While contraction of sarcomeric actomyosin assemblies is well understood, this is not the case for disordered networks of actin filaments (F-actin) driving diverse essential processes in animal cells. For example, at the onset of meiosis in starfish oocytes a contractile F-actin network forms in the nuclear region transporting embedded chromosomes to the assembling microtubule spindle. Here, we addressed the mechanism driving contraction of this 3D disordered F-actin network by comparing quantitative observations to computational models. We analyzed 3D chromosome trajectories and imaged filament dynamics to monitor network behavior under various physical and chemical perturbations. We found no evidence of myosin activity driving network contractility. Instead, our observations are well explained by models based on a disassembly-driven contractile mechanism. We reconstitute this disassembly-based contractile system in silico revealing a simple architecture that robustly drives chromosome transport to prevent aneuploidy in the large oocyte, a prerequisite for normal embryonic development.https://doi.org/10.7554/eLife.31469.001
Muscle contraction forms the basis of animal locomotion, and at the microscopic scale, contraction of actin filament (F-actin) networks drives a plethora of essential cellular processes. Most prominently, the cortex, a thin and entangled network of F-actin underlying the plasma membrane determines the shape of animal cells, and changes in cell shape underlie cell migration, cell division and tissue morphogenesis (Munjal and Lecuit, 2014; Pollard and Cooper, 2009; Salbreux et al., 2012). While myosin motor activity, non-muscle myosin II in particular, is the main driver of these processes, filament dynamics has been shown to play a critical role as well. For example, cell migration is to a large part driven by polymerization of actin filaments at the cell front and depolymerization at the rear (Cramer, 2013; Mseka and Cramer, 2011; Pollard and Borisy, 2003; Ridley, 2011). Cytokinetic ring closure has been shown to be driven by depolymerization of filaments in some species, although myosin activity certainly is the main driver in others (Green et al., 2012; Mendes Pinto et al., 2012; Neujahr et al., 1997).
In addition to its cortical functions, recent studies revealed extensive 3D F-actin networks in the bulk cytoplasm of large oocytes and embryos with essential functions in intracellular transport processes (Field and Lénárt, 2011). For example, mouse oocytes feature a dynamic cytoplasmic F-actin network required to position the nucleus (Almonacid et al., 2015), the meiotic spindle (Almonacid et al., 2014; Azoury et al., 2008; Schuh and Ellenberg, 2008), and to transport vesicles from the cell center to the cortex, including cortical granules to prevent polyspermy (Cheeseman et al., 2016; Schuh, 2011). While myosin II and myosin V motors have been involved in transporting the spindle and vesicles, respectively, filament dynamics appear to have a key role in these processes as well, regulated by Fmn2/Spire1/2 as well as cortical Arp2/3 nucleators, and other factors (Chaigne et al., 2013; Cheeseman et al., 2016; Holubcová et al., 2013; Pfender et al., 2011; Schuh, 2011; Yu et al., 2014).
Generally, while the mechanism of contractile force generation is well understood in sarcomeric-like assemblies, such as muscles or stress fibers, the mechanisms in disordered F-actin networks, such as the above quasi-2D cortical and 3D cytoplasmic networks, remains much more elusive (Lenz et al., 2012; Murrell et al., 2015). Additionally, computational models and theoretical work so far focused largely on disordered F-actin networks formed by non-dynamic filaments inspired by in vitro reconstituted contractile systems composed of stable filaments (Alvarado and Koenderink, 2015; Belmonte et al., 2017; Bendix et al., 2008; Köhler and Bausch, 2012; Murrell and Gardel, 2014). While it is true that these non-dynamic systems contract efficiently, it is also clear that filament dynamics has key contributions in vivo, as illustrated above. Indeed, although much less investigated, theoretical work has shown that F-actin dynamics is able to generate not only protrusive but also contractile stress (Sun et al., 2010; Zumdieck et al., 2007).
In the past, we discovered a prominent example of F-actin-driven intracellular transport in starfish oocytes, whereby a 3D F-actin network collects chromosomes scattered in the large oocyte nucleus and transports them to the forming microtubule spindle (Lénárt et al., 2005). This F-actin network forms in the nuclear region after nuclear envelope breakdown (NEBD), at the onset of meiosis, and is essential to prevent formation of aneuploid eggs (Mori et al., 2011). It contracts homogeneously and isotropically mediating long-range and size-selective transport of chromosomes sterically trapped in the network. While the contraction of the network is isotropic and homogeneous, connections of the network to the cell cortex provide directionality, resulting in transport towards the microtubule spindle, located at the cortex (Mori et al., 2011).
However, the mechanism underlying contraction of this 3D F-actin network remained unknown. To reveal it, we monitored contractile behavior using the chromosomes as natural probes combined with quantitative imaging of filament dynamics in live oocytes. In parallel, we developed physical models of different mechanisms of contraction, and compared the predictions of these models to the experimental system challenged by physical and chemical perturbations. We find that observations are well fitted by models in which contraction is driven by F-actin disassembly, but not by models of motor-driven contractility. Comparisons of experimental observations to in silico reconstruction of this contractile system reveal a novel and remarkably simple architecture to transport chromosomes, to prevent aneuploidy in the large oocyte.
As we showed previously, chromosomes are sterically entrapped and thereby transported by a contracting F-actin network (Mori et al., 2011). Chromosomes can thus be used as probes to monitor the network’s contractile behavior without potential artifacts associated with F-actin labeling. Therefore, we optimized the spatial and temporal resolution to image and track H2B-3mEGFP or -mCherry labeled chromosomes in 3D during actin-driven transport, 2 to 8 min after NEBD (Monnier et al., 2012; Mori et al., 2011) (Figure 1A,B). We reported previously that contractile activity is isotropic and while the network normally contracts to the animal pole, this merely results from passive anchoring of the network to the cell cortex (Mori et al., 2011). Thus, here we analyzed the pair-wise distances between chromosomes to characterize the network contraction rate, a measure that is independent of the location and even existence of the anchor (Figure 1—figure supplement 1).
The trajectories revealed that any pair of chromosomes, irrespective of their initial location, exhibited a constant approach velocity towards one another during transport (Figure 1C). Extrapolation of these linear trajectories showed that all chromosomes would meet at nearly the same time and location, if the network were to contract all the way to a single point (Figure 1C). This implies that pair-wise chromosome velocities depend linearly on their initial separation distance (Figure 1D), indicative of a spatially homogeneous and isotropic contraction (Mori et al., 2011). This means that the ratio () between pair-wise velocities () and initial pair-vise distance () is constant for all pairs of chromosomes:
In theory, could be simply derived as the slope of the fitted line on Figure 1D. In practice, however, we found the initial pair-wise chromosome separation distance difficult to determine, as motions caused by the collapse of the nuclear envelope at NEBD occur at the onset of network contraction. Therefore, we instead determined by extrapolating its value (Figure 1E,F). Generally, at any time t for any one pair of chromosomes the contraction rate is given by:
as the speed is constant, , leading to:
Thus, we calculated the pair-wise approach rate, α(t) on 2 min intervals throughout F-actin-driven transport. As expected, it increases over time according to Equation 3, allowing us to determine the value of α0 (Figure 1F). The value of α0 derived by this method from a large number of oocytes provided a precise estimate of the remarkably invariable initial contraction rate; α0 = 0.085 ± 0.017 min−1 (N = 53 ± S.D.). Note that the reciprocal of this value, t0 = 1/α0= 11.76 ± 0.27 min is the ‘convergence time’, the time that would be required to transport the chromosomes from their initial position to the final meeting point.
