Cooperative ordering of treadmilling filaments in cytoskeletal networks of FtsZ and its crosslinker ZapA

During bacterial cell division, the tubulin-homolog FtsZ forms a ring-like structure at the center of the cell. This Z-ring not only organizes the division machinery, but treadmilling of FtsZ filaments was also found to play a key role in distributing proteins at the division site. What regulates the architecture, dynamics and stability of the Z-ring is currently unknown, but FtsZ-associated proteins are known to play an important role. Here, using an in vitro reconstitution approach, we studied how the well-conserved protein ZapA affects FtsZ treadmilling and filament organization into large-scale patterns. Using high-resolution fluorescence microscopy and quantitative image analysis, we found that ZapA cooperatively increases the spatial order of the filament network, but binds only transiently to FtsZ filaments and has no effect on filament length and treadmilling velocity. Together, our data provides a model for how FtsZ-associated proteins can increase the precision and stability of the bacterial cell division machinery in a switch-like manner.

Regarding the effect of ZapA on the behavior of FtsZ polymers, the study shows that ZapA increases the width of membrane-bound filaments, and stabilizes FtsZ bundles, and also as no effect on the treadmilling properties of FtsZ polymers. Most of the ZapA-related effects occur within a narrow concentration range of ZapA, which is interpreted by the authors as a signature of highly cooperative effects. Besides, the author's results show that the binding of ZapA to the FtsZ polymers is very transiently. From these results, the authors propose a mechanism to explain the role of ZapA on the FtsZ polymers. This model may contribute to the precision and stability of the mode of action of the division machinery. 1) Although the quality of the writing and the experiments is very high, I have some difficulties with the claiming that ZapA acts on the FtsZ filaments in a highly cooperative manner, that will need further clarification as they are central to the main conclusions of the paper.
Pag. 6, line, 125 -the Hill coefficient (n) is 4.12 +/-6.84. With such high uncertainty, one cannot discriminate the mode of association, as a single Langmuir isotherm (n = 1) can also explain the data within the experimental error of the best fit parameter values. The same happens with the Hill coefficient shown in pag. 8, line 199 (N = 4.76 +/-4.21).
I would appreciate if the authors revise this analysis in other to make the claims on the mode of association. For example, I would recommend plotting all the concentration dependence data in log scale of the X-axis, which will help the readers to appreciate the nature of the processes visually.
2) The text in the Results describing that ZapA does not affect treadmilling velocity (pag 9, line 241) is the one only that it is not clear for a general reader. Please improve.
3) In the experiments showing that ZapA binds transiently to the FtsZ filaments (pag 11, line 305) uses a variant of ZapA to monitor the behavior of this protein independently of FtsZ. I would appreciate the authors adding a comment on the controls done to make sure that the variant of ZapA and the original version behaves similarly.
4) The discussion section reads well as it presents the principles, relationships, and generalizations shown by the results. However, my recommendation is to revise the whole mechanistic part, taking into account point 1 of this commentary, which I consider central for the entire study. The comparison with other self-assembling systems, like actin and tubulin polymers, is well done. In this point, I would suggest the authors consider to complete this section by including a comment on the possibility of entropic forces driving force generation of cytoskeletal networks. In this regard, the paper by Braun et al. 2016 Bioessays 38: 474-481 might be useful.

