Cell fusion in yeast is negatively regulated by components of the cell wall integrity pathway.

During mating, Saccharomyces cerevisiae cells must degrade the intervening cell wall to allow fusion of the partners. Because improper timing or location of cell wall degradation would cause lysis, the initiation of cell fusion must be highly regulated. Here, we find that yeast cell fusion is negatively regulated by components of the cell wall integrity (CWI) pathway. Loss of the cell wall sensor, MID2, specifically causes "mating-induced death" after pheromone exposure. Mating-induced death is suppressed by mutations in cell fusion genes ( FUS1, FUS2, RVS161, CDC42), implying that mid2Δ cells die from premature fusion without a partner. Consistent with premature fusion, mid2Δ shmoos had thinner cell walls and lysed at the shmoo tip. Normally, Cdc42p colocalizes with Fus2p to form a focus only when mating cells are in contact (prezygotes) and colocalization is required for cell fusion. However, Cdc42p was aberrantly colocalized with Fus2p to form a focus in mid2Δ shmoos. A hyperactive allele of the CWI kinase Pkc1p ( PKC1*) caused decreased cell fusion and Cdc42p localization in prezygotes. In shmoos, PKC1* increased Cdc42p localization; however, it was not colocalized with Fus2p or associated with cell death. We conclude that Mid2p and Pkc1p negatively regulate cell fusion via Cdc42p and Fus2p.

INTRODUCTION

Cell fusion is an essential process in eukaryotic organisms. In mammals, in addition to fertilization, fusion occurs during a variety of developmental processes, including placental trophoblast fusion during pregnancy, osteoclast formation, and the fusion of myoblasts to form myofibers during skeletal muscle development (Abmayr ; Abmayr and Pavlath, 2012; Alper and Podbilewicz, 2008; Huppertz and Borges, 2008; Ishii and Saeki, 2008; Jansen and Pavlath, 2008; Wassarman and Litscher, 2008; Mugnier ). Inhibition of cell fusion is detrimental; for example, blockage of placental trophoblast fusion has been linked to preeclampsia during pregnancy (Gauster ) and a defect in myoblast fusion causes a congenital myopathy (Di Gioia ). Cell fusion must be temporally and spatially controlled so that cells fuse with the right partner and in the correct location.

During mating in Saccharomyces cerevisiae, two haploid cells fuse to form a diploid zygote (Madhani, 2007). Mating types a and α secrete specific pheromones that are detected by the opposite mating type. The cells polarize their growth along the pheromone gradient forming polarized cells called “shmoos.” During shmooing, cells prepare for fusion, clustering vesicles near the zone of cell fusion that are thought to contain the hydrolases required for cell wall degradation (Baba ; Gammie ). When the mating cells come into contact with each other, continued growth and cell wall remodeling leads to a flat interface between the mating partners, a region called the zone of cell fusion (ZCF). During cell fusion, the cell wall is degraded starting at the center of the ZCF, allowing the plasma membranes to come into close contact. Plasma membrane fusion allows a continuous cytoplasm through which the two nuclei move together and eventually fuse (Gammie ; Ydenberg and Rose, 2008; Merlini ). Because aberrant cell wall degradation in this single-celled organism would lead to lysis and death, the correct timing and location of cell wall degradation is essential for the survival of S. cerevisiae.

Several key regulators of cell fusion have been described. Fus2p and Rvs161p form a heterodimeric amphiphysin that is transported to the shmoo tip, where it is retained by Fus1p, Kel1p, and actin (Stein ; Smith and Rose, 2016). Mutations affecting Fus1p, Fus2p, Kel1p, and Rvs161p cause failure of cell wall removal at the ZCF (Trueheart ; Trueheart and Fink, 1989; Gammie ; Smith and Rose, 2016). In wild-type (WT) mating pairs, after cell contact but before fusion, vesicles are observed to cluster adjacent to the ZCF (Gammie ). Rvs161p and Fus2p are not required to localize the vesicles. In fus1 mutant prezygotes, the vesicles fail to localize, suggesting that Fus1p functions upstream of Rvs161p/Fus2p in the pathway (Trueheart ; Gammie ; Paterson ). Activated Cdc42p binds to Fus2p (Ydenberg ), and the Cdc42p/Fus2p/Rvs161p complex is postulated to catalyze exocytosis to release hydrolases into the ZCF to break down the cell wall (Gammie ; Paterson ). Cdc42p is a highly conserved Rho-GTPase that is involved in several steps in mating and cell fusion (Madden and Snyder, 1998; Johnson, 1999; Richman ; Kozminski ; Adamo ; Perez and Rincon, 2010). In addition to signaling, Cdc42p plays a role in polarity establishment, actin cytoskeleton rearrangement, and secretion (Simon ; Zhao ; Nern and Arkowitz, 1998, 1999; Barale ; Ydenberg ). It has recently been shown that Cdc42p forms a discrete focus at the center of the ZCF during mating, dependent on Fus2p, Rvs161p, Fus1p, and Kel1p (Smith ).

During normal mitotic growth, cell wall damage is rapidly repaired through the actions of the cell wall integrity (CWI) pathway. The CWI pathway contains five plasma membrane–associated integral-membrane proteins (Wsc1p, Wsc2p, Wsc3p, Mid2p, and Mtl1p) each containing highly O-mannosylated domains that extend into the cell wall (Gray ; Verna ; Jacoby ; Ketela ; Rajavel ; Philip and Levin, 2001). The transmembrane proteins are hypothesized to be mechanosensors; changes in the cell wall sensed by the extracellular domains are thought to be transduced across the plasma membrane to intracellular regulatory proteins (Dupres ; Heinisch and Dufrene, 2010). The CWI pathway has been shown to respond to cell wall stress from growth, polarization, heat, and drugs (Gray ; Verna ; Jacoby ; Reinoso-Martin ; Bermejo ). The master regulator of the CWI pathway is Rho1p, a Rho-GTPase related to Cdc42p, which when bound by GTP, activates the sole protein kinase C (Pkc1p) in S. cerevisiae. Rho1p also functions in cell wall deposition, through association with Fks1p, the β-1,3 glucan synthase, and has effects on actin cytoskeleton organization and polarized secretion (Drgonova ; Mazur and Baginsky, 1996; Ozaki ; Qadota ). Pkc1p activates downstream components of the CWI pathway, which culminates in the transcriptional activation of genes involved in cell wall biogenesis and the cell cycle (Levin, 2005, 2011). Pkc1p localizes to sites of polarized growth, where cell wall remodeling is active (Heinisch and Rodicio, 2018). However, in response to localized cell wall damage, Pkc1p rapidly relocalizes to the site of damage (Levin ; Nonaka ; Kamada ; Kono ). Given a very active surveillance of CWI, how do cells allow cell wall degradation to occur at the ZCF during mating? We hypothesize that the CWI pathway must be locally down-regulated during cell fusion.

