Fission yeast TRP channel Pkd2p localizes to the cleavage furrow and regulates cell separation during cytokinesis.

Force plays a central role in separating daughter cells during cytokinesis, the last stage of cell division. However, the mechanism of force sensing during cytokinesis remains unknown. Here we discovered that Pkd2p, a putative force-sensing transient receptor potential channel, localizes to the cleavage furrow during cytokinesis of the fission yeast, Schizosaccharomyces pombe. Pkd2p, whose human homologues are associated with autosomal polycystic kidney disease, is an essential protein whose localization depends on the contractile ring and the secretory pathway. We identified and characterized a novel pkd2 mutant pkd2-81KD. The pkd2 mutant cells show signs of osmotic stress, including temporary shrinking, paused turnover of the cytoskeletal structures, and hyperactivated mitogen-activated protein kinase signaling. During cytokinesis, although the contractile ring constricts more rapidly in the pkd2 mutant than the wild-type cells (50% higher), the cell separation in the mutant is slower and often incomplete. These cytokinesis defects are also consistent with misregulated turgor pressure. Finally, the pkd2 mutant exhibits strong genetic interactions with two mutants of the septation initiation network pathway, a signaling cascade essential for cytokinesis. We propose that Pkd2p modulates osmotic homeostasis and is potentially a novel regulator of cytokinesis.

INTRODUCTION

Force plays a critical role in separating daughter cells during cytokinesis, the last stage of cell division (for review, see Pollard, 2010; Srivastava ). Since the initial discovery (Rappaport, 1967), studies by many groups have established that the actomyosin contractile ring is essential to generate the force required for cytokinesis (Sanger and Sanger, 1980; De Lozanne and Spudich, 1987; Straight ). What remains largely unknown is whether and how dividing cells respond to this force, a potential mechanical stimulus. Some have proposed that the activity of myosin II, a motor protein in the ring, may be force sensitive but the mechanism is far from clear (Effler ; Pinheiro ). On the other hand, few studies have examined whether any other mechanisms could contribute to force sensing during cytokinesis.

Mechanosensitive (MS) channels can open and close in response to mechanical force (for review, see Kung, 2005; Ranade ). The first such channel was identified in bacteria (Martinac ). Since then, many more MS channels have been found in not only prokaryotes but also eukaryotes including humans. Among them, the best studied examples are Escherichia coli MscS and MscL, which modulate intracellular osmolarity in response to hypoosmotic shocks (Sukharev , 1996). Comparably, our understanding of eukaryotic MS channels such as PKD2, Piezo, and NompC is relatively limited, even though they are essential in sensing various forms of mechanical stimuli including flow, touch, and osmolarity (Mochizuki ; Walker ; Coste ). We have not completely understood how MS channels sense force, but two widely accepted models have been proposed (Kung, 2005; Arnadottir and Chalfie, 2010; Ranade ). The “Membrane tension” model hypothesizes that force directly stretches the membrane where MS channels are located to open their ion pores. The “Tethered” model proposes that force is applied to the extracellular matrix and/or the intracellular cytoskeleton to activate the MS channels indirectly. The functions of MS channels in nonsensory cells also remain largely unexplored.

We examined the potential roles of MS channels during cytokinesis using the fission yeast Schizosaccharomyces pombe. Classical genetics studies have identified a large number of important cytokinesis genes in this model organism (Balasubramanian ; Pollard and Wu, 2010; Johnson ). The molecular mechanism of fission yeast cytokinesis has been well studied and found to be largely conserved in higher eukaryotes (Pollard and Wu, 2010). Like animal cells, fission yeast assemble an actomyosin contractile ring at the cell division plane at the beginning of cytokinesis (Wu ). This process requires many essential proteins including Myo2p, Cdc12p, and Cofilin (Chang ; Balasubramanian ; Chen and Pollard, 2011) as well as continuous polymerization of the actin filaments (Courtemanche ). The ring assembly is followed by the contractile ring closure driven by the activities of both type II and type V myosins (Mishra ; Laplante ). The final step of cytokinesis is cell separation (for review, see Sipiczki, 2007; Garcia Cortes ), which includes such key steps as the septum biosynthesis (Liu ; Cortes , 2012; Munoz ) and the cell wall degradation (Martin-Cuadrado ; Dekker ).

Mechanical forces, including that from the contractile ring, the septum, and the turgor, all play important roles during fission yeast cytokinesis. The tension originated from the contractile ring is required for the initiation of cell separation (Balasubramanian ; Mishra ; Stachowiak ). Just as importantly, the ring provides the mechanical cue that guides the septum biosynthesis (Thiyagarajan ; Zhou ). However, a recent work has shown that the ring is not required continuously during cytokinesis (Proctor ). The compression applied by the expanding septum helps both drive the cleavage furrow ingression and anchor the furrow at the cell division plane (Arasada and Pollard, 2014; Thiyagarajan ). This was vividly demonstrated by the observation that a defect in the septum biosynthesis often results in failed cell separation as well as postseparation cell lysis (Cortes ; Munoz ). Finally, the turgor pressure, essential to morphogenesis of yeast cells, plays a key but often underappreciated role in separating daughter cells (Proctor ; Abenza ; Atilgan ). Despite the importance of these forces, we have much to learn about how these mechanical stimuli are sensed and modulated during fission yeast cytokinesis, including the potential role of MS channels.

Here we examined the role of a putative transient receptor potential (TRP) channel Pkd2p during fission yeast cytokinesis. Many eukaryotic MS channels including NompC and human PKD2 belong to the TRP channel family (for review see Christensen and Corey, 2007). Consisting of seven subfamilies, this family of nonselective cation channels has been found in most eukaryotes (Wu ), including yeasts (Palmer , 2005; Ma ). We discovered that a fission yeast TRP channel Pkd2p localizes to the cleavage furrow during cytokinesis. The pkd2-81KD mutation led to strong cytokinesis defects. Our genetic studies also identified the potential interaction between pkd2 and the septation initiation network (SIN) pathway, a Hippo-like signaling cascade that regulates fission yeast cytokinesis (McCollum and Gould, 2001; Hergovich and Hemmings, 2012).

