Analysis of the contribution of phosphoinositides to medial septation in fission yeast highlights the importance of PI(4,5)P2 for medial contractile ring anchoring.

In Schizosaccharomyces pombe, loss of the plasma membrane PI4-kinase scaffold Efr3 leads to sliding of the cytokinetic ring (CR) away from the cell center during anaphase, implicating phosphoinositides (PIPs) in CR anchoring. However, whether other PIP regulators contribute to CR anchoring has not been investigated. Here we report that mutants of other PIP kinases and their regulators divide with off-center septa, similar to efr3∆. Using new biosensors for S. pombe PIPs, we confirm that these mutants have disrupted PIP composition. We extend a previous finding that a mutant known to decrease PI(3,5)P2 levels indirectly affects CR positioning by increasing vacuole size which disrupts nuclear position at the onset of mitosis. Indeed, we found that other mutants with increased vacuole size also disrupt medial division via this mechanism. Although elevated plasma membrane PI(4,5)P2 levels do not affect medial cytokinesis, mutants with decreased levels display CR sliding events indicating a specific role for PI(4,5)P2 in CR anchoring.

INTRODUCTION

Dramatic rearrangements of the plasma membrane (PM) are required during cytokinesis, the final step of cell division (Eggert ; Barr and Gruneberg, 2007; Fededa and Gerlich, 2012). Though phosphoinositides (PIPs) comprise ∼5–10% of PM lipid species in mammalian cells (Wenk ) and are important for cytokinesis (Echard, 2012), the mechanisms by which PIPs promote accurate cytokinesis are not fully defined.

Of the seven PIP species found in the cell, only a subset have been implicated in cell division, including phosphatidylinositol-4-phosphate (PI4P) and phosphatidylinositol-4,5-bisphosphate (PI(4,5)P2). A role for PI4P in cytokinesis was revealed through studies focused on PI4-kinases, the enzymes that generate PI4P from phosphatidylinositol (PI). In Drosophila melanogaster spermatocytes, absence of the type IIIβ PI4-kinase (PI4KIIIβ) encoded by four-wheel drive results in cytokinesis failure and multinucleate spermatids (Brill ). Similarly, the catalytic activity of Schizosaccharomyces pombe PI4KIIIβ Pik1 is essential for normal septation and abscission (Park ). The role of PI4KIIIβ in human cells is less clear, although one study showed that elevated levels of PI4KIIIβ activity and thus higher PI4P levels inhibit cytokinesis, resulting in multinucleate cells (Rajamanoharan ). Type IIIα PI4-kinases (PI4KIIIα) have also been implicated in cytokinesis. In both Saccharomyces cerevisiae and S. pombe, a PI4KIIIα Stt4 is localized to the PM by scaffolds, Efr3 and Ypp1 (Baird ; Snider ). S. pombe cells lacking efr3 do not properly localize Stt4 to the PM, have improper PIP composition, and display cytokinetic ring (CR) sliding from the cell center during anaphase (Snider ). D. melanogaster cells lacking PI4KIIIα also display cytokinesis defects that result in binucleate cells (Eggert ). A role of PI4KIIIα in human cytokinesis has not been reported though the kinase and scaffolding machinery are conserved from yeast to humans (Baird ; Chung ).

Another PIP species implicated in modulating cell division is PI(4,5)P2, the most abundant PIP species in the PM. PI(4,5)P2 is enriched at the division site of mammalian (Emoto ; Field ; Kouranti ; Dambournet ; Abe ), S. pombe (Zhang ; Snider ), and D. melanogaster S2 (El Kadhi ; Roubinet ) cells. In addition, PI5-kinases that generate PI(4,5)P2 from PI4P localize to the division site in human (Emoto ), D. melanogaster S2 (Roubinet ), and S. pombe (Zhang ) cells. PI(4,5)P2 can also be generated by PI3-phosphatases acting on PI(3,4,5)P3 and PI3-phosphatases localize to the division site in Dictyostelium discoideum and S. pombe (Mitra ; Janetopoulos ). In HeLa and Chinese hamster ovary cells, depletion of PM PI(4,5)P2 results in separation of the PM from the actin cytoskeleton and cytokinesis failure (Field ). It is hypothesized that proteins that mediate actin-PM adhesion may require PI(4,5)P2 for this function; candidates relevant to the process of cytokinesis include anillin (Liu ; Sun ) and other regulators of F-actin dynamics such as N-WASP and profilin (Machesky ; Higgs and Pollard, 2000).