Consistent with this view, a dominant hyperactive PKC1 mutation (PKC1-R398P, referred to as a PKC1* in this article) causes a cell fusion defect (Philips and Herskowitz, 1997). One interpretation of this observation was that Pkc1p may negatively regulate cell fusion, and that the negative regulation must be overcome for cells to fuse (Philips and Herskowitz, 1997). If Pkc1p regulates cell fusion, it is likely that upstream components of the CWI pathway do as well. Of the five transmembrane proteins in the CWI pathway, Wsc1p and Mid2p are the most important. In haploids, loss of WSC1 is sufficient to cause death by cell lysis at high temperatures. MID2 stands for “mating-induced death,” because mid2Δ cells lose viability after exposure to mating pheromone (Errede ; Buehrer and Errede, 1997; Gray ; Verna ; Ketela ; Rajavel ). However, the basis of mating-induced death has remained unclear. Suppression of mid2Δ by increased levels of calcium suggested that misregulated calcium signaling might be involved, as it is for the mid1 mutation affecting a calcium transporter (Iida ). In mitotic cells, deletion of WSC1 and MID2 together causes cell death at normal temperatures, which can be suppressed by osmotic support (1M sorbitol; Ketela ; Rajavel ). Even WT cells undergo a degree of pheromone-induced death in the extended absence of mating partners, affecting up to 25% of cells (Zhang ). Death occurs in two waves: a fast wave (∼1.5 h after pheromone addition) and a slow wave (∼9.5 h after pheromone addition). The CWI pathway, as well as Fus1p, Fus2p, and Rvs161p, play a role in the fast cell death, whereas calcium signaling via calcineurin is required for slow cell death (Zhang ). Although the cause of fast cell death was not identified, it was suggested that it entailed increased levels of reactive oxygen species (Zhang ).

Here we examine the role of Mid2p and Wsc1p in cell fusion. We show that the cause of mid2Δ death is unregulated cell fusion, leading to premature cell wall removal in shmoos before prezygote formation. Mid2p and Pkc1p regulate Cdc42p localization in shmoos and prezygotes. Our data suggest a role for Mid2p as a specific negative regulator of Fus2p in cell fusion, whereas Pkc1p regulates Cdc42p localization differently during polarization and fusion.

RESULTS

mid2Δ pheromone-induced death is due to loss of plasma membrane integrity

Cell wall degradation must be tightly regulated during mating to allow cells to properly fuse, while also preventing misplaced or premature fusion. A hyperactive allele of the sole protein kinase C in yeast, Pkc1p (PKC1-R398P, referred to as PKC1*), causes a cell fusion defect, suggesting that Pkc1p acts as a negative regulator of cell fusion (Philips and Herskowitz, 1997). Because Pkc1p is the central kinase in the CWI MAP kinase cascade (Levin ; Nonaka ), we hypothesized that the upstream transmembrane sensor proteins in the CWI pathway (Mid2p, Mtl1p, Wsc1p, Wsc2p, and Wsc3p), could serve as initial signals of this negative regulation of fusion. WSC1 and MID2 are sufficient to support mitotic growth; loss of both makes cells inviable in the absence of osmotic support (Supplemental Figure S1). We tested the effects on mating of deletion mutants in each of the genes; all five mutants exhibited WT levels of mating. However, mid2Δ mutants uniquely showed a mating-specific phenotype, pheromone-induced death in low calcium media, reaching ∼70% inviability within 5 h (Rajavel ; Philip and Levin, 2001). Although the pheromone sensitivity of mid2Δ was known previously (Ono ; de Bettignies ; Rajavel ; Philip and Levin, 2001), the basis of mating-induced death remained unclear. The intact appearance of the cells, and the suppression of death by calcium or PKC1*, but not by osmotic support, suggested that death might be due to a downstream calcium signaling defect (Ketela ). However, a specific target of calcium signaling has not been identified.

To reexamine the nature of mid2Δ cell death, we used microscopy and a variety of vital dyes. Previous work utilized methylene blue staining (de Bettignies ), which has been suggested to be membrane permeable, but decolorized in living cells (Passow ). The fluorescent FUN 1 dye (Millard ) is also membrane permeable and stains the cytoplasm of all cells bright-green. In metabolically active cells, FUN 1 accumulates in the vacuole where it is converted into a red fluorescent molecule that forms distinctive aggregates. After exposure to pheromone, most WT cells exhibited weak green fluorescence in the cytoplasm and bright red aggregates in the vacuoles (Figure 1, top). In contrast, the mid2Δ cells showed only bright-green cytoplasm, confirming that these cells are not metabolically active. Note that under these conditions, WT shmoos acquire an enlarged “peanut” shape, because they have been growing in pheromone for 5 h. In contrast, the mid2Δ cells abruptly stopped growing soon after the initial formation of the shmoo projection and are smaller and more pointed (Figure 1, top).

FIGURE 1:

mid2∆ cells lose integrity upon pheromone exposure. WT (MY8092) and mid2Δ (MY14673) cells were treated with 8 μM α-factor for 5 h and Live/Dead stained (top panel), stained with propidium iodide (middle) or FM4-64 (bottom). For Live/Dead staining cells were treated with 15 μM FUN-1 and 25 μM Calcofluor White, incubated in the dark for 30 min and imaged; live (metabolically active) cells produce red fluorescence within the vacuoles, whereas dead cells stain bright yellow. Cells were stained with 20 μM propidium iodide for 15 min and imaged. Propidium iodide is excluded from living cells. FM4-64 stains the entirety of mid2Δ cells, but is restricted to the plasma membrane of WT cells. n > 100 cells. Scale bar = 2 μm.

