1. Cell Biology
Download icon

Curvature-induced expulsion of actomyosin bundles during cytokinetic ring contraction

  1. Junqi Huang Is a corresponding author
  2. Ting Gang Chew Is a corresponding author
  3. Ying Gu
  4. Saravanan Palani
  5. Anton Kamnev
  6. Douglas S Martin
  7. Nicholas J Carter
  8. Robert Anthony Cross
  9. Snezhana Oliferenko
  10. Mohan K Balasubramanian Is a corresponding author
  1. University of Warwick, United Kingdom
  2. King's College London, United Kingdom
  3. Lawrence University, United States
Short Report
Cited
3
Views
1,836
Comments
0
Cite as: eLife 2016;5:e21383 doi: 10.7554/eLife.21383

Abstract

Many eukaryotes assemble a ring-shaped actomyosin network that contracts to drive cytokinesis. Unlike actomyosin in sarcomeres, which cycles through contraction and relaxation, the cytokinetic ring disassembles during contraction through an unknown mechanism. Here we find in Schizosaccharomyces japonicus and Schizosaccharomyces pombe that, during actomyosin ring contraction, actin filaments associated with actomyosin rings are expelled as micron-scale bundles containing multiple actomyosin ring proteins. Using functional isolated actomyosin rings we show that expulsion of actin bundles does not require continuous presence of cytoplasm. Strikingly, mechanical compression of actomyosin rings results in expulsion of bundles predominantly at regions of high curvature. Our work unprecedentedly reveals that the increased curvature of the ring itself promotes its disassembly. It is likely that such a curvature-induced mechanism may operate in disassembly of other contractile networks.

https://doi.org/10.7554/eLife.21383.001

Introduction

Many eukaryotes assemble a dynamic actomyosin-based contractile ring to divide one cell into two (Pollard and Wu, 2010). During ring contraction, ring components continuously disassemble while maintaining ring contractility (Pollard and Wu, 2010). This is in stark contrast to other actomyosin machineries, particularly the thick and thin filament based sarcomeres in muscle fibers, which undergo contraction and relaxation cycles without disassembly (Murrell et al., 2015). ADF1/Cofilin together with Coronin and WD-repeat protein Aip1 modulate actin turnover by severing actin bundles in many types of actin networks (Jansen et al., 2015). However, their involvement in actomyosin ring contraction and disassembly is still unclear (Chen and Pollard, 2011). The F-actin based molecular motor Myosin II may also regulate disassembly of actin networks during cytokinesis (Guha et al., 2005; Murthy and Wadsworth, 2005), presumably by myosin II mediated actin filaments buckling and breakage (Murrell and Gardel, 2012; Reymann et al., 2012; Vogel et al., 2013). The exact mechanism governing disassembly of actomyosin rings and whether and how the ring disassembly is coordinated to ring contraction have remained unknown so far.

To address these gaps in our knowledge, we studied the dynamics of actomyosin ring contraction and disassembly in the fission yeasts Schizosaccharomyces japonicus and Schizosaccharomyces pombe. We report that micron-scale actin bundles are expelled during contraction of the actomyosin ring in a ring curvature-dependent manner in intact cells, in spheroplasts, and when isolated actomyosin rings are triggered to contract with ATP. We test the curvature-dependence of the actomyosin ring disassembly by physically deforming protoplasts to defined geometries.

Results and discussion

Whilst studying F-actin dynamics in S. japonicus cells expressing LifeAct-GFP, as a marker for all actin structures (Gu et al., 2015; Huang et al., 2012Riedl et al., 2008), large F-actin bundles appeared to be connected with the contracting ring (Figure 1A). We imaged actin rings in the axial plane to achieve higher resolution by preparing spheroplasts, in which actomyosin rings were oriented randomly and slid along the cell membrane during contraction, consistent with previous work in S. pombe and in reconstituted giant unilamellar vesicles (GUVs) (Mishra et al., 2012; Miyazaki et al., 2015). In spheroplasts where the actomyosin ring was positioned close to the imaging plane, actin filaments appeared to project outwards from the contracting ring (Figure 1B). Permeabilized spheroplasts retained substantial numbers of actin filaments that remained attached to the ring, indicating that the actin filaments were connected to the contracting ring (Figure 1C). To ensure that the observed actin filaments were not an artifact of LifeAct-GFP used in these experiments, we fixed and stained wild type cells (that did not express any fusion protein) and wild type cells expressing LifeAct-GFP with Rhodamine-conjugated phalloidin (Figure 1—figure supplement 1A). We also fixed and stained permeabilized spheroplasts generated from these two strains with Rhodamine-conjugated phalloidin. In both cases, we were able to detect actin filaments connected to the contracting actomyosin ring (Figure 1—figure supplement 1A), ruling out the possibility that the observed filaments were an artifact of the tag used. When fixed cells expressing Rlc1p-GFP were stained with Rhodamine-conjugated phalloidin, actin filaments attached to the ring colocalized with Rlc1p, suggesting that Rlc1p-GFP (and by extension Myosin II) was also part of the actin filaments attached to contracting rings (Figure 1—figure supplement 1A). Furthermore, actin filaments linked to the contracting actomyosin ring were detected in S. pombe cells and permeabilized spheroplasts of wild type cells (with no genetic tag) and of wild type cells expressing LifeAct-GFP or Rlc1-GFP (Figure 1—figure supplement 1B), establishing that this phenomenon is not unique to S. japonicus cells and spheroplasts. Actin and myosin II filaments attached to contracting actomyosin rings have been observed by Wollrab and colleagues in S. pombe cells, although the significance of these structures was not investigated in their work (Wollrab et al., 2016). In the rest of the manuscript, unless specifically indicated, experiments were performed in S. japonicus.

Figure 1 with 1 supplement see all
Large actin bundles are expelled during actomyosin ring contraction.

(A) S. japonicus cells expressing LifeAct-GFP. (B) Spheroplasts expressing LifeAct-GFP. (C) Permeabilized spheroplasts (by 0.5% NP-40) stained with CF633-phalloidin. (D) Quantification of the widths of actin rings and bundles by STORM. n = 16 random bundles and 21 random positions of 5 rings. The full-width-half-maximum (FWHM) value of a Gaussian-fitted line was measured to represent the widths. (E) Time-lapse micrographs of spheroplasts expressing LifeAct-GFP. Asterisks: expelling actin bundles. Arrows: rotations of an actin bundle. All scale bars are 5 µm unless specified. Error bars: s.d.

https://doi.org/10.7554/eLife.21383.002

Analysis of actomyosin rings with super resolution microscopy STORM showed it to be 77 ± 16 nm wide for ring diameters ranging 500 nm to 2.8 µm, whilst the ring associated actin filamentous structures were 52 ± 12 nm wide (Figure 1D and Figure 1—figure supplement 1C; enlarged panel), indicating that these structures contained multiple 7 nm-wide actin filaments arranged in a bundle (Takaine and Mabuchi, 2007). On this basis, we now refer to them as ring-associated actin bundles.

Time-lapse microscopic analysis of spheroplasts expressing LifeAct-GFP showed that actin bundles appeared to emerge from the contracting ring, continued to extend in length and were progressively expelled from the sites of their initial appearances (Figure 1E; Video 1). Occasionally, upon the expulsion of actin bundles, the contracting ring seemed to disintegrate at the region of release (Figure 1—figure supplement 1D). Similar observations of actin bundles being expelled from contracting rings were also made in S. pombe (Figure 1—figure supplement 1E). Actin bundles were also expelled during contraction in S. japonicus intact cells expressing LifeAct-GFP in which the GFP intensity was reduced and actin appeared disorganized near the end of cytokinesis (Figure 1—figure supplement 1F and G). The actin intensity measured in permeabilized spheroplasts using CF633-phalloidin staining in large rings was higher than in small rings (Figure 1—figure supplement 1H; Note that the fluorescence intensity from ring-associated bundles were not considered in this analysis). However, the combined actin fluorescence intensity from rings and associated bundles was roughly comparable in large and small rings (Figure 1—figure supplement 1I), suggesting that the expulsion of actin bundles from the ring may represent a mechanism for ring disassembly. To estimate the fraction of ring associated actin filaments that were expelled as bundles during contraction, we measured the fluorescence intensity of the bundle fraction in contracting rings using LifeAct-GFP as a marker (Figure 1—figure supplement 1J). We estimated that ~68% of actin filaments that are lost from the ring during contraction (from diameters of 3.48 ± 1.44 µm to 1.23 ± 0.65 µm) are expelled as bundles.

Video 1
Expulsion of F-actin bundles during ring contraction in a spheroplast expressing LifeAct-GFP.