Although these studies have demonstrated that certain PIPs promote faithful cytokinesis, a comprehensive understanding of the specific PIP species and regulators involved has yet to be obtained in any organism. Here, we took two approaches to define the PIP species contributing to medial cytokinesis in S. pombe. First, we determined which PIP enzymes contribute to proper division by examining septum placement and CR dynamics in a comprehensive set of mutants. Second, we developed and validated a lipid biosensor tool set for S. pombe. Results from these two complementary approaches support the importance of PI(4,5)P2 and its precursor PI4P for proper CR anchoring and thus indicate that a specialized role of PI(4,5)P2 in cytokinesis has been conserved throughout evolution.

RESULTS AND DISCUSSION

Deletion of the S. pombe PI4-kinase scaffold efr3 results in CR anchoring defects, where the CR forms in the cell center as in wild type but then slides away during anaphase in a myosin-dependent manner, resulting in off-center septa (Snider ). To determine whether other regulators of PIP composition contribute to medial cytokinesis, we examined septa placement in strains with mutations in genes that encode PIP kinases, phosphatases, and PIP enzyme binding partners (Figure 1A). As a measure of off-center septation the ratio of short to long daughter cell length at septation was calculated (Figure 1B).

FIGURE 1:

Analysis of septum placement in gene deletions of PIP regulators. (A) Schematic of the S. pombe phosphoinositide pathway. PIPs are green. Proteins that when mutant resulted in off-center septa are in magenta, essential proteins not tested are in gray, and the rest are in black. Essential proteins are underlined; binding partners are indicated in parentheses. (B) Top, Representative images of the indicated strains stained with DAPI and Methyl Blue. Arrows indicate off-center septa. (B) Bottom, Quantification of septum placement in indicated strains. All cells were grown at 32°C except its3-1, which was grown at 25°C and then shifted to 32°C for 2 h. Black bars represent the mean. *, p < 0.05; **, p < 0.01; ****, p < 0.0001; one-way ANOVA. Scale bar = 5 µm.

We first examined septum placement in strains with deletions of genes encoding proteins predicted to modulate the localization or activity of PIP kinases and phosphatases (Figure 1B). Among these are: Opy1, a PM-localized dual pleckstrin homology (PH) domain–containing protein that is mislocalized in S. pombe efr3∆ cells and inhibits a PI5-kinase in S. cerevisiae (Ling ; Snider ); Sfk1, a predicted transmembrane scaffold of PI4KIIIα in S. cerevisiae and human cells (Audhya and Emr, 2002; Chung ); Vac14, which binds both the PI(3,5)P2-5-phosphatase Fig4 and the PI3-5-kinase Fab1 in S. cerevisiae and human cells (Botelho ; Sbrissa ); and Ncs1, a neuronal calcium sensor–­related protein that binds and regulates PI4-kinase Pik1 in S. cerevisiae (Strahl ). Ypp1, another Stt4 PM scaffold, is essential for viability and was not tested (Baird ; Snider ). Septum placement was normal in all tested strains except opy1∆, which had a mild off-center septa phenotype (Figure 1B).

Next, we investigated the roles of PIP phosphatases in medial septation. Single deletions of genes encoding the predicted PI4-phosphatase Sac12; the PI(3,5)P2-5-phosphatase Fig4; the PI3-phosphatase Ymr1; the PI5-phosphatases Syj1, Syj2, and Inp53; and the PI(3,4,5)P3-3-phosphatase Ptn1 did not result in off-center septa (Figure 1B); nor did combined deletions of PI(4,5)P2-5 phosphatases (Supplemental Figure S1D). We were unable to assay septum placement in syj1∆ inp53∆ because this combination is synthetically lethal as previously reported (Supplemental Figure S1E; Kabeche ), or in the absence of the PI4-phosphatase Sac11 because a conditional allele of essential sac11 is not available (Kim ).

Finally, we assessed septum placement in mutants of PIP kinases (Figure 1B). There are three predicted PI4-kinases in S. pombe: essential Stt4 and Pik1 (Park ; Snider ) and nonessential Lsb6. Deletion of lsb6 did not result in off-center septa (Figure 1B). There are no available temperature-sensitive mutants of Stt4 or Pik1 but the hypomorphic allele GFP-stt4 has off-center septa at elevated temperature (Snider ), as does a temperature-sensitive mutant of the essential PI5-kinase its3 (its3-1) analyzed at semipermissive temperature (Figure 1B). As previously reported, deletion of the gene encoding the PI3-5-kinase Fab1 resulted in misplaced septa (Morishita and Shimoda, 2000); deletion of the gene encoding the PI3-kinase Pik3 has a similar phenotype (Figure 1B).

To assess septum placement when both PI4-kinase and PI5-kinase activities are compromised, we attempted to generate efr3∆ its3-1 but found that this combination was synthetically lethal (Supplemental Figure S1A). However, we were able to construct GFP-stt4 its3-1 and found that its off-center septa phenotype is more penetrant than in either GFP-stt4 or its3-1 alone at semipermissive temperature (Supplemental Figure S1, B and C). In contrast, combining efr3∆ with GFP-stt4 did not worsen the off-center septa phenotype of efr3∆, consistent with Stt4 and Efr3 acting in the same complex (Supplemental Figure S1B). Further, although we could not isolate a triple mutant of its3-1 inp53∆ syj1∆ in contrast to a previous report (Kabeche ), deletion of syj1 but not inp53 rescued the off-centered septa defect and temperature sensitivity of its3-1 suggesting that one phosphatase may be a more effective antagonist of Its3 than the other (Supplemental Figure S1, F and G).

We next determined how septa form off-center in PIP kinase mutants. Although fab1∆ cells have misplaced CRs and septa due to off-center nuclei (Morishita and Shimoda, 2000), we wanted to determine whether Pik3 and Fab1 also have a role in CR anchoring. Using live-cell time-lapse microscopy, we visualized CR dynamics in both pik3∆ and fab1∆ strains expressing CR (Rlc1-mNeonGreen [mNG]) and spindle pole body (Sid4-mNG) markers to monitor CR and mitotic events, respectively. We found that although CRs form off-center in both strains, they do not slide away from their initial position (Figure 2). Because PI(3,5)P2 is generated by both Pik3 and Fab1, this indicates that PI(3,5)P2 levels influence medial positioning of the CR, but not CR anchoring. fab1∆ and pik3∆ both have abnormally large vacuoles (Takegawa ; Morishita and Shimoda, 2000) and it was shown in the case of fab1∆ that CRs form off-center due to physical displacement of the nucleus. We simultaneously observed the nucleus using Cut11-GFP (West ) and vacuoles using Cki2-mCherry (Matsuyama ) with live-cell time-lapse imaging. As expected, large vacuoles in pik3∆ prevented medial positioning of the nucleus while small vacuoles in wild-type cells did not (Figure 2D). To determine whether it was solely enlarged vacuoles caused by deletion of pik3 or fab1 rather than a reduction of PI(3,5)P2 per se that causes misplaced CRs, we examined whether abnormally large vacuoles arising independently of PIP misregulation cause off-center septum formation. avt3 encodes a lysosomal amino acid transporter and pxa1 encodes a PhoX homology-associated domain protein; deletion of either gene results in abnormally large vacuoles (Hosomi ; Chardwiriyapreecha ). As predicted, both avt3∆ and pxa1∆ had off-center septa (Supplemental Figure S2A). Time-lapse imaging of cells expressing Rlc1-mNG Sid4-mNG showed that avt3∆ formed CRs off-center that did not move from their original position (Supplemental Figure S2, B and C). In summary, large vacuoles displace nuclei from the cell center, resulting in off-center formation of the CR. This phenomenon can occur independently of PIP misregulation suggesting that the contribution of PI(3,5)P2 to medial cytokinesis is likely indirect via changes to vacuole morphology.

FIGURE 2:

CRs form off-center in mutants of PI(3,5)P2 regulators. (A) Montages of time-lapse imaging of the indicated strains at 25°C. Scale bar = 2 µm. Numbers indicate minutes elapsed; magenta line indicates the cell center. (B) Quantification of CR off-center formation frequency and (C) quantification of sliding event frequency in the indicated strains. Over three independent experiments: wild type, n = 20; fab1∆, n = 23; pik3∆, n = 15. Error bars = SEM. (D) Montages of time-lapse imaging of indicated strains. Scale bar = 2 µm; numbers indicate minutes elapsed.

We next analyzed how off-center septa originate in its3-1 and opy1∆. Using time-lapse microscopy, we found that in its3-1, CRs formed in the cell middle and then moved from center over time, similar to the efr3∆ phenotype (Figure 3). CRs also formed normally in opy1∆ and consistent with its relatively mild off-center septa phenotype (Figure 1), only a small portion of cells displayed CR sliding events (Figure 3). As expected, proteins with reduced localization to the PM in efr3∆ (RhoGEF Rgf1-GFP, Cdc42 GEF Scd1-mNG, and dual PH domain–containing protein Opy1-mNG) were also reduced in its3-1 while the localization of the F-BAR protein GFP-Cdc15 was unaffected (Figure 4; Snider ). This indicates that Rgf1, Scd1, and Opy1 membrane binding is PI(4,5)P2-sensitive. Thus, only loss of Efr3 and Its3, proteins that contribute substantially to PI(4,5)P2 generation, affect CR anchoring significantly.

FIGURE 3:

Mutants of PI(4,5)P2 regulators display CR sliding. (A) Montages of time-lapse imaging of opy1∆ grown at 32°C and its3-1 grown at 25°C and shifted to 32°C for 2 h before imaging. Scale bar = 5 µm. Numbers indicate minutes elapsed; magenta line indicates the cell center. (B) Quantification of CR sliding frequency and (C) quantification of CR off-center formation frequency in the indicated strains. For 32°C, wild type n = 11 and opy1∆ n = 15; two experiments. For 25°–32°C shift, wild type n = 14 and its3-1 n = 11; three experiments. Error bars = SEM.

FIGURE 4:

Localization of membrane-binding proteins in its3-1. Live-cell imaging of GFP-Cdc15 (A), Rgf1-GFP (B), Scd1-mNG (C), and Opy1-mNG (D) in either wild type or its3-1 at 25°C. (A–C) Right, Quantification of fluorescence intensity at the cell division site. (D) Right, Linescan of fluorescence intensity. Measurements in A–C represent three biological replicates. Error bars represent SEM, n ≥ 74. **, p < 0.01; ****, p < 0.0001; Student’s t test. Scale bar = 2 µm.

To further probe the relationship between CR anchoring and changes in PM PIP levels, we developed PIP biosensors for PI4P and PI(3,4,5)P3 that are integrated in the genome and expressed from the constitutive moderate-strength cdc2 promoter like the sensor we previously described for PI(4,5)P2 (Snider ). We measured the fluorescence intensity of these sensors bound to the membranes lining both secondary septa of dividing cells to compare PIP abundance at the PM in different mutants.

To detect changes in PI4P levels, we utilized the P4C domain from Legionella pneumophila SidC (Luo ) fused to GFP. In strains with defective PI4-kinase function (efr3∆, lsb6∆, and efr3∆ lsb6∆), we saw a reduction of PI4P sensor membrane localization in each single mutant compared with wild type, and this reduction was exacerbated in the double mutant (Figure 5A). Consistent with this, linescan analysis of nonseptated cells revealed PI4P enrichment at the cell cortex in wild-type cells, but not in efr3∆ (Supplemental Figure S3A, left panel). These results validate GFP-P4CSidC as a suitable PI4P sensor for S. pombe. Using GFP-P4CSidC, we observed that PI4P levels are also reduced at the membranes lining the secondary septa in its3-1 (Figure 5A), but this reduction was not detected at the cell cortex of interphase cells by linescan analysis (Supplemental Figure S3A, middle panel). It is possible that there are specific defects in delivery of membrane to the division site in its3-1 that account for the differences in PI4P at the secondary septa membranes compared with the rest of the PM. PI4P sensor localization was unaffected in opy1∆ (Figure 5A and Supplemental Figure S3A, right), consistent with the lack of appreciable CR anchoring defects in opy1∆.

FIGURE 5:

Mutants with CR anchoring defects have altered PIP composition. Left, representative images of strains expressing integrated (A) GFP-P4CSidC, (B) GFP-PHAkt, and (C) GFP-2xPHPlc at 25°C, with the exception of its3-1, which was shifted to 32°C for 2 h before imaging. Right, quantification of fluorescence intensity at secondary septal membranes of indicated strains; for all, n > 40. (D) Septa placement in wild-type cells overexpressing the indicated lipid sensor or the empty vector (control). Error bars = SEM, A.U. = arbitrary units, scale bar = 5 µm. One-way ANOVA performed for all except (C) wild type vs. its3-1, for which Student’s t test was performed. *, p < 0.05; **, p < 0.01; ****, p < 0.0001 indicate comparison to wild type for A and C, to ptn1∆ for B, and to empty vector (control) for D.

To visualize PI(3,4,5)P3, we used the PH domain of Akt fused to GFP (Gray ). There was little sensor signal at the PM in wild-type cells but as previously reported using a different approach, deletion of ptn1 (PI(3,4,5)3-phosphatase) results in enrichment of PI(3,4,5)P3 at the PM (Mitra ), and the increased PM sensor signal in ptn1∆ validates its specificity. Because the signal in wild-type cells is so low, we used the ptn1∆ background to examine changes in PI(3,4,5)P3 at the PM in other mutants. PI(3,4,5)P3 was reduced in efr3∆ ptn1∆ and its3-1 ptn1∆ and, interestingly, increased in opy1∆ ptn1∆ at the membranes lining the secondary septa, although the reason for this increase is unknown (Figure 5B and Supplemental Figure S3B). We also noted that pik3∆ ptn1∆ cells had reduced PI(3,4,5)P3 at the cell cortex, consistent with Pik3’s role in generating PI(3,4,5)P3 from PI(4,5)P2 (Figure 5B).

We examined PI(4,5)P2 levels using the previously described sensor, GFP-2xPHPlc (Snider ). After shifting its3-1 to a semipermissive temperature, we detected a 50% reduction in PM PI(4,5)P2 compared with wild type (Figure 5C and Supplemental Figure S3C). As expected from the lack of CR anchoring defects, no significant differences in PI(4,5)P2 levels were detected in opy1∆. We examined GFP-2xPHPlc in PI(4,5)P2-5-phosphatase deletions (syj1∆, syj2∆, inp53∆), none of which had off-center septa. The sensor localized as in wild type in syj1∆ and in syj2∆ (unpublished data), suggesting that there is redundancy between Syj1 and Syj2. However, GFP-2xPHPlc accumulated at membranes to a greater extent in inp53∆ than in wild type. Because inp53∆ does not have off-center septa (Figure 1B), we conclude that increased levels of PI(4,5)P2 do not negatively affect medial cytokinesis, whereas a decrease in PI(4,5)P2 leads to CR anchoring defects.

In conclusion, among the many enzymes that influence PM PIP composition, we detected a CR anchoring defect manifested by its sliding from a central position during anaphase only in efr3∆ and its3-1, indicating a specific role for PI(4,5)P2 in CR anchoring. In accord, overexpression of PI(4,5)P2 and PI4P sensors that are expected to sequester PI(4,5)P2 and PI4P, respectively, result in off-center septa due to CR sliding (Figure 5D; Snider ). Interestingly, overexpression of the PI(3,4,5)P3 sensor also causes septum misplacement, although the penetrance is low. Because Pik3 generates PI(3,4,5)P3 (Mitra ) and the CRs in pik3∆ do not slide, it is unlikely that PI(3,4,5)P3 itself plays a significant role in CR anchoring. Perhaps this less abundant species is dephosphorylated to contribute to the PM PI(4,5)P2 pool that promotes CR anchoring during cytokinesis. These data are consistent with the idea that only PI(4,5)P2, the kinases that generate it, and its precursor PI4P are important for cortical CR anchoring and medial septum placement.

Our work confirmed that PI(3,5)P2 influences medial division in S. pombe but most likely indirectly; reduction of PI(3,5)P2 induces the formation of large vacuoles that prevent normal centering of the nucleus, and other mutants with large vacuoles also exhibit this phenotype. The lack of CR sliding events in fab1∆ and pik3∆ indicates that PI3P and PI(3,5)P2 do not have a role in CR anchoring, further supporting the conclusion that PI(4,5)P2 is the only PIP species that contributes significantly to CR anchoring. PI(4,5)P2 has a conserved role in eukaryotic cytokinesis but it remains to be determined whether mediating CR anchoring to the PM is a conserved function unique to this PIP species.

MATERIALS AND METHODS

Yeast methods

S. pombe strains (Supplemental Table S1) were grown in yeast extract with supplements (YES). rlc1 and sid4 were tagged at the 3′ end of their open reading frames with mNG:kanR or mNG:hygR using pFA6 cassettes as previously described (Wach ; Bähler ). A lithium acetate method (Keeney and Boeke, 1994) was used in S. pombe tagging transformations, and integration of tags was verified using whole-cell PCR and/or micro­scopy. Introduction of tagged loci into other genetic backgrounds was accomplished using standard S. pombe mating, sporulation, and tetrad dissection techniques. Deletions of the sfk1 and ncs1 genes were accomplished as previously described (Chen ).

Constitutively expressed lipid sensors were constructed as previously described (Snider ). Briefly, the cdc2 promoter, sequences encoding GFP, and the desired sensor were PCR amplified and cloned into pJK148 using Gibson assembly. GFP-P4C was PCR amplified from a plasmid provided by the Mao lab (Cornell University, Ithaca, NY) (Luo ) and AKT-PH fragment was PCR amplified from plasmid #67301 from Addgene (Kawano ). These constructs were linearized and inserted into the S. pombe leu1 locus by a lithium acetate method (Keeney and Boeke, 1994).

To overexpress P4CSidC, PHAkt, and 2xPHPlc, sequences encoding each were cloned into pREP1 (Maundrell, 1993). The resulting plasmids were introduced into cells by sorbitol transformation. Cells were fixed in 70% ethanol after induction of expression for 24 h at 32°C.

Microscopy methods

Live-cell images of S. pombe cells were acquired using a Personal DeltaVision (Applied Precision) that includes a microscope (IX71; Olympus), 60× NA 1.42 Plan Apochromat and 100× NA 1.40 U Plan S Apochromat objectives, fixed and live-cell filter wheels, a camera (CoolSNAP HQ2; Photometrics), and softWoRx imaging software (Applied Precision). z-sections were spaced at 0.2–0.5 μm. Images were acquired at indicated temperature and cells were imaged in YES media. Time-lapse imaging was performed on cells in log phase on a YES agar pad at 25–32°C. Images were deconvolved with 10 iterations.

Intensity measurements were made with Fiji software (Schindelin ). For all intensity measurements, the background was subtracted by creating a region of interest (ROI) in the same image where there were no cells (Waters, 2009). The raw intensity of the background was divided by the area of the background, which was multiplied by the area of the ROI. This number was subtracted from the raw integrated intensity of that ROI (Waters, 2009). For intensity quantification of the membranes lining the secondary septa, sum projections were analyzed. An ROI was drawn around the septum and measured for raw integrated density. For linescans, the middle z-slice was analyzed. In Fiji, a line was drawn across the short axis and the fluorescence intensity profile was plotted versus distance. All images used for quantification were not deconvolved.

For quantification of off-center septa, all cells were grown to log phase at 32°C before fixation unless otherwise indicated (its3-1 was grown to log phase at 25°C and then shifted for 2 h to 32°C). For overexpression of lipid sensors, cells were grown in the absence of thiamine for 24 h to induce expression from the nmt1 promoter. For nuclei and cell wall imaging, cells were fixed in 70% ethanol for at least 30 min before 4′,6-diamidino-2-phenylindole (DAPI) and Methyl Blue staining. To quantify off-center septa, the coordinates of the cell tips and septum were logged. Lengths of the shorter and longer cells were calculated from these coordinates and reported as a ratio. For quantification of CR sliding, a line was drawn through the fully formed CR marked by Rlc1-mNG using Fiji software. Any movement of the CR away from the original line placement during the entire length of imaging was scored as a ring sliding event.

Statistics

All statistical analyses of variance (ANOVAs) were followed by Tukey’s post hoc test.

Click here for additional data file.