Microtubule nucleation promoters Mto1 and Mto2 regulate cytokinesis in fission yeast

Abstract

Microtubules of the mitotic spindle direct cytokinesis in metazoans but this has not been documented in fungi. We report evidence that microtubule nucleators at the spindle pole body help coordinate cytokinetic furrow formation in fission yeast. The temperature-sensitive cps1-191 strain (Liu et al., 1999) with a D277N substitution in β-glucan synthase 1 (Cps1/Bgs1) was reported to arrest with an unconstricted contractile ring. We discovered that contractile rings in cps1-191 cells constrict slowly and that an mto2^S338N mutation is required with the bgs1^D277Nmutation to reproduce the cps1-191 phenotype. Complexes of Mto2 and Mto1 with γ-tubulin regulate microtubule assembly. Deletion of Mto1 along with the bgs1^D277N mutation also gives the cps1-191 phenotype, which is not observed in mto2^S338N or mto1Δ cells expressing bgs1⁺. Both mto2^S338N and mto1Δ cells nucleate fewer astral microtubules than normal and have higher levels of Rho1-GTP at the division site than wild-type cells. We report multiple conditions that sensitize mto1Δ and mto2^S338N cells to furrow ingression phenotypes.

INTRODUCTION

Animals, fungi, and amoebas use an actomyosin contractile ring to divide. Coordinating nuclear division and cytokinesis is required to avoid errors during chromosome segregation, but the mechanisms are incompletely understood.

Cells of the fission yeast Schizosaccharomyces pombe assemble a contractile ring from precursors called nodes over 10–15 min following separation of the spindle pole bodies (SPBs) at the onset of anaphase at 23°C (Vavylonis et al., 2008). Furrow ingression starts 35 min after SPB separation (Wu et al., 2003), even if mutations slow the assembly of the ring (Coffman et al., 2009; Roberts-Galbraith et al., 2010; Tebbs and Pollard, 2013; Wang et al., 2014; Laplante et al., 2015). If ring assembly takes longer than 35 min, ingression begins immediately (Chen and Pollard, 2011; Tebbs and Pollard, 2013; Li et al., 2016).

Initiation of furrow ingression by S. pombe depends on an intact contractile ring, a signal from the cell cycle clock and septum synthesis. The enzyme β-glucan synthase 1 (Bgs1) concentrates at the equator where it synthesizes the primary septum (Arellano et al., 1996; Cortés et al., 2007) and helps anchor the ring (Arasada and Pollard, 2014; Davidson et al., 2015). Bgs4 (activated by Rho1) and Ags1 (activated by Rho2) synthesize the secondary septum (Arellano et al., 1996; Calonge et al., 2000; Cortés et al., 2005, 2012). Rho1 activity must decrease after septation to allow cell separation (Nakano et al., 1997).

Tension in the ring promotes septum deposition and maintains the circularity of the pore (Thiyagarajan et al., 2015; Zhou et al., 2015), so mutations compromising contractility slow furrow ingression (Pelham and Chang, 2002; Proctor et al., 2012; Tebbs and Pollard, 2013; Laplante et al., 2015; Li et al., 2016). After ∼50% ingression, septum synthesis alone can complete division (Proctor et al., 2012). Turgor pressure resists ingression, so mutations that reduce pressure or high osmolarity of the medium favor ingression (Proctor et al., 2012; Morris et al., 2019).

Liu et al. (Liu et al., 1999) used chemical mutagenesis to create the S. pombe strain cps1-191 and observed that cells arrested at 36°C with two nuclei and an unconstricted cytokinetic ring. They concluded that colonies did not grow at 36°C due to failed cytokinesis. They found that cps1⁺/bgs1⁺complemented temperature sensitivity of the strain, identified the D277N substitution in the mutated gene, and named the strain cps1-191 (Liu et al., 1999). Time-lapse microscopy of the cps1-191 strain confirmed that the nuclei separate normally but actomyosin rings remain intact and unconstricted for an hour at 36°C (Arasada and Pollard, 2014). Many studies have used the cps1-191 strain to generate cells with nonconstricting actomyosin rings (Pardo and Nurse, 2003; Venkatram et al., 2005; Yamashita et al., 2005; Wachtler et al., 2006; Loo and Balasubramanian, 2008; Roberts-Galbraith et al., 2009; and others).

We found that rings in cps1-191 cells actually constrict very slowly at 36°C and that cells with the bgs1^D227N mutation die from lysis rather than cell cycle arrest. Surprisingly, we found that the cps1-191 constriction phenotype depends on a second point mutation in the gene for the γ-tubulin regulator Mto2, implicating microtubules in the process that drives furrow ingression.

S. pombe has several types of microtubule organizing centers (MTOCs; Sawin and Tran, 2006). During interphase, multiple MTOCs localize along microtubule bundles (Janson et al., 2005). During mitosis, interphase microtubules disassemble and the SPBs nucleate spindle and astral microtubules. During cytokinesis Myp2 recruits MTOCs to the equator (Samejima et al., 2010), where they nucleate the postanaphase array (PAA) of microtubules. These MTOCs depend on Mto1 and, to a lesser extent, Mto2. In the absence of Mto1, no MTOCs form beyond the SPB, which only nucleates microtubules from its inner face (Zimmerman and Chang, 2005). In biochemical experiments, γ-tubulin requires the Mto1/Mto2 complex and MOZART1 homologue Mzt1 to nucleate microtubules (Leong et al., 2019). We find that Mto1/Mto2 complex mutants nucleate fewer astral microtubules and report one consequence: higher than normal Rho1-GTP activity at the cleavage site. Our findings reveal a new link between microtubules and furrow ingression in fission yeast.