Because the FUN 1 dye readily crosses intact membranes, it does not report on the integrity of the plasma membrane. Propidium iodide is a membrane impermeant fluorescent dye that binds tightly to nucleic acids (Deere ). Live cells exclude the dye completely, but cells with damaged membranes stain brightly. As expected, the WT shmoos did not stain with propidium iodide (Figure 1, middle). In contrast, the mid2Δ shmoos showed intense diffuse staining, indicating that the plasma membrane was not intact. As a second way to probe the intactness of the plasma membrane, we used FM4-64, a membrane impermeant fluorescent dye that is often used to visualize vacuoles and endocytosis in live yeast (Fischer-Parton ; Yadav ). In intact cells, FM4-64 first stains the plasma membrane, but over time enters the cell by endocytosis and is transported to the vacuole membrane by vesicle trafficking. In WT shmoos, brief exposure to FM4-64 stained only the plasma membrane. In mid2Δ cells, addition of FM4-64 caused immediate bright staining of the entire cell cytoplasm (Figure 1, bottom). We conclude that the mid2Δ cells are losing viability due to loss of plasma membrane integrity, allowing rapid incorporation of FM4-64 and propidium iodide throughout the cell.

mid2Δ pheromone-induced death is due to premature cell fusion

Given that hyperactivation of the CWI pathway causes a block to cell fusion, we hypothesized that loss of an upstream positive regulator might cause pheromone-induced death due to aberrant or overactive fusion. To test this hypothesis, we deleted FUS1 and FUS2 or introduced cell fusion-specific alleles of RVS161 and CDC42 into the mid2Δ background. Cell death in pheromone was then measured using FM4-64 staining. Mutations in all four cell fusion proteins completely suppressed the mid2Δ pheromone-induced death (Figure 2A). Suppression was specific to genes affecting cell fusion; deletion of KAR5, required for nuclear fusion, had no effect. Fus2p and Rvs161p promote cell wall degradation in prezygotes, suggesting that mid2Δ pheromone sensitivity in shmoos is due to the premature activation of the cell fusion pathway, in the absence of a mating partner.

FIGURE 2:

Components of the CWI pathway negatively regulate cell fusion. (A) mid2Δ death is suppressed by loss of cell fusion proteins. WT (MY8092), mid2Δ (MY14673), mid2Δ cdc42-138 (MY15390), mid2Δ rvs161 (MY15620), mid2Δ fus1Δ (MY15618), mid2Δ fus2Δ (MY15081), or mid2Δ kar5Δ (MY15628) were grown and exposed to pheromone as described in Materials and Methods, and dead cells were assessed using FM4-64 staining after 5 h in pheromone. (B) MID2 and WSC1 negatively regulate cell fusion. Each mating is bilateral, with the indicated deletion in both partners (WT with mid2Δ indicates mid2Δ x mid2Δ mating, fus1Δ with wsc1Δ indicates fus1Δ wsc1Δ × fus1Δ wsc1Δ mating). WT (8092 × 8093), mid2Δ (MY14673 × MY14674), wsc1Δ (MY14305 × MY14306), fus1Δ (JY427 × JY430), fus2Δ (MY9181 × MY9185), fus1Δ fus2Δ (MY10935 × JY429), mid2Δ fus1Δ (MY15618 × MY15619), wsc1Δ fus1Δ (MY15499 × MY15500), mid2Δ fus2Δ (MY15081 × MY15080), wsc1Δ fus2Δ (MY15072 × MY15073), mid2Δ fus1Δ fus2Δ (MY16000 × MY16002), wsc1Δ fus1Δ fus2Δ (MY15489 × MY15490) were grown in YEPD, mated for 3 h at 30°C, stained with FM4-64, and imaged. Fusion was counted as any cell that was fully fused or partially fused (some plasma membrane remaining between the mating partners). Error bars denote SEM. n > 100 cells, in three or more independent experiments.

Given that loss of Mid2p may allow premature fusion, and that hyperactivation of Pkc1p blocks fusion, we reasoned that the CWI pathway might generally suppress cell wall removal in prezygotes. If so, then loss of the CWI sensors should partially compensate for the loss of cell fusion pathways during mating. Fus1p and Fus2p act in partially overlapping pathways to promote cell fusion (Trueheart and Fink, 1989; Gammie ; Ydenberg ). Deletion of either FUS1 or FUS2 causes a partial cell fusion block, with a cell wall remaining between the two mating partners (Gammie ). Deletion of either MID2 or WSC1 in both mating partners partially suppressed the fusion defects in fus1Δ × fus1Δ and fus2Δ × fus2Δ matings (Figure 2B) and the effects of mid2Δ and wsc1Δ are additive (Supplemental Figure S2A). Interestingly, suppression occurred under conditions where the downstream CWI protein kinase, Slt2p, did not appear to be active, suggesting that suppression is related to signaling from upstream components of the pathway (Supplemental Figure S2B). However, when cells lacked both FUS1 and FUS2, fusion was completely blocked, and deletion of MID2 or WSC1 was not able to suppress the fusion defect (Figure 2B). These results suggest that the CWI pathway antagonizes cell wall removal; however, loss of negative regulation cannot overcome complete loss of cell fusion–dependent cell wall removal. These observations support the premise that the pheromone sensitivity of mid2Δ is due to premature cell fusion.

mid2Δ cells lose CWI at their shmoo tip

FM4-64 staining showed that mid2Δ cells lose plasma membrane integrity, but did not test whether the cells also lose CWI, as would be expected from premature cell wall removal. The intact appearance of mid2Δ shmoos by transmitted light microscopy suggested that any loss of CWI must be spatially restricted. To address this, we used live cell imaging of mid2Δ cells as they responded to pheromone. The cells contained cytoplasmic GFP to determine when they lost integrity. In typical videos, GFP was rapidly lost soon after the cells formed shmoo projections; beginning at the shmoo tip. In some images, cellular debris was observed being ejected from the shmoo tip, concomitant with loss of cytoplasmic GFP signal (Figure 3A). These observations strongly suggested that cell death was due to loss of cell integrity at the shmoo tip.

