1. Cell Biology
Download icon

Nanoscale architecture of the Schizosaccharomyces pombe contractile ring

  1. Nathan A McDonald
  2. Abigail L Lind
  3. Sarah E Smith
  4. Rong Li
  5. Kathleen L Gould  Is a corresponding author
  1. Vanderbilt University, United States
  2. Vanderbilt University Medical Center, United States
  3. Stowers Institute for Medical Research, United States
  4. Johns Hopkins University School of Medicine, United States
Research Article
  • Cited 19
  • Views 2,039
  • Annotations
Cite this article as: eLife 2017;6:e28865 doi: 10.7554/eLife.28865

Abstract

The contractile ring is a complex molecular apparatus which physically divides many eukaryotic cells. Despite knowledge of its protein composition, the molecular architecture of the ring is not known. Here we have applied super-resolution microscopy and FRET to determine the nanoscale spatial organization of Schizosaccharomyces pombe contractile ring components relative to the plasma membrane. Similar to other membrane-tethered actin structures, we find proteins localize in specific layers relative to the membrane. The most membrane-proximal layer (0–80 nm) is composed of membrane-binding scaffolds, formin, and the tail of the essential myosin-II. An intermediate layer (80–160 nm) consists of a network of cytokinesis accessory proteins as well as multiple signaling components which influence cell division. Farthest from the membrane (160–350 nm) we find F-actin, the motor domains of myosins, and a major F-actin crosslinker. Circumferentially within the ring, multiple proteins proximal to the membrane form clusters of different sizes, while components farther from the membrane are uniformly distributed. This comprehensive organizational map provides a framework for understanding contractile ring function.

https://doi.org/10.7554/eLife.28865.001

Introduction

In many eukaryotes, including animals and fungi, cell division is accomplished by an actin- and myosin-based contractile apparatus (Green et al., 2012). This complex structure is tightly linked to the plasma membrane and uses myosin motors to constrict an F-actin ring (Cheffings et al., 2016), pulling opposing sides of the membrane together.

Studies in S. pombe have led the cytokinesis field in identifying components and regulators of the contractile ring (Cheffings et al., 2016; Pollard and Wu, 2010; Rincon and Paoletti, 2016; Willet et al., 2015a; Goyal et al., 2011). Genetic screens as well as genome-wide and targeted localization studies have determined a complete ‘parts list’ of protein components that comprise the ring, many of which are conserved in higher eukaryotes (Nurse et al., 1976; Balasubramanian et al., 1998; Chang et al., 1996; Matsuyama et al., 2006). 318 proteins are annotated as localizing to the S. pombe division site (Matsuyama et al., 2006; Wood et al., 2012), which includes both the contractile ring and the lining of the division septum formed during ring constriction. Only a subset of these proteins (38 according to PomBase annotation [Wood et al., 2012]) make up the contractile ring itself.

Though the proteins that comprise the contractile ring have been identified, how these components are knit together into a functional division machine remains unclear despite several substantive efforts towards unraveling this complex question. The S. pombe contractile ring forms in the middle of the cell from precursor ‘nodes’, membrane-tethered protein foci that contain anillin Mid1, IQGAP Rng2, myosin-II heavy chain Myo2 and light chains Cdc4 and Rlc1, F-BAR Cdc15, and formin Cdc12 (Wu et al., 2006). Precursor nodes coalesce into a contiguous ring over ~20 min that recruits many additional components over a further ~20 min before constriction after mitotic exit. The orientation of 5 components within these nodes has been determined (Laporte et al., 2011), and quantitative fluorescence studies have even been used to estimate the number of molecules of many proteins per node as well as in the fully-formed ring (Laporte et al., 2011; Wu and Pollard, 2005). Knowledge from these studies has been incorporated into mathematical models which attempt to understand ring formation and constriction. A search-capture-pull-release model of node condensation was found to recapitulate basic ring formation (Vavylonis et al., 2008; Ojkic et al., 2011), while biophysical tension measurements of the ring have been used to model ring constriction (Stachowiak et al., 2014). Though these models are becoming increasingly complex and explanatory, the field is hampered by sparse information about the fundamental molecular architecture of the ring.

Ultimately, the resolution limit of conventional fluorescence microscopy (~250 nm) restrains the spatial information attainable by studies of S. pombe nodes and contractile rings, each only 100–200 nm in width (Wu et al., 2006; Laplante et al., 2016). At higher resolution, one electron microscopy study revealed that the ring is composed of a dense array of 1000–2000 F-actin filaments with mixed directionality (Kamasaki et al., 2007); however, additional protein components could not be detected with this technique. New super-resolution microscopy technologies, based on the precise (<50 nm) localization of single photoactivated fluorescent molecules (Betzig et al., 2006; Rust et al., 2006; Hess et al., 2006), have the potential to drive our understanding of the contractile ring to a truly molecular level.

Super-resolution methods have recently been effective at determining the molecular architecture and revealing the inner mechanics of multiple cellular structures (Sydor et al., 2015). In focal adhesions, the plasma membrane and F-actin were found to be separated by distinct layers of proteins: an integrin signaling layer, a force transduction layer, and an actin regulatory layer (Kanchanawong et al., 2010), revealing a potential mechanism of force-induced focal adhesion formation and maintenance. At centrosomes, the pericentriolar matrix (PCM) was found to organize into two structural layers: one directly apposed to the centriole wall and a second extending radially outward (Mennella et al., 2012), scaffolded by the N- and C-termini of pericentrin-like protein, respectively. Furthermore, super-resolution microscopy has revealed previously unresolvable structures, such as actin-spectrin periodic repeats which coat the membranes of axons in animals (Xu et al., 2013; He et al., 2016). Super-resolution microscopy has also recently been applied to S. pombe cytokinesis in a study that investigated 6 proteins’ orientations within precursor nodes relative to each other (Laplante et al., 2016).

Using fluorescence photoactivation localization microscopy (fPALM) we approached the question of the molecular architecture of the contractile ring by comprehensively mapping 29 protein components’ spatial organization relative to the underlying plasma membrane. We determined that the contractile ring is composed of layers of protein components at distinct positions interior to the plasma membrane, similar to other membrane-tethered actin structures like focal adhesions (Kanchanawong et al., 2010) and cadherin junctions (Bertocchi et al., 2017), a conclusion verified by fluorescence resonance energy transfer (FRET) experiments. Moreover, we find certain components are uniformly distributed circumferentially within the ring, while others form clusters of various sizes and variable spacing. These data provide a structural framework for understanding the formation, mechanics, and regulation of the contractile ring.

Results

fPALM strategy for determining the spatial organization of contractile ring components

To determine the molecular architecture of the contractile ring using fPALM, we measured the precise spatial distribution of 29 proteins endogenously tagged with mMaple3 relative to the plasma membrane marked with the membrane-binding probe mCling-ATTO647N (Revelo et al., 2014) (Figure 1A). The photoactivatable mMaple3 fluorophore was chosen because it has multiple advantages over the commonly used mEos3.2 (Zhang et al., 2012), most importantly a faster maturation time (49 versus 330 min) (Wang et al., 2014). The mMaple3 fluorophore is ~5 nm in size and was connected to protein termini with a short 11 amino acid linker, minimizing positional uncertainty. In many cases, the spatial distribution of both protein termini was interrogated to determine molecular orientations. We imaged the middle focal plane of cells and utilized optical astigmatism to precisely measure the Z positions of each photoactivated molecule in order to restrict our analysis to a 500 nm Z section (Huang et al., 2008a). This imaging scheme resulted in a significant increase in resolution from conventional fluorescence (Figure 1B) to fPALM images (Figure 1C). Overall, we achieved an average of 34 and 37 nm localization uncertainty for mMaple3 and ATTO647N fluorophores, respectively (Figure 1D).

Figure 1 with 1 supplement see all
fPALM strategy to localize contractile ring proteins relative to the plasma membrane with nanometer resolution.

(A) Schematic of fPALM sample setup. Contractile ring proteins were endogenously tagged with mMaple3 while the plasma membrane was labeled with mCling-ATTO647N. A 3D-fPALM system was used to restrict the imaging plane to a narrow 500 nm Z section. (B) Conventional fluorescence microscopy images of a contractile ring protein and membrane dye. (C) Resulting 2 color fPALM image after localization analysis. Particles are visualized as normalized Gaussians with standard deviation = localization uncertainty. (D) Localization uncertainty achieved with mMaple3 and ATTO647N fluorophores for all data in this study. (E) Example histogram of Cdc15-mMaple3 localization distances from the mCling-ATTO647N membrane edge. A Gaussian curve was fitted to the data from which dcenter and σvert parameters can be extracted. Scale bars = 1 µm.

https://doi.org/10.7554/eLife.28865.002

Before examining contractile ring proteins, we evaluated the mCling-ATTO647N probe’s ability to identify the plasma membrane. First, we verified that mCling-ATTO647N colocalizes with the FM4-64 membrane dye (Figure 1—figure supplement 1A). In fact, mCling is similarly trafficked into intracellular membrane compartments, confirming its membrane incorporation (Figure 1—figure supplement 1A). However, at super-resolution the mCling-ATTO647N signal is wider than the ~10 nm plasma membrane (Figure 1—figure supplement 1B), indicating that mCling-ATTO647N may also bind nonspecifically to the cell wall. Nevertheless, the interior edge of the mCling signal aligns extremely well with the edge of an Acyl-mMaple3 signal, validating the use of mCling to mark the cytosolic face of the plasma membrane (blue line, Figure 1—figure supplement 1B).

To quantify the spatial distribution of each protein component relative to the plasma membrane, we performed an analysis similar to that performed on focal adhesion and cadherin junction proteins (Kanchanawong et al., 2010; Bertocchi et al., 2017). The cytoplasmic edge of the plasma membrane was defined in an unbiased manner as the position where the mCling-ATTO647N signal dropped to 5%. The distance from the defined plasma membrane edge was then calculated for each individual protein localization in a contractile ring side-view ‘spot’. The average distance from the membrane (dcenter), a vertical width parameter (σvert), and a horizontal width parameter (σwidth) were determined as in Kanchanawong et al. (Kanchanawong et al., 2010), and these values were then averaged across multiple rings. We restricted our analysis to fully formed, unconstricted contractile rings that are present in cells for 15–20 min. Cells containing rings of this stage were identified based on the absence of precursor nodes, the presence of a tightly concentrated ring absolutely perpendicular to the cell’s long axis, and the lack of any membrane ingression or septum formation (Figure 1—figure supplement 1C).