RESULTS AND DISCUSSION

Cytokinetic furrows ingress slower in the cps1-191 strain than in cells with the bgs1 ^D277N mutation

Long time-lapse movies at the restrictive temperature of 36°C of the cps1-191 strain with genome-encoded Rlc1-tdTomato (regulatory light chain for both isoforms of myosin-II, Myo2 and Myp2) revealed that the actomyosin ring constricted ∼30-fold slower (median 0.02 μm/min) than in wild-type cells (median 0.62 μm/min; Figure 1, A and B). No rings detached from the plasma membrane (Arasada and Pollard, 2014; Laplante et al., 2015; Cheffings et al., 2019), so Rlc1-tdTomato constriction reflects furrow ingression. Thirty percent of assembled contractile rings slid along the membrane of cps1-191 cells at 36°C, as reported (Arasada and Pollard, 2014; Cortés et al., 2015). Shifting cps1-191 cells from the permissive (25°C) to restrictive (36°C) temperature on the microscope showed that more than 30 min at 36°C before SPB separation was required to compromise furrow ingression (Supplemental Figure S1A).

FIGURE 1:

Both the bgs1^D277N and the mto2^S338N mutations are required to cause the cps1-191 constriction phenotype in a wild-type background. (A) Kymographs of inverted-contrast, maximum-intensity projected images of contractile rings in strains with Rlc1-tdTomato at 36°C. Wild-type cells were imaged at 1-min intervals, and bgs1^D277N and cps1-191 cells were imaged at 5-min intervals. The kymograph of the wild-type cell is displayed (left subpanel) as acquired and (right subpanel) rescaled to match the timescale of the kymographs (other panels) of the cps1-191 and six different bgs1^D277N strains. Horizontal scale bars = 15 min, vertical scale bar = 1 μm. (B) Rates of cytokinetic ring constriction measured from a subset of kymographs in A. The data are not normally distributed, so the median and first and third quartiles are indicated by black bars; n ≥ 55 cells. (C) Log₁₀-transformed cytokinetic ring constriction rates of cells carrying the bgs1^D277N mutation measured from kymographs as in A. The median and first and third quartiles are indicated by black bars; n ≥ 57 cells. Significance was determined by Welch’s ANOVA followed by a Tukey post-hoc test (p < 0.05). (D) Cytokinetic ring constriction rates of cells carrying bgs1⁺ measured from kymographs as in A. The median and first and third quartiles are indicated by black bars. No significant differences were detected by Welch’s ANOVA. (E) Cumulative distribution plots showing accumulation of cells with rings that have (•) assembled, (■) initiated constriction, and (▲) completed constriction in wild-type and bgs1⁺cells with various additional mutations at 36°C. n ≥ 71 cells for C and D.

Furrow ingression was threefold faster (median 0.06 μm/min) in a strain with the bgs1^D277N mutation in a wild-type background than in cps1-191 cells. Both bgs1^D277N and cps1-191 cells have similar growth defects at 36°C, consistent with the 2:2 segregation for this phenotype (Supplemental Figure S1B; Liu et al., 1999). No bgs1⁺ cells lysed during imaging, but both bgs1^D277N and cps1-191 cells lysed frequently (Supplemental Figure S1D), explaining the growth defect at 36°C on solid medium. The lysis frequency varied substantially between replicates, suggesting that this phenotype is sensitive to minute environmental differences. The osmotic stabilizer sorbitol partially rescued the growth of cps1-191 and bgs1^D277N cells at 36°C (Supplemental Figure S1C).

The cps1-191 strain carries a large number of mutations

The complete genome sequence of the cps1-191 strain revealed 384 unique mutations not found in the S. pombe reference genome (Wood et al., 2002; Lock et al., 2019). Most (271) were transition mutations as expected from nitrosoguanidine mutagenesis (Supplemental Figure S1E; Balasubramanian et al., 1998). Ten of these transition mutations substituted amino acids in eight genes spanning all three chromosomes, including the reported bgs1^D277N mutation (Supplemental Figure S1F and Table 1). Two of the eight genes encoding substitutions, mto2 and nup40, have roles in cell cycle regulation or cytokinesis (Lock et al., 2019).

TABLE 1:

Mutations in open reading frames in the cps1-191 strain.

Gene	Description	Mutation(s)	Chromosome	Location
tlh1	RecQ type DNA helicase	A1570T	I	5662
nup40	Nucleoporin	T281I	I	1353659
mto2	γ-Tubulin complex linker	S338N	II	494742
cct1	Chaperonin-containing T-complex α-subunit	R269N	II	2307436
SPBC19G7.04	HMG box protein (predicted)	A211V	II	2349593
bgs1	Linear 1,3-β-glucan synthase catalytic subunit	D277N	II	2356243
mae2	Malic enzyme/malate dehydrogenase	D328G, G332D, N131S	III	277211
ers1	RNA-silencing factor	I209L	III	801197

Mutations in the Mto1/Mto2 complex interact synthetically with the bgs1 ^D277N mutation

To determine whether either or both the mto2 and nup40 mutations affected the rate of furrow ingression in combination with bgs1^D277N, we examined multiple progeny from crosses of cps1-191 strains already confirmed by sequence analysis to contain the bgs1^D277N mutation. Combining the bgs1^D277N and mto2^S338N mutations in a wild-type background reproduced the slow constriction rate of cps1-191 (Figure 1, A and C, and Supplemental Table S1). Ingression rates of bgs1^D277N nup40^T281Idouble mutants were no different from bgs1^D277N alone.

The Mto2 C-terminus is highly phosphorylated (Borek et al., 2015), but neither the mto2^S338C mutation, which precludes phosphorylation of S338, nor the [24A] strain with 24 phosphorylated serines mutated to alanines (Borek et al., 2015), constricted as slowly as the bgs1^D277N mto2^S338N or cps1-191 mutants at 36°C (Figure 1, A and C). Cytokinetic rings in both strains were so stable that we never observed one disassemble.