FIGURE 3:

Loss of MID2 causes cell wall degradation at the shmoo tip. (A) Representative image of mid2Δ cell lysing at the shmoo tip. mid2Δ cells (MY15807) with cytoplasmic GFP (Jin ) were grown to early log phase and plated on an agar pad made with low calcium media and 8 μM α-factor. Cells were imaged live over time, with images taken 1 min apart. Scale bar = 2 μm. (B) mid2Δ cells have thinner cell walls at the shmoo tip than other CWI or fusion pathway mutants. Box and whisker plot shows the ratio of cell wall thickness at the shmoo tip over the opposite side of the cell. Solid line indicates the median values, and dotted lines indicate mean values. WT (MY8092), mid2Δ (MY14673), fus2Δ (MY9181), mid2Δfus2Δ (MY15081), and wsc1Δ (MY14305) cells were exposed to 8 μM α-factor for 3 h and processed for EM imaging (see Materials and Methods for details). (C) Representative transmission electron microscopy images of the strains in B; sections are ∼80 nm thick (see Material and Methods for details on fixing and staining). n > 68 cells for each strain. Scale bar = 800 nm.

To determine where mid2Δ cells were suffering cell wall removal, we used transmission electron microscopy and found that mid2Δ cells have a thinner cell wall at the shmoo tip (Figure 3B). To correct for differences in the angle of the sections during quantification, we measured the thickness of the cell wall at both the shmoo tip and the base of the cell (opposite the shmoo tip) and determined the ratio (cell/shmoo tip) (Figure 3, B and C). In WT shmoos, the mean ratio was 1 ± 0.49, indicating that the cell wall is normally the same thickness at the shmoo tip as elsewhere. In contrast, mid2Δ cells had significantly thinner cell walls at the shmoo tip than WT cells (mean = 3.0 ± 2, p value = 5 × 10−6, Student’s t test; Figure 3, B and C). In many cells the cell wall at the shmoo tip appeared to be completely removed (Figure 3B). The fus2Δ and wsc1Δ mutants were not significantly different from WT (0.9 ± 0.53 and 1.2 ± 0.69, p value = 0.25 and 0.12). Thus, Fus2p and Wsc1p do not have major effects on cell wall thickness in otherwise WT shmoos. In contrast, the dramatic thinning of the shmoo-tip cell wall in the mid2Δ mutant was completely suppressed by fus2Δ, restoring it to WT (ratio = 1.1 ± 0.55, p value = 0.16). We conclude that the cause of mid2Δ pheromone-­induced death is unregulated/premature cell wall removal specifically at the shmoo tip.

Mid2p negatively regulates Cdc42p localization at the shmoo tip

There are two plausible explanations for mid2Δ death. First, in the absence of Mid2p, cells might not sense the cell wall stress associated with shmoo formation. Without an effective CWI response, cell wall deposition might not keep pace with ongoing cell wall removal at the shmoo tip, mediated in part by the cell fusion machinery. Alternatively, Mid2p and the CWI pathway may directly regulate the Fus2p-dependent cell fusion machinery.

In WT cells responding to pheromone, Cdc42p is uniformly distributed over the shmoo-tip cortex for its general role in polarity. In prezygotes, as cell wall removal begins, Cdc42p forms a bright focus at the center of the ZCF, colocalizing with, and dependent on, Fus2p (Smith ). Because mid2Δ appears to cause premature activation of cell wall removal, we examined Cdc42p localization in these cells. In contrast to WT shmoos, which showed relatively uniform distribution of Cdc42p over the cortex, Cdc42p formed a bright focus at the tip of mid2Δ shmoos (Figure 4A). To quantify localization, we normalized for variable levels of expression in each cell by measuring GFP intensity along a line tracing the shmoo-tip cortex; fluorescence intensity at each point is expressed as a ratio to the average intensity at the base of the shmoo tip (Smith ). Overall, GFP expression in whole cells was comparable in WT and mid2Δ cells (ratio of 0.9 mid2Δ/WT). Cdc42p localization in mid2Δ was significantly higher than WT (p value < 0.02; Figure 4, B and C). Cdc42p localization was also assessed in a second CWI pathway mutant, wsc1Δ. In contrast to mid2Δ, wsc1Δ caused somewhat decreased Cdc42p localization (p value < 0.02; Figure 4, A–C). Thus, mid2Δ specifically causes formation of a prezygote-like localization pattern for Cdc42p.

FIGURE 4:

MID2 negatively regulates Cdc42p localization. (A–C) Cdc42p forms a focus in mid2Δ shmoos, dependent on Fus2p. (A) Representative images of WT (MY15747), mid2Δ (MY15910) and wsc1Δ (MY15912), fus2Δ (MY15717), and mid2Δfus2Δ (MY16006) shmoos with GFP-Cdc42p (Dyer) exposed to pheromone for 1.5 h, fixed with 2% formaldehyde for 10 min and imaged. (B, C) Quantification of GFP-Cdc42p fluorescence intensity in strains described in panel A. (B) Fluorescence intensity was measured along the shmoo-tip contour and plotted as a function of distance. Plots are for each of the individual cells in the representative images in panel A. (C) Combined fluorescence measurements for each of the indicated strains. Fluorescence intensity is expressed as the ratio of each point relative to the intensity measured at the edge of the contour. n > 80 shmoos for each strain. (D, E) Cdc42p and Fus2p colocalize in mid2Δ shmoos. WT (MY15747), mid2Δ (MY15910), and wsc1Δ (MY15912) were transformed with a centromere-based plasmid containing Fus2p internally tagged with mCherry (a functional version of Fus2p, MR5821). Cells were exposed to pheromone for 1.5 h, fixed in 2% formaldehyde for 10 min and imaged. Cdc42p and Fus2 do not colocalize in WT or wsc1Δ shmoos, but do in mid2Δ shmoos. (D) Representative images of shmoos quantified in E; error bars represent SEM. n > 30. (F, G) CWI pathway sensor mutants do not affect Cdc42p localization in prezygotes. (F) Representative images of Cdc42p localization in WT (MY15747), mid2Δ (MY15910), or wsc1Δ (MY15912) cells mated to fus1Δfus2Δ (JY429) for 2.5 h, fixed with 2% formaldehyde, and imaged. Scale bars = 2 μm. (G) Quantification of GFP-Cdc42p fluorescence intensity in strains described in panel F. Fluorescence intensities are plotted as a ratio of the center to the average of the edges. Cdc42p formed a focus in WT, mid2Δ, and wsc1Δ prezygotes. n > 86 for all prezygotes.

