now quantified the control groups and added in Figure S8j and amended the figure legends accordingly. Compared to WT cells, there is a mild defect with shorter actin filament in spa2∆ at the steady-state growth, which is added now in Figure S8j.
I do not understand the statement that "full-length Aip5 eluted earlier than the predicted size on gel filtration column chromatography (Fig S4A)". What is the predicted size, and at what size does it elute?
To clarify the confusion, now we added the expected elution volume of Aip5 dimers in the revised Figure  S5a. The predicted elution volume from a gel filtration column (Superdex™ 200 Increase column) based on Aip5 dimer molecular weight (286 kDa) would be around 11-13 ml. However, our purified Aip5 proteins usually eluted from 8-10 ml (larger than 440 kDa), suggesting a higher-order of oligomerization.
The effect of L-arginine on the elution of Aip5 is not very clear.
We added more results to clarify this point now. 1 st , EM negative staining, we found L-arginine break down the Aip5 oligomers from large assemblies (Figure 4a) into a dissolved pattern in Figure S5c. 2 nd , size exclusion chromatography (SEC) did not show a dramatic shift of Aip5 elution peak when treated with L-arginine ( Figure S5a), but with shifted tail lagging behind during elution. It is a typical resolution that SEC can resolve that indicate significant dissolution of large size oligomer ( Figure S5c). Our results indicate a disruption of Aip5 large assemblies by L-arginine into a lower-order of Aip5-oligomer but were not sufficient to be in its monomeric state. This L-arginine effect in elution of Aip5 oligomers was similar to our previously reported profilin oligomerization, in the presence or absence of L-arginine, which is shown in the Figure 2k and 2L of Sun He et. al. (Sun et al., 2018) (also shown below). With L-Arginine, the oligomerization of high-order Arabidopsis profilin 3 is reduced (a shift of early elution peak towards right), but not to the level of monomer completely. Now, we revised the results with clearer description in the Page 4 line 209-212. Xie et al discovered a new member in polarisome, a multiprotein center for actin nucleation in budding yeast and filamentous fungi. The authors named the new member Aip5 and carried out detailed characterization for Aip5 with various assays. Two major conclusions are made: (I) C-terminus folded domain of Aip5 promotes actin nucleation by interacting with G-actin and Bni1; (II) N-terminus disordered domain of Aip5 forms functional high-order oligomer, undergoes condensation when crowded, and demixes with Spa2 for stress adaption. I found conclusion (I) is well supported with data, but not so with conclusion (II). I would not recommend for publication unless the authors address my following concerns.
1. the data does not support the notion that Aip5 itself undergo liquid condensation (Figure 4d).
We agree with Reviewer #2. We did not state that Aip5 undergoes LLPS on its own. Aip5 forms highorder oligomers under the crowding condition, which showed low motility. And it only becomes more fluidic by demixing with the liquid droplet of Spa2 in vitro (Figures 4j.k). Furthermore, we have now examined the dynamic of Aip5's polarized tip localization, as well as the small foci formed in wild type and spa2∆ under energy depletion conditions in vivo (Figures 5d-f). We observed the full recovery of Aip5 protein after photobleaching at the polarized tip region (within 20 sec) where it bound to various polarisome components, Spa2/Bni1. On the contrary, the foci formed in spa2∆ under energy depletion has shown worse dynamic property comparing to the foci formed in wild type, which agreed with the notion that in vitro Aip5 protein phase separated as less-mobile aggregates without its binding partner Spa2. To avoid any potential misleading information, we have now double confirmed our description and phrased them carefully in the revised manuscript.
2. The authors show Aip5 is a client molecule that can be recruited to Spa2 condensates. They looked at the effect on Aip5 dynamics, they should also look at how this affects actin nucleation by Aip5. Without this data, the title is not meaningful, then the authors should change the title.
We appreciate this excellent point from Reviewer #2.
To study the mechanisms by which the Aip5 multivalency, including the dimerization and oligomerization states, directly regulates its activities as nucleator and NPF of Bni1, we performed the following experiments. Firstly, we compared Aip5 activities between the Aip5 dimer and Aip5 monomer. We generated a monomeric version of Aip5-C by truncating away 35 amino acids from its N terminal region (Figures s4a-c). Compared to Aip5-C dimer, the monomeric version of Aip5-C (Aip5-mC) showed a noticeable reduction in biochemical activities for assembling actin and promoting Bni1-mediated nucleation (Figures s4d-f), suggesting the importance of the dimeric interaction of Aip5 for its biochemical activity. Secondly, to determine the changes of Aip5 activities by a high-ordered oligomerization state through Spa2 recruitment, we examined Aip5-mediated actin polymerization rate, in the absence or presence of Spa2-LC, with or without crowding agent ( Figure R1). Unfortunately, we did not observe the significant change of the Aip5 activities from the enhanced multivalent interaction with Spa2. At this moment, without a full reconstitution of polarisome complex proteins, including Bni1, Spa2, Aip6, and Bud6, which comprise all necessary interactions domains for orchestrating a multivalent complex, we would not explicitly claim the regulatory mechanism of Aip5 activities by demixing into Spa2 droplet.
As suggested, we changed the title to "Macromolecular condensation of actin assembly factor Aip5 in polarisome complex". We also added these discussions in the first paragraph of the Discussion section. We wish Reviewer #2 endorses our opinion that such title change does not affect the quality, the conclusion, and the significance of our report.
