It has long been known that the centralspindlin complex and CPC are highly enriched in the midzone during cytokinesis (Glotzer, 2009); however, previous studies in Drosophila and mammalian cells as well as Xenopus embryos have reported MT plus-tip localization of the centralspindlin complex as well as the CPC component ABK (Breznau et al., 2017; Nishimura and Yonemura, 2006; Vale et al., 2009). While there has been significant investigation of midzone populations of cytokinesis regulators, very little attention has been given to MT plus-end localized components. Thus, we sought to further investigate the spatio-temporal dynamics of the tip-localization properties of centralspindlin and ABK by live-cell TIRF microscopy of dividing Drosophila S2 cells expressing fluorescently tagged Dm MKLP1, RacGAP50C, and ABK.

To specifically observe the MT plus-end localization properties of these regulators, S2 cells, which are semi-adherent, were seeded on concanavalin A (Con A) coated glass-bottom dishes to adhere and flatten them. We had previously shown that contractile myosin rings assembled normally on Con A (Ye et al., 2016), but since such a treatment could interfere with furrow formation, the effects of Con A on cytokinesis was further assessed by overnight time-lapse imaging of S2 cells seeded on Con A. Importantly, we found that 97% of dividing cells began to ingress furrows in the mid-cell region within ~10 min of anaphase onset. Of these cells, 32% completed cytokinesis normally (as defined by clearly delineated mono-nucleated daughter cells) while 68% of cells regressed their furrows 39.4 ± 7.6 min. after furrow ingression initiated (Video 1, n = 28, mean ± SD values are reported). Thus, the Con A concentration used in these experiments negatively impacted the latest stages of cytokinesis (most likely cytokinetic abscission) but did not affect the processes that we aimed to study, namely cleavage furrow establishment and ingression.

Video 1

Download asset

After observing that Con A didn’t interfere with the early stages of cytokinesis, we began by imaging Dm MKLP1. In metaphase, Dm MKLP1 diffusely localized throughout the cytosol without evident spindle MT enrichment (Video 2); however, within 2–3 min. of anaphase onset, Dm MKLP1 localized to and tip-tracked on the plus-ends of astral MTs (Figure 1A; Figure 1—figure supplement 1; Video 2). In agreement with prior observations (Vale et al., 2009), Dm MKLP1 localized to astral MT plus-tips, both polar and equatorial, before being lost from the polar tips and becoming preferentially patterned onto equatorial tips ~ 10 min. following anaphase onset. We next visualized FP-tagged RacGAP50C since homodimers of MKLP1/ZEN-4 and MgcRacGAP/CYK-4 interact to form a functional centralspindlin complex (Mishima et al., 2002; Pavicic-Kaltenbrunner et al., 2007). In agreement with a recent report of MgcRacGAP localization to the MT plus-ends in Xenopus laevis (Breznau et al., 2017), we also observed MT plus-end localization and tip-tracking activity of RacGAP50C during cytokinesis (Figure 1B; Video 3) Like Dm MKLP1, RacGAP50C decorated polar and equatorial astral MT plus-ends within ~2 min. of anaphase onset before becoming patterned onto equatorial MTs. We also observed RacGAP50C and Dm MKLP1 accumulation on midzone MTs and the equatorial cortex (Figure 1A,B; Videos 2 and 3). Thus, the two components of the centralspindlin complex, Dm MKLP1 and RacGAP50C, exhibit the same dynamic localization pattern on midzone MTs, the cortex, and astral MT plus-tips as cells progressed through anaphase and cytokinesis.

Figure 1 with 1 supplement see all

Download asset Open asset

Video 2

Download asset

Video 3

Download asset

Since our transgenes are expressed on top of the endogenous proteins, we examined the level of transgene expression to assess whether the appearance of CS-TIPs was a result of overexpressing centralspindlin components. To address this issue we performed: 1) quantitative western blotting of cells expressing Dm MKLP1- and RacGAP50C-EGFP to access the extent of overexpression and 2) TIRF and/or spinning disk confocal imaging under conditions where the endogenous copy of either Dm MKLP1 or RacGAP50C was depleted using RNAi. Western blotting indicated that the Dm MKLP1-EGFP transgene was expressed ~5 fold lower than its endogenous counterpart (Figure 1C). Importantly, Dm MKLP1-EGFP localized to MT plus-ends, midzones, and the cortex when endogenous Dm MKLP1 was depleted and rescued the severe cytokinesis failure phenotype that resulted from depletion of Dm MKLP1. (Figure 1D; Figure 1—figure supplement 1). Spinning disk confocal imaging of Dm MKLP1-EGFP cells depleted for the endogenous protein revealed that CS-TIPs are not restricted to the ventral surface of the cell as the decoration of CS-TIPs was observed throughout the cell volume (up to 4.6 µm towards the dorsal surface) (Figure 1D; Video 5). Like Dm MKLP1-EGFP, RacGAP50C-EGFP expression was less than its endogenous counterpart (Figure 1—figure supplement 1). Expression of RacGAP50C-EGFP largely rescued the failure of cytokinesis following RacGAP50C depletion and the RacGAP50C-EGFP localized to CS-TIPs, midzones, and the cortex when the endogenous RacGAP50C was depleted (Figure 1—figure supplement 1). Altogether, these data lead us to conclude that the spatio-temporal properties of the MT plus-tip population of centralspindlin were not a consequence of transgene overexpression.

Video 5

Download asset

Consistent with previous reports (Landino et al., 2017; Vale et al., 2009), we also observed ABK localization on polar and equatorial astral MT plus-tips, midzones, and the equatorial cortex (Figure 1E). Possibly due to intense and rapid ABK enrichment at the cortex, ABK localization on plus-ends was evident for less time (~1 min.) than the centralspindlin components, which typically lasted for 2–10 min. on polar MTs and >10 min. on equatorial MTs. Polo kinase becomes enriched on midzones during cytokinesis and its activity is required for furrow initiation and successful cytokinesis (Brennan et al., 2007; Burkard et al., 2009; Carmena et al., 1998; Lénárt et al., 2007; Petronczki et al., 2007). We next applied TIRF microscopy to Polo-EGFP expressing cells to examine its spatio-temporal dynamics during cytokinesis. As expected, Polo localized to the overlap zone of the midzone MT array. However, Polo kinase also localized to the MT plus-ends within ~2–3 min. of anaphase onset before becoming patterned onto equatorial astral tips ~10 min. post anaphase onset (Figure 1F; Video 4). Polo kinase uniformly localized along the length of MTs as cells advanced into late telophase, which was expected since Polo kinase localizes to interphase MTs in Drosophila cells (Archambault et al., 2008). Thus, prior to telophase, Polo kinase exhibited an identical localization pattern as the centralspindlin complex. While the centralspindlin plus-tip localization is clearly a conserved phenomenon as it has now been observed in Drosophila, Xenopus, and human cells, it would be worthwhile to determine if the plus-tip localization of Polo kinase is also conserved or rather a Drosophila-specific phenomenon related to the fact that, unlike many other model systems, the kinase localizes robustly to interphase MTs. In cells expressing Polo-EGFP we often observed inward moving zones of Polo-EGFP exclusion (Figure 1G; Video 4). Since MTs could not penetrate these exclusion zones and were pushed inward as the zones expanded, we interpreted the Polo-EGFP exclusion zones as indicative of local regions of actomyosin contractility. Since cortical contact by Polo-positive MT plus-tips consistently triggered local contractility at both the equatorial region and in the vicinity of polar astral MT plus-ends prior to patterning, we hypothesized that: 1) Polo kinase, like centralspindlin, localizes to CS-TIPs in S2 cells, and 2) these specialized MT plus-tips trigger cortical contractility.