RESULTS

A putative TRP channel Pkd2p localizes to the cleavage furrow

To identify MS channels that may play a role during cytokinesis, we first determined whether any of the fission yeast TRP channels localize to the cell division plane (Palmer ; Aydar and Palmer, 2009; Ma ). The fission yeast genome encodes for the genes of three putative TRP channels, Pkd2p, Trp663p, and Trp1322p (Pombase). We tagged each of the endogenous proteins with GFP at its C-terminus (unless specified, mEGFP was used throughout this study). Only Pkd2p-GFP localized to the cell division plane prominently during cytokinesis (Figure 1A, Supplemental Figure S1A, and Supplemental Movie S1). The pkd2::pkd2-GFP cells exhibited similar morphology and viability to wild-type cells (unpublished data), indicating that Pkd2p-GFP is a functional replacement of the endogenous protein. We concluded that Pkd2p is a putative TRP channel localized at the cleavage furrow and it may have a role in cytokinesis.

FIGURE 1:

Localization of a TRP channel Pkd2p at the cell division plane. (A–C) Pkd2p localization during cytokinesis. (A, B) Fluorescence micrograph of cells expressing Pkd2p-GFP (green) and Rlc1p-tdTomoato (red), a marker for the contractile ring. Unless specified, maximum intensity projections are shown in all figures. (A) Fluorescence micrographs showing Pkd2p and Rlc1p colocalized to the contractile rings. Right, merged image. (B) Time-lapse micrographs of the division plane of a cell (head-on view and merged color). Numbers represent time in minutes. (C) Plot showing the time course of Pkd2p-GFP fluorescence at the cell division plane after the separation of SPBs (time zero). Pkd2p-GFP appeared at the division plane at the start of ring constriction (shaded area) and the fluorescence intensities peaked when the ring closure was completed. (D) Pkd2p localization during interphase. Left, fluorescence micrographs of six cells expressing Pkd2p-GFP (numbered from 1 to 6 based on their length). Average intensity projections of three center Z-slices are shown. Right, line scans based on the micrographs of cell 1 (top) and 6 (bottom). Pkd2p localized equally to the two cell tips (blue arrowheads). Dashed lines: median plane of the cells. (E, F) Regulation of Pkd2p localization. (E) Fluorescence micrographs of cells expressing Pkd2p-GFP (green) and Rlc1p-tdTomato (red), treated for 1 h with either control (dimethyl sulfoxide [DMSO], left) or 10 µM latrunculin A (LatA, right). Disassembly of the contractile ring displaced Pkd2p-GFP to the cortex clumps (red brackets). (F) Fluorescence micrographs of wild-type (cdc12) and cdc12-112 cells expressing Pkd2p-GFP at 36°C. Pkd2p was displaced from the division plane at the restrictive temperature. Bars represent 5 µm. Error bars represent SD.

We determined Pkd2p localization throughout cell cycle using live fluorescence microscopy. During cell division, Pkd2p-GFP first appeared at the cell division plane during telophase, ∼30 min after separation of the spindle pole bodies (SPBs; Figure 1, B and C). Its molecular number gradually increased, eventually peaking at ∼1200, when the ring closure was completed (Supplemental Figure S1C). Pkd2p-GFP localized to the intracellular vesicles and organelles. It was also found at the plasma membrane of cell tips (Figure 1D and Supplemental Movie S1) but it did not show a preference for either one of the tips (Figure 1D). We concluded that Pkd2p localization at the cell division plane is dependent upon cell-cycle progression.

We next examined how Pkd2p is recruited to the cleavage furrow during cytokinesis. First, we determined whether the actin or microtubule cytoskeletal structures are directly required for Pkd2p localization during cytokinesis. The actin cytoskeletal structures were disassembled within 5 min in the cells treated with 50 µM latrunculin A (Coue ; Chen and Pollard, 2013). However, Pkd2p-GFP remained at the cell division plane in these cells (Supplemental Figure S1D). Similarly, after the microtubules had been depolymerized in the cells treated by methyl benzimidazol-2-carbamate (MBC; Sawin and Snaith, 2004), Pkd2p localization at the division plane was unchanged (Supplemental Figure S1E). These observations demonstrated that neither actin filaments nor microtubules are required to maintain Pkd2p localization at the cleavage furrow. Next, we determined whether the contractile ring was required for Pkd2p localization. We dissembled the contractile ring by treating the cells with 10 µM latrunculin A for an extended period of time. In these cells, Pkd2p-GFP dispersed from the cleavage furrow to discrete puncta on the plasma membrane (Figure 1E). Consistent with this finding, Pkd2p also failed to localize to the cleavage furrow in cdc12-112, a mutant of the formin required for the ring assembly (Chang ), at the restrictive temperature (Figure 1F). Finally, we determined whether the secretory pathway is required for Pkd2p localization. Brefeldin A (BFA) was used to inhibit the secretory pathway (Misumi ). In these BFA-treated cells, Pkd2p-GFP localization to the cell division plane was inhibited (Supplemental Figure S1F). Together, our data suggested that Pkd2p localization at the cleavage furrow during cytokinesis requires both the contractile ring and the secretory pathway.

Pkd2 is an essential gene required for both cell growth and cell division

Pkd2p is the only fission yeast homologue of polycystins (Palmer ), an evolutionally conserved subfamily (TRPP) of TRP channels (Wu ). There are two human homologues of Pkd2p: PKD1 and the MS channel PKD2 (Hanaoka ; Nauli ). Mutations of either one of the human genes lead to a hereditary renal disorder, autosomal polycystic kidney disease (ADPKD). As a putative TRP channel, fission yeast Pkd2p possesses an N-terminal signal peptide, an extracellular lipid binding (ML-like) domain, a central transmembrane domain (TRP domain), and a C-terminal cytoplasmic coiled-coil domain (Figure 2, A and B). Pkd2p had been proposed to be a part of the cell wall integrity pathway (Palmer ; Aydar and Palmer, 2009) but its role in cytokinesis had not been characterized.