3. The authors argue that interacting with Spa2 helps Aip5 stress adaption by looking at Aip5 'aggregates' under stress conditions and Sap2 deletion background. Authors did no characterization with Aip5 puncta in stressed cells and yet implied that they are in similar aggregating state within that in vitro ( Figure 4d which the authors called condensates instead of aggregates, without any justification). Based on the reversibility and sensitivity to 1,6 hexanediol, Aip5 assembly in stressed cells represent liquid condensates. The authors should further characterize the material properties for Aip5 assembly in vitro and in stressed cells before implying they are the same. One alternative explanation for the observations under stress condition is Aip5 goes to liquid stress bodies and Spa2 has an effect on the formation of such stress body. To rule out this possibility, the authors should also look at how Spa2 condensation and the recruitment of Aip5 to Spa2 condensates are affected by pH in vitro, and also characterize Spa2 localization in stressed cells, its relation to Aip5 foci and the material properties.
We appreciate Reviewer #2's for the constructive suggestions and apologize for the missed justification. We have now added the suggested experiments and with a better description, explanation, and discussion.
Firstly, to better understand the in vivo Aip5 foci property under stressed conditions, we applied FRAP to examine the protein dynamic (Figures 5d, e). Indeed, the result agreed with 1,6 hexanediol observation where Aip5 foci formed in wild type dissolved earlier than those formed in spa2∆ ( Figure  s8a), which Aip5 foci recovered faster and better with Spa2 protein within the cells. Furthermore, the rapid reversal of in vivo Aip5 condensates (around 50% recovery within 1 min, Figures 5d,e) and the slow FRAP recovery of in vitro Aip5 assembly on its own (less than 20% recovery within 4 min, Figures. 4j, k) suggest that the dynamic behaviors of pure recombinant Aip5 is not identical to the in vivo Aip5 assemblies . One possibility for the different behavior between in vitro and in vivo of Aip5 is that Aip5 has other binding partners that help Aip5 in a better fluidic state, even in the spa2∆ strain. We have demonstrated that Bni1 (Figures S1c-f; Figures 2i-m) is the binding partner of Aip5 which affected the polarized Aip5 tip localization in vivo. On the other hand, deletion of Bud6 diminished Aip5 tip localization, implying the Bud6 involvement in partial recruitment of Aip5 in vivo (Figures 1a, b). In addition, when Aip5 localized to the polarized tip where it binds to various polarisome proteins, Aip5 showed the most dynamic behavior from FRAP experiment (Figures 5d, e). Taken together, our newly added FRAP assay and 1,6 hexanediol suggested Aip5 exhibits the properties expected for biomolecular liquid-like condensates in vivo. Similar characterizations were also reported for liquid-like condensates in nuclear (Sabari et al., 2018).
Secondly, we performed the suggested in vitro pH experiment for Spa2-Aip5. The renew results are added in Figures S8c-g. We found Spa2-LC still formed droplet at pH range 4.4-7.5, though some droplets with irregular shape were observed at pH4.4 (Figure s8c). We next tested the recruitment of Aip5 into condensates Spa2-LC at physiological pH conditions. We chose the same pH conditions as we used for in vivo stress treatment in Figures 5j,k. Though Aip5 can be recruited into Spa2-LC droplet under all the three pH condition, the Aip5 dynamics in each droplet are different (Figures s8d-g). At pH5.5, demixing with Spa2-LC did not improve Aip5 motility. On the other hand, Aip5 showed the better fluidic property in Spa2-droplets at both pH6.5 and pH7.5 when compared to Aip5 assemblies without Spa2. This new in vitro result was consistent with the in vivo observation that Aip5 formed more pronounced condensates at pH5.5 than under pH6.5/pH7.5 in wild type cells (Figures 5j,k). It indicates pH is another regulator factors for Aip5 phase behavior. Under a low pH of 5.5, a presence of Spa2 could not alleviate the decreased motility of Aip5 assemblies. We further confirmed in vitro pH-dependent experiment in vivo by examining Aip5 dynamic using FRAP at pH 5.5 ( Figures S8h, i), where there was also no significant difference in the recovery rate of Aip5-GFP in wild type and spa2∆ in vivo. In contrast, the Aip5 foci formed in spa2∆ at pH6.5 recovered faster than that under pH5.5, both in vivo and in vitro (Figures S8f,h,i). Above data also suggest that other cellular components besides Spa2 play roles in regulating the dynamics of Aip5 in vivo condensates. Aip5 condensates in the SPA2 were barely detectable under pH6.5 and pH7.5. Thus, we were not able to examine their dynamic recovery rates through FRAP.
Third, in vivo Spa2 localization experiment is an excellent suggestion. We have examined Spa2 localization together with Aip5 during stress treatment. a), under energy depletion condition, Aip5 condensates highly colocalized with Spa2 (Figure 5f), which is consistent with the function of Spa2 in regulating Aip5 motility (Figures 5d, e). b), the Spa2 condensate formed during energy depletion condition also displayed a recovery property similar to Aip5, both of which recovered to around 50% in 60 seconds after photo-bleaching (Figures 5g, h). c), [redacted] 4. The authors made a lot of assumptions when using different terms to describe material properties of protein assemblies. For example, Line 194: coalescence. No data is given to shown the growth is via coalescence. Line 229: glassy, the FRAP data only show it's not dynamic, no data is given to show the material is a glass.
First, we appreciate Reviewer #2's for pointing out the inappropriate word "coalescence" we used. We have removed this word. Second, we agree that we need to be careful about the word "glassy" state. In the previous report from Rohit V. Pappu group, glassy intra-condensates are used to describe the condensates that have a lag time for disassembly (Holehouse and Pappu, 2018). In material science, the term "glass" is restricted to the properties of a formation of amorphous material through a decrease of temperature, and a reversible transition by a temperature increase. We do not have such physicalchemical characterization approaches to determine the material properties, such as calorimetric ideal glass transition temperature T0c, etc. Now to prevent any misleading, we have now revised the term "glassy" state to "less-motile assemblies".