Mto2 regulates MTOCs other than SPBs in S. pombe through interactions with its binding partner Mto1 and γ-tubulin (Sawin and Tran, 2006). Neither mto1⁺ nor mto2⁺ are essential genes, but mutations of mto1⁺ are more severe than mutations of mto2⁺ (Sawin et al., 2004; Venkatram et al., 2005). In a bgs1⁺ background with Rlc1-tdTomato to label contractile rings and Pcp1-mEGFP to label SPBs, all mutant strains of mto1 and mto2 ingressed furrows at the wild-type rate with cytokinesis timelines similar to wild-type cells (Figure 1, D and E). Deletion of Mto1 in the bgs1^D277N strain slowed furrowing to the rate observed in cps1-191 cells, while bgs1^D277N mto2Δ cells furrowed at an intermediate rate (Figure 1C). Therefore, the bgs1^D277N mutation sensitizes cells to a compromised Mto1/Mto2 complex. One mechanism might be reduced recruitment of Bgs1 to the division site, but bgs1⁺mto1Δ and bgs1⁺mto2^S338N cells recruited Bgs1 molecules to the equator normally (Supplemental Figure S1G).

The finding that bgs1^D277N mto2^S338N has a more severe phenotype than bgs1^D277N mto2Δ suggests that the mto2^S338N mutation may exert a dominant-negative effect on the function of Mto1 that is more severe than the loss of Mto2. We attempted to assess this hypothesis by ectopically introducing the mto2^S338N coding sequence in addition to the endogenous mto2⁺ gene, but progeny from the crosses exhibited germination defects.

The mto1Δ and mto2 ^S338N mutations disrupt astral microtubules

Wild-type cells nucleated multiple, short-lived astral microtubules from each SPB with a median duration of 20 s (Figure 2A and Supplemental Figure S1H), while mto2^S338N cells produced fewer astral microtubules and most mto1Δ cells nucleated no astral microtubules, as reported (Figure 2, B and C; Sawin et al., 2004). We measured the numbers of microtubules nucleated by SPBs only during anaphase B, to avoid intranuclear “astral” microtubules nucleated earlier (Zimmerman et al., 2004). Immunofluorescence demonstrated similar defects in astral microtubule nucleation in mto1Δ and mto2Δ cells (Sawin et al., 2004; Janson et al., 2005; Samejima et al., 2005). Furthermore, mitotic spindles reached their maximum lengths and disassembled later than normal in both the mto2^S338N and mto1Δ strains (Figure 2D).

FIGURE 2:

The mto2^S338N and mto1Δ mutations perturb astral microtubules and the time of spindle breakage. (A) Time series of inverted-contrast maximum-intensity projected fluorescence micrographs of wild-type cells at 36°C with astral microtubules labeled with GFP-Atb2 (α tubulin). Arrows of the same color mark the same astral microtubule in consecutive frames. Scale bar = 2 μm. (B) Representative fluorescence micrographs of three strains (wild-type, mto2^S338N, and mto1Δ) at 36°C with spindle microtubules labeled with GFP-Atb2. Scale bar = 2 μm. (C) The number of astral microtubules nucleated from a single SPB over the course of mitosis at 36°C in wild-type, mto2^S338N, and mto1Δ cells; n ≥ 21 spindle poles. Significance was determined using pairwise K-S tests with Bonferroni correction. (D, F) Cumulative distribution plots comparing mitotic and cytokinetic outcomes in wild-type, mto2^S338N, and mto1Δ cells with GFP-Atb2 to mark microtubules, Rlc1-tdTomato to mark the cytokinetic ring, and Sfi1-mCherry to mark SPBs at 36°C; n ≥ 21 cells. (D) Time course after time zero of the accumulation of cells with broken mitotic spindles, which occurred later than normal in mto2^S338N and mto1Δ strains (p < 0.05 by pairwise K-S tests with Bonferroni correction). (E) Log₁₀-transformed cytokinetic ring constriction rates in strains with Rlc1-tdTomato at 36°C lacking either the PAA microtubules (mto1-427) or astral microtubules (mto1(1-1085)) in combination with bgs1^D277N. The median and first and third quartiles are indicated by black bars; n ≥ 57 cells. Significance was determined by Welch’s ANOVA followed by a Tukey post-hoc test (p < 0.05). (F) Time course after time zero of the accumulation of cells with rings that have (•) assembled, (■) initiated constriction, and (▲) completed constriction. Ring assembly was delayed in cells carrying GFP-Atb2 when combined with the mto2^S338N and mto1Δ mutations but not wild-type cells (p < 0.05 by pairwise K-S tests with Bonferroni correction; n ≥ 21 cells).

γ-Tubulin-Mto1/Mto2 complexes associated with the cytokinetic ring nucleate PAA microtubules (Samejima et al., 2010). We confirmed that mto1Δ cells do not form PAAs (Sawin et al., 2004) but observed that PAA microtubules formed at the normal time in mto2^S338N cells (Supplemental Figure S1J). This suggested that the loss of astral microtubules, not loss of PAA microtubules, causes a synthetic phenotype with bgs1^D277N. Experiments with mto1 mutants that selectively lack either PAA or astral microtubules (Samejima et al., 2010) supported this hypothesis. Only the mto1(1-1085) mutant lacking astral microtubules exhibited the synthetic effect on constriction rate when combined with the bgs1^D277Nmutation (Figure 2E).

Astral microtubules may serve as signaling platforms or recruit cell cycle regulators to the SPB, but functions of Mto1 and Mto2 independent of microtubule nucleation may contribute to the synthetic interaction with bgs1^D277N. To assess this possibility, we tested the effects of depolymerizing microtubules with carbendazim, but microtubules repolymerized within 1 h of continuous exposure to a high dose (250 μg/ml), so insufficient time was available to measure furrow ingression in bgs1^D277N cells without microtubules.

Other stresses sensitize cells to mutation of mto1 and mto2

Unconventional type-II myosin myp2 (Bezanilla et al., 1997; Okada et al., 2019) recruits Mto1 to the ring to generate the PAA (Samejima et al., 2010), and stressing myp2∆ cells with 0.6 M KCl delays ring constriction, so we tested whether salt has a similar effect on mto1Δ and mto2^S338N cells. The stress of 0.6 M KCl in minimal medium (EMM5S) had similar effects on wild-type, mto2^S338N, and mto1Δ cells, slowing the time courses of cells completing all three steps in cytokinesis and delaying the onset of ring constriction (Figure 3A), although this delay was less for mto1Δ cells than for wild-type and mto2^S338N cells (Figure 3C). The mto1Δ and mto2^S338N phenotypes differ from the myp2Δ phenotype under salt stress (Okada et al., 2019), so the loss-of-function phenotypes differ for the Mto1/Mto2 complex and Myp2. This is further evidence that the PAA is not involved in the phenotype observed in cps1-191 cells.