Spatial distribution of structural contractile ring components

We first determined the spatial distribution of 19 structural components of the contractile ring relative to the membrane (Figure 2A–B, Figure 2—figure supplement 1). Endogenous mMaple3 fusions of these proteins were functional and had no impact on growth or division (Figure 2—figure supplement 2). The resolution of the mMaple3 probe (Figure 1D) was sufficient to distinguish differences in distance from the membrane between components. Overall, we found the fully-formed contractile ring is made up of components 182 ± 26 nm wide and extends from the plasma membrane 293 ± 64 nm into the cytoplasm.

Figure 2 with 2 supplements see all
Nanoscale organization of contractile ring structural components.

(A) Representative fPALM images of structural contractile ring components. Scale, 100 nm. Particles are visualized as normalized Gaussians with standard deviation = localization uncertainty. Dashed lines indicate plasma membrane edges. (B) Distance from the plasma membrane (dcenter) of structural contractile ring components. Box plots depict 1st and 3rd quartiles and median; Whiskers, minimum and maximum; Notches, 95% confidence intervals. Lines connect proteins labeled on opposite termini.

https://doi.org/10.7554/eLife.28865.004

Multiple membrane-binding proteins including the anillin Mid1, the F-BAR proteins Cdc15 and Imp2, and the septin Spn3 are positioned within 80 nm of the plasma membrane. Both termini of Mid1 are within 50 nm of the membrane, consistent with direct membrane binding through its C2 domain (Sun et al., 2015; Rincon and Paoletti, 2012; Celton-Morizur et al., 2004). Cdc15 and Imp2 contain N-terminal F-BAR domains that also directly bind the membrane (Roberts-Galbraith et al., 2010; McDonald et al., 2016; McDonald et al., 2015). We find Spn3 near the membrane, consistent with its assembly into membrane-bound oligomers with additional septin subunits (An et al., 2004). In addition to membrane-binding proteins, the C-terminal tail of the essential contractile ring myosin-II Myo2 (Kitayama et al., 1997; Palani et al., 2017) is near the membrane, as was observed in cytokinesis precursor nodes (Laporte et al., 2011; Laplante et al., 2016). The N-terminus of the formin Cdc12 is also found near the membrane, in accord with direct binding to the F-BAR domain of Cdc15 (Willet et al., 2015b; Carnahan and Gould, 2003). The C-terminus of Cdc12 is found at a similar distance, further supporting its localization to a membrane-proximal stratum.

Slightly further from the membrane (80–160 nm) we find the IQGAP Rng2, a scaffolding protein critical for contractile ring formation and constriction (Padmanabhan et al., 2011; Tebbs and Pollard, 2013). Multiple accessory proteins important for ring integrity are also present in this layer and are linked to the underlying layer through direct protein-protein interactions: Cyk3, Spa2, Blt1, paxillin Pxl1, Nod1, and Fic1 (Roberts-Galbraith et al., 2010; Bohnert and Gould, 2012; Ren et al., 2015; Pollard et al., 2012; Zhu et al., 2013; Guzman-Vendrell et al., 2013; Pinar et al., 2008; Ge and Balasubramanian, 2008). Pxl1’s localization to an intermediate layer between the membrane and F-actin is comparable to paxillin’s position between membrane-embedded integrins and F-actin in focal adhesions (Kanchanawong et al., 2010). At similar intermediate distances from the membrane, we find the tail domains of two additional myosins that contribute to S. pombe cytokinesis: the non-essential myosin-II, Myp2 (Bezanilla et al., 1997; Mulvihill and Hyams, 2003), as well as a myosin-V, Myo51 (Win et al., 2001; Tang et al., 2016). Though both are non-essential, Myp2 is implicated in ring constriction (Palani et al., 2017; Laplante et al., 2015) while Myo51 contributes to ring formation and constriction (Tang et al., 2016).

Interestingly, we find that the C-termini of three F-BAR proteins (Cdc15, Imp2, and Rga7) extend away from their membrane-bound F-BAR domains into the intermediate layer of the ring (lines, Figure 2B). These proteins contain central regions predicted to be unstructured, which may permit the SH3 domains of Cdc15 and Imp2 to extend inward ~100 nm distances and connect with their multiple interactors also found at this layer including Pxl1, Spa2, Fic1, Cyk3 (Figure 2), and Rgf3 (Figure 3) (Roberts-Galbraith et al., 2010; Ren et al., 2015; Roberts-Galbraith et al., 2009).

Figure 3 with 2 supplements see all
Nanoscale organization of contractile ring signaling components.

(A) Representative fPALM images of signaling contractile ring components. Scale, 100 nm. Particles are visualized as normalized Gaussians with standard deviation = localization uncertainty. Dashed lines indicate plasma membrane edges. (B) Distance from the plasma membrane (dcenter) of signaling contractile ring components. Box plots depict 1st and 3rd quartiles and median; Whiskers, minimum and maximum; Notches, 95% confidence intervals.

https://doi.org/10.7554/eLife.28865.007

At the farthest distances from the membrane (160–350 nm), we find F-actin, the motor domains of the type II and type V myosins Myo2, Myp2, and Myo51, as well as Fim1, an F-actin crosslinker (Figure 2). Since actin is not functional as a fluorescent fusion in S. pombe (Wu and Pollard, 2005), we imaged F-actin with two probes: LifeAct-mMaple3 and Phalloidin-Alexa488. Measurements using these two methods were consistent, placing F-actin’s dcenter at 219 ± 64 (LifeAct-mMaple3) or 228 ± 75 nm (Phalloidin; Figure 2—figure supplement 1B) from the membrane. The N-terminal head domains of Myo2, Myp2, and Myo51 are level with F-actin, while the light chain Rlc1 is ~30 nm closer to the membrane, in accord with its binding to the neck domains of myosin motors that are immediately proximal to the head domains (Naqvi et al., 2000). Two actin crosslinking proteins are present in the contractile ring, fimbrin Fim1 and α-actinin Ain1 (Wu et al., 2001), with Ain1 being the predominant cytokinetic crosslinker (Wu et al., 2001; Li et al., 2016). We were unable to produce a functional mMaple3-Ain1, but anticipate it would localize in the same region as F-actin, similar to Fim1.

Spatial distribution of contractile ring signaling components

The contractile ring contains a plethora of signaling proteins that influence the behavior of its structural elements including kinases, phosphatases, GTPase exchange factors (GEFs) and GTPase activating proteins (GAPs) (Wood et al., 2012; Bohnert and Gould, 2011). We examined the spatial distribution of 10 such components that are present in fully formed unconstricted rings (Figure 3A–B, Figure 3—figure supplement 12). The kinases Orb2/Pak1/Shk1, Sid2, and Pck1 are relatively close to the membrane, on average <110 nm. In focal adhesions, regulators such as Focal Adhesion Kinase (FAK) are also found at an intermediate layer between the membrane and F-actin (Kanchanawong et al., 2010). In contrast, the N-terminus of the kinase Pom1 was found farther away from the membrane (149 nm) despite the presence of a nearby membrane-binding motif (Hachet et al., 2011). It may be that Pom1 localization to the contractile ring utilizes a different mechanism than when targeted to the cell tip cortex (Hachet et al., 2011). It is noteworthy that Pom1 contains multiple PxxP motifs and may bind the SH3 network established by Cdc15 and Imp2. The RhoGEF Rgf3, a known partner of Cdc15 and Imp2 SH3 domains (Ren et al., 2015), is also present at an intermediate height.

Phosphatases Clp1 (Cdc14-related) and Ppb1 (Calcineurin), RhoGAP Rga2, and RhoGEF Gef2 were also found in an intermediate stratum. Gef2 is in complex with Nod1 (Zhu et al., 2013) and is found at a similar elevation to its binding partner, 145 nm and 140 nm, respectively. Rad24, a 14-3-3 protein, while not technically a signaling molecule, binds to many phosphoproteins to control signaling pathways (Mackintosh, 2004; Aitken, 2006). We find the majority of Rad24 high in elevation, similar to F-actin, though it also localizes throughout the cytoplasm.

Distinguishing contractile ring layers with dual fluorophores

To corroborate the relative positions of proteins determined individually, we next imaged multiple strains containing the mMaple3 tag on two separate proteins. First, we imaged two proteins in the same layer: mMaple3-Cdc15 and Mid1-mMaple3 in the proximal layer or Cdc15-mMaple3 and Imp2-mMaple3 in the intermediate layer. Confirming their overlapping localization, distributions relative to the plasma membrane remained single narrow peaks (Figure 4A–B, Figure 4—figure supplement 1). Next, we combined tagged proteins from adjacent layers: mMaple3-Cdc15 and Imp2-mMaple3, as well as mMaple3-Cdc15-mMaple3 with tags on its two termini in proximal and intermediate layers. In these cases, a single peak was still observed, but the peak was significantly wider, matching predictions (green dashed lines, Figure 4A). Finally, we examined strains containing proteins in proximal and distal layers: mMaple3-Cdc15 and mMaple3-Myo2, along with mMaple3-Cdc15 and LifeAct-mMaple3. In these cases, either an even wider peak or two resolvable peaks were observed, confirming the spatial separation of proximal and distal components (Figure 4A–B).

Figure 4 with 1 supplement see all
Distinguishing contractile ring layers with pairs of mMaple3-tagged proteins.

(A) Comparison of single mMaple3-tagged proteins in similar or disparate localizations in the contractile ring. Orange and blue lines represent average localization of individual proteins from Figures 23, while dashed green lines represent the theoretical localization of two individual proteins imaged simultaneously. Grey histograms depict actual localization data from the indicated dual mMaple3-tagged strains. (B) Comparison of σvert (localization spread) parameters between single and double tagged strains reveals wider peaks when two proteins are separated spatially. One way ANOVA tests were performed between the indicated samples. *p<0.05; **p<0.01; ****p<0.0001.

https://doi.org/10.7554/eLife.28865.010

Distinct layers around F-BAR protein scaffolds

The fPALM analysis of Cdc15 and Imp2 F-BAR proteins indicated that their C-termini were positioned in a different stratum of the contractile ring than their N-terminal membrane-bound F-BAR domains. To verify the existence of distinct layers of F-BAR- and SH3-scaffolded components with a different approach, we placed a mCherry fluorophore on either the N-terminus (near the F-BAR domain) or C-terminus (near the SH3 domain) of Cdc15 in combination with GFP- or mNeonGreen (mNG)-conjugated ring components and analyzed each strain by fluorescent resonance energy transfer (FRET) (Figure 5A–B, Figure 5—figure supplement 1A).