Taken together, analysis of chromosome trajectories as probes embedded in the contracting network reveals a remarkably regular behavior indicative of a homogeneous and isotropic contraction characterized by a single parameter, the initial contraction rate, .
To visualize F-actin dynamics underlying this very regular contractile behavior, we expressed the fluorescent F-actin marker UtrCH (mEGFP3-UtrCH or mCherry3-UtrCH) in oocytes (Burkel et al., 2007). In addition, at the time of NEBD we spiked in a small amount of fluorescent phalloidin in a different color. Phalloidin bound rapidly and irreversibly to filaments present at the time of injection allowing us to distinguish the initial network present at NEBD from F-actin that polymerize later (Figure 2A).
Analysis of these pulse-label recordings revealed that while the network in the nuclear region contracts, new filaments are polymerized on nuclear envelope remnants (NER) forming a new network filling up the space around the contracting initial network (Figure 2A,C, Video 1). Intensity profiles showed that polymerization occurred predominantly at the boundary defined by NER, and to a limited extent within the network (Figure 2B). Correspondingly, we distinguish the network initially filling the nuclear space at NEBD, referred hereafter as ‘old network’, from the ‘new network’ that is polymerized from the NER boundary after NEBD.
To quantify the rate of contraction of the old network, we fitted a circle to the phalloidin-labeled region and recorded its diameter, D over time (Figure 2E). This revealed that the old network contracted to approximately half of its initial size, and with a nearly constant speed. The rate of contraction derived this way was similar, although somewhat lower than the rate determined by tracking chromosomes (compare Figures 1F and 2E). We then performed direct comparisons on a single cell basis combining phalloidin pulse-labeling with labeling chromosomes (Figure 2D). We found the contraction rate extracted from the pair-wise chromosome approach and the rate derived by measuring the diameter of the phalloidin-labeled ‘old network’ to correlate precisely (Figure 2F). Thus, similar contraction rates are derived by the two methods, however, it appears that phalloidin used to visualize the old network slightly slows contraction and increases variability, likely dependent on the amount of phalloidin injected.
Altogether, visualizing filament dynamics identified two major components in this contractile system: the contraction of the initial ‘old network’ and the polymerization of the ‘new network’ along the NER boundary, filling the space as the old network contracts. Chromosomes are embedded and transported by the old network, thus the contraction rate, derived from pair-wise chromosome approach is a direct measure of the old network’s contraction rate.
Having identified the two main activities, polymerization at the NER and contraction of the network, we next wondered whether chromosome transport is driven by polymerization compressing the network, or contractile forces generated within the network. To distinguish these two possibilities, we ablated the F-actin network in a 3D volume using a pulsed infrared laser. We expected the network to collapse into the ablated region, were pushing forces generated by polymerization to dominate. Instead, a recoil response would evidence tension and thus active contractile forces within the network (Figure 3A).
We observed a prominent recoil response with the network rapidly retracting away from the ablated region (Figure 3B–D, Video 2). This shows that the network is under tension and suggests that pushing by polymerization does not have a major contribution. After retraction, the gap did not close, but was filled progressively by F-actin suggesting that in response to the perturbation, filament assembly may occur throughout the network (Figure 3B,D). The recoil response did not depend on the direction of the cut, consistent with isotropic contraction (Figure 3—figure supplement 1).
To test that the recoil observed after laser ablation is not an artifact, we also performed this experiment in oocytes where F-actin was stabilized by injecting a large amount of recombinant, fluorescently labeled UtrCH. Similar to mouse oocytes (Holubcová et al., 2013), high concentration of UtrCH effectively stabilized the network and prevented its contraction (Figure 3B–D, Video 2). Stabilization also prevented the recoil response confirming the specificity of laser ablation.
We thus conclude that production of F-actin at the NER boundary is not the major driver of contraction. Instead, our data show that active forces generated within the network drive contractility. Importantly, the dramatic effects of UtrCH stabilization, along with the minor slowing effect of phalloidin injection above, suggest that F-actin dynamics has a key role in contractile force generation.
To investigate a possible role for filament dynamics in contraction, we quantified changes in filament amounts in space and time. We used the fluorescence intensity of mEGFP3-UtrCH, which binds to filamentous actin with high affinity, as a proxy for filament mass (Figure 4A) (Belin et al., 2014; Burkel et al., 2007; Chaigne et al., 2013; Singh et al., 2017). To distinguish the ‘old’ and ‘new’ networks we assumed the old network to be a sphere with a radius that decreases according to the contraction rate derived from chromosome approach (Figure 4A):
Such quantifications extrapolated to the 3D volume revealed a continuous decrease of F-actin mass within the old network through the contraction process (Figure 4B, for F-actin densities see Figure 4—figure supplement 1A). To relate this measure to the rate of filament disassembly, we expressed the mass of F-actin as the number of filaments per unit volume, and defined the rate of disassembly, as:
where is the initial rate of F-actin disassembly, the dispersion of the length distribution of F-actin and the normalized average length of F-actin (Figure 4B,C, see Supplementary file 1 for details). We found that Equation 5 fitted well the evolution of F-actin mass allowing us to extract the values of and (Figure 4B,C). For untreated oocytes, we found = 0.157 and = 0.6. These parameters are dimensionless as neither the length nor the number of filaments are known in absolute terms.
Next, we explored potential mechanisms that can explain contraction under the constraints of the observed filament dynamics. F-actin networks are commonly modeled as active viscoelastic gels (Hannezo et al., 2015; Hawkins et al., 2009; Julicher et al., 2007; Kruse et al., 2006; Marchetti et al., 2013). For example, this approach was successfully used to model F-actin contraction in droplets of Xenopus egg extracts, with size and time scales comparable to our experimental system (Lewis et al., 2014). Thus, we developed a similar model assuming that effects of filament orientation can be neglected (i.e. assuming isotropic stresses), which is justified especially when no large-scale ordering of the system is observed (Hannezo et al., 2015; Hawkins et al., 2014; Lewis et al., 2014). We further assumed that viscosity mainly arises from network rearrangements rather than viscous drag, or in other words, that the main factor opposing contraction is the unbinding kinetics of cross-linkers (Blanchoin et al., 2014; Salbreux et al., 2012).
We modeled the F-actin network as a viscoelastic gel in which viscous stress (originating from friction within the network) and elastic stress (originating from F-actin deformation) oppose the network contractile stress . We further generalized the model to allow viscosity and elasticity of the system to depend on F-actin concentration, noted A.