FIGURE 3:

The mto2^S338N and mto1Δ mutations sensitize cells to stresses. (A–C) Effects of 0.6 M KCl on cytokinesis in wild-type, mto2^S338N, and mto1Δ cells at 25°C. (A) Rates of cytokinetic ring constriction. The median and first and third quartiles are indicated by black bars and significance (p < 0.05) was determined using Welch’s ANOVA followed by a Tukey post-hoc test. (B, C) Time courses of cytokinetic events in cells expressing Rlc1-tdTomato to mark the cytokinetic ring and Pcp1-mEGFP to mark SPBs. Cumulative distribution plots show the accumulation of cells with rings that have (•) assembled, (■) initiated constriction, and (▲) completed constriction in medium with (B) no added KCl or (C) 0.6 M KCl. Dashed gray lines in B and C indicate the time when 50% of cells in B reach each milestone to facilitate comparison. n ≥ 20 cells for A–C. (D–F) Effects of the Rho1-GTP biosensor Pkc1(HR1-C2) on cytokinesis at 36°C. (D) Rates of cytokinetic ring constriction. The median and first and third quartiles are indicated by black bars and significance (p < 0.05) was determined using Welch’s ANOVA followed by a Tukey post-hoc test. (E, F) Cumulative distribution plots comparing mitotic and cytokinetic outcomes in cells (E) without and (F) with the Pkc1(HR1-C2) Rho1 biosensor. Plots show the accumulation of cells with rings that have with (•) assembled, (■) initiated constriction, and (▲) completed constriction. Dashed gray lines in E and F indicate the time when 50% of cells in F reach each milestone to facilitate comparison. n ≥ 21 cells for D–F.

Although the cytokinetic timelines of mto2^S338N and mto1Δ cells were indistinguishable from wild-type cells at 36°C (Figure 1E), expression of GFP-tagged α-tubulin (Atb2) in either strain slightly delayed ring assembly but not in wild-type cells (Figure 2F). GFP-tagged α-tubulin also delayed the completion of ring constriction in mto1Δ cells, substantiating that additional stressors can perturb cytokinesis in cells with the mto1Δ and mto2^S338N mutations.

The septation initiation network is intact in mto1Δ and mto2 ^S338N cells

The septation initiation network (SIN; Hippo in humans and mitotic exit network in budding yeast) promotes formation of the contractile ring and cytokinesis (Simanis, 2015). High SIN activity can even promote septum formation during interphase (Minet et al., 1979). SPBs are hubs for SIN and other cell cycle signals, and proteins localized to SPBs during mitosis are markers for cell cycle progression (Tatebe et al., 2001; Grallert et al., 2004; Simanis, 2015). Therefore, the loss of astral microtubule nucleation associated with mutations of mto1 and mto2 might compromise SIN signaling.

However, our data and the original characterization of cps1-191 cells (Liu et al., 1999) suggest that the activity of the SIN is normal. Early in mitosis, Cdc7 first localizes to both SPBs and then only one SPB during anaphase B (Sohrmann et al., 1998). Liu et al. observed the SIN kinase Cdc7 localized to one SPB in cells with an unconstricted ring. Time-lapse imaging showed that the time courses and numbers of Cdc7-mEGFP molecules at each SPB are normal in mto2^S338N and mto1Δ cells (Supplemental Figure S1, K–N).

Contractile rings disassemble before constricting generating multinucleate cells in strains with low SIN activity (Mitchison and Nurse, 1985; Hachet and Simanis, 2008; Dey and Pollard, 2018), but rings never disassembled during our observations of cps1-191 or bgs1^D277Nmto2^S338N cells, further evidence that the SIN activity is normal. Interphase mto2^S338N cells never had two nuclei (0 of 234 cells), and we confirmed this occurs rarely (6 of 220 cells) in mto1Δ cells (Zimmerman and Chang, 2005). However, both nuclei can migrate to one side of the contractile ring after it forms at the normal time, creating one half of the cell with two nuclei and the other half lacking nuclei.

Cells lacking Mto1 have altered chromatid cohesion and DNA repair, likely due to loss of nuclear movements dependent on microtubules (Zhurinsky et al., 2019), but it is still unclear how chromosomal abnormalities might be communicated to the division machinery independent of the SIN. Answers should come from research on how microtubules influence prerequisites for furrow formation: an intact contractile ring; a signal from the cell cycle clock; and initiation of septum synthesis.

The concentration of Rho1-GTP at the cleavage site is high in mto1Δ and mto2 ^S338N cells

To explore one pathway that might be influenced by defects in astral microtubules, we localized S. pombe Rho1-GTP using the Pkc1(HR1-C2)-mEGFP biosensor (Davidson et al., 2015). This probe, derived from Saccharomyces cerevisiae Pkc1, contains Rho1-GTP binding HR1 domains and a plasma-membrane targeting C2 domain fused to mEGFP (Kono et al., 2012). Active GTP-bound RhoA concentrates in the cell cortex of metazoan cells (Michaelson et al., 2001) through the C-terminal CAAX motif, which is thought to promote activation by GEFs.

The fluorescence of this probe is the same when free or bound to Rho1-GTP, so it can measure local accumulation but not the total active Rho1-GTP. Binding of biosensors can compromise functions of the target protein, so expression of the biosensor may perturb cytokinesis if Rho1-GTP regulates furrow formation. Therefore, we expressed Pkc1(HR1-C2)-mEGFP from the thiamine-repressible 3nmt1 promoter in medium lacking thiamine (i.e., inducing conditions) for only ∼15 h before imaging, the minimum time for maximum expression (Maundrell, 1990). Conveniently, expression from the P3nmt1 promoter was highly variable, giving cells with a range of concentrations of Pkc1(HR1-C2)-mEGFP.