Figure 5 with 1 supplement see all
FRET confirms distinct layers around F-BAR proteins in the contractile ring.

(A–B) Quantification of FRET between mCherry-Cdc15 (A) or Cdc15-mCherry (B) and GFP- or mNG-conjugated contractile ring components. See also Figure 5—figure supplement 1 for direct mCherry-Cdc15:Cdc15-mCherry comparisons. % FRET is the increase in GFP donor signal following mCherry acceptor photobleaching. One way ANOVA tests were performed for each strain against an Rlc1-GFP negative control. *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001. (C–D) Cdc12-mNG and Fic1-GFP FRET signals at the contractile ring are specific to mCherry-Cdc15 (C) or Cdc15-mCherry (D). Scale = 4 µm.

https://doi.org/10.7554/eLife.28865.012

Supporting the hypothesis that Cdc15 and probably other F-BARs scaffold distinct layers of proteins in the contractile ring, we detected specific FRET signals between mCherry-Cdc15 and membrane-bound Acyl-GFP, GFP-Imp2, GFP-Rga7, and both mNG-Cdc12 and Cdc12-mNG (Figure 5A). A strong FRET signal between mCherry-Cdc15 and Acyl-GFP is consistent with the binding of Cdc15’s F-BAR to anionic phospholipids within the plasma membrane (Roberts-Galbraith et al., 2010; McDonald et al., 2015). Furthermore, the FRET signal of mCherry-Cdc15 with two other F-BAR domains (GFP-Imp2 and GFP-Rga7) indicates that all three cytokinesis F-BARs are packed in close proximity upon the membrane, despite the fact that they do not heterodimerize (Roberts-Galbraith et al., 2009; Martín-García et al., 2014). The FRET signal of mCherry-Cdc15 with both termini of Cdc12 is consistent with the membrane-proximal spatial distribution of both Cdc12 termini observed by fPALM analysis. Importantly, the FRET signals of membrane-proximal ring components with mCherry-Cdc15 were specific, as these particular GFP- and mNG-tagged components did not display significant FRET with Cdc15 tagged at its C-terminus with mCherry (Figure 5C–D).

Cdc15-mCherry, on the other hand, showed significant FRET signals with Imp2-GFP, Rga7-GFP, and five known binding partners of its SH3 domain: Fic1-GFP, Pxl1-GFP, GFP-Spa2, GFP-Rgf3, and GFP-Cyk3 (Figure 5B) (Roberts-Galbraith et al., 2010; Ren et al., 2015; Roberts-Galbraith et al., 2009). The detection of FRET between Cdc15-mCherry and the C-terminal domains of Imp2 and Rga7 supports the spatial distribution of these molecules proposed from our fPALM analysis and indicates these domains, like their F-BAR domains, are positioned close enough together to support FRET. The detection of FRET signals between known Cdc15 SH3 domain-binding partners and Cdc15-mCherry was expected. Moreover, the specificity of the signal for Cdc15-mCherry over mCherry-Cdc15 (Figure 5—figure supplement 1B) supports the extended orientation of Cdc15 proposed from our fPALM analysis. We detected no significant FRET signal between mCherry-labeled Cdc15 and multiple additional contractile ring proteins (Figure 5—figure supplement 1C), again consistent with the CR being composed of distinct strata.

Circumferential clustering capability of contractile ring scaffolds

Our fPALM analysis of the molecular architecture of the contractile ring described above utilized a ‘side-view’, designed to easily measure distances from the membrane. With this view, information about how protein components organize along the circumferential axis of the ring is missing. Therefore, we performed additional fPALM imaging at the top plane of cells (Figure 6A) to capture a circumferential view of the contractile ring and assess the organization of a subset of contractile ring components. To quantify the level of clustering, we calculated the Inverse Difference Moment (IDM) of fPALM images along the ring circumferential axis (Haralick et al., 1973). IDM is a measure of local homogeneity; values close to 1 have homogenous local intensities (uniform) and those further from 1 have heterogeneous local intensities (clustered). We simulated a completely uniform and a clustered distribution for comparison (top two panels, Figure 6B–D).

Lateral organization of components in the contractile ring.

(A) Schematic of fPALM setup with Z focus at the top of cells. (B) Representative fPALM image of the top of contractile rings with the indicated components labeled. Simulated clustered or smooth ring examples are included for comparison. (C) Linescans through the center of images in (A). (C) Quantification of the Inverse Difference Moment (IDM) of a grey level co-occurance matrix along the contractile ring’s circumferential axis. IDM is a measure of local homogeneity, see Materials and methods. Lines indicate a one way ANOVA performed against mMaple3-Cdc15; ****p<0.0001.

https://doi.org/10.7554/eLife.28865.014

Surprisingly, we detected multiple distinct patterns of circumferential spatial distribution among the subset of contractile ring proteins examined. Some appeared uniform in distribution while others adopted variable clustered organizations; even different domains within a single molecule were distributed differently (Figure 6B–D). Distally, F-actin, detected by LifeAct-mMaple3, was distributed uniformly along the ring circumference, consistent with electron micrographs of the ring (Kamasaki et al., 2007). Also in the distal layer, the N-terminus of Myo2 was uniformly distributed. Within the intermediate layer, the C-termini of Myp2, Fic1, and Cdc15 were evenly distributed. In contrast, the domains of Cdc15 and Myo2 close to the membrane appeared to form clusters spaced at regular intervals along the circumferential axis with diameters of 71 ± 17 and 104 ± 26 nm, respectively. Also in the proximal layer, Rng2 and Mid1 were present in larger, unevenly spaced clusters 201 ± 49 and 200 ± 44 nm in diameter, respectively. Though these clustered components all reside in the proximal layer, not all proximal components were clustered. Spn3 displayed a uniform distribution which might be expected for dense septin filament bundles that are not entirely in register (Oh and Bi, 2011) and because septins are never present in early cytokinesis nodes (Tasto et al., 2003; Berlin et al., 2003).

Cdc15’s N-terminal F-BAR forms linear oligomers (McDonald et al., 2015); to test if oligomerization was responsible for organization of the observed larger scale clusters, we imaged a mMaple3-Cdc15(E30K E152K) oligomerization mutant. Indeed the oligomerization mutant prevented cluster formation. Interestingly, the oligomerization mutation also resulted in a widened mMaple3-Cdc15 ring thickness, suggesting oligomerization may also help concentrate Cdc15 in the ring. In contrast, the deletion of mid1 did not prevent Cdc15 cluster formation.

Discussion

Using fPALM super-resolution imaging, we have determined the spatial organization of 29 central components of the contractile ring relative to the membrane. We found that these components organize into distinct connected layers rising from the plasma membrane and extending up to 350 nm towards the interior of the cell (Figure 7 and Video 1). We also discovered that certain components adopt distinct spatial distributions along the circumference of the contractile ring ranging from uniform to large, irregularly sized and spaced clusters. Our measurements establish a strong foundation for a comprehensive understanding of the molecular architecture and function of a eukaryotic contractile ring.

Schematic model of the S. pombe contractile ring molecular architecture.

Depicted protein positions were calculated experimentally and are color-coded in the Z dimension. Note that this model does not incorporate stoichiometry. See also Video 1.

https://doi.org/10.7554/eLife.28865.015
Video 1
Animated schematic model of the S. pombe contractile ring molecular architecture.

Depicted protein positions were calculated experimentally and are color-coded in the Z dimension. Note that this model does not incorporate stoichiometry.

https://doi.org/10.7554/eLife.28865.016

Structural layers within the contractile ring

Our analysis of 29 components of the contractile ring by fPALM indicates the existence of distinct layers tethering F-actin to the plasma membrane, similar to two previously investigated membrane-tethered F-actin structures, focal adhesions (Kanchanawong et al., 2010) and cadherin junctions (Bertocchi et al., 2017).

Nearest to the membrane (0–80 nm), proteins with membrane-binding domains scaffold additional components through a network of protein-protein interactions. The anillin Mid1, tethered to the membrane via its C2 domain, binds the tail of the essential type II myosin, Myo2 (Motegi et al., 2004), as well as extending to bind IQGAP Rng2 (Padmanabhan et al., 2011), which in turn stabilizes the Myo2-Mid1 interaction (Laporte et al., 2011). The F-BAR domain of Cdc15, also bound directly to the membrane, scaffolds the formin Cdc12 (Willet et al., 2015b; Carnahan and Gould, 2003). The membrane-binding F-BAR domain of Rga7 may also have a ring binding partner, Rng10 (Liu et al., 2016), suggesting that F-BAR domains may scaffold additional ring components, linking their partners indirectly but closely to the membrane. It is not surprising that the integrity of this proximal layer of the contractile ring is critical for strong attachment to the plasma membrane. In fact, additional time is required to assemble the ring when Mid1 is missing (Huang et al., 2008b; Hachet and Simanis, 2008; Wu et al., 2003), while rings can slide and disassemble when Cdc15 abundance is reduced (McDonald et al., 2015; Arasada and Pollard, 2014). It is surprising that the ring remains associated with the plasma membrane in the majority of cells lacking Mid1 or fully functional Cdc15 or both (McDonald et al., 2015), indicating that other ring components important for establishing this tight plasma membrane-contractile ring linkage must exist, especially during constriction when Mid1 leaves the ring (Wu et al., 2003).

At 80–160 nm away from the membrane, we find the C-termini of Cdc15 and Imp2, which contain their SH3 domains. These domains are functionally redundant (Ren et al., 2015), but having one or the other is essential for cytokinesis and cell viability (Roberts-Galbraith et al., 2009), demonstrating the importance of an SH3-scaffolded network within the contractile ring. Multiple binding partners of Cdc15's and Imp2’s SH3 domains including Pxl1, Fic1, Spa2, Rgf3, and Cyk3 (Roberts-Galbraith et al., 2010; Ren et al., 2015; Roberts-Galbraith et al., 2009) concentrate in this intermediate zone. Cyk3 itself contains an SH3 domain, likely further strengthening this interaction network. Cdc15 and Imp2 SH3 domains bind a K/RxLPxΦP motif that is found in a number of additional contractile ring proteins (Ren et al., 2015) that may also be present in this zone. It is interesting to note that the functions of the majority of these proteins remains unclear, and that their functions are not strictly essential. However, given the multiple protein interaction domains present in these components and their spatial organization, we speculate that eliminating pairs or subsets of these elements would undermine the networking of this zone. This may result in weakening the linkage between the strata of the contractile ring ultimately leading to loss of ring integrity and cell viability, irrespective of any mechanistic contributions of these components to ring function. In support of this hypothesis, synthetic lethality has been observed in several cases when null alleles of these genes were combined (Ren et al., 2015; Pinar et al., 2008; Roberts-Galbraith et al., 2009; Liu et al., 2016; Chen et al., 2016). Indeed in the complete absence of Cdc15 and its SH3 interaction network, F-actin rings can still form but quickly disintegrate (Laporte et al., 2011), underscoring the stabilizing function of the intermediate zone.

Farthest from the membrane (160–350 nm), we find the center of F-actin localization. The elevated localization and ~200 nm thickness of F-actin matches the limited electron micrographs of S. pombe rings that are available (Kamasaki et al., 2007). It is notable that classical electron microscopy studies of animal cells have also measured the contractile ring as 250–500 nm thick vertically from the membrane (Schroeder, 1973). Interestingly, both termini of the formin Cdc12, which nucleates and elongates F-actin in the ring de novo, are significantly closer to the membrane. This localization suggests that F-actin may be nucleated and polymerized by Cdc12 close to the membrane, but filaments are pulled upward and incorporated into the main F-actin density by tension from myosin motors and actin crosslinkers. Indeed we found the head domains of all 3 contractile ring myosins and the actin crosslinker Fim1 at a high elevation apposed to F-actin. The essential myosin-II Myo2 is oriented with its tail close to the membrane and motor extended, an orientation also seen in precursor nodes (Laporte et al., 2011; Laplante et al., 2016). This implies a different organization than the canonical bipolar filaments that are observed in sarcomeres and mammalian cleavage furrows (Fenix et al., 2016; Henson et al., 2017; Pollard et al., 2017). Though bipolar filaments are not formed, an orientation with Myo2’s tail bound near the membrane would provide a critical anchor for force generation by its distal motor domain. Myo2 has additionally been implicated in ring formation independent of its motor activity (Palani et al., 2017), a function that may be contained within domains near the membrane and other early ring components. We find the tails of the non-essential myosins Myp2 and Myo51 at an intermediate layer. Myp2 and Myo51 are single-headed myosin motors (Tang et al., 2016; Bezanilla and Pollard, 2000); embedding their tail domains in the intermediate layer may also aid in force generation upon the distal F-actin. It is likely that other components such as α-actinin and tropomyosin that depend on F-actin for ring localization are also present in the most interior layer.

We also quantified the geometry of each ring component within the ring, both in terms of ring width and in terms of strata width, e.g. spread from the center of the ring and distance from the membrane. Relative to the membrane, ring components tended to organize into similarly ‘thick’ strata of ~150–200 nm. In terms of width within the ring, components organized in widths between 120–220 nm, with Fim1-mMaple3 being widest and mMaple3-Imp2 being thinnest, though in general widths did not significantly correlate with components’ distance from the membrane. Clearly, the membrane proximal layer is no wider than the actin layer or vice versa.

Signaling within the contractile ring

The nanoscale spatial organization of signaling components of the contractile ring has not been studied previously, though we estimate that there are ~10 protein kinases, ~3 phosphatases, and >20 GTPases and associated GAPs and GEFs in the contractile ring. Though some of these modulators have known binding partners that may help recruit them to the appropriate layer (e.g. the protein kinase Sid2 by Blt1 (Goss et al., 2014), the GTPase exchange factor Gef2 by Nod1 (Zhu et al., 2013), and the Rho1-exchange factor Rgf3 by Cdc15 and Imp2 [Ren et al., 2015]), it is also possible that these enzymes accumulate in distinct zones due to more transient interactions with their substrates. We did find that most localized in the correct areas for modification of their reported substrates. Sid2 is localized at a low elevation (88 nm), properly placed to access its substrates Cdc12 (Bohnert et al., 2013), Spa2 (Gupta et al., 2013), and Clp1 (Chen et al., 2008). The protein kinase Pom1 is in the intermediate layer (175 nm), overlapping with the localization of 3 known substrates: Imp2, Cdc15, and Rga7 (Kettenbach et al., 2015). Furthermore, the Clp1 phosphatase is present at a low-to-intermediate level (136 nm), near its scaffold Mid1 (Clifford et al., 2008) and known and predicted ring substrates Cdc15, Blt1, Rng2, Cyk3, Myp2, and Orb2 (Clifford et al., 2008; Chen et al., 2013).

Protein clusters in the contractile ring

When we examined a circumferential view of the ring, we found that components adopted a range of spatial organizations. Mid1 and Rng2 form irregularly spaced 200 nm clusters, which resemble the size and shape of cytokinesis precursor nodes previously measured with different techniques (conventional microscopy and live cell fPALM) (Wu et al., 2006; Laporte et al., 2011). The membrane-proximal termini of Myo2 and Cdc15 form smaller 70–100 nm diameter regularly-spaced clusters along the circumference of the ring. All four of these proteins are present in precursor nodes (Wu et al., 2006), which have recently been suggested to be maintained in the fully developed contractile ring (Laplante et al., 2016). However, the very different diameter and regularity of Cdc15 and Myo2 clusters from those of Mid1 and Rng2 argues that at the least, node structures change as the ring develops. This is perhaps to be expected since Mid1, responsible for setting up precursor nodes, dissociates from the ring as it constricts (Wu and Pollard, 2005). Additionally, the relative stoichiometry of certain components in the ring change during constriction: Myo2 is concentrated during constriction while Cdc15, Rng2, and Cdc12 remain at a constant concentration (Wu and Pollard, 2005; Wu et al., 2003). Therefore, reorganization of the membrane-proximal level within the ring must occur. Future two-color super-resolution imaging could determine if large clusters contain combinations of components that persist from precursor nodes as well as any additional novel clustering behavior in the fully formed ring.

Interestingly, we found components that are further away from the membrane are uniform in their distribution, even compared to their opposite termini. In particular, the F-actin signal from LifeAct-mMaple3 is uniform, in accord with electron microscopy studies of contractile ring F-actin which show a relatively even distribution in both S. pombe and animal cells (Kamasaki et al., 2007; Henson et al., 2017). In contrast to their membrane-proximal termini, the opposite termini of Myo2 and Cdc15 appear quite uniform. A long flexible region between Cdc15’s F-BAR and SH3 domains likely acts to eliminate any F-BAR-oligomerization-mediated clustering once the SH3 domain is reached. The even distribution of Myo2’s head domain suggests a generally even binding of actin throughout the ring. This distribution is in contrast to myosin-II ‘stacks’ formed in animal cells, where antiparallel myosin units are assembled into large-scale clusters parallel to F-actin filaments (Fenix et al., 2016; Henson et al., 2017).

Building the ring bottom up

Together with prior knowledge of when proteins are recruited to the ring, the spatial architecture of the contractile ring determined here suggests possible mechanics of ring formation. In fact, the different layers seen in our analysis generally correlate with the order of assembly of their components. Of the earliest ring components which are recruited to precursor nodes (Mid1, Rng2, Myo2, Cdc15, and Cdc12) (Wu et al., 2006; Laporte et al., 2011), 4/5 contain termini present within the proximal membrane layer (0–80 nm), and 2/5 contain direct membrane binding domains. The precise order of assembly of Rng2, Myo2, and Cdc12 does not appear to matter for ring formation (Tao et al., 2014), but it seems likely that the initial setup of the ring involves the construction of a membrane-anchored scaffold. F-actin is formed de novo at nodes and the ring by Cdc12 (Pelham and Chang, 2002) as well as incorporated from longitudinal actin cables (Huang et al., 2012). F-actin, though turning over quickly in the ring (Henson et al., 2017), may be principally held in its elevated location by the tension from the motor activity of myosins. Circumferential tension due to myosin may produce force on F-actin and other components toward the center of the cell, as in ‘purse-string’ models of ring function (Henson et al., 2017). Once a contiguous ring is formed, many additional components are recruited. Components that join the ring later are generally present within the intermediate layer. Many of these components, therefore, may ‘fill in’ between the membrane-bound proximal layer and F-actin to form a robust structure that withstands the forces of constriction, couples ring constriction with other events of mitosis, and connects the ring to cell wall formation that occurs coincidentally.

Materials and methods

Yeast methods and strain construction

Request a detailed protocol

S. pombe strains (Supplementary file 1) were grown in yeast extract (YE) media at 25°C, with the exception of pnmt1-Acyl-mMaple3 (Figure 1—figure supplement 1B) and nmt1-Acyl-GFP (Figure 5A) which were grown in Edinburgh minimal medium (EMM) lacking thiamine to induce expression (Basi et al., 1993). S. pombe transformations were performed with a lithium acetate method (Keeney and Boeke, 1994). mMaple3 (Wang et al., 2014) was cloned into a pFA6a vector at AscI and PacI sites. C-terminal fluorophore endogenous fusion strains were created by transforming a pFA6a integration cassette amplified with gene specific primers for insertion at the 3’ end of open reading frames (Bähler et al., 1998). N-terminal mMaple3 or GFP fusion proteins were created with two methods. For nonessential genes (Cyk3, Imp2, Mid1, Pom1, Pxl1, Rga7, and Spa2), constructs containing (1) a 500 bp 5’ flank, (2) mMaple3 or GFP with a GGGGSGGGGSG C-terminal linker, (3) the coding sequence, and (4) a 500 bp 3’ flank were assembled into a pSK or pIRT2 vector (Gibson et al., 2009). These cassettes were amplified from their 5’ and 3’ flanks, transformed into ura4 + deletion strains of the targeted gene, and selected for ura4 + loss and integration by resistance to 5-FOA. For essential genes (Cdc12, Cdc15, Myo2, Rng2, and Rgf3), constructs containing (1) a 500 bp 5’ flank, (2) mMaple3 or GFP with a GGGGSGGGGSG C-terminal flexible linker, (3) the coding sequence, (4) a KanR cassette from pFA6a, and (5) a 500 bp 3’ flank were assembled into a pSK vector. These cassettes were digested before and after their 5’ and 3’ flanks and transformed into wildtype S. pombe. G418 resistant transformants were screened for correct integration by PCR. pAct-LifeAct-mMaple3 was constructed to be identical to the previously described pAct-LifeAct-GFP (Huang et al., 2012). A 1 kb promotor from act1, the LifeAct peptide (MGVADLIKKFESISKE) (Riedl et al., 2008), a short GGPGG linker, and mMaple3 were assembled into a pJK148 plasmid which was subsequently digested by NruI and integrated into the leu1-32 locus (Keeney and Boeke, 1994). Acyl-mMaple3 was constructed with the N-terminal acyl sequence from Gpa1 (MGCMSSKYADTSGGEV) and mMaple3 in a pREP1 vector (Maundrell, 1993). All strains were confirmed by PCR and sequencing.

Sample preparation for fPALM imaging

Request a detailed protocol

To enrich for cells undergoing cytokinesis for imaging, newly separated short cells containing mMaple3-tagged proteins were isolated from a 7–30% lactose gradient, grown for 80 min in YE, labeled with 0.5 mM mCling-ATTO647N (Revelo et al., 2014) (Synaptic Systems, Goettingen, Germany) for 10 min, and fixed. The resulting fixed sample was enriched for cells containing fully formed but unconstricted contractile rings. Cells were fixed with 3.7% paraformaldehyde directly in media for 20 min and subsequently washed 6X with PBS. Staining of actin with phalloidin was performed on fixed cells with 3.3 µM Phalloidin-Alexa488 in PBS + 0.1% NP-40 for 30 min, followed by 3 washes with PBS. Fixed and washed cells were resuspended in a small volume of PBS containing 1:100 diluted 80 nm gold particles (Microspheres-Nanospheres, Cold Spring, NY) previously sonicated for 1 hr to disperse clusters. The cell and gold particle suspension was mounted on a 3% agarose pad to prevent cell drifting during imaging.

fPALM imaging

Request a detailed protocol

fPALM imaging (Figures 14 and 6) was performed on a Nikon dSTORM 4.0 system which included a Nikon Eclipse Ti microscope, 405, 488, 561, and 647 nm solid state lasers, a Hamamatsu ORCA-Flash4.0 camera, and a 100X CFI HP Apochromat TIRF 1.49NA objective with a 1.5X tubelens (resulting in 110 nm pixels). Imaging was performed at a 0° laser angle (‘straight through’) focused in Z at the center of cells for Figures 14 and the top of cells for Figure 6. Z drift was minimized using a Nikon Perfect Focus system. The mMaple3 channel was imaged with simultaneous 0.5% 405 nm activation and 7.5% 561 nm excitation lasers, filtered through a polychroic mirror (ZT405/488/561/647rpc, Chroma, Bellows Falls, VT) and emission filter (ET585/65 m, Chroma), and captured with 30 ms exposures over 10–20 k frames. The ATTO647N channel was imaged with simultaneous 0.2% 405 nm activation and 2% 647 nm excitation lasers, filtered through a polychroic mirror (ZT405/488/561/647rpc, Chroma) and emission filter (ET705/75 m, Chroma), and captured with 10 ms exposures over 15 k frames. The Alexa488 channel for Phalloidin-Alexa488 was imaged with simultaneous 0.2% 405 nm activation and 2% 488 nm excitation lasers, filtered through a polychroic mirror (ZT405/488/561/647rpc, Chroma) and emission filter (ET525/50 m, Chroma), and captured with 10 ms exposures over 15 k frames. Laser powers and exposure times were optimized for single photoactivated localizations per ring in each frame.

fPALM analysis

Request a detailed protocol

fPALM images were analyzed using the ImageJ plugin ThunderSTORM (Ovesný et al., 2014). Images were pre-filtered with a wavelet B-spline filter and molecules were approximately localized with the 8-neighborhood local maximum method. Molecules were sub-pixelly localized using the elliptical Gaussian weighted least squares method to identify precise lateral and Z positions. A Z-calibration was performed with 0.1 µm Tetraspek beads (ThermoFisher) for accurate Z positioning. Axial drift, though minimal with this microscope setup, was corrected in post-processing by tracking fiducial gold bead markers as well as cross-correlation analysis. fPALM images are visualized with localizations as normalized Gaussians, where each peak’s standard deviation (FWHM) equals its localization uncertainty (Betzig et al., 2006).

Processed and aligned 2-color images were restricted to a 500 nm plane through the center of the cells. Fully formed, unconstricted rings were identified visually through the following criteria: (1) a lack of precursor nodes, (2) a fully formed contiguous ring perpendicular to the long cell axis, and (3) a lack of membrane and septum ingression detected in the mCling channel. These criteria place all analyzed rings in a short window of 15–20 min between ring formation and constriction (Wu et al., 2003). The edge of the plasma membrane was identified in an unbiased manner using an automated threshold method to determine where the signal from the mCling-ATTO647N channel drops to 5% of its maximum plasma membrane intensity (Figure 1—figure supplement 1B). The distance of each individual mMaple3 particle in the contractile ring to this edge was calculated. These data were fit with a Gaussian curve and the distance from the membrane (dcenter) and a vertical width parameter (σvert or FWHM) were determined using R (Figure 1E), as performed previously in a study of focal adhesion proteins (Kanchanawong et al., 2010). A horizontal width parameter (σwidth) was also calculated to describe the FWHM of ring localizations parallel to the plasma membrane. dcenter, σvert, and σwidth values from multiple rings (see Figure 1—figure supplement 1 and Figure 2—figure supplement 1 for n values and other statistics) were plotted using ggplot2 in R.

The contractile ring horizontal width dimension reported in the text (182 ± 26 nm) represents an average of all components’ σwidth. The contractile ring maximum extension from the plasma membrane figure reported in the text (293 ± 64 nm) represents LifeAct-mMaple3’s (the most distal component) dcenter + ½(σvert), approximating the ring’s extension into the cytoplasm.

Analysis of local homogeneity in Figure 6 was performed with downconverted 8-bit images using the ‘Texture Analysis’ ImageJ plugin (https://imagej.nih.gov/ij/plugins/texture.html). This plugin computes a grey-level co-occurrence matrix for all the pixels in the ROI of a contractile ring p(i, j), and calculates multiple of Haralick’s textural features (Haralick et al., 1973) from this matrix. We utilized the Inverse Difference Moment (IDM), a measure of local homogeneity:

i=1Ngj=1Ng11+i-j2p(i,j)

where Ng is the maximum grey value and p(i, j) is the grey level co-occurrence matrix. This analysis was performed specifically in the direction of the ring circumferential axis.

FRET imaging

Request a detailed protocol

FRET imaging (Figure 5) was performed on a Perkin Elmer Ultraview Vox spinning disk system equipped with a Zeiss Axio Observer microscope, 488 and 561 nm solid state lasers with a PhotoKinesis bleaching module, a Yokogawa CSU-X1 spinning disk, a 63X C-Apochromat objective, and a Hamamatsu ImageEM C9100-13 EMCCD camera. An acceptor photobleaching method was employed for FRET imaging in live cells. The mCherry-Cdc15 or Cdc15-mCherry fluorophores were bleached throughout the cells with 100% 561 nm laser power for 20 cycles. 10 frames of single Z slices in the donor GFP channel were acquired pre- and post-bleach. FRET percentages were calculated by first correcting for background and photobleaching over the 10 pre- and post-bleach frames, and subsequently calculating the percentage increase in GFP donor fluorescence at contractile ring ROIs of a consistent size. Statistical tests in Figure 5A–B and Figure 5—figure supplement 1B and S5B are ANOVA tests versus Rlc1-GFP a negative control which does not FRET with either mCherry-Cdc15 or Cdc15-mCherry, with uncorrected p values reported in Figure 5—figure supplement 1A–B. Statistical tests in Figure 5—figure supplement 1C were Fisher’s exact tests between the mCherry-Cdc15 and Cdc15-mCherry conditions.

References

  1. 1
  2. 2
  3. 3
  4. 4
    Isolation and characterization of new fission yeast cytokinesis mutants
    1. MK Balasubramanian
    2. D McCollum
    3. L Chang
    4. KC Wong
    5. NI Naqvi
    6. X He
    7. S Sazer
    8. KL Gould
    (1998)
    Genetics 149:1265–1275.
  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
    Isolation and characterization of fission yeast mutants defective in the assembly and placement of the contractile actin ring
    1. F Chang
    2. A Woollard
    3. P Nurse
    (1996)
    Journal of Cell Science 109:131–142.
  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
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
    Efficient targeted integration at leu1-32 and ura4-294 in Schizosaccharomyces pombe
    1. JB Keeney
    2. JD Boeke
    (1994)
    Genetics 136:849–856.
  43. 43
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
  54. 54
  55. 55
  56. 56
  57. 57
  58. 58
  59. 59
  60. 60
  61. 61
  62. 62
  63. 63
  64. 64
  65. 65
  66. 66
  67. 67
  68. 68
  69. 69
  70. 70
  71. 71
  72. 72
  73. 73
  74. 74
  75. 75
  76. 76
  77. 77
  78. 78
  79. 79
  80. 80
  81. 81
  82. 82
  83. 83
  84. 84
  85. 85
  86. 86
  87. 87
  88. 88
  89. 89
  90. 90
  91. 91
    Two type V myosins with non-overlapping functions in the fission yeast Schizosaccharomyces pombe: Myo52 is concerned with growth polarity and cytokinesis, Myo51 is a component of the cytokinetic actin ring
    1. TZ Win
    2. Y Gachet
    3. DP Mulvihill
    4. KM May
    5. JS Hyams
    (2001)
    Journal of Cell Science 114:69–79.
  92. 92
  93. 93
  94. 94
  95. 95
  96. 96
  97. 97
  98. 98
  99. 99

Decision letter

  1. Mohan K Balasubramanian
    Reviewing Editor; University of Warwick, United Kingdom

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 "Nanoscale architecture of the Schizosaccharomyces pombeS. pombe contractile ring" for consideration by eLife. Your article has been reviewed by three peer reviewers, one of whom is a member of our Board of Reviewing Editors and the evaluation has been overseen by Vivek Malhotra as the Senior Editor. The reviewers have opted to remain anonymous.

The referees were impressed with the use of super-resolution approach to understand organization of the cytokinetic apparatus in fission yeast, which has become an excellent system to study cytokinesis. The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission. We hope you will be able to submit the revised version within two months.

We request that you address the following issues (essential revisions), but you might also want to consider other points in the referee comments (nearly all of which are reasonable) in preparing your revisions.

Essential revisions:

First, the accuracy of your estimate of the location of cell membranes was called into question in points 3 and 4 of referee 1 and point 4 of referee 2. Please address these with alternative approaches suggested by the referees, or other approaches that may be available to you.

Second, and linked to the above point, it will be important to generate some double-tagged strains (e.g. Mid–mMaple3 Cdc15–mMaple3), and then to test for two separate gaussian peaks. If they showed 3-4 examples of double peaks for combined tags, then this would address these concerns (point 5 of referee 1).

Finally, please also address why the cell cycle stages you describe are reliable considering the caveats in the synchrony used.

Reviewer #1:

By using the super resolution localization microscopy technique fPALM, McDonald et al., measured the distances between cell membrane and various cytokinetic ring proteins in the fixed S. pombe cells. The authors further validated the proximity of the F–BAR protein Cdc15 to its interacting proteins using FRET technique. In addition, the authors also measured the distances of several regulatory proteins such as kinases and GTPases that participate in cytokinesis relative to cell membrane. Together, the authors provided a 2–dimensional architecture of the cytokinetic ring at the nanometer scale.

The cytokinetic ring is a complex contractile machinery, and how the ring proteins are organized to collectively executing cytokinesis is an important but not yet well studied question in the field. The question addressed by McDonald et al., is important towards the understanding of cytokinetic ring function, and fits the scope of eLife. However, due to the technical caveats particularly the ambiguity of the cytokinesis stages and the labelling of cell membrane (major comments below), I am not very convinced by the accuracy of their findings. Since the relative molecular distances of ring proteins are at the nanometer scale, a good definition of cell membrane and the cytokinesis stages is essential to this study. Thus, these concerns should be addressed before further consideration for eLife.

1) The authors mentioned that they collected synchronized cells for their fPALM experiments (subsection “Sample preparation for fPALM imaging”). However, there is no supporting evidence for a good cell synchronization. The cells shown in Figure 1B do not seem to be at the same cell cycle stages (cells with different ring widths could be observed). Since the localizations of ring proteins may vary in early assembled rings, in matured rings, or in rings ready for constriction, a reliable indicator of cytokinesis stages will be required when performing the fPALM study. It would be useful to include mitotic markers such as SPB distance or spindle length for more precise indexing of cytokinesis progression.

2) The authors only map the ring protein localization at one single time point (+90 minutes after synchronization). To substantiate their findings, the relative localizations of several key proteins such as Cdc15, F-actin, Myo2 at different stages of cytokinesis (ready for constriction vs. constricting) should be profiled at higher temporal resolution.

3) A reliable cell membrane marker is critical to the accuracy of measurements in this study. It is not entirely clear if mCling is an appropriate marker for cell membrane in S. pombe as the dye binds to cell wall strongly (>100 nm thick structure in Figure 1—figure supplement 1B). How the authors be certain that the cell membrane was indeed labelled by mCling? The punctate staining of mCling in Figure 1C also suggests that not all membrane/cell wall could be stained. The different sizes of the mCling 'blobs' in Figure 2A and 3A suggest uneven binding to cell wall, which may result in the bias and inaccurate cell membrane edge determination used by authors.

4) It is unclear how the internal edge of cell membrane in the study is determined. Was it mathematically defined for every single measurement or was it 'visually' defined? This is especially important as precision of fPALM nanoscale mapping of ring proteins strongly dependent on precision of localization of plasma membrane. A detailed description of how this was done is required in the Materials and methods section.

5) To substantiate the relative distances of ring proteins, the authors could express and resolve two mMapple3 tagged proteins such as Mid1–mMapple3/mMapple3–Cdc15 (proteins close to cell membrane) and mMapple3–Myo2/LifeAct–mMapple3 (actin rich domain) with fPALM in the same cells. Ideally, two Gaussian peaks could be detected compared to cells expressing only one kind of tagged proteins.

Reviewer #2:

Constriction of the plasma membrane during cytokinesis depends on acto-myosin based contractile ring at the cell centre. Although we know the components of CR, lack of information regarding the nature of molecular arrangement and interaction has severely limited our understanding of the process. In S.pombe, a conserved actomyosin ring of 400nm thickness composed of 130 proteins drives constriction. Such molecular complexity is beyond the scope of diffraction limited optics. This paper (along with a recent work from Pollard lab) attempts at elucidating the structure of actomyosin ring using super-resolution microscopy. The authors suggest a stratified structure of membrane binding components proximal to the membrane and force generators distal to it. While both the scope of the study and its methodical execution deserves much credit, the manuscript suffers from lack of clarity and adequate discussion.

1) The authors show that the components of the Contractile rings are arranged in three distinct zones relative to plasma membrane in the contractile ring prior to ring constriction. My major concern is with lack of temporal information in this manuscript. Is this organization (protein-map) is altered during constriction? Do distribution of some components of Proximal zone change cluster-like to homogenous as the rings constrict or do they retain their conformation?

2) The authors rightly argue that establishment of ultrastructural detail of the actomyosin ring is important to understand their function. However, they do not discuss the important implications of their observation. Myo2 tail is localised to membrane proximal zone where as its head domain is more distal and uniform. Does this mean that type 2 myosin is not arranged as a bipolar filament in contractile ring? Does this distribution change when the rings undergo constriction?

3 "At similar intermediate distances from the membrane, we find the tail domains of two other myosins: the non-essential myosin–II, Myp2, as well as a myosin–V, Myo51."

The manuscript doesn't provide any information on the head of these two motors. Given the recent implications of myosin motors in force generation and mutually exclusive localisation in the contractile ring, I think it may be worthwhile to investigate relative positioning of myo2 and myp2 (Laplante et al., 2015).

4) fPALM strategy for determining the spatial organization of contractile ring components

"The plasma membrane was defined as the edge of the mCling–ATTO647N signal, and the distance from this line was calculated for each individual localization in a contractile ring side-view "spot".

I am unable to fathom the accuracy of the plasma membrane position. If a line was drawn at the edge of the signal of plasma membrane manually it leaves much scope for errors. Signals such as the ones in Figure 1C and Figure 1—figure supplement 1B, clearly indicates a diffuse signal near the edge. How did the authors trace such a signal? Was it based on any particular threshold? Given, that this is the point of reference for the entire study, I hope the authors illustrate exactly how they performed their analysis.

5) "We imaged synchronized cells that contained fully formed contractile rings (e.g. no precursor nodes remaining) that had not begun constriction (scored by membrane ingression and septum formation)".

The synchronisation achieved in lactose gradient selection of G1 cells, is not mentioned in the text. Was there any particular reason for G1 synchronisation and not G2?

In the absence of a cell-cycle marker and given the maturation period of S.pombe rings, it seems the authors imaged rings at least 20 minutes apart in their maturation. Certainly, this is enough scope for molecular rearrangements. Since incorporating a third fluorophore is challenging with the current set up, could the authors at least investigate simultaneous position and orientation of two ring components with each other and suggest they do not differ significantly from single labelling?

6) "Therefore, we performed additional fPALM imaging at the top plane of cells (Figure 5A) to capture a lateral view of the contractile ring and assess the organization of a subset of contractile ring components".

I am unable to follow how the authors determine "top plane of cells". In contrast to the medial plane where the contractile ring serves as a reliable landmark with two points of maximal separation, "top of the cell" has no such landmarks. How does the authors reliably find the top focal plane? It would be useful if the authors can give an approximation of the circumferential length the 500nm Z section encompasses.

Reviewer #3:

In this study, McDonald et al., used fPALM to map the organization of proteins within the cytokinetic ring. Specifically, they measured the average distance of 29 ring proteins from the plasma membrane, revealing vertical zones of enrichment that make sense given the known functions and interactions of these proteins. For several factors, these positions are further confirmed by FRET. This work follows two previous studies that mapped a more limited set of proteins in cytokinetic ring precursors by SHREC (Laporte, 2011) and fPALM (Laplante, 2016). This McDonald et al., study is an impressive body of work that provides the most detailed account of protein organization in the cytokinetic ring, at least to my knowledge. It will be a reference point for future mechanistic studies. My main concern is that the work is entirely descriptive, it is strictly a localization study. Ideally, one would like to see the function of these localization results tested – what aspects of this vertical organization are important for cytokinesis? As such, it is more of a "resource" than a "research article." Nonetheless, it represents a major step forward for the field, and is likely to draw broad interest. I have a couple of comments that the authors might address to strengthen the manuscript:

Figure 1: Are all proteins distributed in the same width within the ring? The authors measure distance from the membrane and vertical distribution (i.e. spread in distance away from the membrane) for each protein. However, it would be interesting to know if all proteins are distributed along the same horizontal width or not. Presumably, these data are readily available from the images, and the authors state in the text an overall ring width of 181nm, but I did not see how this number was calculated. Measurement for each component would further refine the molecular architecture of the ring by adding additional confinement for each protein.

Figure 5: The data suggest that ring proteins are clustered near the membrane, but not clustered in the more interior strata. Particularly striking are examples like Myo2 and Cdc15, where the cortical domains are clustered, but their 'inner' domains are unclustered. As the authors know, the Pollard lab published fPALM data showing that both termini of these same constructs are clustered in the ring. In fact, all 6 constructs (4 proteins, 2 of which were tagged at each terminus) examined in the Pollard study were clustered. One difference between the two studies is that the Pollard lab imaged live cells, while the McDonald study imaged fixed cells, raising the question of whether fixation affected clustering (there are other difference and possibilities as well). I wonder if the authors could address this discrepancy by imaging lateral organization in live cells. They could also consider imaging more examples of cortical versus interior proteins, to support further this general trend of clustered at the membrane but diffuse at the interior. Finally, measurements of horizontal spread in middle focal planes (previous comment) might show tight clusters at the cortex with broad spread at the interior.

https://doi.org/10.7554/eLife.28865.019

Author response

Essential revisions:

First, the accuracy of your estimate of the location of cell membranes was called into question in points 3 and 4 of referee 1 and point 4 of referee 2. Please address these with alternative approaches suggested by the referees, or other approaches that may be available to you.

We have addressed this point (see detailed explanations below) by confirming mCling is a genuine membrane marker, and implementing an unbiased automated method to define the edge of the membrane. Importantly, using this method has not changed any of our major conclusions.

Second, and linked to the above point, it will be important to generate some double-tagged strains (e.g. Mid–mMaple3 Cdc15–mMaple3), and then to test for two separate gaussian peaks. If they showed 3-4 examples of double peaks for combined tags, then this would address these concerns (point 5 of referee 1).

We have now imaged and analyzed multiple double tagged strains with different combinations of proximal, intermediate, and distal components. Confirming our individual measurements, we find double tags in the same layer form a single peak, while double tags in different layers form either double peaks or extra-wide single peaks, depending on their distance apart.

These data have been included in a new Figure 4 and Figure 4—figure supplement 1.

Finally, please also address why the cell cycle stages you describe are reliable considering the caveats in the synchrony used.

We unfortunately misled the reviewers regarding the purpose of the lactose gradient. In fact, it was utilized simply to enrich for cells in a single cell cycle stage so that more cells could be found at the appropriate stage in each field of view. Fully formed, unconstricted rings were subsequently identified visually based on very strict criteria. In other words, only that fraction of rings corresponding to fully formed but unconstructed ring in any single field of view were analyzed. We have clarified the methodology in the manuscript and added examples to illustrate how appropriate rings were identified (Figure 1—figure supplement 1C). See our response to reviewer 1 point 1 and reviewer 2 point 5 below for details.

Reviewer #1:

By using the super resolution localization microscopy technique fPALM, McDonald et al., measured the distances between cell membrane and various cytokinetic ring proteins in the fixed S. pombe cells. The authors further validated the proximity of the F–BAR protein Cdc15 to its interacting proteins using FRET technique. In addition, the authors also measured the distances of several regulatory proteins such as kinases and GTPases that participate in cytokinesis relative to cell membrane. Together, the authors provided a 2–dimensional architecture of the cytokinetic ring at the nanometer scale.

The cytokinetic ring is a complex contractile machinery, and how the ring proteins are organized to collectively executing cytokinesis is an important but not yet well studied question in the field. The question addressed by McDonald et al., is important towards the understanding of cytokinetic ring function, and fits the scope of eLife. However, due to the technical caveats particularly the ambiguity of the cytokinesis stages and the labelling of cell membrane (major comments below), I am not very convinced by the accuracy of their findings. Since the relative molecular distances of ring proteins are at the nanometer scale, a good definition of cell membrane and the cytokinesis stages is essential to this study. Thus, these concerns should be addressed before further consideration for eLife.

1) The authors mentioned that they collected synchronized cells for their fPALM experiments (subsection “Sample preparation for fPALM imaging”). However, there is no supporting evidence for a good cell synchronization. The cells shown in Figure 1B do not seem to be at the same cell cycle stages (cells with different ring widths could be observed). Since the localizations of ring proteins may vary in early assembled rings, in matured rings, or in rings ready for constriction, a reliable indicator of cytokinesis stages will be required when performing the fPALM study. It would be useful to include mitotic markers such as SPB distance or spindle length for more precise indexing of cytokinesis progression.

This comment, also raised by reviewer 2, reflects a misunderstanding of the purpose of the lactose gradient methodology in our study. We apologize for the confusion. The lactose gradient was used to simply enrich for cells in the same cell cycle phase but we did not analyze all cells obtained. This step simply allowed us to decrease the number of experiments required to image a sufficient number of cells at the appropriate stage in cytokinesis. We made no assumption about exact cell cycle stage based on the enrichment protocol.

Rather, cells and rings were individually selected for analysis based on criteria that place them within a short window between full ring formation and the beginning of constriction (which is generally considered 15-20 minutes (Wu and Pollard, 2003)). These visual criteria are: (1) a lack of precursor nodes, meaning the first cytokinesis proteins have condensed into a ring, (2) the presence of a fully formed and tightly concentrated ring absolutely perpendicular to the cell length, and (3) the absence of detectable ring constriction or membrane ingression. We have reworded the methods and main text to clarify our use of the lactose gradient, and added a panel in Figure 1—figure supplement 1C that explicitly lays out the criteria used for selecting mature pre-constriction rings with examples.

It is possible that certain early ring components reorganize during the short period between initial ring formation and ring constriction, which is on the order of a few minutes. One fact to note is that the majority of proteins we have analyzed are not present at the division site in nodes or during the initial stages of ring formation meaning that if they are present in rings, the rings are fully formed and ready to constrict. Furthermore, visualizing mitotic markers such as spindle pole bodies will not allow us to define any more precise temporal window post- anaphase B onset when the majority of proteins are assembled into the ring and constriction. We have gathered data on a large number of rings using our criteria; temporally these ring populations are either all mature for late-arriving proteins, or for some proteins span full assembly through to maturation. Therefore, the data represent an average across this very short time window. That no single protein has a large deviation from its mean and that the deviations from the mean are similar across all proteins suggests significant individual protein rearrangements do not occur.

We believe the criteria we have employed to analyze protein localization exclusively in fully formed but not yet constricting rings achieves excellent temporal resolution and the best accuracy currently technically feasible in fPALM fixed samples.

2) The authors only map the ring protein localization at one single time point (+90 minutes after synchronization). To substantiate their findings, the relative localizations of several key proteins such as Cdc15, F–actin, Myo2 at different stages of cytokinesis (ready for constriction vs. constricting) should be profiled at higher temporal resolution.

In the future we hope to determine if ring architecture changes during constriction. However, we believe this is beyond the scope of the current study. For one, such experiments will require significant additional technical optimization because the mCling dye does not uniformly label the “ingressing” membrane, possibly because new membrane is added primarily from vesicular trafficking and would require much longer labelling with the dye to achieve uniform membrane labeling in the furrow.

3) A reliable cell membrane marker is critical to the accuracy of measurements in this study. It is not entirely clear if mCling is an appropriate marker for cell membrane in S. pombe as the dye binds to cell wall strongly (>100 nm thick structure in Figure 1—figure supplement 1B). How the authors be certain that the cell membrane was indeed labelled by mCling? The punctate staining of mCling in Figure 1C also suggests that not all membrane/cell wall could be stained. The different sizes of the mCling 'blobs' in Figure 2A and 3A suggest uneven binding to cell wall, which may result in the bias and inaccurate cell membrane edge determination used by authors.

mCling is currently the only membrane marker compatible with super-resolution imaging and fixation (see its extensive characterization in Revelo et al., 2014). We recognize this membrane dye is not perfect due to its binding to the cell wall, but to our knowledge it is only current option for this analysis.

To support this probe’s use as a membrane marker, we have compared the mCling label with the widely used FM4–64 membrane dye using conventional resolution imaging (Figure 1—figure supplement 1A). We find that these two membrane probes colocalize strongly and both are trafficked into intracellular membrane compartments, leading us to conclude mCling is a bona fide membrane marker. Comparing the mCling dye’s localization with a membrane tethered acyl-mMaple3 with fPALM also reveals an exact apposition of these two signals (Figure 1—figure supplement 1B), supporting mCling’s use as a membrane marker.

4) It is unclear how the internal edge of cell membrane in the study is determined. Was it mathematically defined for every single measurement or was it 'visually' defined? This is especially important as precision of fPALM nanoscale mapping of ring proteins strongly dependent on precision of localization of plasma membrane. A detailed description of how this was done is required in the Materials and methods section.

We originally identified the internal edge of the membrane visually. However, due to the concern about the manual nature of this approach, we implemented an automated method to identify the membrane edge. This method uses the position where the mCling intensity drops by 95% to mark the edge of the plasma membrane in an unbiased manner. All the data has been re-analyzed using this automated method. Satisfyingly, implementing the new method did not result in any significant changes in dcenter or σvert values.

5) To substantiate the relative distances of ring proteins, the authors could express and resolve two mMapple3 tagged proteins such as Mid1-mMapple3/mMapple3-Cdc15 (proteins close to cell membrane) and mMapple3-Myo2/LifeAct-mMapple3 (actin rich domain) with fPALM in the same cells. Ideally, two Gaussian peaks could be detected compared to cells expressing only one kind of tagged proteins.

We thank the reviewer for this excellent suggestion. We have performed this analysis on multiple pairs of proteins found in similar and distinct layers. As predicted, we find double tags in the same layer form single peaks. Supporting our individual measurements of proteins in different layers, we find double tags in different layers visualize as extra-wide single or resolvable double peaks, depending on their distance apart. We present examples of these double tags compared to their individual components in a new Figure 4A and Figure 4—figure supplement 1. We also compare the σvert values (a measure of the vertical width of the gaussians) of the double tags to their individual counterparts to show a “widening” of peaks when the components are far apart in a new Figure 4B. These data have strengthened our conclusions about the spatial separation of ring components relative to the plasma membrane.

Reviewer #2:

Constriction of the plasma membrane during cytokinesis depends on acto-myosin based contractile ring at the cell centre. Although we know the components of CR, lack of information regarding the nature of molecular arrangement and interaction has severely limited our understanding of the process. In S.pombe, a conserved actomyosin ring of 400nm thickness composed of 130 proteins drives constriction. Such molecular complexity is beyond the scope of diffraction limited optics. This paper (along with a recent work from Pollard lab) attempts at elucidating the structure of actomyosin ring using super-resolution microscopy. The authors suggest a stratified structure of membrane binding components proximal to the membrane and force generators distal to it. While both the scope of the study and its methodical execution deserves much credit, the manuscript suffers from lack of clarity and adequate discussion.

1) The authors show that the components of the Contractile rings are arranged in three distinct zones relative to plasma membrane in the contractile ring prior to ring constriction. My major concern is with lack of temporal information in this manuscript. Is this organization (protein-map) is altered during constriction? Do distribution of some components of Proximal zone change cluster-like to homogenous as the rings constrict or do they retain their conformation?

These are very interesting points and we intend to pursue the question of possible rearrangements during constriction in the future. As mentioned in our response to reviewer 1 point 2 above, technical challenges have so far limited our analysis to pre-constriction contractile rings. We must develop new methods to interrogate the constricting plasma membrane reliably.

2) The authors rightly argue that establishment of ultrastructural detail of the actomyosin ring is important to understand their function. However, they do not discuss the important implications of their observation. Myo2 tail is localised to membrane proximal zone where as its head domain is more distal and uniform. Does this mean that type 2 myosin is not arranged as a bipolar filament in contractile ring? Does this distribution change when the rings undergo constriction?

Indeed, Myo2 is not arranged as a bipolar filament in nodes or the contractile ring and this has been noted previously (Laporte et al., 2011; Laplante et al., 2016; Pollard et al., 2017). We have added a discussion point noting this difference from mammalian cells (subsection “Yeast methods and strain construction). As described above, examining constricting rings for any changes is beyond the scope of this study.

3 "At similar intermediate distances from the membrane, we find the tail domains of two other myosins: the non–essential myosin–II, Myp2, as well as a myosin-V, Myo51."

The manuscript doesn't provide any information on the head of these two motors. Given the recent implications of myosin motors in force generation and mutually exclusive localisation in the contractile ring, I think it may be worthwhile to investigate relative positioning of myo2 and myp2 (Laplante et al., 2015).

We have created the endogenous N–terminal mMaple3 fusions of Myo51 and Myp2 and added analysis of these motor domains. As expected, we found these domains in the same layer as F-actin where they can perform their motor activities, distal to the plasma membrane. In the Laplante et al., study, using conventional microscopy it was concluded that Myo2 and Myp2 co-localize before constriction but not during constriction. We agree with the findings that the motor domains co-localize prior to constriction.

4) fPALM strategy for determining the spatial organization of contractile ring components.

"The plasma membrane was defined as the edge of the mCling–ATTO647N signal, and the distance from this line was calculated for each individual localization in a contractile ring side-view "spot".

I am unable to fathom the accuracy of the plasma membrane position. If a line was drawn at the edge of the signal of plasma membrane manually it leaves much scope for errors. Signals such as the ones in Figure 1C and sup Figure 1B, clearly indicates a diffuse signal near the edge. How did the authors trace such a signal? Was it based on any particular threshold? Given, that this is the point of reference for the entire study, I hope the authors illustrate exactly how they performed their analysis.

We originally identified the internal edge of the membrane visually. However, due to the concern about the manual nature of this approach, we implemented an automated method to identify the membrane edge. This method uses the position where the mCling intensity drops by 95% to mark the edge of the plasma membrane in an unbiased manner. All the data has been re-analyzed using this automated method. Satisfyingly, using the new method did not result in any significant changes in dcenter and σvert values.

5) "We imaged synchronized cells that contained fully formed contractile rings (e.g. no precursor nodes remaining) that had not begun constriction (scored by membrane ingression and septum formation)".

The synchronisation achieved in lactose gradient selection of G1 cells, is not mentioned in the text. Was there any particular reason for G1 synchronisation and not G2?

In the absence of a cell-cycle marker and given the maturation period of S.pombe rings, it seems the authors imaged rings at least 20 minutes apart in their maturation. Certainly, this is enough scope for molecular rearrangements. Since incorporating a third fluorophore is challenging with the current set up, could the authors at least investigate simultaneous position and orientation of two ring components with each other and suggest they do not differ significantly from single labelling?

Please see response to reviewer 1, point 1 above. We again apologize for the wording in our original text.

6) "Therefore, we performed additional fPALM imaging at the top plane of cells (Figure 5A) to capture a lateral view of the contractile ring and assess the organization of a subset of contractile ring components".

I am unable to follow how the authors determine "top plane of cells". In contrast to the medial plane where the contractile ring serves as a reliable landmark with two points of maximal separation, "top of the cell" has no such landmarks. How does the authors reliably find the top focal plane? It would be useful if the authors can give an approximation of the circumferential length the 500nm Z section encompasses.

The top focal plane was determined by the presence of a ring “arc”, which is an equally reliable landmark as the two ring “spots” seen in the middle plane. Assuming a 3.5µm cell diameter, a 500nm Z section centered at the top of the cell should reveal at maximum a 1.9µm contractile ring “arc” at the top of the cell. This is very similar to the size of ring arcs we analyzed.

Reviewer #3:

In this study, McDonald et al., used fPALM to map the organization of proteins within the cytokinetic ring. Specifically, they measured the average distance of 29 ring proteins from the plasma membrane, revealing vertical zones of enrichment that make sense given the known functions and interactions of these proteins. For several factors, these positions are further confirmed by FRET. This work follows two previous studies that mapped a more limited set of proteins in cytokinetic ring precursors by SHREC (Laporte, 2011) and fPALM (Laplante, 2016). This McDonald et al., study is an impressive body of work that provides the most detailed account of protein organization in the cytokinetic ring, at least to my knowledge. It will be a reference point for future mechanistic studies. My main concern is that the work is entirely descriptive, it is strictly a localization study. Ideally, one would like to see the function of these localization results tested – what aspects of this vertical organization are important for cytokinesis? As such, it is more of a "resource" than a "research article." Nonetheless, it represents a major step forward for the field, and is likely to draw broad interest. I have a couple of comments that the authors might address to strengthen the manuscript:

Figure 1: Are all proteins distributed in the same width within the ring? The authors measure distance from the membrane and vertical distribution (i.e. spread in distance away from the membrane) for each protein. However, it would be interesting to know if all proteins are distributed along the same horizontal width or not. Presumably, these data are readily available from the images, and the authors state in the text an overall ring width of 181nm, but I did not see how this number was calculated. Measurement for each component would further refine the molecular architecture of the ring by adding additional confinement for each protein.

We thank the reviewer for the suggestion, and have added this recommended analysis of ring width (σwidth) to Figures 2 and 3—figure supplement 1A. Widths appear to vary between proteins, without much correlation with distance from the membrane. We have added a discussion point to this effect (subsection “Structural layers within the contractile ring”).

We also now report a more accurate ring with value in the text (182nm) that represents an average of the 29 proteins’ widths.

Figure 5: The data suggest that ring proteins are clustered near the membrane, but not clustered in the more interior strata. Particularly striking are examples like Myo2 and Cdc15, where the cortical domains are clustered, but their 'inner' domains are unclustered. As the authors know, the Pollard lab published fPALM data showing that both termini of these same constructs are clustered in the ring. In fact, all 6 constructs (4 proteins, 2 of which were tagged at each terminus) examined in the Pollard study were clustered. One difference between the two studies is that the Pollard lab imaged live cells, while the McDonald study imaged fixed cells, raising the question of whether fixation affected clustering (there are other difference and possibilities as well). I wonder if the authors could address this discrepancy by imaging lateral organization in live cells. They could also consider imaging more examples of cortical versus interior proteins, to support further this general trend of clustered at the membrane but diffuse at the interior. Finally, measurements of horizontal spread in middle focal planes (previous comment) might show tight clusters at the cortex with broad spread at the interior.

We have altered the language of the discussion to include the difference in methods to qualify our conclusions. It is unfortunately not possible to image circumferential organization in live cells with our microscope setup, as image acquisition takes 10+ minutes in each channel.

There are certainly caveats to the use of fixed cells, which is why we minimized the extent and length of fixation and used FRET methodology in live cells to corroborate our findings. Still, we cannot rule out that the evenness of signal in the interior of the ring is somehow affected by fixation. However, there are also significant caveats with the use of live cells for fPALM. Though fixed cells are disadvantageous due to loss of the dynamics of the cytokinesis process, fixing is critical to allow superresolution fluorophores to be exhaustively photoconverted and excited over multiple minutes. Uneven conversion and excitation becomes quite possible with the fast and short acquisitions necessary in live cells, perhaps biasing the results of image analysis towards punctate patterns. We also note that the methodology reported in the Pollard study lacked any fiducial marking for drift correction which could also contribute to the development of punctate images.

To expand this analysis as suggested, we imaged and analyzed more ring components in the circumferential dimension. In the proximal layer, we added septin Spn3 and found a uniform distribution, and in the intermediate layer we have added Fic1 and the tail of Myp2, finding that both are uniform. The trend remains consistent that intermediate and distal components are uniformly distributed, while certain proximal components (generally those that were present in precursor nodes) cluster.

https://doi.org/10.7554/eLife.28865.020

Article and author information

Author details

  1. Nathan A McDonald

    Department of Cell and Developmental Biology, Vanderbilt University, Nashville, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Visualization, Methodology, Writing—original draft, Writing—review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-2716-3881
  2. Abigail L Lind

    Department of Biomedical Informatics, Vanderbilt University Medical Center, Nashville, United States
    Contribution
    Formal analysis, Validation
    Competing interests
    No competing interests declared
  3. Sarah E Smith

    Stowers Institute for Medical Research, Kansas City, United States
    Contribution
    Resources, Methodology
    Competing interests
    No competing interests declared
  4. Rong Li

    Department of Cell Biology, Johns Hopkins University School of Medicine, Baltimore, United States
    Contribution
    Resources, Supervision, Methodology
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-0540-6566
  5. Kathleen L Gould

    Department of Cell and Developmental Biology, Vanderbilt University, Nashville, United States
    Contribution
    Conceptualization, Resources, Formal analysis, Supervision, Funding acquisition, Writing—original draft, Project administration, Writing—review and editing
    For correspondence
    kathy.gould@vanderbilt.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-3810-4070

Funding

National Institutes of Health (Research Grant)

  • Rong Li
  • Kathleen L Gould

American Heart Association (Graduate Student Fellowship)

  • Nathan A McDonald

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

Acknowledgements

We thank members of the Gould lab for critical review of the manuscript, Xiaowei Zhuang for the mMaple3 constructs, and Chunkai Zhou for FRET microscope assistance. fPALM imaging was performed in the Vanderbilt Cell Imaging Shared Resource and FRET imaging was performed in the Stowers Institute Microscopy Center.

Reviewing Editor

  1. Mohan K Balasubramanian, University of Warwick, United Kingdom

Publication history

  1. Received: May 22, 2017
  2. Accepted: September 14, 2017
  3. Accepted Manuscript published: September 15, 2017 (version 1)
  4. Version of Record published: September 26, 2017 (version 2)

Copyright

© 2017, McDonald 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

  • 2,039
    Page views
  • 407
    Downloads
  • 19
    Citations

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

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

    1. Cell Biology
    2. Neuroscience
    Veronica Cloud et al.
    Research Article Updated
    1. Cell Biology
    2. Physics of Living Systems
    Maximilian AH Jakobs et al.
    Tools and Resources