We assumed the elastic stress to be isotropic and neo-Hookean (Lewis et al., 2014):
in which is the strain and is the elastic modulus, with characterizing the power law dependence of elasticity on A.
The viscous stress is written as:
where is the viscosity, with µ = being the power law dependence of viscosity on A.
where is the contraction coefficient and is the power law dependence of contraction on A (Figure 4D).
Unlike in a motor-driven process, we expect F-actin disassembly to generate a contractile stress dependent on the density of filament ends, and on their disassembly rate. Given that corresponds to the density of ends and the disassembly rate, the contractile stress can be written as (Figure 4D; see Supplementary file 1 for details):
At these scales, we assume inertia to be negligible, thus the gradient of stresses is zero:
We determine the velocity field of network displacement, which is set by force balance, by integrating Equation 10 in space. Then, we obtain , and by integrating the equations of mass balance over time:
in which is the strain relaxation rate. We integrated Equations 10, 11, 12 and 13 numerically (see Supplementary file 1 for details), and from the solution calculated using Equation 2. Since we could not find an analytical solution to the system of equations, we are unable to provide a simple formula for that is derived from first-principles. In addition, we defined the parameter describing adhesion at the nuclear boundary, such that the adhesive stress at the boundary is − (see Supplementary file 1).
In the above framework of the active-gel theory, we first investigated whether a motor-driven (referred hereafter as model M) or a disassembly-mediated process (referred hereafter as model D) could possibly account for the observed decrease in F-actin mass during contraction. We determined the viscoelastic parameters by randomly sampling parameter values and fitting the integration results to the contraction rate measured experimentally (Table 1 and Figure 1F). We find that both models can recapitulate the observed behavior (Figure 4E). However, the models additionally predict the contraction rate to depend directly on disassembly in model D, whereas this dependence is indirect and additionally depends on motor activity for model M. Thus, it is expected that the dependence of contraction rate on disassembly rate will differentiate the two models. Motivated by this prediction, we experimentally manipulated motor activity and the disassembly rate to assess whether model M or D can predict the observed contraction rate in various perturbation experiments.
If the contractile process is motor-driven, non-muscle myosin II (NMYII) is the prime candidate to mediate it, as myosin II is the only myosin type that is able to assemble into minifilaments and generate force by pulling filaments towards one another (Murrell et al., 2015; Sellers, 2000; Syamaladevi et al., 2012). Indeed, in whole-oocyte proteomics analyses we found NMYII to be abundant in the starfish oocyte (data not shown). Thus, to test the possible role of NMYII we inhibited its motor activity by the small-molecule inhibitor, blebbistatin. Further, we over-expressed myosin regulatory light chain (MRLC), which we found in starfish oocytes to enhance NMYII activity (Bischof et al., 2017). To our surprise, these perturbations did not affect contraction rates derived from pair-wise chromosome approach (Figure 5A,B). To confirm the effectiveness of the treatments, we followed the same oocytes until the end of the first meiotic division, when the polar body is extruded, accompanied by a global surface contraction wave driven by NMY-2 (Bischof et al., 2017). Consistent with our earlier results, the amplitude of the contraction wave was significantly decreased upon blebbistatin treatment, while it was increased in oocytes over-expressing MRLC (Figure 5C,D). Thus, chromosome transport and derived contraction rates were unaffected by NMY-2 perturbations, which were effective as assessed by their effects on surface contraction waves in the same oocytes. Consistently, we saw no effects on chromosome transport following inhibition of the two common pathways of NMYII activation through myosin light chain kinase and Rho kinase, by injecting ML-7 and Y-27632, respectively (Figure 5A–D).
The only alternative scenario we could envisage is a mechanism similar to that driving long-range vesicle transport, nuclear and spindle positioning in mouse oocytes (Chaigne et al., 2013; Holubcová et al., 2013; Schuh, 2011). In this case, directional transport is thought to result from polymerization of actin filaments on vesicles by Fmn2/Spire1/2 nucleators and transport of vesicles on these filaments by myosin V motors. However, a similar mechanism is unlikely to function is starfish oocytes, as this mechanism is strictly dependent on vesicles and in the nuclear region of starfish oocytes we did not observe any membranous structures (Figure 5—figure supplement 1A), and the time scales do not match with the process running an order of magnitude slower in mouse as compared to starfish oocytes. Additionally, when we expressed a dominant-negative tail construct of myosin Vb (Schuh, 2011), we observed no effect on chromosome transport and derived contraction rates (Figure 5—figure supplement 1B,C).
As shown above, the mass of F-actin in the old network continuously decreases, and filament stabilization inhibits contraction, suggesting a direct link between F-actin disassembly and network contraction. To test this, we titrated in a stabilizing activity by injecting increasing amounts UtrCH using a quantitative microinjection method (Jaffe and Terasaki, 2004). Analysis of pair-wise chromosome approach showed that UtrCH slowed chromosome transport in a dose-dependent manner (Figure 6A,C). We derived the contraction rates for each condition as for Figure 1E,F, and using the contraction rate we calculated the size of the old network and the F-actin mass contained within, as for Figure 4A. Analysis of F-actin mass in the old network revealed that UtrCH slows net F-actin disassembly in a dose-dependent manner (Table 2, Figure 6B,D). Quantification of changes in mean UtrCH intensity in a fixed size region on the other hand revealed similar F-actin density despite the very different contraction rates (Figure 6E and Figure 4—figure supplement 1B, respectively). In other words, stabilization by UtrCH did not render the network denser or sparser, it merely slowed its contraction. As a result, contraction rates () and decrease in F-actin mass were reduced in a correlated manner, indicating a direct coupling between disassembly and contraction.
We next compared these quantitative observations to the predictions of models of motor- and disassembly-driven contractile mechanisms. We first determined the viscoelastic parameters from control experiments (i.e. oocytes injected with PBS; Supplementary file 2) that were very similar to those for untreated oocytes (Table 1). We then used the measured disassembly rate and the dispersion of filament length to predict the contraction rate for each condition (without changing any other viscoelastic parameters) (Table 2). The two models, motor- and disassembly-driven, predicted very distinct behaviors in response to increasing UtrCH concentrations. For intermediate and high UtrCH levels, the motor-driven model predicted a fast increase of contraction rate, in contrast to the disassembly-driven model predicting contraction rate to remain nearly constant over time, well matching experimental observations (Figure 6F). Consistently, the initial contraction rate, derived from the disassembly-driven, but not from the motor-driven model, predicted contraction rates in response gradual stabilization (Table 3).
This is explained by the fact that in the disassembly-driven mechanism contraction and disassembly rates are inherently coupled. In the motor-driven model, decreasing disassembly rate caused an increase in F-actin concentration that resulted in more effective motor action, leading to the observed increase of contraction rate over time. Similar results were obtained with a purely viscous model (Figure 6—figure supplement 1 and Supplementary file 1).