Asterisks mark the elongation of an actin bundle. Left panel is presented in 'Fire' lookup table and its calibration bar. Spheroplasts are outlined by dotted lines. Time zero indicates the start of the video.

https://doi.org/10.7554/eLife.21383.004

Other components of actomyosin rings such as the regulatory light chain of myosin Rlc1p, tropomyosin Cdc8p (co-staining of these shown in Figure 2A), IQGAP protein Rng2p, and F-BAR protein Cdc15p (Figure 2A) were present in actin bundles associated with contracting rings in spheroplasts. We investigated the dynamics of ring contraction using each of these proteins as a marker and found that GFP-Rng2p, Cdc15p-GFP, and Rlc1p-GFP were associated with the expelling/expelled bundles (Figure 2B, Video 2). In time-lapse imaging experiments, Rlc1p-GFP was also found in bundles associated with contracting rings in S. pombe (Figure 2—figure supplement 1A). When the contracting ring was analyzed in spheroplasts of S. japonicus, actomyosin bundles became detectable when the ring contracted to a diameter of ~2 µm and were expelled from the ring at the rate of 2.19 ± 0.61 µm/min, (Figure 2C,D; Figure 2—figure supplement 1B).

Figure 2 with 1 supplement see all
Quantitative analysis of actomyosin bundle expulsion.

(A) Top panel: localizations of tropomyosin Cdc8p and Rlc1p-GFP in permeabilized spheroplasts. Both proteins were detected by immunofluorescence microscopy. Bottom panel: spheroplasts expressing GFP-Rng2p, and Cdc15p-GFP were imaged to show the localization of these proteins on ring-associated bundles. (B) Time-lapse micrographs of spheroplasts expressing GFP-Rng2p, Cdc15p-GFP, and Rlc1p-GFP, respectively. The time is indicated at bottom right corner in minutes. (C) Quantification of the Rlc1p-GFP ring diameter during initial bundle expulsion (n = 11 rings) in spheroplasts. The ring diameters were calculated by fitting the measured perimeters in Figure 2—figure supplement 1B with Diameter = Perimeter / π. (D) Quantification of the Rlc1p-GFP bundle-released rate (n = 15 bundles) in spheroplasts. (E) Quantitative analysis of bundle expulsion in permeabilized spheroplasts. i: quantification of the number of bundles as a function of ring perimeter; ii: total bundle lengths as a function of ring perimeter; iii: number of bundles associated with small or large rings; iv: average bundle length associated with small or large rings; v: sum of total bundle length (T.l.) and ring perimeter (Peri.) as a function of ring perimeter. Large: ring perimeter ≥ 11 µm; Small: ring perimeter < 11 µm; n = 75 (Large), 46 (Small), respectively. Graphs ii and v: each data point represents one permeabilized spheroplast, n = 121. (F) Kymographs of contracting rings at 1 min interval in intact cells expressing Rlc1p-GFP. Asterisks: actomyosin bundles. (G) Quantifications of the ring diameter during initial bundle expulsion (n = 28 rings), time of bundle expulsion (n = 12 rings), traveling rate of expelled bundles (n = 15 bundles), and total ring contraction time (n = 12 rings) in intact cells. Scale bar: 5 µm. Error bars: s.d.

https://doi.org/10.7554/eLife.21383.005
Video 2
Expelled filaments/bundles contain GFP-Rng2p, Cdc15p-GFP, and Rlc1p-GFP.

Asterisks indicate expelling bundles. Time zero indicates the start of the video.

https://doi.org/10.7554/eLife.21383.007

Profiling the distribution of ring-associated bundles in spheroplasts indicated that there was an inverse correlation (Figure 2Ei) between the number of actin bundles (measured using tropomyosin Cdc8p as a marker for actin bundles) or bundle lengths with the perimeter of the actomyosin ring (Figure 2Eii). It appeared that there were on average 3.5 ± 1.8 actin bundles attached to rings of less than 11 µm perimeter (corresponding to about 3.5 µm ring diameter), whereas there were 0.7 ± 1.2 actin bundles associated to rings of larger perimeter (Figure 2Eiii). When the length of individual ring-associated actin bundles was measured, we found that on average the bundle was longer in small rings (4.16 ± 3.30 µm for perimeter of <11 µm) than in larger rings (1.42 ± 1.86 µm for perimeter of >11 µm) (Figure 2Eiv). The total length of ring perimeters and actin bundles associated with the ring was higher in smaller rings compared to larger rings (Figure 2Ev). These analyses indicated that ring disassembly and bundle expulsion was coupled to ring contraction.

In mitotic cells expressing Rlc1p-GFP (Figure 2F, Figure 2—figure supplement 1C) or GFP-Rng2p (Figure 2—figure supplement 1D), actomyosin bundles were also not detected in rings with a larger diameter. The bundles associated with the actomyosin ring became more prominent when the actomyosin ring contracted to a diameter of 2.26 ± 0.52 µm (Figure 2G; Figure 2—figure supplement 1E; Video 3; initial ring diameter was 5.49 ± 0.37 µm). On average, the ring-associated bundles became detectable after 12.3 ± 3.2 min of initiation of actomyosin ring contraction, and continued to be expelled from the actomyosin ring until the end of contraction at a rate of 1.91 ± 0.55 µm/min (Figure 2G; full contraction time was 18.21 ± 3.18 min). Hence, it appeared that the expulsion of actomyosin bundles became obvious when rings contracted to ~40% of the initial diameter.

Video 3
Ring contraction in intact S. japonicus cells expressing Rlc1p-GFP.

Asterisks indicate actomyosin bundles are expelled from the contracting rings. Time zero indicates the start of the video.

https://doi.org/10.7554/eLife.21383.008

We then asked if cytoplasmic factors were required for the expulsion of actin bundles during ring contraction (Figure 3A). To investigate this, we established a S. japonicus cell ghost system (depleted of cytosol) which contained contractile actomyosin rings, as previously described in S. pombe and S. cerevisiae (Mishra et al., 2013Young et al., 2010). Treating S. japonicus cell ghosts with ATP induced contraction of isolated actomyosin rings (Figure 3—figure supplement 1A). Interestingly, the actin bundles remained associated with the contracting and fully contracted rings (Figure 3B, Video 4) although the F-actin signal as visualized with LifeAct-GFP was lesser than in cells and spheroplasts. To quantitate aspects of actin bundle expulsion in cell ghosts, we fixed and stained contracting and contracted actomyosin rings with CF633-phalloidin (Figure 3C). The ring-associated bundles appearing in cell ghosts after ATP addition were similar to those observed during in vivo ring contraction (Figure 2E). We observed 4.3 ± 1.4 bundles associated with the rings in cell ghosts, with individual average length of 3.9 ± 1.3 µm (Figure 3D; total bundle length 16.1 ± 6.7 µm). The sum F-actin intensity of cell ghosts (ring actin + bundle actin) following ATP addition showed no significant changes, whereas there was a decrease of actin intensity in the rings (Figure 3E). We estimated protein levels of actin and the actin filament binding protein tropomyosin Cdc8p in the supernatant and pellet fractions following centrifugation of contracting actomyosin rings in cell ghosts (treated with ATP for 20 or 40 min). We did not find a significant increase in the amount of actin and Cdc8p in the supernatant in cell ghosts treated with ATP for 20 or 40 min (Figure 3F), consistent with the idea that ring contraction resulted in denser actin filaments containing binding proteins such as tropomyosin that remain in the pellet fraction. These results indicated that in cell ghosts treated with ATP for 20 to 40 min, the rings reduced in actin amounts. However, this did not lead to an increase in soluble G-actin during ring contraction, suggesting that most actin was expelled in filaments contained in bundles during this period of time. Furthermore, these experiments established that the expulsion of actin bundles was a result of ring contraction and was independent of continuous presence of cytosol. Finally, the absence of cytosolic components such as G-actin also suggests that the actomyosin bundles are expelled as a result of contraction rather than the possibility that these bundles were newly polymerized actin cables that were not incorporated into the highly curved actomyosin ring (Figure 3A).

Figure 3 with 1 supplement see all
Expulsion of actomyosin bundles during contraction of isolated rings.

(A) i, bundle expulsion mediated by cytosolic proteins; ii, newly polymerized actin filaments that are unable to incorporate into a highly curved actomyosin ring. (B) Time-lapse micrograph shows contraction of an actomyosin ring in a cell ghost after addition of LifeAct-GFP (concentration: ~4 ng/µl) and 0.5 mM ATP. Arrows: expulsion of actomyosin bundles. (C) Cell ghosts containing Rlc1p-GFP were treated with 0.5 mM ATP for 20 min or 40 min respectively, and stained with CF633-phalloidin to visualize F-actin structures. (D) Quantification of the number of expelled bundles, total bundle length, and average bundle length in each cell ghost after 40 min ATP treatment (n = 32 cell ghosts, 137 bundles). (E) Quantification of the sum actin intensity in cell ghosts (ring + bundles) or ring (ring) after treatment with ATP. To measure the total actin intensity in a cell ghost stained with CF633-phalloidin, a square to cover the actomyosin ring and the associated bundles as the region-of-interest was selected, and the sum intensity of the region-of-interest was measured. For the actin intensity of the ring in cell ghosts, a line along the ring circumference to cover the ring area was selected as the region-of-interest for sum intensity measurement. n = 23, 24 cell ghosts for 20 min, 40 min time point, respectively. ns: no significant difference. ***p<0.001. (F) Immunoblots of actin and Cdc8p from cell ghosts treated with ATP for 20 min and 40 min. S, Supernatant: soluble G-actin or short F-actin. P, Pellet: actomyosin rings and associated bundles. Four blots were used for quantifications. Scale bar: 5 µm. Error bars: s.d.

https://doi.org/10.7554/eLife.21383.009
Video 4
Expulsion of actin bundles in 0.5 mM ATP-treated isolated rings in cell ghosts.