Video 4

Download asset

Since cortical contractility is triggered by the GTP-bound state of the small GTPase RhoA (Bement et al., 2006; Glotzer, 2004; Jordan and Canman, 2012), we next examined the behavior of the RhoA-GEF ECT2 (Pebble in Drosophila) by visualizing dividing cells co-expressing Dm ECT2-EGFP and Tag-RFP-T-α-tubulin by TIRF microscopy. During interphase Dm ECT2 localized to the nucleus, but a cortical pool was also evident that remained throughout mitosis. Following anaphase onset, Dm ECT2 localized to midzone MTs and became enriched on the nearby equatorial cortex (Figure 2A; Video 6; n = 22/23). Dm ECT2 did not robustly tip-track on astral MTs (although rare events were observed), rather, cortical Dm ECT2 co-localized with MT plus-tips within 10.5 ± 1.0 s. (mean ± SEM, n = 21) of contact (Figure 2B), which was followed by amplification of Dm ECT2 recruitment that peaked 1.5 ± 0.1 (mean ± SEM, n = 18) minutes after the plus-tips contacted the cortex (Figure 2C; Video 6).

Figure 2

Download asset Open asset

Video 6

Download asset

We next hypothesized that MT plus-end-mediated recruitment of Dm ECT2 would result in local activation of RhoA. TIRF-based visualization of cells co-expressing TagRFP-T-α-tubulin and Rhotekin, a reporter for active RhoA-GTP (Bement et al., 2005; Benink and Bement, 2005), N-terminally tagged with EGFP revealed that a subset of astral MT plus-tips, which we presumed to be CS-TIPS, locally activated RhoA within seconds of contacting the cortex (Figure 2B,D; Video 7, n = 18/18). The Rhotekin (active RhoA) signal initially appeared within 14.3 ± 1.7 s. of cortical contact by the astral MT plus-tips and local enrichment peaked within 2.5 ± 0.2 min. of first contact (Figure 2B,C; n = 18). Due to the high spatial and temporal resolution of the live-cell TIRF microscopy, both MT plus-tip-based and midzone-derived RhoA activation signals could be visualized in Rhotekin-expressing cells. The two activation signals often arrived simultaneously, although MT plus-tip signaling sometimes preceded evident midzone-localized RhoA activation (Figure 2D, Video 7). We confirmed that centralspindlin and Polo kinase were enriched on the MT plus-tips that triggered RhoA activation by visualizing the co-localization of cortically enriched TagRFP-T-Rhotekin with both Dm MKLP1-EGFP (not shown) and Polo-EGFP (Figure 2E; Video 8, n = 20/22).

Video 7

Download asset

Video 8

Download asset

During cytokinesis, activation of RhoA results in assembly and contraction of an actomyosin ring. We reasoned that if CS-TIPs were involved in RhoA activation near the cortex, then it would result in myosin accumulation in their vicinity. To test this hypothesis, we made a stable cell line co-expressing Polo-EGFP (to identify CS-TIPs) and myosin regulatory light chain (MRLC) tagged with TagRFP-T. Dual-color TIRF imaging of this cell line revealed a strong enrichment of MRLC near the Polo-positive MT plus-ends (Figure 2F; Video 9) that peaked within 5.5 ± 0.54 min. (Figure 2C; mean ± SEM, n = 19) of persistent CS-TIP contact with the cortex. MRLC enrichment also mediated localized cortical contractility as the CS-TIPs were clearly pulled inward when the cortical contacts persisted, which resulted in the coalescence of CS-TIPs within a band of enriched MRLC localized several microns interior to the initial MT contact points at the cortex (Figure 2F; Video 9). CS-TIPs were required to maintain RhoA activation (Video 8) and cortical contractility (Video 9) as polar sites of Rhotekin and MRLC enrichment were lost upon equatorial patterning and disassembly of polar Polo-positive CS-TIPs. Identical results for the establishment and maintenance of RhoA activation and MRLC enrichment were attained in cells expressing Dm MKLP1 (n = 12). Altogether, these results support the conclusion that CS-TIPs recruit Dm ECT2 to activate cortical RhoA thereby resulting in localized myosin accumulation and induction of cortical contractility.

Video 9

Download asset

Given the central role ABK and Polo kinase play in cytokinesis, we proceeded to investigate the contribution of ABK and Polo kinase activities to CS-TIP assembly. Wash-in experiments with the Drosophila specific ABK inhibitor, Binucleine 2, and the Polo kinase inhibitor BI 2536 (Eggert et al., 2004; Lénárt et al., 2007; Smurnyy et al., 2010) were conducted to isolate cytokinesis-specific contributions of each kinase. Polo-EGFP or Dm MKLP1-EGFP cells were continuously imaged from metaphase to anaphase onset and Binucleine 2 was directly introduced into the imaging chamber when CS-TIPs first appeared. Interestingly, wash-in of Binucleine 2 resulted in a complete loss of Polo kinase and Dm MKLP1 from MT plus-tips within 2.5 ± 0.43 min. of introducing the inhibitor (Figure 3, A,B; Videos 10 and 11; n = 12). Introducing the Polo kinase inhibitor BI 2536, which we and others have shown effectively inhibits Dm Polo kinase (Carmena et al., 2014; Kachaner et al., 2017; Lénárt et al., 2007), did not affect the CS-TIP localization of either Polo kinase or Dm MKLP1 (Figure 3, C,D; Videos 12 and 13; n = 10). Taken together, the data demonstrate that ABK activity is required for centralspindlin and Polo kinase to localize to CS-TIPs while Polo kinase activity is not.

Figure 3 with 1 supplement see all

Download asset Open asset

Video 10

Download asset

Video 11

Download asset

Video 12

Download asset

Video 13

Download asset

PRC1 (Feo in Drosophila) is a well-characterized midzone MT component that preferentially bundles anti-parallel MTs (Bieling et al., 2010). PRC1 forms a complex with kinesin-4 (Klp3A in Drosophila) that regulates midzone MT length and stabilization (Bieling et al., 2010; Hu et al., 2011; Subramanian et al., 2013; Zhu and Jiang, 2005). Klp3A also localize to the central spindle and midbody. Mutations in the gene that encode Klp3A protein in Drosophila caused defects in central spindle assembly during late anaphase/early telophase (Williams et al., 1995). Notably, defects in the central spindle assembly didn’t have a major impact on spindle length and cells exhibited normal mitosis (Williams et al., 1995). To investigate if the PRC1-Klp3A complex contributes to CS-TIP assembly, Klp3A was depleted from Polo-EGFP or Dm MKLP1-EGFP expressing cells. In agreement with prior work (D'Avino et al., 2007), Polo kinase failed to localize to midzone MTs in cells with ≥97% of Klp3A depleted (Figure 3E; Figure 3—figure supplement 1; Video 14; n = 25); however, Polo kinase assembled normally onto CS-TIPs (Figure 3E). Klp3A was next depleted from cells expressing Dm MKLP1-EGFP and neither midzone MT localization nor CS-TIP localization was affected (Figure 3—figure supplement 1). These data support the conclusion that Dm MKLP1 and Polo kinase do not require the PRC1-Kinesin-4 complex to assemble onto CS-TIPs.