To further test the active role of disassembly in driving contraction, we used Latrunculin A, a toxin that sequesters actin monomers. Therefore, Latrunculin A is expected to block actin filament assembly and to enhance depolymerization by reducing the effective monomer concentration (Spector et al., 1989). Thus, in a dynamically assembling and disassembling network the net effect will be shifting the equilibrium towards disassembly.
We observed a dramatic increase in contraction rate upon acute treatment with a high dose of Latrunculin A (Figure 7A, Video 3). Starting approx. 50 s after Latrunculin A addition, the network diameter begun to decrease at rates approx. three-fold higher than DMSO controls (Figure 7B). Quantification of F-actin mass in the rapidly contracting old network revealed that F-actin amount decreased at a correspondingly faster rate (Figure 7C, Figure 4—figure supplement 1C and Table 2).
Similar to above, we next determined the viscoelastic parameters of the network in the DMSO control (Supplementary file 3, note that this time was derived from fitting the network diameter since chromosomes escape and thus cannot be used for the analysis, Figure 7—figure supplement 1). We then used the measured disassembly rate and the dispersion of filament length to predict by the two models (Table 2). While the motor-driven model predicted the contraction rate to change little upon increased disassembly, the disassembly-driven model predicts an increase in contraction rate closely matching experimentally observed values (Figure 7D).
In summary, the observed changes in contraction due to perturbations of motor activities, as well as stabilizing and destabilizing filaments, are well predicted by a model in which contraction is driven by disassembly, but not by a motor-driven model. While in many aspects the two models are similar (e.g. both models include the ‘fluidization’ of the network by disassembly), the principal difference is that in the disassembly-driven model disassembly and contraction are directly coupled.
Our data suggest that the F-actin network transporting chromosomes in starfish oocytes contracts by a mechanism mediated by filament disassembly. To explore the mechanism at the filament level, we implemented an agent-based model in the Cytosim software (Nedelec and Foethke, 2007). We based the model on recently published cytosim simulations, used to support a general theoretical framework of contractility in disordered filament networks, albeit in that case with static filaments (Belmonte et al., 2017). As in the published model, we used a circular 2D geometry, assuming a ‘nuclear region’ with a radius of 10 μm initially filled by filaments having the rigidity of F-actin. We then added a few features to model the starfish chromosome transport system: first, disassembly occurring at the same rate for all filaments throughout the network; and second new filament assembly at the boundary. We also introduced ‘chromosomes’, passive objects embedded in the network for the purpose of producing simulated trajectories.
We first tested whether the combination of disassembly and cross-linking could drive network contraction. Consistent with previous theory and experiments (Belmonte et al., 2017; Bendix et al., 2008; Braun et al., 2016), in the absence of molecular motors the disassembling network did not contract over a broad range of parameters (number of filaments, cross-linkers and their unbinding rates) tested (Figure 8—figure supplement 1A–C).
This suggested that contraction in our system either requires a molecular motor or a specific molecular activity that is capable of harnessing F-actin disassembly to produce contractile force. This latter can be a molecule capable of tracking the depolymerizing end of a filament and forming a connection to a neighboring filament at the same time (Zumdieck et al., 2007). Introducing such elements into the simulations, referred to as end-binding cross-linkers, resulted in robust large-scale contraction of the disassembling network in the absence of molecular motors (Figure 8—figure supplement 1A,B).
We then tested the effects of assembly of new filaments at the network boundary. This reproduced well the observed behavior, whereby newly produced filaments fill up the space left by the ‘old’ contracting network. Interestingly, we found that contraction rate is rather insensitive to the rate of filament assembly at the boundary: insufficient or excess filament production rates led to inhomogeneity in the network, but this did not significantly affect the rate of contraction (Figure 8—figure supplement 1D). This suggests that the contraction of the old network, and new filament assembly at the NER boundary do not necessarily need to be coupled, and this is consistent with experimental observations. First, we observed an accumulation of filaments at the boundary of old and new networks in oocytes in which contraction and filament production was imbalanced by UtrCH stabilization (Figure 8—figure supplement 1E). Second, in untreated oocytes we observed a gradual accumulation of filaments at the NER boundary from the vegetal to the animal pole. This can be explained by constant rate of filament production all along the boundary, and while these produced filaments are distributed to fill in the relatively larger gap at the vegetal pole, filaments accumulate at the animal pole where there is no gap (Figure 8—figure supplement 1E). Thus it appears that filament production does not contribute to force production; it rather serves to fill in the gap between the ‘old network’ and cytoplasmic F-actin networks, mechanically isolating the contractile system in the nuclear region from the cytoplasm (Figure 8—figure supplement 1E).
As the final step, we combined these components of the contractile system, and set the rate of global disassembly and assembly at the boundary to match experimental observations. We then simulated two scenarios: (i) contraction driven by an end-binding cross-linker harnessing depolymerization for generating contractile force; (ii) myosin II-like motor activity driving contraction of the disassembling network (in presence of cross-linkers, but without end-binding cross-linkers) (Figure 8A,B). Both simulations produced robust contraction. However, while contraction mediated by the end-binding cross-linker closely reproduced several key features of the experimental system, motor-driven contraction showed a distinct behavior not observed experimentally (Figure 8C, Video 4). First, the disassembly-driven mechanism invariably produces contraction with a constant speed, as observed experimentally, while motor-driven contraction typically accelerates (Figure 8C). Second, while a strictly quantitative comparison of the 2D simulations and the 3D experimental system is not possible, network density remains nearly constant during contraction in the disassembly-driven simulations, as in experiments, while during motor-driven contraction network density continuously increases (Figure 8D). Further, the simulations reproduce the fact that contraction rate for a disassembly-driven mechanism is tightly coupled to the rate of disassembly (Figure 8E).
Taken together, the simulations are consistent with the active gel theory, and additionally identify the ‘disassembly harnessing factor’ as the key component driving the contractile system. By searching the literature for molecular candidates for such factor with features predicted by the simulations, we found that a formin-like molecule may fit these requirements. It has been shown that in vitro and under specific conditions the formin, mDia1 is able to track depolymerizing ends and generate pulling forces on the disassembling filament end (Jégou et al., 2013; Mizuno et al., 2007). In addition, formins have been shown to have cross-linking activities on their own or by binding to other actin regulators such as Spire (Esue et al., 2008; Montaville et al., 2014). In this way a hypothetical formin may serve as an effective ‘harnessing factor’.
We tested this hypothesis by inhibiting all formin-like activities in the oocyte by SMIFH2, a small-molecule inhibitor of the FH2 domain contained and essential for the activity of all known formins (Rizvi et al., 2009), and which has already been evaluated in mouse and starfish oocytes (Kim et al., 2015; Ucar et al., 2013). We thus quantified the contraction rate and F-actin mass in oocytes treated with SMIFH2 and compared to DMSO controls (Figure 9A–D and Figure 4—figure supplement 1D, Table 2). Impairing formin activity led to a significant decrease of network contraction rate and a reduction of F-actin mass loss, similar to stabilization by UtrCH, and well predicted by the disassembly-driven model (Figure 9E; Table 4 and Supplementary file 4).