Asterisks indicate expelled actomyosin bundles during contraction. Time zero indicates the start of the video.

https://doi.org/10.7554/eLife.21383.011

What is the mechanism of expulsion of actin bundles during ring contraction? We had previously noted that actomyosin bundles were expelled at later stages of contraction. As the actomyosin bundles were expelled from contracting rings of smaller diameters, which have higher curvatures, we speculated that the local curvature of rings might induce bundle expulsion. In spheroplasts expressing Rlc1p-GFP, the expelled bundles became visible and their numbers increased with the progression of time and increasing the curvature (Figure 4A). Although rings started with curvatures (1/radius) of even smaller than 0.4 µm−1 (translates to a ring diameter of >5 µm in spheroplasts), the mean curvature at which bundles began to be expelled was 1.1 ± 0.4 µm−1 (translates to a ring diameter of ~2 µm) (Figure 4A). Given that bundle expulsion was only observed after a significant reduction in ring diameter, suggesting that bundle expulsion in rings is promoted by increased curvature. Furthermore, even in partially contracted and curved actomyosin rings in ATP treated cell ghosts, the majority (~68%) of ring-associated bundles were detected at regions of increased local curvature where the ring appeared to be bent (Figure 4—figure supplement 1A).

Figure 4 with 1 supplement see all
Increased ring curvature promotes actomyosin bundle expulsion.

(A) Analysis of the curvature of sliding rings in spheroplasts expressing Rlc1p-GFP (n = 11 rings). The ring curvatures were color-coded, and mean curvatures of each stage of contraction are shown in the lower panel. The mean ring curvature during initial bundle expulsion was plotted in the right graph. Asterisk: an expelled bundle. (B) Distorted actomyosin rings after spheroplasts were compressed mechanically (n = 11 rings). d: flat; h: curved regions. The degree of compression was quantified by compression factors. (C) Micrographs of bundle expulsion from the curved regions of rings. The ring curvatures are color-coded. Asterisks: expelled bundles. (D) Quantification of the number of ring-associated bundles, and fraction of bundles expelled from curved regions (n = 94 bundles from 18 rings). (E) Quantification of the local curvatures at the initial sites of bundle expulsion and the mean ring curvatures during bundle expulsion (n = 25 bundles from 11 rings). (F) Quantification of the ring perimeter during initial bundle expulsion and the rate of perimeter change in contracting rings (n = 11 rings). (G) Quantification of the traveling rate of expelled bundles (n = 16 bundles) (H) Rates of contraction in flat (△d) and curved (△h) regions (n = 11 rings). Scale bar: 5 µm. Error bars: s.d.

https://doi.org/10.7554/eLife.21383.012

To directly test the hypothesis that the ring disassembly by bundle expulsion depends on ring curvature, we compressed the spheroplasts expressing Rlc1p-GFP to artificially produce rings with distinctive curved and flat regions (Figure 4B; compression factor ~0.54 ± 0.08). Strikingly, about 95% of the actomyosin bundles were expelled from the curved regions of compressed rings compared to flat regions (Figure 4C,D; Figure 4—figure supplement 1B and Video 5; average 5.2 ± 0.9 bundles per ring), and the local curvature at the sites of expelled bundles was 1.5 ± 0.3 µm−1 (Figure 4E; mean curvature of compressed rings was 0.7 ± 0.2 µm−1). Whereas actomyosin bundles were expelled when uncompressed rings contracted to a perimeter of 6.3 ± 1.5 µm (Figure 2—figure supplement 1B), majority of compressed rings started releasing actomyosin bundles when contracted to a perimeter of only 12.5 ± 2.4 µm, indicating that the curvature change induced expulsion of actomyosin bundles (Figure 4F). Curiously, following compression, the ring contraction rate (△perimeters/time) was 0.37 ± 0.09 µm/min, which was significantly less than the 1.26 ± 0.57 µm/min observed in uncompressed spheroplasts (Figure 4F and Figure 2—figure supplement 1B). The bundles were expelled at a reduced rate of 1.18 ± 0.42 µm/min as compared to 2.19 ± 0.61 µm/min in uncompressed spheroplasts (Figure 4G and Figure 2D). Furthermore, shortening of the flat regions (△d) of compressed rings occurred faster than the decrease of ring heights (△h) (Figure 4H; △d/min= 0.16 ± 0.04 µm/min vs. △h/min = 0.07 ± 0.02 µm/min). Thus, the actomyosin bundles were constrained by the local curvature of contracting rings, and their expulsion depends on the curvature changes during ring contraction. Our results showed that ring disassembly is coupled physically to ring contraction in a ring-geometry coordinated manner.

Video 5
Curvature-induced expulsion of actomyosin bundles in distorted rings.

Curvature of distorted rings is color-coded. Ring contraction is shown in three different angles of projections. Time zero indicates the start of the video.

https://doi.org/10.7554/eLife.21383.014

The expulsion of micron-scale actomyosin bundles may serve as a major mechanism to disassemble actomyosin rings during cytokinesis (Figure 4—figure supplement 1C). It is presently unclear if other mechanisms, such as cofilin and myosin II mediated filament severing / breakage and loss of actin monomers also contribute to actin disassembly during cytokinesis. New probes exhibiting reduced photobleaching and directly reporting actin cytoskeleton dynamics will be required to help estimate if other mechanisms are in operation as well. The coupling of bundle expulsion mediated-ring disassembly to ring contraction appears intrinsic to the ring itself, and is potentially driven by a physical mechanism that is independent of cytosolic factors. It is possible that the affinity of binding of unidentified actin binding proteins with F-actin is reduced upon increased curvature of actin filaments. Thus, actomyosin bundles may not be able to bend sufficiently to accommodate the high ring curvature at later times in contraction, leading to their expulsion from the ring. It is likely that attainment of a curved morphology is an energetically unfavorable conformation since actin filaments have a persistence length of ~10 microns (which can increase further with bundling) (Ott et al., 1993; Riveline et al., 1997).

Although we were able to visualize actin filaments, Rlc1p-GFP, GFP-Rng2p, and Cdc15p-GFP on expelled bundles in living cells and spheroplasts, we were only able to visualize expulsion of actin filaments in bundles in contracting actomyosin rings within cell ghosts (Figure 3—figure supplement 1A and data not shown for GFP-Rng2). It is possible that other ring proteins do not associate tightly to actin bundles during ring contraction in a less crowded environment (such as permeabilized cell ghosts), whereas they are able to associate with expelled bundles in a cellular environment of increased molecular crowding. Alternatively, it is possible that, in the presence of cytosol, as in a cellular context (Pelham and Chang, 2002; Wong et al., 2002), loss of ring proteins from actin bundles in the ring is balanced by replenishment from the cytosolic pool. However, in cell ghosts, such replenishment of ring proteins does not occur due to the absence of cytosol. Future work should investigate these possibilities.

Several types of actin filament behaviors such as branching and bundling are coupled to the curvature of the actin filament network (Liu et al., 2008; Risca et al., 2012). Expulsion of the actomyosin bundles that occurs when the cytokinetic actomyosin ring acquires high curvature suggests that curvature-induced disassembly could be a general mechanism for disassembly of actomyosin networks.

Materials and methods

Yeast strains, medium, and culture conditions

S. japonicus and S. pombe strains used in this study are listed in Table S1 (Supplementary file 1). S. japonicus strains were constructed as previously described (Gu et al., 2015). S. japonicus and S. pombe cells were cultured in YEA medium (5 g/l yeast extract, 30 g/l glucose, 225 mg/l adenine) at 24°C until mid-log phase for physiological analysis.

S. japonicus spheroplasts and cell ghosts preparation, and ATP-dependent contraction of isolated rings

Isolated actomyosin rings of S. japonicus in cell ghosts were prepared using similar published methods (Huang et al., 2016; Mishra et al., 2013). Briefly, S. japonicus cells were cultured in YEA medium overnight to mid-log phase, and 15 ml of cells were spun down and washed once with an equal volume of SCS buffer (20 mM sodium citrate [pH 5.8], 1 M sorbitol). Cells were then digested in 1 ml SCS buffer containing 0.1 g/ml lytic enzymes Lallzyme MMX (Lallemand) (Flor-Parra et al., 2014). The resulting spheroplasts were then washed and recovered in YEA medium containing 1 M sorbitol until the actomyosin rings were assembled. Spheroplasts were then washed once with wash buffer (20 mM PIPES–NaOH [pH 7.0], 0.8 M sorbitol, 2 mM EGTA, 5 mM MgCl2), and then permeabilized with isolation buffer (50 mM PIPES–NaOH [pH 7.0], 0.16 M sucrose, 50 mM EGTA, 5 mM MgCl2, 50 mM potassium acetate) containing 0.5% Nonidet P40 (NP-40) to obtain cell ghosts containing isolated rings. The suspension was washed twice with reactivation buffer (0.16 M sucrose, 5 mM MgCl2, 50 mM potassium acetate, 20 mM MOPS–NaOH [pH 7.0], pH adjusted to 7.5). To induce ATP-dependent actomyosin ring contraction, the cell ghosts were treated with reactivation buffer containing 0.5 mM ATP (Sigma; A6559).