Video 14

Download asset

During cytokinesis, ABK localizes to midzone MTs, astral MT plus-tips, and the cortex; however, the contribution of different ABK pools to furrow formation is unclear. In many model systems, including Drosophila, CPC localization to the midzone is mediated by MKLP2 (Subito in Drosophila) (Cesario et al., 2006; Gruneberg et al., 2004; Kitagawa et al., 2013; Nguyen et al., 2014). To investigate whether midzone-localized ABK contributed to CS-TIP assembly, Dm MKLP2 was depleted from cells expressing Dm MKLP1-EGFP. Depletion of Dm MKLP2 resulted in a significant reduction in midzone levels of active phosphorylated ABK (Figure 3F; Figure 3—figure supplement 1; n = 30), but Dm MKLP1 still localized robustly to CS-TIPs and cells completed cytokinesis normally (Figure 3G), (Ye et al., 2015). The results from the Binucleine 2 wash-ins and Dm MKLP2 depletions, taken together, support the conclusion that while global ABK activity is essential for CS-TIP assembly, the midzone-based ABK pool is largely dispensable. Since cortical ABK becomes evident following CS-TIP assembly (Figure 1C), we favor the interpretation that a soluble cytosolic pool of active ABK promotes assembly of CS-TIPs after anaphase onset.

We next examined the specific roles of ABK and Polo kinase in CS-TIP-mediated activation of RhoA by conducting wash-in experiments with Binucleine 2 and BI 2536. Dual color TIRF imaging of anaphase cells expressing TagRFP-T-Rhotekin and EGFP-α-tubulin revealed a dramatic loss of active RhoA from the vicinity of astral MT plus-tips within 3.16 ± 0.77 min. of introducing Binucleine 2 (Video 15; n = 10). Identical results were attained in cells co-expressing TagRFP-T-Rhotekin and Dm MKLP1-EGFP (Video 16; n = 10). Loss of localized Rhotekin and Dm MKLP1 from MT plus-tips occurred concomitantly and within ∼2–3 min. of Binucleine 2 addition, demonstrating that ABK activity is required for both CS-TIP assembly and downstream activation of RhoA.

Video 15

Download asset

Video 16

Download asset

Polo kinase activity is required for targeting ECT2 to the midzone by promoting a physical association between ECT2 and centralspindlin (Burkard et al., 2009; Petronczki et al., 2007). Thus, we next examined the effects of Polo kinase inhibition on downstream signaling to RhoA activation by CS-TIPs. Introduction of BI 2536 to anaphase cells expressing TagRFP-T- Rhotekin and EGFP-α-Tubulin resulted in a dramatic reduction in active RhoA enrichment near the astral MT plus-tips within ~4 min. of adding the inhibitor (Video 17; n = 8). The same experiment was also conducted in cells co-expressing Dm MKLP1-EGFP and TagRFP-T-Rhotekin. In agreement with our prior observations (Figure 3D), Polo kinase inhibition did not affect the CS-TIP localization of Dm MKLP1-EGFP; however, Rhotekin signal was lost in the vicinity of the Dm MKLP1-positive MT plus-tips (Video 18; n = 8). Thus, Polo kinase activity is required for CS-TIP signaling, but not for CS-TIP assembly. We reasoned that since Polo kinase activity is required for CS-TIP-mediated activation of RhoA, then Polo inhibition should alter the localization of Dm ECT2 in response to physical contact by CS-TIPs. Indeed, Polo kinase inhibition resulted in the loss of localized cortical Dm ECT2 from CS-TIP contact points (Video 19; n = 4).

Video 17

Download asset

Video 18

Download asset

Video 19

Download asset

Over the course of visualizing hundreds of dividing cells by TIRF microscopy we frequently noticed MT structures that assembled in close proximity to the cortex as cells progressed through cytokinesis. We referred to these structures as satellite MT arrays because they formed de-novo and were not evidently linked to the spindle apparatus. Interestingly, two classes of satellite arrays were observed, which were deemed midzone and astral brush satellites because their behavior and organization were analogous to conventional midzone and astral MTs respectively.

Similar to prior observations, we noted that Dm MKLP1 on astral MT plus-tips bundled both parallel and anti-parallel MTs (Vale et al., 2009). Midzone satellites were comprised of anti-parallel bundles of equatorial astral MTs that crossed paths by growing towards the cortex, often buckling upon contact, and eventually becoming bundled by Dm MKLP1 and likely other cross-linking factors such as PRC1 (Video 2). Midzone satellites assembled in close proximity to the cortex and the ensuing cortical contractility further bundled the MTs and enriched Dm MKLP1 (and Polo kinase) in the overlap regions as MTs were moved inward. The process resulted in midzone satellites being in direct contact with the contractile machinery throughout cytokinesis (Figure 4A; Video 2). In most cases satellite midzones were eventually pulled into the conventional midzone that was positioned microns away from the cortex when contractility had initiated.

Figure 4

Download asset Open asset

Astral brush satellites formed de novo from cytosolic nucleation centers producing dynamic MTs that became organized over time into branched, spindle pole-like structures with MTs oriented toward the cortex (Figure 4B). While exhibiting astral-like properties, astral brushes typically assembled ~5–10 microns away from the spindle apparatus and the MTs comprising astral brushes did not emanate from spindle poles or centrosome. Furthermore, the MT plus-tips of astral brushes were often enriched with centralspindlin and activated RhoA, especially prior to patterning of CS-TIPs onto equatorial MTs (Figure 4B).

Imaging astral brushes by TIRF microscopy was limited to the portions of MTs that were close to the plasma membrane. To better assess the properties of astral brush satellites we next employed correlative TIRF-spinning disk confocal microscopy on the same cell in order to visualize MTs outside of the TIRF field (Figure 4C). In this approach a cell would be imaged by time-lapse TIRF microscopy until the assembly of astral brushes became evident at which point a 2-color spinning disk confocal Z-section would be rapidly acquired. This approach revealed the existence of oblique astral brushes with MTs, only the tips of which were evident in the TIRF field, oriented at a steep angle towards the bottom of the cell (Figure 4C – inset 1) as well as brushes consisting of multiple MTs in the TIRF field that were oriented nearly parallel and closer to the plasma membrane (Figure 4C – inset 2). In analyzing both the confocal Z-slices (Video 5) as well as 3D reconstructions (Videos 20 and 21), it was evident that oblique brushes originated considerably further up into the cell volume than the astral brushes in which MTs could be seen in the TIRF field. For example, the vertex of the brush in inset 1 (Figure 4C) was ~4 μm from the bottom of the cell versus ~1.4 μm for the astral brush in inset 2. In both scenarios the base of the astral brush structures were >5 μm from the centrosomes and midzones thereby qualifying them as legitimate ‘satellite’ structures per our description of this phenomenon.