While these results suggest that a formin-like molecular activity may be harnessing the free energy originating from disassembly of actin filaments, unfortunately the experimental tools currently available in starfish oocytes are limited, and thus identification of this factor will remain a challenge for the future.
Animal oocytes typically have an exceptionally large nucleus storing nuclear proteins such as histones needed for early embryonic development. Oocytes also divide very asymmetrically in order to retain these nutrients in a single egg. The requirement for this highly asymmetric division constrains the size of the meiotic spindle. Therefore, across animal species, the meiotic spindle is small and this necessitates additional mechanism to collect and transport chromosomes scattered in the large nuclear volume to the forming microtubule spindle.
Here, we propose and reconstruct in silico a simple and robust architecture for an F-actin-based contractile system adapted to the essential function of transporting chromosomes in the large oocyte. First, the process is triggered by NEBD opening the previously inaccessible nuclear compartment to cytoplasmic actin, actin assembly factors and cross-linkers resulting in the rapid assembly of a cross-linked F-actin network in the nuclear region. As revealed by our simulations, the only additional components necessary are: (i) filament production along the boundary to separate the nuclear region from cytoplasmic F-actin networks allowing contraction towards the center, and (ii) overall filament dynamics shifted towards disassembly. In the presence of a ‘depolymerization harnessing factor’ this system undergoes robust contraction exhibiting features that closely match the experimental observations. These include constant speed, stable network density during contraction (well matched to the size of chromosomes [Mori et al., 2011]), and graded response to filament stabilization and destabilization, but no effect of motor inhibition. We find that the rate of disassembly is the key parameter controlling the rate of contraction, while contraction rate is broadly insensitive to filament production at the boundary. Finally, an additional elegant feature of the system is that the disassembling network will eventually release the transported chromosomes, as it loses connectivity and stops contracting. By contrast, in the absence of disassembly, a contracting network would necessarily densify, in which case the chromosomes would be surrounded by a tight meshwork of F-actin at the end of contraction, which would likely interfere with their capture by spindle microtubules.
In conclusion, we investigated here a contractile system adapted to transport chromosomes in large oocytes. Both experiments and simulations point to an F-actin disassembly-driven mechanism driving this process, as a simple and robust solution to the challenge posed by the size and geometry of the starfish oocyte. Network contraction can be seen here as a large-scale treadmilling system, where F-actin disassembles in bulk and is replenished by filaments polymerized at the NER boundary. However, the evidences lead us to conclude that de-novo filament assembly at the NER boundary does not participate in force generation, which rather originates in the bulk of the network.
Key questions remain open regarding the detailed molecular mechanism underlying disassembly-driven force generation. The first question is the exact source of free energy driving the process. While actin polymerization is well established to produce force, at cell protrusions, for example (Footer et al., 2007; Krause and Gautreau, 2014), force production by disassembly has so far only been evidenced in vitro for specific actin concentration (Jégou et al., 2013). It is clear that microtubule depolymerization can produce force (Grishchuk et al., 2005), that is harnessed for example to drive separation of sister chromatids in anaphase (Cheeseman and Desai, 2008; Westermann et al., 2006). Similar to tubulin, actin polymerization is tightly coupled to nucleotide hydrolysis driving the system out of equilibrium that may enable generation of force by depolymerization (Braun et al., 2016). Alternatively, entropic forces generated either by crowding agents surrounding filaments or by diffusible cross-linkers confined in filament overlaps could drive contraction (Braun et al., 2016), but we have not tested this possibility.
Another important question is the identity and nature of the factor harnessing the free energy originating from disassembly of actin filaments, and converting it into contractile force. Intriguingly, experiments in vitro and our initial experiments in starfish oocytes hint at formins as potential candidates for this function. These results need to be confirmed, since they rely on a small-molecule inhibitor, SMIFH2, which has been shown to inhibit formins in starfish oocytes (Ucar et al., 2013), but its specificity and selectivity profile is yet to be characterized. Secondly, the observed effect may be indirect resulting from the altered balance between other formin-driven actin assembly pathways and Arp2/3 nucleated structures, as recently showed in yeast (Burke et al., 2014). Finally, although Jégou and colleagues clearly demonstrated tracking of filament ends by mDia1 (Jégou et al., 2013), we are at present unable to assess how similar or different these in vitro conditions may be from those in starfish oocytes, where (dis)assembly dynamics at filaments ends is likely to be regulated by multiple, so far unidentified factors.
It will be very exciting to test in the future whether formins, so far primarily considered to promote polymerization, may be also involved in transmitting force at disassembling filaments ends in vivo. However, at this point we are not able to exclude other scenarios, such as rapid re-binding of a cross-linker to the shrinking filament end (Mendes Pinto et al., 2012). Further, the rate of disassembly could be affected by severing, generating more depolymerizing ends, and the actual rate of depolymerization at the end set by capping and depolymerizing factors. More elaborate models are necessary to explore these scenarios. In either case, the contractile 3D network we analyzed here, adapted to the task of transporting chromosomes in the large starfish oocyte, will continue to be a valuable system to further explore the important mechanistic details of the so far poorly understood motor-independent mechanisms of contraction.
Starfish (P. miniata, also known as Asterina miniata) were obtained from Southern California Sea Urchin Co., Marinus Scientific, South Coast Bio-Marine, or Monterey Abalone Co. and maintained in seawater tanks at 15°C at the European Molecular Biology Laboratory (EMBL) Marine Facility. Oocytes were isolated and injected using mercury-filled needles as described elsewhere (Jaffe and Terasaki, 2004). Recombinant protein markers were injected shortly (10–15 min) after initiation of meiosis, whereas mRNA injections were done the day before and incubated overnight at 14°C to obtain sufficient levels of protein expression. Meiotic maturation was triggered by the addition of 10 µM 1-methyladenine (ACROS Organics). NEBD typically started 25 min after hormone addition, and only oocytes starting NEBD between 15 and 35 min were analyzed. Oocyte centrifugation was performed at 2500 rpm for 20 min (Multifuge 3; Heraeus) at 4°C, as detailed elsewhere (Matsuura and Chiba, 2004; Mori et al., 2011).