Sample preparation for microscopy imaging

To image cells in suspensions, 1 ml of a yeast culture at mid-log growth phase was first spun down at 450 xg for 2 min, and ~16 µl of concentrated samples were loaded onto an Ibidi µ-Slide 8-Well glass bottom dish (Cat. No. 80827). To image spheroplasts in suspension, 1 ml of spheroplasts (regenerated in YEA medium containing 1 M sorbitol for 1.5 to 3 hr) was spun down at 450 xg for 2 min and ~16 µl of the concentrated suspension was loaded onto an Ibidi µ-Slide 8-Well glass bottom dish. To image contraction of rings in cell ghosts, equal volume of 1 mM ATP (final concentration 0.5 mM ATP) was added to cell ghosts and imaged in an Ibidi µ-Slide 8-Well glass bottom dish. All imaging dishes were sealed with an adhesive film membrane or mineral oil to prevent evaporation during imaging.

To generate actomyosin rings of distorted shapes in Figure 4B,C and Figure 4—figure supplement 1B, spheroplasts were prepared as described above, and were compressed by capillary forces pulling the coverslip and slide containing an agarose pad together. The slide was then sealed with VALAP prior to imaging. All samples were imaged at room temperature.

The CellASIC ONIX Microfluidic system was used to immobilize cells for imaging in Figure 1—figure supplement 1F and G, Figure 2F,G, and Figure 2—figure supplement 1C and E.

To estimate the fraction of actin filaments that were expelled as bundles from the contracting rings in Figure 1—figure supplement 1J, two Z-stacks (0.2 µm step size) of the same spheroplasts expressing LifeAct-GFP were acquired consecutively at a 12 min interval to minimize photobleaching. To quantitate the effect of photobleaching during imaging, the neighboring spheroplasts in the imaging field were selected and measured for their sum fluorescence intensity of LifeAct-GFP between two time points. The ratio of the sum fluorescence intensity of late versus early time points of selected spheroplasts was 1.018 ± 0.044, indicating that the photobleaching effect was minimal. Only spheroplasts without obvious actin bundles at the beginning of imaging (average ring diameter is 3.48 ± 1.44 µm) were chosen.

Immunofluorescence microscopy and F-actin staining

Spheroplasts were first permeabilized with isolation buffer containing 0.5% NP-40, and washed twice with reactivation buffer. Permeabilized spheroplasts were then fixed with 3.7% formaldehyde for 12 min at room temperature and washed twice with reactivation buffer. A rabbit primary antibody recognizing S. pombe tropomyosin Cdc8p (Balasubramanian et al., 1992) and a mouse CF633 anti-GFP (Sigma; SAB4600146) were added at 1:400 and 1:200, respectively and incubated overnight at 4°C. The samples were then washed twice with reactivation buffer, followed by treatment with the Donkey AlexaFluor 555 anti-Rabbit (Abcam; ab150074) at 1:400 for 2 hr at room temperature. After two washes with reactivation buffer, the samples were mounted on a slide and visualized using the spinning-disk confocal microscopy (Figure 2A and E).

To visualize actin structures in Figure 1C, Figure 1—figure supplement 1H and I, Figures 3C,D,E, and Figure 4—figure supplement 1A, spheroplasts were permeabilized and fixed as described above, and treated with CF633-phalloidin (Biotium; #00046; dissolved in water) at 1:20 for 10 min. Samples were washed twice with reactivation buffer before visualization.

In Figure 1—figure supplement 1A and B, cells and permeabilized spheroplasts were fixed with 3.7% formaldehyde for 12 min at room temperature. The fixed cells were permeabilized with PBS containing 1% Triton X-100 for 1 min, washed twice with PBS, and then treated with Rhodamine-conjugated phalloidin (Life Technologies; R415) at 1:20. Similarly, fixed and permeabilized spheroplasts were treated with Rhodamine-conjugated phalloidin at 1:20 to visualize actin structures.

Spinning-disk confocal microscopy

All images, except Figure 1D and Figure 1—figure supplement 1C, were captured using the Andor Revolution XD spinning disk confocal system, which was equipped with a Nikon ECLIPSE Ti inverted microscope, Nikon Plan Apo Lambda 100×/1.45NA oil immersion objective lens, a spinning-disk system (CSU-X1; Yokogawa), and an Andor iXon Ultra EMCCD camera. Images were acquired at the pixel size of 80 nm/pixel using the Andor IQ3 software. Three Laser lines at wavelengths of 488 nm, 561 nm, and 640 nm were used for excitation. All images were acquired with Z-step sizes of 0.2 µm, 0.3 µm or 0.5 µm, at varied interval times in individual time-lapse microscopy experiments.

STORM

Super resolution microscopy (Figure 1D and Figure 1—figure supplement 1C) was performed using a custom-built TIRF widefield microscope with enhanced stability (http://wosmic.org). The STORM microscope has been tested to have an axial resolution (x-y direction) of 22–28 nm, done as previously described (Nieuwenhuizen et al., 2013). For sample preparation, permeabilized spheroplasts were fixed with 3.7% formaldehyde, and washed twice with reactivation buffer. Recombinant GST-LifeAct-GFP protein (1.3 µg/µl) was then added at 1:100 into the fixed permeabilized spheroplasts and incubated overnight at 4°C. The sample was then washed twice with reactivation buffer, and further treated with the AlexaFluor 647 anti-GFP antibody (Life Technologies; A-21447) at 1:200 for 4 hr at room temperature, followed by two washes with reactivation buffer. The sample was resuspended in imaging buffer (80 mM PIPES [pH 6.7]; 10 mM MgCl2; 1 mM EGTA; 50% v/v glucose, 10% v/v b-mercapthoethanol). Prior to imaging with STORM, 0.1 µl of GOC mix (5 mg/ml glucose oxidase and 1 mg/ml catalase) was added to 9.9 µl of samples, and then sandwiched between two coverslips. The STORM images were reconstructed using GDSC SMLM plugins from the University of Sussex installed in Fiji (Schindelin et al., 2012). To measure the bundle or ring width from STORM images, a portion of a bundle or a ring was selected and straightened (Fiji/Edit/Selection/Straighten) using the segmented line function of Fiji. A vertical line was drawn every 200 nm along the straighten line. The intensity of pixels along the vertical line was measured (Fiji/Analyze/Plot Profile). The width was measured by taking the Full-Width-Half-Maximum (FWHM) value after fitting the vertical line profile to a Gaussian curve function (Fiji/Analyze/Tools/Curve Fitting). These processes were repeated using a macro written in Fiji.

Image analysis

Images were analyzed using Fiji (Schindelin et al., 2012) and Imaris (Bitplane). The image stacks were projected along the Z-axis (sum intensity or maximum intensity) for analysis and for representation. The sum intensity projections were performed in Figure 1—figure supplement 1G,H,I,J, Figure 2F, and Figure 3E. The maximum intensity projections were performed in Figure 1A,B,C,E, Figure 1—figure supplement 1A,B,D,E,F, Figure 2A,B, Figure 2—figure supplement 1A,C,D, Figure 3B,C, Figure 3—figure supplement 1A, Figure 4A,B,C, and Figure 4—figure supplement 1A,B.

The background of all microscopy images, except Figure 1D, Figure 1—figure supplement 1C, and Figures 4B,C, and Figure 4—figure supplement 1B, was subtracted in Fiji (Fiji/Process/Subtract Background).

All time-lapse microscopy images, except Figure 1—figure supplement 1G and Figure 4C, were corrected for photo-bleaching in Fiji (Fiji/Image/Adjust/Bleach Correction).

Imaris was used to facilitate three-dimensional measurements and representation. The ring perimeters (Figure 2EFigure 4F), the number of bundles (Figure 2Ei,Eiii, Figure 3D, Figure 4D and Figure 4—figure supplement 1A), the rate of bundle expulsion from the ring (Figure 2D,2G, Figure 4G), and the length of bundles (Figure 2Eii,Eiv,Ev, Figure 3D), were quantified after loading raw image stacks without prior processing.

To measure ring contraction rate in Figure 4F,H and Figure 2—figure supplement 1B, the ring perimeter was calculated as the total length of the ring skeleton using the Matlab function 'regionprops'. The procedure was then repeated for each time frame of the time-lapse series. The resulting data were fitted to a linear regression model using Matlab function 'fitlm'. The ring contraction rate was then measured as the slope of a fitted line.