Video 20

Download asset

Video 21

Download asset

Having observed that CS-TIPs activate RhoA via recruitment of cortical Dm ECT2, we next turned our attention to the mechanism by which CS-TIP components localize to plus-tips. The MT plus-end-tracking protein, EB1 is known to autonomously tip-track, and interact with other proteins to recruit them to polymerizing MT plus-ends (Akhmanova and Steinmetz, 2008; Duellberg et al., 2014; Honnappa et al., 2009; Strickland et al., 2005). EB1 was depleted by RNAi (Figure 5—figure supplement 1) to examine its contribution to CS-TIP assembly and signaling. Consistent with previous work (Rogers et al., 2002), we observed a reduction in spindle length in EB1-depleted cells compared to control cells (control RNAi: 7.0 ± 0.86 µm, EB1 RNAi: 6.1 ± 0.69 µm; n = 73, p=4.31 E-12); (Figure 5—figure supplement 1). While >80% of control cells exhibited robust CS-TIPs, EB1 depletions, which reproducibly yielded >85% knock down efficiencies (Figure 5—figure supplement 1), significantly compromised tip-tracking of both Polo kinase and Dm MKLP1 as <10% of EB1-depleted cells exhibited MT plus-tip localization of these components (Figure 5, A,B; Video 22; n = 31 and 41 for control and EB1 depleted cells respectively). Since EB1 depletion compromised the plus-tip localization of centralspindlin, we reasoned that it would also result in loss of RhoA activation by MT plus-tips. To test this hypothesis, we depleted EB1 in cells expressing Rhotekin-EGFP and Tag RFP-T-α-Tubulin, and performed dual color TIRF imaging. Interestingly, while midzone RhoA activation was detectable in nearly all EB1 depleted cells (19/21), EB1 depletion resulted in the loss of evident RhoA activity in the vicinity of MT plus-tips in 20/21 cells (Figure 5—figure supplement 1; Video 23).

Figure 5 with 2 supplements see all

Download asset Open asset

Video 22

Download asset

Video 23

Download asset

Furrow initiation defects have been previously reported in dividing sea urchin eggs upon injection of either an inhibitory antibody against EB1 or a dominant negative EB1 truncation (Strickland et al., 2005). Furthermore, depletion of both EB1 and EB3 from HeLa cells caused a 5-fold increase in cytokinesis failure (Ferreira et al., 2013). To assess the functional contributions of EB1 to cytokinesis in S2 cells, we depleted EB1 from cells expressing Tag-RFP-T-α-tubulin and subjected them to overnight imaging. Interestingly, EB1 depletion resulted in furrow initiation failure in 15.5% mitotic cells, which is 4.7-fold higher than the control RNAi cells where only 3.3% mitotic cells exhibited furrow initiation failure. (Figure 5C; n = 120 and 110 for control and EB1 RNAi cells respectively). Cytokinesis failure results in binucleated cells and, accordingly, analysis of fixed cells revealed that EB1-depleted cells exhibited a ~3.5–5 fold increase in the number of binucleated cells with respect to control cells (Figure 5D and Figure 5—figure supplement 1; three independent RNAi experiments, n = 570 and 603 cells counted in total for control- and EB1 RNAi-treated cells respectively). Thus, EB1 depletion compromised CS-TIP assembly and signaling, and resulted in functional consequences as cells did not initiate furrow ingression and failed cytokinesis at higher rates than control cells.

We next tested if restoring centralspindlin to MT plus-tips could rescue the effects of EB1 depletion by generating cells expressing an EB1-RacGAP50C-EGFP hybrid. EB1-RacGAP50C-EGFP constitutively tip-tracked on polymerizing MTs and this behavior was retained when endogenous EB1 was depleted by RNAi targeting the 3’UTR of EB1, which was not present in the hybrid sequence (Figure 5—figure supplement 1). While depletion of EB1 resulted in an increase in binucleated WT cells, there was not a statistically significant difference in the percentage of binucleates between control and EB1 depleted cells expressing EB1-RacGAP50C-EGFP (Figure 5D and Figure 5—figure supplement 1; two independent RNAi experiments, n = 431 and 550 cells counted in total for control and 3’ UTR EB1 RNAi cells respectively).

Our characterization of the EB1-dependent nature of CS-TIP tracking is consistent with recent work in Xenopus laevis embryonic cells describing the plus-tip tracking behavior of the centralspindlin component MgcRacGAP via a SxIP EB1-interaction motif during cytokinesis (Breznau et al., 2017). The SxIP motif in X. laevis MgcRacGAP is not conserved beyond Xenopus species despite the fact that centralspindlin plus-tip tracking is clearly conserved in Drosophila and human cells. Our group recently identified unconventional EB1-interaction motifs in the Drosophila chromokinesin NOD (deemed ‘PT’ motifs) (Ye et al., 2018) that are similar to a newly characterized MT plus-tip localization sequence (Kumar et al., 2017; Manatschal et al., 2016; Stangier et al., 2018). From our work and others, we propose a PT motif consensus of hxxPTxh (‘h’, hydrophobic a.a.; x, any amino acid). Sequence alignments revealed a conserved, with the possible exception of C. elegans, putative hxxPTxh motif in the C-terminal region of MgcRacGAP/Cyk-4 (Figure 5E). After observing RacGAP50C localization on CS-TIPs (Figure 1B), we generated a C-terminal truncation of RacGAP50C (1-526) lacking the hxxPTxh motif (a.a. 530–536; IFGPTIV), but retaining the ability to associate with Dm MKLP1 via its N-terminus (Mishima et al., 2002; Pavicic-Kaltenbrunner et al., 2007). ∆C-RacGAP50C resulted in a severe reduction in tip-tracking efficiency; however, its localization to the midzone and equatorial cortex was unaffected (Figure 5F,H; Video 24; n = 20). We next generated a more targeted deletion mutant of RacGAP50C in which only four amino acids (PTIV^533-536) were deleted from the hxxPTxh motif. The PTIV deletion does not remove any amino acids that are required for GAP activity and structural simulations (Biasini et al., 2014) did not reveal a significant change in the structure of the deletion mutant (Figure 5—figure supplement 2). Like the ∆C-RacGAP50C mutant, ∆PTIV-RacGAP50C localized normally to midzone MTs and the equatorial cortex (Figure 5G; Video 25; n = 26). However, the tip-tracking activity of ∆PTIV-RacGAP50C was compromised as the mutant exhibited tip-tracking in only ~23% of cells that were transfected compared to ~80% of cells expressing WT RacGAP50C (Figure 5H). It should be noted that the RacGAP50C mutants had to be studied through transient transfections because attempts to make stable cell lines expressing the ΔC or ΔPTIV RacGAP50C mutants were unsuccessful.

