|Home | About | Journals | Submit | Contact Us | Français|
Cytokinesis requires the coordination of many cellular complexes, particularly those involved in the constriction and reconstruction of the plasma membrane in the cleavage furrow. We have investigated the regulation and function of vesicle transport and fusion during cytokinesis in budding yeast. By using time-lapse confocal microscopy, we show that post-Golgi vesicles, as well as the exocyst, a complex required for the tethering and fusion of these vesicles, localize to the bud neck at a precise time just before spindle disassembly and actomyosin ring contraction. Using mutants affecting cyclin degradation and the mitotic exit network, we found that targeted secretion, in contrast to contractile ring activation, requires cyclin degradation but not the mitotic exit network. Analysis of cells in late anaphase bearing exocyst and myosin V mutations show that both vesicle transport and fusion machineries are required for the completion of cytokinesis, but this is not due to a delay in mitotic exit or assembly of the contractile ring. Further investigation of the dynamics of contractile rings in exocyst mutants shows these cells may be able to initiate contraction but often fail to complete the contraction due to premature disassembly during the contraction phase. This phenotype led us to identify Chs2, a transmembrane protein targeted to the bud neck through the exocytic pathway, as necessary for actomyosin ring stability during contraction. Chs2, as the chitin synthase that produces the primary septum, thus couples the assembly of the extracellular matrix with the dynamics of the contractile ring during cytokinesis.
In animal cells, amoebas, and yeast, cytokinesis occurs centripetally through the formation of a cleavage furrow (reviewed in Guertin et al., 2002 ). Ingression of the cleavage furrow leads to close juxtaposition of the equatorial plasma membrane and homotypic fusion of this membrane physically divides the cell into two. The force that drives ingression of the cleavage furrow has been mainly attributed to the actomyosin-based contractile ring, which assembles at the onset of cytokinesis along a plane perpendicular to the elongating mitotic spindle. This mechanism contrasts with that for cytokinesis in plant cells, where formation of the dividing cell plate structure occurs centrifugally through local targeting and fusion of vesicles (Field et al., 1999 ). However, new membrane addition at the cleavage furrow was observed in early studies of cytokinesis in Xenopus eggs (Bluemink and de Laat, 1973 ; Byers and Armstrong, 1986 ). Recent work in echinoderm, Drosophila, and Caenorhabditis elegans embryos further confirmed targeted membrane insertion as a general phenomenon in early embryonic cell divisions (Burgess et al., 1997 ; Jantsch-Plunger and Glotzer, 1999 ; Skop et al., 2001 ; Shuster and Burgess, 2002 ; Xu et al., 2002 ). Pharmacological or biochemical disruption of the secretory pathway resulted in failed cytokinesis, suggesting that membrane addition at the cleavage furrow comes from targeted vesicular transport and fusion and is required for cytokinesis, even in cells that use the contractile ring to draw membrane inward (Skop et al., 2001 ). In conjunction with membrane addition, targeted vesicle fusion in the cleavage furrow may provide timed delivery of factors required for completion of cytokinesis. This was suggested based on results involving disruption of the furrow microtubule arrays that are required for the delivery of vesicles containing specific cargoes to the cleavage furrow (Jesuthasan, 1998 ). However, there is little data on the function of any proteins and lipid factors delivered by membrane trafficking during cytokinesis or whether membrane addition is similarly important for the much smaller somatic cell to complete cytokinesis.
In this study, we exploit the budding yeast Saccharomyces cerevisiae as a model for understanding the function and regulation of targeted exocytosis during cytokinesis. This choice was motivated first by the fruitful studies of vesicular transport carried out using this organism and the availability of a large number of well characterized conditional mutants affecting almost all known aspects of membrane trafficking and recycling (Novick et al., 1980 ; Schekman and Novick, 2004 ). Much of the work on vesicular trafficking in budding yeast has been in the context of polarized growth during bud formation. During this process, Golgi-derived vesicles, which contain the Rab GTPase Sec4, are transported via Myo2, a type V myosin, along oriented actin cables into the nascent bud (Pruyne and Bretscher, 2000 ). Fusion of these vesicles with the plasma membrane requires the conserved exocyst complex (TerBush et al., 1996 ; Guo et al., 1999 ). Late in the cell cycle, it has been observed that many of the proteins involved in polarized secretion are retargeted to the bud neck, consistent with a possible role in cytokinesis (Finger et al., 1998 ).
Cytokinesis in budding yeast, as in animal cells, also involves an actomyosin-based contractile ring, which constricts during cytokinesis (reviewed in Tolliday et al., 2001 ). The assembly of the contractile ring occurs in several steps, starting early in the cell cycle with Myo1 localizing to a ring around the bud neck, in a septin-dependent manner (Bi et al., 1998 ; Lippincott and Li, 1998b ). Actin assembles into the contractile ring during mitotic anaphase in a Cyk1- and formin-dependent manner (Lippincott and Li, 1998b ; Tolliday et al., 2002 ). The initiation of actomyosin ring contraction is governed by the mitotic exit network (MEN), a signaling pathway that controls the exit from mitosis through the release of Cdc14, a protein phosphatase critical for the inactivation of mitotic Cdk1, from the nucleolus where Cdc14 is held inactive (reviewed in Bardin and Amon, 2001 ). Contraction of the ring constricts the membrane and guides the synthesis of a chitin-based primary septum catalyzed by the chitin synthase Chs2, a transmembrane protein. The production and bud neck localization of Chs2 is tightly restricted to late cell cycle stages, in contrast to the other chitin synthases (reviewed in Cabib, 2004 ). It has been suggested that the formation of the primary septum behind the contractile ring may provide additional support needed for closing the gap between mother and daughter cells (Schmidt et al., 2002 ).
We first determined the precise time in the cell cycle where the secretory pathway is targeted to the site of cell division and demonstrated that this targeting event is controlled by mitotic exit, but not directly by the MEN. We found that targeted vesicle fusion is important for the completion of cell division. Analysis of contractile ring dynamics suggested that the secretory pathway is required for the localization of Chs2, a membrane-bound factor necessary for the stability of the actomyosin ring during the contraction process.