H2B-3mEGFP, H2B-mCherry (Neumann et al., 2006), mEGFP3- and mCherry3-UtrCH (Burkel et al., 2007) and MRLC-mEGFP (Bischof et al., 2017) were subcloned into pGEMHE for in vitro transcription as described elsewhere (Lénárt et al., 2003). Mouse sequence of Myosin Vb Tail was used to search for homologs in the P. miniata transcriptome by BLASTP (Altschul et al., 1990). Hits were considered homologs when the e-value was < 10−20. Identified cDNA sequence was synthetized by GENEWIZ, fused to mEGFP and subcloned into pGEMHE for in vitro transcription. Capped mRNAs were synthesized from linearized templates using the AmpliCap-Max T7 High Yield Message Maker and extended with poly(A) tails using the Poly(A) Tailing kit (CellScript). mRNAs were dissolved in 11 µl RNase-free water (typically at 8–12 µg/µl) and injected to 1–5% of the oocyte volume. Histone H1 from calf thymus (Merck) was labeled with AlexaFluor 568 succinimidyl ester (Invitrogen) according to the manufacturer’s instructions, purified, concentrated by Vivaspin columns (10,000 MW, Sartorius) and was injected to 0.1% of the oocyte volume. DiIC16(3) (Invitrogen) was dissolved in vegetable oil to saturation and injected into oocytes as in (Lénárt et al., 2003). The methanol stock of phalloidin-AlexaFluor 568 (Invitrogen) was dried and dissolved in PBS (pH 7.2) and amounts corresponding to 10−5 units were injected directly into the nuclear region of oocytes ~ 1 min after NEBD. Utrophin CH domain (UtrCH) (Burkel et al., 2007) was subcloned into pET24d(+) for expression in E. coli. The recombinant protein was expressed for 1 hr at 37°C and purified on a Ni-NTA resin (QIAGEN). The purified protein was labeled by AlexaFluor 488 and AlexaFluor 568 succinimidyl ester (Invitrogen) according to the manufacturer’s instruction, purified by gel filtration (PD-10, GE Healthcare) and concentrated by Vivaspin columns (10,000 MW, Sartorius) before injection to oocytes. Fluorescently labeled recombinant UtrCH was injected 5–10 min prior to NEBD.
For drug treatments, oocytes were transferred into plastic dishes (Ibidi #80131). To inhibit myosin II ATPase activity, oocytes were incubated with a final concentration of 300 µM Blebbistatin (Abcam) for at least 3 hr before starting maturation; with ML-7 (Tocris) and Y-27632 (Enzo Life Sciences) at a final concentration of 100 µM for 1 hr. To induce acute F-actin depolymerization, oocytes were first matured and then treated with Latrunculin A (Abcam) after NEBD, as NEBD is known to depend on actin polymerization in starfish oocytes (Mori et al., 2014). To inhibit formin FH2 activity, we incubated oocytes with SMIFH2 (Tocris) at a final concentration of 50 µM for at least 2 hr.
Microscopy was done on a Leica SP5 or SP8 confocal microscope equipped with a fast Z-focusing device (SuperZ galvo stage) and using a 40x HCX PL AP 1.10 NA water immersion objective lens (Leica Microsystems). To record chromosome transport, starfish oocytes were imaged in 3D (a Z-step of 2 µm over 70 µm) over time (time step of 5 s) using a square frame of 256 × 256 pixels at a pixel size of 447 nm. To monitor F-actin network contraction, starfish oocytes were imaged in 3D centered on the median plane (plane of biggest surface) with a Z-step of 2 µm covering 10–20 µm over time (time step 10 s) using a square frame of 512 × 512 pixels at a pixel size of 223 nm. All imaging was performed at room temperature (20–22°C).
A pulsed two-photon laser (Chameleon, Coherent) was interfaced to a LSM780 confocal microscope (Zeiss) through a customized optical path aligned with the optical axis of the microscope. We sequentially imaged (a Z-step of 2 µm over 20 µm every 10 s) and performed 3D ablation (typical duration ~ 1 s) by imaging a stack of 1 × 50×60 µm, with a Z-step of 2 µm at 800 nm (30% laser power). Imaging was started ~ 1 min before the 3D ablation.
Image processing was done using Fiji (Schindelin et al., 2012). Figures were assembled in Adobe Illustrator. Data analyses were performed using Matlab (MathWorks). Image data are shown after brightness and contrast adjustment and application of a 2D Gaussian blur filter (sigma value: 0.2–1.0). Some image data were registered using the ‘StackReg’ plugin implemented in Fiji to correct for oocyte movements during experiments. Panels show either single Z-slices, maximum or sum intensity projections as indicated in the figure legends. Colored time projections were done using the TimeRGBcolorcode macro in Fiji. Chromosome tracking and data analysis were done by in-house routines written in Matlab. Segmentation and tracking algorithms are described elsewhere (Monnier et al., 2012). Pair-wise relative chromosome velocities were defined as the slope of the best-fit line (R2 > 0.9) to the separation distance between the pair versus time. Similarly, pair-wise chromosome approach rate was defined as the slope of the linear fit to the separation distance between the pair for every 2 min interval during the actin-driven transport phase (from 2 to 8 min after NEBD).
To control for photobleaching, we measured fluorescence intensity in a small region in the cytoplasm during the time course of the experiment. Oocytes in which the fluorescence intensity decrease exceeded 10%, were not included in further analysis.
To measure the mass of F-actin, a custom-written Fiji macro was first used to automatically segment the nuclear region based on either the DiIC16(3) fluorescence or the bright field channel. This pipeline included a FeatureJ Hessian filtering, followed by a 2D Gaussian filtering and automatic thresholding using the ‘Otsu’ or ‘Huang’ algorithm. The segmented area was then used on the F-actin channel to measure the density (‘mean’) (Figure 4A and Figure 4—figure supplement 1). To estimate the mass of F-actin within the contracting network over time, we used the initial outline of the nucleus at the onset of the actin-driven phase, and applied iteratively the function ‘Enlarge’ with a factor equals to to extrapolate the area and further the volume of the contracting network. Thus, the mass over time was measured as the product of the density and the extrapolated volume at each time point. For Latrunculin A-treated oocytes, we proceeded in two steps: before the addition of Latrunculin A, we extrapolated the size of the old network using derived from chromosome tracking. After drug addition, we measured the size of contracting old network, and derived the contraction rate similar to Figure 2E.
Surface curvature measurements were performed using an algorithm written in Matlab and described elsewhere (Bischof et al., 2017). The strength of surface contraction waves was defined by the maximum variance of the radius of curvature. The Matlab script is available in the Supplementary file 6. Statistical tests used were unpaired nonparametric tests (Mann-Whitney and Kruskal-Wallis) implemented in Matlab.
The contractile networks were simulated with Cytosim (Nedelec and Foethke, 2007) using a Langevin dynamics approach. Actin filaments are represented by incompressible bendable filaments of rigidity 0.075 pN.μm2 (persistence length of 18 µm) in a medium of viscosity 0.3 Pa.s (Belmonte et al., 2017; Gittes et al., 1993). Filaments have an initial length of 1.5 µm and are discretized into segments of 100 nm. Disassembly occurs from one end at a speed of 0.0015 µm/s. The other end is static. Cross-linkers are modeled as Hookean springs of 10 nm resting length and rigidity of 250 pN/µm with a binding rate kon = 10 s−1 (binding range of 0.01 μm) and an unbinding rate koff = 0.1 s−1 that is independent of the force exerted by the link. Motors are also Hookean springs of similar characteristics as cross-linkers, but motors additionally move on filaments with a speed of 0.02 µm/s, with a linear force-velocity relationship characterized by a maximal stall force of 6 pN. Chromosomes are modeled as point-like particles to trace the network evolution, but without a drag they do not affect its behavior. Chromosomes can bind a single filament with a binding rate kon = 10 s−1 (range of 0.1 μm) and a force-independent unbinding rate koff = 0.0001 s−1, and are also able to track the depolymerizing end of filaments. Actin disassembly, cross-linker and motor binding and unbinding events were modeled as first-order stochastic processes (Gillespie, 1977).