Reviewer #2 (Remarks to the Author):
In this manuscript, the authors studied how the well-conserved protein ZapA organizes FtsZ filament structures and treadmilling dynamics using a previously established supported lipid bilayer system. The authors found that ZapA cooperatively increases the spatial order of the filament network: FtsZ bundles become thicker, straighter, and more correlated in time and space above a certain ZapA concentration. Curiously, the binding of ZapA to FtsZ is transient and does not change FtsZ's treadmilling dynamics at all. While the results did not come as a surprise because in vivo studies have shown that ZapA is non-essential and has little effect on FtsZ's treadmilling dynamics (and hence one may argue what new insight this work could bring), I strongly encourage the publication of this work because it provided the most quantitative, comprehensive characterizations of the structure and kinetics of the incredibly dynamic polymer network formed by FtsA, FtsZ and ZapA. These quantitative measurements are impossible in vivo. They pointed to a model in which ZapA can increase the precision and stability of the Z-ring, facilitate its assembly during cell division, and hence supporting the scaffolding function of the Z-ring for divisome assembly. Additionally, the analyses developed by the authors to extract meaningful biochemical measurements from the complex filament network will greatly benefit other researchers in the field. I think the manuscript can be accepted as it is now, but do wish the authors to spend a bit more time to address how the transient interactions of ZapA with FtsZ could increase the stiffness of FtsZ filaments. I found the reasoning of liquid droplets-like weak interactions not convincing. Additionally, how do the authors reconcile that previous biochemical data showing that ZapA reduces FstZ's GTPase activity, but here in this work ZapA does not change FstZ's treadmilling dynamics, which are dependent on GTP hydrolysis? Perhaps previous work was measured in solution and here it is on supported lipid bilayers? 1 The manuscript by Caldas et al., which reads very well, aims at understanding the impact of the FtsZ-2 associated protein ZapA on the structural and dynamic organization of the polymers of FtsZ, the 3 main element of the bacterial division machinery in most bacteria. To achieve this aim, the authors 4 used a very elegant in vitro reconstitution approach in supported bilayers, combined with front-line 5 quantitative single-molecule imagining technologies. This experimental approach has previously 6 been successfully used by the group of Martin Loose to establish the treadmilling behavior of FtsZ 7 polymers in the presence of FtsA. The latter result was very influential in the field. In this new study, 8 the authors have made a significant effort to improve the quality of the imaging analysis. The latter 9 is a remarkable achievement, as it sets the standards for future studies on protein assembly systems 10 at membrane surfaces. 11 Regarding the effect of ZapA on the behavior of FtsZ polymers, the study shows that ZapA increases 12 the width of membrane-bound filaments, and stabilizes FtsZ bundles, and also as no effect on the 13 treadmilling properties of FtsZ polymers. Most of the ZapA-related effects occur within a narrow 14 concentration range of ZapA, which is interpreted by the authors as a signature of highly cooperative 15 effects. Besides, the author's results show that the binding of ZapA to the FtsZ polymers is very 16 transiently. From these results, the authors propose a mechanism to explain the role of ZapA on the 17 FtsZ polymers. This model may contribute to the precision and stability of the mode of action of the 18 division machinery. 19 1) Although the quality of the writing and the experiments is very high, I have some difficulties with 20 the claiming that ZapA acts on the FtsZ filaments in a highly cooperative manner, that will need 21 further clarification as they are central to the main conclusions of the paper. scale of the X-axis, which will help the readers to appreciate the nature of the processes visually. 29 We thank the reviewer for pointing this out. First, we want to emphasize that in our study, we used 30 the Hill coefficient merely as a global measure of cooperative behavior. The Hill coefficient we 31 obtained does not allow to provide a molecular-level interpretation of our data. 32 behavior of the FtsZ-ZapA system: two different stable states and a critical ZapA concentration that 36 triggers the transition between them. These three features are inconsistent with a graded response. 37 38 Next, following the reviewer's advice, we have improved our analysis and fitting procedure. 39 Previously, the non-linear regression we used in our analysis relied on the assumption that the Hill 40 coefficient relies on a normal distribution. Now, we have implemented a bootstrap method for the 41 robust estimation of fitting parameters without underlying assumptions. After the initial fitting 42 procedure is performed, this method generates new sets of data points by reshuffling randomly the 43 residuals of the fit. The new data sets generated are then used to perform new fits and obtain a 44 distribution of fitting parameters. Using this method, we could robustly calculate the 90% 45 confidence interval of the distribution, which consistently shows that the corresponding Hill 46 coefficient is much larger than 1. Interestingly, we found that the distribution of values obtained for 47 the Hill coefficient was bimodal, presumably because the range and number of concentrations used 48 in our titration experiments was limited. 49 Finally, to convince the reader that fitting a Hill equation with a Hill coefficient nH > 4 provides a 50 better fit than with nH = 1, we show the corresponding curves in the supplemental data 51 (Supplementary Figure 4) and used two objective evaluation criteria: the reduced ci-squared values 52 and Bayesian Info Criterium (BIC). Importantly, by calculating ΔBIC = BIC(nH=1) -BIC(nH > 4) to 53 compare different Hill coefficients, we can objectively exclude nH = 1 as the best possible fit (see 54 Supplementary Table 5 and Supplementary Figure 4). 55 56 We want to thank the reviewer again for these comments, as it helped use to improve our analysis 57 and to more convincingly show that the effect of ZapA on FtsZ is highly cooperative. 58 2) The text in the Results describing that ZapA does not affect treadmilling velocity (pag 9, line 241) 59 is the one only that it is not clear for a general reader. Please improve. 60 We thank the reviewer for pointing this out. We believe that the message of this section is now 61 clearer (see lines 412). appreciate the authors adding a comment on the controls done to make sure that the variant of 65 ZapA and the original version behaves similarly. 66 We thank the reviewer for this comment since confirming the activity of the fluorescently labeled 67 protein is a very important control. Indeed, we had not provided a quantitative comparison for the 68 effect of non-labelled and Cy5-labelled ZapA on the FtsZ/FtsA pattern. 69 We provide now an additional table (Supplementary Table 4  We thank the reviewer for his positive comments and bringing the very interesting paper by the Diez 88 and Dogic lab to our attention. We read this review as well as the corresponding research paper by 89 Lansky et al. 1 carefully and can conclude that the biggest difference between the ZapA-FtsZ and 90 Ase1-microtubule system is that Ase1, in contrast to ZapA, shows a very long residence time on the 91 filament lattice. This persistent binding time allows Ase1 to diffuse along the microtubules lattices, 92 and to generate entropic forces. In contrast, the residence time of ZapA is simply too short for such 93 kind of effect on FtsZ filaments. We think this is a very important difference between these two 94 proteins systems and therefore include it in our discussion (lines 489 ff. We want to thank the reviewer for such positive feedback on our work! The reviewer is correct that 120 previous studies found a decreased GTPase activity of FtsZ in the presence of ZapA and that this 121 effect would imply decreased treadmilling dynamics. In contrast, we found that ZapA has no effect 122 on the treadmilling velocity. We want to point out that a reduction of GTPase activity was only found 123 at a non-physiological pH of 6.5, which promotes intrinsic bundling of FtsZ 2-4 . In contrast, the effect 124 was almost absent at a physiological pH of 7.5 (ref. 5). Furthermore, as the reviewer pointed out 125 correctly, the potential effect of the membrane on the GTPase activity was not addressed in these 126 previous studies. In fact, these conflicting reports were one of the reasons why it was important to 127 find out how ZapA actually affects the treadmilling behavior of membrane-bound FtsZ (see lines 49-128 51 and 280-288). Our results now unambiguously show that ZapA has no effect on the 129 polymerization dynamics of membrane-bound FtsZ filaments, while it changes the architecture of 130 the filament network. 131 How ZapA can increase the stiffness of FtsZ filaments is an interesting point. We believe that the 132 reviewer is referring to the study by Dajkovic et