Yeast cell culture and genetic techniques were performed as described in Sherman et al. (1974 ). Media were prepared as follows: yeast extract, peptone, dextrose (YPD) contained 2% glucose, 1% yeast extract, and 2% Bactopeptone (Difco. Detroit, MI). YPR contained 2% raffinose, 1% yeast extract, and 2% Bactopeptone. YPGR contained 2% galactose, 2% raffinose, 1% yeast extract, and 2% Bactopeptone. Synthetic complete (SC) medium was prepared as described in Kaiser et al. (1994 ).
All strains used in this study are listed in Table 1. pNT138 was generated by digesting the pGFP-TUB1 insert from pAFS125 (Straight et al., 1997 ) with KpnI and NotI for ligation into pRS304 (Sikorski and Hieter, 1989 ) digested with KpnI and NotI. To generate a strain expressing Myo1-green fluorescent protein (GFP) and GFP-Tub1, pNT138 was digested with SnaBI and transformed into RLY1450 (Tolliday et al., 2003 ) to produce RLY1994. Labeling of the endogenous copy of SEC3 was achieved by a PCR-mediated one-step tagging technique (Longtine et al., 1998 ). Briefly, pFA6a-GFP-HIS3MX6 was used as a template for PCR with primers containing sequences flanking the 3′ end of SEC3. The resultant PCR product was used to transform RLY261 to produce RLY1995, which was then transformed with pNT138 cut with SnaBI to produce RLY1996. To obtain a strain expressing GFP-Tub1 along with Sec4 GFP, RLY261 was transformed first with SnaBI-digested pNT138, and then with pRC556 (pEGFP-SEC4) (Schott et al., 2002 ), to produce RLY1997. The MYO2 (RLY1998) and myo2-16 (RLY1999) strains were obtained from the Bretscher laboratory (Schott et al., 1999 ). These strains were transformed with pRC566 (pEGFP-SEC4) to produce RLY2000 and RLY2001. To obtain Sec3-GFP and Sec4-GFP strains expressing the GAL-inducible clb2 with deletion of the destruction boxes (Surana et al., 1993 ), RLY1995, RLY1996, and RLY1997 were transformed with pGAL-CLB2ΔDB digested with NcoI to produce RLY2066, RLY2002, and RLY2003, respectively. As controls for GAL induction, RLY1995, RLY1996, and RLY1997 were transformed with pRL62 (Lippincott and Li, 1998a ) digested with XcmI to produce RLY2065, RLY2004, and RLY2005, respectively. To analyze Sec3-GFP localization in the mob1-77 (Luca and Winey, 1998 ) background, with over expression of Sic1, RLY656 was mated to RLY1996 and then sporulated to produce MOB1 and mob1-77 strains with Sec3-GFP. The resultant strains RLY2039 and RLY2040 were transformed with a 2μ plasmid containing SIC1 (Luca et al., 2001 ) to produce RLY2042 and RLY2041, respectively. The exocyst mutant strains sec3-2 (RLY1685) and sec10-2 (RLY1689) were backcrossed three times with RLY387 to put the mutations in the W303 strain background. The resultant crosses produced mutant and congenic wild-type strains RLY1838, RLY1840, RLY2006, and RLY2007, respectively. To obtain a wild-type and sec3-2 strains expressing Sec4-GFP, RLY1840, and RLY1838 were transformed with pRC566 (Schott et al., 2002 ) to produce RLY2009 and RLY2008, respectively. The GAL-SEC4 (RLY1900) and dominant negative GAL-SEC4N34 (RLY1901) strains were obtained from Ruth Collins (Cornell University, Ithaca, NY) (Walworth et al., 1989 ). Cdc14-GFP expression in wildtype and sec3-2 cells was obtained by transformation of RLY1840 and RLY1838 with pCDC14-5XGFPIT (Yoshida et al., 2002 ) digested with StuI, to produce RLY2013 and RLY2012, respectively. For expression of Myo1-GFP and GFP-Tub1, RLY1838, RLY1840, RLY2006, and RLY2007 were first transformed with pNT119 (Tolliday et al., 2003 ) digested with BclI and then pNT138 digested with SnaB1 to produce RLY1899, RLY1898, RLY2011, and RLY2010, respectively. MYO2, myo2-16, SEC10, and sec10-2 strains expressing Chs2-GFP were created by transforming RLY1998, RLY1999, RLY2007, and RLY2006 with pLP31 (Chs2-GFP) (Tolliday et al., 2003 ) to create RLY2060, RLY2061, RLY2062, and RLY2063, respectively. RLY2062 and RLY2063 also were expressing Tub1-GFP from transformation with pNT138. The chs2Δ strain (RLY1993) was obtained from E. Cabib (National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health [NIH], Bethesda, MD) and was transformed with pNT119 to produce RLY2064 (chs2Δ, Myo1-GFP).
RLY1994 (Myo1-GFP, GFP-Tub1), RLY1996 (Sec3-GFP, GFP-Tub1), RLY1997 (Sec4-GFP, GFP-Tub1), RLY2062 (Chs2-GFP, GFP-Tub1), RLY1450 (Myo1-GFP), and RLY2064 (chs2Δ, Myo1-GFP) and were grown overnight in the appropriate SC + 2% dextrose or YPD + 0.2 mg/ml adenine sulfate to mid-logarithmic phase and placed in a growth chamber for imaging, as adapted from Maddox et al. (2000 ). Briefly, a 25% gelatin (Sigma-Aldrich, St. Louis, MO) solution was prepared in the appropriate SC + 2% dextrose liquid media and heated to mix the gelatin and media. Then, 200 μl of the heated gelatin mix was pipetted onto a glass slide and covered with a second slide perpendicularly, being supported by two slides, with tape for spacers, parallel to the first. After polymerization, the slides were carefully pried apart, and the layer of gelatin on one slide was trimmed to the size of the coverslip. One milliliter of the overnight cultures was pelleted and resuspended in 50 μl of the appropriate SC + 2% dextrose media. Five microliters of the resuspended cells was pipetted onto the gelatin slab and covered with a coverslip. The coverslip was sealed to the slide by using 1:1:1 Vaseline/lanolin/paraffin (VALAP). Fluorescence images were collected with a PerkinElmer Ultraview spinning disk confocal on a Nikon TE2000 inverted microscope, by using a 100× 1.4 numerical aperture Plan Apo objective lens. The 488-nm wavelength from a krypton-argon laser was selected with a Chroma 488/10-nm bandpass excitation filter. A Chroma single wavelength, 488-nm transmitting dichroic mirror, and HQ550 long-pass emission filter were used. Z-series optical sections were acquired with a Hamamatsu ORCA ER-cooled charge-coupled device camera and with a Prior Proscan focus motor. Image acquisition was initiated every 30 s, and each time point consisted of a stack of 5 z-series images with a distance of 0.5 μm apart, totaling a distance of 2 μm. Exposures of 0.7–0.8 s were used and images were collected by binning 2 × 2 to increase signal over camera noise. MetaMorph imaging software (Universal Imaging, Downingtown, PA) was used for controlling the hardware during imaging, image analysis, and production of two-dimensional (2D) projections of the z-images. The average intensity of Chs2-GFP signal was determined by drawing a line over the Chs2-GFP signal for each time point where the signal was present at the bud neck.
For strains to be induced by galactose, cells were cultured in YPR. Temperature-sensitive (ts) strains were cultured in YPD. All were grown overnight at room temperature. At ~OD260 = 0.200, cells were pelleted and resuspended in either YPGR for GAL induction or YPD for the ts strains with 20 μg/ml nocodazole and grown for ~3 h at room temperature until culture is >90% arrested in metaphase. For GAL induction, cells were pelleted, washed twice in YPGR, and then resuspended in YPGR. For ts strains, after arrest was obtained, cultures were pelleted and resuspended in YPD + 20 μg/ml nocodazole warmed to 37°C. Cells were incubated at 37°C for 30 min, pelleted, washed two times with warmed YPD, and resuspended in 37°C YPD. For analysis of Sec3-GFP and Sec4-GFP localization in cells with an inducible block in mitosis or the MEN, 1-ml aliquots were taken from the culture every 15 min after release from nocodazole arrest, and live cells were quantified for specific GFP-tagged protein localization with a Nikon Eclipse E800 microscope with either a Plan Apo 100×/1.40 or a Plan Apo TIRF 100×/1.45 oil differential-interference contrast objective to assay the number of cells with Chs2-GFP signal at the bud neck. Images were obtained with filtering fluorescent light through a GFP filter set (Chroma Technology, Brattleboro, VT) with a cooled Interline camera (Princeton Scientific Instruments, Monmouth Junction, NJ) by using MetaMorph software. Exposures ranged between 0.2 and 0.7 s. For all the localization experiments, 200 cells were analyzed for each time point and each experiment was repeated at least twice. A Nikon labophot-2 microscope with a Plan 40 0.5 ELWD objective was used to view the morphology of the cells to determine the status of the cell cycle. To analyze cytokinesis defects in ts strains and the GAL-inducible Sec4 strains, samples were taken at time 0 after nocodazole washout and 2 h after. Two milliliter samples were fixed in 5% formaldehyde for 1 h at room temperature and washed two times with and resuspended in 1× phosphate-buffered saline (PBS). Samples were treated with 0.2 mg/ml zymolyase 20T (Seikagaku America, Rockville, MD) in sorbitol buffer (1 M sorbitol in 50 mM KPO4, pH 7.5) containing 2 mM dithiothreitol for 10 to 20 min at 30°C. After cells lost refractile appearance, indicating removal for cell wall, the morphology of the cells were analyzed for completion of cytokinesis by comparing the number of large-budded cells (undivided) versus unbudded cells or small-budded (divided) cells. For assaying actin ring formation, 2-ml samples were collected every 15 min after release from arrest and fixed and zymolyase treated as described above and stained with rhodamine-conjugated phalloidin (Molecular Probes, Eugene, OR) as described in Pringle et al. (1991 ). Cells expressing Cdc14-GFP were treated similarly as described above in relation to arrest and release from nocodazole. Two-milliliter samples were taken every 15 min for 60 min from nocodazole washout, fixed briefly in 1% formaldehyde for 30 min, washed two times with 1× PBS, adhered to a poly-lysine–coated slide, and covered with a mounting solution containing 4,6-diamidino-2-phenylindole (DAPI) to visualize DNA (Pringle et al., 1991 ). Two hundred cells were counted for release of Cdc14-GFP into the cytoplasm for each time point, with this experiment being performed twice. For analysis of Chs2-GFP in wild-type (WT), myo2-16, and sec10-2 cells, 2-ml aliquots were taken every 15 min after release from nocodazole arrest, lightly fixed in 1% formaldehyde for 30 min, and imaged and quantified for Chs2-GFP localization with the fluorescent microscope described above. Three-dimensional (3D) images and movies of Chs2-GFP in WT and sec10-2 cells were produced by imaging lightly fixed cells with the spinning disk confocal microscope described previously, imaging 13 sections in the z-plane, with a spacing of 0.2 μm, with exposures of 0.7 s. Each stack of z-series images was deconvolved and rendered into 3D images and movies by using Autodeblur deconvolution software (AutoQuant Imaging, Troy, NY).
RLY1838 and RLY1840 were cultured overnight in YPD, pelleted, and treated as described above for arrest and release in nocodazole. After washout of nocodazole, cells were collected at 30 min after the release to enrich for cells in cytokinesis. Ten milliliters of cells was pelleted and prepared for electron microscopy (EM) as described in Li (1997 ).
RLY1898, RLY1899, RLY2010, and RLY2011 were prepared and treated with nocodazole as described above. At nocodazole washout, cultures were washed in warmed YPD and resuspended in SC-Leu + 2% glucose at 37°C. One milliliter of these cultures was pelleted and resuspended in 50 μl of warmed SC-Leu media. Five microliters of the resuspension of cells was placed on freshly prepared growth slab as described above but prepared with 1.5% agarose instead in SC-Leu + 2% glucose media. Briefly, 250 μl of the heated mixed agarose solution was placed onto a slide. A second slide is placed on top of the agarose and left to polymerize at room temperature for ~10 min. Slides were carefully pried apart, and the remaining agarose slab was trimmed to the size of the cover glass, which is placed on top of the slab after cells are added and sealed with VALAP. The slides were kept at 37°C and immediately set for imaging on a similar spinning disk confocal microscope as described above but with a 37°C incubation chamber (Solvent Scientific, Segensworth, United Kingdom) encompassing the stage of the microscope to allow for imaging at 37°C. Images were collected and analyzed as above. MetaMorph was used to measure the diameter of the Myo1-GFP ring in each frame of the time-lapse series. The kymographs were produced using the MetaMorph add-on for kymographs, by selecting the signal area (Myo1-GFP ring) with a line, to be recorded in the kymograph. Each horizontal line of the kymograph represents fluorescence across the contractile ring at each time point of the time-lapse series.
To determine the precise timing of localization of the exocytic machinery to the site of cytokinesis (the bud neck), cells expressing GFP-fusion proteins were analyzed using timelapse confocal microscopy. To obtain a baseline of cytokinetic events under our experimental conditions, cells expressing both Myo1-GFP (myosin II), under the native MYO1 promoter and Tub1-GFP (α-tubulin), under the native TUB1 promoter (Straight et al., 1997 ), were analyzed. The signals of Myo1-GFP at the bud neck and Tub1-GFP along the spindle are distinct and separate, thus allowing for imaging of spindle breakdown, which indicates completion of mitosis, and contraction of the actomyosin ring, which indicates the onset of cytokinesis, in the same cell. Videomicroscopy was performed on cells in anaphase in an asynchronous culture, as indicated by their elongated spindles. Figure 1A (top) shows selected images from a representative time-lapse series (also see Supplemental Movie 1), where the spindle disassembled in less than a minute after the onset of actomyosin ring contraction (time point 4.5). Analysis of 10 time-lapse series showed that the average time separating the two events was 0.4 ± 0.3 min. The entire process of contraction took 4.7 ± 1.3 min (Figure 1B).
The above-mentioned analysis suggests that the onset of contraction and spindle breakdown are nearly concurrent, allowing the use of Tub1-GFP and spindle breakdown as a timing marker for the onset of cytokinesis. To determine the timing of exocytosis relocation to the bud neck relative to the onset of cytokinesis, cells expressing fusion proteins Sec3-GFP (under the native SEC3 promoter) and Tub1-GFP were analyzed using time-lapse microscopy. Sec3 is a component of the exocyst complex thought to be important for marking the site to which the rest of the complex localizes. Sec3 localizes to the bud tip during the early parts of the cell cycle then relocates to the bud neck late in mitosis (Finger et al., 1998 ). To identify the timing of Sec3-GFP localization to the bud neck, cells from an asynchronous culture with elongated spindles but not yet showing bud neck localized Sec3 were chosen for time-lapse imaging. Localization of Sec3-GFP to the bud neck was defined as when the protein accumulated at the bud neck or formed a stable ring-like structure. Figure 1A (middle) and Supplemental Movie 2 show an example of the cells imaged, where Sec3-GFP arrived at the bud neck at time point 0.5, several frames before spindle breakdown at time point 5.5. Analysis of 10 time-lapse series demonstrates that the average time of arrival of Sec3-GFP to the bud neck occurred at 3.9 ± 1.9 min before spindle breakdown (Figure 1B).
Next, we determined the time of vesicle arrival at the bud neck. Sec4 is a Rab GTPase required for exocytosis and was previously used as a marker for detecting Golgi-derived vesicles in vitro and in vivo. yEGFP-Sec4 was expressed under the YOP1 promoter carried on the CEN plasmid pRS315 (Schott et al., 2002 ). As before, cells in anaphase were chosen for imaging where Sec4-GFP had not yet localized to the bud neck and the same criteria used for the localization of Sec3-GFP were applied for Sec4-GFP. As shown in Figure 1A (bottom) and Supplemental Movie 3, Sec4-GFP arrived at the bud neck at time point 0.5, also several frames before spindle breakdown (time point 4). Data from six time-lapse series showed that Sec4-GFP arrived at the bud neck 3.4 ± 1.0 min before spindle breakdown (Figure 1B). These results show that localization of exocytic vesicles as well as the complex required for vesicle fusion at the bud neck occurred 3–4 min before the onset of cytokinesis.
During bud formation, vesicles are targeted to the site of polarized growth through myosin V/actin cable-based transport (Pruyne et al., 1998 ). To test whether this is also true for trafficking of cytokinesis-associated vesicles, we tested whether Sec4 accumulation at the bud neck was affected by myo2-16, a temperature-sensitive mutation in the cargo-binding domain of the major yeast myosin V, which has been shown to depolarize Sec4 from growing bud tips (Schott et al., 1999 ). To see how the myo2-16 mutation effects localization of Sec4-GFP specifically at cytokinesis, wild-type and mutant cells were synchronized in metaphase with nocodazole, a microtubule-depolymerizing drug, as described in the Materials and Methods. Sec4-GFP localization was analyzed in live cells in late anaphase (around 30–45 min after release from arrest) by fluorescent microscopy. Although bud neck localization of Sec4-GFP was not affected at the nonpermissive temperature of 37°C in wild-type cells, in myo2-16 cells, Sec4-GFP did not accumulate at the bud neck in anaphase cells but instead occurred in punctuate structures throughout the cytoplasm (Figure 1C), as was seen previously in small-budded cells stained for Sec4 (Schott et al., 1999 ). Therefore, vesicle trafficking to the cytokinesis site requires the actin cable/myosin V-based transport system.
Studies in sea urchin embryos suggested that targeting secretion to the cleavage furrow is dependent on the exit from mitosis, a cell cycle transition that requires degradation of the B-type mitotic cyclin (Shuster and Burgess, 2002 ). In budding yeast, the known events triggered by mitotic exit are spindle breakdown and onset of actomyosin ring contraction (Tolliday et al., 2001 ). Because localization of Sec3 and Sec4 to the bud neck occurred several minutes before these events, we tested whether targeted secretion to the cytokinesis site is universally dependent on mitotic exit by examining the effect of blocking mitotic exit with a nondegradable mitotic cyclin (Clb2ΔDB) on the localization of Sec3 and Sec4 to the bud neck. Clb2ΔDB was previously shown to arrest cells in anaphase (Surana et al., 1993 ). Cells expressing Sec3-GFP and Clb2ΔDB under the control of the inducible Gal1 promoter, or the vector (pRL62, no Clb2ΔDB) control, were cultured in YPR overnight and then shifted into YPGR media containing nocodazole. Under these conditions, Clb2ΔDB expression was induced while cells accumulated in metaphase. After 3 h, nocodazole was washed out, and cells were released into YPGR media. Samples were taken every 15 min, and live cells were quantified for localization of Sec3-GFP at the bud neck. As shown in Figure 2A, Sec3-GFP did not localize to the bud neck in cells expressing Clb2ΔDB and arrested in anaphase as expected, whereas cells without Clb2ΔDB exhibited Sec3-GFP localization to the bud neck as they underwent cytokinesis. The cells expressing Clb2ΔDB formed normal actin rings (our unpublished data), as also observed in other cells with mutations in mitotic exit regulators (Lippincott et al., 2001 ; Luca et al., 2001 ).
Similar experiments were carried out on cells expressing GFP-Sec4, Tub1-GFP, and Clb2ΔDB. After release from the nocodazole arrest, Sec4-GFP localization to the bud neck was observed in cells with and without Clb2ΔDB expression (Figure 2B). However, Sec4-GFP localization in the former occurred less frequently, did not seem as intense as in the control cells, and the signal was never seen as a full Sec4-GFP ring (Figure 2C, a). Because blocking mitotic exit prevented Sec3 localization to the bud neck, it was possible that the observed defect in Sec4 accumulation was directly due to a failure of vesicles to fuse with the plasma membrane, followed by vesicle dispersal. To test this possibility, we examined the effect of the sec3-2 temperature-sensitive mutation on Sec4-GFP localization to the bud neck. sec3-2 at the nonpermissive temperature blocks secretion and results in accumulation of vesicles in the cytoplasm (Finger and Novick, 1997 ). In anaphase sec3-2 cells at the nonpermissive temperature, accumulation of Sec4-GFP at the bud neck was even more prominent than in wild-type cells: Sec4-GFP localized not in a compact region but to a wider area around the bud neck (Figure 2C, b). Electron microscopy of sec3-2 cells in anaphase also showed that vesicles accumulated in a broad area around the bud neck, in contrast to wild-type cells (Figure 2C, c). This result suggests that the partial defect in Sec4-GFP localization observed in Clb2ΔDB arrested cells was unlikely to be due to a fusion defect. Taken together, these results suggest that localization of the vesicle fusion machinery and, to a lesser extent, transport of vesicles to the bud neck require inactivation of Cdk1 at exit from mitosis.
The MEN inactivates mitotic Cdk1 by releasing the phosphatase Cdc14 from the nucleolus to activate APCCdh1, which degrades cyclin, and to promote the transcription and stabilization of the Cdk1 inhibitor Sic1 (Bardin and Amon, 2001 ). However, it was previously shown that the MEN has a separate control on cytokinesis apart from Cdk1 inactivation (Lippincott et al., 2001 ; Luca et al., 2001 ; Menssen et al., 2001 ). To better understand the regulation of exocytosis in anaphase, we examined whether the localization of Sec3-GFP is directly controlled by the MEN. Here, we bypassed the block in mitotic exit caused by a temperature-sensitive MEN mutant, mob1-77, at the nonpermissive temperature by the overexpression of Sic1 from a high copy 2μ plasmid. mob1-77 cells at the nonpermissive temperature can form actomyosin contractile rings (Luca et al., 2001 ). With Sic1 overexpression, cell cycle progression proceeded due to bypass of the mitosis block in mob1-77 cells, but cytokinesis fails due to lack of contraction of the actomyosin ring, resulting in the formation of chains of cells (Luca et al., 2001 ). MOB1 2μ + SIC1, mob1-77 + 2μ SIC1, and mob1-77 cells were arrested with nocodazole in metaphase. During arrest, cells were warmed to the nonpermissive temperature and then released from arrest. Samples were assayed every 15 min for visualization of Sec3-GFP localization at the bud neck. The results in Figure 2D show that the localization of Sec3-GFP to the bud neck occurred normally in MOB1+ 2μ SIC1 and mob1-77 + 2μ SIC1 cells, in contrast to that in mob1-77 cells, which arrested in anaphase and were unable to localize Sec3-GFP. Hence, targeted exocytosis at the bud neck is dependent on Clb2 degradation but not on the MEN.
Having established that vesicle targeting and localization of the exocyst are events that immediately precede cytokinesis, we wanted to test whether these events are required for the completion of cytokinesis. Temperature-sensitive myo2 (myosin V), sec3, sec10 mutations and a dominant negative SEC4 (Sec4N34) construct were used. Because these mutations also block bud growth, cells were first synchronized in metaphase by nocodazole treatment at the permissive temperature to allow bud growth and then released from the arrest into the nonpermissive temperature (for testing dominant negative Sec4, the cells were arrested with nocodazole and released into media, allowing expression of Sec4N34). The percentage of cells that completed cytokinesis was determined as the reduction of the large-budded population over time. Because we were specifically interested in cytokinesis as opposed to cell separation, the cells were fixed and the cell was wall removed by zymolyase treatment before scoring (Lippincott and Li, 1998b ). The values presented in Figure 3A represent the percentage of decrease in the large-budded population from 0 to 2 h after release from the nocodazole arrest under nonpermissive conditions. In the wild-type culture, ~60% of the cells divided 2 h after release, compared with <10% in various mutants, suggesting that most of the mutant cells were unable to undergo cytokinesis. Chromosome segregation in these cells, however, occurred normally as seen by staining cells with DAPI after release from the nocodazole arrest (our unpublished data).
It was possible, however, that blocking secretion activates a cell cycle checkpoint response that prevents mitotic exit. To test whether this were true, we examined the release of Cdc14 from the nucleolus, a critical event downstream of the MEN (Bardin and Amon, 2001 ). Wild-type and mutant cells were transformed with a plasmid that expressed GFP-tagged Cdc14 under the native CDC14 promoter (Cdc14-5 × GFP) (Yoshida et al., 2002 ). Previous studies showed that Cdc14 is initially released from the nucleolus at early anaphase through the FEAR network, maintained in the cytosol during late anaphase by the MEN, and then returned to the nucleolus in the next G1 (Bardin and Amon, 2001 ). Cdc14-5xGFP cells were released from the nocodazole arrest, and samples were taken every 15 min after release and prepared for visualization of the GFP signal and DNA by DAPI staining. Cdc14-GFP localization was quantified for each time point, and the results are shown in Figure 3B. In SEC3 cells, Cdc14 resides in the nucleolus in metaphase cells, released into the cytoplasm in anaphase cells at the ~30-min time point, and returned to the nucleolus ~60 min. The same timing of Cdc14 release and resequestration was observed in sec3-2 cells at the nonpermissive temperature, suggesting that the mutation did not block cytokinesis by preventing mitotic exit. Together, these results suggest that disruption of vesicle transport and fusion prevents the completion of cytokinesis.
Another possible explanation for the effect of vesicle transport and fusion mutations on cytokinesis was that the secretory pathway is required for the assembly of the actomyosin ring. In particular, previous work showed that the formation of the actin ring is dependent upon the activity of Rho1, which activates formin family actin nucleators (Tolliday et al., 2002 ). Rho1 is a membrane-associated GTPase shown to interact with the exocytic machinery (Guo et al., 2001 ), and it might require membrane-based transport system for its localization and function. To test this possibility, we tested whether the mutations that disrupt vesicle targeting and fusion affect the assembly of the actin ring. Again, cells were released from the nocodazole arrest into the nonpermissive temperature. Samples of cells were taken every 15 min, fixed, and stained with rhodamine-phalloidin. As shown in Figure 4, sec10-2, sec3-2, and myo2-16 (myosin V) mutant cells had no defect in actin ring formation, compared with their congenic wild-type strains. Similarly, Myo1 (myosin II) localization to the bud neck also was unaffected in these mutants (Figure 5). Therefore, disruption of vesicle transport and fusion do not affect the formation of the contractile ring.
Next, we asked whether the exocyst mutations affect the contraction of the actomyosin rings. This might be expected if actomyosin ring contraction could not overcome the membrane tension in the absence of vesicle addition. To analyze ring contraction, time-lapse confocal microscopy was carried out on wild-type, sec3-2, or sec10-2 cells expressing Myo1-GFP and Tub1-GFP (α-tubulin). The cells were released from nocodazole arrest at 37°C after a 30-min incubation at 37°C. The cells were then transferred to an imaging platform enclosed in a 37°C chamber. Figure 5A (top) and Supplementary Movie 4 show a typical example of ring contraction in a wild-type cell where contraction continued until Myo1-GFP signal narrowed to a point near the center of the bud neck, which subsequently disappeared. Ring contraction was visibly different in sec3-2 and sec10-2 mutant cells. Typical examples are shown in Figure 5A (middle and bottom) and Supplemental Movies 5 and 6, where contraction started with normal timing, i.e., concurrent with spindle disassembly, however, ring disassembly occurred before the contraction was complete. This difference was better depicted in kymographs where fluorescence profiles across the 3D projections of the Myo1 ring of all frames of the movie were compiled into one image, starting ~2 min before the initiation of contraction (Figure 5B). There was no difference between the average size or intensity of the Myo1-GFP ring in the wild-type and mutants at the beginning of each time-lapse series (Table 2). In wild-type cells, the contraction phase of the kymograph was centered and cone-shaped with a pointy apex, whereas those in the mutants had flat, and sometimes off-centered apexes. Table 2 shows that the average width of the apex in the wild-type was ~1.5 times smaller then that of sec3-2 and sec10-2 cells before complete ring disassembly. t tests showed that the differences in apex width between wild-type and sec3-2 (t = 0.016) and between wild-type and sec10-2 (t = 0.002) were significant. Therefore, the mutations that block vesicle fusion led to instability of the contractile ring during cytokinesis.
One possible explanation for the instability of the actomyosin ring in the exocyst mutants could be failed delivery of a membrane-bound stabilization factor. Of all proteins known to localize to the bud neck during anaphase, Chs2, a transmembrane protein synthesized and localized to the bud neck late in the cell cycle (Choi et al., 1994 ; Chuang and Schekman, 1996 ) is a likely candidate. The activity of Chs2 is required for the production of the primary septum that is laid down behind the contractile ring during cytokinesis (Shaw et al., 1991 ), and Δchs2 mutation phenocopies Δmyo1 mutation (Schmidt et al., 2002 ). To test this possibility, we first examined the precise timing of Chs2 delivery to the bud neck. By time-lapse imaging of cells expressing Chs2-GFP (under the native CHS2 promoter on the CEN plasmid, pRL73; Tolliday et al., 2003 ) and Tub1-GFP (α-tubulin), we found that Chs2-GFP arrives at the bud neck 2.4 ± 0.8 min before spindle breakdown, which was slightly later than Sec3 and Sec4 localization (Figure 1). Further analysis of the relative intensity (as percentage of the maximum intensity) of the Chs2-GFP signal showed that in seven of 10 cells recorded, Chs2-GFP at the bud neck peaked after spindle disassembly, indicating that delivery of Chs2-GFP to the plasma membrane continued after contraction initiation (Figure 6A). In the other three cells, Chs2-GFP intensity peaked just before or at the time of spindle disassembly. This result coincides with fluorescence recovery after photobleaching data of Chs2, which showed rapid recovery, consistent with continued delivery to the bud neck (Dobbelaere and Barral, 2004 ).
The delivery of Chs2 to the bud neck was shown not to be dependent on the chitosome, the organelle that targets the other chitin synthases to the plasma membrane (Chuang and Schekman, 1996 ). To determine whether Chs2 is delivered via Golgi-derived vesicles and the actin cable/myosin V-based transport system, Chs2-GFP localization was analyzed in wild-type and myo2-16 cells. Cells cultured overnight at room temperature were arrested in media containing nocodazole and then released from the arrest at the nonpermissive temperature of 37°C. Samples were taken every 15 min and quantified for Chs2-GFP localization at the bud neck. In myo2-16 cells, most Chs2 localized to punctate structures distributed in both the mother and bud (Figure 6B), a pattern similar to the localization of Sec4-GFP in myo2-16 cells (Figure 1C). This suggests that Chs2 is delivered to the bud neck using the actin cable–myosin V-based transport pathway.
Using the same procedure as described above, we also examined the effects of the sec10-2 mutation on Chs2 localization at the nonpermissive temperature. Chs2-GFP could localize to the bud neck in sec10-2 cells, at a slightly lower rate (Figure 6C). However, instead of localizing to a tight band at the bud neck as observed in wild-type cells, Chs2 localizes broadly at the bud neck of the mutant cells (Figure 6, C and D, and Supplemental Movies 7 and 8). Similar changes in localization of other vesicle-associated proteins have been observed in other exocyst mutants (Finger et al., 1998 ; Figure 2B). To further characterize this abnormal localization, cells from the 45-min time point after nocodazole washout were observed using a spinning disk confocal microscope. A stack of images for each cell were collected along the z-axis and were deconvolved to produce 3D images of Chs2-GFP localization in wild-type and sec10-2 mutant cells (Figure 6D and Supplemental Movies 7 and 8). Chs2-GFP clearly localized to a ring in the bud neck plasma membrane in the wild-type, whereas in sec10-2 cells, Chs2-GFP localizes to the interior of the bud neck in a dispersed pattern, suggesting that Chs2-GFP was not incorporated in the plasma membrane after being delivered to the bud neck. Together, these results demonstrate that Chs2 is deposited at the bud neck before and during actomyosin ring contraction in a myosin V- and exocyst-dependent manner.
We next asked whether Chs2 was acting as a stability factor during the contraction of the actomyosin ring. To analyze the role of Chs2 in actomyosin ring contraction, we used a strain with a deletion of CHS2 (chs2Δ) expressing Myo1-GFP for time-lapse microscopy. chs2Δ cells are viable and able to divide but do so inefficiently by the formation of aberrant secondary septa (Shaw et al., 1991 ). We imaged Myo1 ring contraction in chs2Δ cells, along with wild-type and sec10-2 cells. In these experiments we imaged the Myo1-GFP ring that happened to be in a plane parallel or diagonal to that of the slide. This allowed clear visualization of ring shrinkage during contraction in wild-type cells (Figure 7 and Supplemental Movie 9; Bi et al., 1998 ). In contrast, in both the exocyst mutant sec10-2 and the chs2Δ cells, the ring abruptly broke at one or two points at the time of cytokinesis and then disassembled in a zipping manner from the broken ends (Figure 7 and Supplemental Movies 10–12). The previous movies (Figure 5 and Supplemental Movies 5 and 6) with the ring in a plane perpendicular to the slides would not have distinguished between this end-directed disassembly from contraction. This result suggests that in Δchs2 and sec10 mutant cells, the actomyosin ring does not contract but instead breaks and then rapidly disassembles.
Spatially and temporally precise insertion of new membrane at the cleavage furrow has been a well documented phenomenon since the 1970s (Bluemink and de Laat, 1973 ); however, understanding the molecular basis of this fascinating cellular process has only recently become a focus of the cytokinesis field. Above, we described the characterization of targeted secretion during budding yeast cytokinesis. We show that secretion is targeted to the site of cell division at a precise stage in the cell cycle and is negatively regulated by the mitotic kinase Cdk1. This targeted secretion is mediated through an actin/myosin V-based transport system and the plasma membrane-associated exocyst complex. Inhibition of membrane trafficking during mitosis blocks the completion of cytokinesis, and this effect correlates with a decreased stability of the actomyosin ring during the contraction phase. Further analysis suggested that the membrane protein Chs2, a chitin synthase, is an essential stability factor for the contractile ring. These results establish budding yeast as a model for understanding the regulation and mechanistic roles of membrane trafficking during cytokinesis.
Data from studies of embryonic cell divisions clearly indicated that membrane addition to the equatorial region is a late cell cycle event that occurs independently of the contractile ring activity (Bluemink and de Laat, 1973 ; Drechsel et al., 1997 ; Shuster and Burgess, 2002 ). Previous work in yeast also has suggested that the secretory pathway, oriented early in the cell cycle toward the growing bud tip, is targeted to the bud neck region late in the cell cycle (Finger et al., 1998 ). It was unclear, however, whether this event is associated with cytokinesis or with the process of septum formation and resolution, which follows the closure of the plasma membrane. We have observed that localization of the exocyst component Sec3, and the post-Golgi vesicle-associated Rab, Sec4, to the bud neck occurs 3–4 min before the onset of actomyosin ring contraction. The membrane protein Chs2 is deposited at the bud neck through the secretory pathway just before and during the actomyosin ring contraction. These results suggest that vesicle transport and fusion at the bud neck occur just before and during cytokinesis.
Transport of secretory vesicles toward the site of polarized growth during bud formation is mediated by the actin cable and myosin V-based cytoskeletal system (Pruyne et al., 1998 ). Thus, it is not surprising that delivery of secretory vesicles, marked with GFP-Sec4, to the bud neck during cytokinesis also is achieved through this transport system. An interesting question is how this transport system reorients late in mitosis. Here, two events are likely to be important: 1) disassembly of bud tip-oriented actin cables at mitotic entry; and 2) assembly of bud neck-directed cables at mitotic exit. It is interesting to note that the peak for the assembly of bilateral arrays of actin cables around the bud neck occurs rather late during the cell division process, i.e., after disassembly of the contractile ring (Tolliday et al., 2002 ). However, a few bud neck-oriented cables can be found even before mitotic exit, which could be important for the initial targeting of Sec4-positive vesicles toward the bud neck. Although it is thought that in most metazoan cell types, furrow-bound vesicles are transported mostly along microtubules, there is no definitive evidence ruling out an involvement of actin-based transport mechanisms. Most studies looking at vesicle transport used cytochalasin D (Bluemink and de Laat, 1973 ; Shuster and Burgess, 2002 ), a drug that does not always result in loss of actin filaments (Kolega et al., 1991 ). It is possible that actin plays a role in short-range deposition of secretory vesicles and recycled endocytic vesicles within the cleavage furrow.
It is significant that targeted secretion and actomyosin ring contraction seem to be regulated by different cell cycle pathways (Figure 8A). Previous work showed that the MEN possibly controls contractile ring activation in two ways: 1) inactivation of the mitotic Cdk1 (Bardin and Amon, 2001 ); and 2) activation of actomyosin ring contraction without involving Cdk1 (Lippincott et al., 2001 ; Luca et al., 2001 ; Menssen et al., 2001 ). The latter was suggested by the observations that the MEN mutants still exhibited a cytokinesis block with a fully assembled contractile ring even when mitotic Cdk1 was inactivated. In fact, mitotic Cdk1 may indirectly inhibit contractile ring activation by inhibiting the MEN. The cell cycle control of cytokinesis-specific secretion, on the other hand, seems to be the opposite: whereas the MEN does not seem to be directly required for localizing the exocyst (Sec3) to the bud neck, nondegradable Clb2 completely blocks the localization of Sec3 and partially prevented targeting of vesicles (Sec4) to the bud neck. This finding is surprisingly consistent with how cytokinetic events are controlled in dividing sea urchin zygotes. It was found that whereas contractile ring activation, as indicated by the ingression of the cleavage furrow, was not strictly dependent on mitotic cyclin degradation, secretion of the extracellular matrix protein hyalin was completely blocked by nondegradable cyclin (Shuster and Burgess, 2002 ). These findings suggest that, despite the morphological and size differences between budding yeast and sea urchin zygotes, as well as the difference between their transport pathways (microtubules vs. actin-based), the mechanisms by which major cytokinetic events are controlled during the cell cycle seem to be more conserved than one might expect.
In plant cells, new membrane addition is obviously important for cytokinesis; fusion of vesicles leads to the growth of a cell plate that divides one cell into two (reviewed in Field et al., 1999 ). When considering why new membranes might be needed for cytokinetic processes that involve actomyosindriven furrow formation, two possible roles have been discussed (Finger and White, 2002 ), one mechanical and one biochemical. A mechanical role for membrane insertion could prevent the buildup of membrane tension generated during ingression of the cleavage furrow due to an increase in surface area. A biochemical role could lead to changes in the molecular composition at the cleavage site through delivery of membrane bound factors such as specific lipids and proteins. These factors could be important for force generation by the contractile ring or could facilitate the final membrane fusion that disconnects the two progeny cells. Recent work suggested that membrane trafficking could play a role in the formation of lipid raft domains in the cleavage furrow. Raft components are present in isolated midbodies (Skop et al., 2004 ) and disruption of these domains impaired cytokinesis (Wachtler et al., 2003 ).
For a yeast cell, where the bud neck is already very narrow at the onset of cytokinesis (1–2 μm in diameter), cell division does not involve a drastic increase in surface area. It remains possible, however, that the turgor pressure on yeast plasma membrane (Harold, 1990 ) is too strong to be overcome by the contractile force provided from the actomyosin ring alone. From our results, it seems that membrane insertion definitely performs a biochemical role because the addition of Chs2 by vesicular fusion is required for the stability of the actomyosin ring during the contraction phase. A recent study by Dobbelaere and Barral (2004 ) found that the rate of Myo1 ring contraction was significantly slowed in sec3-4 cells compared with the wild-type (Dobbelaere and Barral, 2004 ). These differences could be due to a difference in strain background, mutant alleles, or experimental conditions in the two studies. We note, however, that even in the single contraction kymograph shown in that study, the intensity of the Myo1 ring declined faster in the sec3-4 mutant than in the wild-type (Dobbelaere and Barral, 2004 ), suggesting that the stability of the Myo1 ring also might be affected. Additionally, without imaging the contractile ring parallel or diagonal to the plane of the slide, it can be difficult to distinguish between contraction and disassembly.
The yeast Myo1 (myosin II) ring forms early in the cell cycle, and its assembly and stabilization require the septin hourglass-like structure that serves as a scaffold for many bud neck-associated components (Bi et al., 1998 ; Lippincott et al., 2001 ). At the onset of cytokinesis, the septin hourglass splits into two rings that sandwich the now mature contractile ring (Lippincott et al., 2001 ). Chs2 could be deposited at this point to assume the septins' role in contractile ring stabilization and also may be important for tethering the contractile ring to the plasma membrane for effective force generation (Figure 8B). Presently, we do not know whether the function of Chs2 in contractile ring stabilization is mediated through a direct interaction of its cytoplasmic domain with the contractile ring or through its enzymatic activity that leads to chitin deposition. The newly deposited chitin could serve as an extracellular scaffold that recruits a factor that stabilizes the contractile ring; or alternatively, reinforcement of the furrowing plasma membrane by chitin, together with insertion of new membrane, could ease the tension on the actomyosin ring and prevent its breakage. A defect in contractile ring stability could explain the late cytokinesis failure observed in C. elegans embryos treated with brefeldin A (Skop et al., 2001 ). Premature disassembly of the contractile ring could result in regression of the cleavage furrow due to tension in the plasma membrane. Clearly, Chs2 is a yeast specific factor for stabilization of the contractile ring, but proteins with a function linking the contractile ring with the extracellular matrix may exist in animal cells.
We thank C. Kaiser, A. Bretscher, R. Collins, F. Luca, A. Toh-e, E. Cabib, and N. Tolliday for providing strains and plasmids. We are grateful to J. Waters and the Nikon Imaging Center at Harvard Medical School for being very helpful with confocal microscopy, image acquisition, and analysis, and to M. Ericsson for help with electron microscopy. We also thank J. Hastings and A. DeGiacomo for technical assistance, and I. Lister, P. Marina Losada, and C. Field for critical reading of the manuscript and helpful discussions. This work is supported by NIH grant GM-059964 (to R. L.).
This article was published online ahead of print in MBC in Press (http://www.molbiolcell.org/cgi/doi/10.1091/mbc.E04-12-1090) on March 16, 2005.
The online version of this article contains supplemental material at MBC Online (http://www.molbiolcell.org).