If Cdc42p localization in mid2Δ shmoos is a manifestation of premature cell fusion, then Cdc42p should colocalize with and be dependent on Fus2p. Formation of the Cdc42p focus at the shmoo tip in the mid2Δ mutant was completely blocked by fus2Δ, restoring localization to levels indistinguishable from WT (p value = 0.45; Figure 4, A–C). Therefore, Cdc42p shmoo-tip localization in a mid2Δ cell is Fus2p dependent. Interestingly, in fus2Δ shmoos, there was a small, but significant, decrease in Cdc42p shmoo-tip localization relative to WT (p value < 0.01), with a general increase in localization at more distal cortical regions (Figures 4, A–C and 5, F and G). Thus, a small fraction of the overall localization of Cdc42p at the shmoo tip is affected by Fus2p.

FIGURE 5:

Pkc1p regulates Cdc42p in shmoos and prezygotes. (A, B, C) Cdc42p localization in PKC1* prezygotes is partially Fus2p dependent. (A) Representative images of Cdc42p localization in WT (MY15747), PKC1* (MY15902), fus2Δ (MY15717), and PKC1*fus2Δ (MY16005) mated to a fus1Δ fus2Δ (JY429) strain. (B) Fluorescence intensity measurements for each individual prezygote in panel A measured across the ZCF. (C) Combined fluorescence measurements across the ZCF, plotted as the ratio of intensity at each point relative to the edges; n > 91 for all prezygotes. (D, E) Cdc42p and Fus2p do not colocalize in PKC1* prezygotes. Representative images and quantification of WT (MY15747) and PKC1* (MY15902) cells transformed with a centromere-based plasmid containing Fus2p internally tagged with mCherry (MR5821) mated to a fus1Δ fus2Δ (JY429) strain; error bars denote SEM of three independent experiments. (F, G) PKC1* increases Cdc42p localization to the shmoo tip. (F) Representative images and (G) quantification of Cdc42p focus at the shmoo tip. PKC1* Cdc42-GFP (MY15902) has a Cdc42p focus at the shmoo tip compared with WT (MY15747). The PKC1* Cdc42p focus is partially Fus2p dependent with decreased focus formation in fus2Δ (MY15717) and PKC1* fus2Δ (MY16005); WT and fus2Δ data shown in Figure 4, B and C, as well. Fluorescence intensity was measured as described in Smith ; fluorescence measured around the shmoo tip and plotted as a function of distance. Cells were incubated with pheromone for 1.5 h, fixed with 2% formaldehyde, and imaged. n > 86. (H, I) Cdc42p and Fus2p do not colocalize in PKC1* shmoos. Quantification and representative images of Cdc42p and Fus2p colocalization in PKC1* shmoos. WT (MY15747), PKC1* (MY15902), fus2Δ (MY15717), and PKC1* fus2Δ (MY16005) cells were exposed to pheromone for 1.5 h, fixed in 2% formaldehyde for 10 min, and imaged. n > 30. Scale bar = 2 μm.

A hallmark of the Cdc42p focus at the ZCF is the precise colocalization with Fus2p (Smith ). In contrast, in WT shmoos Cdc42p and Fus2p are adjacent but not colocalized; in most cells (58%) there is no overlap between the proteins. This can be seen also by the average separation between the peaks of fluorescence intensity measured along an axis orthogonal to the cortex (Figure 4, D and E; Smith ). In the majority of mid2Δ shmoos (74%) Cdc42p and Fus2p were colocalized, with average peak intensities within ∼0.1 µm (Figure 4, D and E). Colocalization was specific to mid2Δ; wsc1Δ shmoos were not significantly different from WT (Figure 4, D and E).

To determine whether these phenotypes carried over into prezygotes, we assessed Cdc42p localization during mating. We used fus1Δ fus2Δ partners to slow cell fusion and allow capture of this transient event (Smith ). In prezygotes, both mid2Δ (p value = 0.49) and wsc1Δ (p value = 0.62) showed a Cdc42p focus similar to WT, as expected from the lack of phenotypes in prezygotes (Figure 4, F and G). Moreover, mid2Δ fus2Δ prezygotes did not form a Cdc42p focus (Supplemental Figure S3). We infer that deletion of MID2 and WSC1 suppressed the fusion defects by limiting the cell’s ability to rebuild the cell wall, rather than by increasing cell wall removal.

Deletion of downstream components of the CWI pathway, SLT2 and RLM1, also shows some pheromone-induced death, 49 and 31% respectively, after 5 h in pheromone. However, unlike mid2Δ, slt2Δ pheromone-induced death is suppressed by the addition of osmotic support (1M sorbitol) to the media (Errede ). Suppression of slt2Δ by sorbitol suggests that death is due to reduced cell wall deposition during shmoo formation in the absence of an intact CWI pathway.

Pkc1p affects Cdc42p localization in shmoos and prezygotes

Hyperactive Pkc1p causes a moderate block to cell fusion (∼20%; Philips and Herskowitz, 1997). Given the effect of mid2Δ, we reasoned that hyperactive PKC1* might negatively affect Cdc42p localization in prezygotes. When a PKC1* strain with GFP-Cdc42p was mated to a fus1Δ fus2Δ strain there was a significant decrease in Cdc42p at the ZCF compared with WT (p value = 0.047; Figure 5, A–C). We previously showed that curvature of the ZCF induced by fps1Δ has a negative effect on Cdc42p localization in prezygotes (Smith ). Alone fps1Δ and PKC1* cause 20–26% fusion failure; together the mutations cause cell fusion failure in 86% of zygotes (Philips and Herskowitz, 1997). This suggests that curvature and the CWI pathway play independent roles in regulating cell fusion.

Given that Cdc42p is recruited by Fus2p to form the focus at the ZCF (Smith ), we asked whether the effect was due solely to decreased recruitment by Fus2p. If so, then fus2Δ should be epistatic to PKC1*, and the fus2Δ PKC1* double mutant should be identical to fus2Δ. As previously reported, the fus2Δ prezygotes showed a large decrease in Cdc42p localization relative to WT (p value < 0.001). The PKC1* fus2Δ prezygotes showed a slight further decrease in Cdc42p localization relative to fus2Δ (Figure 5, A–C). These data suggest that the PKC1*-mediated decrease in Cdc42p localization acts largely through regulation of recruitment by Fus2p. To confirm this finding, we examined their colocalization in prezygotes. In WT, the Cdc42p focus is precisely colocalized with Fus2p at the center of the ZCF (Figure 5, D and E), with peak fluorescence intensities separated by <0.06 µm, on average. In contrast, colocalization was perturbed in the PKC1* prezygotes. Although both Fus2p and Cdc42p localized near the cortex, they were not strongly colocalized; the fluorescence intensity peaks were separated by almost 0.12 µm on average (Figure 5, D and E). Given that Fus2p was properly localized in the PKC1* prezygotes, these observations suggest that Pkc1p may regulate Cdc42p localization by affecting the interaction with Fus2p.

We next examined the effect of PKC1* in shmoos. Because mid2Δ and PKC1* should have opposing effects on the CWI signaling pathway, we expected that Cdc42p would also show reduced localization in shmoos. Surprisingly, the PKC1* mutant showed a bright Cdc42p focus at the shmoo tip, similar to mid2Δ, and significantly brighter than WT (p value = 0.01; Figure 5, F and G). However, because Cdc42p has multiple roles during shmoo formation, and because the PKC1* shmoos do not lyse, we hypothesized that Cdc42p localization might reflect these alternate functions. If so, then increased localization of Cdc42p in PKC1* shmoos should be at least partially independent of Fus2p. Introduction of PKC1* into the fus2Δ mutant caused a significant increase in Cdc42p localization to the shmoo tip, relative to fus2Δ alone (p value < 0.001), although a focus did not form (Figure 5F). Taken together these data suggest that hyperactivation of Pkc1p increases localization of Cdc42p to the shmoo tip, in a manner that is mostly independent of Fus2p. However, when Fus2p is present, the combined effects cause formation of a Cdc42p focus.

It remained possible that Pkc1p positively regulates Fus2p localization (and/or Cdc42p interaction) in shmoos, unlike its negative role in prezygotes. To test this, we examined Fus2p and Cdc42p colocalization in PKC1* shmoos. As previously reported, Cdc42p did not colocalize with Fus2p in WT shmoos (Figures 4, D and E, and 5, H and I; Smith ). In PKC1* shmoos, Cdc42p also did not colocalize with Fus2p (Figure 5, H and I). These observations imply that Pkc1p regulates Cdc42p localization independent of recruitment by Fus2p, and that, in shmoos, Fus2p plays an indirect role in Cdc42p localization. Taken together with the decrease seen in wsc1Δ, we conclude that Cdc42p localization to the shmoo tip is driven in part by cues from cell wall remodeling, and this response is hyperactivated in PKC1* and different from what occurs in mid2Δ.

DISCUSSION

Negative regulation of cell fusion

Yeast cell fusion requires the concerted removal of the cell wall that separates the mating partners. Errors in the temporal and spatial regulation of cell wall removal would be catastrophic, implying the existence of multiple levels of regulation to insure the fidelity of mating. Previous work has largely focused on positive regulation of cell fusion, identifying multiple proteins required for cell wall removal. Several proteins (Fus1p, Spa2p, Bni1p) are thought to aid in the clustering of vesicles at the ZCF, whereas others are thought to facilitate their fusion with the plasma membrane and the subsequent release of hydrolases (Fus2p, Rvs161p, Cdc42p, and Kel1p; Trueheart ; Trueheart and Fink, 1989; Gammie ; Paterson ; Ydenberg ; Smith and Rose, 2016).

To mate, cells must sense when they are in contact with a mating partner to initiate cell wall removal. At the same time, cells monitor damage to their cell walls and respond by regulating proteins involved in repairing damage (Kono ). The CWI pathway is activated in response to stresses on the cell wall during the polarized growth associated with budding and shmoo formation (Buehrer and Errede, 1997; Gray ; Ketela ). However, activation of the CWI pathway should interfere with the cell wall degradation that must occur for cell fusion to proceed. Consistent with this, PKC1*, encoding a hyperactive form of Pkc1p, causes a cell fusion defect (Philips and Herskowitz, 1997). It remained unclear whether the cell fusion defect results from increased cell wall repair or misregulation of the cell fusion machinery. To explore this question, we examined whether the mechanosensory transmembrane proteins, which surveil the cell wall and activate the CWI pathway, act as negative regulators of cell fusion (Dupres ; Heinisch and Dufrene, 2010).

Wsc1p and Mid2p are the sole essential CWI pathway sensors; cells lacking both proteins cannot survive without osmotic support (Rajavel ). Cells appear to grow normally when Wsc1p and Mid2p are the only sensors in the cell, implying that the remaining three sensors are not absolutely required (Supplemental Figure S1). In contrast, Mid2p has a unique function during mating, to regulate cell fusion. As a result, mid2∆ cells die in response to pheromone. In this work, we show that the pheromone-induced death of mid2∆ cells is caused by premature cell wall removal, specifically at the shmoo tip.

It was previously known that mid2∆ cells become sensitive to pheromone when grown in low calcium, but the mechanism remained unclear (Ketela ; Rajavel ). Mid2p contains a putative calcium binding domain (Ono ) and was thought to be similar to Mid1p, a calcium transporter (Iida ). Both proteins were identified in a screen for mutants that were sensitive to pheromone in low calcium, but unlike mid1∆, mid2∆ cells showed no defect in calcium uptake (Iida , 1994). Further confusing the issue, the mid2∆ pheromone-induced death is not suppressed by osmotic support, which suggested that death is not due to a cell wall defect (Rajavel ), but instead reflects aberrant calcium signaling. Although the role of calcium remains unclear, suppression of mid2∆ pheromone-induced death by mutations affecting cell wall removal during mating (Figure 2A) supports the view that death is due to unregulated cell fusion.

Wsc1p localizes to sites of polarized growth during both shmoo formation and vegetative growth. In contrast, Mid2p becomes localized at the shmoo tip only during the pheromone response; otherwise, it is localized around the cell cortex (Delley and Hall, 1999; Ketela ; Rajavel ; Hutzler ). If both sensors are located at the shmoo tip, why does wsc1∆ not cause pheromone-induced death? One possibility is that Wsc1p signaling is down-regulated during shmoo formation, so that Mid2p is the sole monitor of the cell wall in shmoos. Thus, a mid2∆ cell would have reduced activity of both major cell wall sensors, which would be a lethal combination. We think this is unlikely for two reasons. First, wsc1Δ decreased Cdc42p localization to the shmoo tip, suggesting that Wsc1p is indeed active. Second, in an assay for downstream activation of the CWI pathway, both wsc1Δ and mid2∆ cause reduced phosphorylation of Slt2p in response to pheromone (to 71 and 41% WT levels, respectively; Supplemental Figure S2B). An alternate possibility is that Mid2p has a unique function during mating, blocking Fus2p localization, possibly independent of its role in CWI signaling. Wsc1p and Mid2p occupy different microdomains within the plasma membrane, perhaps allowing for their differing functions (Kock ). Regardless of the mechanism, Mid2p would prevent premature cell wall removal, and must be down-regulated in prezygotes to allow cell fusion.

Loss of either Wsc1p or Mid2p suppressed the fusion defect seen in fus1∆ × fus1∆ or fus2∆ × fus2∆ matings (Figure 2B), supporting the idea that the CWI pathway negatively regulates fusion and demonstrating that in prezygotes, both sensors are active. Suppression of fus1∆ and fus2∆ further suggests that cell fusion is mediated in part by the balance between cell wall removal and cell wall repair. Loss of either Fus1p or Fus2p would cause a reduced rate of cell wall removal, which can be easily counteracted by the CWI pathway. Reducing cell wall repair, in turn, by loss of a CWI sensor, would allow cell fusion to proceed by the remaining cell fusion machinery. However, some mating-enhanced cell wall removal is still required for fusion; loss of a CWI sensor is not sufficient to drive fusion when cells lack both Fus1p and Fus2p (Figure 2B).

It is not known at what level components of the CWI pathway regulate cell fusion. However, we think it unlikely that regulation entails activation of downstream signaling components. Under high osmolarity conditions, Slt2p phosphorylation was undetectable, yet suppression of fus2∆ was not greatly affected (Supplemental Figure S2, A and B). Therefore, the CWI pathway most likely regulates cell fusion upstream of Slt2p.

Regulation of Cdc42p localization

Cdc42p is recruited to a focus at the ZCF specifically in prezygotes (Smith ). In mid2∆ shmoos, a Cdc42p focus forms at the shmoo tip, dependent on and colocalizing with Fus2p (Figure 4). Given that Fus2p localizes to the ZCF independent of the interaction with Cdc42p (Smith ), and that the proteins do not colocalize away from the ZCF, Mid2p may regulate the localization of Fus2p. Mid2p would thereby prevent the premature assembly of a Fus2p/Cdc42p complex which might initiate improper cell wall degradation.

In prezygotes, Cdc42p localization was found to be affected by curvature at the ZCF (Smith ). When the ZCF is positively curved (projecting outward as in shmoos), Fus2p was localized, but formation of the Cdc42p focus was suppressed. One hypothesis to explain the sensitivity to curvature is that the amphiphysin-like Fus2p/Rvs161p heterodimer adopts a different conformation when it binds to flat membranes, conferring the ability to bind and recruit Cdc42p. An alternative possibility is that Mid2p, as a mechanosensor, may directly signal the changing curvature of the cell wall at the ZCF. In shmoos, and prezygotes with curved ZCFs, Mid2p would negatively regulate Fus2p association with the membrane and/or Cdc42p recruitment, preventing premature cell wall degradation. However, when cells come into contact and form a flat interface at the ZCF, negative regulation by Mid2p would be relieved, allowing for cell wall degradation and fusion to proceed. When mid2∆ cells found a mating partner and formed a prezygote before premature cell wall degradation, Cdc42p localization was WT, and fusion was unimpaired (Figure 4, F and G). It is likely, however, that both positive and negative regulation occur during cell fusion. Altering the curvature of the ZCF and hyperaction of PKC1 have additive effects (Philips and Herskowitz, 1997), suggesting that they may regulate cell fusion by independent pathways.

The role of Pkc1p in regulating Cdc42p localization is complex, apparently having different functions in shmoos and prezygotes. In shmoos, PKC1* caused the formation of a Cdc42p focus, superficially similar to the phenotype of mid2∆. However, unlike mid2∆, the Cdc42p focus did not colocalize with Fus2p (Figure 5, B–D), and PKC1* did not cause lysis in pheromone. Therefore, it is likely that PKC1*-driven localization reflects one of Cdc42p’s several other functions, and not cell fusion–specific cell wall degradation. If so, it is confusing that PKC1*-driven Cdc42p localization is affected at all by Fus2p. One possibility is that, although Fus2p does not form a focus at the cortex, there is some Fus2p-dependent cell wall removal in shmoos. In support, β-1,3 glucan deposition at the tip of fus2∆ shmoos is increased, as assayed by aniline blue staining (Fitch ). In addition, fus2∆ caused reduced Cdc42p localization to the shmoo tip in otherwise WT cells (Figure 4C). We speculate that Cdc42p localization in shmoos partially reflects cell wall assembly, as part of normal morphogenesis and the response to CWI signaling of cell wall degradation (Buehrer and Errede, 1997). In the presence of hyperactive Pkc1*p, some Fus2p-dependent cell wall degradation may lead to an exaggerated CWI response, observed as a large increase in Cdc42p localization.

In contrast, in prezygotes PKC1* caused decreased localization of Cdc42p to the ZCF. Decreased localization of Cdc42p, together with up-regulation of the CWI pathway, may explain the cell fusion defect caused by PKC1*. Interestingly, Fus2p localization to the ZCF occurred normally in PKC1* prezygotes, suggesting that Pkc1p may regulate the interaction between Fus2p and Cdc42p. In contrast, Mid2p appeared to regulate the localization of Fus2p. Thus, the different components of the CWI pathway may regulate cell fusion by different mechanisms. It is interesting to note, as well, that during fission yeast mating, cells show a similar focus formation driving cell wall removal that is dependent on a MAPK cascade (Dudin ). Fission yeast also show a dynamic patch of active Cdc42p at polarity sites; however, it was not possible to distinguish a role in fusion from roles in polarization and signaling (Merlini ).

Understanding the role of cell wall degradation in yeast mating provides general insights into other cell fusion events. Even cells in higher eukaryotes must remove the intervening extracellular matrix to allow plasma membrane fusion with neighboring cells. Given the importance of such events and the detriment of improper fusion, it is likely that cells employ multiple levels of regulation to ensure that fusion proceeds safely. One general component of cell fusion regulation may be the application of a “brake” under conditions of cell surface stress or before successful engagement with a fusion partner.

MATERIALS AND METHODS

General yeast techniques

Yeast media, general methods, and transformations were performed as described previously (Amberg ) with minor modifications. Strains and plasmids are listed in Supplemental Tables 1 and 2. All strains and plasmids are available upon request. Deletion strains were either created via PCR amplification of selective markers and homologous recombination at the locus of interest, or via crossing of various deletion strains followed by sporulation and tetrad dissection. All strains were grown at 30°C. For pheromone induction experiments, early exponential cells growing in selective media were treated for 90 min (unless otherwise specified) with synthetic α-factor (Department of Molecular Biology Syn/Seq Facility, Princeton University) added to a final concentration of 8 μg/ml.

For cells grown in low calcium media, yeast nitrogen base was made according to the formula from BD Sciences (San Jose, CA), but replacing all calcium salts with equivalent sodium salts.

Microscopy

All images were acquired at 23°C using a deconvolution microscopy system (DeltaVision; Applied Precision, LLC) equipped with an inverted microscope (TE200; Nikon) and a 100× objective with numerical aperture of 1.4. Deconvolution and image analysis were performed using Precision softWoRx and ImageJ (National Institutes of Health).

For imaging with propidium iodide, cells were prepared as described above and imaged after 3 h in pheromone. Cells were spun down and resuspended in phosphate-buffered saline (PBS), propidium iodide was added to a final concentration of 20 µm, and cells were incubated for 15 min at 23°C in the dark before imaging.

The Live/Dead Yeast Viability Kit (Invitrogen) was used according to protocol. Cells were prepared as above after 3 h in pheromone. A final concentration of 15 µm FUN-1 and 25 µm Calcofluor White were used before imaging.

Microscopic assays of FM4-64–stained mating mixtures and pheromone-induced cells were performed as described previously (Grote, 2008). Pheromone-induced cells were prepared as described and then resuspended in 1 ml of TAF buffer (20 mM Tris-HCl, 20 mM NaN3, 20 mM NaF in water) and kept on ice. FM4-64 (Molecular Probes/Invitrogen) was added to cells to a final concentration of 4 μM and stained shmoos were imaged as above. For fusion assays, FM4-64 staining was used. An equal OD600 (0.5) of each mating type in log phase was mixed, concentrated on 25 mm 0.45 μm nitrocellulose filter disks (Millipore), and incubated on rich media plates for 2.5–3 h at 30°C.

For imaging of pheromone-induced cells with fluorescent proteins (Figures 4 and 5), cells were induced with pheromone in selective synthetic media for 90 min, fixed for 10 min with 2% formaldehyde at 30°C, washed in 1 X PBS, and then imaged in a smaller volume of PBS. A two-tailed, two-sample unequal t test of central peak values was used to determine significance.

For imaging of prezygotes containing fluorescent proteins, mating mixtures were prepared as described above, resuspended in synthetic media, fixed for 10 min with 2% formaldehyde at 30°C, and imaged as above. Quantification of fluorescence was performed using ImageJ software (Schindelin , 2015). Fluorescence intensity was measured along a line contour corresponding to the ZCF or shmoo tip as described in Figure 1 of Smith . For prezygotes the contour extends the length of the ZCF; for shmoos the ends of the contour were defined by the middle of the inward curve of the shmoo neck. To account for cell to cell variation, fluorescence is expressed as the ratio of intensity divided by the average values for the 5 pixels (0.3 μm) at each end of the contour. The average end values represent regions in which Cdc42p did not appear to be concentrated. To combine data, contour measurements across the shmoo tip or ZCF were centered relative to one another to account for cell to cell variation in length. Two-tailed, heteroscedastic t tests were used to obtain p values for localization curves. Quantification of colocalization of Cdc42p and Fus2p was performed by quantifying fluorescence across a line perpendicular to the shmoo tip or ZCF using the ImageJ software. The peak was defined as the brightest pixel in the central focal plane for each cell. Error bars represent the standard error of the mean for all cells imaged. A two-tailed, two-sample unequal t test was used to determine significance.

Live imaging was performed by placing 0.04 OD600 of mid2∆ cells on a 2% agarose pad containing 8 μg/ml α-factor at 23°C. Images were taken at 1-min intervals using the DeltaVision system described above.

Electron microscopy

Cells were prepared for transmission electron microscopy as described in Gammie and Rose, 2002. Briefly, the workflow included a glutaraldehyde fixation, potassium permanganate staining, sodium periodate treatment, uranyl acetate staining, and embedding in LR White resin. Specifically, 40 ml of mid–log phase cells were grown in synthetic media without calcium, induced with 8 μg/ml α-factor at 30°C for 3 h, spun down, and fixed in 2% glutaraldehyde for 30 min at room temperature. Cells were stained with 4% potassium permanganate for 4 h at 4°C. Ultrathin sections (∼80 nm) were placed on a nickel slotted grid (Formvar film, FF-2010-Ni; Electron Microscopy Sciences, Hatfield, PA) and imaged directly, without lead citrate staining.

Click here for additional data file.