All time-lapse videos were edited and saved in MP4 format with H.264 compression. Graphs were made with Prism 6 (GraphPad). The figures were arranged with Adobe Illustrator.

Curvature analysis

To obtain the local curvature of Rlc1p-GFP rings in Figure 4A,C,E, and Figure 4—figure supplement 1B, a similar approach was developed as previously described (Driscoll et al., 2012). Briefly, raw Z stacks containing the actomyosin ring were de-noised with non-local means filter using a plugin installed in Fiji (plugin source: http://imagej.net/Non_Local_Means_Denoise). To facilitate shape measurement, the 3D stack was converted to 2D maximum intensity projection along the normal to the ring plane (Fiji/Image/Stacks/3D_project) in each time point. As the ring plane was often tilted to the X-Y plane, the projection angle was determined manually for each ring prior to the projection. The shape of the ring at each time point was extracted from 2D projections using segmentation via auto-thresholding by Otsu (Fiji/Image/Adjust/Thresholding) and subsequent skeletonization of the resulting mask (Fiji/Process/Binary/Skeletonize). The centres of skeleton pixels were defined as skeleton points.

Further processing of ring skeletons was done in Matlab. In order to measure the local curvature of the ring skeleton at the point-of-interest (POI), two skeleton points that were 10 pixel points (2 µm apart along the skeleton) away from the POI in clockwise and counter-clockwise directions were assigned. Next, a circle with radius R was fitted to these three points. The local curvature at the POI was then derived as the inverse of the radius of that circle (1/radius). For representation in figures, the curvature was smoothened over 10 skeleton points.

Purification of the recombinant LifeAct-GFP fusion protein

Expression of GST-LifeAct-GFP was induced in E. coli BL21 (DE3-pLysS) (Promega) using 0.5 mM IPTG at 30°C for 4 hr. The recombinant protein was purified on Glutathione sepharose 4B beads according to the manufacture’s instructions (GE Healthcare). The elution buffer containing glutathione was exchanged to reactivation buffer (0.16 M sucrose, 5 mM MgCl2, 50 mM potassium acetate, 20 mM MOPS–NaOH [pH 7.0], pH adjusted to 7.5) using PD Minitrap G-10 columns (GE Healthcare). The purified recombinant proteins were stored at −80°C.

Immunoblotting

The immunoblot in Figure 3F was prepared as follows: cell ghost pellet and supernatant were separated by centrifugation (15000 xg, 5 min at 4°C) after 0.5 mM ATP addition for 20 min and 40 min. One millilitre of TCA precipitation buffer (250 mM NaOH, 7.5% TCA) was added to resuspend the cell ghost pellet and supernatant, and the suspension was incubated on ice for 10 min. Samples were centrifuged at 17,000 xg for 25 min at 4°C. Precipitated proteins were resuspended in HU-DTT (200 mM Tris-HCl [pH 6.8], 8 M urea, 5% SDS, 0.1 mM EDTA, 0.005% bromophenol blue, and 15 mg/ml DTT) as previously described (Palani et al., 2012). Samples were heated for 5 min at 95°C before loading on SDS-PAGE gels. Antibodies used were goat anti-actin (generous gifts from John Cooper, Washington University School of Medicine, St Louis, USA) and rabbit anti-Cdc8 (Balasubramanian et al., 1992). Secondary antibodies used were rabbit anti–goat, goat anti–rabbit, IgGs coupled to horseradish peroxidase (NEB; Jackson ImmunoResearch Laboratories).

Quantification of protein intensities

The signal intensities of indicated protein bands on immunoblots were measured using Fiji. Signal intensities were corrected against the gel background signal. The band intensity of the supernatant (S) and pellet (P) lanes was measured separately and summed up. To quantitate the fraction, the band intensity of supernatant lanes was divided by the sum of supernatant 'S' and pellet 'P' (fraction = band intensity/sum; sum = S + P). Four measurements were taken from blots derived from four independent experiments in Figure 3F.

Statistical analysis

Statistical significance was determined using Student’s t-test in Figure 1D, Figure 1—figure supplement 1H,I, Figure 2Eii,Eiv, Figure 3E,F, and Figure 4H. Calculations of mean, standard deviation (s.d.), and statistical significances, were done using Prism 6.0 (GraphPad).

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31

Decision letter

  1. Anthony A Hyman
    Reviewing Editor; Max Planck Institute of Molecular Cell Biology and Genetics, Germany

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.

[Editors’ note: a previous version of this study was rejected after peer review, but the authors submitted for reconsideration. The first decision letter after peer review is shown below.]

Thank you for submitting your work entitled "Curvature-Induced Expulsion of Actomyosin Bundles During Cytokinetic Ring Contraction" for consideration by eLife. Your article has been reviewed by two peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Richard Losick as the Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Michael Glotzer (Reviewer #1).

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife at this point.

Although both reviewers and the reviewing editor found your work of interest, they were unable to come to a conclusion until your data was better quantified. In particular we agree that a major flaw is the lack of a quantitative analysis of the relationship of the extruded actin to the ring actin.

We do note that this is a report, describing the initial finding of an exciting new mechanism for ring disassembly. Therefore, while the primary conclusions must be rigorously established, I think that some of the less central points could be stated less strongly and strengthened in future publications.

Therefore, I would recommend that you follow the suggestions of the reviewers to further quantify your work. If you feel that this still supports your conclusions, eLife would be happy to see a new manuscript.

Reviewer #1:

This interesting report presents strong evidence that under some conditions, bundles of actin are expelled by the contractile ring during constriction, rather than being depolymerized in place. The expulsion of actin bundles is more prominent as constriction occurs, can be seen in cell ghosts, and mechanical perturbations reveal that this specifically occurs at regions of high local curvature. The images provided are striking and the data are well quantified. These findings will be of interest for the readership of eLife.

1) While the authors do demonstrate this phenomenon in a wide variety of contexts including two different Schizosaccharomyces species, different labels, intact cells, spheroplasts, cell ghosts, upon seeing these images it is striking that this was not seen previously in all the extensive work on S. pombe cytokinesis, particularly as investigators have been visualizing cytokinesis en face quite frequently in recent years. Indeed, en face views with a similar/identical label (Rlc1) does not appear to show this expulsion behavior (PMID 25355954). In addition, some, but not all the experiments in this paper have been performed with LifeAct which can induce filament bundling. To ensure that this phenomenon is general and robust and intrinsic to the contractile ring, it would be important to visualize evidence of this behavior in intact cells. It would be best to fix LifeAct-GFP, Rlc1-GFP, and unlabeled cells of both S. pombe and S. japonicus and stain them with phalloidin. It would be critical to detect these extruded filaments in cells that do not express a fluorescent protein (performing the experiment with the FP will allow the authors to assess that these bundles can be preserved and labeled).

Reviewer #2:

It is well-established that the cytokinetic ring closes via actomyosin dynamics. A more poorly understood aspect of ring ingression is that, as its diameter decreases, ring components are likely eliminated to allow continued contraction. In this manuscript, Huang et al. propose that actin and myosin containing filaments can be continuously expulsed from the contractile ring contributing to the completion of cytokinesis in two yeast species. Since the authors only observe such filaments in late stages of cytokinesis they suggest that this represents a separate phase during cytokinesis where filament expulsion promotes ring closure. Furthermore, the authors claim that this expulsion phenomenon is dependent on changes in cell curvature and is driven by mechanical forces generated by high curvature independent of cytoplasmic regulators or polymerization driven elongation of these filaments.

While the model proposed by the authors is intriguing and has the potential to expand our understanding of the mechanism of contractile ring closure, in its current form, the manuscript fails to support many of the authors' assertions and requires significant revisions and improvements before being suitable for publication in eLife.

In general, much of the data are poorly annotated in the figure, figure legend, text, or all three places. Some data also appear to be contradictory to the conclusion drawn by the authors or lacking quantification to support the conclusions. The following is a list of comments in chronological order. I cannot predict whether a revision will be suitable for publication before these points are addressed.

1) Many graphs presented in this manuscript contain the axis label relative intensity (AU) without any mention of what the intensity is normalized to. Typically, when relative values are shown in a graph, one of the data sets presented has the average of 1, and thus indicates that it is the reference point. This is not the case for any of the graphs (non-exhaustive list of figures affected: Figure 1—figure supplement 1C, Figure 1—figure supplement 1F, Figure 3C, Figure 3D). The notation of figures, in general, needs to be improved. "B.r. and s.r." are not appropriate abbreviations – writing out "big and small" would be adequate and far less cumbersome for the reader. The protein imaged should always be part of the image notation. Spectrum bars need to be annotated on the figure.

2) The data presented in the top and bottom panels of Figure 1—figure supplement 1D are not consistent. The bottom panel shows drastic reduction in fluorescence in the ring over time while levels of ring fluorescence in the top seems to change very little. Is the top series saturated? Which one is representative of properly imaged ring dynamics?