To limit the computational cost and focus on the effect of the mode of contraction, we limited the simulation to a circular two-dimensional system of constant radius 10 µm and we neglected the steric interaction between filaments. The disc was populated initially by 5000 filaments of 1.5 µm long, 20,000 cross-linkers and 20,000 motors for the motor-driven mechanism. For the disassembly-driven mechanism, we added 40,000 end-tracking cross-linkers to the simulated system. All filaments disassembled at the same rate from one end while the other end remained stable. The disassembly rate was set at 0.0015 μm/s to match the experimentally measured contractile characteristics. To simulate the polymerization of newly-assembled filaments at the edge of the nuclear region, filaments were added with an initial length of 1.5 µm at the boundary at a rate of 12 filaments per second. Because there is no edge or steric interactions, adding new filaments per se does not produce compressive force on the old network. Cytosim is an Open Source project hosted on https://github.com/nedelec/cytosim. The configuration file is provided as Supplementary file 1.
The Supplementary Material includes 7 figure supplements, 4 Supplementary Movies, 6 Supplementary Files including the detailed description of the viscoelastic model (Supplementary file 1), 3 tables (Supplementary file 2–4) and the configuration file for the Cytosim model (Supplementary file 5) to be run in Cytosim available at https://github.com/nedelec/cytosim (Nédélec, 2018; copy archived at https://github.com/elifesciences-publications/cytosim). The Matlab scripts used to analyze chromosome tracks is also available (Supplementary file 6).
Active diffusion positions the nucleus in mouse oocytesNature Cell Biology 17:470–479.https://doi.org/10.1038/ncb3131
Actin-based spindle positioning: new insights from female gametesJournal of Cell Science 127:477–483.https://doi.org/10.1242/jcs.142711
Reconstituting cytoskeletal contraction events with biomimetic actin-myosin active gelsMethods in cell biology 128:83–103.https://doi.org/10.1016/bs.mcb.2015.02.001
A theory that predicts behaviors of disordered cytoskeletal networksMolecular Systems Biology 13:941.https://doi.org/10.15252/msb.20177796
A quantitative analysis of contractility in active cytoskeletal protein networksBiophysical Journal 94:3126–3136.https://doi.org/10.1529/biophysj.107.117960
A cdk1 gradient guides surface contraction waves in oocytesNature Communications 8:849.https://doi.org/10.1038/s41467-017-00979-6
Actin dynamics, architecture, and mechanics in cell motilityPhysiological Reviews 94:235–263.https://doi.org/10.1152/physrev.00018.2013
Versatile fluorescent probes for actin filaments based on the actin-binding domain of utrophinCell Motility and the Cytoskeleton 64:822–832.https://doi.org/10.1002/cm.20226
A soft cortex is essential for asymmetric spindle positioning in mouse oocytesNature Cell Biology 15:958–966.https://doi.org/10.1038/ncb2799
Molecular architecture of the kinetochore-microtubule interfaceNature Reviews Molecular Cell Biology 9:33–46.https://doi.org/10.1038/nrm2310
Mechanism of cell rear retraction in migrating cellsCurrent Opinion in Cell Biology 25:591–599.https://doi.org/10.1016/j.ceb.2013.05.001
The filamentous actin cross-linking/bundling activity of mammalian forminsJournal of Molecular Biology 384:324–334.https://doi.org/10.1016/j.jmb.2008.09.043
Bulk cytoplasmic actin and its functions in meiosis and mitosisCurrent Biology 21:R825–R830.https://doi.org/10.1016/j.cub.2011.07.043
Exact stochastic simulation of coupled chemical reactionsThe Journal of Physical Chemistry 81:2340–2361.https://doi.org/10.1021/j100540a008
Flexural rigidity of microtubules and actin filaments measured from thermal fluctuations in shapeThe Journal of Cell Biology 120:923–934.https://doi.org/10.1083/jcb.120.4.923
Cytokinesis in animal cellsAnnual Review of Cell and Developmental Biology 28:29–58.https://doi.org/10.1146/annurev-cellbio-101011-155718
The evolution of the actin binding NET superfamilyFrontiers in Plant Science 5:254.https://doi.org/10.3389/fpls.2014.00254
Vesicles modulate an actin network for asymmetric spindle positioningNature Cell Biology 15:937–947.https://doi.org/10.1038/ncb2802
Formin mDia1 senses and generates mechanical forces on actin filamentsNature Communications 4:1883.https://doi.org/10.1038/ncomms2888
Steering cell migration: lamellipodium dynamics and the regulation of directional persistenceNature Reviews Molecular Cell Biology 15:577–590.https://doi.org/10.1038/nrm3861
Contractility and retrograde flow in lamellipodium motionPhysical Biology 3:130–137.https://doi.org/10.1088/1478-3975/3/2/005
Requirements for contractility in disordered cytoskeletal bundlesNew Journal of Physics 14:033037.https://doi.org/10.1088/1367-2630/14/3/033037
Actin-myosin spatial patterns from a simplified isotropic viscoelastic modelBiophysical Journal 107:863–870.https://doi.org/10.1016/j.bpj.2014.06.041
Unequal cell division regulated by the contents of germinal vesiclesDevelopmental Biology 273:76–86.https://doi.org/10.1016/j.ydbio.2004.04.038
Bayesian approach to MSD-based analysis of particle motion in live cellsBiophysical Journal 103:616–626.https://doi.org/10.1016/j.bpj.2012.06.029
Actomyosin sliding is attenuated in contractile biomimetic corticesMolecular Biology of the Cell 25:1845–1853.https://doi.org/10.1091/mbc.E13-08-0450
Forcing cells into shape: the mechanics of actomyosin contractilityNature Reviews Molecular Cell Biology 16:486–498.https://doi.org/10.1038/nrm4012
Collective Langevin dynamics of flexible cytoskeletal fibersNew Journal of Physics 9:427.https://doi.org/10.1088/1367-2630/9/11/427
Cytosim, version 6686989GitHub.
Fiji: an open-source platform for biological-image analysisNature Methods 9:676–682.https://doi.org/10.1038/nmeth.2019
A new model for asymmetric spindle positioning in mouse oocytesCurrent Biology 18:1986–1992.https://doi.org/10.1016/j.cub.2008.11.022
An actin-dependent mechanism for long-range vesicle transportNature Cell Biology 13:1431–1436.https://doi.org/10.1038/ncb2353
Latrunculins--novel marine macrolides that disrupt microfilament organization and affect cell growth: I. Comparison with cytochalasin DCell Motility and the Cytoskeleton 13:127–144.https://doi.org/10.1002/cm.970130302
Structural and functional insights on the Myosin superfamilyBioinformatics and Biology Insights 6:BBI.S8451–21.https://doi.org/10.4137/BBI.S8451
Pekka LappalainenReviewing Editor; University of Helsinki, Finland
In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.
Thank you for submitting your article "Disassembly-driven contraction of an F-actin network transports chromosomes in starfish oocytes" for consideration by eLife. Your article has been reviewed by three peer reviewers, one of whom, Pekka Lappalainen (Reviewer #1), is a member of our Board of Reviewing Editors, and the evaluation has been overseen by Andrea Musacchio as the Senior Editor.
The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.
Contractile actomyosin structures are critical for many morphogenetic events and transport processes in cells. Recent studies demonstrated that a three-dimensional contractile actin network collects chromosomes and transports them to the forming microtubule spindle in starfish oocytes. However, the mechanism by which this network contracts has remained elusive. Here, Bun et al., provide evidence that myosin activity is not necessary for the network contraction, but that actin filament disassembly instead appears to drive the network contractility.
All reviewers found this study very interesting and stated that the experimental part is novel and convincing. However, because the experiments are not sufficiently well connected to the model (and the model was not rigorously tested experimentally), we cannot publish the manuscript, at least in its present form. However, if you can significantly strengthen the manuscript along the lines suggested by the reviewers, we will be glad to consider an adequately revised version of the manuscript for publication in eLife.
1) The first part of the study presents some high quality experimental work with the tools available in starfish, and the second part of the study presents an interesting model. However, the main weakness of this manuscript is that the model is not really constrained by the experimental work, and that the prediction of the model was not challenged by some additional experimental work (since the putative cross-linker was not identified). Thus, the study would be much stronger if the authors could experimentally test their model. At minimum, alternative models to explain the data, and the limitations of the study, should be extensively discussed.
2) A central argument for favoring a disassembly driven mechanism over a motor driven mechanism is the comparison of the fits in Figure 4F and 6F to a viscoelastic model assuming either motor-driving or disassembly. Since this is so essential, the authors should give an expression for the contraction rate α(t) for the two models in the main text and make clear which measured quantities go into the fit and what other parameters are fitted. Importantly they should also give the value of all the fitting parameters. The authors should compare these parameters wherever possible to measured or measurable values (e.g. for filament length and depolymerization rate). In the current version, it is unclear if and how much these parameters vary from fit to fit and if they are within reasonable limits.
3) Along this line: the Motor Model depends on the actin concentration. This concentration does not change a lot for various UtrCH concentration (Figure 6E). Yet, the fit of the 'low' to 'high' UtrCH concentrations (dashed red, green, blue lines, Figure 6F) is drastically different from the control (black dashed line). Where does this sudden change come from if the input is so similar?
4) It appears that the only difference between the two viscoelastic models (D versus M) is the σ_c term. For the Motor-driven model there is a linear or quadratic dependency of A (actin density), for the disassembly driven model there is a linear dependency of A. Furthermore σ_c^D depends on k_depol and <l>. At first sight it appears surprising how this could lead to the large differences in the fits in Figure 6F, but A (and perhaps also <l>?) are time dependent. It would be helpful to explain in more detail which variables are time dependent, and also to show <l> (t) in real units (i.e. micrometers). Is k_depol also a function of time?
5) Comparing the two models: Model M and D both get the actin concentration A as input (linearly or quadratically). However, in Figure 6F the two fits go into different directions. While in M the fit increases faster and gets a peak, the fit for model D becomes smaller in value and flatter. This suggest that the bigger effect on the fit of model D is given by the <l> and k_depol parameters. Again, it is important to give the value and potentially even plot these two parameters for the different cases. This would be also informative in the light of the claim that these quantities are gradually changed with the UtrCH injection.
[Editors' note: further revisions were requested prior to acceptance, as described below.]
Thank you for resubmitting your work entitled "A disassembly-driven mechanism explains F-actin-mediated chromosome transport in starfish oocytes" for further consideration at eLife. Your revised article has been favorably evaluated by Andrea Musacchio (Senior editor) and three reviewers, one of whom is a member of our Board of Reviewing Editors.
The manuscript has been improved but there are some remaining issues that need to be addressed before acceptance, as outlined below:
This is a revised version of a manuscript, which provides evidence that actin filament disassembly drives the contraction of a three-dimensional actin network that collects chromosomes and transports them to the forming microtubule spindle in starfish oocytes. The authors have satisfactorily addressed the majority of the previous concerns, and the manuscript is significantly improved. Also, the experimental and simulation details are much better clarified now.
However, the formin inhibitor data were not considered particularly strong, and therefore we feel that you should tone down the conclusions concerning the role of formin(s) as a cross-linker in this process (please, see specific comments below). All proposed changes can be addressed textually and do not require new experiments.
1) The authors used SMIFH2 for inhibiting formins. To justify the use of SMIFH2 inhibitor, they referred to (Ucar et al., 2013). However, characterization of the effects of the inhibitor in this study were not particularly thorough, and this should be considered in the Discussion.
2) Even if the inhibitor works against formins, the effect of the inhibitor on the observed actin dynamics could be indirect e.g. due to altering the balance between Arp2/3 complex and formin -driven actin assembly pathways (see e.g. Burke et al., 2014). This possibility should be considered in the interpretation of the results.
3) In subsection “In silico reconstitution of the contractile system predicts a ‘depolymerization harnessing factor’ essential to drive contraction”, the authors test if the combination of disassembly and crosslinking could drive network contraction. This has already been done on a dynamic actin network (see: Ennomani et al., 2016), and should be discussed in the manuscript.
4) In the same subsection, the authors mention that mDia1 is able to track depolymerizing filament ends. However, these in vitro experiments (Jegou et al., 2013) were performed under specific conditions, where the critical concentration of actin was below the barbed end Cc. Thus, this argument cannot be used here.https://doi.org/10.7554/eLife.31469.034
- Serge Dmitrieff
- François J Nédélec
- Péter Lénárt
- Philippe Bun
- Julio M Belmonte
- Serge Dmitrieff
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
We would like to acknowledge members of the Lenart lab for reagents and support, Johanna Bischof in particular for sharing plasmid constructs. We would like to additionally acknowledge the essential support of EMBL’s Advanced Light Microscopy Facility, Protein Expression and Purification Facility, IT support, and Laboratory Animal Resources, Kresimir Crnokic in particular. The work was supported by the European Molecular Biology Laboratory, and specifically by the EIPOD Interdisciplinary Postdoctoral Programme to PB and JB, co-funded by the Marie-Curie Actions of the European Commission. SD was in part supported by the Center for Modeling in Biosciences (BIOMS).
- Pekka Lappalainen, Reviewing Editor, University of Helsinki, Finland
© 2018, Bun et al.
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.