FIGURE 2:

An essential gene pkd2 is required for cell growth and cell division. (A, B) Putative TRP channel Pkd2p. (A) Domain structure of Pkd2p. *: the signal peptide. ML-like: MD-2–related lipid recognition domain. TRP-like: the transmembrane domain. CC: coiled-coil domain. (B) Predicted topology of Pkd2p on the plasma membrane (orange). Red: the signal peptide. Pkd2p is projected to contain extensive extracellular and cytoplasmic domains. (C, D) pkd2 is an essential gene. (C) Tetrad dissections of sporulated pkd2/pkd2Δ cells, germinated on either YE5s (top) or YE5s plus 1.2 M sorbitol plates (bottom). Red circles: expected pkd2Δ spores. The pkd2 deletion mutant is not viable. (D) Representative micrographs of pkd2Δ spores germinated on YE5s for more than 1 wk. Most mutant spores died during the outgrowth (top) although a small number died after one to two rounds of cell division (bottom). Compared to wild-type spores that initiated polarized outgrowth (J), pkd2Δ spores grew isotropically. (E–J) Characterization of a hypomorphic pkd2 mutant. (E) Bar graph comparing the average number of Pkd2p-GFP molecules per cell among pkd2::P81nmt1-pkd2-GFP (P81nmt1), pkd2-GFP (Ppkd2), and pkd2::P3nmt1-pkd2-GFP (P3nmt1) in YE5s (n > 60). Replacing the endogenous pkd2 promoter with P81nmt1 reduced the number of Pkd2p molecules by 70%, while P3nmt1 promoter increased the molecular number onefold. (F) Micrographs of pkd2-81KD cells in either YE5s (suppressing) or EMM5s (inducing) media. Under suppressing condition, many mutant cells were multiseptated (arrowheads) and some appeared to be lysed (arrows). Both morphological defects were rescued under inducing condition. (G) Transmission electron micrographs of the cell wall of a wild-type (top) and a pkd2-81KD (bottom) cell. (H) Box plot showing the average thickness of wild-type and pkd2-81KD cell walls (n > 8 cells). There was no significant difference between them (P > 0.05). (I) Fluorescence micrographs of bodipy-phallacidin stained wild-type (left) and pkd2-81KD (right) cells. Distribution of the actin patches (arrow) was less polarized in the mutant cells, even though their actin cytoskeletal structures appeared to be similar to those of the wild-type cells. (J) Left, representative time-lapse micrographs of a wild-type (top) and a pkd2-81KD (bottom) spore after germination (time zero). Numbers represent time in hours. Right, histogram showing the elapsed times before the initiation of polarized outgrowth (arrowhead). The outgrowth of pkd2-81KD mutant (red) was delayed, compared with the wild-type spores (blue; n > 30). All error bars represent SD.

To determine the function of Pkd2p, we first attempted to construct a deletion mutant. However, pkd2Δ was not viable (Figure 2C), which agreed with the previous finding (Palmer ). In fact, microscopic examination of pkd2Δ spores found that the majority of them (24/26) failed to divide and died before initiating polarized outgrowth (Figure 2D), although few of the mutant spores (2/26) did divide one to two rounds before death (Figure 2D). We attempted to rescue the deletion mutant with an osmotic stabilizer because of its proposed role in the cell wall integrity pathway. However, pkd2Δ did not survive in the presence of 1.2 M sorbitol (Figure 2C). We concluded that pkd2 is an essential gene, required for cell growth and cell division.

Next, we attempted to construct a viable pkd2 mutant. We replaced the endogenous pkd2 promoter with a series of inducible promoters (Maundrell, 1990; Basi ). Among the three resulting pkd2 mutants, the one with the strongest growth defect is pkd2::P81nmt1-pkd2 (referred to as pkd2-81KD; Supplemental Figure S2A), in which a weak inducible 81nmt1 promoter replaced the endogenous promoter. The pkd2-81KD mutant was also hypersensitive to high concentration of salts including CaCl2 (Supplemental Figure S2B). To quantify the depletion or overexpression of Pkd2p in the mutants, Pkd2p was tagged with GFP in both pkd2-81KD and pkd2::P3nmt1-pkd2 to measure average numbers of Pkd2p molecules per cell (Supplemental Figure S2C and Supplemental Movie S2). The number of Pkd2p molecules decreased by ∼70% in the pkd2-81KD cells but doubled in pkd2::P3nmt1-pkd2, when the nmt1 promoters were suppressed (Figure 2E). The depletion of Pkd2p led to strong morphological defects of pkd2-81KD cells, which could be reversed by removing thiamine to restore the expression of pkd2 (Figure 2F). In contrast, neither trp663Δ nor trp1322Δ cells exhibited an apparent morphological defect (Supplemental Figure S1B). We concluded that pkd2-81KD is a novel hypomorphic pkd2 mutant that can be used in further studies of this gene.

We examined the pkd2-81KD mutant for its other defects. The mutant cells were significantly wider than the wild-type cells (Supplemental Figure S2D). The cell wall of this pkd2 mutant, examined with transmission electron microscopy, was similar in appearance and thickness to that of the wild type (Figure 2, G and H). Compared to the wild-type cells, the actin cytoskeletal structures of the mutant appeared to be similar, but distribution of the interphase actin patches was less polarized in the mutant cells, suggesting a potential cell polarity defect (Figure 2I). The requirement of Pkd2p was not limited to vegetative cells. The pkd2 mutant spores initiated their outgrowth more slowly than the wild-type ones did (Figure 2J), reminiscent of the defect observed in pkd2Δ spores noted above. This germination defect of the pkd2-81KD spores was also apparent in their usual slow growth following tetrad dissections (unpublished data). We concluded that Pkd2p likely plays an essential role in cell morphogenesis.

Temporary deflation of the pkd2 mutant cells

When examining pkd2-81KD cells with bright-field microscopy, we noticed that some of them appeared to be lysed. Because cell lysis is often associated with a defect in the cell wall biosynthesis, we characterized this phenotype in further details (Cortes ; Munoz ; Davidson ; Sethi ). To our surprise, these mutant cells did not lyse completely but they merely shrank temporarily, which we termed as “Deflation,” before quickly recovered. In all, 16% of the mutant cells underwent deflation and reinflation (Figure 3A and Supplemental Movie S3). In contrast, no wild-type cell shrank under the same condition. The temporary deflation of the pkd2 mutant cells lasted ∼20 min (20 ± 8 min, average ± SD, n = 12) during which the cell size was reduced by ∼20% (Figure 3D). Interestingly, the majority of deflated cells were not undergoing either cell division or septation (80%, n = 15), suggesting that neither triggered the deflation. To our knowledge, this unique phenotype has not been described before for any other mutant. The quick loss of cell volume and the following recovery to some extent resembled the response of wild-type cells to hypertonic stress.