3) What is the difference between Figure 1—figure supplement 1C and 1F? 1C shows a 3-fold decrease in fluorescence intensity while the intensities in Figure 1—figure supplement 1F are barely statistically significantly different. (How is total actin detected and distinguished from actin in the ring in Figure 1—figure supplement 1?) In fact, claiming that there is a significant difference between small and large rings in 1F is tenuous at best since the difference is minimal and could well be explained by a few outlying data points.

4) The following conclusion is not supported by the data presented: "Cdc8p, IQGAP protein Rng2p, and F-BAR protein Cdc15p were also present in actin bundles associated with contracting rings (Figure 1—figure supplement 1G), which suggested that the bundles associated with contracting actomyosin rings signified disassembly of the entire ring as a single unit". All that is shown is that the bundles contain these components. Whether they have been part of a disassembling ring or just associated with filaments after expulsion is not clear. Along the same line, the authors observe bi-directional movement of myosin in expulsed bundles. Is this behavior observed in the ring? If not, wouldn't this suggest that these filaments are structurally distinct from filaments in the ring?

5) The biphasic nature of ring closure speed appears to be a minor, poorly substantiated observation, since it is entirely possible that expulsion starts to happen earlier but that the filaments are short and therefore not detectable. The really interesting observation in my opinion is the relationship between ring perimeter and bundle length. There appears to be a pretty good correlation in phase 2, suggesting that expulsion is a compensatory mechanism for reduction in ring size. However, the quantification of this in Figure 1F is confusing and contradictory. What are the units in the graphs? It appears that total length of bundles is larger than ring perimeter (max value 58 vs. 28) but T.I. + perimeter is smaller than perimeter alone? In addition, biphasic behavior of ring intensity (constant in phase 1, decreasing in phase 2), would further substantiate this biphasic concept.

6) The following statement is not supported by the data presented "These experiments showed that loss of function of actomyosin ring proteins tested did not affect the timing or the ability of these cells to expel actomyosin bundles during ring contraction". All that is shown are two pictures for each condition showing that small rings have bundles but no quantification is provided. All one can conclude from this is that neither of those proteins is essential for formation of these bundles. What about the kinetics, number of bundle, length or dynamics? Given the amount of redundancy in regulating cytokinesis, it would not be surprising if any one perturbation does not completely abolish the formation of bundles but the protein in question may still contribute to bundle formation and dynamics.

7) When making cell ghosts it appears that rlc1p-gfp is no longer associated with actin bundles. Is this just not evident from the picture shown or does this mean that myosin is lost from the bundle?

8) How is total actin intensity measured in cell ghost experiments? Just the sum of intensities on ring and filaments or total fluorescence in entire field of view? I presume Cdc8p is a loading control for western blots but this needs to be explicitly stated in the text. What is the quantification of the western blot comparing? Supernatant to pellet levels, loading control to actin monomers, filaments? It appears that most actin in the preparation is monomeric so that a significant reduction in filamentous actin may not be easily detected by changes in monomeric actin.

9) The authors claim that expulsion depends on ring curvature in non-compressed cells The data presented are insufficient to support this conclusion. All I see in Figure 4A is a picture. No quantitative data except for an arbitrary value of 68% of something, with no explanation of what is considered high curvature. In fact, based on the data presented it would be reasonable to conclude that high curvature is not essential for expulsion since a large fraction of filaments form in regions of "low" curvature (32%). In fact, maximum expulsion seems to occur at intermediate curvatures not at high curvature and small rings with high curvature appear to have fewer expulsion events.

10) The description of experimental details in the Materials and methods section needs significant revision. Just one example: "Unless specified otherwise, all spinning-disk images were shown by 2D maximum/sum intensity projections." Which one is it? There is a critical difference between sum and max projections. What do the authors mean by 2D projection? Do they mean Z-projection?

[Editors’ note: what now follows is the decision letter after the authors submitted for further consideration.]

Thank you for resubmitting your work entitled "Curvature-Induced Expulsion of Actomyosin Bundles During Cytokinetic Ring Contraction" for further consideration at eLife. Your revised article has been favorably evaluated by Richard Losick (Senior editor), a Reviewing editor, and one reviewer.

The authors have made significant changes to their manuscript that documents that during cytokinesis actin bundles with its associated binding proteins are expelled from the constricting contractile ring, particularly as it reaches small diameters. The authors make a compelling case that this is a common occurrence and that it constitutes a mechanism by which a significant portion of actin is eliminated from the ring.

But there are some remaining issues that need to be addressed before acceptance, as outlined below:

The data shown here, indicates that as the ring shrinks from a perimeter of ~26 µm to ~13 µm, relatively few bundles are observed, despite ~half of the f-actin being lost during this first period of constriction. There are three implications from this finding. First, it suggests that at early stages, "classical" disassembly is probably responsible for the loss of actin and ABPs. Second, unless ~1/3 of the actin is lost by extrusion during these early stages, the value of 68% of total ring lost by extrusion seems quite high (the solution to this appears to be that this value is calculated from rings that have already constricted significantly – a 3.5 µm diameter ring is not a typical early ring, other data in the manuscript indicates that an early ring is ~ 8 µm in diameter). Thus, this calculation appears somewhat misleading. Third, the manuscript should discuss how these two processes likely work together during ring constriction.

https://doi.org/10.7554/eLife.21383.016

Author response

[Editors’ note: the author responses to the first round of peer review follow.]

Reviewer #1:

[…]

1) While the authors do demonstrate this phenomenon in a wide variety of contexts including two different Schizosaccharomyces species, different labels, intact cells, spheroplasts, cell ghosts, upon seeing these images it is striking that this was not seen previously in all the extensive work on S. pombe cytokinesis, particularly as investigators have been visualizing cytokinesis en face quite frequently in recent years. Indeed, en face views with a similar/identical label (Rlc1) does not appear to show this expulsion behavior (PMID 25355954). In addition, some, but not all the experiments in this paper have been performed with LifeAct which can induce filament bundling. To ensure that this phenomenon is general and robust and intrinsic to the contractile ring, it would be important to visualize evidence of this behavior in intact cells. It would be best to fix LifeAct-GFP, Rlc1-GFP, and unlabeled cells of both S. pombe and S. japonicus and stain them with phalloidin. It would be critical to detect these extruded filaments in cells that do not express a fluorescent protein (performing the experiment with the FP will allow the authors to assess that these bundles can be preserved and labeled).

We agree with reviewer 1 that bundle expulsion that we report here was not obvious in the Fred Chang lab paper (Zhou et al:PMID 25355954). However, recently (while our paper was under review / revision) a similar expulsion phenomenon was noticed in S. pombe cells in a paper from Daniel Riveline’s group (Wollrab et al:PMID 27363521), although they did not characterize this bundle expulsion phenomenon. Rather they focused on a different aspect of actomyosin ring dynamics.

As suggested by reviewer 1, we performed formaldehyde fixation of untagged wild type cells and spheroplasts and wild-type cells and spheroplasts expressing LifeAct- GFP or Rlc1p-GFP in both S. japonicus and S. pombe (to ensure that the ring- associated bundles observed in our study was not an artifact of LifeAct-GFP or Rlc1p-GFP). After staining the actin structures with Rhodamine-conjugated phalloidin in these fixed samples, we were able to observe actin bundles associated with contracting actomyosin rings, including in untagged wild type cells. We have included these images in Figure 1—figure supplement 1A and 1B. This experiment clearly established that the bundles expelled during ring contraction were not an artifact of using non-native fluorescent markers, such as LifeAct-GFP or Rlc1p-GFP.

Reviewer #2:

[…]

While the model proposed by the authors is intriguing and has the potential to expand our understanding of the mechanism of contractile ring closure, in its current form, the manuscript fails to support many of the authors' assertions and requires significant revisions and improvements before being suitable for publication in eLife.

In general, much of the data are poorly annotated in the figure, figure legend, text, or all three places. Some data also appear to be contradictory to the conclusion drawn by the authors or lacking quantification to support the conclusions. The following is a list of comments in chronological order. I cannot predict whether a revision will be suitable for publication before these points are addressed.

We thank reviewer 2 for his/her detailed comments. We have carefully gone through and re-annotated the figures, and re-written/expanded figure legends, main text and Materials and methods. We believe these changes improve the clarity and accuracy of presentation.

1) Many graphs presented in this manuscript contain the axis label relative intensity (AU) without any mention of what the intensity is normalized to. Typically, when relative values are shown in a graph, one of the data sets presented has the average of 1, and thus indicates that it is the reference point. This is not the case for any of the graphs (non-exhaustive list of figures affected: Figure 1—figure supplement 1C, Figure 1—figure supplement 1F, Figure 3C, Figure 3D). The notation of figures, in general, needs to be improved. "B.r. and s.r." are not appropriate abbreviations – writing out "big and small" would be adequate and far less cumbersome for the reader. The protein imaged should always be part of the image notation. Spectrum bars need to be annotated on the figure.