Video 24

Download asset

Video 25

Download asset

Since deletion of either the C-terminus or hxxPTxh motif in RacGAP50C compromised the MT plus-tip localization of RacGAP50C, we sought to further assess centralspindlin functionality when RacGAP50C localization was restored to plus-tips via direct fusion to EB1. Unlike prior failed attempts to make stable cell lines expressing the ΔC--RacGAP50C mutants, we were able to make stable cell lines expressing the mutants fused to EB1. Depletion of Dm MKLP1, which results in co-depletion of RacGAP50C (D'Avino et al., 2006), resulted in the absence of obvious cortical regions of localized RhoA activation in 11/13 Dm MKLP1-depleted cells due to an apparent loss of both midzone derived and MT plus-tip-based spatial cues (Figure 5—figure supplement 2; Video 26). Interestingly, depletion of EB1 resulted in the specific loss of MT plus-tip-proximal, but not midzone-based, RhoA activation (Figure 5—figure supplement 1). When the ∆C-RacGAP50C mutant was fused to EB1, it constitutively localized to growing plus-tips throughout the cell cycle and also localized to the midzone during anaphase/cytokinesis. Interestingly, the EB1-∆C-RacGAP50C hybrid supported MT plus-tip-based RhoA activation in the absence of endogenous RacGAP50C in 3/3 cells that were imaged (Figure 5I, Figure 5—figure supplement 2). The finding indicates that the RhoA activating capacity of CS-TIP-based centralspindlin, likely via Dm ECT2 recruitment, is mediated by the N-terminal 526 a.a. of RacGAP50C.

Video 26

Download asset

The positioning of the contractile region has been variously proposed to involve positive equatorial cues derived from the central spindle, astral MTs, or both; and inhibitory factors acting at poles (Canman et al., 2003; Mogilner and Manhart, 2016). Direct evidence of RhoA activation by astral MTs is lacking. Here we show that Dm ECT2 is recruited to cortical CS-TIP contact sites within seconds, which, in turn, leads to rapid localized RhoA activation in the vicinity of MT plus-tips (Figure 2). How do the molecular constituents of a CS-TIP, namely Polo kinase and centralspindlin, carry out this function (Figure 6D)? It is presently unknown which component of a CS-TIP is responsible for physically recruiting cortical Dm ECT2, but based on prior work (Burkard et al., 2009; Petronczki et al., 2007; Somers and Saint, 2003; Wolfe et al., 2009), we postulate that it is mediated by the N-terminus of RacGAP50C once it has been phosphorylated by Polo kinase. Our data also show that the C-terminus of RacGAP50C and EB1 are required for centralspindlin’s plus-tip enrichment. What then is the necessity of the other centralspindlin component Dm MKLP1 to CS-TIP based signaling? An appealing model (discussed further below) is that the motor domain of Dm MKLP1 is also necessary, although not sufficient, for robust plus-end localization of centralspindlin. As for the role of plus-tip localization of Polo kinase, it is unclear whether its plus-tip localization is necessary for CS-TIP-based signaling or if the kinase’s plus-end localization during cytokinesis is even conserved beyond Drosophila. Although, global Polo kinase activity is required for CS-TIP-based signaling. It is conceivable that centralspindlin is the only constituent of CS-TIPs necessary to recruit cortical ECT2 and that this role is conserved in cell types where centralspindlin has been observed at MT plus-ends during cytokinesis. Once ECT2 is recruited to a CS-TIP contact point, more of the GEF becomes locally amplified via an unknown process that may involve a poorly understood positive feedback mechanism (Bement et al., 2015; Graessl et al., 2017; Tyson et al., 2003).

Due to the high spatial and temporal resolution of the TIRF-based imaging assays we employed, both MT plus-tip-proximal and midzone-derived RhoA activation signals could be observed in Rhotekin-expressing cells. The CS-TIPs and midzone-derived RhoA activation signals typically ‘arrived’ simultaneously, although CS-TIP-based signaling sometimes preceded evident midzone-derived RhoA activation (Figure 2D). The observation of distinct RhoA activating signals is in accordance with the notion that the MT plus-tip-based pathway described here functions concurrently with a furrow positioning signal(s) that can ‘act at a distance’ to activate RhoA and myosin such as has been described in cells in which astral MTs have been depleted (von Dassow et al., 2009). Reliance on a diffusional midzone-derived signal to the cortex could be problematic for the fidelity of cytokinesis as the nature of such a signal would be inexorably linked to variable features of the midzone MT array such as distance from the cortex as well as the proper localization and levels of gradient-producing midzone components. In contrast, we propose that CS-TIP-based signaling should provide more consistent spatial cues than midzone-mediated signals since it would depend primarily on MT polymerization rates, lengths, and the cortical orientation of astral MTs (emanating from centrosomes, spindle poles, or satellite astral brushes) – features that are highly reproducible between bipolar anaphase spindles. Nonetheless, given the essential nature of cytokinesis, most cells likely employ both plus-tip-based (action via direct physical contact) and midzone-derived (action at a distance) RhoA activation pathways.

Interestingly, in drug-triggered monopolar cytokinesis, numerous cytokinesis regulators (PRC1, MKLP1, ECT2, Plk1) localized to MT tips (Hu et al., 2008). The work provided further evidence that spindle bipolarity is not required for assembly of a MT array capable of positioning the cleavage furrow even in the absence of conventional midzones with antiparallel overlap MTs. We observed a comparable localization pattern of key cytokinesis factors on astral MT plus-tips in the unperturbed bipolar cytokinesis events studied here. Thus, our proposed mechanism of RhoA activation by regulatory factors on MT plus-tips is likely a key contributor to furrow positioning in both experimentally-induced monopolar cytokinesis and naturally occurring bipolar cytokinesis.

A midzone derived ABK gradient has been proposed to contribute to proper spatial regulation of RhoA activation (Fuller et al., 2008; Tan and Kapoor, 2011); and in addition to ABK, Polo kinase has also been shown to induce furrow initiation by recruiting ECT2 to the central spindle (Petronczki et al., 2007), indicating that both ABK and Polo kinase activities are required for furrow positioning. We show that CS-TIP assembly and signaling was unaffected by treatments that mis-localize midzone pools of active ABK (Figure 3, F and G) and Polo kinase (Figure 3E). While global inhibition of either ABK or Polo kinase leads to severe cytokinesis defects, furrow positioning and cytokinesis was not evidently impacted in this study by mis-localization of midzone pools of either kinase (Figure 3). Thus, midzone-based ABK and Polo kinase activity gradients are either dispensable for furrow positioning in Drosophila or redundant with MT plus-end--based furrow positioning cues and/or other pathways.

Our conclusion that MT plus-tips activate RhoA by targeting a cortical pool of ECT2 are generally consistent with a recent study in which artificial targeting of ECT2 to the plasma membrane rescued cytokinesis defects of ECT2 depletion. Although this study also found that furrow positioning and ingression were unaffected in cells expressing an ECT2 mutant that failed to localize to midzones as a result of reduced binding to MgcRacGAP (Kotýnková et al., 2016). On its face, this finding contradicts our proposal that recruitment of cortical ECT2 by CS-TIPs requires centralspindlin - specifically RacGAP50C (Figure 6D); however, the ECT2 mutant developed in this study exhibited reduced, but still detectable enrichment at the equatorial cortex and, importantly, MgcRacGAP was still required for successful cytokinesis in cells expressing the ECT2 mutant. In light of our model, it is possible that a cortical population of the ECT2 mutant used in this study may still bind to MgcRacGAP with sufficient affinity to support furrowing upon MT contact (Basant and Glotzer, 2018).

Successful cytokinesis requires that the cleavage furrow not only be established in the correct position, but also maintained over tens of minutes. During our high-resolution imaging, we often observed ‘’satellite MT arrays’’ that were in close proximity to the cortex and assembled after contractility had been established. Near the equatorial cortex, midzone satellites were built from equatorial astral MTs that became bundled and forced into anti-parallel MT arrays by the contractile machinery. Importantly, once contractility was established, the midzone satellites maintained direct physical contact with the cortex such that any midzone-based cues would no longer have to act over long distances to signal to the cortex. The other type of satellite MT array - astral brushes – also assembled after contractility had begun, but via apparent branching MT nucleation from astral MTs. The astral brushes, in which MT plus-ends were oriented toward the cortex, locally activated RhoA by creating a high density of new CS-TIPs. We posit that midzone and astral brush MT satellites could fulfill two critical roles during cytokinesis as autonomous signaling centers that can 1) act at distances far from their analogous structures (midzones and poles/centrosomes) in the spindle proper, and 2) maintain contractility after it has been established. It is worth noting that in large cells amplification of an astral-like MT array (Helmke et al., 2013; Ishihara et al., 2016; Mitchison et al., 2012), likely similar to the satellite astral brushes observed here (Figure 4, A-C), would be required for MT tips to act over distances of tens to hundreds of microns.

We observed that depletion of the tip-tracking protein EB1 resulted in furrow ingression defects and increased incidence of cytokinesis failure. Our results are generally consistent with earlier reports where injection of an inhibitory antibody against EB1 or a dominant negative EB1 truncation into dividing sea urchin eggs delayed furrow ingression (Strickland et al., 2005), and the finding that EB1 and EB3 co-depletion in HeLa cells resulted in a 5-fold increase in cytokinesis failure (Ferreira et al., 2013). However, differences in the extent of mitotic phenotypes in more recent knockout studies of EB proteins in HeLa cells have been reported. (McKinley and Cheeseman, 2017; Yang et al., 2017). McKinley et al. reported spindle abnormalities in ∼35% of mitotic cells upon double knockout of EB1 and EB3; however, Yang et al. reported seemingly normal mitotic spindles in cells where all three EB proteins (EB1, EB2, and EB3) were knocked out. Since these studies didn’t focus intently on cytokinesis it would be worthwhile to further investigate the contribution of EB1 to cytokinesis keeping in mind that redundant pathways likely exist and that the relative contribution of these pathways may differ between model systems and cell types. Experimental conditions that isolate the MT plus-end-based pathway from redundant pathways in EB depleted/knockout cells would be informative.

Xenopus laevis MgcRacGAP localized to MT plus-ends in embryonic cells during cytokinesis in a C-terminal SxIP motif-dependent manner (Breznau et al., 2017). Interestingly, the SxIP motif identified in X. laevis MgcRacGAP was not conserved beyond Xenopus species despite the fact that centralspindlin plus-tip localization occurs in flies and human cells. Thus, there are likely other EB1 interfaces in the centralspindlin complex. In this study, we report that deletion of the C-terminus of RacGAP50C compromised MT plus-tip localization and we identified a conserved motif (hxxPTxh) in RacGAP50C that also reduced the plus-end localization of RacGAP50C when deleted. The hxxPTxh motif in RacGAP50C is similar to recently identified EB1-interaction motifs in budding yeast Kar9, fission yeast Dis1/TOG, and the Drosophila chromokinesin NOD (Kumar et al., 2017; Manatschal et al., 2016; Matsuo et al., 2016; Ye et al., 2018). The characteristics of centralspindlin and NOD have notable similarities and differences. Both NOD and centralspindlin localize to polymerizing MT plus-ends and contain motor domains, albeit it is noteworthy that NOD exhibited evident MT plus-end directed motility along MTs (Ye et al., 2018) while motile centralspindlin puncta were only occasionally observed in this study. The region in NOD containing the non-conventional EB1-interaction motifs and the NOD motor domain were each necessary but not sufficient for tip-tracking in cells. Similarly, it is possible that both the Dm MKLP1 motor domain and EB1 interaction motifs in RacGAP50C are required to support robust centralspindlin plus-tip-localization (Figure 6D). Finally, the hxxPTxh motifs in NOD are in a region of the protein that is predicted to be disordered, which is ideal for supporting higher affinity interactions with EB1. To the contrary, the hxxPTxh motif in RacGAP50C is contained within a structured region of the protein’s GAP domain. While our data show that this motif contributes to MT plus-tip localization we cannot conclude that it directly interacts with EB1 nor do we rule out that other EB1 interaction motifs exist in the C-terminus of RacGAP50C or in other CS-TIP components. Thus, a more thorough investigation of EB1-CS-TIP component interactions is warranted.

The observation that Dm MKLP1, RacGAP50C, ABK, and Polo kinase localize to the plus-ends of polar and equatorial astral MT tips before becoming patterned onto equatorial astral MTs raises an important question: how do CS-TIPs selectively disassemble from the polar astral MTs? Since we have shown that CS-TIPs activate cortical contractility, any patterning mechanism that disassembles polar CS-TIPs would inherently contribute to the polar relaxation phenomenon. A recent study in C. elegans found that TPXL-1 (C. elegans TPX-2 homologue)-based activation of Aurora A kinase contributed to the clearance of anillin and F-actin (Mangal et al., 2018) from the polar cortex. A polar Aurora A kinase activity gradient that regulates kinetochore-MT attachments exists in Drosophila and mammalian cells (Chmátal et al., 2015; Ye et al., 2015) although its role in cytokinesis is unclear. In Drosophila, a kinetochore based PP1 phosphatase activity gradient derived from a PP1-SDS22 complex has been proposed to regulate polar relaxation (Kunda et al., 2012; Rodrigues et al., 2015). It is entirely possible that phosphatase and/or kinase activity gradients in Drosophila contribute to the loss of CS-TIPs from polar astral MTs. Alternatively, a chromosome-based RanGTP gradient has been shown to locally reduce Anillin at the cortex (Kiyomitsu and Cheeseman, 2013), and could also impede CS-TIPs-based signaling from the polar astral MTs as the chromosomes approach the polar cortex in late anaphase, but prior to nuclear envelope reformation. While we presently favor the hypothesis that polar- or chromosome/kinetochore-derived activity gradients regulate CS-TIP patterning, a detailed investigation is necessary to resolve the question of how polar CS-TIPs are selectively disassembled as cells progress through cytokinesis.

The Drosophila MKLP1 gene (known as Pavarotti in Drosophila, CG1258) and ECT2 (known as Pebble in Drosophila, CG8114) were PCR amplified from cDNA clones RE22456 and SD01796 respectively with a 5′ SpeI site and a 3′ XbaI site. The resulting PCR products of the individual genes were then inserted into the 5′ SpeI and 3′ XbaI sites of the pMT/V5 His-B vector (Invitrogen) containing in-frame EGFP gene at the 3’ end, cloned between 5’ XbaI and 3’ SacII sites, and the CENPC promoter at the 5’ end, cloned between single Kpn1 site. The regulatory light chain of the non-muscle type two myosin (Spaghetti squash in Drosophila, CG3595) was amplified from the cDNA clone LD14743. The resulting PCR product was cloned between 5’ SpeI and 3’ XbaI of the pMT vector containing in-frame RFP gene at the 3’ end and Mis12 promoter at the 5’ end. EGFP-Rhotekin was PCR amplified from the pCS2-eGFP-rGBD plasmid (gift from the Bement lab (UW-Madison) via Wadsworth lab (UMass Amherst)). The resulting PCR product was cloned between 5’ XhoI and 3’ApaI sites of the pMT vector containing CENPC promoter at the 5’ end cloned between single KpnI site. A second version of Rhotekin was also created in the pMT vector where we swapped the EGFP fluorophore with RFP. Polo kinase under its own promoter was cloned between 5’ KpnI and 3’ SpeI sites of the pMT vector containing in-frame EGFP gene, cloned between 5’ SpeI and 3’ ApaI sites. The RacGAP50C (also known as Tumbleweed in Drosophila, CG13345) gene was amplified from genomic DNA of Drosophila. The resulting PCR product was cloned in a similar fashion as described for MKLP1. A four-amino acid deletion (PTIV^533-536) in the RacGAP50C gene was generated by Q5 site directed mutagenesis kit (NEB, MA, USA), following manufacturer’s protocol. C-terminal truncation of the RACGAP50C gene was created by amplifying amino acids (1-526) from the full length RacGAP50C (a.a. 1–625) gene. The resulting PCR product was inserted into the 5′ SpeI and 3′ XbaI sites of the pMT vector containing in-frame EGFP gene at the 3’ end, and the CENPC promoter at the 5’ end. EB1-RacGAP50C-GFP fusion was created by inserting EB1 PCR product (amplified from EB1 cDNA) into the SpeI site of the existing RacGAP50C-GFP construct. EB1 fusions to the RacGAP50C mutants (∆C and ∆PT) were created in a similar way. See Table 1 for the primers used in the cloning.

Drosophila Schneider (S2) cell line was obtained from American Type Culture Collection (ATCC). Identity of the cell line was confirmed by genomic DNA isolation and PCR amplification of many genes, and cells were periodically screened for mycoplasma contamination. All cell lines were grown at 25°C in Schneider’s medium (Life Technologies, Carlsbad, CA), containing 10% heat-inactivated fetal bovine serum (FBS) and 0.5X antibiotic-antimycotic cocktail (Gibco, Life Technologies, NY, USA). Cell lines were generated by transfecting the DNA constructs containing the gene of interest with Effectene transfection reagent (Qiagen, Hilden, Germany), following the manufacturer’s protocol. Four days after transfection, expression of EGFP/RFP tagged proteins was checked by fluorescence microscopy. To make a stable cell line, cells were selected in the presence of Blasticidin S HCl (Thermo Fisher Scientific, Waltham, MA) and/or Hygromycin (Sigma-Aldrich) until there was no observable cell death. Thereafter, cell lines were either frozen down or maintained in the S2 media @ 25°C without Blasticidin and/or Hygromycin B. Transient cell lines were imaged after 4–5 days of transfection, and they were not subjected to antibiotics selection. ABK-GFP and mCherry-α-tubulin cell line, was a generous gift from Dr. Eric Griffis, University of Dundee.

Around 500 base pairs of DNA templates containing T7 promoter sequence at the 5’ end for Dm MKLP2 (also known as Subito in Drosophila, CG12298), EB1 (CG3265), and Dm Kinesin-4 (also known as Klp3A in Drosophila, CG8590) were generated by PCR. Double-stranded RNAs (dsRNAs) were synthesized from the respective DNA templates at 37°C using the T7 RiboMax Express Large-Scale RNA Production System (Promega Corp., Madison, WI), following the manufacturer’s protocol. For RNAi experiments, cells at about 25% confluency were incubated in a 35 × 10 mm tissue culture dish for an hour. Thereafter, media was carefully aspirated off the dish and 1 ml of serum-free S2 media containing 20 μg of dsRNA was added to the dish. After 1 hr, 1 ml of fresh S2 media containing FBS was added to the dish and incubated for 2–4 days at 25°C. Depletion of endogenous RacGAP50C was achieved by PCR amplification of 270 base pairs of 3’ UTR and 200 base pairs of the coding sequence. Depletion of the endogenous EB1 was carried out by amplification of 600 bp of the EB1 3’UTR from a cDNA clone (FI12926 – DGRC). Following PCR amplification, double-stranded RNAs (dsRNAs) were synthesized as described above, and 20 μg of dsRNA was used for RNAi experiments. Cells were analyzed after 4 days of incubation by quantitative western blotting and binucleates counting. See Table 2 for primer sequences.

Drosophila S2 cells were plated on an acid-washed Concanavalin A (Sigma-Aldrich) coated coverslips for 30 min, then cells were quickly rinsed with 1X BRB80 buffer. Subsequently, cells were fixed with 10% paraformaldehyde for 10 min. Cells were then permeabilized with phosphate-buffered saline (PBS) containing 1% Triton X-100 for 8 min, rinsed three times with PBS plus 0.1% Triton X-100, and blocked with 5% boiled donkey serum ((Jackson Immuno Research Laboratories, Inc, West Grove, PA) for 60 min. All primary antibodies were diluted in 5% boiled donkey serum. Anti-Phospho-aurora A/B/C (Cell Signaling Technology, Danvers, MA) was used at a concentration of 1:1000, and anti-tubulin antibody (DM1A; Sigma-Aldrich) at 1:2000. All secondary antibodies (Jackson Immuno Research Laboratories, Inc, West Grove, PA) were diluted in boiled donkey serum at 1:200. After secondary antibody incubation, coverslips were washed three times with PBS plus 0.1% Triton X-100, followed by incubation with 4′,6′-diamidino-2-phenylindole (DAPI) at a concentration of 1:1000 for 10 min, and three additional washes with PBS plus 0.1% Triton X-100. Coverslips were sealed in a mounting media containing 20 mM Tris, pH 8.0, 0.5% N-propyl gallate, and 90% glycerol. Four-color Z-series consisting of ∼30 planes at 0.2 μm intervals were acquired for GFP, RFP, Cy5, and DAPI channels on a TIRF-Spinning Disk Confocal system assembled on a Nikon Ti-E microscope equipped with a 100 × 1.4 NA DIC Apo Oil immersion objective, two Hamamatsu ORCA-Flash 4.0 LT digital CMOS camera (C11440), four laser lines (447nm, 488nm, 561nm, and 641nm), and a MetaMorph software. Fluorescence intensities were obtained by drawing a 50 × 100 grid manually around the maximum-intensity projection of the Z-series images. The local background was estimated by placing the 50 × 100 region to a nearby place. To obtain the ratio intensities, identical regions were drawn manually on the Dm MKLP1-EGFP channel and transferred to a phosphorylated ABK (pABK) channel.

Equal amount of proteins was loaded into a 10% or 8% SDS-PAGE gel. After running the gel, proteins were transferred to a nitrocellulose membrane using the Trans-Blot Turbo transfer system (Bio-Rad Laboratories, Inc, Hercules, CA) for 10 min. Subsequently, the membranes were incubated in 5% milk (V/V made in Tris-buffered saline with 0.1% Tween) for one hour. Following blocking, the membranes were incubated with their respective primary antibodies (α-EB1 - 1:5000, α-Klp3A - 1:500, α-Tubulin – 1:10,000, α-Pavarotti/Dm MKLP1 – 1:5,000, α-RacGAP50C – 1:4,000) for 1 hr, followed by three times wash and secondary antibody incubation at 1:5000 dilution for 1 hr. Following secondary antibody incubation, membranes were washed three times in Tris-buffered saline with 0.1% Tween (TBST) and imaged with a G:BOX system controlled by GeneSnap software (Syngene, Cambridge, U.K.). Images were further quantified to estimate the knock down efficiency with the help of Fiji/Image J software (Schindelin et al., 2012). To obtain the intensity values, identical regions were drawn over all the bands and their integrated intensity was recorded. Intensity values were normalized to their respective loading controls (Tubulin) to estimate the knockdown efficiency.

Around 500 to 700 µl of cells, expressing the gene of interest were seeded on a 35 mm glass bottom dish with 20 mm bottom well (Cellvis, CA, USA) coated with Concanavalin A for 30–60 min. Before imaging, 2 ml of fresh S2 media containing FBS was added to the dish. Live-cell TIRF movies were acquired on a Nikon Ti-E microscope equipped with a 100 × 1.4 NA DIC Apo Oil immersion objective, a Hamamatsu ORCA-Flash 4.0 LT digital CMOS camera (C11440), four laser lines (447nm, 488nm, 561nm, and 641nm), and a MetaMorph software. Live-cell TIRF movies of all the constructs (MKLP1-EGFP and Tag-RFP-T-α-tubulin; RacGAP50C-EGFP and Tag-RFP-T-α-tubulin; Polo Kinase-EGFP; ABK-GFP and Tubulin mCherry; EGFP-Rhotekin and Tag-RFP-T-α-tubulin; Tag-RFP-T-Rhotekin and Dm MKLP1-EGFP; Tag-RFP-T-Rhotekin and MRLC-EGFP; Dm ECT2-EGFP and Tag-RFP-T-α-tubulin; EB1-RacGAP50C-EGFP and Tag-RFP-T-α-tubulin; EB1-∆C-RacGAP50C-EGFP and Tag-RFP-T-α-tubulin; EB1-∆C-RacGAP50C-EGFP and Tag-RFP-T-Rhotekin; EB1-∆PTIV-RacGAP50C-EGFP and Tag-RFP-T-α-tubulin) were acquired at the following settings: 561 exposure time 200 ms, 488 exposure time 200 ms, temperature 25°C, and frame acquisition rate 5 s. To observe cytokinesis defects, overnight imaging was performed on control and EB1 depleted cells. Cells were seeded on Concanavalin A coated 35 mm glass bottom dish as described above. A total of 76 time-points were acquired for 50 stage positions at the interval of 4 min @ 25°C on a Nikon Eclipse Ti-E microscope (Nikon, Tokyo, Japan) equipped with 40X/1.30 Oil Nikon Plan Fluor DIC N2 objective (Nikon), Andor iXon₃ EMCCD camera (Andor Technology, Belfast, U.K.), and Metamorph software (Molecular Devices, Sunnyvale, CA). Images were further analyzed with the help of MetaMorph software.

Cells were seeded on Concanavalin A (Con A) coated glass-bottom dishes. After 30–45 min. of incubation, cells were imaged on a TIRF-Spinning Disk system assembled on an Eclipse Ti-E inverted microscope (Nikon), equipped with a Borealis (Andor Technology, Ltd., Belfast, U.K.) retrofitted CSU-10 (Yokogawa Electric Corp., Tokyo, Japan) spinning disk head and two ORCA-Flash4.0 LT Digital CMOS (Hamamatsu Corp., Bridgewater, NJ), using a 100 × 1.49 NA Apo differential interference contrast objective (Nikon). Metamorph software was used to control the imaging system. For correlative TIRF-spinning disk confocal imaging, cells were followed by time-lapse TIRF imaging; once the cell had assembled CS-TIPs (based on localization of Dm MKLP1-GFP on MT +TIPs) the system was switched to spinning disk mode and 0.2 µm confocal z-sections were taken. All quantifications and 3D-reconstructions were made with the help of Metamorph software.

Cells expressing Dm MKLP1-EGFP and Tag-RFP-T-α-Tubulin; Polo-EGFP; Rhotekin-EGFP and Tag-RFP-T-α-Tubulin; Rhotekin-RFP and Dm MKLP1-EGFP; Dm ECT2-EGFP and Tag-RFP-T-α-Tubulin were subjected to ABK or Polo inhibition. Cells were seeded on Concanavalin A coated 35 mm glass bottom dish with 20 mm bottom well (Cellvis, CA, USA) for 30–60 min. Before imaging, 2 ml of fresh S2 media containing FBS was added to the dish. Live-cell TIRF movies were acquired on a Nikon Ti-E microscope as described above. The whole field of view under the microscope was scanned to find mitotic cells. Mitotic cells expressing the gene of interest were imaged at the interval of 20–30 s. until the anaphase onset. After anaphase onset, when the CS-TIPs were visible, S2 media in the glass bottom dish was exchanged with S2 media containing Binucleine 2 (40µM) for ABK inhibition or BI 2536 (1µM) for Polo inhibition. Thereafter, imaging was continued at 5 s. intervals. Images were further analyzed with the help of Metamorph software (CA, USA).

GFP/RFP intensities before and after the addition of inhibitor (Binucleine two or BI 2536) were quantified to estimate the effects of inhibitor on CS-TIPs patterning and signaling. Identical regions were drawn over the GFP/RFP puncta to measure the fluorescence intensity. The local background was estimated by placing the same region to a nearby place. Background intensity values were subtracted from GFP/RFP values to estimate the actual GFP/RFP fluorescence intensity. Histograms and dot plots were generated using Prism software (Graphpad, CA, USA), and figures were assembled using Adobe Illustrator software (Adobe, CA, USA). Statistical analysis was performed using Prism software (Graphpad, CA, USA).

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

Thank you to Margaux Audett, UMass, for help with generating reagents. We acknowledge Patricia Wadsworth, UMass and Alex Mogilner, NYU for sharing insightful conversations. We are also tremendously grateful to Eric Griffis, University of Dundee, for the kind gift of ABK cell line; Bill Bement, UW-Madison for the kind gift of Rhotekin plasmid via Patricia Wadsworth; Stephen Rogers, UNC Chapel Hill for the kind gift of α-EB1A antibody; Jonathan Scholey, UC Davis for the kind gift of α-Klp3A antibody; and David Glover, University of Cambridge for his generosity in sharing α-Pavarotti and RacGAP50C antibodies. This work was supported by an NIH grant (GM107026) to TJM.

1. Received: June 7, 2018
2. Accepted: February 10, 2019
3. Accepted Manuscript published: February 13, 2019 (version 1)
4. Accepted Manuscript updated: February 15, 2019 (version 2)
5. Version of Record published: March 4, 2019 (version 3)

© 2019, Verma and Maresca

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.