FIGURE 3:

Deflation of the pkd2 mutant cells. (A) Time-lapse micrographs of pkd2-81KD cells. Both cells 1 and 2 shrank temporarily (arrowheads). Numbers represent time in minutes. (B) Time-lapse micrographs of a pkd2-81KD cell expressing Rlc1p-tdTomato. The ring transformed into a “wavy band” (bottom) during the period of temporary deflation (top, red underline). (C) Time-lapse micrographs of a pkd2-81KD cell expressing Sty1p-GFP. Temporary shrinking (top, red underlined) of the cell was concomitant with increased Sty1p-GFP localization in the nucleus (bottom). Number represents time in minutes. (D) Plot showing the time courses of average sizes (red line, normalized) and nuclear Sty1p-GFP fluorescence (blue line) of pkd2-81KD cells (n = 7). The shrinking reduced the average cell size by ∼20%. Concurrently, Sty1p-GFP fluorescence in the nucleus increased by about threefold. Error bars represent SD.

We determined whether deflation of the pkd2 mutant cells was a result of osmotic stress by first examining turnovers of their cytoskeletal structures which would stop temporarily in response to osmotic stresses (Robertson and Hagan, 2008). We first examined the contractile ring and the actin patches, two actin cytoskeletal structures. Upon deflation of the mutant cells, their contractile ring stopped constricting abruptly and converted temporarily from a circular ring into a “wavy band” (Figure 3B and Supplemental Movie S4). The deflation triggered a stop in turnover of the endocytic actin patches as well (Supplemental Figure S3A). Not surprisingly, turnover of the microtubules stopped temporarily in the deflated cells too (Supplemental Figure S3B and Supplemental Movies S5 and S6). We concluded that turnover of the cytoskeletal structures was stopped temporarily in the deflated pkd2 mutant cells as in osmotic stressed cells.

We next determined whether deflation of the pkd2 mutant cells also activated the Sty1p-mediated mitogen-activated protein kinase (MAPK) pathway, another hallmark of the cells under osmotic stresses. Activation of this MAPK pathway shuttles Sty1p into the nucleus where the kinase mediates the stress-activated gene expressions (Millar ; Shiozaki and Russell, 1995; Degols ). We found significantly more Sty1p-GFP molecules in the nuclei of pkd2-81KD cells than the wild-type cells (Supplemental Figure S3, C and D). When the pkd2 mutant cells shrank, nuclear localization of Sty1p-GFP increased concomitantly by up to three folds, compared with predeflation (Figure 3, C and D, and Supplemental Movie S7). We concluded that the Sty1p-mediated MAPK pathway was hyperactivated during deflation of the pkd2 mutant cells as was found in osmotically stressed cells. In addition, we also found that the deflation of pkd2 mutant cells could be rescued by the osmotic stabilizer sorbitol (n > 500 cells; Supplemental Figure S3E). In summary, we concluded that Pkd2p most likely regulates osmotic homeostasis during cell growth.

Pkd2p regulates the ring closure during cytokinesis

To examine the role of Pkd2p during cytokinesis, we first measured the contractile ring assembly in pkd2-81KD cells. Overall, the ring assembled normally in the mutant cells (Supplemental Movies S8 and S9). The pkd2 mutation did not disrupt the placement of rings at the cell division plane (Figure 4A). The rings were positioned properly in the pkd2 mutant cells as well, unlike the oblique angled rings found in those mutants with a defect in the septum biosynthesis (Munoz ). We did notice that the mutant cells were significantly shorter than the wild-type cells during cytokinesis (Figure 4B). The pkd2 mutant cells also experienced a slight but significant delay in ring assembly and maturation, compared with the wild-type cells (Figure 4C; 29 ± 4 vs. 25 ± 4 min, average ± SD, P < 0.001). We concluded that Pkd2p is not essential for either the assembly or maturation of the contractile ring.

FIGURE 4:

Increased ring closure rate in pkd2-81KD cells. (A) Micrographs of wild-type (pkd2, left) and pkd2-81KD (right) cells expressing Rlc1p-tdTomato. Placement of the contractile rings in the mutant was similar to that in the wild-type cells. (B) Bar graph showing the average length of wild-type and pkd2-81KD cells during cytokinesis. The analysis included only cells with one complete contractile ring. The pkd2 mutant cells were significantly shorter than the wild-type cells (n > 50, P < 0.001). (C) Ring assembly in pkd2-81KD cells. Left, time-lapse fluorescence micrographs of a wild-type (top, pkd2) and a pkd2-81KD (bottom) cell. Both cells expressed Rlc1p-tdTomato (red) and Sad1p-GFP (green), a marker for SPBs (arrowheads). Ring assembly was slower in the mutant (∼6 min) than the wild-type cells. Asterisks indicate the appearance of a complete ring. Numbers represent time in minutes. Right, box plot showing the duration of ring assembly plus maturation in wild-type (gray) and pkd2-81KD cells (red). Ring assembly and maturation was slower in the mutant compared with the wild-type cells (P < 0.0001). (D–F) Ring closure in pkd2-81KD cells. Time-lapse fluorescence micrographs of the cell division plane in a wild-type (top) and a pkd2-81KD (bottom) cell. Frame interval is 2 min. The rings constricted more rapidly in the mutant than in the wild-type cells. (E) Fluorescence kymographs of ring closure in a wild-type (top) and a pkd2 mutant (bottom) cell. Lines: Start of ring closure. (F) Box plot showing the average ring closure rates in wild-type (gray) and pkd2-81KD (red) cells. The rate increased by ∼50% in the mutant. Bars represent 5 µm. Error bars represent SD.