We thank reviewer 2 for pointing out the incorrect axis label. This is unfortunately a mistake on our part in choosing the incorrect terminology. The referee is right in that these are not relative units, but they are the raw intensity of fluorescence of the proteins described. We have replaced all incorrect axis labels by showing the identity of the protein measured (e.g. actin intensity). All the micrographs are labeled with the protein being imaged and spectrum / calibration bars have also been included in the micrographs in the current manuscript. Abbreviations “b.r. and s.r.” have been replaced by “large and small”, as suggested.

2) The data presented in the top and bottom panels of Figure 1—figure supplement 1D are not consistent. The bottom panel shows drastic reduction in fluorescence in the ring over time while levels of ring fluorescence in the top seems to change very little. Is the top series saturated? Which one is representative of properly imaged ring dynamics?

The inconsistency of top and bottom panels in the previous Figure 1—figure supplement 1D was due to an unfortunate and erroneous swap of the bleach corrected panel with a bleach-uncorrected panel (we generated both versions during preparation of the manuscript). We now show the appropriate bleach corrected micrographs in the top and bottom panels that reveals the phenomenon of bundle expulsion even better than in the original manuscript (shown in Figure 1—figure supplement 1F in the revised manuscript).

3) What is the difference between Figure 1—figure supplement 1C and 1F? 1C shows a 3-fold decrease in fluorescence intensity while the intensities in Figure 1—figure supplement 1F are barely statistically significantly different. (How is total actin detected and distinguished from actin in the ring in Figure 1—figure supplement 1?) In fact, claiming that there is a significant difference between small and large rings in 1F is tenuous at best since the difference is minimal and could well be explained by a few outlying data points.

The previous Figure 1—figure supplement 1C showed the intensity of actin present in the ring (excluding ring-associated actin bundles) in the analysis. The previous Figure 1—figure supplement 1F showed the combined actin intensity (total actin) from rings and associated bundles. Please note that the Figure 1—figure supplement 1C and 1F are now Figure 1—figure supplement 1H and 1I in the current manuscript.

To measure the sum actin fluorescence intensity, the permeabilized spheroplasts were stained with CF633-phalloidin to label the actin structures. The total actin was measured by drawing a square to cover the actomyosin ring and the associated bundles as the region-of-interest, whereas the ring associated actin was measured by drawing a line along the ring circumference without including ring-associated bundles as the region-of-interest. These details are provided in the corresponding figure legend.

In the previous Figure 1—figure supplement 1F we were not claiming any difference between total actin in large and small rings. Rather, based on the statistics presented, we were claiming that they were of comparable intensity (as the referee agrees from his/her own analysis). Our apologies that this was described confusingly. This has been clearly rewritten now.

4) The following conclusion is not supported by the data presented: "Cdc8p, IQGAP protein Rng2p, and F-BAR protein Cdc15p were also present in actin bundles associated with contracting rings (Figure 1—figure supplement 1G), which suggested that the bundles associated with contracting actomyosin rings signified disassembly of the entire ring as a single unit". All that is shown is that the bundles contain these components. Whether they have been part of a disassembling ring or just associated with filaments after expulsion is not clear. Along the same line, the authors observe bi-directional movement of myosin in expulsed bundles. Is this behavior observed in the ring? If not, wouldn't this suggest that these filaments are structurally distinct from filaments in the ring?

We have performed time-lapse microscopy to document the dynamics of two cytokinetic ring proteins: GFP-Rng2p and Cdc15p-GFP (in Figure 2B), in addition to Rlc1p-GFP, which was presented in the original submission. Consistent with our original description from single images, these proteins were found to be associated with the expelling / expelled bundle in contracting rings using time-lapse microscopy (presented in the revised manuscript, Figure 2B). Despite this, we have toned down the statement (for reasons described in the response to #7 below) that “the bundles associated with contracting actomyosin rings signified disassembly of the entire ring as a single unit”, by stating that several cytokinetic ring proteins associate with expelling / expelled bundles in living cells and discussed this in the manuscript (Results and Discussion, tenth paragraph).

We have removed the data showing bi-directional movements of Rlc1p-GFP on expelled bundles as, in our opinion, it is less central to our study describing bundle expulsion. This is partly also due to the suggestion from the editors to remove data not central to the main point of our manuscript. However, we have been able to detect clusters moving in the ring, although this point has been better described in a recent paper from Daniel Riveline and colleagues that was published (Woolrab et al; PMID 27363521) while our paper was under review/revision.

5) The biphasic nature of ring closure speed appears to be a minor, poorly substantiated observation, since it is entirely possible that expulsion starts to happen earlier but that the filaments are short and therefore not detectable. The really interesting observation in my opinion is the relationship between ring perimeter and bundle length. There appears to be a pretty good correlation in phase 2, suggesting that expulsion is a compensatory mechanism for reduction in ring size. However, the quantification of this in Figure 1F is confusing and contradictory. What are the units in the graphs? It appears that total length of bundles is larger than ring perimeter (max value 58 vs. 28) but T.I. + perimeter is smaller than perimeter alone? In addition, biphasic behavior of ring intensity (constant in phase 1, decreasing in phase 2), would further substantiate this biphasic concept.

We have removed the description of the biphasic ring contraction behaviour in the current manuscript, again to streamline and focus on the major point that ring disassembly occurs in large part through bundle expulsion. The graphs have been updated with the appropriate units (Figure 2E). We show individual graphs in Figure 2E without insets so that data points are clearer.

Thanks to Reviewer 2 for pointing out the confusing data points, i.e. total bundle length + ring perimeter in contracted rings is smaller than the initial ring perimeter. We investigated the graph again, and realized that the upper limit for y-axis (inset of 3rd graph in our previously submitted manuscript) was unfortunately set at a lower value erroneously, which eliminated some of the important points. We have rectified this mistake and the revised graph fully supports the conclusions that we have made (Figure 3Ev).

6) The following statement is not supported by the data presented "These experiments showed that loss of function of actomyosin ring proteins tested did not affect the timing or the ability of these cells to expel actomyosin bundles during ring contraction". All that is shown are two pictures for each condition showing that small rings have bundles but no quantification is provided. All one can conclude from this is that neither of those proteins is essential for formation of these bundles. What about the kinetics, number of bundle, length or dynamics? Given the amount of redundancy in regulating cytokinesis, it would not be surprising if any one perturbation does not completely abolish the formation of bundles but the protein in question may still contribute to bundle formation and dynamics.

We have removed this section on the analysis of mutants, as these experiments pertain to S. pombe, whereas the vast majority of the study is on S. japonicus. Also, as pointed out by the referee, while our data are suggestive they are not conclusive, which requires generation of a large number of double and triple mutants in S. pombe, and crucially generation of many single mutants in S. japonicus, all of which are beyond the scope of this study.

7) When making cell ghosts it appears that rlc1p-gfp is no longer associated with actin bundles. Is this just not evident from the picture shown or does this mean that myosin is lost from the bundle?

We have been able to detect Rlc1p, Cdc15p, Rng2p, and F-actin in expelled bundles in cells and spheroplasts, but only F-actin was clearly detected in expelled bundles in cell ghosts. Although the reasons for this are presently unclear, it is likely that ring proteins undergo constant turnover in expelled actin bundles (as they do in actomyosin rings) in cells and spheroplasts in which a continuous supply of a molecularly crowded cytosol is available, whereas this is not available in cell ghosts. Alternatively, the molecular crowding itself may help retain ring proteins on actin bundles whereas in the absence of such crowding, ring proteins may be lost from actin bundles. We have discussed this in the revised manuscript (Results and Discussion, eleventh paragraph).

8) How is total actin intensity measured in cell ghost experiments? Just the sum of intensities on ring and filaments or total fluorescence in entire field of view? I presume Cdc8p is a loading control for western blots but this needs to be explicitly stated in the text. What is the quantification of the western blot comparing? Supernatant to pellet levels, loading control to actin monomers, filaments? It appears that most actin in the preparation is monomeric so that a significant reduction in filamentous actin may not be easily detected by changes in monomeric actin.

To measure the sum actin fluorescence intensity in cell ghosts, they were stained with CF633-phalloidin to label the actin structures. The total actin intensity was measured by drawing a square to cover the actomyosin ring and the associated bundles as the region-of-interest, whereas the ring associated actin was measured by drawing a line along the ring circumference without including ring-associated bundles as the region-of-interest. These details are provided in the corresponding figure legend.

The western blots compare actin and an actin-associated protein (Cdc8p- tropomyosin) in the supernatant and pellet. Cdc8p was not used as a loading control. We have changed the axis labeling to “supernatant fraction”. This fraction is calculated by dividing the protein band intensity of supernatant lanes to the sum intensity of the bands of supernatant lanes and pellet lanes (fraction = supernatant intensity / (supernatant intensity + pellet intensity). We have expanded the descriptions of protein band intensity quantification in the Materials and methods section of the current manuscript.

9) The authors claim that expulsion depends on ring curvature in non-compressed cells The data presented are insufficient to support this conclusion. All I see in Figure 4A is a picture. No quantitative data except for an arbitrary value of 68% of something, with no explanation of what is considered high curvature. In fact, based on the data presented it would be reasonable to conclude that high curvature is not essential for expulsion since a large fraction of filaments form in regions of "low" curvature (32%). In fact, maximum expulsion seems to occur at intermediate curvatures not at high curvature and small rings with high curvature appear to have fewer expulsion events.