Expression of the biosensor did not affect the time course of contractile ring assembly or constriction rate at 36°C of wild-type, mto2^S338N, or mto1Δ cells, but delayed the times when populations of these cells initiated and completed ring constriction (Figure 3, D–F). The populations of mto2^S338N and mto1∆ cells completed ring constriction later than wild-ype cells expressing the biosensor (Figure 3F). Pkc1(HR1-C2) expression levels were not normally distributed, so we used a Spearman correlation test to show that constriction rates were not strongly correlated with Pkc1(HR1-C2) level (Supplemental Figure S1I).

Pkc1(HR1-C2)-mEGFP concentrates at growing cell tips during interphase and at the cytokinetic furrow at 36°C (Figure 4, A and B, and Supplemental Movie S1) as reported at 25°C (Davidson et al., 2015). However, at 36°C the fraction of cells with Pkc1(HR1-C2)-mEGFP concentrated at one or both ends was higher in wild-type than in mto2^S338N and mto1∆ mutant cells (Figure 4D). Poor growth of bgs1^D277N cells expressing Pkc1(HR1-C2)-mEGFP at 36°C precluded localization of GTP-Rho1. Ectopic concentration of Pkc1(HR1-C2)-mEGFP at other cortical locations was more frequent in mto2^S338N (49%) and mto1Δ (46%) mutants than wild-type (23%) cells (Figure 4C and Supplemental Movie S1), especially in cells with low concentrations at their poles (Figure 4D). Thus, Pkc1(HR1-C2)-mEGFP compromises some cellular functions and cps1-191 cells are particularly sensitive.

FIGURE 4:

The mto2^S338N and mto1Δ mutations disrupt Rho1 signaling. (A–C) Inverted-contrast maximum-intensity projected fluorescence micrographs of cells at 36°C expressing the Rho1 biosensor Pkc1(HR1-C2)-mEGFP and Pcp1-mEGFP to mark SPBs. Scale bars 2 μm. (A) Wild-type cells. Active Rho1 concentrates at poles of interphase cells (arrowheads) and at the equator of cells during cytokinesis and septation. Times are minutes after SPB separation. (B) mto2^S338N and mto1Δ cells exhibiting Rho1 concentration at cell ends during interphase (each left) and cell equators during cytokinesis (each right). (C) Representative images of mto1Δ and mto2^S338N cells during interphase showing fluctuating, ectopic concentrations of the biosensor (arrows) at times denoted in minutes after the start of image acquisition. Ectopic accumulations were observed in 23% of wild-type, 49% of mto2^S338N, and 46% of mto1Δ cells; n ≥ 72 cells; p = 0.007 and 0.006, respectively, from the wild-type population by a chi-squared test with Bonferroni correction for multiple comparisons. (D) The percent of cells exhibiting bipolar, monopolar, or nonpolar growth in an asynchronous population of cells expressing Pkc1(HR1-C2)-mEGFP shifted to 36°C. Asterisk denotes significant difference (p = 0.008) from the wild type as determined by a chi-squared test with Bonferroni correction for multiple comparisons; n ≥ 143 cells. (E) Time course of the integrated density of the Rho1 biosensor at the equator of the cell in A during cytokinesis normalized to interphase value. The main peak (dark blue) is during furrow formation and the short second peak (light blue) is at cell separation. (F, G) Rho1 biosensor signals at the equators of populations of ≥21 cells. Significance (p < 0.05) was determined using Welch’s ANOVA and Tukey’s post-hoc analysis. (F) Peak Rho1 biosensor signal at the equator during cytokinesis and separation. (G) Duration of Rho1 biosensor peak during cytokinesis.

Movie S1.

Download video file ^{(1.1MB, mov)}

Rho biosensor Pkc1(HR1-C2)-mEGFP localization during 1) wild-type S. pombe cytokinesis, 2) mto2S338N interphase, and 3) mto1Δ interphase. Times expressed relative to SPB separation, scales 2 μm.

The Pkc1(HR1-C2)-mEGFP signal at the cleavage site peaked during furrow ingression (Figure 4C), followed by a brief decrease and second peak during cell separation (single light-blue data point in Figure 4C). The normalized peaks of Pkc1(HR1-C2)-mEGFP fluorescence during cytokinesis and at separation were more intense in mto2^S338N and mto1Δ cells than in wild-type cells (Figure 4D). Furthermore, the peak of Pkc1(HR1-C2)-mEGFP fluorescence during cytokinesis was prolonged in mto1Δ cells (Figure 4E). Nevertheless, the timing of cytokinesis was normal in mto1Δ and mto2^S338N cells relative to mitotic events (Figure 1E). An additional insult (e.g., bgs1 mutation, salt stress, Rho1 biosensor, or GFP-Atb2 expression) is required to sensitize these cells and generate a phenotype. More work is required to assess which Rho1 regulators contribute to the observed hyperactivation, how microtubules regulate any of these proteins, and to determine how high Rho1 accumulation influences furrow ingression.

Excess Rho1-GTP may compromise the activities of other GTPases through a mechanism involving GDP-dissociation inhibitors (GDIs). The bgs1^D227N mutation in the wild- type background slows furrow ingression by 90%, and combining the bgs1^D227N mutation with mto1 or mto2 mutations slows furrowing another threefold in spite of normal Bgs1 recruitment to the division site in mto2^S338N and mto1Δ cells. The abnormally thick cell walls and secondary septa in cells depleted of Bgs1 and cps1-191 cells suggest that other glucan synthases are overactive to compensate for the loss of Bgs1 activity (Cortés et al., 2007, 2015; Sethi et al., 2016). In mammalian cells, small GTPases (e.g., Rho, Rac, Cdc42) compete for GDIs, so changes in RhoA activity can affect other GTPases (Boulter et al., 2010). S. pombe has a single Rho GDI, Gdi1, so high Rho1-GTP at the division site of bgs1^D277N, mto2^S338N, or bgs1^D277N mto1Δ cells may release Gdi1 to interact with other GTPases involved in cytokinesis such as Rho2, which stimulates Ags1 to synthesize the secondary septum. Reducing GTP-Rho2 would compromise the ability of Ags1 to compensate for low Bgs1 activity.

CONCLUSIONS

For two decades, the cps1-191 strain has been used to generate S. pombe cells arrested with unconstricting actomyosin rings. We find that this phenotype depends on both the originally identified bgs1^D277N mutation and the mto2^S338N mutation. Caution is advised when interpreting experiments on cps1-191 cells that examine the “arrest” phenotype, particularly if the cps1-191 strain is crossed with other strains. The bgs1 and mto2 loci are ∼2 Mb apart on chromosome II, and we frequently observed recombination events between these genes, so the two mutations can be separated.

Our work demonstrates cross-talk between the proteins that nucleate astral microtubules and the machinery driving furrow ingression. Compromising this communication results in excess active Rho at the cleavage site. RhoA determines the division site in metazoan cells by activating formins to assemble actin filaments, stimulating Rho-associated protein kinase to activate myosin-II, and promoting midbody ring maturation through citron kinase (Pollard, 2017; El-Amine et al., 2019). Other mechanisms position cytokinetic rings in S. pombe (Pollard, 2017), but Rho1 contributes to full activation of the SIN to drive ring formation (Alcaide-Gavilán et al., 2014) and activates glucan synthases to form the septum (Arellano et al., 1996). Thus, Rho signaling is an ancient mechanism to coordinate cytoskeletal elements, but it has been adapted for a range of functions to regulate cytokinesis during the divergence from the last eukaryote common ancestor.

METHODS AND MATERIALS

Growth conditions and strain construction

Supplemental Table S2 lists the S. pombe strains used in this work. Oligonucleotide primers were from Millipore Sigma. Restriction enzymes were from New England BioLabs. Plasmid purification was performed using an E.Z.N.A. Plasmid Mini Kit (Omega Bio-tek) with Miraprep modifications (Pronobis et al., 2016). The Keck DNA Sequencing Lab of Yale University performed routine DNA sequencing. All cloning was done with high-efficiency NEB 5-alpha Competent Escherichia coli from New England BioLabs. Standard methods were used for genetic crosses (Moreno et al., 1991).

mto2^S338N:KanMX6 and mto2^S338C:KanMX6 strains of S. pombe were generated by homologous integration (Bähler et al., 1998). First, pFA6a-mto2^S338N-C-KanMX6 was generated. A 245-base fragment of the end of the mto2 coding sequence was amplified from cps1-191 (containing the mto2^S338N mutation) with flanking homology to the pFA6a vector using the Expand High Fidelity PCR system (Millipore Sigma). The vector backbone was amplified from pFA6a-GFP-KanMX6 using Phusion polymerase (New England BioLabs). Both PCR reactions were digested with DpnI and purified using an E.Z.N.A. Cycle Pure Kit (Omega Bio-tek). pFA6a-mto2^S338N-C-KanMX6 was generated using NEBuilder HiFi DNA Assembly Master Mix (New England BioLabs) in an assembly containing 5 fmol of each fragment. The mto2^S338N segment was verified by sequencing. pFA6a-mto2^S338C-C-KanMX6 was generated by site-directed mutagenesis. Mutagenic primers were used to amplify pFA6a-mto2^S338N-C-KanMX6 using the Expand High Fidelity PCR system (Millipore Sigma). The PCR product was digested with DpnI and transformed into NEB 5-alpha Competent E. coli. The mto2^S338Csegment was verified by sequencing.

PCR-based homologous integration cassettes were amplified using the Expand High Fidelity PCR system (Millipore Sigma) from the appropriate plasmid with 80 base pairs flanking homology to replace the 3′ end of mto2 (primers designed with PPPP at www.bahlerlab.info/resources/; Bähler et al., 1998). Lithium acetate transformation was used to introduce 5–10 μg of integration cassette DNA (Murray et al., 2016). Integration at the correct locus was verified by colony PCR using LongAmp Taq 2X Master Mix (New England BioLabs), and the mto2 gene was sequenced in its entirety to verify the S338N or S338C point mutations.

The bgs1^D277N strain was generated by CRISPR (Fernandez and Berro, 2016). A two-step PCR was used to assemble pUR19-adh1-Cas9-rrk1-sgRNA:Fex1⁺ with a guide sequence targeting bgs1 using Q5 High-Fidelity polymerase (New England BioLabs). pUR19-adh1-Cas9-rrk1-sgRNA:Fex1⁺ and a 400–base pairs repair guide fragment containing the bgs1^D277N mutation were introduced into fex1Δ fex2Δ S. pombe cells by lithium acetate transformation (Murray et al., 2016). The bgs1 gene was sequenced in its entirety to verify the bgs1^D277N mutation.

For serial dilution assays, 10-ml cultures of each strain were maintained in exponential phase for 36 h in YE5S (rich) medium in 50-ml baffled flasks. All cultures were diluted to an OD₅₉₅ of 0.5. Five 10-fold serial dilutions were made for each strain, yielding a total of six cellular densities. Five microliters of each concentration was spotted on YE5S + 1.8% agar or YE5S 1.2 M sorbitol + 1.8% agar plates. Each plate was grown either at 25°C or 36°C and imaged each day for 4 d to monitor growth.

Whole-genome sequencing

S. pombe genomic DNA was isolated using a Quick-DNA Fungal/Bacterial Miniprep kit and purified by a Genomic DNA Clean & Concentrator-25 kit (both from Zymo Research). Genomic DNA was sequenced on a NovaSeq 6000 (Illumina) with 150–base pairs paired reads at the Yale Center for Genome Analysis. Sequence data quality was examined using FastQC v 0.11.7 (Babraham Bioinformatics). The S. pombe reference genome was downloaded from PomBase (Lock et al., 2019) and reads were aligned using MagicBLAST (Zhang et al., 2015). Aligned reads were sorted, duplicates removed, and indexed using SAMtools (Li et al., 2009). Deviations from the reference genome were identified using Pilon (Walker et al., 2014), and alignments were viewed using IGV 2.4.10 (Robinson et al., 2017). Identified mutations were cross-referenced with PomBase to determine location, eliminate previously-reported variants, and identify protein-coding mutations (Lock et al., 2019).

Microscopy

Cells were grown in 50-ml baffled flasks in exponential phase at 25°C in YE5S (rich) medium for 24 h, then washed in EMM5S (low-fluorescence) medium for an additional 12 h. For experiments at 36°C, the culture was shifted to 36°C 2 h before mounting and imaging. To image, cells were isolated by centrifugation at 2300 × g for 30 s, washed with fresh EMM5S, and resuspended at ∼20-fold higher concentration than in culture. Then 1.5–2.0 μl of concentrated cells were mounted on an EMM5S + 2% agarose pad freshly poured on a Therminator block (Davies et al., 2014) and a 24 × 50 mm no. 1.5 coverslip was applied (Globe Scientific). Then 150 μl of EMM5S was injected beneath the coverslip in contact with the agarose pad and the edges of the chamber were sealed using VALAP to reduce shrinkage of the pad. For salt stress experiments, cells at 25°C were washed and resuspended in EMM5S + 0.6 M KCl, then immediately mounted on an EMM5S + 0.6 M KCl + 2% agarose pad on a Therminator block and imaged at 25°C. The average time between KCl addition and initiation of imaging was 5 min.

Cells were visualized using an Olympus IX-71 microscope with a 100×/ NA 1.4 Plan Apo lens (Olympus) and a CSU-X1 (Andor Technology) confocal spinning-disk confocal system equipped with an iXON-EMCCD camera (Andor Technology). OBIS 488-nm LS 20 mW and OBIS 561-nm LS 20 mW lasers (Coherent) were used for excitation. The desired temperature was maintained during acquisition using a home-built version of the Therminator (Davies et al., 2014). Images were acquired using Micromanager 1.4 (Edelstein et al., 2014).

Acquisition settings for Pcp1-mEGFP Rlc1-tdTomato: 12 Z-slices were acquired with 0.5 μm spacing; 22 μW, 561 nm laser power and 50-ms exposures; 35 μW, 488 nm laser power and 50-ms exposures (power measured at the objective). An emission filter suitable for both tdTomato and mEGFP fluorescence was used to reduce the number of filter changes. All bgs1⁺cells were imaged at 4 XY positions and 1-min intervals. Cells carrying bgs1^D277N were imaged at 20 XY positions and 5-min intervals.

Acquisition settings for Rlc1-tdTomato Sfi1-mCherry mEGFP-Atb2: For cytokinesis timelines to determine the timing of spindle breakage, 12 Z-slices were acquired with 0.5 μm spacing at three XY positions and 1-min intervals using 150-ms exposures with 32 µW, 561 nm laser power, and 200-ms exposures with 89 μW, 488 nm laser power at the objective. Separate emission filters were used for tdTomato/mCherry and mEGFP fluorescence to avoid bleedthrough of the fluorophores with different spectral characteristics. To measure astral microtubule numbers and longevities, only mEGFP-Atb2 was imaged using 89 µW, 488 nm laser power and 200-ms exposures of 10 Z-slices with 0.5 μm spacing at one XY position at 10-s intervals.

Acquisition settings for GFP-Bgs1 Rlc1-tdTomato Sfi1-mCherry: 12 Z-slices with 0.5 μm spacing were acquired at three XY positions and 1-min intervals using 150-ms exposures with 32 μW, 561 nm laser power, and 100-ms exposures with 89 μW, 488 nm laser power at the objective. Separate emission filters were used for tdTomato/mCherry and mEGFP fluorescence to avoid bleedthrough of the fluorophores with different spectral characteristics. Strains bearing GFP-tagged proteins for the calibration curve were imaged under identical conditions. For camera noise correction, 100 images were acquired with 100-ms exposures and the shutter closed; for uneven illumination correction, dilute fluorescein was mounted on a 2% agarose EMM5S pad on the Therminator and 100 images were acquired using 50-ms exposures with 34 μW, 488 nm laser power.

Image analysis

All image visualization and analyses were done using Fiji (Schindelin et al., 2012). Macros used for processing can be downloaded from https://github.com/SDundon/ImageJ-Processing-Macros. Before analysis, the contrast of each image was adjusted so that the black point was 5% below extracellular background signal and the white point was 20% above maximum signal intensity. Representative images were selected based on similarity to median values of each data set.

Constriction rates and cytokinesis timeline measurements: Maximum-intensity projections were produced for each time-lapse acquisition and a kymograph was first produced for each ring. If necessary, the StackReg plugin (Thévenaz et al., 1998) was used to register images across all time points. A line was drawn parallel to the ring, the “reslice” tool in Fiji used, and the resliced image was maximum-intensity projected to create a kymograph. The ring circumference for each time point was calculated using the 180912_AutoCircum.ijm macro. Circumference was plotted against time as a scatter plot in Microsoft Excel, and a linear trendline fit to the slope to calculate the constriction rate. For analysis of cells bearing the bgs1^D277N mutation (including cps1-191), cells were excluded from analysis if SPB separation occurred more than 2 h after the start of acquisition. We found that rings formed after this time always constricted at an even slower rate, most likely due to effects of phototoxicity.

Cytokinesis milestones were defined as follows: “SPB separation” is the first frame in which two SPBs labeled with Pcp1-mEGFP could be resolved; “Complete assembly” is the first frame in which no Rlc1-tdTomato–labeled nodes were visible adjacent to the ring, because all had coalesced at the equator; “Constriction onset” is the first frame in which the Rlc1-tdTomato–labeled ring was smaller than its initial diameter and continued to decrease in size; “Constriction complete” is the frame in which the Rlc1-tdTomato signal reached its smallest size at the end of constriction. All milestones were temporally aligned with SPB separation set as time 0 and plotted as cumulative distributions. Spindle breakage was defined as the first time point in which there was no longer a single microtubule structure marked with GFP-Atb2 connecting the two SPBs. PAA appearance was defined as the first frame in which microtubules appeared at the equator distinct from the mitotic spindle.

Astral microtubule measurements: Astral microtubules were considered to be the same structure across consecutive frames (10-s intervals) if they 1) originated from the same SPB and 2) were oriented in the same direction ±45° relative to the SPB in consecutive frames (Figure 3A). The number of astral microtubules that originated from a single SPB during anaphase B was determined to exclude the intranuclear “astral” microtubules that form during earlier phases of mitosis (Zimmerman et al., 2004).

Rho biosensor and GFP-Bgs1 measurements: A region of interest (ROI) was selected for bleach correction reference using the following criteria: 1) must stay within the boundaries of one cell for the duration of the acquisition; 2) must be located within a cell that does not divide for the duration of the movie (to prevent differences in cytosolic concentration due to recruitment to the division site); and 3) Pcp1-mEGFP–marked SPB must never enter the ROI. The same ROI was used for bleach correction of all channels. Bleach correction was conducted using the Fiji plugin for Exponential Fitting Method for the Bleach Correction. The intensity from all 12 Z-slices (0.5 μm apart) was then summed.

Rho biosensor measurements: Before analysis, the contrast of each image was adjusted so that the black point was 5% below extracellular background signal and the white point was 20% above maximum signal intensity of Pcp1-mEGFP, which was more consistent than Pkc1(HR1-C2)-mEGFP signal. As previously observed, expression from the 3nmt1 promoter was highly variable. Therefore, cells were analyzed that had an average Pkc1(HR1-C2)-mEGFP pixel intensity of more than 175,000 A.U. (arbitrary units) at SPB separation. This was calculated by drawing a spline-fit polygon ROI around the cell at the last time point before SPB separation. The integrated density was measured for this ROI, as well as a circular ROI covering the SPB (Supplemental Figure S1O). The SPB integrated density was subtracted from the cellular integrated density, and this value divided by the cell area with SPB area subtracted. This value (175,000 A.U.) was selected because cells of all genotypes below this threshold were found to have very low peak Rho/interphase Rho ratio values with a very different SD than cells above this threshold. We reasoned that cells below this threshold express insufficient Pkc1(HR1-C2)-mEGFP to enable reliable measurement of Rho1-GTP enrichment at the cell equator. After elimination of cells below this threshold, the data were homoscedastic when the ratio was plotted against average pixel intensity.

To determine cellular polarity, a cell end was counted as “polarized” if it exhibited visible Pkc1(HR1-C2)-mEGFP enrichment and/or was observed to grow over the course of the time-lapse experiment. Cells that exhibited no growth from either end were scored as “nonpolar,” growth from one end was scored as “monopolar,” and growth from both ends was “bipolar.” Cells undergoing cytokinesis were excluded from this analysis, as cell ends cease growth during this time.

To calculate the peak Rho/interphase Rho ratio, a 3.75 × 1.95 μm ROI was positioned with the cytokinetic ring in the center (long axis to match the cell width; Supplemental Figure S1O). The integrated density was measured for this ROI for each time point. These values were divided by the integrated density of this ROI for the time point immediately before SPB separation to normalize the intensity by Pkc1(HR1-C2)-mEGFP expression. Normalized intensity was plotted against time using Microsoft Excel; all time points in which the SPB was located within the ROI were excluded (Figure 3E). To measure peak duration, the peak fluorescence start was defined as the last time point of consistent signal increase across more than three time points. Peak end was defined as the last time point before consistent signal decrease below the peak start value across more than three time points. For cytokinesis peak signal, the maximum ratio between these points was determined. For separation peak signal, the maximum normalized signal within two time points of cell separation was determined.

Measurements of Bgs1 and Cdc7 molecule numbers: Bleach-corrected summed intensity images were corrected for camera noise and uneven illumination. To generate the calibration curve, a spline-fit polygon ROI was drawn around wild-type (unlabeled) cells and the average intensity in the ROI was measured (n = 74 cells). Subsequently, ROIs were similarly drawn around cells bearing seven different GFP-tagged proteins for the calibration curve (n > 65 cells). For each cell, the wild-type background fluorescence was subtracted and the integrated density was calculated. These integrated density measurements were plotted against the average number of molecules per cell for each protein reported in Wu and Pollard, 2005 and fit with a linear regression (Wu and Pollard, 2005).

For Bgs1: A 1.0 μm × 3.75 μm (cell width) ROI was positioned to cover the equatorial GFP-Bgs1 signal and the integrated density measured from cell cycle time zero until 30 min past the completion of constriction. The same ROI was used to measure cytoplasmic GFP-Bgs1 signal adjacent to the ring at constriction onset. Cytoplasmic GFP-Bgs1–integrated density was subtracted from the equatorial signal until the onset of constriction. The average constriction rate and average ring size were used to calculate a correction factor for the progressively smaller cytoplasmic volume within the plane of the constricting ring, which was applied to the value of the cytoplasmic signal and subtracted from the equatorial signal until constriction completion.

For Cdc7: A 1.0 μm ×1.0 μm ROI was positioned over each SPB, and the integrated density measured from the first time point where signal was visible above background (typically at SPB separation as determined by Sfi1-mCherry signal). The same ROI was used to measure cytoplasmic Cdc7-mEGFP at SPB separation, and this was subtracted from the signal measurements to correct for the large volume of cytoplasm not containing the SPB.

Statistical analysis

All statistical tests were carried out in R (R Core Team, 2018). Plotting was done using tidyverse and ggbeeswarm packages (Clarke and Sherrill-Mix, 2017; Wickham, 2017), except for pie charts in Figure 2, which were generated using Microsoft Excel.

Abbreviations used:

GDI

GDP-dissociation inhibitor

MTOC

microtubule organizing center

PAA

postanaphase array

SIN

septation initiation network

SPB

spindle pole body.