Next, we determined whether Pkd2p regulates ring closure, a more likely role considering its localization during cytokinesis. Surprisingly, ring closure on average took less time in the mutant than in the wild-type cells (Figure 4D), in spite of our earlier observation that the mutant cells were wider than the wild-type ones (Supplemental Figure S2D). As expected, the rings in the mutant cells contracted more than 50% faster than they did in the wild-type cells (Figure 4, E and F, and Supplemental Figure S4A; 0.47 ± 0.07 vs. 0.30 ± 0.04 µm/min, P < 0.0001). Our measurement of ring closure in the wild-type cells, closely matched the previously published values (Chen and Pollard, 2011; Proctor ; Laplante ). We did not observe any sliding of the rings during their contraction in the mutant (n > 50), which would have occurred if the ring had not been anchored properly (Cortes ; Arasada and Pollard, 2014). Therefore, it was unlikely that this rapid ring closure in the mutant was a result of failure to couple the ring to the plasma membrane or the cell wall. Unexpectedly, overexpression of Pkd2p did not slow the ring closure as one would have predicted (Supplemental Figure S4, B and C), indicating that Pkd2p is necessary, but not sufficient in modulating the rate of ring closure. We concluded that Pkd2p modulates ring closure during cytokinesis but is dispensable for ring assembly, maturation, or anchoring.

Pkd2p is required for separation of daughter cells

We next examined the role of Pkd2p in cell separation, the final phase of fission yeast cytokinesis. Two lines of evidence indicated that Pkd2p may play an important role in cell separation. First, the fraction of pkd2-81KD cells with septa was about three times higher than the wild-type cells (Figure 5, A and B). In addition, close to 9% of the mutant cells were multiseptated (Figure 5B). We also noticed that the septum in these multiseptated mutant cells frequently swung from one side to another at a rate of 0.05 ± 0.03 µm/min (n = 12). Second, the septa in many mutant cells were curved, deposited ectopically at the cell tips or unusually thick (Figure 5A). We postulated that Pkd2p is likely essential for cell separation.

FIGURE 5:

Separation defect of pkd2-81KD cells during cytokinesis. (A) Fluorescence micrographs of calcofluor-stained wild-type (pkd2, left) and pkd2-81KD (right) cells. Septum of the mutant cells were often curved (arrow), thicker than normal (asterisk), or deposited ectopically (arrowhead). (B) Bar graph showing the percentage of wild-type and pkd2-81KD cells with at least one septum (n > 500). The percentage of multiseptated cells is shown in red. The fraction of septated pkd2-81KD cells was almost three times higher than that of the wild-type cells. (C, D) Separation defect of the pkd2 mutant cells. (C) Time-lapse micrographs of a wild-type (top) and two pkd2-81KD (middle and bottom) cells. All three cells expressed Rlc1p-tdTomato. Left, fluorescence micrographs of the cell when the ring closure was completed (time zero). Numbers represent time in minutes. Frame intervals are 4 (top), 6 (middle), and 10 (bottom) min, respectively. The wild-type cell took 32 min to separate (red arrowheads). In contrast, the two pkd2-81KD cells took 60 and 120 min, respectively. (D) Dot plot showing the durations of cell separation in wild-type and pkd2-81KD cells (n > 15). The horizontal lines represent averages. The mutant cells separated more slowly than the wild-type cells.

We examined separation of the pkd2 mutant cells during cytokinesis with live microscopy. The wild-type cells reliably separated in ∼30 min (28 ± 5 min, average ± SD, n = 20) after completion of the ring closure (Figure 5C). In contrast, the mutant cells took more than twice as long (Figure 5D; 75 ± 21 min, n = 17) with some of them taking as long as 120 min to separate (Figure 5C). Furthermore, some mutant cells failed to separate completely, explaining the many multiseptated pkd2 mutant cells. Wild-type cells typically separate symmetrically by detaching both sides of the new ends from each other simultaneously (Figure 5C). In contrast, most mutant cells separated asymmetrically by usually detaching one side of the new ends first (Figure 5C, bottom row). We concluded that Pkd2p plays an essential role in separating the new ends during cell separation.

We next determined whether Pkd2p was required for either the septum biosynthesis or the cell wall degradation, two critical steps of cell separation. We first examined the localization of two glucan synthases Bgs1p and Bgs4p, both of which are required for the septum biosynthesis (Le Goff ; Liu ; Cortes ). Both GFP-Bgs1p and GFP-Bgs4p still localized to the cell division plane in the pkd2-81KD cells, as they did in the wild-type cells (Figure 6, A and B). Furthermore, the time course of recruitment and dispersal of either Bgs1p or Bgs4p at the division plane in the mutant remained largely unchanged as well (Figure 6, C and D). The septum did close more rapidly in the mutant compared with the wild-type cells, consistent with the quicker ring closure in the mutant (Supplemental Figure S4, D and E). As a result, the number of Bgs1p molecules at the cell division plane of the mutant decreased slightly compared with the wild-type cells (Figure 6C). In contrast, the number of GFP-Bgs4p molecules at the division plane rose slightly (Figure 6D). We concluded that Pkd2p is dispensable for the localization of these two glucan synthases during cell separation. Next, we determined whether Pkd2p is required for the localization of Eng1p, a glucanase that degrades the septum at the end of cell separation (Martin-Cuadrado ; Sipiczki, 2007). The pkd2 mutation did not alter the localization of Eng1p-GFP at the cell division plane (Supplemental Figure S4, F and G). The number of Englp-GFP molecules at the division plane remained unchanged, compared with that in the wild-type cells. We concluded that Pkd2p is not required for the localization of Eng1p either and the most likely role of Pkd2p during cell separation is to regulate the turgor pressure in the separation of new ends.

FIGURE 6:

Localization of the glucan synthases Bgs1p and Bgs4p in pkd2-18KD cells. (A) Fluorescence micrographs of wild-type (pkd2, top) and pkd2-81KD (bottom) cells expressing GFP-Bgs1p. The pkd2 mutation did not alter GFP-Bgs1p localization at the cell division plane. (B) Time-lapse micrographs of a wild-type (top) and a pkd2-81KD (bottom) cell expressing GFP-Bgs1p. The wild-type cells separated at 36 min (white arrowheads) after the ring closure (time zero), but the mutant cells failed to separate. Nevertheless, GFP-Bgs1p localization at the cell division plane appeared unchanged in the mutant. Frame interval is 4 min. Numbers represent time in minutes. (C, D) Line plots showing the time courses of GFP-Bgs1p (C) and GFP-Bgs4p (D) molecules localized at the division plane of either wild-type (blue line) or pkd2-81KD cells (red line). The number of GFP-Bgs1p molecules in the pkd2 mutant was slightly reduced but the number of GFP-Bgs4p molecules was increased, compared with the wild-type cells (n > 5). Error bars represent SD.

Genetic interactions between pkd-81KD and the other cytokinesis mutants

Pkd2p localization at the cleavage furrow and its role in cell separation prompted us to examine whether it interacts with the other cytokinesis genes. We carried out a targeted screen to identify genetic interactions between pkd2-81KD and 11 other cytokinesis mutants (Figure 7A). Four of them interacted negatively with the pkd2 mutant (Figure 7A), including the temperature-sensitive mutants of myo2 (type II myosin), cdc12 (formin), rng2 (IQGAP1 homologue), and cdc15 (a F-BAR protein), respectively. All four genes are required for the ring assembly (Fankhauser ; Chang ; Kitayama ; Eng ). On the other hand, cps1-191, the temperature-sensitive mutant of bgs1, exhibited no genetic interaction with pkd2-81KD mutant. We concluded that Pkd2p plays a synergistic role with Myo2p, Cdc12p, Rng2p, and Cdc15p during cytokinesis.

FIGURE 7:

Genetic interactions between pkd2-81KD and the other cytokinesis mutants. (A) Table summarizing the genetic interactions (GI) between pkd2-81KD and 11 other cytokinesis mutants. Positive interactions are designated by either “++” (strong), whereas negative interactions are designated by either “--” (strong) or “-” (weak). “NO” indicates no genetic interactions. (B) Tenfold dilution series of yeast cells at the indicated temperatures. pkd2-81KD mutant had positive genetic interactions with mob1-R4 and sid2-250 but negative interactions with myo2-E1 and cdc12-112. It had no genetic interactions with cps1-191. (C, D) Micrographs of sid2-250, sid2-250 pkd2-81KD, mob1-R4, and mob1-R4 pkd2-81KD cells grown at 36°C. Top, bright-field micrographs of the live cells. Bottom, fluorescence micrographs of the calcofluor-stained cells. Arrows: lysed cells. Arrowheads: the septum. The SIN mutants frequently lysed and few of them contained septum at the restrictive temperature, defects that were partially rescued by the pkd2 mutation. (E, F) Bar graphs showing the percentage of either lysed (E) or septated (F) cells at 36°C. The pkd2 mutation largely rescued both the lysis and septation defects of the SIN mutants. Error bars represent SD.

Two SIN mutants, mob1-R4 and sid2-250, had the strongest genetic interactions with pkd2-81KD. They are the temperature-sensitive mutants of Sid2p kinase and its activator Mob1p, respectively (Balasubramanian ; Hou ). Growth of mob1-R4 was rescued by the pkd2 mutation at the restrictive temperature (Figure 7B). In comparison, the pkd2 mutation only rescued the growth of sid2-250 at semirestrictive temperature. The SIN temperature-sensitive mutants failed in septation and eventually lysed, at the restrictive temperature (Figure 7, C and D; Balasubramanian ; Salimova ; Krapp ). The pkd2-81KD mutation prevented the lysis of both SIN mutant cells (Figure 7E and Supplemental Figure S5C). Additionally, the relatively mild septation defect of mob1-R4 cells was also rescued by pkd2-81KD (Figure 7F and Supplemental Figure S5D). In comparison, the septation defect of sid2-250 mutant was only partially rescued by the pkd2 mutation at both semirestrictive and restrictive temperature (Figure 7F and Supplemental Figure S5D). Additionally, the septa of sid2-250 pkd2-81KD cells were much thinner than those of wild-type cells (Figure 7C). We concluded that Pkd2p and the SIN pathway play antagonistic roles in septation and maintaining cell integrity.

DISCUSSION

Forces, generated by the contractile ring, the septum, and the turgor, are essential for fission yeast cytokinesis. However, we know relatively little about how these forces can be sensed and modulated by cells. Here we showed that a putative force-gated channel Pkd2p has several important functions during cytokinesis, including modulating the ring closure and promoting cell separation. Defects of the pkd2-81KD mutant cells suggest that the most likely role of this TRP channel is to regulate osmotic homeostasis. Not surprisingly, Pkd2p is well placed at the cleavage furrow during cytokinesis to execute this function. The genetic interactions between pkd2-81KD and the other cytokinesis mutants also support that this transmembrane protein may be a novel component of the cytokinetic machinery.

Localization of Pkd2p channel at the cleavage furrow

To our knowledge, Pkd2p is the first putative ion channel that localizes to the cell division plane of fission yeast cells. Our discovery is compatible with the previous finding that Pkd2p was mostly found at the intracellular organelles and the plasma membrane (Palmer ; Aydar and Palmer, 2009). Pkd2p is one of the few TRP channels that have been found at the cleavage furrow so far. TRP3 channel of the zebrafish Danio rerio, which mediates the influx of calcium at the furrow, is the other notable example (Chan ). Many molecules in the contractile ring may interact with Pkd2p directly to recruit it to the cleavage furrow. This mechanism will be consistent with our observation that the contractile ring is required for Pkd2p localization. The septum may also interact with Pkd2p directly, considering that the initial appearance of Pkd2p at the furrow coincides with the start of primary septum synthesis. Future studies will be needed to determine the molecular mechanism of Pkd2p recruitment to the cell division plane.

Based on our study of fission yeast Pkd2p, it will be interesting to determine the localization of other Pkd2p homologues during cytokinesis. Budding yeast homologues of Pkd2p localize to both the ER and plasma membrane (Protchenko ) but their localization during cytokinesis has not been examined. Among the two human Pkd2p homologues, PKD1 is primarily localized at the plasma membrane (Hughes ) and PKD2 is mostly found on the ER membrane (Gonzalez-Perrett ), although PKD2 does localize to the mitotic spindles through its interaction with formin mDia1 in cultured kidney epithelial cells (Rundle ). Additionally, most Pkd2p homologues, including those of humans, Caenorhabditis elegans, and Drosophila melanogaster, also localize to the cilia of nondividing cells (Barr and Sternberg, 1999; Barr ; Pazour ; Yoder ; Gao ; Nauli ; Watnick ). Future work would be necessary to determine the cytokinetic localization of Pkd2p in dividing animal cells.

Molecular functions of Pkd2p during cytokinesis

Our study suggests that Pkd2p most likely regulates osmotic homeostasis during cytokinesis as well as cell morphogenesis. This model is supported by at least three lines of evidence. First, the unique deflation defect of the pkd2 mutant could be due to a defect in the regulation of turgor pressure. The temporary shrinking of pkd2-81KD cells activated the stress response pathway mediated by Sty1p. The shrinking also stopped turnovers of all cytoskeletal structures that we examined. All of these phenotypes are consistent with misregulated turgor pressure. In addition, the septa of multiseptated pkd2-81KD cells often swung from side to side, suggesting that turgor pressure fluctuates drastically in these mutant cells. Second, the morphology of pkd2-81KD cells also imply a defect in the osmotic regulation during cell morphogenesis. Compared to wild-type cells, the pkd2 mutant cells are shorter but wider and the distribution of their actin patches is partially depolarized. Furthermore, both pkd2Δ and pkd2-81KD spores either failed or delayed their outgrowth during germination. These observations are consistent with a defect in the regulation of turgor pressure, which is essential to fission yeast cell growth (Atilgan ; Davi ). Third, the cytokinesis defects of pkd2-81KD cells are also consistent with misregulated turgor pressure. Both rapid ring and septum closure in the mutant cells can be a result of low turgor pressure, which could reduce resistance to the furrow ingression and increase the ring closure rate, compared with the wild-type cells. This hypothesis is consistent with the important role of turgor in determining ring closure rate (Proctor ). On the other hand, slow or failed separation of the pkd2 mutant cells may be attributed to misregulated turgor at the cell division plane (Sipiczki, 2007; Abenza ; Atilgan ). Although future studies will be necessary to directly measure the putative abnormities of turgor pressure in the pkd2 mutation cells, our observations strongly support that Pkd2p plays an essential role in osmotic homeostasis during cytokinesis and cell morphogenesis.

As a TRP channel, Pkd2p could potentially regulate the turgor pressure through a number of ways. It may control the ion influx to directly modulate the turgor. As a permissive cation channel, Pkd2p could allow the calcium influx to regulate the turgor pressure, considering its localization on the plasma membrane (Palmer ). In this way, Pkd2p may be similar to another TRP channel TRPV4 that regulates osmolarity of animal cells (Colbert ; Strotmann ; Liedtke and Friedman, 2003). Alternatively, Pkd2p may regulate the intracellular osmolarity indirectly through interacting with the other signaling pathways including the SIN pathway. It is worth noticing that both Pkd2p and the Sid2p/Mob1p complex localize to the cell division plane during late cytokinesis (Sparks ; Goss ). More studies are needed to determine how Pkd2p acts through these potential mechanisms to regulate the turgor pressure.

Molecular functions of the Pkd2p family proteins

Our study identified a novel function of the fungal Pkd2p family proteins in regulating cytokinesis. This family of transmembrane proteins can be found among many fung speciies and is one of 17 essential fungal protein families (Hsiang and Baillie, 2005). Four Saccharomyces cerevisiae Pkd2 homologues, Flc1 (flavin carrier 1), Flc2, Flc3, and YOR365c, have been identified. Unlike fission yeast pkd2, none of these budding yeast homologues is essential (Protchenko ; Rigamonti ; Vazquez ). However, deletion of both flc1 and flc2 is lethal. These four proteins appear to play a role in osmotic regulation (Protchenko ), heme transportation (Protchenko ) and calcium homeostasis (Rigamonti ), although the molecular mechanism of their functions is unknown. It also remains to be determined whether any of these four budding yeast genes is required for cytokinesis. Other fungal Pkd2p homologues include spray of Neurospora crassa (Bok ), FlcA/B/C of Aspergillus fumigatus (de Castro ), and CaFlc1-3 of Candida albicans (Protchenko ). Interestingly, all of them regulate hyphal growth of the fungi, similar to the fission yeast Pkd2p regulation of cell growth. Our proposed role of Pkd2p in regulating osmotic homeostasis would be consistent with the central role of turgor in the fungal growth (Lew, 2011).

In multicellular eukaryotic organisms, Pkd2p is required for several developmental processes but the underlying molecular mechanism is far from clear. Both D. melanogaster and C. elegans homologues of Pkd2p contribute to the locomotion of sperms and the pkd2 mutations lead to male sterility (Barr and Sternberg, 1999; Gao ; Watnick ). The human Pkd2p homologues PKD1 and PKD2 are required for cardiac development and left-right orientation (Wu ; Pennekamp ). Their function in the kidney has been studied most extensively, because loss of function mutations of either genes result in a hereditary human disease, ADPKD (for review, see Chapin and Caplan, 2010). This renal disorder affects more than 600,000 Americans alone. It is characterized by progressive proliferation of liquid-filled cysts in the kidney that often culminates to a terminal renal failure (Mochizuki ). Despite years of work, it still remains largely unknown how the mutations of these human Pkd2p homologues lead to ADPKD. Study of fission yeast Pkd2p and its function in osmotic homeostasis will likely provide us with critical insights into the molecular functions of human PKD1 and PKD2 as well as the pathogenesis of ADPKD.

MATERIALS AND METHODS

Yeast genetics

Yeast cell culture and genetics procedures were carried out according to the standard methods. A SporePlay+ dissection microscope (Singer, England) was used for tetrad dissections. Most yeast strains were constructed through PCR-based homologous recombination (Bahler ). To construct trp663::trp663-GFP, trp1322::trp1322-GFP, and pkd2::pkd2-GFP, we integrated the mEGFP sequence at the endogenous locus of each gene. We constructed trp1322/trp1322Δ by removing one copy of the open reading frame (ORF). The diploid cells were then sporulated and selected to isolate trp1322Δ. The trp663Δ mutant is from a fission yeast deletion library (Bioneer, South Korea) and was confirmed by PCR. To examine the viability of the pkd2Δ mutant, we constructed the diploid pkd2 by deleting one copy of the ORF. The diploid was then sporulated and dissected into more than 20 tetrads that were deposited onto either YE5s or YE5s plus 1.2 M sorbitol agar plates. For microscopy of pkd2Δ spores, a small block of agar containing the spore was removed from the agar plate and deposited on a glass-bottomed Petri dish for observation. To replace the endogenous pkd2 promoter, we integrated inducible nmt1 promoters at the endogenous locus, preceding the start codon of the pkd2 ORF. To construct pkd2::P81nmt1-pkd2-GFP-ura, we constructed pkd2::pkd2-GFP-ura first, before replacing the endogenous promoter with P81nmt1. The resulting yeast strains were confirmed by both PCR and the Sanger sequencing method. The pkd2-81KD spores were obtained by digesting the tetrads with glusulase II (PerkinElmer) and observed under a YE5s agar pad for 8–12 h with bright-field microscopy.

Microscopy

For microscopy, exponentially growing yeast cells at 25°C with a density between 5.0 × 106/ml and 1.0 × 107/ml in YE5s liquid media (unless specified), were harvested by centrifugation at 4000 rpm for 1 min and resuspended in YE5s. The cells were then applied to a 25% gelatin + YE5s pad and sealed under the coverslip with VALEP (a mix of an equal amount of Vaseline, lanolin, and paraffin). Live microscopy was carried out on an Olympus IX71 microscope equipped with both a 100× (NA = 1.41) and a 60× (NA = 1.40) objective lens, a confocal spinning-disk unit (CSU-X1; Yokogawa, Japan), a motorized XY stage, and a Piezo Z Top plate (ASI). The images were captured on an Ixon-897 EMCCD camera controlled by iQ3.0 (Andor, Ireland). Solid-state lasers of 488 and 561 nm were used in the confocal fluorescence microscopy, at a power of no more than 5 mW (<10%). Unless specified, we imaged cells by acquiring a Z-series of 15 slices at a step size of 0.5 µm. Live microscopy was conducted in a room where the temperature was maintained at around 20°C. To minimize the temperature variations, we usually imaged both wild-type and the pkd2 mutant cells on the same date. For temperature-sensitive mutants, we inoculated the cultures at the restrictive temperatures for 4 h before either imaging them at room temperature immediately or fixing the cells before imaging. For visualizing the cell wall, we stained cells with 10 µg/ml calcofluor (Sigma) and used an Olympus IX81 microscope equipped with a CCD camera and a mercury lamp for the epifluorescence microscopy. For visualizing the actin cytoskeletal structure, we fixed cells with 16% EM-grade paraformaldehyde (EMS) and stained them with bodipy-phallacidin (Invitrogen), as described previously (Chen ).

Electron microscopy

Transmission electron microscopy was carried out at the Microscopy and Image Analysis Laboratory core at the University of Michigan (Ann Arbor, MI). Briefly, exponentially growing cells were harvested and rinsed with 0.1 M phosphate buffer (pH = 7.4) before being fixed with 2.5% glutaraldehyde (EMS). The fixed cells were stained with 1% osmium tetroxide to increase contrast and embedded in epoxy resin overnight at 4°C. The samples were then sectioned at a thickness of 70 nm (Leica EM UC7). The EM specimens were imaged with a JEM-1400 (Joel, Japan) transmission electron microscope. The thickness of the cell wall was measured as the average distance between the plasma membrane and the outer boundary of the cell wall, at more than 10 randomly selected sites.

Image processing

We used ImageJ (National Institutes of Health) to process all the images, with either freely available or customized macros/plug-ins. For quantitative analysis, the fluorescence micrographs were corrected for X-Y drifting using the StackReg plug-in (Thevenaz ) and for photobleaching using the EMBLTools plug-in (Rietdorf, EMBL Heidelberg). Average intensity projections of Z-slices were used for quantifications. The contractile ring localization of Pkd2p-GFP was quantified by measuring the GFP fluorescence in a 3.6 µm × 0.8 µm (36 × 8 pixels) rectangle centering on the cell division plane, corrected with background subtraction through measuring the background fluorescence intensities in two 3.6 µm × 0.2 µm rectangles (36 × 2 pixels) adjoined to the division plane. The cell width was measured using both live and fixed cells, which yielded similar results. The length of cytokinetic cells was measured using bright-field images of the live cells expressing Rlc1p-tdTomato.

The nuclear localization of Sty1p-GFP was measured by quantifying GFP fluorescence intensities in the nuclei with background subtraction. To simplify the analysis, we assumed the nucleus as a circle of 2 µm diameter that centers on the nucleus localization of Sty1p-GFP. The background fluorescence was determined by measuring the fluorescence intensities of Sty1p-GFP in the cytoplasm surrounding the nucleus. The total area of a cell was measured based upon the cytoplasmic fluorescence of Styp1-GFP after the cell was segmented semiautomatically using the Binary conversion and Analysis Particles tools of ImageJ.

The rate of ring closure is measured by analyzing fluorescence kymographs of a ring during cytokinesis. The kymograph was constructed from the time-lapse fluorescence micrographs (Figure 4E), which was then used to determine both the start and end of ring closure. The diameter of a ring was determined at the start of ring closure. For quantitative fluorescence microscopy, our spinning-disk confocal microscope was calibrated using a previously published method (Wu and Pollard, 2005). Briefly, we imaged wild-type and seven fission yeast strains expressing GFP-tagged proteins to produce a calibration curve (R2 > 0.95). Slope of the calibration curve was used to determine the ratio of fluorescence intensities to molecule numbers. The figures were made with Canvas X (ACDsee systems, Canada). The domain structure of Pkd2p is based on the prediction by Pfam and the National Center for Biotechnology Information (NCBI). The predicted topology of Pkd2p was created by Protter (http://wlab.ethz.ch/protter/), a protein topology data aggregation server (Omasits ).

Click here for additional data file.