The data that support the conclusion that expulsion depends on ring curvature was/is presented in Figure 4A in the original and revised manuscript. In the representative movie shown (n = 11) in Figure 4A, expelled bundles become visible from ~ the 60-second time point, and with progression of time and increasing curvature the expelled bundle number increases. We find that although rings can start with curvatures (1/radius) of even smaller than 0.4 µm-1 (translates to a ring diameter of > 5 µm in spheroplasts), the mean curvature at which bundles begin to be expelled is ~ 1.1 µm-1 (translates to a ring diameter of ~ 2 µm). Given that bundle expulsion is only observed after a significant reduction in ring diameter, we believe our data strongly support the view that bundle expulsion in uncompressed rings is promoted by increased curvature. We have then tested this hypothesis with the spheroplast compression experiment.

Our apologies for the confusing description of Figure 4—figure supplement 1A. In this panel we are looking at a partially contracted ring in a cell ghost. Rings in cell ghosts do not undergo symmetric contraction. Rather, they undergo irregular contraction with regions of sharp edges. What we were indicating in this figure was that, even within a curved and contracted ring in a cell ghost ~ 68% of actin bundles are expelled from regions where the ring is bent. In other words, the entire ring has an increased curvature, due to partial contraction, but nearly 68% of the cables are found preferentially in regions in which the rings are bent, which means that the rings have a further increased local curvature. We were unable to perform curvature analysis on contracting isolated rings in cell ghosts because the rings do not contract on a flat x-y plane. The poor z-resolution affects the accuracy of curvature analysis. We took precaution not to over-estimate the number, thus, the 68% is a conservative value of this quantification.

Together, Figure 4 and Figure 4—figure supplement 1, establish that bundle expulsion is promoted by increased ring curvature even in uncompressed cells.

10) The description of experimental details in the Materials and methods section needs significant revision. Just one example: "Unless specified otherwise, all spinning-disk images were shown by 2D maximum/sum intensity projections." Which one is it? There is a critical difference between sum and max projections. What do the authors mean by 2D projection? Do they mean Z-projection?

We fully appreciate that maximum intensity and sum intensity projections are very different. Our apologies for not making this point clear in the original manuscript. This has been fully rectified in the revised manuscript. We have clearly stated whether an image shows a maximum intensity or sum intensity in the Methods section, which clarifies the panels in which maximum intensity was used and those in which sum intensity was used. The referee is right in that we are using Z- projection to show a 2-dimensional image in the X-Y plane. This has now been corrected to “projected along the Z-axis”.

[Editors' note: the author responses to the re-review follow.]

The authors have made significant changes to their manuscript that documents that during cytokinesis actin bundles with its associated binding proteins are expelled from the constricting contractile ring, particularly as it reaches small diameters. The authors make a compelling case that this is a common occurrence and that it constitutes a mechanism by which a significant portion of actin is eliminated from the ring.

But there are some remaining issues that need to be addressed before acceptance, as outlined below:

The data shown here, indicates that as the ring shrinks from a perimeter of ~26 µm to ~13 µm, relatively few bundles are observed, despite ~half of the f-actin being lost during this first period of constriction. There are three implications from this finding. First, it suggests that at early stages, "classical" disassembly is probably responsible for the loss of actin and ABPs. Second, unless ~1/3 of the actin is lost by extrusion during these early stages, the value of 68% of total ring lost by extrusion seems quite high (the solution to this appears to be that this value is calculated from rings that have already constricted significantly – a 3.5 µm diameter ring is not a typical early ring, other data in the manuscript indicates that an early ring is ~ 8 µm in diameter). Thus, this calculation appears somewhat misleading. Third, the manuscript should discuss how these two processes likely work together during ring constriction.

We have read this comment and our manuscript carefully, and do not understand how the reviewer arrived at ~ half of the F-actin being lost during this first period of constriction. We chose early rings as those with a diameter of ~ 3.5 µm since in other experiments in Figure 2C, we only observed bundles when rings reached a diameter of ~ 2.5 µm. Since the goal of the experiment suggested by reviewer 1 was to estimate the fraction of the ring-associated actin lost through bundle expulsion, an initial diameter of ~ 3.5 µm seemed like a good starting point. We therefore do not believe the reported calculations are misleading. As to the issue of whether actin is also lost through other “classical” disassembly pathways, this is an interesting and open question. Further analysis, using better probes for the actin cytoskeleton will be required to fully address this possibility.

As suggested by this reviewer, we have made a mention of these considerations in the revised manuscript.

https://doi.org/10.7554/eLife.21383.017

Article and author information

Author details

  1. Junqi Huang

    Division of Biomedical Sciences, Warwick Medical School, University of Warwick, Coventry, United Kingdom
    Contribution
    JH, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article
    Contributed equally with
    Ting Gang Chew
    For correspondence
    junqi.huang@warwick.ac.uk
    Competing interests
    No competing interests declared.
  2. Ting Gang Chew

    Division of Biomedical Sciences, Warwick Medical School, University of Warwick, Coventry, United Kingdom
    Contribution
    TGC, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article
    Contributed equally with
    Junqi Huang
    For correspondence
    t.g.chew@warwick.ac.uk
    Competing interests
    No competing interests declared.
  3. Ying Gu

    Randall Division of Cell and Molecular Biophysics, King's College London, London, United Kingdom
    Contribution
    YG, Generated yeast strains and edited the manuscript, Contributed unpublished essential data or reagents
    Contributed equally with
    Saravanan Palani
    Competing interests
    No competing interests declared.
  4. Saravanan Palani

    Division of Biomedical Sciences, Warwick Medical School, University of Warwick, Coventry, United Kingdom
    Contribution
    SP, Acquisition of data, Analysis and interpretation of data
    Contributed equally with
    Ying Gu
    Competing interests
    No competing interests declared.
  5. Anton Kamnev

    Division of Biomedical Sciences, Warwick Medical School, University of Warwick, Coventry, United Kingdom
    Contribution
    AK, Acquisition of data, Analysis and interpretation of data
    Competing interests
    No competing interests declared.
  6. Douglas S Martin

    Department of Physics, Lawrence University, Appleton, United States
    Contribution
    DSM, Acquisition of data, Analysis and interpretation of data
    Competing interests
    No competing interests declared.
  7. Nicholas J Carter

    Division of Biomedical Sciences, Warwick Medical School, University of Warwick, Coventry, United Kingdom
    Contribution
    NJC, Acquisition of data, Analysis and interpretation of data
    Competing interests
    No competing interests declared.
  8. Robert Anthony Cross

    Division of Biomedical Sciences, Warwick Medical School, University of Warwick, Coventry, United Kingdom
    Contribution
    RAC, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article
    Competing interests
    No competing interests declared.
    ORCID icon 0000-0002-0004-7832
  9. Snezhana Oliferenko

    Randall Division of Cell and Molecular Biophysics, King's College London, London, United Kingdom
    Present address
    Francis Crick Institute, London, United Kingdom
    Contribution
    SO, Drafting or revising the article, Contributed unpublished essential data or reagents
    Competing interests
    No competing interests declared.
  10. Mohan K Balasubramanian

    Division of Biomedical Sciences, Warwick Medical School, University of Warwick, Coventry, United Kingdom
    Contribution
    MKB, Conception and design, Analysis and interpretation of data, Drafting or revising the article
    For correspondence
    m.k.balasubramanian@warwick.ac.uk
    Competing interests
    MKB: Reviewing editor, eLife
    ORCID icon 0000-0002-1292-8602

Funding

Wellcome Trust (103895/Z/14/Z)

  • Robert Anthony Cross

Wellcome Trust (103741/Z/14/Z)

  • Snezhana Oliferenko

Wellcome Trust (WT101885MA)

  • Mohan K Balasubramanian

Royal Society

  • Mohan K Balasubramanian

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We would like to acknowledge members of the MKB. laboratory for discussion. This work was supported by Wellcome Trust Senior Investigator Awards to MKB. (WT101885MA), RAC. (103895/Z/14/Z), and SO (103741/Z/14/Z). MKB. was also supported by a Royal Society Wolfson Merit Award.

Reviewing Editor

  1. Anthony A Hyman, Reviewing Editor, Max Planck Institute of Molecular Cell Biology and Genetics, Germany

Publication history

  1. Received: September 9, 2016
  2. Accepted: October 12, 2016
  3. Accepted Manuscript published: October 13, 2016 (version 1)
  4. Version of Record published: October 24, 2016 (version 2)
  5. Version of Record updated: October 25, 2016 (version 3)

Copyright

© 2016, Huang 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.

Metrics

  • 1,836
    Page views
  • 501
    Downloads
  • 3
    Citations

Article citation count generated by polling the highest count across the following sources: PubMed Central, Crossref, Scopus.

Comments

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading